Surgical-site infection after anterior cruciate ligament (ACL) reconstruction accounts for about 1% of cases. Symptoms most often appear in postoperative weeks 2–3; the earliest published cases began on day 3. Joint swelling, worsening pain, and regression of range of motion are equally common in patients who recover slowly, have hemarthrosis, or are overloaded, and fever is present in only some of those with infection. Results such as C-reactive protein (CRP) must be interpreted against the distribution among uncomplicated patients at the same postoperative time point. A single CRP value in the first postoperative week cannot separate infection from the surgical response; judgment at that stage rests mainly on the wound, effusion, and systemic findings. Beyond the second week, CRP that remains clearly elevated, or that rises again after falling, requires a search for a cause. A normal CRP, a normal peripheral white blood cell count, a low synovial fluid white cell count, or a single negative culture cannot alone exclude infection. When infection is clinically suspected, synovial fluid should be obtained by aspiration without delay; management of acute infection does not wait for a second CRP or for culture results. Published CRP and synovial fluid white cell cutoffs in the ACL reconstruction population all have design or statistical problems that affect their conclusions and cannot be used as decision thresholds for an individual patient.
1. Interpreting CRP by Postoperative Interval; Prompt Aspiration When Infection Is Suspected
The decision sequence below was assembled by the author from the material in this chapter. It is not a guideline recommendation and has not been prospectively validated. It treats the upper tail of published uncomplicated cohorts as a prompt for rechecking and further investigation, neither as a threshold that confirms infection nor as one that excludes it. The ESSKA and EBJIS recommendations of 2023 state the point directly: a normal CRP cannot exclude infection, and an elevated CRP cannot confirm it[64]. The sequence is framed for primary reconstruction with a hamstring autograft. CRP around day 3 may be higher after bone–patellar tendon–bone (BPTB) grafts, and is also higher in patients who undergo concurrent microfracture, those whose preoperative CRP is already elevated, and men (Sections 2 and 3); revision and meniscal surgery showed no independent effect in the multivariable analyses of the Ruiz-Ibán and Uchida cohorts. The infection-case values in this chapter come from series that mix BPTB, revision, and pediatric patients and are not the same population as the uncomplicated cohorts.
Management when infection is suspected. Worsening or persistent pain (including night pain), recurrent or persistent effusion, fever, stalled recovery of range of motion, locally elevated skin temperature, and drainage from the incision or portals—any of these at any time point warrants prompt examination of the wound, the joint, and the systemic status, regardless of the CRP at that moment; these findings are not specific, and a large noninfectious hematoma can produce them as well[64]. When examination raises suspicion of infection and an effusion is present, aspirate promptly; send the fluid for microbiology, cell count with differential, and crystal analysis. In patients without sepsis, antibiotics are deferred until after sampling so as not to interfere with culture; in patients with sepsis, antimicrobial therapy is not delayed for aspiration.
When acute infection is suspected, begin empiric antibiotics immediately after sampling and proceed to arthroscopic irrigation and debridement as soon as clinical suspicion is established, without waiting for culture or white cell count results, and without waiting for a second CRP. When chronic infection is suspected, antibiotics may be started after intraoperative sampling[64,69]. Synovial fluid from wound or harvest-site infection may be nearly normal; the wound itself must be examined directly (Section 6). The CRP values discussed by interval below are intended for patients in whom clinical suspicion is not high and who are systemically stable.
Postoperative days 1–3. After uncomplicated reconstruction CRP rises universally. In cohorts that report day-3 values, the mean peaks around day 3: hamstring cohorts, day-3 means of 29.4 and 47.3 mg/L; BPTB cohorts, 64.7 and 95.3 mg/L; individual patients may exceed 100 mg/L[31-33]. The earliest published infection cases began on postoperative day 3[5,33]; most present around two weeks. At this stage the height of a single CRP mainly reflects the surgical response and cannot distinguish infection. A second value one or two days later shows the direction of change. The plasma half-life of CRP is about 19 hours[22]; after the stimulus subsides, concentration should fall day by day, but postoperative synthesis does not stop abruptly, and the rate of decline is also affected by the timing of blood draw, resorption of hematoma, rehabilitation load, and medication. There is no standardized daily decrement (Section 2). Failure to fall after the peak, or continued rise after day 3, indicates ongoing stimulus; re-examine the wound and effusion and measure temperature, without waiting for the second week. Early, highly virulent infection is uncommon but has been reported: a young man after hamstring reconstruction with repair of both menisci became ill in the first postoperative week (the Abstract and case presentation say day 5; Table 1 and the Discussion say day 3), with fever and wound erythema and swelling, CRP 183.6 mg/L (original 18.36 mg/dL), synovial fluid white cells 62,667/µL, and culture growing Klebsiella pneumoniae[75]. Such infections are recognized by symptoms and wound findings, not by the CRP trajectory.
Postoperative days 5–7. Mean CRP in hamstring cohorts falls to 10–23 mg/L; the upper tail remains wide. In the Uchida cohort on day 7, 24/439 (5.5%) reached 30 mg/L, with a maximum of 114.8 mg/L[35]; in the Ruiz-Ibán cohort on day 7 (±3 days), among 113 samples, 11.5% exceeded 25 mg/L and 4.4% exceeded 50 mg/L[34]. Values above 50 mg/L on day 7 occur in uncomplicated patients; published data contain no single CRP value that separates infection from normal recovery in this interval. Effusion is common at this stage and is usually bloody. Among Paci’s 31 noninfected patients aspirated for painful effusion, at a mean of 5.5 days after surgery mean CRP was 61 mg/L, synovial fluid white cells reached as high as 62,000/µL, and the neutrophil fraction could exceed 80%; in the 6 cases with white cells above 16,200/µL, red cells all exceeded 2,000,000/µL[37]. This is a symptom-selected population and does not represent the distribution of routine recovery. In grossly bloody synovial fluid, some of the white cells are carried in with blood; the original authors advised reliance on clinical judgment when hemarthrosis is obvious, gave no correction formula, and no validated correction method exists after ACL reconstruction. Differential count, crystals, Gram stain, and culture must then be judged together with the clinical course. In patients without wound abnormality, fever, or systemic symptoms who are otherwise stable, CRP may be repeated within 24–48 hours with re-examination of the wound and effusion; when effusion is substantial, the author’s practice is to aspirate and send fluid for testing—this is the author’s habit and is not supported by comparative studies as a routine. When examination raises suspicion of infection, manage as above.
Postoperative day 14. The cohorts come closest together at this time point. In the Uchida cohort of 439, 1 patient reached 30 mg/L and none reached 50 mg/L[35]; in the Ruiz-Ibán cohort, among 107 samples on day 14, none exceeded 25 mg/L[34]. The Margheritini cohort on day 15 was 11.2±9.5 mg/L, and the Calvisi cohort on day 15 had a mean of 7.4 mg/L; neither study reports the proportion above 30 mg/L. From 11.2±9.5 it is entirely possible that individual patients exceed 30 mg/L, and Calvisi’s Figure 2 at this time point also shows an outlier that reads above 30 mg/L on the graph[31,32]. Symptoms of infection most often appear around this time point (Section 5). The author investigates patients whose CRP remains above 30 mg/L on day 14 as suspected infection until examination provides another explanation. This is the author’s practice and has not been validated: the four cohorts differ in assay methods and populations and cannot be pooled into a single denominator; 30 mg/L is also not an exclusion threshold—infections presenting with CRP below this level have likewise been reported (Section 5).
Postoperative days 21–30. In the Ruiz-Ibán cohort, among 101 samples on day 21, 2.0% still exceeded 25 mg/L; among 102 samples on day 28, none exceeded 25 mg/L, but 23.5% exceeded 5 mg/L[34]. Cohort means on days 28–30 were 3.4–6.1 mg/L[31,32,34]. A value above 25 mg/L on day 30 already exceeds every observation in Ruiz-Ibán at that time point; the other cohorts do not report proportions in this bracket. Subacute, low-virulence infection may not be diagnosed until 1–2 months after surgery; a single negative culture cannot exclude it. Beyond 1 month, only one Indian randomized trial (90 randomized, 87 completed) has reported medians and interquartile ranges (Table 3): at 4 weeks CRP medians were 3.8–4.7 mg/L, and at 12 weeks ESR medians were 7–10 mm/h, without proportions above any given value[38]. In the author’s view, CRP still above 10 mg/L beyond 1 month, or ESR still above 20 mm/h at 3 months, requires an explanation; these two figures come from the author’s clinical experience and have no published distribution behind them.
Any time-point rise after a fall. A CRP that has fallen and then rises again is more informative than any absolute value; Margheritini already made this point in 2001[31]. Falls during rehabilitation, joint aspiration, venous thrombosis, and upper respiratory or urinary tract infection can all raise CRP again and should be asked about and examined for one by one (Section 6).
Synovial fluid. Synovial fluid white cell count and neutrophil fraction are the most commonly used means of distinguishing bloody effusion from infection, but in the first postoperative week the two overlap substantially. A low count likewise cannot alone exclude infection: in the Costa cohort, one case of Staphylococcus aureus infection on postoperative day 10, without antibiotics before aspiration, had a synovial fluid white cell count of 1,765/µL[65]. Existing synovial fluid cutoffs all come from small samples and from different derivation populations. Before interpreting a negative culture, confirm whether antibiotics were given before aspiration and how long before (Section 5).
Other blood markers. Peripheral white blood cell count is normal in most infected patients; ESR in the early postoperative period overlaps with normal recovery; procalcitonin in culture-positive joint infection can be as low as 0.05 ng/mL (Sections 4 and 5). Markedly elevated procalcitonin suggests a stimulus outside the surgery; normal results for these three markers cannot cancel an already formed clinical suspicion.
Imaging. MRI can show complex layered effusion, fluid within tunnels, peri-tunnel bone resorption, focal soft-tissue fluid collections, and sinus tracts, and can be used to choose aspiration sites and the extent of debridement. Available material cannot show that MRI distinguishes infection from hematoma: the only MRI study of infection after ACL reconstruction had no noninfected controls, and the proportions of effusion and peri-tunnel edema in uncomplicated early knees are likewise unreported[13].
Tests most likely to change the judgment. In patients in whom clinical suspicion is not high, the tests most likely to change judgment are a second CRP 24–48 hours later; synovial fluid cell count, differential, crystals, and culture; and checks for alternative explanations—wound and subcutaneous hematoma, calf swelling and venous ultrasound, pulmonary and urinary symptoms, newly started medications—and, when needed, MRI of tunnel and effusion morphology. In the author’s experience, most patients can be assigned after these checks either to continued observation or to management as infection. For the remainder who remain difficult to judge, available material provides no data-supported sequence of next steps: the consequences of infection and of delayed diagnosis have been quantified; the cost of misclassifying a case as infection has not (Section 5). “Treating as suspected infection is the lower-cost choice” is a common clinical habit, not evidence.
2. CRP Changes After Uncomplicated Reconstruction
Kinetics of CRP
CRP is synthesized by hepatocytes (other local sites of synthesis have also been reported); transcription is driven mainly by IL-6. After a single stimulus, serum concentration exceeds 5 mg/L by about 6 hours and peaks around 48 hours. The plasma half-life is about 19 hours and is unchanged in healthy persons and across disease states, so circulating concentration depends only on the rate of synthesis[21]. This half-life comes from the in vivo clearance study of Vigushin et al. using ¹²⁵I-labeled CRP: in 8 healthy subjects the mean (SD) was 18.8 (3.9) hours, and patients with active rheumatoid arthritis, bacterial infection, chronic osteomyelitis, and malignancy did not differ from healthy subjects (ANOVA P=0.382)[22]. The same paper cites prior clinical observations that elevated CRP falls by at most 50% per day after the stimulus subsides[22]. If synthesis stopped completely, pure exponential clearance with a 19-hour half-life would lower concentration by about 58% in 24 hours; the actual postoperative decline is slower than this, indicating that synthesis continues. Neither 50% nor 58% is a daily decrement that patients must meet. In healthy blood donors the median CRP is 0.8 mg/L, the 90th percentile 3.0 mg/L, and the 99th percentile 10 mg/L; in the general population the median is slightly higher than in donors and rises with age and BMI[21].
These facts are the premise for reading postoperative data. The peak on postoperative days 2–3 relates to the amount of tissue injury in the first two days; that relationship comes from graded data in fracture and arthroplasty (see the end of this section) and has not been measured directly in ACL reconstruction. Slow or absent decline after the peak indicates only that stimulus continues; it does not by itself support a diagnosis. Patients with a higher baseline (obesity, older age) may also have higher absolute postoperative values; among ACL patients only Uchida’s BMI data support this point (Section 3).
Five Direct Studies
Five studies have reported CRP at successive time points after uncomplicated ACL reconstruction, totaling 744 reconstructions (Ruiz-Ibán: 119 reconstructions in 117 patients), published 2001–2024, from Rome, L’Aquila, Beijing, Madrid, and Osaka. Their blood-draw schedules, grafts, assay methods, and summary statistics differ (Tables 1 and 2).
Table 1. Design of the five uncomplicated cohorts
| Study | n | Graft | Blood-draw schedule | CRP method and lower limit of detection | Reporting |
|---|---|---|---|---|---|
| Margheritini 2001, Rome[31] | 45 | All BPTB, single surgeon | Preoperative; postoperative days 1 (or 2), 3, 7, 15, 30 | Immunoturbidimetry; normal <6 mg/L | Mean±SD |
| Calvisi 2008, L’Aquila[32] | 58 | BPTB 13, hamstring 45 | Preoperative; postoperative days 1, 3, 7, 15, 30 | Immunoprecipitation; lower limit 0.6 mg/dL | Mean±SD (mg/dL) |
| Wang 2014, Beijing[33] | 83 | All hamstring, 5 surgeons | Postoperative days 3, 5 | Method not stated; normal 0–10 mg/L | Mean± (SD or SE not specified) |
| Ruiz-Ibán 2015, Madrid[34] | 119 | Soft-tissue grafts (autograft 107, allograft 12), 6 surgeons | Preoperative; postoperative day 1; days 7, 14, 21, 28 (last four each ±3 days) | Sensitivity 0.01 mg/L; normal <5 mg/L | Mean (SD); also proportions above 5, 25, 50 mg/L |
| Uchida 2024, Osaka two centers[35] | 439 | Hamstring 209, BPTB 230 | Preoperative; postoperative days 1, 7, 14 | Latex agglutination; lower limit 0.05 mg/dL | Mean±SD and range (mg/dL) |
Table 2. CRP at each time point in the five cohorts (converted uniformly to mg/L)
| Time point | Margheritini[31] | Calvisi overall / BPTB / hamstring[32] | Wang[33] | Ruiz-Ibán[34] | Uchida[35] |
|---|---|---|---|---|---|
| Preoperative | <6 (entry criterion; Figure 1 has points, no values) | ≤6 (below detection limit) | <8 (entry criterion) | 1.8 (2.6) | 0.6±1.9, range 0–26.9 |
| Day 1 | Figure 1 has points, no values | 17.8±12.6 / 23.8±15.5 / 16.0±11.2 | — | 8.5 (11.6) | 5.4±5.8, range 0–43.8 |
| Day 3 | 95.3±51.6 (peak) | 37.3±27.1 / 64.7±24.4 / 29.4±22.4 (peak) | 47.3±5.5 | Not drawn | Not drawn |
| Day 5 | — | — | 16.9±2.4 | — | — |
| Day 7 | Below day 3 (Figure 1, no values) | 23.2±19.9 / 29.9±23.8 / 21.3±18.5 | — | 10.5 (17.0) (mean peak) | 9.2±12.2, range 0–114.8 (mean peak) |
| Days 14–15 | 11.2±9.5 | 7.4±5.0 / 7.4±5.0 / 7.4±5.1 | — | 4.5 (3.4) | 2.3±3.5, range 0–31.1 |
| Day 21 | — | — | — | 4.4 (7.6) | — |
| Days 28–30 | 4±1 | 6.1±0.8 / ≤6 / 6.1±0.9 | — | 3.4 (3.0) | — |
Notes to Table 2: Calvisi and Uchida originally reported mg/dL; values here are multiplied by 10. Ruiz-Ibán’s Abstract writes mg/ml; the body and laboratory description use mg/L; this table uses mg/L. For Wang, the figure after “±” is not specified in the original as SD or SE; see the Wang 2014 subsection below.
Taken together, CRP rises after every uncomplicated reconstruction, with the peak usually around day 3. In the three studies that cover day 3, hamstring cohort day-3 means were 29.4 and 47.3 mg/L, and BPTB means 64.7 and 95.3 mg/L; the three differ in assay method, the hours encompassed by “day 3,” and concurrent procedures. Calvisi’s BPTB group had only 13 patients, Wang’s dispersion figure is questionable, and between-technique differences can be taken only in direction. Day-7 means were 9.2–23.2 mg/L, days 14–15 2.3–11.2 mg/L, and days 28–30 3.4–6.1 mg/L. Individual variation far exceeds differences among means: SDs commonly approach or exceed the mean, and the Uchida day-7 range was 0–114.8 mg/L. None of the five studies reported P90 or P95, and all reported only postoperative day, without recording hours from the end of surgery to blood draw. The elapsed time corresponding to the same “day 3” varies with counting method and clock time of sampling; CRP is near its peak in that window, and earlier or later sampling is enough to change whether a given patient is “above threshold.” No data set yet meets all of the following: hamstring graft, consistent surgeon and perioperative medication, sample size over 100, coverage of day 3 and day 30, a modern assay with a low detection limit, and values compatible with existing cohorts. Data from China are limited to Wang 2014’s 83 cases on days 3 and 5. These gaps are the reason this book argues for rebuilding reference distributions by percentile and by exact postoperative hour (Section 7).
Margheritini 2001: BPTB Cohort, Highest Day-3 Peak
Forty-five consecutive patients (22 men, 23 women; mean age 28 years), all with chronic isolated ACL injury, underwent fully arthroscopic reconstruction with patellar tendon by a single surgeon; epidural anesthesia, cefazolin 2 g preoperatively and 1 g every 6 hours for 2 days postoperatively, drains removed within 2 days, NSAIDs withheld, cryotherapy used. Entry required preoperative CRP <6 mg/L and ESR ≤15 mm/h; diabetes and other metabolic disease, prior knee surgery, and concurrent lesions such as meniscal injury or synovitis were excluded; 6-month follow-up was free of complications[31]. CRP peaked on day 3 and was 4±1 mg/L on day 30, not statistically different from preoperative (P=0.286); ESR peaked on day 7 at 68.3±21.0 mm/h and on day 30 was still 17.4±9.8 mm/h, above preoperative (P<0.001). The authors considered that CRP not approaching normal beyond 2 weeks, or rising again, should raise suspicion of a serious complication and prompt further investigation[31].
This is the only cohort with a single surgeon and a single technique that also covers both day 3 and day 30. It represents BPTB reconstruction with every concurrent procedure excluded; patients who undergo concurrent meniscal repair or microfracture cannot take its values directly. The day-3 SD exceeds half the mean; mean±SD does not describe the upper tail of the distribution. CRP is clearly right-skewed; the paired t-test P values in the original can serve only as rough reference.
Calvisi 2008: BPTB–Hamstring Difference Not Tested Between Groups
Fifty-eight consecutive patients (46 men, 12 women; age 15–52 years, median 25), all with chronic ACL insufficiency more than 6 weeks after injury; concurrent meniscal or focal cartilage lesions were allowed. Acute tears, Outerbridge grade III–IV diffuse cartilage injury, grade III concurrent ligament injury, revision, abnormal preoperative CRP, and infection, tumor, rheumatologic, and hematologic disease were excluded. BPTB 13, hamstring 45; no infection at 6 months; blood draws were timed to discharge and to brace-unlock follow-up visits[32]. Overall, CRP peaked on day 3 and by day 15 no longer differed from preoperative (P=0.18); the shape of the two curves was the same[32].
The authors wrote that the higher CRP after BPTB “may be related to greater harvest trauma.” In that study this is only a description: the paper performed only within-group Friedman tests over time, with no between-group comparison and no confidence interval for the difference. The day-3 mean difference of 35.3 mg/L (64.7 minus 29.4) can be calculated directly from Table I; its uncertainty is not given in the original. Two points can be used: the hamstring day-3 mean of about 30 mg/L is clearly lower than Margheritini’s BPTB cohort; and although the day-15 mean no longer differed statistically from preoperative, Figure 2 at that time point still shows several outliers beyond ±2 SD, one of them above 30 mg/L (read from the figure; the original gives no value)—not everyone returns to the detection limit. The study’s detection limit was 6 mg/L; preoperative values were uniformly recorded as “≤6,” so “return to baseline” means no difference from an undetectable preoperative value. It cannot distinguish 1 mg/L from 5 mg/L and therefore cannot judge whether 8 mg/L on day 30 is a departure[32].
Wang 2014: Day-3 and Day-5 Values Usable; the 41 mg/L Cutoff Is Not
Peking University Third Hospital reviewed 1,054 arthroscopic ACL reconstructions from August 2007 to July 2008; only 84 had both day-3 and day-5 CRP and ESR. After excluding 1 concurrent PCL reconstruction, 83 formed the noninfection group (59 men; mean age 27.4 years): all hamstring autograft, 5 senior surgeons, tourniquet and spinal anesthesia, postoperative antibiotics for 3–5 days, preoperative CRP <8 mg/L and ESR <20 mm/h. Day-3 CRP was 47.3±5.5 mg/L and day-5 16.9±2.4 mg/L (P<0.01); ESR was 22.7±3.0 mm/h on day 3 and 22.3±3.3 mm/h on day 5, with no difference between the two days[33]. The infection group comprised 39 cases of septic arthritis (0.51%) among 7,679 reconstructions from 1997–2010, defined as positive synovial fluid culture, or cell count >10,000/mL with matching symptoms; 30 (76.9%) were culture-positive, including coagulase-negative staphylococci in 20 and Staphylococcus aureus in 6. Apart from 4 chronic infections, symptoms appeared at a mean of 13.7±6.0 days (range 3–26) after surgery; pretreatment CRP was 101.9±18.0 mg/L and ESR 57.1±8.2 mm/h. Table 1 lists infection-group temperature 38.8°C (37.9–40) and noninfection-group 37.0°C (36.4–37.6), with a footnote stating that both groups were measured on postoperative day 3, yet infection-group symptoms appeared at a mean of day 13.7; the original does not clarify the infection group’s measurement day. These temperatures cannot be used as evidence that infected patients are already generally febrile on postoperative day 3[33].
Two reservations apply to the values in Table 2. Neither group’s “±” is specified as SD or SE. If the noninfection group’s day-3 CRP SD were only 5.5 mg/L in 83 patients, that would differ too widely from Calvisi’s hamstring 22.4 mg/L and Margheritini’s 51.6 mg/L and would look more like a standard error; the infection group’s 101.9±18.0 mg/L likewise cannot be taken as dispersion. This chapter records the means as stated. The 83 cases are only 8% of the contemporaneous 1,054; the authors acknowledge this in the Limitations, explaining that most patients had already been discharged by days 3–5. Who remained is not described and need not be those with the smoothest recovery.
On ROC analysis the authors proposed CRP 41 mg/L and ESR 32 mm/h as “optimal cutoffs” (CRP area under the curve 0.992). This cutoff cannot be used to judge an individual patient. The noninfection group used day-5 values; the infection group used “pretreatment” values, and the infection group’s symptoms appeared at a mean of postoperative day 13.7: one side is normal patients on day 5 whose surgical inflammation has not yet settled, the other is infected patients with symptoms around two weeks, and the normal backgrounds of the two groups differ (noninfection day-5 mean 16.9 mg/L; days 14–15 cohort means 2.3–11.2 mg/L)[33]. The figure 0.992 cannot be read as “41 mg/L on day 5 separates infection from normal,” because the infection-group values were not measured on day 5; how large a contribution the time mismatch makes, and in which direction, cannot be calculated from the original data. If applied to day 3, the noninfection mean of 47.3 mg/L already exceeds 41 mg/L; what proportion exceeds it is not given as a distribution in the original. If applied to day 14, 1/439 in the Uchida cohort reached 30 mg/L; against that distribution, 41 mg/L is generous for day 14[35].
The original has several further internal inconsistencies. Sensitivities in Table 3 back-calculate to a denominator of 34 rather than 39 (e.g., 0.941=32/34, 0.706=24/34), and specificities back-calculate to 82 rather than 83 (e.g., 0.488=40/82, 0.976=80/82); the implied analysis counts do not match the table headers of 39 and 83, and the source of the difference is not explained. The cutoff was derived and evaluated in the same set of patients, with no validation set. The Discussion also swaps the false-negative and false-positive rates for 41 mg/L as 2.4% and 5.9%; from Table 3 sensitivity 0.941 and specificity 0.976 they should be 5.9% and 2.4%.
The infection definition’s “cell count >10,000/mL with symptoms” overlaps the synovial fluid white cell range of noninfectious bloody effusion (Sections 5 and 6). Whether any of the 9 culture-negative “infections” were sterile effusions is not discussed in the original; in the author’s view it cannot be excluded[33,37].
What can be taken from this study is: mean CRP on days 3 and 5 after hamstring reconstruction in China; mean pretreatment CRP and time to onset in infection cases in China; and the order in which CRP falls before ESR after treatment (Section 5).
Ruiz-Ibán 2015: The Only Cohort Reporting Upper-Tail Proportions; Day 3 Missing
At Ramón y Cajal Hospital in Madrid, 147 patients were enrolled prospectively from November 2010 to April 2012. After excluding 7 who did not undergo reconstruction, 20 with fewer than 4 blood draws, and 1 diagnosed with septic arthritis on day 10 (0.71%), 119 reconstructions in 117 patients entered analysis (93 men, 24 women; mean age 31.6 years; operative time 58.3±20.0 minutes). All received soft-tissue grafts (autograft 107, allograft 12); primary 107, revision 12; femoral side transtibial 82, anteromedial portal 22, double-bundle 15; concurrent lateral meniscal surgery 55, medial meniscal surgery 48, cartilage procedures 19 (microfracture 10). Of 6 surgeons, 1 performed 50–150 cases per year and the other 5 performed 4–20 per year. Blood was drawn preoperatively, on postoperative day 1, and on days 7, 14, 21, and 28 (the last four each allowed ±3 days); laboratory sensitivity 0.01 mg/L, normal <5 mg/L; all 6 draws were completed in 72 patients (60.5%)[34]. Proportions above 5 mg/L were 41.8% on day 1, 41.6% on day 7, 33.6% on day 14, 18.8% on day 21, and 23.5% on day 28; proportions above 25 mg/L were 8.6% on day 1, 11.5% on day 7, and 2.0% on day 21, and 0 at the remaining time points; proportions above 50 mg/L occurred only on day 1 (1.7%) and day 7 (4.4%)[34].
This is the first ACL cohort designed to the reference-interval sample-size requirement (IFCC and CLSI’s 120) and the only study to report proportions above 5, 25, and 50 mg/L at each time point. The authors themselves stated the shortcomings: day 3 was deliberately not sampled (on the grounds that infection rarely begins in the first 3–4 days), so the peak window is missing; there is no BPTB; hematoma and hemarthrosis were not assessed; and there were no infection cases, so sensitivity of any cutoff cannot be estimated. The authors also wrote that specificity cannot be estimated; that claim is too broad. The proportions above 5, 25, and 50 mg/L at each time point in this cohort are precisely the positive proportions among these noninfected patients at those time points; subtracting them from 100% gives the specificity within this sample—though it cannot be applied directly to symptomatic patients in whom infection is suspected. Five surgeons performed only 4–20 cases per year, and the series mixes 12 allografts, 12 revisions, and 19 cartilage procedures; the means represent a mixed population.
Uchida 2024: 439 Cases, Lowest Means, Zero Complications
Two Osaka hospitals prospectively enrolled 439 patients from April 2012 to May 2017 (215 men, 224 women; mean age 24 years), excluding combined ligament injury, liver disease, and missing data; primary 391, revision 48; anatomic double- or triple-bundle hamstring 209, rectangular-tunnel BPTB 230; meniscectomy 74, meniscal repair 139, microfracture 3, cartilage transplantation 2; surgeons had a mean of 20 years’ experience (range 11–38). Operative workflow, perioperative medication, and rehabilitation were the same at both hospitals: cefmetazole 2 g/day for 3 days, loxoprofen 180 mg/day for 1 week, ice during the hospital stay, brace immobilization for 2 weeks, partial weight-bearing at 3 weeks, full weight-bearing at 4 weeks. CRP was measured by latex agglutination with a lower limit of 0.05 mg/dL; values below the limit were recorded as 0. Blood was drawn preoperatively and on postoperative days 1, 7, and 14[35]. Proportions above 3 mg/L were 4.7% preoperatively, 57.6% on day 1, 69.9% on day 7, and 21.4% on day 14; numbers reaching 30 mg/L were 4 on day 1, 24 on day 7, and 1 on day 14; none reached 50 mg/L on day 14[35].
This is currently the only large-sample upper-tail description. On that basis the authors recommended early action for day-14 values above 30 mg/L; that recommendation comes from the upper tail of the empirical distribution and aligns in direction with Margheritini’s judgment that values “not approaching normal beyond 2 weeks” warrant concern. In the Abstract and Discussion the authors also stated that these data help exclude postoperative complications including septic arthritis; a zero-infection cohort cannot do that. It describes the distribution among uncomplicated patients and has no infected patients for comparison, so it cannot show below which value infection can be excluded. All patients received 1 week of NSAID (loxoprofen) postoperatively; the low means may relate to that, or to body habitus and technique—the original cannot separate these. Without day-3 sampling, the day-7 mean is only the highest among the measured time points; the true peak time is unknown, and the authors acknowledge that day-3 values may be higher. The number excluded for “missing data” is not given; enrollment is inadequately described[35].
Xiang 2024: Values from 135 Cases Not Yet Merged into Table 2
An observational study from Changsha compared all-inside (73 cases) with complete tibial tunnel (62 cases) hamstring reconstruction by a single surgeon without a tourniquet. The body reports only that at 1 month postoperatively the all-inside group had lower CRP, ESR, white blood cells, D-dimer, and neutrophils, and that at 3 months only white blood cells still differed; all numerical values are in the supplementary material[36]. Supplementary Tables 1–5 in fact give five time points. CRP (mg/L) for the complete-tunnel group at postoperative day 3, 1 week, 2 weeks, 1 month, and 3 months was 45.9±10.2, 34.7±7.2, 30.6±5.5, 14.9±4.6, and 8.1±2.4; for the all-inside group 33.4±13.7, 24.5±5.2, 21.2±4.9, 9.9±4.1, and 7.4±2.3. ESR (mm/h) for the complete-tunnel group was 53.5, 51.5, 45.7, 40.2, and 18.1; for the all-inside group 34.4, 34.6, 27.1, 21.7, and 18.2. At 1 month, white blood cells were 9.22±1.42 versus 5.82±1.19×10⁹/L. Except for CRP, ESR, D-dimer, and neutrophils at 3 months, between-group P values at every time point were less than 0.001[36].
Hamstring graft, more than 100 cases, single surgeon, covering day 3 to 3 months—by design this is precisely the reference data that have been lacking in China. This chapter does not yet merge it into Table 2 and does not use it as a reference for patients in China. Its values are clearly higher than existing cohorts: week-2 CRP means of 21–31 mg/L, against 2.3–11.2 mg/L in the four uncomplicated cohorts at the same interval; in Uchida’s 439, only 1 reached 30 mg/L on day 14, whereas here both groups’ week-2 means sit near that level. Month-1 means of 10–15 mg/L are likewise higher than Calvisi’s day-30 6.1 mg/L and Ruiz-Ibán’s day-28 3.4 mg/L. Week-2 CRP SDs were 18% and 23% of the mean, against 68%–152% in the four uncomplicated cohorts at the same interval. Almost every marker at every time point had a between-group P<0.001. The groups also differed by more than the tibial tunnel: the all-inside group harvested only semitendinosus, the complete-tunnel group semitendinosus plus gracilis; on the femoral side one used a guide-pin retrograde socket and the other drilled through an anteromedial portal; fixation also differed. At 1 month the white cell difference was 3.4×10⁹/L and the ESR difference 18 mm/h; how much came from number of tendons harvested, how much from drilling method or fixation, the original cannot separate. The same paper also has internal inconsistencies: the Abstract and opening of the Discussion state that markers differed only at 1 month and that at 3 months only white cells remained different, yet Supplementary Tables 1–5 show between-group P<0.001 for all five markers at day 3, 1 week, and 2 weeks, and the later Discussion then says all inflammatory markers were lower throughout the first postoperative month. The body states assay method and instrument (CRP by immunoturbidimetry, Roche COBAS 8000) but does not state the blood-draw arrangements for the three early time points, whether patients were still in hospital, missingness, clock time of sampling, or the detection limit. Until the authors supply this information and the raw distributions, these values cannot be checked and therefore are not merged with the other cohorts; that is not a judgment that the results are wrong—high values and tidy SDs alone are not enough for that conclusion.
ACL Studies with Blood Markers as Secondary Outcomes
Another ten ACL studies measured postoperative blood markers, mostly to compare a treatment (tourniquet, irrigation fluid, anticoagulant, dressing, physiotherapy, supplement, tunnel technique, or intra-articular injection), not to establish reference values (Table 3).
Table 3. ACL studies with blood markers as secondary outcomes
| Study | n and population | Compared treatment | Markers and time points | Principal values (CRP converted to mg/L) | Marker-related conclusion |
|---|---|---|---|---|---|
| Mendias 2013, USA[43] | 18; BPTB auto 15, allo 3; single surgeon | None (descriptive) | 18 markers, including plasma high-sensitivity CRP and IL-6; postoperative day 3 and weeks 2, 5, 12, 18, 26 | Values only on figures: CRP rose >10-fold on day 3; returned to preoperative by week 2 (Results) or week 5 (Discussion); IL-6, TNF-α, IL-1 unchanged throughout | Temporal pattern of plasma CRP consistent with serum data; first postoperative draw on day 3 missed the IL-6 peak |
| Barker 2009, USA[44] | 19 men; mostly hamstring; tourniquet; femoral nerve block; postoperative celecoxib + aspirin for 3 weeks | Vitamins E+C vs placebo | High-sensitivity CRP, IL-6, IL-10, etc.; preoperative, 90 min after tourniquet release, 72 h, 7 days | Whole-group median (IQR): CRP preoperative 2.5 (1.6–3.5), 72 h 36.7 (27.3–58.7), 7 days 15.7 (12.0–28.4) mg/L; IL-6 at 72 h 30.0 (25.8–48.6), at 7 days 27.4 (16.7–43.6) pg/mL | No detected difference in CRP or IL-6 between supplement and placebo; upper quartile at 72 h of 58.7 mg/L is P75, not P95; antioxidant and placebo donated by Carlson Laboratories; USANA partly funded |
| Zheng 2026, Chengdu[45] | 72 high VTE-risk patients; hamstring; tourniquet; postoperative diclofenac + parecoxib | Fondaparinux vs enoxaparin | CRP, ESR, IL-6, D-dimer; preoperative, days 1, 3, 7, 13 | Preoperative CRP about 11 mg/L; postoperative values only on figures; fondaparinux group lower CRP on days 3 and 13, lower IL-6 on days 1 and 3, lower ESR on day 3 (P<0.05) | Choice of anticoagulant may alter early inflammatory markers; preoperative CRP 11 mg/L indicates a population with already elevated baseline |
| Nagashima 2023, Japan[46] | 132; single-bundle hamstring; single surgeon | No tourniquet (77) vs tourniquet (55, historical control) | White cells, creatine kinase, CRP; preoperative, days 1, 7 | Day-1 CRP no tourniquet 14.0±11.2 mg/L, tourniquet 9.1±7.6 mg/L (P=0.016); no difference on day 7, values not reported | Slightly higher day-1 CRP without tourniquet; Discussion attributes this to synovitis from hemostatic trauma to synovium, Abstract to blood loss and soft-tissue injury; historical control; meniscal repair stitch counts differed (3.2 vs 2.3), unadjusted |
| Xiang 2024, Changsha[36] | 135; hamstring; no tourniquet; single surgeon | All-inside (73) vs complete tibial tunnel (62) | CRP, ESR, white cells, D-dimer, neutrophils; supplementary tables: day 3, 1 week, 2 weeks, 1 month, 3 months; body reports only 1- and 3-month between-group comparisons | Supplementary CRP: complete tunnel 45.9, 34.7, 30.6, 14.9, 8.1 mg/L; all-inside 33.4, 24.5, 21.2, 9.9, 7.4 mg/L; ESR at 1 month 40.2 vs 21.7 mm/h | Values clearly higher than other cohorts; blood-draw arrangements and raw distributions not stated and cannot be checked; not adopted in this chapter (see above) |
| Wang 2021, Qingdao[47] | 716; single-bundle; nonacute injury | Irrigation volume (0, 1, 3 L) and temperature (room, 36°C, 4°C) | CRP days 1, 3; IL-1, IL-6, IL-10 at 6, 12, 24 h postoperatively | Group CRP day 1 1.18–1.23, day 3 1.20–1.22 (original labeled mg/L); IL-6 about 110 at 6 h, about 79 at 24 h (units not labeled); no between-group differences | Irrigation volume and temperature do not affect inflammatory markers; CRP magnitude and “no rise from day 1 to day 3” contradict every other cohort; CRP absolute values not adopted in this chapter |
| Oestervemb 2025, Poland[48] | 60; semitendinosus + gracilis double-bundle | Deep oscillation therapy vs conventional physiotherapy | Capillary fingertip CRP, D-dimer (point-of-care device); postoperative day 3 (baseline) and 2 and 4 weeks later (about postoperative days 17 and 31) | Day-3 CRP 34.4±51.0 and 26.8±15.8 mg/L; control at 2 weeks 15.4±9.3, at 4 weeks 7.3±5.1 mg/L; treatment at 2 weeks 6.3±10.9, at 4 weeks 1.8±1.1 mg/L; also reports quartiles and ranges | Day-3 means consistent with other cohorts; accuracy of point-of-care fingertip assay not validated; values after 2 weeks not merged with venous-blood data |
| Yonetani 2022, Japan[49] | 60; BPTB; 3 surgeons; drain 2 days | Film dressing vs gauze dressing (historical control) | Hemoglobin, CRP; days 1, 7 | Day-1 CRP gauze 7.7±6.5, film 3.9±4.2 mg/L (P=0.009); day 7 20.4±22.4 and 11.7±14.3 mg/L (P=0.077) | BPTB day-7 means 12–20 mg/L; dressing difference on day 1 is small |
| Orrego 2005, Chile[50] | 25 mixed arthroscopic ACL reconstruction or osteotomy, all with tourniquet | None (listed alongside meniscectomy, arthroplasty, spine surgery) | CRP; preoperative, days 1, 2, 3, 7, 15 | Means (SD not reported): preoperative 3.0, day 1 18.6, day 2 45.7, day 3 42.6, day 7 35.9, day 15 10.5 mg/L; 2 complications (cellulitis, venous thrombosis) day 7 109.5±71.4 vs uncomplicated 29.5±29.5 mg/L | Peak on days 2–3; signal in complicated patients is failure to fall—or a rise—on day 7; mixes osteotomy and cannot alone represent ACL reconstruction |
| Balagod 2025, India[38] | 90; hamstring; single surgeon; randomized double-blind; 87 completed | Early or delayed intra-articular hyaluronic acid vs placebo | ESR, CRP; preoperative, postoperative weeks 4, 8, 12, 52 (Abstract says monthly to 6 months) | Median (IQR): CRP at 4 weeks in three groups 3.85 (2.5–5.2), 4.7 (3.1–6.6), 3.8 (2.1–5.6) mg/L; at 12 weeks 2.6–3.2; at 52 weeks 1.8–2.0; ESR at 4 weeks 14.5–22, at 12 weeks 7–10, at 52 weeks 6–8 mm/h | Injection does not change ESR or CRP; provides medians from 1–12 months postoperatively—rare ACL data covering beyond 1 month |
Notes to Table 3: Nagashima and Yonetani originally reported mg/dL; values here are multiplied by 10. Wang 2021 originally reported mg/L and is transcribed as such, with concerns noted in the last column. Orrego originally reported mg/dL; values here are multiplied by 10. Xiang values are taken from Supplementary Tables 1 and 2.
Barker’s 72-hour median is compatible with Calvisi’s hamstring day-3 mean; the upper quartile of 58.7 mg/L is P75, not the P95 needed to judge abnormality[32,44]. Oestervemb’s point-of-care fingertip assay and Balagod’s data beyond 1 month also report quartiles[38,48]. Day-1 CRP lies between 3.9 and 18.6 mg/L; between-group differences in the two historical-control studies of tourniquet and dressing are only a few milligrams per liter, and neither adjusted for confounding[46,49]. Perioperative medication alters markers: choice of anticoagulant changes day-3 CRP and IL-6[45]; Barker’s and Zheng’s patients were both on NSAIDs—Barker’s authors acknowledge that cytokine data were affected, Zheng’s original does not discuss this[44,45]. In the Indian trial the three groups did not differ at any time point; by 1 month most patients’ CRP was already below 5 mg/L, and ESR medians at 3 months were 7–10 mm/h[38]. Wang 2021’s absolute CRP values are far below those of every other cohort at the same intervals and show no rise from day 1 to day 3; this chapter does not adopt them (Table 3)[47].
In Barker’s trial the antioxidant and placebo were donated by Carlson Laboratories, and study funding came partly from the supplement company USANA; Orrego’s 600 CRP assay reagents were donated by Roche[44,50]. These relationships do not by themselves prove bias, nor do they guarantee no influence; this chapter uses the values from these two studies only descriptively.
Comparison with Fracture and Arthroplasty
The CRP response to surgical trauma relates to the amount of tissue injury; this has been studied most clearly in fracture and arthroplasty data. Neumaier et al. analyzed 349 uncomplicated proximal femoral fracture operations: every technique peaked on postoperative day 2—percutaneous screws mean 87 mg/L, dynamic hip screw 120 mg/L, proximal femoral nail 140 mg/L, hemiarthroplasty and total hip arthroplasty both 160 mg/L; opening the medullary canal, extruding marrow, and inserting a stem or nail all raised CRP substantially[24]. The same group’s 787 uncomplicated fracture operations showed day-2 medians by site from ankle 39 mg/L to femur 136 mg/L (90th percentiles 90 and 221 mg/L), returning to normal around postoperative day 12; they plotted 10th, 50th, and 90th percentile curves. The curves of 17 deep infections separated from the uncomplicated curves only after day 4; after day 4, values above 96 mg/L had sensitivity 92% and specificity 93%[25]. Among Larsson et al.’s 193 uncomplicated elective operations in 1992, primary total hip arthroplasty (109) peaked on day 3 at 116±43 mg/L, revision (9) at 136±58 mg/L, unicompartmental knee arthroplasty (39) on day 2 at 140±46 mg/L, and minimally invasive lumbar disc excision (36) on day 2 at 48±27 mg/L. Decline came in two phases—fast on days 3–5, then slower—returning to normal or near normal by days 14–21; by day 21, 95% of primary total hips were already normal, the remaining 6 at 10–15 mg/L, and all were normal by day 42. Anesthetic technique, blood loss, transfusion, operative time, antibiotics and anti-inflammatory drugs, age, and sex were unrelated to the CRP response; the authors considered CRP still high beyond 3 weeks after hip or knee arthroplasty abnormal[26].
Arthroplasty values are higher still. In Park et al.’s 320 TKA knees, CRP after unilateral replacement (108 knees) had means of 151 mg/L on day 2, 72 on day 5, 44 on day 7, 13 on day 14, and 3.7 on day 42; Windisch et al.’s 1,034 TKAs were 170±68 mg/L on day 2 and 168±81 on day 3, still 55±34 on day 10; Bilgen et al.’s THA and TKA day-2 values were 232 and 260 mg/L[27-29]. Peak CRP after ACL reconstruction is clearly lower than these figures; the hamstring day-3 mean is close to the day-2 median after ankle fracture fixation (39 mg/L), with BPTB in between. This is only an order-of-magnitude analogy and is not evidence that the two classes of surgery involve comparable injury. Cutoffs derived from arthroplasty and fracture data have not been validated in the ACL reconstruction population, and the normal peaks of the two populations differ by several fold; they cannot be transferred directly.
Windisch et al. retrospectively reviewed 1,068 primary unilateral TKAs at a single center; CRP values and trajectories in the first 5 postoperative days did not distinguish in advance the patients who later developed infection. Thirty-four (3.1%) underwent revision for surgical-site or periprosthetic joint infection; CRP measured daily in the first 10 postoperative days did not differ on any day between infected and uncomplicated patients. None of the 34 infected patients had a day-5 value higher than day 2, whereas 24 of 1,034 uncomplicated patients (2.3%) had day 5 higher than day 2 without infection; preoperative CRP above 5 mg/L was instead a predictor of infection[30]. The original does not report time to onset of these infections; the author surmises that most infections may have begun after day 5, so that blood drawn in the first 5 days did not yet carry an infection signal. Infection after ACL reconstruction most often begins in weeks 2–3 (Section 5); a single CRP in the first 5 days is likewise unlikely to identify it. Windisch 2016 and 2017 report the same cohort; this chapter does not treat them as two independent observations[29,30].
3. Factors Affecting Postoperative CRP Levels
Among direct ACL reconstruction data, Ruiz-Ibán, Uchida, and Calvisi have analyzed influencing factors. Sex was consistent in direction across two cohorts; concurrent microfracture, surgeon experience, and preoperative CRP were associated with higher CRP only in the Ruiz-Ibán cohort, and BMI only in the Uchida cohort; graft-type data are inconsistent; tourniquet has only one ACL historical-control comparison (Nagashima, Table 3), and NSAIDs, induction dexamethasone, tranexamic acid, drains, and vancomycin soaking have no dedicated ACL reconstruction analyses. Ruiz-Ibán’s Methods call the analysis multivariable logistic regression, the statistical note says linear regression after log transformation, and Table 3 calls the exponentiated coefficients odds ratios; the dependent variables are log-transformed CRP peak and area under the curve, so the fold-changes cited below should be read as ratios of geometric means and used only for order of magnitude[34].
Sex. In the Ruiz-Ibán cohort, men’s CRP peak was 1.7 times women’s (P<0.001) and area under the curve 1.9 times (P=0.014); Table 3 writes the peak-effect 95% CI as 1.8–2.5, with the point estimate outside the interval[34]. In the Uchida cohort men were higher on days 1, 7, and 14, and on day 14 sex was the only factor that remained significant (β 0.304, P<0.001)[35]. The two studies differ in population, technique, and assay, but agree in direction and support building reference distributions stratified by sex. The authors explain this by larger male knees and greater surgical trauma; neither was measured in the originals. In healthy populations women’s baseline CRP is instead higher than men’s: in the Dallas Heart Study of 2,749 persons, women’s median was 3.3 mg/L and men’s 1.8 mg/L[58]. Higher postoperative values in men suggest that the difference comes from the surgical response rather than continuation of baseline; no ACL study in the same cohort has simultaneously adjusted for preoperative CRP, knee size, and harvest trauma. This explanation is the most reasonable; the mechanism is not yet confirmed. CRP measurement in the Dallas study was supported by Roche Diagnostics, and the last author, de Lemos, received research funding from that company[58].
BMI. In the Uchida cohort, higher BMI meant higher CRP on day 1 (β 0.320, P<0.001) and day 7 (β 0.113, P=0.017), not significant on day 14[35]; Ruiz-Ibán did not put BMI in the model. The positive association of BMI with baseline CRP in healthy populations is established. In NHANES III of 16,616 US adults, taking CRP ≥2.2 mg/L as elevated, adjusted odds ratios for overweight (BMI 25.0–29.9) and obesity (≥30) were 1.41 and 2.13 in men and 2.23 and 6.21 in women; restricted to nonsmoking young adults aged 17–39 without inflammatory disease, obese men were still 2.85 and obese women 12.9[23]. In the Dallas study BMI was the continuous variable most strongly related to CRP, with correlation coefficients higher in women (0.51–0.58) than in men (0.29–0.32)[58]. In the early postoperative period among ACL patients, only Uchida provides data for this association.
Age. In the Uchida cohort younger patients were slightly higher on day 7 (β −0.106, P=0.027), nonsignificant at other time points; mean age in that cohort was 24 years with a limited age span[35]. Age was nonsignificant in the Ruiz-Ibán model[34].
Surgeon. In the Ruiz-Ibán cohort, patients of surgeons who performed only 4–20 cases per year had CRP peak and area under the curve about 2.5 times those of senior surgeons’ patients (peak P=0.007, AUC P<0.001); operative time was nonsignificant in the model. The authors conjectured that the difference came from the trauma of notchplasty, fat-pad resection, and similar maneuvers[34]. Uchida’s surgeons all had more than 11 years’ experience; surgeon experience was nonsignificant. Between the two hospitals there was a difference only on day 1 (β −0.257, P<0.001), not on days 7 or 14[35]. This association comes from one cohort of 6 surgeons. The model already included meniscal and cartilage procedures but had no data on tourniquet, anesthesia and analgesia, or operative trauma volume; the 2.5-fold figure may mix these factors and cannot be taken as an independent trauma effect. That experienced surgeons do not differ from one another rests on Uchida alone.
Concurrent procedures. In the Ruiz-Ibán cohort, 10 microfracture cases were associated with about a 2-fold higher CRP peak (P=0.021; Abstract 1.9-fold, Table 3 2.0-fold); primary versus revision, medial and lateral meniscal surgery, and chondral debridement entered in the same model showed no significant effect[34]. In the Uchida cohort there were only 3 microfracture cases; meniscectomy (74), meniscal repair (139), and revision (48) were nonsignificant at all three time points[35]. The microfracture effect comes from 10 cases; the direction is reasonable, the fold-change unstable.
Graft. Calvisi’s 13 BPTB cases had a day-3 mean more than twice that of hamstring, without a between-group test[32]. Uchida’s 230 BPTB and 209 hamstring cases did not differ on days 1, 7, or 14; the authors thought this might relate to differences in technique or bone-block size[35]. Calvisi offers only an untested descriptive difference; Uchida tested and found none—the two do not constitute opposite directions. Uchida had no day-3 sampling; whether the BPTB peak is higher cannot be settled from available data. Bone-block size and technique are explanations the authors proposed and did not test.
Tunnel technique. The Changsha cohort reported lower markers at 1 month with all-inside than with complete tibial tunnel, but its values are clearly higher than other cohorts and cannot be checked, and the groups also differed in number of tendons harvested, femoral drilling method, and fixation; the between-group difference cannot be attributed to tunnel technique alone. This chapter does not treat it as evidence that tunnel technique affects CRP (Section 2)[36]. Variables such as tunnel length and whether the far cortex is breached can only be recorded items.
Preoperative CRP. In the Ruiz-Ibán cohort, patients with preoperative CRP above 5 mg/L accounted for 8.4% and also had higher postoperative peaks (1.5-fold, P<0.001)[34]. Margheritini, Calvisi, and Wang all directly excluded those with abnormal preoperative values; their figures do not apply to patients with mild preoperative elevation[31-33].
Factors Supplemented from Adjacent Populations
Operative time. In Windisch et al.’s factor analysis of 1,034 TKAs, ASA class, operative time, tourniquet ischemia time, age, weight, height, and BMI showed no contribution to CRP variation; only sex did—men higher on days 2–5 and 7–8, with the largest difference on day 4 (170 vs 125 mg/L, P=0.019)[29]. Chung et al.’s spine surgery cohort found that age, male sex, diabetes, hypertension, operative time, surgical site, and instrumentation all affected CRP change (repeated-measures ANOVA interaction P from <0.001 to 0.047); BMI had no effect[51]. Conclusions on operative time disagree between the two data sets; neither found a BMI effect. In ACL reconstruction, Uchida found a BMI effect and Ruiz-Ibán found no operative-time effect[34,35]. These two factors behave differently across operations and cohorts; each center needs its own data.
Induction dexamethasone. In Xu et al.’s randomized double-blind trial of 108 TKAs, the two-dose low-dose dexamethasone group had lower CRP and IL-6 than placebo at 24, 48, and 72 hours postoperatively (P<0.001 at all three time points). Both groups’ CRP peaked at 48 hours; the IL-6 peak was at 24 hours in the placebo group and delayed to 48 hours with dexamethasone. Values at each time point appear only on Figure 2 (y-axis units mg/L and pg/mL; placebo CRP at 48 hours reads about 60 mg/L, dexamethasone about 40 mg/L); the body gives none[59]. Backes et al.’s trial of 120 arthroplasties is often cited as evidence that “dexamethasone lowers CRP and IL-6 after arthroplasty,” but that trial measured only white cells and glucose: the three groups did not differ in the rise of postoperative white cells, and diabetic patients had higher day-2 glucose after dexamethasone (152.9 vs 137.3 mg/dL, P=0.012); the Discussion’s “50% reduction in CRP” cites Hall, Smith, and others[60]. Induction dexamethasone can depress CRP in the first 3 days after arthroplasty; ACL reconstruction commonly uses a single small dose and involves far less trauma—how much it depresses CRP has not been measured in ACL, and dose and timing should be recorded case by case. Reference distributions should be built among patients on a consistent medication regimen.
NSAIDs. The entire Uchida cohort received 1 week of loxoprofen postoperatively and has the lowest means of the five cohorts[35]; Margheritini withheld NSAIDs and has the highest day-3 peak[31]. Between these two also stand differences in graft, technique, ethnicity, assay method, and rehabilitation immobilization; the relative height of the two means cannot be read as evidence that NSAIDs depress CRP. Direct data on the effect of NSAIDs on CRP after ACL reconstruction do not exist; they can serve only as a recorded variable.
Anticoagulants and VTE prophylaxis. In the Chengdu randomized trial, the fondaparinux group had lower CRP on days 3 and 13 and lower IL-6 on days 1 and 3 than the enoxaparin group (Table 3)[45]. In the Osaka cohort that used only mechanical prophylaxis, the day-7 thromboembolism rate was 11%[42]. Whether prophylaxis changes the marker or the outcome cannot be separated from available data; both should be recorded case by case.
Graft vancomycin soaking. Many centers have already adopted this as routine (Section 5). Whether it changes postoperative blood or synovial fluid markers has no human data. When building reference distributions, whether soaking was used should be recorded, and comparisons with historical unsoaked cohorts should note it[3].
4. ESR, White Blood Cell Count, IL-6, D-Dimer, Procalcitonin, and Synovial Fluid Cytokines
ESR
ESR changes more slowly than CRP. In the Margheritini cohort it peaked on day 7 (68.3±21.0 mm/h) and on day 30 was 17.4±9.8 mm/h, still above preoperative[31]. In the Wang cohort, days 3 and 5 were 22.7 and 22.3 mm/h, with no difference between the two days (P=0.838)[33]. In Paci’s 31 noninfected patients aspirated for painful effusion, ESR at a mean of 5.5 days postoperatively was 58.8±23.9 mm/h, falling in the same range as the weighted mean of 66.4 mm/h among published infection cases[37]. Arthroplasty data show the full course of ESR: in Park et al.’s TKA, ESR peaked on day 5 (69 mm/h), remained above 20 mm/h on day 42, and approached preoperative only by day 90; among 214 knees in unilateral and staged bilateral groups, 93 (43%) had an atypical trajectory of CRP or ESR in at least one, atypical ESR accounting for 32.2% and atypical CRP only 11.7%[27]. In Bilgen et al.’s THA and TKA, ESR was about 100 mm/h on day 5 and still 21–22 mm/h at 12 months postoperatively, above the preoperative 14–16 mm/h[28]. In Larsson et al.’s arthroplasty patients ESR peaked around day 5, with timing and amplitude less stable than CRP, and on day 42 35% were still more than 10 mm/h above preoperative[26]. In Honsawek et al.’s TKA, ESR peaked only in week 2[39]. ACL reconstruction and arthroplasty data agree on two points: a single early postoperative ESR overlaps values in infected patients; and weeks to months after surgery it has not yet returned to preoperative (medians beyond 1 month after ACL reconstruction are in Section 2)[38]. Using ESR to judge infection within the first postoperative month has no validated basis. Wang’s proposed ESR cutoff of 32 mm/h has the same time-point mismatch as the CRP cutoff.
Peripheral White Blood Cell Count
Post-ACLR peripheral white cell values come mainly from Paci’s 31 cases (single time point). These patients were aspirated for painful effusion and ultimately judged noninfected—a symptom-selected population that cannot represent uncomplicated patients in routine recovery. At a mean of 5.5 days postoperatively, white cells were 8.4±2.0×10⁹/L; the weighted mean among published infection cases was 9.3×10⁹/L (range 4.2–16.5). The two overlap; the authors considered that white cells cannot distinguish infection[37]. In Table 3, Nagashima reported day-1 values of 9.7 versus 9.1×10⁹/L[46]; Zheng 2026’s values appear only on figures[45]; Xiang’s values are not adopted[36]. In early case series recounted by Margheritini, only 1 of 4 infections had an elevated white cell count[31]. White cells and the neutrophil fraction may rise on postoperative day 1 because of surgical stress and dexamethasone; this has not been measured directly after ACL reconstruction.
IL-6
IL-6 rises within hours after surgery, earlier than CRP. In Honsawek et al.’s 49 TKAs, IL-6 was 57.0±21.2 pg/mL preoperatively, peaked at 184.5±53.5 pg/mL at 24 hours, and returned to preoperative by 2 weeks. The same patients’ CRP was 85.0±21.7 mg/L at 24 hours, 6.9±4.8 mg/L at 2 weeks, and 4.4±2.5 mg/L at 6 weeks, returning to the normal range. The body and Table 1 of that paper labeled CRP units as mg/dL; the authors later published a formal erratum stating that CRP values had been labeled 10-fold too high and that units throughout should be mg/L (Int Orthop. 2011;35(11):1749). Values read with the erratum match the known magnitude after TKA[39]. Circulating IL-6 after ACL reconstruction comes only from two studies that used it as a secondary outcome, and their peak times do not agree: Wang 2021 first drew blood at 6 hours postoperatively, when levels were highest, then fell successively at 12 and 24 hours[47]; Barker found levels only slightly above preoperative at 90 minutes after tourniquet release, and 2–3 times preoperative at 72 hours and 7 days[44]. The two differ in assay method, analgesia, and sampling schedule; what can be established is only that IL-6 rises earlier than CRP and remains above preoperative within the first week. Mendias’s first postoperative sample was on day 3; plasma IL-6 was unchanged throughout[43]. A 2024 systematic review pooled 16 studies and 492 patients and found associations of serum IL-6, serum and synovial MMP-3, and urinary and synovial CTX-II with postoperative pain, laxity, and function scores; the authors considered that none of these markers is yet ready for clinical use. The review did not address infection[40]. Whether IL-6 can identify infection after ACL reconstruction has no data.
D-Dimer
D-dimer rises sharply on postoperative day 1. In Azboy et al.’s 80 primary THAs or TKAs, preoperative values were 412±260 ng/mL, day 1 4,575±1,844 ng/mL (the original says about 9-fold; by Table 1 means about 11-fold), day 3 1,706±1,066 ng/mL, and days 5, 15, and 45 means 1,523, 2,026, and 1,662 ng/mL. Against the international consensus chronic periprosthetic joint infection threshold of 860 ng/mL, “above threshold” rates among uncomplicated patients were 100% on day 1, 87.5% on day 3, 81.2% on day 5, 88.7% on day 15, and 77% on day 45[41]. In these uncomplicated primary arthroplasty patients, 860 ng/mL produced large numbers of positives at all 5 measured time points and cannot alone judge infection; the study had no infection group and cannot answer whether another early postoperative cutoff is usable.
The clear postoperative use of D-dimer is to suggest venous thrombosis. Direct ACL reconstruction data come from 136 Osaka patients managed with mechanical prophylaxis only: contrast-enhanced CT on day 7 found venous thromboembolism in 15 (11.0%), 1 of them symptomatic. Day-7 D-dimer was 3.41±1.0 µg/mL with thrombosis and 2.01±1.2 µg/mL without; a cutoff of 2.8 µg/mL had sensitivity 80.0% and specificity 83.5% (area under the curve 0.875); the two groups did not differ preoperatively[42]. The day-7 mean without thrombosis of about 2 µg/mL is a mean, not a defined normal range. The 2.8 µg/mL cutoff was derived internally among the 15 thromboses in that cohort, with no validation set. The original writes the cutoff unit as both µg/mL and mg/mL; judging from the preoperative mean of 0.59 µg/mL it should be µg/mL. Sensitivity and specificity for the same cutoff are written later in the Discussion as 92.3%/82.2%[42]. This cutoff is for judging thrombosis and is unrelated to recognition of infection.
Procalcitonin
Procalcitonin (PCT) rises only mildly after elective orthopedic surgery. In Chung et al.’s 103 elective spine operations, preoperative PCT was below 0.05 ng/mL in all; on postoperative day 1, 27 rose to a quantifiable level (mean 0.17±0.17 ng/mL) and the rest remained below 0.05 ng/mL; on day 3, 20 were quantifiable (mean 0.16±0.22 ng/mL); on day 5, 6 (mean 0.15±0.15 ng/mL). All but 2 were below 0.25 ng/mL (Abstract) or 0.5 ng/mL (Discussion). In the same patients, day-3 CRP mean was 26.7 mg/L and varied with age, sex, diabetes, operative time, site, and instrumentation; PCT was unrelated to these factors[51]. Hunziker et al. prospectively observed 103 patients with new fever within 10 days after orthopedic surgery; 45 (44%) had confirmed infection, mainly respiratory and urinary, with 4 joint infections and 2 surgical-site infections. On the day of fever, median CRP was 129 mg/L in noninfected and 179 mg/L in infected patients and did not separate the groups (area under the curve 0.56); PCT was 0.27 versus 0.34 ng/mL on the day of fever and 0.14 versus 0.30 ng/mL on day 3, area under the curve 0.62–0.71, and was an independent predictor in multivariable analysis (odds ratio 2.3–3.3). The authors stated clearly that “normal” postoperative PCT does not lie within the healthy-person normal range and that cutoffs must be set separately[52].
Three studies of instrumented spine surgery disagree. In Inose et al.’s 242 cases with 10 surgical-site infections, PCT on postoperative days 6–7 did not differ between infected and noninfected (P=0.223); neutrophil fraction and neutrophil-to-lymphocyte ratio had only moderate discriminatory ability[53]. In Chen et al.’s 106 cases with 8 early infections, day-7 PCT was 0.42±0.22 versus 0.10±0.05 ng/mL, and a cutoff of 0.2 ng/mL had sensitivity 87.5% and specificity 94.1%; day-7 CRP was 108.8±37.1 versus 43.2±40.3 mg/L, and a cutoff of 64.1 mg/L had sensitivity 100% and specificity 76.5%. The authors themselves noted that CRP rises in the first 3 postoperative days with or without infection, that the noninfection group included only patients with complete data, and that the infection rate does not represent true incidence[54]. In Zheng et al.’s 90 cases with 7 infections, day-3 PCT was 0.19±0.13 versus 0.12±0.96 ng/mL and CRP 94.5±45.2 versus 54.3±42.6 mg/L. The six sets of cutoff sensitivity and specificity given in the original (CRP day 3 87.5%/62.5%, day 6 94.1%/52.3%; PCT day 3 78.6%/61.1%; ESR day 6 80.0%/68.5%; temperature day 3 58.8%/76.7%, day 6 52.9%/71.2%) cannot be calculated from these case counts: with 7 positives, sensitivity can only be multiples of 1/7 (14.3%, 28.6%, 42.9%, 57.1%, 71.4%, 85.7%, 100%), yet 87.5% implies a denominator of 8, 94.1%, 52.9%, and 58.8% imply 17, and 78.6% implies 14; specificities likewise fail to match 83 cases. The original does not explain the algorithm; diagnostic performance cannot be verified, and this chapter does not adopt it[55]. All of these cutoffs were derived internally in small samples; the populations are instrumented spine surgery; time to diagnosis of infection was days 9–27 in Chen, days 7–24 in Zheng, and days 7–119 in Inose. What can be borrowed for ACL reconstruction is only the direction: elective orthopedic surgery itself raises PCT only mildly (among the 27 who rose on day 1 after spine surgery, mean about 0.17 ng/mL); clear elevation suggests a stimulus outside the surgery; a low value does not exclude local infection. Case-by-case PCT values in infection after ACL reconstruction appear only in the Saudi series of 12 (Section 5); no cutoff has been validated in the ACL population. The 0.25–0.5 ng/mL figures mentioned by Chung and Hunziker are only observations within their own cohorts.
Neutrophil-to-Lymphocyte Ratio
Among 245 patients readmitted within 90 days after arthroplasty, Yu et al. excluded inflammatory arthropathy, superficial infection (24), fever of other cause (12), malignancy (4), and missing data, and analyzed 121: IL-6 (area under the curve 0.814), neutrophil-to-lymphocyte ratio (0.802), and CRP (0.793) had similar ability to distinguish early periprosthetic joint infection; ESR (0.744) and white cells (0.632) were poorer. Median CRP was 66.6 mg/L in the infection group and 8.6 mg/L in the noninfection group[56]. This is a single comparison on the morning after readmission, not sequential postoperative sampling; it reflects which marker discriminates better once symptoms have appeared and cannot show which marker can warn in advance.
Synovial Fluid Cytokines
Cytokines in synovial fluid are already extremely high after surgery. A 2026 systematic review pooled 44 studies, 2,428 patients, and 4,721 synovial fluid samples, taking the healthy-knee range of 0–10 pg/mL as baseline. After ACL reconstruction, weighted mean synovial IL-6 was 3,519.6 pg/mL within 1 week, 612.7 pg/mL from 1 week to 1 month, 255.6 pg/mL at 3–6 months, 107.5 pg/mL at 6–12 months, and 12.5 pg/mL beyond 1 year. IL-1β was 1,881.5 pg/mL within 1 week and rose further to 4,045.7 pg/mL from 1 week to 1 month. IL-8 was 1,239.0 pg/mL within 1 week and still 19.2 pg/mL at 1 year[57]. These are all values from noninfected patients. Studies report dispersion differently; the authors performed no formal meta-analysis, and weighted means can be read only for order of magnitude. The review does not discuss infection. In the first postoperative month, synovial IL-6 two to three orders of magnitude above baseline is the norm; when synovial cytokines are used to distinguish infection, the comparator should be the level produced by surgery itself in the same postoperative interval. Two authors of that review disclosed research support from Elite Orthopedics and consulting fees and fellowship funding from Arthrex and other companies[57].
5. Infection Incidence, Onset Timing, and Markers at Presentation
Incidence and Risk Factors
A systematic review and meta-analysis published in 2023 included 23 observational studies and 469,441 ACL reconstructions: pooled surgical-site infection incidence about 1% (95% CI 0.7%–1.2%, I²=98%), of which deep infection about 0.6% (0.4%–0.8%); six studies reported time from surgery to symptom onset, with pooled mean 17.1 days (95% CI 13.2–21.0 days)[1]. Another systematic review the same year included 17 studies and 141,991 cases[2]. Risk factors from the two analyses are in Table 4.
Table 4. Infection risk factors in two meta-analyses
| Factor | Zhao 2023[1] (23 studies, 469,441 cases) | Zhang 2023[2] (17 studies, 141,991 cases) |
|---|---|---|
| Male sex | OR 1.78 (1.43–2.21) | OR 1.90 (1.33–2.73) |
| Diabetes | OR 3.40 (1.56–7.41), high heterogeneity | OR 2.69 (1.66–4.35) |
| Glucocorticoid use | OR 4.80 (2.61–8.84) | OR 7.83 (3.68–16.63) |
| Revision | OR 2.05 (1.03–4.06) | OR 2.31 (1.22–4.37) |
| Hamstring autograft | vs BPTB, OR 2.83 (2.22–3.60) | OR 2.51 (2.03–3.10) |
| Concurrent lateral extra-articular tenodesis | OR 3.92 (1.96–7.84) | OR 3.45 (1.63–7.28) |
| BMI | BMI >30: OR 1.82 (1.30–2.55) | Comparison of mean BMI in infected vs noninfected (3 studies, 104 infections): nonsignificant |
| Operative time | Infected mean 8 minutes longer | — |
| Professional athlete | — | Significant |
| Age, smoking, meniscal repair, outpatient surgery | — | All nonsignificant |
Direction in the two analyses is largely consistent. Conclusions on BMI differ (one significant, one not—not opposite directions) because the objects of analysis differ: Zhao compared infection proportions among those with BMI above 30; Zhang compared mean BMI between infected and noninfected. The two algorithms cannot be compared directly; BMI is not yet an established risk factor[1,2]. These are associations at the incidence level. A high OR does not mean high infection probability for an individual patient; risk factors can only adjust the pretest judgment and cannot replace symptoms, local findings, and synovial fluid examination at the time. Whether sex and BMI change postoperative CRP in uncomplicated patients is a separate question (Section 3).
Most of the incidence figures above come from an era before routine graft vancomycin soaking. At one Croatian center, after 1,329 soakings (5 mg/mL) from 2018–2023 there was 1 infection (0.07%), against 1.4% in the preceding 1,891 cases (OR 0.05, 95% CI 0.0073–0.40); this is a historical control, and the authors themselves note temporal bias[3]. A 2023 meta-analysis pooled 13 studies and 31,150 cases (soaked 11,437, control 19,713, mostly retrospective): infection rate after soaking 0.09%, without soaking 0.74% (OR 0.17, 95% CI 0.10–0.30). Definitions of infection and septic arthritis were inconsistent across studies; heterogeneity appears only in the forest plot (I²=0%) and is not stated in the body[4]. Centers that soak routinely should use the lower incidence when estimating pretest probability and planning infection registries (Section 7); previously registered historical cases are unaffected.
Onset Timing and Clinical Presentation
Single-center series give onset times consistent with the pooled value, with means or medians between 9.5 and 23.5 days after surgery (Table 5). Two further figures record something other than onset: among 17 infected knees that underwent MRI in the New York HSS registry, median time from surgery to diagnosis was 35 days (9–411)[13]; among 17 cases in Caen from 2010–2022, mean time from reconstruction to arthroscopic irrigation was 24.2±18.5 days (5–63)[14]. A 2014 systematic review of 17 papers and 196 infections found mean time from reconstruction to presentation 23.5±46.8 days (2–455); among 123 with classification data, 92.6% were acute or subacute, with mean onset 16.8±10.5 days (2–56) after surgery[15].
Putting these series together yields a rough time distribution: onset in the first 3–4 postoperative days is only individual cases (Beijing series earliest day 3, Swedish series day 4); cases appear successively on days 4–7; weeks 2–3 are the peak; and scattered subacute cases still appear at 1–2 months[5,10,15]. This is the author’s synthesis of onset times across series; no study gives a day-by-day distribution. The relationship between pathogen and timing disagrees between the systematic review and the German series: in the systematic review, time from reconstruction to presentation was later for Staphylococcus aureus than for coagulase-negative staphylococci (29.6±31.1 vs 13.5±7.4 days, P=0.02)[15]; in the German series the two onset times were similar (14.9±10.3 vs 17.3±8.8 days, P=0.552)[6]. Subacute infection caused by Cutibacterium acnes and low-virulence staphylococci may not be diagnosed until 1–2 months after surgery[6,15]. The Beijing 2009 and 2014 infection groups come from the same institute, with observation windows 1997–2007 and 1997–2010; the former falls entirely within the latter, and the latter also cites the former. The two cannot be treated as independent series; how many cases overlap cannot be determined from the originals[5,33].
Fever is not required. Temperature above 38.0°C among infected patients accounted for 38.9% in the German series[6], 61.5% (≥38.0°C) in the Croatian series[7], and 85% in the Naples series[16]; all 21 in the Beijing 2009 series were febrile at presentation (38.0–39.7°C)[5]. Measurement conditions for these proportions differ (timing of presentation, antipyretics, site of measurement) and cannot be converted into a negative likelihood ratio for “absence of fever.” Worsening pain, effusion, and regression of range of motion are present in almost all infected patients[6]; patients with postoperative hematoma and those who overload also show the same findings.
Table 5. Onset timing and markers at presentation in series of infection after ACL reconstruction
| Series | Reconstructions / infections | Symptom onset (postoperative days) | CRP (mg/L) | ESR (mm/h) | Synovial WBC (/µL) | Culture positive |
|---|---|---|---|---|---|---|
| Beijing 2009[5] | 4,068/21 | 13.3 (3–29) | Units uncertain; not listed | 59.7 | — | 16/21 |
| Beijing 2014[33] | 7,679/39 | 13.7±6.0 (3–26), excluding 4 chronic | 101.9±18.0 | 57.1±8.2 | — | 30/39 |
| Germany[6] | 7,096/36 | 17.1±9.3 (4–37) | 138±84 (27–417) | — | — | 30/36 |
| Croatia[7] | 1,891/26 | 23.5±21.0 (6–100) | 124.1±51.4 (45.7–230) | 63.7 | — | 11/26 |
| Saudi Arabia[8] | 836/12 | 20.2±14.2 (6–47) | 194.8±115.2 (100.5–445) | 64.7 | 68,609 (53,000–91,000) | 12/12 |
| US military hospital[9] | 1,615/11 (intra-articular) | 14.2 (6–45) | Units unclear; not listed | — | 52,735 (25,400–72,800) | — |
| Sweden[10] | 575/10 | 9.5 (4–20) | 91.8 (10–199) | 62.3 | — | 2 negative |
| France FAST[11] | 1,809/7 | 15.7±5.5 (7–21) | — | — | — | — |
| Boston Children’s Hospital[12] | 5,638/13 | Median 15 (6–632) | Units uncertain; not listed | Median 75 | Median 58,000 (35,000–140,000) | 13/13 |
| Naples[16] | —/39 | — | — | — | Median 80,710 | 28/39 |
| Harbin[61] | —/20 (readmissions) | — | 12.5–97.1; 75% above 50 | — | — | — |
| Canadian cluster[62] | —/9 | — | 55.0 (17.1–104) | 50.3 | 77,186 (32,230–222,000), neutrophils 94% | Coagulase-negative staphylococci 5; no growth 4 |
| Korea[63] | —/6 | — | 56.9–147 | — | 22,400–102,000, neutrophils 80%–90% | All S. epidermidis |
Notes to Table 5: Values are mean±SD (range) or median (range), transcribed as in the original. For the Beijing 2014 series, “±” is not specified as SD or SE (see Section 2). The Korean series originally reported CRP as 5.69–14.7 mg/dL; values here are multiplied by 10. CRP units for Beijing 2009, the US military hospital, and Boston Children’s Hospital are discussed below. The Beijing 2009 and 2014 observation windows overlap and are not independent series.
Blood Markers at Presentation
In series with clear units, CRP at presentation in infected patients is mostly 100–200 mg/L (Table 5). The 2014 systematic review of 196 cases found CRP above 10 mg/L in 98%, pooled mean 183 mg/L (original 18.3±4.8 mg/dL); ESR above 30 mm/h in 89%, pooled mean 62.5±6.5 mm/h; the original notes that the ± figures here are 95% confidence intervals[15]. In the German series, CRP at presentation was higher in Staphylococcus aureus than in coagulase-negative staphylococcal infection (224 vs 133 mg/L, P=0.011), and fever was also more common (71.4% vs 30.0%, P=0.071, not statistically significant)[6]. In the systematic review, those diagnosed more than 7 days after suspected onset had CRP 415.8±254.4 mg/L, versus 119.8±163.9 mg/L when diagnosed within 7 days (P=0.005); the original authors thought the 7 tuberculous infections in the late-diagnosis group might explain that group’s later presentation after reconstruction (31.4±33.9 vs 14.4±10.7 days) and higher CRP[15].
Low values likewise occur. The Swedish series had one case at 10 mg/L, the German series a minimum of 27 mg/L, and the Canadian series a minimum of 17 mg/L; subacute infection from low-virulence organisms can show only mild elevation[6,10,62]. Four further sources have uncertain CRP units. Beijing 2009 Table 2 labels CRP as mg/mL with normal range 0–0.8 and mean 8.3 (4.1–17.8); by the normal range this should be mg/dL, i.e., about 83 mg/L; the same table also prints fibrinogen as mg/mL (normal range 200–400)[5]. The US military hospital series Table 5 gives no unit, laboratory normal “<0.5,” mean 13.7 (5.5–19.4)—likewise more like mg/dL[9]. Boston Children’s Hospital reports a median CRP of 11.23 mg/L (6.77–23.17) among 9 with values[12]; the Costa study reports infection-group serum CRP 12.4±11.5 mg/L and noninfection 6.8±6.1 mg/L, and 9±7 mg/L with antibiotics before aspiration versus 15±12.9 mg/L without[65]. The last two sources are an order of magnitude lower than other series; neither original explains this. If the unit is truly mg/L, infections presenting with only mild CRP elevation are more common than the series above show; that possibility cannot be excluded. Comparisons in this chapter use only series with clear units.
ESR in infected patients is mostly 50–75 mm/h (Table 5), overlapping the 58–68 mm/h of uncomplicated or noninfectious-effusion patients on postoperative days 5–7[31,37]. Peripheral white cell count is not high in most infected patients: 7 of 21 in Beijing 2009 exceeded normal[5]; the systematic review’s pooled elevated proportion was 40%[15]. The Saudi series’ 14.2±4.2×10⁹/L is an exception; 83% of that series had diabetes and mean BMI was 37[8]. In Beijing 2009 the neutrophil fraction was 71.7% (60.3%–82.9%) versus 60.9% in controls, and fibrinogen 774.7 versus 580.1 mg/dL (P=0.007)[5]. Procalcitonin in infection after ACL reconstruction has case-by-case data only from the Saudi series of 12: mean 1.77±2.76 ng/mL, minimum 0.05 ng/mL, maximum 9.66 ng/mL[8]. Culture-positive joint infection can have procalcitonin as low as 0.05 ng/mL, consistent with the spine data in Section 4.
Synovial Fluid White Cell Count: Costa 2021
Synovial fluid white cells in infected patients are mostly 50,000–80,000/µL (Table 5); low values likewise occur: among Costa’s 24 infected patients the lowest was 1,765/µL, while 2 noninfected patients reached 32,900 and 28,070/µL[65].
The only ACL data that directly compare aspirated infected and noninfected patients are those of Costa et al. The Rizzoli Institute in Bologna retrospectively reviewed 38 patients readmitted and aspirated for suspected infection from 2009–2019 (24 infected, 14 noninfected); 36 came from that hospital’s contemporaneous 3,408 reconstructions, and 2 had been reconstructed elsewhere. Infection was defined as two specimens growing the same pathogen, or meeting three of the following four: CRP >10 mg/L plus ESR 30 mm/h, positive histology, intraoperative purulence, and a single positive culture[65]; the ESR item has no inequality sign in the Methods, and the supplementary table header is >30 mm/h. Synovial fluid white cells were 54,430±37,488/µL in the infection group and 12,543±12,352/µL in the noninfection group (P=0.0003; original units printed as cells/mL, which by magnitude should be /µL). Neutrophil fraction 87.9% versus 77.9% and ESR 75.7 versus 78.4 mm/h did not differ. Four infected patients in the supplementary table lacked synovial white cell counts; ROC analysis was based on 20 infected and 14 noninfected. Area under the ROC curve for synovial white cells was 0.89; at 28,100/µL, sensitivity 0.80 (16/20), specificity 0.93 (13/14), accuracy 0.85; at 40,000/µL, sensitivity 0.65 (13/20), specificity 1.00 (14/14); neutrophils ≥91% had sensitivity 0.63 and specificity 0.93. In the multivariable model only synovial white cells were an independent predictor[65].
These two cutoffs can serve only as partition points within that cohort. The authors stated four limitations: retrospective design; aspiration only when infection was suspected, with no asymptomatic synovial fluid as background; small numbers and predominantly male; and a diagnostic standard that includes culture results while culture is itself one of the tests under evaluation, thereby inflating specificity[65]. Synovial white cells and neutrophil fraction are outside the diagnostic definition and are unaffected by that last point, but three further problems remain: the cutoffs were derived and evaluated in these 34 counts, with no validation set; mean time to readmission was 36.1±39.7 days in the infection group and 13.7±7.5 days in the noninfection group, so the two groups sit at different postoperative times and have different normal backgrounds for synovial white cells; and 4 of the 24 were chronic infections more than 2 months after surgery. The figure 28,100/µL is the best partition point for distinguishing infection from noninfection among patients at this hospital who were suspected of infection and underwent aspiration; it cannot serve as a decision threshold at an arbitrary postoperative time point. The research protocol at the author’s hospital uses these two figures as pre-specified “supports infection” grading criteria, not as diagnostic standards. A contemporaneous editorial commentary treated 15,000/µL as a personal reference line for deciding return to the operating room for irrigation, and acknowledged that because case numbers are so small a unified decision value may never be obtained[66].
ESSKA and EBJIS recommendations give no unified post-ACLR cutoff. Thresholds in the literature are 17,500–50,000/µL for native-joint infection and 2,000–10,000/µL for prosthetic joints; low-grade infection with low-virulence pathogens is common after ACL reconstruction, and a lower threshold than for native joints is advised. The same recommendations state that synovial white cells below 2,000/µL can in most situations exclude infection, that a neutrophil fraction above 90% has a higher likelihood ratio, and that Gram stain has high specificity but sensitivity of only 29%–50%; synovial glucose, protein, lactate dehydrogenase, D-lactate, calprotectin, and α-defensin have not been validated in the ACL reconstruction population. The authors state frankly that the physiologic course of synovial white cells after ACL reconstruction is currently unknown[64]. The 2,000/µL exclusion rule cannot be applied to an individual patient after ACL reconstruction: in Costa’s supplementary table, a 16-year-old man febrile on postoperative day 10, without antibiotics before aspiration, had synovial white cells 1,765/µL and neutrophils 78%, with Staphylococcus aureus grown from both synovial fluid and intraoperative tissue, positive histology, and pus at surgery[65]. Among historical infection cases pooled by Paci there was also one methicillin-sensitive Staphylococcus aureus infection on postoperative day 9 with synovial white cells 750/µL[37]. These two cases prove only that high-virulence infection with a low count has occurred; they cannot estimate the miss rate. In the first postoperative week the white cell background of bloody effusion often exceeds 2,000/µL, so the rule has limited meaning for early effusion in any case (Section 6). When recounting Ruiz-Ibán 2015, the recommendations write “a 10-fold CRP rise has specificity for infection of 98% on day 4 and 96% on day 7,” yet Ruiz-Ibán had no day-4 sampling. Ruiz-Ibán set brackets at 25 and 50 mg/L as 5 and 10 times the laboratory upper normal of 5 mg/L; “10-fold rise” means above 50 mg/L. In its Table 2, proportions above 50 mg/L were 1.7% on day 1 and 4.4% on day 7, corresponding to proportions not exceeding 50 mg/L among these noninfected patients of 98.3% and 95.6%[34,64]. These two figures can be viewed as specificity within that noninfected sample; they are not diagnostic performance validated in patients with suspected infection. The companion prevention and surgery paper from the same series writes only that a rise to ten times normal in the first postoperative week suggests infection, without giving a number[69].
By contrast, a 2025 narrative review still writes that “synovial white cells above 20,000/µL suggest empyema, above 50,000/µL without crystals basically diagnose septic arthritis, and neutrophil fraction above 90% has sensitivity 94% and specificity 100%”—an entire paragraph hanging on a single incidence study as its source, without stating which patients the cutoffs came from[67]. A 2026 review recounts Costa’s 28,100 as the “most reliable” cutoff and still prints the unit as cells/mL[68]. The coexistence of both sets of numbers reflects the absence of validated post-ACLR synovial fluid cutoffs. Integer cutoffs that appear in reviews should have their source population checked before citation.
Noninfectious Painful Effusion: Paci 2010 and 16,200/µL
Paci 2010 retrospectively reviewed 151 reconstructions by 2 surgeons over 2 years; 31 underwent aspiration at a mean of 5.5 days postoperatively for painful effusion and were ultimately confirmed noninfected (negative culture and Gram stain; resolution without antibiotics). Mean synovial white cells in the 31 were 9,600/µL (SD 15,200), neutrophils 66% (SD 34), range 0–62,000/µL; all 6 with counts above 16,200/µL had red cells above 2,000,000/µL and were grossly visible hemarthroses. Concurrent ESR was 58.8±23.9 mm/h, CRP 6.1±4.9 mg/dL (i.e., 61 mg/L), and peripheral white cells 8.4±2.0×10⁹/L[37]. The authors treated the upper bound of the 98% confidence interval around the mean of the 31 (16,200/µL) as a “normal upper limit,” then compared it with pooled values from 95 published infection cases (Table 2 footnote notes partial missing data, so usable n for each marker is less than 95), obtaining sensitivity 86%, specificity 92%, and positive likelihood ratio 10.4[37].
This cutoff cannot be used, because the statistical concept is misapplied: a confidence interval around a mean describes where the mean may lie, not the range of individual values. Six of 31 (19%) exceeded 16,200/µL; the authors’ own data already refute a specificity of 92%. In Table 4, raising the cutoff increases sensitivity (86% at 16,200/µL, 92% at 50,000/µL) and decreases specificity (from 92% to 74%)—impossible in the same set of patients; that table’s calculations are likewise wrong. The authors also claim that removing the 6 hemarthrosis false positives would raise sensitivity to 97%, yet sensitivity is calculated only among the infected, and removing noninfected cases cannot change it. What can be used are the descriptive facts: painful effusion within the first postoperative week is mostly bloody; synovial white cells in noninfected patients can reach 62,000/µL, with neutrophil fraction above 80%; CRP can sit around 60 mg/L and ESR near 60 mm/h[37]. These figures are direct grounds for considering hematoma first when interpreting any early postoperative marker.
Culture-Negative Cases and Antibiotics Before Aspiration
Culture-positive rates across series range from 42% to 100% (Table 5); one narrative review gives a culture-negative proportion of 15%–30%[70]. Antibiotics before aspiration are a documented cause of culture-negative results: among 6 negatives in the German series, 2 had already received oral antibiotics before submission, and another 5 had received antibiotics before aspiration at an outside hospital[6]; among 5 Canadian cases that grew coagulase-negative staphylococci, 2 had already been given oral antibiotics by a family physician before referral, and another 4 had no growth—the authors’ conclusion was not to give any antibiotic before aspiration[62]; among 2 Swedish negatives, 1 had taken oral cloxacillin for a week before aspiration[10]. In the Costa study, among 9 infected patients who had received antibiotics before aspiration, synovial culture was falsely negative in 5 (55.6%), versus 4 of 15 (26.7%) who had not; corresponding culture sensitivities were 44% and 73%, a difference not reaching statistical significance (P=0.21). Synovial culture sensitivity was 0.63 and intraoperative tissue culture 0.96 (P=0.0045)[65]. Coagulase-negative staphylococci are the leading pathogen in most series (2014 systematic review 45.6%, Staphylococcus aureus 23.8%; 2025 meta-analysis of 333 cases 42.3% and 23.1%, culture-negative 18.3%)[15,18]. These organisms grow slowly and have low virulence; CRP and the proportion with fever at presentation may both be lower (see the German series above)[6].
When culture is negative, management as infection still rests on criteria that differ by center. The German series relied on intraoperative findings (pus, fibrin deposition, synovial thickening) plus clinical presentation[6]. Costa’s diagnostic standard combines CRP and ESR, histology, intraoperative purulence, and culture; synovial white cells are outside the definition[65]. ESSKA and EBJIS recommend sending synovial fluid simultaneously for microbiology, cell count with differential, and crystal analysis, because postoperative gout or pseudogout can coexist with infection or alone produce synovial fluid with a very high white cell count and negative culture[64]. The prevention and surgery paper writes the sequence as a rule: suspected infection requires aspiration, and aspiration must precede any antibiotic[69]; the diagnosis paper notes that this principle applies to patients without sepsis[64]. Actual practice is not always like this: among 17 infections in Caen from 2010–2022, the 9 “uncertain” cases were aspirated, and the rest were irrigated directly after diagnosis based on fever, stiffness, pain, limited motion, and elevated inflammatory markers, with specimens taken intraoperatively[14].
Consequences of Infection
A 2024 systematic review and meta-analysis included 13 studies, of which 8 compared 129 cases of septic arthritis after ACL reconstruction with 6,644 uncomplicated patients: infected patients scored 7.53 points lower on Lysholm (95% CI 3.20–11.86), 1.42 lower on Tegner (1.07–1.76), and 10.45 points lower on subjective IKDC; return-to-sport rates were 40/76 (53%) among infected and 1,427/1,873 (76%) among uncomplicated, with pooled RR 1.56 (1.24–1.95) as the ratio of uncomplicated to infected. Objective stability and radiographic osteoarthritis incidence did not show statistically significant differences between groups[17]. In a 2025 meta-analysis of 20 studies and 333 cases, the pooled graft-retention rate after debridement plus antibiotics was 92% (95% CI 88%–94%)[18]. In the 2014 systematic review, graft removal occurred in 7/25 when more than 7 days elapsed from suspected onset to diagnosis, versus 0/31 when diagnosed within 7 days[15]. The late-diagnosis group included 7 tuberculous infections, all of whose grafts were retained; the 7 removals came from the remaining patients. This comparison made no adjustment and can show only that delayed diagnosis is associated with graft removal; it cannot quantify the effect of delay itself. Another 2025 systematic review (body 23 studies, 269 cases; Abstract 24 studies, 307 cases) compared retention with removal: removal had subjective IKDC 13.6 points lower (66.6 vs 80.2, P=0.02) and greater KT-1000 laxity (3.30 vs 1.6 mm). Pseudomonas aeruginosa infection (removal rate 60% vs 14.4%) and allografts were more often removed; the authors state they cannot separate stronger pathogen virulence from greater surgeon inclination to remove[19].
Long-term outcome has one matched controlled study. Among 2,315 reconstructions in Caen, 20 infections (0.86%); 17 were treated with arthroscopic irrigation, graft retention, and 6 weeks of antibiotics, matched 1:1 to noninfected patients by age, sex, BMI, meniscal and cartilage injury, graft type, follow-up length, and smoking. At a mean follow-up of 7 years, Lysholm (89.1 vs 92.1), subjective IKDC (82.9 vs 88.5), KOOS (89.9 vs 91.7), and Tegner (4.9 vs 5.5) showed no statistically significant differences, yet return to sport was about 2 months later (10.2 vs 8.1 months, P=0.01), and return to preinjury level was 53% versus 65% (P=0.49). Of the 17, 1 needed a second irrigation and none had the graft removed; the authors note the small sample and low power[14]. These data quantify the consequences of infection and of delayed diagnosis; the cost of misclassifying a case as infection—an unnecessary debridement, weeks of antibiotics, even graft removal—has not been quantified in the material included in this chapter, and the two cannot be weighed on the same scale.
Fall of Markers After Treatment
CRP responds to treatment faster than ESR. In Wang 2014’s infection group, CRP peaked on day 1 after treatment (95.0±64.6 mg/L), was 13.5±13.5 mg/L on day 14, and 8.6±10.7 mg/L on day 21; ESR peaked on day 3 (64.2±30.4 mm/h) and fell to 19.5±16.8 mm/h only on day 28[33]. Using CRP to monitor treatment response is faster than using ESR; that point can be adopted. The French FAST cohort’s management rule was to irrigate and debride again if CRP at 48-hour intervals remained above 100 mg/L or failed to fall[11]; the German series rechecked CRP every other day during the hospital stay and again at 1 week and 4–6 weeks after stopping antibiotics[6]. These two are center-specific operating habits and have not been validated. Statements in reviews such as “stop antibiotics when CRP returns to the normal range,” “switch to oral after CRP falls,” and “consider revision only after CRP is normal” likewise have no numerical basis[18,67,68].
6. Conditions Easily Confused with Infection
Hematoma and bloody effusion. This is the most common early postoperative alternative explanation and the one that most readily overlaps infection. All 31 of Paci’s noninfectious painful effusions were bloody (Section 5)[37]. The Croatian series classified 3 patients with CRP 40–70 mg/L, negative culture, and rapid improvement after aspiration plus antibiotics as sterile effusion[7]. In an Indian Chandigarh series of 26 suspected infections, 9 were ultimately classified as sterile effusion: at onset CRP 45.2 mg/L, ESR 31 mm/h, synovial white cells 4,200/µL, neutrophils 21%; CRP 37.6 mg/L at 48 hours and 10.0 mg/L at 1 week. The 17 infected patients had onset CRP 92–93 mg/L, synovial white cells 71,350–78,700/µL, and neutrophils 72%–84%[71]. That series’ partition of “sterile effusion” versus “infection” depended in part on whether treatment led to improvement and is not a judgment independent of treatment; criteria for “moderate/severe virulence” grouping are also undefined in the Methods[71]. Among 1,053 US quadriceps-tendon reconstructions, 4 patients returned to the operating room for irrigation and debridement, 2 of them for harvest-site hematoma; all synovial cultures were negative[72]. Hematoma and infection are also not mutually exclusive: in the French FAST cohort’s multivariable analysis, immediate postoperative hemarthrosis was an independent risk factor for infection (OR 127.2, 95% CI 16.0–1,011.3; the interval spans two orders of magnitude, events are few, and only direction can be taken), and prior knee surgery had OR 15. In that cohort, 5.6% of patients had adverse events in the first week, including 82 popliteal hematomas and 12 hemarthroses[11]. How much hematoma itself raises CRP and synovial white cells has never been quantified separately; “hematoma provides a culture medium for bacteria” is a common conjecture. When bloody effusion is accompanied by elevated markers, blood markers cannot distinguish the two; the safe sequence is to aspirate first, send culture, cell count, and crystal analysis, and then decide whether to observe.
Infection of the incision and harvest site. Synovial fluid from extra-articular infection may be nearly normal. Among Judd et al.’s 12 extra-articular infections, 4 were aspirated, with cell counts 1,133/µL (427–2,850), versus 52,735/µL (25,400–72,800) for intra-articular infection; elsewhere in the body the figures are written as 4,170 versus 52,000/µL—the two places disagree numerically, but a more than tenfold difference between extra-articular and intra-articular counts holds in both[9]. Patients with wound drainage or harvest-site swelling and pain whose synovial fluid is unremarkable will be missed by an “aspirate only if there is effusion” approach; the wound must be examined directly.
What imaging can add. Among 11,451 reconstructions in the New York HSS registry from 2010–2018 there were 48 infections (0.42%), of which 17 had MRI around the time of debridement (median 1 day from debridement, range 8 days before to 10 days after; hamstring 11, BPTB 4, allograft 2). Both radiologists saw joint effusion in all 17 knees. The following intervals are the two readers’ respective proportions: complex layered effusion 70.6%–76.5% (κ 0.86), peri-tunnel edema 94.1%–100%, subcutaneous edema 82.4%–100%, fluid within tunnels 47.1%–52.9% (κ 0.91, more often femoral), and tunnel bone resorption 47.1%–58.8% (κ 0.83). All 8 acute infections (onset within 30 days) had complex layered effusion; simple effusion was seen only in chronic cases. Graft type and pathogen were unrelated to MRI findings[13]. This study had no noninfected controls; the authors themselves state that specificity of these findings for infection cannot be shown (Section 1).
Inflammatory arthropathy and gout. A 2026 case report records an inflammatory arthropathy long managed as culture-negative infection. A 21-year-old man developed knee swelling 12 weeks after BPTB reconstruction with repair of both menisci; aspiration cell count 94,500/µL, neutrophils 92%, ESR 68, CRP 68 (units not labeled); Gram stain, culture, and crystals all negative. Three arthroscopic debridements followed; the hip and ankle successively developed effusions and underwent open debridement. All intraoperative cultures, blood cultures (one positive judged contaminant), and 12 pathogen studies were negative; a third aspiration showed cell count 169,000/µL, neutrophils 89%. Mildly elevated rheumatoid factor and complement C3 finally pointed to seronegative inflammatory arthropathy; methotrexate plus prednisone brought complete relief, and at 19 months postoperatively he returned to the soccer field[73]. A case report can only prove that such a situation has occurred; the authors also state that a causal link between surgery and polyarthritis was not established. However high the synovial cell count and neutrophil fraction, they only support infection and cannot alone confirm it. When cultures are repeatedly negative, multiple joints are involved, and there is no response to antibiotics and debridement, inflammatory arthropathy should enter the differential. Postoperative gout can likewise produce synovial fluid of purulent appearance; in one Beijing series a patient with concurrent gout attack had the graft removed[74].
Venous thrombosis. Calf swelling, low-grade fever, and a renewed CRP rise can come from deep vein thrombosis. Among uncomplicated primary arthroplasty patients, D-dimer above 860 ng/mL accounted for the majority at all 5 time points measured within 45 days postoperatively; that cutoff alone cannot judge infection[41]. One of 2 complicated patients in the Chilean cohort had venous thrombosis, with CRP rising rather than falling on day 7[50]. In the Osaka cohort of ACL reconstructions managed with mechanical prophylaxis only, day-7 CT found thromboembolism in 11%; 14 of 15 thromboses were asymptomatic. Classification in the Results is written as isolated pulmonary embolism 7, isolated deep vein thrombosis 5, and both 3; case-by-case Table 2 instead has 5, 7, and 3, with those including pulmonary embolism 10 versus 8; the original does not explain[42]. Absence of calf swelling cannot exclude thrombosis. When CRP rises again without new joint effusion, perform venous ultrasound first; day-7 D-dimer may be referenced against that cohort’s derived 2.8 µg/mL—a cutoff that has not been validated (Section 4)[42].
Infection at other sites and drug reaction. Among patients with new fever within 10 days after orthopedic surgery, the leading infectious causes are respiratory and urinary; joint infection accounts for only a minority[52]. Upper respiratory infection, urinary tract infection, and drug fever from antibiotics or other medications can all raise CRP again. These situations are approached first by history and examination, then by targeted tests along the clues—urinalysis, chest radiograph, or respiratory pathogen testing. Such directed checks change the judgment more than another CRP.
Intra-articular injection. In the Indian randomized trial of 90 hamstring reconstructions, early or delayed postoperative intra-articular hyaluronic acid versus placebo produced no difference in ESR or CRP medians at weeks 4, 8, 12, or 52; the delayed-injection group had 1 harvest-site infection on postoperative day 5 that the authors judged unrelated to injection[38]. Whether intra-articular injection itself causes a transient rise in markers cannot be answered by this trial’s sparse blood-draw schedule.
7. Why Postoperative CRP Cutoffs Are Missing, and Establishing Reference Distributions
Available material cannot answer how high CRP after ACL reconstruction must be to count as abnormal.
No upper-tail percentiles. Judging where a value sits among those with normal recovery requires upper-tail statistics such as P90 and P95; the five cohorts give only mean±SD. Mean±2SD is an interval derived under a normal distribution, whereas postoperative CRP is clearly right-skewed; an “upper limit” calculated this way may be higher or lower than the true P95 depending on the shape of the distribution, has no fixed relationship to it, and cannot be used as P95. The same algorithm’s “lower limit” can also fall below zero, which itself shows that the premise does not hold. The proportions above 5, 25, and 50 mg/L reported by Ruiz-Ibán and the numbers reaching 30 mg/L reported by Uchida are the only upper-tail data[34,35].
No exact blood-draw time. CRP changes by tens of milligrams per liter each day between 24 and 96 hours after surgery. For the same “day 3,” elapsed time from surgery still varies with the clock times of surgery and of sampling. All five studies report only postoperative day, without stating the counting method; Ruiz-Ibán’s last four draws also allow ±3 days[34]. “Day 3” across studies cannot be compared directly, and even within one study values at the same labeled time point mix different elapsed times.
Mixed populations and techniques. Day-3 means for BPTB and hamstring differ by a factor of two[32]; patients of less experienced surgeons are about 2.5 times higher, microfracture about 2 times, men 1.7 times, and those with elevated preoperative CRP 1.5 times—the last four being ratios of geometric means in the Ruiz-Ibán cohort[34]. The mean of a mixed population cannot represent any of its subgroups; that is the reason for estimating distributions stratified by sex, technique, and perioperative pathway.
No same-time-point comparison with infection cases. The only study that attempted a comparison, Wang 2014, put day-5 normal values and day-13.7 infection values together for ROC (Section 2)[33]. CRP in infection cases is drawn when symptoms trigger sampling; timing and number of tests vary, “day of onset” is the day symptoms appear, and from which day the marker begins to diverge is unknowable. Matching uncomplicated cohorts to the same calendar day as the symptom day still has a time mismatch, only smaller than Wang’s. Available ACL material therefore cannot reliably estimate sensitivity and specificity on a fixed postoperative day. A retrospective cohort after open colorectal cancer resection offers a contrast: Warschkow et al. extracted CRP for each postoperative day from the records of 1,187 patients and analyzed by day. Among 314 with CRP on day 4, 103 developed inflammatory complications; at 123 mg/L, sensitivity was 0.66, specificity 0.77, area under the curve 0.76; areas under the curve on days 2 and 3 were 0.64–0.69[76]. The authors considered day-4 CRP usable for screening but not discriminating enough to yield a single cutoff, and that CRP cannot serve as a black-or-white decision standard; they also acknowledged bias from the retrospective design and recommended confirmation in prospective work. The design point is same-day comparison of infected and noninfected patients, with cutoffs calculated separately by day; the outcome was all inflammatory complications, broader than joint infection. Inflammatory complication incidence in that cohort approached 30%, so such a comparison accumulates easily. Infection after ACL reconstruction is under 1%; accumulating 50 infections requires about 5,000 reconstructions at 1% incidence and about 10,000 at 0.5%, before counting cases missing a blood draw on the designated day. The upper tail of an uncomplicated cohort cannot substitute for this design; Section 1 treats the upper tail as a prompt for rechecking, which is not the same as a day-specific derived cutoff.
Different laboratory methods. Calvisi 2008 had a detection limit of 6 mg/L[32]; Margheritini 2001 used immunoturbidimetry and stated only normal <6 mg/L[31]; Ruiz-Ibán 2015 laboratory sensitivity was 0.01 mg/L[34]; Uchida 2024 detection limit was 0.5 mg/L[35]. Readings in the low range from different methods are not directly interchangeable. When assay methods and reference populations are comparable, another laboratory’s reference interval can be adopted after verification[78]; if postoperative CRP distributions are borrowed across centers, assay method and perioperative pathway must first be checked and the distribution verified with local data.
Reference-interval methodology already has ready treatments for the upper tail and for stratification. Clinical laboratory reference intervals usually take the central 95% of the reference population, i.e., P2.5 to P97.5. The recommendation retained by CLSI and IFCC is at least 120 reference individuals, nonparametric estimation of the limits and their confidence intervals, and 120 per stratum when stratified by sex[77,78]. Smaller numbers can still estimate reference limits, but confidence intervals may be very wide; Ozarda’s review cites that estimating the 90% confidence interval of a reference limit by bootstrap requires at least about 80 cases for an acceptably narrow interval[78]. The figure 120 is a sample-size recommendation and does not guarantee precision of the upper-tail estimate: drawing 120 from the same population, the expected number above the true P95 is 6 and above P97.5 is 3; upper-tail estimates of a postoperative dynamic distribution are decided by a few high values. The same methodological review distinguishes reference intervals from clinical decision limits: the former describe the distribution in a defined population; the latter are action thresholds derived by ROC curves and predictive values from comparison of patients and controls. The two must not be mixed[78]. Wang’s 41 mg/L and Paci’s 16,200/µL belong to the latter class, and the controls used to derive them were unsuitable. Reference-interval methods address a static distribution at a single time point; postoperative CRP changes with time, and direct application would mix values from different moments.
For reference values that change with time, the growth-curve field has two methods. Cole and Green’s 1992 LMS method summarizes the distribution with three age-varying smooth curves: L is the Box-Cox power (skewness), M the median, and S the coefficient of variation; the premise is that measurements are positive and approximately normal after power transformation. Curves are fitted as cubic splines by penalized likelihood, with smoothness expressed as equivalent degrees of freedom[79]. Postoperative CRP may be 0 or below the detection limit, and a single power transformation may not make the distribution approximately normal; the author therefore does not take LMS as the primary scheme. Neither methodological paper has compared the two methods on postoperative CRP. Quantile regression makes no distributional assumption, estimates conditional quantiles at each time directly, and uses splines to describe change over time; covariates are easily included (the original uses prior growth history and parental height as examples—corresponding here to sex, BMI, and preoperative CRP). Standard errors can be obtained by bootstrap, with the original resampling each child’s entire longitudinal record as a unit; a patient’s repeated postoperative draws should likewise be one unit. Compared with LMS it is more flexible and more readily exposes mismatch between data and model assumptions[80].
Combining the two tools, a feasible design takes exact postoperative hours as the x-axis, estimates empirical percentiles and continuous percentile curves at each time point among uncomplicated patients on a consistent perioperative pathway, then overlays this center’s infection cases’ CRP one by one on that band to describe when and how they diverge. Two points must be written into the protocol: judgment of “uncomplicated” cannot use CRP itself, or the reference population will already have been screened by CRP and the resulting distribution will no longer represent all uncomplicated patients; and at centers that soak routinely with vancomycin, infection counts will be fewer than the 0.5%–1% estimate, so the registry period must be extended accordingly[3]. What this yields is the distributional range of CRP during normal recovery; it does not answer “above what value is infection.” The latter requires comparison of infected and noninfected cases at the same postoperative time point—a different research question. Until such data appear, in the author’s view the figures from the uncomplicated cohorts in Section 2—especially the proportions above fixed values reported by Ruiz-Ibán and Uchida—can only be used to judge whether a value clearly exceeds the published upper tail of normal recovery, and cannot be used to exclude infection. Paci’s symptom-aspirated noninfected cases and the infection series in Section 5 describe the range of overlap of these two patient classes with normal recovery and do not participate in defining the normal upper tail.
8. Data Discrepancies in the Source Literature
The table below summarizes numerical, unit, and wording discrepancies in the cited studies and how this chapter takes their values.
| Study | Problem found | How read |
|---|---|---|
| Wang 2014[33] | Table 2 “±” for both groups unspecified as SD or SE; post-treatment blood-draw day written as day 10 in Methods and day 9 in Results; Table 3 rate values imply analysis counts of 34 infected and 82 controls, not matching group sizes 39 and 83, difference unexplained; Discussion swaps false-negative and false-positive rates for 41 mg/L (2.4%, 5.9%) relative to Table 3, and 4.3%/34.0% for 32 mm/h do not match Table 3’s 8.8%/19.5%; Table 1 temperature footnote says both groups measured on postoperative day 3, inconsistent with infection-group symptom onset; infection group (day 13.7) and noninfection group (day 5) time-point mismatch, from which the 41 mg/L and 32 mm/h cutoffs were derived; same institute as Wang 2009, overlapping infection observation windows | Adopt day-3 and day-5 means and descriptive infection-group data; do not adopt cutoffs or temperature comparison |
| Xiang 2024[36] | Abstract and opening Discussion say “no between-group difference except at 1 month; at 3 months only white cells remain,” yet Supplementary Tables 1–5 show P<0.001 for all five markers at day 3, 1 week, and 2 weeks, and later Discussion says “all lower within the first month” (internal inconsistency); blood-draw arrangements and missingness for the three earlier time points not stated; week-2 and month-1 CRP and ESR means higher than every other cohort (week 2 is 1.9–13 times contemporaneous cohort means; month 1 is 1.6–4.4 times); month-1 white cell difference 3.4×10⁹/L; almost every between-group comparison P<0.001; week-2 CRP SD 18% and 23% of the mean (four uncomplicated cohorts contemporaneously 68%–152%); sampling clock time and detection limit not stated | Values not adopted: blood-draw arrangements and raw distributions not stated and cannot be checked—not a judgment that results are wrong; existence recorded only (Section 2, Table 3). Groups also differ in number of tendons harvested, femoral drilling, and fixation; between-group difference cannot be attributed to tibial tunnel alone |
| Wang 2021[47] | Day-1 and day-3 CRP 1.18–1.23 mg/L in 716 cases is anomalous in magnitude and does not rise; case-selection figure differs by 45 at two places (1,178−451 should be 727, figure writes 772; after subtracting 11 missing, group totals are 716); cytokine units unlabeled | Absolute CRP not adopted; use only “irrigation has no effect” and IL-6 timing |
| Paci 2010[37] | Treats 98% CI around the mean as a reference range; 6/31 above 16,200/µL contradicts specificity 92%; Table 4 sensitivity at 50,000/µL higher than at 16,200/µL; Abstract and Results disagree on CI (2,800–16,200 vs 2,900–16,200/µL), neutrophil fraction interval (58%–84% vs 51%–81%), painful-effusion rate (23.5% vs 20.8%), and aspiration-time SD (Results 1.6 days; Table 1 and Discussion 1.5 days); Table 2 unit g/dL; claim that removing 6 hemarthroses raises sensitivity to 97%, yet sensitivity is calculated only among the infected and is unaffected; usable n for each marker among 95 historical infection cases varies | Adopt descriptive values from the 31; do not adopt the cutoff |
| Mishra 2018[20] | Infection incidence written 0.01% (actually 17/1,152=1.48%); double-bundle white cell mean 129000 (table converts to about 12,900); single-bundle S. aureus 2 cases (50%) disagrees with Table 1b’s 3; median age and KT-1000 means disagree with tables | Values not adopted in this chapter |
| Honsawek 2011[39] | Original labels CRP units as mg/dL; authors’ 2011 formal erratum (Int Orthop. 2011;35(11):1749) states entire paper should be mg/L | Adopt CRP and IL-6 timing as mg/L per erratum |
| Backes 2013[60] | Trial did not measure CRP or IL-6; Discussion cites CRP-reduction data from others’ work, yet secondary literature often cites them as that trial’s results | For CRP/IL-6 evidence cite Xu 2018 instead; do not treat as CRP evidence |
| Konstantinou 2025[67] | Synovial fluid cutoff paragraph (>20,000, >50,000, neutrophils >90% sensitivity 94%/specificity 100%) cites only one incidence study; cutoff source unclear; culture-negative rate worded differently in two places | Use only as an example of coexisting cutoffs; do not adopt its cutoffs |
| Margheritini 2001[31] | Abstract blood-draw schedule “days 1, 3, 7, 15, 30,” Methods “days 2, 3, 7, 15, 30”; Figure 1 has a day-1 data point (reads about 13 mg/L) but no numerical value; right-skewed data analyzed with paired t-tests | Do not list figure-read values; P values for reference only |
| Calvisi 2008[32] | Units mg/dL (e.g., day-3 overall 3.73±2.71); body and Abstract never use mg/L—conversion needed when citing; BPTB vs hamstring not tested between groups | Convert to mg/L; between-group difference descriptive only |
| Ruiz-Ibán 2015[34] | Abstract 117 patients, 119 reconstructions; Results and Table 1 write 119 patients; Abstract unit mg/ml; “24 (27.4%)” should be 33.3%; sex effect 1.7 (1.8–2.5) interval excludes the point estimate; microfracture 1.9 (Abstract) vs 2.0 (Table 3); Methods say multivariable logistic regression, statistical note says linear regression after log transformation, Table 3 calls exponentiated coefficients for a continuous dependent variable odds ratios; authors say specificity cannot be estimated without infection controls—wording too broad | Adopt means, proportions, and factor direction; effect sizes for order of magnitude only |
| Uchida 2024[35] | Mean age body 24.2, Table 1 24.0; Table 4 meniscectomy row interval printed “−0.041 to −0.130” (recomputed from SE 0.043 and β 0.048 should be −0.041 to 0.130); Table 3 same row printed “−0.445 to 0.53” (from SE 0.152 and β −0.045 should be −0.445 to 0.153); β column is standardized coefficients, CI is interval for unstandardized coefficients, original does not explain (Table 4 sex row β 0.304, CI 0.140–0.282: unstandardized coefficient recomputed from SE 0.036 is 0.211; × sex SD 0.50 ÷ day-14 CRP SD 0.35 = 0.30, compatible with β; Table 3 sex row β 0.316, CI 0.525–1.020, SE 0.126: 0.773×0.50÷1.22=0.32 likewise compatible); Abstract and Discussion claim these zero-complication data help exclude postoperative complications—data cannot support that | Adopt principal values; do not use to exclude infection; sex-row β and CI can be reconciled and are not treated as a typesetting problem |
| Mendias 2013[43] | Time for CRP to return to preoperative: Results say 2nd follow-up visit, Discussion 3rd | Qualitative citation only |
| Zheng 2026[45] | Figure 4 time points “preoperative, days 5, 10, 15” disagree with Methods and Figures 1 and 2; platelet unit 10¹²/L | Direction only |
| Nagashima 2023[46] | Attribution of day-1 CRP difference: Abstract (blood loss and soft-tissue injury) differs from Discussion (hemostatic trauma to synovium causing synovitis); historical control unadjusted | Direction only |
| Oestervemb 2025[48] | CRP unit Methods and Table IV mg/L, Results body μg/ml (equivalent, only inconsistent notation); fingertip point-of-care assay not validated | Adopt day-3 means only; values after 2 weeks not merged with venous-blood data |
| Costa 2021[65] | Diagnostic definition includes CRP, ESR, histology, purulence, and single culture—the culture under evaluation is inside the definition (incorporation bias stated by the authors themselves; synovial white cells and neutrophil fraction are outside the definition and unaffected by this); infection and noninfection groups differ by about 22 days in time to readmission; synovial CRP P value Table 1 0.0059, body 0.0068; synovial white cell unit printed cells/mL; 4 infected patients lack synovial white cell counts, ROC based on 20 infected and 14 noninfected; Table 1 and Table 5 serum CRP (12.4 vs 6.8; 9 vs 15 mg/L) far below other series, unexplained in original | 28,100 and 40,000/µL described only as partition points within that cohort; not adopted as cutoffs |
| Renz 2023 (ESSKA/EBJIS)[64] | Recounting Ruiz-Ibán 2015 writes day-4 specificity 98%, day-7 96%; that study had no day-4 sampling—corresponds to day 1 | 98.3% and 95.6% are only proportions not exceeding 50 mg/L in that noninfected sample; not validated diagnostic performance |
| Wang 2009[5] | Infection rate Abstract and Results 0.52%, Discussion 0.51% (21/4,068=0.516%); intravenous antibiotic days disagree in three places (Methods mean 19.4 days, 13–38; Table 3 mean of 21 cases 21.8 days; Results surgical group 19.4 days, 13–28); CRP and fibrinogen unit mg/mL, by normal range should be mg/dL; observation window 1997–2007 falls inside Wang 2014 infection group’s 1997–2010, same institute | CRP values do not enter comparisons; not treated as an independent series from Wang 2014 |
| Schollin-Borg 2003[10] | Diagnostic delay body 4–36 days, Table 2 case-by-case 0–34 days; ESR unit printed mm Hg; synovial white cell reference range applies peripheral-blood norms | Adopt Table 2 case-by-case values |
| Judd 2006[9] | Study period Abstract 1994–2002, Methods January 1994–December 2001; infection rate body 0.68% of 1,615, Table 8 writes 0.62%, all 11 occurring among 418 cases in 1999–2001 (2.6%); extra-articular infection synovial cell count body 4,170, Table 5 1,133; CRP unlabeled; Table 5 footnote CPR | CRP values do not enter comparisons |
| Kunovac 2020[7] | Culture-negative proportion written 58.69%, positive 37.98%, disagreeing with 15/26 and 11/26; white cells and ESR unlabeled; PubMed record reverses author names | Use flowchart 15/26, 11/26; authors per PDF first page |
| Ma 2024[61] | Enrollment start year Abstract 2011, Methods 2010; CRP unit diagnostic criteria g/L, Abstract mmol/L, Table 1 mg/L | Use Table 1 mg/L |
| Chona 2025[12] | Peripheral white cells and platelets unit printed cells/mL; Table 4 subgroup onset-time ranges disagree with body; CRP median 11.23 mg/L far below other series | CRP values do not enter comparisons |
| Ascione 2019[16] | Diagnostic cutoff unit /µL, Results unit cells/mL; Table I synovial white cell interval “52735-11300” lower bound greater than upper (body 52–113×10³); antibiotic course range Table 6–10 weeks, body 8–10 weeks; Table I prints CRP as PCR | Adopt median and culture-positive rate only |
| Kim 2012[63] | Observation window Abstract and Methods 2006–2008, Discussion 2003–2005 | Adopt case-by-case values |
| Lin 2026[75] | Onset day Abstract and case presentation day 5, Table 1 and Discussion day 3; “day 5” CRP case presentation 18.360 mg/dL, page 3 elsewhere 4.810 mg/dL | List both onset days side by side, take neither alone; CRP per case presentation |
| Miltenberg 2026[73] | Table 1 white cells, ESR, CRP unlabeled | Case only |
| Cassano 2024[70] | Same paragraph citing the same source gives RR 0.23 and 1.035 in opposite directions for autograft vs allograft; PubMed record missing article number | Cite only culture-negative proportion and pathogen composition |
| Gupta 2018[71] | “Moderate/severe virulence infection” grouping criteria undefined; sterile-effusion judgment depends on treatment response; “4 of 5 (66.7%)” arithmetic error | Adopt descriptive values |
| Musso 2005[62] | Debridement count in same paragraph written as 8 and as all 9 | Adopt case-by-case laboratory values |
| Wang 2014 systematic review[15] | Graft-removal rate when diagnosis delayed >7 days: Table 4 P=0.002, body 0.02; included studies’ definitions of septic arthritis inconsistent | Adopt table values |
| Seilern und Aspang 2024[72] | Time from reconstruction to debridement Abstract and body 18–23 days, Table 2 case-by-case 22, 27, 20, 18 days; PDF 2024 online first, PubMed record 2025 | Cite only the 2 hematoma cases |
| Vigushin 1993[22] | Inpatient count Abstract 35, Methods 33, Table I groups sum to 40; self-citation at head of paper volume 90 disagrees with masthead 91 (record takes 91 per masthead and PubMed); “falls by at most 50% per day” is a citation of prior observation | Cite only the 8 healthy subjects’ half-life and decline rate |
| Neumaier 2006[24] | Abstract calls Table 2 means (8.7, 12, 14, 16, 16 mg/dL) medians; Table 2 median column is actually 5.7, 9.8, 14, 17, 15 mg/dL; Methods write mean (SEM), Table 2 columns are mean, CI, quartiles, and median | Use Table 2 means |
| Neumaier 2008[25] | Abstract writes ankle day-2 median 45 mg/L; table 39 mg/L | Use table |
| Windisch 2016 and 2017[29,30] | Two reports of the same cohort (same institution, same period, same ethics approval number; full-cohort CRP values identical) | Not two independent observations: 2016 for sex difference, 2017 for inability of early trajectory to identify infection |
| Azboy 2021[41] | TKA vs THA day-3 D-dimer P value Results 0.0019, Discussion 0.01; Discussion values do not match Table 1 units; age-group unit ng/dL; Results say day-1 rise “about 9-fold,” by Table 1 means 11-fold | Adopt Table 1 whole-cohort values only |
| Chung 2011[51] | PCT cutoff Abstract 0.25, Discussion and Conclusions 0.5 ng/mL; Table 2 day-5 ESR SD 0.61 (day 3 is 11.82) suspect; Figure 1 legend hypertension interaction P printed 0047 | Directional conclusions only |
| Hunziker 2010[52] | CRP detection limit Methods mg/dL, Table II mg/L | Directional conclusions only |
| Zheng 2020[55] | Six cutoff sensitivities (87.5%, 94.1%, 78.6%, 80.0%, 58.8%, 52.9%) cannot be calculated from 7 infections; specificities also fail to match 83 cases; algorithm unexplained in original; PCT reference unit ng/L; noninfection PCT SD 0.96 greater than mean 0.12; day-6 ESR P body 0.002, Table 4 0.006 | Diagnostic performance and cutoffs not adopted; only direction of between-group mean comparison |
| Zhang 2023[2] | Table 2 and body disagree on meniscal-repair OR (0.92 vs 0.98) | Use Table 2 |
| Kottek 2025[3] | Data-availability statement contradicts retrospective extraction; control group cited from Kunovac 2020; sole infection case’s CRP written only “markedly elevated” | Adopt incidence only |
| Lin 2024[17] | Body writes KT-1000 difference 0.13 (95% CI 0.75–1.01), missing the minus on the lower bound; Figure 6A is 0.13 (−0.75 to 1.01), P=0.77; KOOS total I²=96% | KT-1000 per Figure 6A; no between-group difference detected |
| Hashimoto 2023[42] | Cutoff unit mixes µg/mL and mg/mL; sensitivity/specificity two sets 80.0/83.5 and 92.3/82.2; thrombus subtype proportions Results (PE 5.1%, DVT 3.7%, both 2.2%) disagree with Discussion (6.2% and 5.4%); Results classification isolated PE 7, isolated DVT 5, both 3; Table 2 case-by-case 5, 7, and 3 (those including PE 10 vs 8) | Totals, symptoms, and cutoff per Abstract and tables; PE counts listed side by side, take neither alone |
| Mustamsir 2025[18] | Body prints heterogeneity-test P as 2.0948; Figure 3 labels 2.0948 as τ² and writes P=0.69, I²=0%; Tegner means 4.08 and 5.08; incidence lower bound 0.14% and 0.4% | Adopt Figure 3 retention rate (306/333, 92%) and pathogen composition |
| Zhang & Gong 2026[68] | Costa cutoff unit printed cells/mL; Table 1 mixes animal and in vitro studies | Example of recounting only |
| Meza 2025[13] | “Fluid traversing the graft” Methods say only 1 excluded (should leave 16), Table 2 footnote calculates on 15; Table 2 peri-tunnel edema 94.1%/100%, body elsewhere writes femoral peri-tunnel edema 64.7%—whether the same variable is unclear; no noninfected controls | Adopt Table 2 proportions and agreement; make no specificity judgment |
| Balagod 2025[38] | Abstract says ESR and CRP measured monthly to 6 months; Table 4 only weeks 4, 8, 12, 52; injury-to-surgery written “mean 6 weeks (range 10 weeks–2 years),” mean outside the range; Table 4 has no time-point P values | Adopt Table 4 medians only as reference beyond 1 month |
| Cole & Green 1992[79] | M written first as median then as mean on the same page; US girls example SD score mean printed “0901” | Method points only |
| Warschkow 2011[76] | Body says white cells “differed at every time point except day 2,” Table 3 day 2 P=0.025; day-4 cutoff for all inflammatory complications 123 mg/L, for anastomotic leak 143 mg/L—distinguish when citing | Cite only day-4 results (103 complications among 314 with CRP) and authors’ conclusions as a design contrast |
| Hu 2023[4] | Body says level of evidence unreported in 3 papers, yet Table 1 gives Carrozzo 2022 level III; I² (0%) appears only in Figure 2 forest plot, not in body; included studies’ infection definitions inconsistent | Adopt pooled incidence and OR; note mostly retrospective historical controls |
| Lewis 2025[19] | Total cases Abstract 24 studies 307 cases, body 23 studies 269 cases, Table 3 sums to 307; ROM and revision-subgroup P values disagree between body and tables; Abstract KT-1000 “3.30 ± 134” missing decimal | Adopt body’s IKDC, KT-1000, and pathogen associations |
| Larsson 1992[26] | Arthroplasty anesthesia counts written neuraxial 134 (86%), general 21 (14%), sum 155, disagreeing with three arthroplasty groups totaling 157; Abstract says CRP “usually normal within 21 days,” Results “normal or near normal” by days 14–21 | Adopt each technique’s peak, timing, normalization proportions, and authors’ conclusions |
| Tripon 2026[14] | Tegner comparison P Abstract 0.18, Results and Table 5 0.19; Table 2 graft type hamstring 16 (57%), BPTB 12 (43%) sum 28, disagreeing with total 34; body writes 16/34, 12/34 yet gives the same percentages; infection-group follow-up range Table 2 2–11 years, Results 1–11 years | Adopt both groups’ outcome scores, return-to-sport time, and irrigation and aspiration counts |
Data Scope
Material is current to 18 September 2026 and includes literature that directly studies blood or synovial fluid markers after ACL reconstruction, together with adjacent material from arthroplasty, spine, fracture and colorectal surgery, native-joint infection, and healthy populations. The latter class is used only to illustrate biologic patterns and methodology; its numerical values are not standards for ACL reconstruction. Of the 80 cited items, CLSI EP28-A3c was not obtained in full text and is used only to state the sample-size requirement for reference intervals.
References
- [1]
Zhao D, Liang GH, Pan JK, Zeng LF, Luo MH, Huang HT, et al. Risk factors for postoperative surgical site infections after anterior cruciate ligament reconstruction: a systematic review and meta-analysis. Br J Sports Med. 2023;57(2):118-128. DOI: 10.1136/bjsports-2022-105448. PMID: 36517215 .
- [2]
Zhang L, Yang R, Mao Y, Fu W. A Systematic Review and Meta-analysis of Risk Factors for an Infection After Anterior Cruciate Ligament Reconstruction. Orthop J Sports Med. 2023;11(10):23259671231200822. DOI: 10.1177/23259671231200822. PMID: 37846316 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [3]
Kottek T, Kunovac B, Vrgoč G, Ivković A, Janković S. Vancomycin-soaking of the graft significantly reduces the incidence of septic arthritis following anterior cruciate ligament reconstruction: comparison of cohorts from the Single-Centre database. Int Orthop. 2025;49(9):2097-2102. DOI: 10.1007/s00264-025-06594-1. PMID: 40586927 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [4]
Hu M, Zhang Y, Shang G, Guo J, Xu H, Ma X, et al. Vancomycin presoak reduces infection in anterior cruciate ligament reconstruction: a systematic review and meta-analysis. BMC Musculoskelet Disord. 2023;24(1):267. DOI: 10.1186/s12891-023-06331-y. PMID: 37020216 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [5]
Wang C, Ao Y, Wang J, Hu Y, Cui G, Yu J. Septic arthritis after arthroscopic anterior cruciate ligament reconstruction: a retrospective analysis of incidence, presentation, treatment, and cause. Arthroscopy. 2009;25(3):243-9. DOI: 10.1016/j.arthro.2008.10.002. PMID: 19245985 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [6]
Schuster P, Schulz M, Immendoerfer M, Mayer P, Schlumberger M, Richter J. Septic Arthritis After Arthroscopic Anterior Cruciate Ligament Reconstruction: Evaluation of an Arthroscopic Graft-Retaining Treatment Protocol. Am J Sports Med. 2015;43(12):3005-12. DOI: 10.1177/0363546515603054. PMID: 26403209 .
- [7]
Kunovac B, Ivković A, Vrgoc G, Janković S. Knee infection following anterior cruciate ligament reconstruction: a cohort study of one thousand, eight hundred and ninety one patients from the single-centre database. Int Orthop. 2020;44(5):869-875. DOI: 10.1007/s00264-020-04500-5. PMID: 32114657 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [8]
El-Kady RAE, ElGuindy AMF. Septic Arthritis Complicating Arthroscopic Anterior Cruciate Ligament Reconstruction: An Experience from a Tertiary-Care Hospital. Infect Drug Resist. 2022;15:3779-3789. DOI: 10.2147/IDR.S369240. PMID: 35859912 .
- [9]
Judd D, Bottoni C, Kim D, Burke M, Hooker S. Infections following arthroscopic anterior cruciate ligament reconstruction. Arthroscopy. 2006;22(4):375-84. DOI: 10.1016/j.arthro.2005.12.002. PMID: 16581449 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [10]
Schollin-Borg M, Michaëlsson K, Rahme H. Presentation, outcome, and cause of septic arthritis after anterior cruciate ligament reconstruction: a case control study. Arthroscopy. 2003;19(9):941-7. DOI: 10.1016/j.arthro.2003.09.004. PMID: 14608312 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [11]
Bohu Y, Klouche S, Herman S, de Pamphilis O, Gerometta A, Lefevre N. Professional Athletes Are Not at a Higher Risk of Infections After Anterior Cruciate Ligament Reconstruction: Incidence of Septic Arthritis, Additional Costs, and Clinical Outcomes From the French Prospective Anterior Cruciate Ligament Study (FAST) Cohort. Am J Sports Med. 2019;47(1):104-111. DOI: 10.1177/0363546518810527. PMID: 30481480 .
- [12]
Chona D, Kay J, Tavabi N, Quinn M, Kramer D, Yen YM, et al. Septic Arthritis After Anterior Cruciate Ligament Reconstruction: A Comparative Study of Pediatric, Adolescent, and Young Adult Populations. Orthop J Sports Med. 2025;13(6):23259671251334591. DOI: 10.1177/23259671251334591. PMID: 40496597 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [13]
Meza BC, Marom N, Greditzer H, Bogner E, HSS ACL Registry, Marx RG. Findings of magnetic resonance imaging in the knee with postreconstruction infection of the anterior cruciate ligament: A descriptive and reliability study. J ISAKOS. 2025;11:100382. DOI: 10.1016/j.jisako.2024.100382. PMID: 39756796 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [14]
Tripon M, Bougard M, Gras A, Mergenthaler G, Maroteau G, Hulet C. Functional results of septic arthritis after anterior cruciate ligament reconstruction treated by arthroscopic lavage with preservation of the graft: Case-control study at 7 years follow-up. Orthop Traumatol Surg Res. 2026;112(2):104455. DOI: 10.1016/j.otsr.2025.104455. PMID: 41072725 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [15]
Wang C, Lee YH, Siebold R. Recommendations for the management of septic arthritis after ACL reconstruction. Knee Surg Sports Traumatol Arthrosc. 2014;22(9):2136-44. DOI: 10.1007/s00167-013-2648-z. PMID: 24061716 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [16]
Ascione T, Balato G, Mariconda M, Rosa D, Rizzo M, Pagliano P. Post-arthroscopic septic arthritis of the knee. Analysis of the outcome after treatment in a case series and systematic literature review. Eur Rev Med Pharmacol Sci. 2019;23(2 Suppl):76-85. DOI: 10.26355/eurrev_201904_17477. PMID: 30977874 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [17]
Lin AP, Nguyen BTT, Tran SQ, Kuo YJ, Huang SW, Chen YP. Does septic arthritis after anterior cruciate ligament reconstruction lead to poor outcomes? A systematic review and meta-analysis of observational studies. Knee Surg Relat Res. 2024;36(1):45. DOI: 10.1186/s43019-024-00248-z. PMID: 39639400 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [18]
Mustamsir E, Aulia Pandu A, Fernando A. The effectiveness of arthroscopic irrigation and debridement in the management of septic arthritis following anterior cruciate ligament reconstruction: A Systematic Review and Meta-Analysis. Acta Orthop Traumatol Turc. 2025;59(1):7-17. DOI: 10.5152/j.aott.2025.24070. PMID: 40337953 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [19]
Lewis DC, McNamara NE, Tabish EM, Featherall JT, Rawson HW, Micicoi G, et al. Patients treated for infection following ACL reconstruction with graft removal have poorer outcomes than those treated with graft retention: A systematic review. J Exp Orthop. 2025;12(1):e70147. DOI: 10.1002/jeo2.70147. PMID: 40059961 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [20]
Mishra P, Lal A, Mohindra M, Mehta N, Joshi D, Chaudhary D. Incidence, management and outcome assessment of post operative infection following single bundle and double bundle acl reconstruction. J Clin Orthop Trauma. 2018;9(2):167-171. DOI: 10.1016/j.jcot.2017.09.017. PMID: 29896022 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [21]
Pepys MB, Hirschfield GM. C-reactive protein: a critical update. J Clin Invest. 2003;111(12):1805-12. DOI: 10.1172/JCI18921. PMID: 12813013 .
- [22]
Vigushin DM, Pepys MB, Hawkins PN. Metabolic and scintigraphic studies of radioiodinated human C-reactive protein in health and disease. J Clin Invest. 1993;91(4):1351-7. DOI: 10.1172/JCI116336. PMID: 8473487 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [23]
Visser M, Bouter LM, McQuillan GM, Wener MH, Harris TB. Elevated C-reactive protein levels in overweight and obese adults. JAMA. 1999;282(22):2131-5. DOI: 10.1001/jama.282.22.2131. PMID: 10591334 .
- [24]
Neumaier M, Metak G, Scherer MA. C-reactive protein as a parameter of surgical trauma: CRP response after different types of surgery in 349 hip fractures. Acta Orthop. 2006;77(5):788-90. DOI: 10.1080/17453670610013006. PMID: 17068712 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [25]
Neumaier M, Scherer MA. C-reactive protein levels for early detection of postoperative infection after fracture surgery in 787 patients. Acta Orthop. 2008;79(3):428-32. DOI: 10.1080/17453670710015355. PMID: 18622849 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [26]
Larsson S, Thelander U, Friberg S. C-reactive protein (CRP) levels after elective orthopedic surgery. Clin Orthop Relat Res. 1992;(275):237-42. PMID: 1735220 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [27]
Park KK, Kim TK, Chang CB, Yoon SW, Park KU. Normative Temporal Values of CRP and ESR in Unilateral and Staged Bilateral TKA. Clin Orthop Relat Res. 2008;466(1):179-88. DOI: 10.1007/s11999-007-0001-x. PMID: 18196391 .
- [28]
Bilgen O, Atici T, Durak K, Karaeminoğullari, Bilgen MS. C-reactive protein values and erythrocyte sedimentation rates after total hip and total knee arthroplasty. J Int Med Res. 2001;29(1):7-12. DOI: 10.1177/147323000102900102. PMID: 11277348 .
- [29]
Windisch C, Brodt S, Roehner E, Matziolis G. The C-reactive protein level after total knee arthroplasty is gender specific. Knee Surg Sports Traumatol Arthrosc. 2016;24(10):3163-3167. DOI: 10.1007/s00167-016-4289-5. PMID: 27535675 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [30]
Windisch C, Brodt S, Roehner E, Matziolis G. C-reactive protein course during the first 5 days after total knee arthroplasty cannot predict early prosthetic joint infection. Arch Orthop Trauma Surg. 2017;137(8):1115-1119. DOI: 10.1007/s00402-017-2709-8. PMID: 28488017 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [31]
Margheritini F, Camillieri G, Mancini L, Mariani PP. C-reactive protein and erythrocyte sedimentation rate changes following arthroscopically assisted anterior cruciate ligament reconstruction. Knee Surg Sports Traumatol Arthrosc. 2001;9(6):343-5. DOI: 10.1007/s001670100232. PMID: 11734870 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [32]
Calvisi V, Lupparelli S. C-reactive protein changes in the uncomplicated course of arthroscopic anterior cruciate ligament reconstruction. Int J Immunopathol Pharmacol. 2008;21(3):603-7. DOI: 10.1177/039463200802100313. PMID: 18831927 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [33]
Wang C, Ao Y, Fan X, Wang J, Cui G, Hu Y, et al. C-reactive protein and erythrocyte sedimentation rate changes after arthroscopic anterior cruciate ligament reconstruction: guideline to diagnose and monitor postoperative infection. Arthroscopy. 2014;30(9):1110-5. DOI: 10.1016/j.arthro.2014.03.025. PMID: 24863405 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [34]
Ruiz-Ibán MA, Díaz Heredia J, Martínez Val IC, Alonso Güemes S, Cuéllar Gutiérrez R, Sastre Solsona S. Evolution of C-reactive protein values in the first month after anterior cruciate ligament reconstruction: reference values. Knee Surg Sports Traumatol Arthrosc. 2015;23(3):763-9. DOI: 10.1007/s00167-013-2702-x. PMID: 24114354 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [35]
Uchida R, Ueda Y, Iuchi R, Takao R, Kanamoto T, Tanaka Y, et al. C-reactive Protein Levels After Anterior Cruciate Ligament Reconstruction. Cureus. 2024;16(10):e71473. DOI: 10.7759/cureus.71473. PMID: 39539898 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [36]
Xiang S, Ao Z, Ren Z, Cai G, Hu Z, Hu W, et al. Comparison of a full tibial tunnel or an all-inside technique for anterior cruciate ligament reconstruction: An observational study. Medicine (Baltimore). 2024;103(42):e40135. DOI: 10.1097/MD.0000000000040135. PMID: 39432663 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [37]
Paci JM, Schweizer SK, Wilbur DM, Sutton LG, Werner FW, Scuderi MG, et al. Results of laboratory evaluation of acute knee effusion after anterior cruciate ligament reconstruction: what is found in patients with a noninfected, painful postoperative knee?. Am J Sports Med. 2010;38(11):2267-72. DOI: 10.1177/0363546510374573. PMID: 20699428 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [38]
Balagod R, Tripathy SK, Pradhan SS, Varghese P, Ramasubbu MK, Srinivasan A, et al. Early Intra-articular Hyaluronic Acid Injection After Anterior Cruciate Ligament Reconstruction Provides Short-Term Pain Relief and Improves Early Postoperative Function With No Clinical Benefits at 6 and 12 Months: A Randomized Controlled Trial. Arthroscopy. 2025;41(9):3371-3381. DOI: 10.1016/j.arthro.2025.02.010. PMID: 39983793 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [39]
Honsawek S, Deepaisarnsakul B, Tanavalee A, Sakdinakiattikoon M, Ngarmukos S, Preativatanyou K, et al. Relationship of serum IL-6, C-reactive protein, erythrocyte sedimentation rate, and knee skin temperature after total knee arthroplasty: a prospective study. Int Orthop. 2011;35(1):31-5. DOI: 10.1007/s00264-010-0973-0. PMID: 21203883 . Erratum in: Int Orthop. 2011;35(11):1749. DOI: 10.1007/s00264-011-1327-2. (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [40]
Batty LM, Mackenzie C, Landwehr C, Webster KE, Feller JA. The Role of Biomarkers in Predicting Outcomes of Anterior Cruciate Ligament Reconstruction: A Systematic Review. Orthop J Sports Med. 2024;12(10):23259671241275072. DOI: 10.1177/23259671241275072. PMID: 39380669 .
- [41]
Azboy I, Çatal B, Başarır K, Mutlu M, Bilgen ÖF, Parvizi J. The Natural Course of Serum D-Dimer, C-Reactive Protein, and Erythrocyte Sedimentation Rate Levels After Uneventful Primary Total Joint Arthroplasty. J Arthroplasty. 2021;36(9):3118-3122. DOI: 10.1016/j.arth.2021.04.031. PMID: 34088567 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [42]
Hashimoto Y, Komiya E, Nishino K, Nishida Y, Masuda A, Nakamura H. Postoperative D-dimer levels predict venous thromboembolisms detected with contrast-enhanced computerized tomography in patients undergoing anterior cruciate ligament reconstruction. BMC Musculoskelet Disord. 2023;24(1):95. DOI: 10.1186/s12891-023-06212-4. PMID: 36740690 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [43]
Mendias CL, Lynch EB, Davis ME, Sibilsky Enselman ER, Harning JA, Dewolf PD, et al. Changes in circulating biomarkers of muscle atrophy, inflammation, and cartilage turnover in patients undergoing anterior cruciate ligament reconstruction and rehabilitation. Am J Sports Med. 2013;41(8):1819-26. DOI: 10.1177/0363546513490651. PMID: 23739685 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [44]
Barker T, Leonard SW, Trawick RH, Martins TB, Kjeldsberg CR, Hill HR, et al. Modulation of inflammation by vitamin E and C supplementation prior to anterior cruciate ligament surgery. Free Radic Biol Med. 2009;46(5):599-606. DOI: 10.1016/j.freeradbiomed.2008.11.006. PMID: 19111610 .
- [45]
Zheng P, Che G, Zhang X, You H, Peng J, Huang Y, et al. Anticoagulant and anti-inflammatory effects of fondaparinux sodium during anterior cruciate ligament reconstruction in high risk VTE patients: A randomised controlled, triple blinded, prospective study. Knee Surg Sports Traumatol Arthrosc. 2026;34(9):3257-3268. DOI: 10.1002/ksa.70081. PMID: 41020563 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [46]
Nagashima M, Sasaki R, Tanaka K, Takeshima K. The use of tourniquet is useful in terms of blood loss and soft tissue damage in arthroscopic anterior cruciate ligament reconstruction: a retrospective study. Sci Rep. 2023;13(1):17798. DOI: 10.1038/s41598-023-45159-3. PMID: 37853067 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [47]
Wang C, Yang P, Zhang D, Jeon IH, Yu T, Zhang Y, et al. Effects of Temperature and Volume of Intraoperative Normal Saline Irrigation on Postoperative Pain, Swelling, and Serum Markers of Inflammation in Patients After Elective, Arthroscopic, Single-Bundle Surgical Reconstruction of the Anterior Cruciate Ligament: A Retrospective, Single-Center Study. Med Sci Monit. 2021;27:e930954. DOI: 10.12659/MSM.930954. PMID: 34403406 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [48]
Oestervemb K, Trybulski R, Szczygieł E, Szczygielska-Babiuch A, Kacprzak B, Hagner-Drengowska M. Deep oscillation therapy enhances early rehabilitation after ACL reconstruction: a randomized controlled trial. J Rehabil Med. 2025;57:jrm44416. DOI: 10.2340/jrm.v57.44416. PMID: 41058126 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [49]
Yonetani Y, Kurokawa M, Amano H, Kusano M, Kanamoto T, Tanaka Y, et al. The Wound Dressing Influenced Effectiveness of Cryotherapy After Anterior Cruciate Ligament Reconstruction: Case-Control Study Comparing Gauze Versus Film Dressing. Arthrosc Sports Med Rehabil. 2022;4(3):e965-e968. DOI: 10.1016/j.asmr.2022.01.010. PMID: 35747645 .
- [50]
Orrego L M, Pérez C M, Pérez Y M, Cheyre E J, Mardones P R. [Plasma C reactive protein in elective orthopedic surgery]. Rev Med Chil. 2005;133(11):1341-8. DOI: 10.4067/s0034-98872005001100010. PMID: 16446858 .
- [51]
Chung YG, Won YS, Kwon YJ, Shin HC, Choi CS, Yeom JS. Comparison of Serum CRP and Procalcitonin in Patients after Spine Surgery. J Korean Neurosurg Soc. 2011;49(1):43-8. DOI: 10.3340/jkns.2011.49.1.43. PMID: 21494362 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [52]
Hunziker S, Hügle T, Schuchardt K, Groeschl I, Schuetz P, Mueller B, et al. The value of serum procalcitonin level for differentiation of infectious from noninfectious causes of fever after orthopaedic surgery. J Bone Joint Surg Am. 2010;92(1):138-48. DOI: 10.2106/JBJS.H.01600. PMID: 20048106 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [53]
Inose H, Kobayashi Y, Yuasa M, Hirai T, Yoshii T, Okawa A. Procalcitonin and Neutrophil Lymphocyte Ratio After Spinal Instrumentation Surgery. Spine (Phila Pa 1976). 2019;44(23):E1356-E1361. DOI: 10.1097/BRS.0000000000003157. PMID: 31725684 .
- [54]
Chen T, Chen R, Zhang H, Feng Q, Cai L, Li J. Postoperative laboratory markers as predictors of early spinal surgical site infections: A retrospective cohort study. Chin J Traumatol. 2025;28(6):412-417. DOI: 10.1016/j.cjtee.2025.08.002. PMID: 40962683 .
- [55]
Zheng S, Wang Z, Qin S, Chen JT. Usefulness of inflammatory markers and clinical manifestation for an earlier method to diagnosis surgical site infection after spinal surgery. Int Orthop. 2020;44(11):2211-2219. DOI: 10.1007/s00264-020-04567-0. PMID: 32435956 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [56]
Yu BZ, Fu J, Chai W, Hao LB, Chen JY. Neutrophil to lymphocyte ratio as a predictor for diagnosis of early Periprosthetic joint infection. BMC Musculoskelet Disord. 2020;21(1):706. DOI: 10.1186/s12891-020-03704-5. PMID: 33109144 .
- [57]
Touhey DC, Knapik DM, Tartibi S, Gopinatth V, Boghosian T, Matava MJ, et al. Synovial Fluid Biomarkers Associated With Anterior Cruciate Ligament Injury and Reconstruction: A Systematic Review. Orthop J Sports Med. 2026;14(5):23259671261443910. DOI: 10.1177/23259671261443910. PMID: 42221203 .
- [58]
Khera A, McGuire DK, Murphy SA, Stanek HG, Das SR, Vongpatanasin W, et al. Race and gender differences in C-reactive protein levels. J Am Coll Cardiol. 2005;46(3):464-9. DOI: 10.1016/j.jacc.2005.04.051. PMID: 16053959 .
- [59]
Xu B, Ma J, Huang Q, Huang ZY, Zhang SY, Pei FX. Two doses of low-dose perioperative dexamethasone improve the clinical outcome after total knee arthroplasty: a randomized controlled study. Knee Surg Sports Traumatol Arthrosc. 2018;26(5):1549-1556. DOI: 10.1007/s00167-017-4506-x. PMID: 28474088 .
- [60]
Backes JR, Bentley JC, Politi JR, Chambers BT. Dexamethasone reduces length of hospitalization and improves postoperative pain and nausea after total joint arthroplasty: a prospective, randomized controlled trial. J Arthroplasty. 2013;28(8 Suppl):11-7. DOI: 10.1016/j.arth.2013.05.041. PMID: 23937923 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [61]
Ma Y, Guo J, Lv S, Cheng S, Liang C, Xie C, et al. Standardized treatment of infection after anterior cruciate ligament reconstruction. Sci Rep. 2024;14(1):22332. DOI: 10.1038/s41598-024-65546-8. PMID: 39333207 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [62]
Musso AD, McCormack RG. Infection after ACL reconstruction: what happens when cultures are negative?. Clin J Sport Med. 2005;15(5):381-4. DOI: 10.1097/01.jsm.0000171256.77779.46. PMID: 16162999 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [63]
Kim YM, Joo YB. Clinical Presentation of Staphylococcus epidermidis Septic Arthritis Following Anterior Cruciate Ligament Reconstruction. Knee Surg Relat Res. 2012;24(1):46-51. DOI: 10.5792/ksrr.2012.24.1.46. PMID: 22570852 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [64]
Renz N, Madjarevic T, Ferrari M, Becker R, Ravn C, Vogely C, et al. Recommendations on diagnosis and antimicrobial treatment of infections after anterior cruciate ligament reconstruction (ACL-R) endorsed by ESSKA and EBJIS. J Infect. 2023;86(6):543-551. DOI: 10.1016/j.jinf.2023.03.021. PMID: 37019288 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [65]
Costa GG, Grassi A, Lo Presti M, Cialdella S, Zamparini E, Viale P, et al. White Blood Cell Count Is the Most Reliable Test for the Diagnosis of Septic Arthritis After Anterior Cruciate Ligament Reconstruction: An Observational Study of 38 Patients. Arthroscopy. 2021;37(5):1522-1530.e2. DOI: 10.1016/j.arthro.2020.11.047. PMID: 33278527 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [66]
Paci JM. Editorial Commentary: Synovial Fluid White Blood Cell Is the Gold Standard to Detect Infection After Anterior Cruciate Ligament Reconstruction: Don't Hesitate to Aspirate. Arthroscopy. 2021;37(5):1531-1533. DOI: 10.1016/j.arthro.2020.12.221. PMID: 33896505 .
- [67]
Konstantinou E, Pfeiffer T, Rocca MS, Grandberg C, Dias K, Musahl V. Prevention and Management of Postoperative Infection After Anterior Cruciate Ligament Reconstruction: A Narrative Review. J Clin Med. 2025;14(2):336. DOI: 10.3390/jcm14020336. PMID: 39860341 . (Data problems affect the main conclusions; its cutoffs or inferences should not be applied directly; see Section 8.)
- [68]
Zhang S, Gong J. Therapeutic Strategies and Clinical Outcomes of Infections Following Anterior Cruciate Ligament Reconstruction: A Narrative Review. Orthop Surg. 2026;18(4):619-630. DOI: 10.1111/os.70298. PMID: 41848531 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [69]
Pérez-Prieto D, Totlis T, Madjarevic T, Becker R, Ravn C, Monllau JC, et al. ESSKA and EBJIS recommendations for the management of infections after anterior cruciate ligament reconstruction (ACL-R): prevention, surgical treatment and rehabilitation. Knee Surg Sports Traumatol Arthrosc. 2023;31(10):4204-4212. DOI: 10.1007/s00167-023-07463-3. PMID: 37243789 .
- [70]
Cassano GD, Moretti L, Vicenti G, Buono C, Albano F, Ladogana T, et al. Infection after Anterior Cruciate Ligament Reconstruction: A Narrative Review of the Literature. Healthcare (Basel). 2024;12(9):894. DOI: 10.3390/healthcare12090894. PMID: 38727451 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [71]
Gupta R, Sood M, Malhotra A, Masih GD, Raghav M, Khanna T. Incidence, Risk Factors, and Management of Infection Following Anterior Cruciate Ligament Reconstruction Surgery. Indian J Orthop. 2018;52(4):399-405. DOI: 10.4103/ortho.IJOrtho_379_17. PMID: 30078899 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [72]
Seilern Und Aspang J, Kopriva J, Garry J, Haley R, Collins A, Lamplot JD, et al. Zero incidence of culture-positive septic arthritis and low infection rate following ACLR with all-soft tissue quadriceps tendon autograft: An analysis of 1053 cases. Knee Surg Sports Traumatol Arthrosc. 2025;33(8):2793-2798. DOI: 10.1002/ksa.12540. PMID: 39506607 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [73]
Miltenberg B, O'Mara L, Cautela F, Tjoumakaris FP, Brahmabhatt S. Inflammatory Arthropathy After Anterior Cruciate Ligament Reconstruction: A Case Report. JBJS Case Connect. 2026;16(1):e25.00570. DOI: 10.2106/JBJS.CC.25.00570. PMID: 41642977 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [74]
Wang C, Meng LY, Chen NY, Li D, Wang JQ, Ao YF. [Management algorithm for septic arthritis after anterior cruciate ligament reconstruction]. Beijing Da Xue Xue Bao Yi Xue Ban. 2021;53(5):850-856. DOI: 10.19723/j.issn.1671-167X.2021.05.007. PMID: 34650284 .
- [75]
Lin ZH, Lin CH, Ding YS, Chiang CH. Acute Klebsiella pneumoniae infection of the knee joint following anterior cruciate ligament reconstruction in a young adult: A case report. J Int Med Res. 2026;54(8):3000605261473109. DOI: 10.1177/03000605261473109. PMID: 42545209 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [76]
Warschkow R, Tarantino I, Torzewski M, Näf F, Lange J, Steffen T. Diagnostic accuracy of C-reactive protein and white blood cell counts in the early detection of inflammatory complications after open resection of colorectal cancer: a retrospective study of 1,187 patients. Int J Colorectal Dis. 2011;26(11):1405-13. DOI: 10.1007/s00384-011-1262-0. PMID: 21701807 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [77]
Clinical and Laboratory Standards Institute. Defining, establishing, and verifying reference intervals in the clinical laboratory; approved guideline, third edition. CLSI document EP28-A3c. Wayne, PA: CLSI; 2010.
- [78]
Ozarda Y. Reference intervals: current status, recent developments and future considerations. Biochem Med (Zagreb). 2016;26(1):5-16. DOI: 10.11613/BM.2016.001. PMID: 26981015 .
- [79]
Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med. 1992;11(10):1305-19. DOI: 10.1002/sim.4780111005. PMID: 1518992 . (The original has typographic, unit, or internal inconsistencies; see Section 8.)
- [80]
Wei Y, Pere A, Koenker R, He X. Quantile regression methods for reference growth charts. Stat Med. 2006;25(8):1369-82. DOI: 10.1002/sim.2271. PMID: 16143984 .