The Journal of Arthroplasty 30 (2015) 439–443
Contents lists available at ScienceDirect
The Journal of Arthroplasty journal homepage: www.arthroplastyjournal.org
Diabetes Mellitus, Hyperglycemia, Hemoglobin A1C and the Risk of Prosthetic Joint Infections in Total Hip and Knee Arthroplasty Hilal Maradit Kremers, MD a, Laura W. Lewallen, MD b, Tad M. Mabry, MD b, Daniel J. Berry, MD b, Elie F. Berbari, MD c, Douglas R. Osmon, MD c a b c
Department of Health Sciences Research, Mayo Clinic, Rochester, Minnesota Department of Orthopedic Surgery, Mayo Clinic, Rochester, Minnesota Department of Medicine, Division of Infectious Diseases, Mayo Clinic, Rochester, Minnesota
a r t i c l e
i n f o
Article history: Received 5 September 2014 Accepted 7 October 2014 Keywords: total joint arthroplasty infection diabetes mellitus hyperglycemia hemoglobin A1c
a b s t r a c t Diabetes mellitus is an established risk factor for infections but evidence is conflicting to what extent perioperative hyperglycemia, glycemic control and treatment around the time of surgery modify the risk of prosthetic joint infections (PJIs). In a cohort of 20,171 total hip and knee arthroplasty procedures, we observed a significantly higher risk of PJIs among patients with a diagnosis of diabetes mellitus (hazard ratio [HR] 1.55, 95% CI 1.11, 2.16), patients using diabetes medications (HR 1.56, 95% CI 1.08, 2.25) and patients with perioperative hyperglycemia (HR 1.59, 95% CI 1.07, 2.35), but the effects were attenuated after adjusting for body mass index, type of surgery, ASA score and operative time. Although data were limited, there was no association between hemoglobin A1c values and PJIs. © 2014 Elsevier Inc. All rights reserved.
Total hip arthroplasty (THA) and total knee arthroplasty (TKA) are the fastest growing and among the most costly procedures in the United States with over 1.2 million procedures annually [1]. Although considered relatively safe, surgical site infections (SSIs), and in particular, prosthetic joint infections (PJIs), are a rare but devastating complication of THA and TKA [2]. Diabetes mellitus has been considered a well-established risk factor for SSIs, following many surgical procedures [3–5]. Yet, evidence is conflicting as to the extent the disease itself, or perioperative hyperglycemia control, or diabetes mellitus management around the time of surgery modifies the risk of SSIs following THA or TKA [6–9]. Studies in patients undergoing orthopedic procedures suggest that both the presence of diabetes mellitus and perioperative hyperglycemia are associated with an increased risk of SSIs, but evidence is inconsistent with respect to the role of diabetes control on the risk of SSIs. In a nationwide administrative data-based study [10], uncontrolled diabetes mellitus, as defined by the ICD-9 codes, was associated with more than doubling of the risk of SSIs. This finding could not be replicated in two recent studies that relied on hemoglobin A1c (HbA1C) as a measure of glycemic control [11,12]. A larger body of literature in non-orthopedic surgeries suggests that the relationship between hyperglycemia,
glycemic control and SSI outcomes may differ depending on the type of surgery and the presence or absence of a diagnosis of diabetes mellitus [8,9,13]. In some studies, surgical stress related postoperative hyperglycemia was found to be common and an independent risk factor for SSIs, even among patients without a diabetes diagnosis with a dose– response relationship [8,14–17]. The same was also demonstrated in orthopedic trauma patients [18]. Some suggest that tight glycemic control may be more beneficial in patients without diabetes mellitus when compared to those with diabetes. Adding to this confusion, some clinical trials and meta-analyses have failed to provide sufficient evidence for strict perioperative glycemic control for the prevention of SSIs and other adverse outcomes outside the intensive care unit [19–21]. There is a paucity of evidence to guide clinical management of diabetes mellitus and hyperglycemia in patients undergoing THA and TKA. Prevention strategies and perioperative glucose monitoring may vary depending if the target population is patients with diabetes mellitus or any patient with perioperative hyperglycemia. We therefore sought to determine the association between diabetes mellitus, preoperative and postoperative hyperglycemia, glycemic control, diabetes medication use and the risk of PJI in a large contemporary cohort of patients undergoing THA and TKA. Material and Methods
This work was presented in abstract format at the 2014 annual meeting of the American Academy of Orthopedic Surgeons. The Conflict of Interest statement associated with this article can be found at http:// dx.doi.org/10.1016/j.arth.2014.10.009. Reprint requests: Hilal Maradit Kremers, MD, MSc, Mayo Clinic, 200 First Street SW, Harwick 6-69, Rochester, MN, 55905. http://dx.doi.org/10.1016/j.arth.2014.10.009 0883-5403/© 2014 Elsevier Inc. All rights reserved.
The study population comprised 20,171 THA and TKA procedures (both primary and revision) performed at the Mayo Clinic, Rochester, Minnesota campus between 1/1/2002 and 12/31/2009. Hip hemiarthroplasty, uni/bicompartmental knee arthroplasty procedures,
440
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443
revision procedures for PJIs and aseptic revisions with unexpected positive intraoperative cultures were excluded from the data set. We also excluded patients who had denied authorization to participate in research. Data Collection The Institutional Total Joint Registry, electronic health records and administrative records were used for data collection. Surgery related clinical data were obtained from the Total Joint Registry and included surgery type, surgical indications, anesthesia type, operative time and American Society of Anesthesiologists (ASA) score. Presence of diabetes mellitus was defined using diagnoses recorded in administrative database and institutional medical index. Diagnoses were not validated through manual review of the electronic health records. Electronic health records were used to capture height, weight, body mass index (BMI) measurements and all blood glucose and HbA1c laboratory values within ± 1 week of surgery. A total of 116,898 blood glucose values (85% point-of-care testing) were available on 9129 patients (68% of the 9129 patients with at least one blood glucose value did not have a diagnosis of diabetes mellitus). A total of 2605 HbA1C values were available for 1513 patients (21% of the 1513 patients with HbA1C values did not have a diagnosis of diabetes mellitus). In patients with multiple preoperative and postoperative glucose values, we calculated the median preoperative and postoperative glucose values and used the median values in our analyses. Glucose and HbA1c values were evaluated on a continuous scale and also dichotomized using a glucose cut-off value N180 mg/dL (hyperglycemia) and an HbA1c cut-off value N 7% (as a measure of glucose control status). Natural language processing tools were used to retrieve all current medications from the electronic health records [22]. Medications were coded using the Rxnorm codes and were further classified using the National Drug File—Reference Terminology categories. For the purposes of this study, we included all medications under drug class code HS500—Blood Glucose Regulating Agents. Clinical Endpoints PJI events were limited to events that occurred during the one year time window following surgery. Medical records of all potential cases of PJIs were reviewed manually to validate diagnosis and to classify PJIs using the Infectious Diseases Society of America (IDSA) criteria [23] which included (a) presence of a sinus tract that communicates with the prosthesis, (b) presence of acute inflammation on histopathologic examination, (c) presence of purulence without another known etiology surrounding the prosthesis, and (d) two or more intraoperative cultures or a combination of preoperative aspiration and intraoperative cultures that yields the same organism. Statistical Analysis Among subjects who underwent a THA or TKA, a priori selected measures of diabetes (diabetes diagnosis, preoperative and postoperative glucose levels, hyperglycemia, HbA1c levels and diabetes treatments) were each evaluated for an association with the outcome of PJIs using Cox proportional hazards regression. A robust sandwich covariance estimator was used to correct the working independence model for within-subject correlation in those who had multiple joint arthroplasty surgeries. To control for potential confounding, multiple variable Cox proportional hazards models were fit to assess the effect of each diabetes measure on PJIs while adjusting for: a) age and gender; b) age, gender, BMI, type of surgery (THA versus TKA, primary versus revision), ASA score and operative time; and c) the previous covariates plus diagnosis of diabetes. Adjusted hazard ratios (HRs) from these models represent ratios of instantaneous PJI rates (e.g., between those with and without diabetes) over 1-year follow-up, holding the adjusting covariates in the model fixed.
Each diabetes measure had varying degrees of missing data, and thus the regression analyses were limited to respective patients with observed values. For laboratory data, only the hemoglobin A1C measurement closest to surgery date (if any available) was used, while any number of glucose measurements per patient were summarized based on the median value during the one week (and one day) before and/or after surgery. Also, due to the skewed distributions of glucose and hemoglobin A1C values, analyses were performed on log transformed data for each. Diagnostics of multicollinearity, such as variance inflation factor (VIF) and condition index, were inspected to ensure that including diabetes diagnosis as an adjusting covariate in the model was appropriate for each of the other diabetes-related predictors. Effect modification by diabetes diagnosis was evaluated from its interaction term with continuous log-transformed glucose and hemoglobin A1C values (preoperative, postoperative, perioperative median values). Data were analyzed with the use of SAS statistical software (version 9.3, SAS Institute, Cary NC), at a two-tailed alpha level of 0.05. Results The study included 20,171 (9720 THA and 10,451 TKA) procedures (Table 1). The mean (±standard deviation) age of the entire cohort at the time of surgery was 66.2 (± 12.6) years and 45% of the patients were male. About one-fifth of the procedures were revisions, including 26% of THA and 14% of TKA surgeries. Available for nearly all patients, BMI levels were on average slightly higher in those who underwent TKA (32.1 ± 6.6) versus THA (29.5 ± 6.2). Similarly, a higher proportion of the TKA patients had a diagnosis of diabetes mellitus than the THA patients (20% versus 14%). The mean (±standard deviation) operative time of all procedures was 148.9 (± 68.5) min, and 38% of patients had an ASA score of 3 or 4. A total of 4654 (23%) surgeries had at least one preoperative blood glucose measurement within 1 week of surgery, with a median (interquartile range [IQR]) value of 101 (93–114) mg/dL. Preoperative blood glucose values 1 day before surgery were available for 1379 (7%) surgeries. A total of 5061 (25%) surgeries had at least one postoperative glucose measurement within 1 week following surgery, and the postoperative levels tended to be higher than the preoperative values (median, 136 mg/dL; IQR, 115–162). Postoperative glucose values 1 day after surgery were available for 4399 (22%) surgeries. Combining the preoperative and postoperative glucose measurements, 9129 (45%) patients had at least one blood glucose measurement within ±1 week, of whom 2508 (27%) were classified as having perioperative hyperglycemia based on at least one blood glucose value N180 mg/dL. The proportion of patients with diabetes mellitus and hyperglycemia were examined for a temporal trend over the 8-year study period. The proportion of patients with diabetes mellitus was about 17% and did not change significantly over time, whereas the proportion of patients with hyperglycemia (among those with glucose measurements) ranged from about 26% to 30% in between 2002 and 2008 before dropping to 21% in 2009, which coincided with an increase in glucose testing in that year. A total of 368 SSIs occurred during the 1-year window following surgery, of which 192 met IDSA criteria of PJIs. The 1 year cumulative incidence of PJIs was 0.66%, 1.97%, 0.84% and 1.75% for primary THA, revision THA, primary TKA and revision TKA, respectively. Table 2 shows multivariable adjusted hazard ratios for each variable from Cox models that included: a) age and gender (first column); b) age, gender, BMI, type of surgery (THA versus TKA, primary versus revision), ASA score and operative time (second column); and c) the previous covariates plus diagnosis of diabetes (last column). Adjusted for age and gender, diabetes mellitus was associated with a higher risk of PJIs (HR 1.55; 95% CI 1.11, 2.16), though the effect was attenuated and no longer significant when further adjusting for BMI, type of surgery, ASA scores and operative time (HR 1.23; 95% CI 0.87, 1.74). Similarly, based on 9129 procedures (109 PJI events) with at least one glucose
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443
441
Table 1 Baseline Characteristics of the Study Cohorts.
Age, years, mean (±SD) Male gender Prior joint arthroplasty Body Mass Index (BMI), kg/m2, mean (±SD) Diabetes mellitus diagnosis Operative time (min), mean (±SD) ASA score 1 2 3 4 Preop glucose within 1 wk of surgery (mg/dL)a Median (Interquartile range) Postop glucose within 1 wk of surgery (mg/dL)a Median (Interquartile range) Hyperglycemia ± 1 week of surgery Hemoglobin A1C (%)a Any diabetes medication use Insulin use Oral hyperglycemic medication use
Data Available (n)
Total Hip Arthroplasty (n = 9720)
Total Knee Arthroplasty (n = 10,451)
20,171 20,171 20,171 20,156 20,171 20,171 19,817
64.6 ± 14.1 4613 (47%) 2515 (26%) 29.5 ± 6.2 1407 (14%) 148.7 ± 71.7
67.7 ± 10.7 4564 (44%) 1522 (15%) 32.1 ± 6.6 2100 (20%) 149.1 ± 65.3
66.2 ± 12.6 9177 (45%) 4037 (20%) 30.9 ± 6.6 3507 (17%) 148.9 ± 68.5
421 (4%) 5653 (59%) 3421 (36%) 91 (1%) 101.0 (93.0, 113.0) 133.0 (113.0, 159.0) 1063 (24%) 6.1 (5.6, 6.6) 1112 (11%) 883 (9%) 737 (8%)
190 (2%) 6094 (60%) 3885 (38%) 62 (1%) 101.0 (93.5, 115.0) 138.0 (117.5, 166.0) 1445 (30%) 6.3 (5.8, 6.9) 1531 (15%) 1203 (12%) 1134 (11%)
611 (3%) 11,747 (59%) 7306 (37%) 153 (1%) 101.0 (93.0, 114.0) 136.0 (115.0, 162.0) 2508 (27%) 6.2 (5.7, 6.8) 2643 (13%) 2086 (10%) 1871 (9%)
4654 5061 9129 1513 20,171 20,171 20,171
Total (n = 20,171)
a At least one preop or postop glucose value was available for all of the patients with diabetes mellitus. At least one HbA1c value was available for 34% of the patients with diabetes mellitus.
value ± 1 week of surgery, hyperglycemia (HR 1.59; 95% CI 1.07, 2.35) was associated with a higher risk of PJIs in age and gender-adjusted analyses. The effect of hyperglycemia was slightly attenuated and no longer significant (HR 1.52; 95% CI 0.92, 2.53) when further adjusted for BMI, type of surgery, ASA score, operative time and diabetes diagnosis (third column). Use of any diabetes medication (HR 1.56; 95% CI 1.08, 2.25), insulin (HR 1.59; 95% CI 1.06, 2.37) or oral hypoglycemic medications (HR 1.23; 95% CI 0.87, 1.74) was each significantly associated with a higher risk of PJIs independent of age and gender, but these effects were attenuated when further adjusted for BMI, type of surgery, ASA score and operative time (second column) or diabetes diagnosis (third column). In general, level of blood glucose (log-transformed) was not significantly associated with developing PJIs from the multiple variable-adjusted models. However, we observed a significant interaction between diabetes mellitus diagnosis and 1-day preoperative blood glucose levels, indicating a differential effect of high glucose values on PJI risk between patients with and without a diabetes diagnosis (P value for interaction, 0.026). Elevated 1-day preoperative blood glucose values were associated with a lower risk of PJIs in patients with diabetes mellitus (HR 0.09; 95% CI 0.02, 0.36). In contrast, elevated
1-day preoperative blood glucose values were associated with a higher (non-significant) risk of PJIs in patients without a diagnosis of diabetes mellitus (HR 1.86; 95% CI 0.21, 16.37). A similar analysis assessing an interaction between diabetes diagnosis and hyperglycemia demonstrated that the effect of hyperglycemia on PJIs did not differ significantly in patients with and without diabetes mellitus (P value for interaction, 0.492). Although data were limited to 1513 patients (19 PJI events) with HbA1c values, there was no significant association with developing PJIs for HbA1c, either as a continuous variable (HR 0.05; 95% CI 0.00, 5.24) or dichotomized at a cut-off value of 7% (HR 0.30; 95% CI 0.04, 2.26). Discussion In this large contemporary retrospective cohort of patients with THA and TKA, we demonstrated that the presence of diabetes mellitus and/or hyperglycemia was associated with an increased PJI risk, though some of this effect was confounded by other known risk factors for PJIs such as obesity, ASA score and procedure duration. Although data were limited, poor glycemic control (as assessed by HbA1c levels) was not a good discriminator of PJI risk among patients with diabetes mellitus.
Table 2 Adjusted Hazard Ratios (95% Confidence Intervals) for the Risk of Prosthetic Joint Infections According to Presence of Diabetes Mellitus, Hyperglycemia and Preoperative and Postoperative Glucose Values. Variable
Diabetes Mellitus Hyperglycemia (glucose N 180, ±1 week) Any diabetes medication use Insulin use Oral hypoglycemic use Laboratory values Glucose, 1 day prea Glucose, 1 day posta Glucose, 1 week prea Glucose, 1 week posta Glucose, ± 1 daya Glucose, ± 1 weeka Hemoglobin A1C a Hemoglobin A1C N7% a
Hazard Ratio (95% Confidence Intervals) Age and Gender-Adjusted
Age, Gender, BMI, Type of Surgery, ASA, Operative Time Adjusted
1.55 (1.11, 2.16) 1.59 (1.07, 2.35) 1.56 (1.08, 2.25) 1.59 (1.06, 2.37) 1.83 (1.23, 2.72)
1.23 (0.87, 1.74) 1.31 (0.87, 1.96) 1.15 (0.78, 1.68) 1.19 (0.79, 1.79) 1.38 (0.91, 2.08)
1.23 (0.87, 1.74) 1.52 (0.92, 2.53) 0.97 (0.63, 1.51) 1.05 (0.66, 1.66) 1.31 (0.79, 2.17)
0.71 (0.14, 3.67) 1.18 (0.44, 3.12) 1.38 (0.40, 4.78) 0.88 (0.29, 2.62) 0.76 (0.34, 1.67) 1.56 (0.75, 3.24) 0.05 (0.00, 5.24) 0.30 (0.04, 2.26)
0.31 (0.08, 1.26) 0.96 (0.36, 2.54) 0.63 (0.16, 2.45) 0.66 (0.21, 2.11) 0.57 (0.24, 1.36) 0.99 (0.44, 2.23) 0.04 (0.00, 4.11) 0.28 (0.04, 2.16)
With DM: 0.09 (0.02, 0.36) Without DM: 1.86 (0.21, 16.37) 0.94 (0.30, 2.94) 0.43 (0.09, 2.03) 0.68 (0.18, 2.65) 0.51 (0.18, 1.42) 0.94 (0.35, 2.49) 0.03 (0.00, 3.68) 0.29 (0.04, 2.18)
Analyses were performed with log-transformed values.
Age, Gender, BMI, Type of Surgery, ASA, Operative Time and Diabetes Adjusted
442
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443
Actual preoperative and postoperative glucose values were not significantly associated with the PJI risk, but there was some evidence to suggest that the effect of 1-day preoperative blood glucose values on PJI risk differs among patients with and without a diagnosis of diabetes mellitus. We conclude that, once various clinical risk factors (such as BMI and surgical factors) are taken into account, actual glucose values do not add any additional information for the purpose of PJI risk stratification or prevention. Our findings build upon the current knowledge on the relationship between diabetes mellitus, perioperative glucose levels and SS and PJI outcomes in patients undergoing joint arthroplasty. Studies in surgical cohorts (mostly cardiovascular surgery) as well as arthroplasty cohorts reported significant associations between both preoperative and postoperative glucose levels and SSIs, but findings were inconsistent and evidence appeared less strong for preoperative levels than postoperative levels [4,8,10,13–20,24–39]. It has been suggested that hyperglycemia may be a stronger predictor of SSIs than diabetes mellitus itself. Yet, studies that found significant associations were typically small, underpowered and did not adequately adjust for potential confounders [15,25,29,35]. Similar to our findings, previous studies noted that the association between hyperglycemia and SSI risk was no longer significant when other important risk factors are considered in multivariable analyses [13]. At least 3 studies in joint arthroplasty reported significant association between perioperative glucose values and SSIs or PJIs [24,25,28] but all three studies were small and underpowered. For example, in a cohort of 7181 primary THA and TKA [24], the risk of PJIs increased with increasing preoperative and postoperative glucose levels among patients without diabetes mellitus, but this was based on 32 PJIs among 5805 patients. In the same study, the type of diabetes mellitus medication was not associated with the risk of PJIs. Other measures, such as glycemic variability [40], may be more important than the actual blood glucose values in predicting the risk of SSIs or PJIs in joint arthroplasty. Prospective studies with serial glucose measurements on all patients are needed to answer some of these questions. Regarding glycemic control, our findings are consistent with the majority of previous studies that did not demonstrate any association between HbA1c levels and SSI risk [11,12,14,29,41] but not all [28,42]. Important clinically relevant questions are whether glucose and HbA1c testing should be performed for all joint arthroplasty candidates, irrespective of a prior diagnosis of diabetes mellitus, whether surgery should be delayed for patients with abnormal values, what the target values should be and whether glucose monitoring should be routine during the postoperative period. The American Diabetes Association endorses glucose monitoring in all patients with diabetes mellitus and those receiving therapies associated with hyperglycemia, whereas the Endocrine Society guidelines recommend that all surgical patients be tested for blood glucose and HbA1c levels irrespective of a diabetes mellitus diagnosis [43,44]. This is based on compelling evidence that the prevalence of inpatient hyperglycemia is high and the hospitalization period provides an opportunity to diagnose unrecognized diabetes mellitus. Indeed, it is conceivable that part of the association between elevated glucose levels and SSIs observed in some studies is mediated by unrecognized diabetes mellitus which we were unable to assess in this study. As such, identifying unrecognized diabetes mellitus during the perioperative period could still be a rationale for blood glucose testing on all arthroplasty patients who are typically obese and at a higher risk of diabetes mellitus. Moreover, PJIs are only one of many potential complications of joint arthroplasty surgery. Hyperglycemia can negatively impact other infections (e.g. pneumonia, bacteremia, urinary tract infections), surgical complications and mortality [7,38,45]. Therefore, despite our findings, hyperglycemia may still be a potential target in all patients undergoing THA and TKA. Although a target glucose range depends upon the patient and comorbidities, considering the potential risks associated with glycemic overtreatment [46], the optimal perioperative glucose range for PJI risk is unclear. Unfortunately, results of interventions for strict glucose control are currently
inconclusive and further randomized controlled trials are needed in the arthroplasty population. Our study has several potential limitations. First, it is limited to procedures performed at a single hospital, reducing the ability of generalizing the findings to other institutions. Yet, we believe this is also a strength of our study because confounding by both unknown and known healthcare delivery factors, such as infection prevention practices, is minimized. Second, due to its retrospective design, blood glucose values were not available for all patients. The majority of blood glucose values were point-of-care measurements which are not as accurate as other methods of testing [47]. Only a quarter of the subjects had at least one preoperative and a quarter had one postoperative glucose value, reducing the study power. Furthermore, glucose testing was probably not random, i.e., physicians selectively requested glucose testing for patients with a high likelihood of glucose abnormalities and less frequently for patients with a low likelihood. Also, one abnormal glucose value would likely trigger serial follow-up glucose testing. Although we are unable to determine the extent of this testing bias, it is unlikely to affect our risk estimates because comparisons were made only among individuals with at least one corresponding glucose value. Third, we relied on electronically available administrative data to ascertain diabetes diagnosis. Therefore, we may have missed diabetes diagnoses not recognized around the time of index hospitalization. We also could not account for the severity of diabetes. Otherwise, the percentage of patients with diabetes in our cohort is similar to other large series. We also did not have information on the type of diabetes mellitus, and how blood glucose was monitored and controlled during perioperative period (hospital team, or diabetes service, order set or sliding scale). Despite these limitations, this is possibly one of largest orthopedic studies to date examining the role of hyperglycemia on the risk of PJIs in joint arthroplasty. Prospective studies with glucose values on all patients would be valuable to better document the impact of hyperglycemia and glucose control on PJI risk. Fourth, our analyses with medication use should be interpreted with caution as we could not distinguish ongoing diabetes medications from those used for acute glucose control during hospitalization. Of note, 18% of insulin use in our cohort was among patients without a clinical diagnosis of diabetes mellitus. Finally, confounding by indication limits interpretation of our results with medication use as insulin use is mostly likely a marker of higher risk rather than a risk factor itself. Unique strengths of our study are large sample size and complete ascertainment and chart-review based validation of all PJI outcomes. We were able to ascertain all PJIs during the 1-year time window, irrespective of when and where they occurred. In conclusion, diabetes mellitus, hyperglycemia and antidiabetic medication use are all associated with a higher risk of PJIs in THA and TKA patients but these associations are largely mediated by other risk factors for PJIs, particularly obesity, operative time and ASA scores. No significant association was found with HbA1c levels, suggesting that HbA1c is not a reliable measure for predicting the risk of PJIs in patients undergoing THA and TKA. Accounting for diabetes mellitus sufficiently identifies high risk patients without the need for using glucose, HbA1c values or medication use for risk stratification. Acknowledgments The authors thank Dr. Victor Montori for providing input on an earlier version of this manuscript. References 1. Kurtz SM, Lau E, Ong K, et al. Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030. Clin Orthop Relat Res 2009;467(10):2606. 2. Berbari EF, Osmon DR, Lahr B, et al. The Mayo prosthetic joint infection risk score: implication for surgical site infection reporting and risk stratification. Infect Control Hosp Epidemiol 2012;33(8):774.
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443 3. Chapter 1: mitigation and education. In: Parvizi J, Gehrke T, editors. Proceedings of the International Consensus Meeting on Periprosthetic Joint Infection. Brooklandville MA: Data Trace Publishing Company; 2013. p. 1. 4. Golden SH, Peart-Vigilance C, Kao WHL, et al. Perioperative glycemic control and the risk of infectious complications in a cohort of adults with diabetes. Diabetes Care 1999;22(9):1408. 5. Moghissi ES, Korytkowski MT, DiNardo M, et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Diabetes Care 2009;32(6):1119. 6. Kao LS, Phatak UR. Glycemic control and prevention of surgical site infection. Surg Infect (Larchmt) 2013;14(5):437. 7. Murad MH, Coburn JA, Coto-Yglesias F, et al. Glycemic control in non-critically ill hospitalized patients: a systematic review and meta-analysis. J Clin Endocrinol Metab 2012;97(1):49. 8. Kwon S, Thompson R, Dellinger P, et al. Importance of perioperative glycemic control in general surgery: a report from the Surgical Care and Outcomes Assessment Program. Ann Surg 2013;257(1):8. 9. Abdelmalak BB, Lansang MC. Revisiting tight glycemic control in perioperative and critically ill patients: when one size may not fit all. J Clin Anesth 2013;25(6):499. 10. Marchant Jr MH, Viens NA, Cook C, et al. The impact of glycemic control and diabetes mellitus on perioperative outcomes after total joint arthroplasty. J Bone Joint Surg Am 2009;91(7):1621. 11. Adams AL, Paxton EW, Wang JQ, et al. Surgical outcomes of total knee replacement according to diabetes status and glycemic control, 2001 to 2009. J Bone Joint Surg Am 2013;95(6):481. 12. Iorio R, Williams KM, Marcantonio AJ, et al. Diabetes mellitus, hemoglobin A1C, and the incidence of total joint arthroplasty infection. J Arthroplasty 2012;27(5):726 [e721]. 13. Jeon CY, Furuya EY, Berman MF, et al. The role of pre-operative and post-operative glucose control in surgical-site infections and mortality. Plos One 2012;7(9):e45616. 14. Latham R, Lancaster AD, Covington JF, et al. The association of diabetes and glucose control with surgical-site infections among cardiothoracic surgery patients. Infect Control Hosp Epidemiol 2001;22(10):607. 15. Ata A, Lee J, Bestle SL, et al. Postoperative hyperglycemia and surgical site infection in general surgery patients. Arch Surg 2010;145(9):858. 16. Ramos M, Khalpey Z, Lipsitz S, et al. Relationship of perioperative hyperglycemia and postoperative infections in patients who undergo general and vascular surgery. Ann Surg 2008;248(4):585. 17. Gandhi GY, Nuttall GA, Abel MD, et al. Intraoperative hyperglycemia and perioperative outcomes in cardiac surgery patients. Mayo Clin Proc 2005;80(7):862. 18. Richards JE, Kauffmann RM, Obremskey WT, et al. Stress-induced hyperglycemia as a risk factor for surgical-site infection in nondiabetic orthopedic trauma patients admitted to the intensive care unit. J Orthop Trauma 2013;27(1):16. 19. Gandhi GY, Nuttall GA, Abel MD, et al. Intensive intraoperative insulin therapy versus conventional glucose management during cardiac surgery - A randomized trial. Ann Intern Med 2007;146(4):233. 20. Kao LS, Meeks D, Moyer VA, et al. Peri-operative glycaemic control regimens for preventing surgical site infections in adults. Cochrane Database Syst Rev 2009;3: CD006806. 21. Buchleitner AM, Martinez-Alonso M, Hernandez M, et al. Perioperative glycaemic control for diabetic patients undergoing surgery. Cochrane Database Syst Rev 2012; 9:CD007315. 22. Pathak J, Murphy SP, Willaert BN, et al. Using RxNorm and NDF-RT to classify medication data extracted from electronic health records: experiences from the Rochester Epidemiology Project. AMIA Annu Symp Proc 2011;2011:1089. 23. Osmon DR, Berbari EF, Berendt AR, et al. Diagnosis and management of prosthetic joint infection: clinical practice guidelines by the Infectious Diseases Society of America. Clin Infect Dis 2013;56(1):e1. 24. Jamsen E, Nevalainen P, Eskelinen A, et al. Obesity, diabetes, and preoperative hyperglycemia as predictors of periprosthetic joint infection: a single-center analysis of 7181 primary hip and knee replacements for osteoarthritis. J Bone Joint Surg Am 2012;94(14):e101.
443
25. Jamsen E, Nevalainen P, Kalliovalkama J, et al. Preoperative hyperglycemia predicts infected total knee replacement. Eur J Intern Med 2010;21(3):196. 26. Olsen MA, Nepple JJ, Riew D, et al. Risk factors for surgical site infection following orthopaedic spinal operations. J Bone Joint Surg Am 2008;90A(1):62. 27. Richards JE, Kauffmann RM, Zuckerman SL, et al. Relationship of hyperglycemia and surgical-site infection in orthopaedic surgery. J Bone Joint Surg Am 2012;94(13): 1181. 28. Stryker LS, Abdel MP, Morrey ME, et al. Elevated postoperative blood glucose and preoperative hemoglobin A1C are associated with increased wound complications following total joint arthroplasty. J Bone Joint Surg Am 2013;95(9):808 [S801–802]. 29. King Jr JT, Goulet JL, Perkal MF, et al. Glycemic control and infections in patients with diabetes undergoing noncardiac surgery. Ann Surg 2011;253(1):158. 30. Guvener M, Pasaoglu I, Demircin M, et al. Perioperative hyperglycemia is a strong correlate of postoperative infection in type II diabetic patients after coronary artery bypass grafting. Endocr J 2002;49(5):531. 31. Jackson RS, Amdur RL, White JC, et al. Hyperglycemia is associated with increased risk of morbidity and mortality after colectomy for cancer. J Am Coll Surg 2012;214(1):68. 32. Estrada CA, Young JA, Nifong LW, et al. Outcomes and perioperative hyperglycemia in patients with or without diabetes mellitus undergoing coronary artery bypass grafting. Ann Thorac Surg 2003;75(5):1392. 33. Swenne CL, Lindholm C, Borowiec J, et al. Peri-operative glucose control and development of surgical wound infections in patients undergoing coronary artery bypass graft. J Hosp Infect 2005;61(3):201. 34. Trick WE, Scheckler WE, Tokars JI, et al. Modifiable risk factors associated with deep sternal site infection after coronary artery bypass grafting. J Thorac Cardiovasc Surg 2000;119(1):108. 35. Vriesendorp TM, DeVries JH, Hulscher JB, et al. Early postoperative hyperglycaemia is not a risk factor for infectious complications and prolonged in-hospital stay in patients undergoing oesophagectomy: a retrospective analysis of a prospective trial. Crit Care 2004;8(6):R437. 36. Vriesendorp TM, Morelis QJ, Devries JH, et al. Early post-operative glucose levels are an independent risk factor for infection after peripheral vascular surgery. A retrospective study. Eur J Vasc Endovasc Surg 2004;28(5):520. 37. Zerr KJ, Furnary AP, Grunkemeier GL, et al. Glucose control lowers the risk of wound infection in diabetics after open heart operations. Ann Thorac Surg 1997;63(2):356. 38. Frisch A, Chandra P, Smiley D, et al. Prevalence and clinical outcome of hyperglycemia in the perioperative period in noncardiac surgery. Diabetes Care 2010;33 (8):1783. 39. Sehgal R, Berg A, Figueroa R, et al. Risk factors for surgical site infections after colorectal resection in diabetic patients. J Am Coll Surg 2011;212(1):29. 40. Mendez CE, Mok KT, Ata A, et al. Increased glycemic variability is independently associated with length of stay and mortality in noncritically ill hospitalized patients. Diabetes Care 2013;36(12):4091. 41. Harris AH, Bowe TR, Gupta S, et al. Hemoglobin A1C as a marker for surgical risk in diabetic patients undergoing total joint arthroplasty. J Arthroplasty 2013;28(8 Suppl.):25. 42. Dronge AS, Perkal MF, Kancir S, et al. Long-term glycemic control and postoperative infectious complications. Arch Surg 2006;141(4):375 [discussion 380]. 43. Standards of medical care in diabetes—2011. Diabetes Care 2011;34(Suppl. 1):S11. 44. Umpierrez GE, Hellman R, Korytkowski MT, et al. Management of hyperglycemia in hospitalized patients in non-critical care setting: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2012;97(1):16. 45. Ascione R, Rogers CA, Rajakaruna C, et al. Inadequate blood glucose control is associated with in-hospital mortality and morbidity in diabetic and nondiabetic patients undergoing cardiac surgery. Circulation 2008;118(2):113. 46. Berkowitz SA, Aragon K, Hines J, et al. Do clinical standards for diabetes care address excess risk for hypoglycemia in vulnerable patients? A systematic review. Health Serv Res 2013;48(4):1299. 47. Desachy A, Vuagnat AC, Ghazali AD, et al. Accuracy of bedside glucometry in critically ill patients: influence of clinical characteristics and perfusion index. Mayo Clin Proc 2008;83(4):400.
Contents lists available at ScienceDirect
The Journal of Arthroplasty journal homepage: www.arthroplastyjournal.org
Diabetes Mellitus, Hyperglycemia, Hemoglobin A1C and the Risk of Prosthetic Joint Infections in Total Hip and Knee Arthroplasty Hilal Maradit Kremers, MD a, Laura W. Lewallen, MD b, Tad M. Mabry, MD b, Daniel J. Berry, MD b, Elie F. Berbari, MD c, Douglas R. Osmon, MD c a b c
Department of Health Sciences Research, Mayo Clinic, Rochester, Minnesota Department of Orthopedic Surgery, Mayo Clinic, Rochester, Minnesota Department of Medicine, Division of Infectious Diseases, Mayo Clinic, Rochester, Minnesota
a r t i c l e
i n f o
Article history: Received 5 September 2014 Accepted 7 October 2014 Keywords: total joint arthroplasty infection diabetes mellitus hyperglycemia hemoglobin A1c
a b s t r a c t Diabetes mellitus is an established risk factor for infections but evidence is conflicting to what extent perioperative hyperglycemia, glycemic control and treatment around the time of surgery modify the risk of prosthetic joint infections (PJIs). In a cohort of 20,171 total hip and knee arthroplasty procedures, we observed a significantly higher risk of PJIs among patients with a diagnosis of diabetes mellitus (hazard ratio [HR] 1.55, 95% CI 1.11, 2.16), patients using diabetes medications (HR 1.56, 95% CI 1.08, 2.25) and patients with perioperative hyperglycemia (HR 1.59, 95% CI 1.07, 2.35), but the effects were attenuated after adjusting for body mass index, type of surgery, ASA score and operative time. Although data were limited, there was no association between hemoglobin A1c values and PJIs. © 2014 Elsevier Inc. All rights reserved.
Total hip arthroplasty (THA) and total knee arthroplasty (TKA) are the fastest growing and among the most costly procedures in the United States with over 1.2 million procedures annually [1]. Although considered relatively safe, surgical site infections (SSIs), and in particular, prosthetic joint infections (PJIs), are a rare but devastating complication of THA and TKA [2]. Diabetes mellitus has been considered a well-established risk factor for SSIs, following many surgical procedures [3–5]. Yet, evidence is conflicting as to the extent the disease itself, or perioperative hyperglycemia control, or diabetes mellitus management around the time of surgery modifies the risk of SSIs following THA or TKA [6–9]. Studies in patients undergoing orthopedic procedures suggest that both the presence of diabetes mellitus and perioperative hyperglycemia are associated with an increased risk of SSIs, but evidence is inconsistent with respect to the role of diabetes control on the risk of SSIs. In a nationwide administrative data-based study [10], uncontrolled diabetes mellitus, as defined by the ICD-9 codes, was associated with more than doubling of the risk of SSIs. This finding could not be replicated in two recent studies that relied on hemoglobin A1c (HbA1C) as a measure of glycemic control [11,12]. A larger body of literature in non-orthopedic surgeries suggests that the relationship between hyperglycemia,
glycemic control and SSI outcomes may differ depending on the type of surgery and the presence or absence of a diagnosis of diabetes mellitus [8,9,13]. In some studies, surgical stress related postoperative hyperglycemia was found to be common and an independent risk factor for SSIs, even among patients without a diabetes diagnosis with a dose– response relationship [8,14–17]. The same was also demonstrated in orthopedic trauma patients [18]. Some suggest that tight glycemic control may be more beneficial in patients without diabetes mellitus when compared to those with diabetes. Adding to this confusion, some clinical trials and meta-analyses have failed to provide sufficient evidence for strict perioperative glycemic control for the prevention of SSIs and other adverse outcomes outside the intensive care unit [19–21]. There is a paucity of evidence to guide clinical management of diabetes mellitus and hyperglycemia in patients undergoing THA and TKA. Prevention strategies and perioperative glucose monitoring may vary depending if the target population is patients with diabetes mellitus or any patient with perioperative hyperglycemia. We therefore sought to determine the association between diabetes mellitus, preoperative and postoperative hyperglycemia, glycemic control, diabetes medication use and the risk of PJI in a large contemporary cohort of patients undergoing THA and TKA. Material and Methods
This work was presented in abstract format at the 2014 annual meeting of the American Academy of Orthopedic Surgeons. The Conflict of Interest statement associated with this article can be found at http:// dx.doi.org/10.1016/j.arth.2014.10.009. Reprint requests: Hilal Maradit Kremers, MD, MSc, Mayo Clinic, 200 First Street SW, Harwick 6-69, Rochester, MN, 55905. http://dx.doi.org/10.1016/j.arth.2014.10.009 0883-5403/© 2014 Elsevier Inc. All rights reserved.
The study population comprised 20,171 THA and TKA procedures (both primary and revision) performed at the Mayo Clinic, Rochester, Minnesota campus between 1/1/2002 and 12/31/2009. Hip hemiarthroplasty, uni/bicompartmental knee arthroplasty procedures,
440
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443
revision procedures for PJIs and aseptic revisions with unexpected positive intraoperative cultures were excluded from the data set. We also excluded patients who had denied authorization to participate in research. Data Collection The Institutional Total Joint Registry, electronic health records and administrative records were used for data collection. Surgery related clinical data were obtained from the Total Joint Registry and included surgery type, surgical indications, anesthesia type, operative time and American Society of Anesthesiologists (ASA) score. Presence of diabetes mellitus was defined using diagnoses recorded in administrative database and institutional medical index. Diagnoses were not validated through manual review of the electronic health records. Electronic health records were used to capture height, weight, body mass index (BMI) measurements and all blood glucose and HbA1c laboratory values within ± 1 week of surgery. A total of 116,898 blood glucose values (85% point-of-care testing) were available on 9129 patients (68% of the 9129 patients with at least one blood glucose value did not have a diagnosis of diabetes mellitus). A total of 2605 HbA1C values were available for 1513 patients (21% of the 1513 patients with HbA1C values did not have a diagnosis of diabetes mellitus). In patients with multiple preoperative and postoperative glucose values, we calculated the median preoperative and postoperative glucose values and used the median values in our analyses. Glucose and HbA1c values were evaluated on a continuous scale and also dichotomized using a glucose cut-off value N180 mg/dL (hyperglycemia) and an HbA1c cut-off value N 7% (as a measure of glucose control status). Natural language processing tools were used to retrieve all current medications from the electronic health records [22]. Medications were coded using the Rxnorm codes and were further classified using the National Drug File—Reference Terminology categories. For the purposes of this study, we included all medications under drug class code HS500—Blood Glucose Regulating Agents. Clinical Endpoints PJI events were limited to events that occurred during the one year time window following surgery. Medical records of all potential cases of PJIs were reviewed manually to validate diagnosis and to classify PJIs using the Infectious Diseases Society of America (IDSA) criteria [23] which included (a) presence of a sinus tract that communicates with the prosthesis, (b) presence of acute inflammation on histopathologic examination, (c) presence of purulence without another known etiology surrounding the prosthesis, and (d) two or more intraoperative cultures or a combination of preoperative aspiration and intraoperative cultures that yields the same organism. Statistical Analysis Among subjects who underwent a THA or TKA, a priori selected measures of diabetes (diabetes diagnosis, preoperative and postoperative glucose levels, hyperglycemia, HbA1c levels and diabetes treatments) were each evaluated for an association with the outcome of PJIs using Cox proportional hazards regression. A robust sandwich covariance estimator was used to correct the working independence model for within-subject correlation in those who had multiple joint arthroplasty surgeries. To control for potential confounding, multiple variable Cox proportional hazards models were fit to assess the effect of each diabetes measure on PJIs while adjusting for: a) age and gender; b) age, gender, BMI, type of surgery (THA versus TKA, primary versus revision), ASA score and operative time; and c) the previous covariates plus diagnosis of diabetes. Adjusted hazard ratios (HRs) from these models represent ratios of instantaneous PJI rates (e.g., between those with and without diabetes) over 1-year follow-up, holding the adjusting covariates in the model fixed.
Each diabetes measure had varying degrees of missing data, and thus the regression analyses were limited to respective patients with observed values. For laboratory data, only the hemoglobin A1C measurement closest to surgery date (if any available) was used, while any number of glucose measurements per patient were summarized based on the median value during the one week (and one day) before and/or after surgery. Also, due to the skewed distributions of glucose and hemoglobin A1C values, analyses were performed on log transformed data for each. Diagnostics of multicollinearity, such as variance inflation factor (VIF) and condition index, were inspected to ensure that including diabetes diagnosis as an adjusting covariate in the model was appropriate for each of the other diabetes-related predictors. Effect modification by diabetes diagnosis was evaluated from its interaction term with continuous log-transformed glucose and hemoglobin A1C values (preoperative, postoperative, perioperative median values). Data were analyzed with the use of SAS statistical software (version 9.3, SAS Institute, Cary NC), at a two-tailed alpha level of 0.05. Results The study included 20,171 (9720 THA and 10,451 TKA) procedures (Table 1). The mean (±standard deviation) age of the entire cohort at the time of surgery was 66.2 (± 12.6) years and 45% of the patients were male. About one-fifth of the procedures were revisions, including 26% of THA and 14% of TKA surgeries. Available for nearly all patients, BMI levels were on average slightly higher in those who underwent TKA (32.1 ± 6.6) versus THA (29.5 ± 6.2). Similarly, a higher proportion of the TKA patients had a diagnosis of diabetes mellitus than the THA patients (20% versus 14%). The mean (±standard deviation) operative time of all procedures was 148.9 (± 68.5) min, and 38% of patients had an ASA score of 3 or 4. A total of 4654 (23%) surgeries had at least one preoperative blood glucose measurement within 1 week of surgery, with a median (interquartile range [IQR]) value of 101 (93–114) mg/dL. Preoperative blood glucose values 1 day before surgery were available for 1379 (7%) surgeries. A total of 5061 (25%) surgeries had at least one postoperative glucose measurement within 1 week following surgery, and the postoperative levels tended to be higher than the preoperative values (median, 136 mg/dL; IQR, 115–162). Postoperative glucose values 1 day after surgery were available for 4399 (22%) surgeries. Combining the preoperative and postoperative glucose measurements, 9129 (45%) patients had at least one blood glucose measurement within ±1 week, of whom 2508 (27%) were classified as having perioperative hyperglycemia based on at least one blood glucose value N180 mg/dL. The proportion of patients with diabetes mellitus and hyperglycemia were examined for a temporal trend over the 8-year study period. The proportion of patients with diabetes mellitus was about 17% and did not change significantly over time, whereas the proportion of patients with hyperglycemia (among those with glucose measurements) ranged from about 26% to 30% in between 2002 and 2008 before dropping to 21% in 2009, which coincided with an increase in glucose testing in that year. A total of 368 SSIs occurred during the 1-year window following surgery, of which 192 met IDSA criteria of PJIs. The 1 year cumulative incidence of PJIs was 0.66%, 1.97%, 0.84% and 1.75% for primary THA, revision THA, primary TKA and revision TKA, respectively. Table 2 shows multivariable adjusted hazard ratios for each variable from Cox models that included: a) age and gender (first column); b) age, gender, BMI, type of surgery (THA versus TKA, primary versus revision), ASA score and operative time (second column); and c) the previous covariates plus diagnosis of diabetes (last column). Adjusted for age and gender, diabetes mellitus was associated with a higher risk of PJIs (HR 1.55; 95% CI 1.11, 2.16), though the effect was attenuated and no longer significant when further adjusting for BMI, type of surgery, ASA scores and operative time (HR 1.23; 95% CI 0.87, 1.74). Similarly, based on 9129 procedures (109 PJI events) with at least one glucose
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443
441
Table 1 Baseline Characteristics of the Study Cohorts.
Age, years, mean (±SD) Male gender Prior joint arthroplasty Body Mass Index (BMI), kg/m2, mean (±SD) Diabetes mellitus diagnosis Operative time (min), mean (±SD) ASA score 1 2 3 4 Preop glucose within 1 wk of surgery (mg/dL)a Median (Interquartile range) Postop glucose within 1 wk of surgery (mg/dL)a Median (Interquartile range) Hyperglycemia ± 1 week of surgery Hemoglobin A1C (%)a Any diabetes medication use Insulin use Oral hyperglycemic medication use
Data Available (n)
Total Hip Arthroplasty (n = 9720)
Total Knee Arthroplasty (n = 10,451)
20,171 20,171 20,171 20,156 20,171 20,171 19,817
64.6 ± 14.1 4613 (47%) 2515 (26%) 29.5 ± 6.2 1407 (14%) 148.7 ± 71.7
67.7 ± 10.7 4564 (44%) 1522 (15%) 32.1 ± 6.6 2100 (20%) 149.1 ± 65.3
66.2 ± 12.6 9177 (45%) 4037 (20%) 30.9 ± 6.6 3507 (17%) 148.9 ± 68.5
421 (4%) 5653 (59%) 3421 (36%) 91 (1%) 101.0 (93.0, 113.0) 133.0 (113.0, 159.0) 1063 (24%) 6.1 (5.6, 6.6) 1112 (11%) 883 (9%) 737 (8%)
190 (2%) 6094 (60%) 3885 (38%) 62 (1%) 101.0 (93.5, 115.0) 138.0 (117.5, 166.0) 1445 (30%) 6.3 (5.8, 6.9) 1531 (15%) 1203 (12%) 1134 (11%)
611 (3%) 11,747 (59%) 7306 (37%) 153 (1%) 101.0 (93.0, 114.0) 136.0 (115.0, 162.0) 2508 (27%) 6.2 (5.7, 6.8) 2643 (13%) 2086 (10%) 1871 (9%)
4654 5061 9129 1513 20,171 20,171 20,171
Total (n = 20,171)
a At least one preop or postop glucose value was available for all of the patients with diabetes mellitus. At least one HbA1c value was available for 34% of the patients with diabetes mellitus.
value ± 1 week of surgery, hyperglycemia (HR 1.59; 95% CI 1.07, 2.35) was associated with a higher risk of PJIs in age and gender-adjusted analyses. The effect of hyperglycemia was slightly attenuated and no longer significant (HR 1.52; 95% CI 0.92, 2.53) when further adjusted for BMI, type of surgery, ASA score, operative time and diabetes diagnosis (third column). Use of any diabetes medication (HR 1.56; 95% CI 1.08, 2.25), insulin (HR 1.59; 95% CI 1.06, 2.37) or oral hypoglycemic medications (HR 1.23; 95% CI 0.87, 1.74) was each significantly associated with a higher risk of PJIs independent of age and gender, but these effects were attenuated when further adjusted for BMI, type of surgery, ASA score and operative time (second column) or diabetes diagnosis (third column). In general, level of blood glucose (log-transformed) was not significantly associated with developing PJIs from the multiple variable-adjusted models. However, we observed a significant interaction between diabetes mellitus diagnosis and 1-day preoperative blood glucose levels, indicating a differential effect of high glucose values on PJI risk between patients with and without a diabetes diagnosis (P value for interaction, 0.026). Elevated 1-day preoperative blood glucose values were associated with a lower risk of PJIs in patients with diabetes mellitus (HR 0.09; 95% CI 0.02, 0.36). In contrast, elevated
1-day preoperative blood glucose values were associated with a higher (non-significant) risk of PJIs in patients without a diagnosis of diabetes mellitus (HR 1.86; 95% CI 0.21, 16.37). A similar analysis assessing an interaction between diabetes diagnosis and hyperglycemia demonstrated that the effect of hyperglycemia on PJIs did not differ significantly in patients with and without diabetes mellitus (P value for interaction, 0.492). Although data were limited to 1513 patients (19 PJI events) with HbA1c values, there was no significant association with developing PJIs for HbA1c, either as a continuous variable (HR 0.05; 95% CI 0.00, 5.24) or dichotomized at a cut-off value of 7% (HR 0.30; 95% CI 0.04, 2.26). Discussion In this large contemporary retrospective cohort of patients with THA and TKA, we demonstrated that the presence of diabetes mellitus and/or hyperglycemia was associated with an increased PJI risk, though some of this effect was confounded by other known risk factors for PJIs such as obesity, ASA score and procedure duration. Although data were limited, poor glycemic control (as assessed by HbA1c levels) was not a good discriminator of PJI risk among patients with diabetes mellitus.
Table 2 Adjusted Hazard Ratios (95% Confidence Intervals) for the Risk of Prosthetic Joint Infections According to Presence of Diabetes Mellitus, Hyperglycemia and Preoperative and Postoperative Glucose Values. Variable
Diabetes Mellitus Hyperglycemia (glucose N 180, ±1 week) Any diabetes medication use Insulin use Oral hypoglycemic use Laboratory values Glucose, 1 day prea Glucose, 1 day posta Glucose, 1 week prea Glucose, 1 week posta Glucose, ± 1 daya Glucose, ± 1 weeka Hemoglobin A1C a Hemoglobin A1C N7% a
Hazard Ratio (95% Confidence Intervals) Age and Gender-Adjusted
Age, Gender, BMI, Type of Surgery, ASA, Operative Time Adjusted
1.55 (1.11, 2.16) 1.59 (1.07, 2.35) 1.56 (1.08, 2.25) 1.59 (1.06, 2.37) 1.83 (1.23, 2.72)
1.23 (0.87, 1.74) 1.31 (0.87, 1.96) 1.15 (0.78, 1.68) 1.19 (0.79, 1.79) 1.38 (0.91, 2.08)
1.23 (0.87, 1.74) 1.52 (0.92, 2.53) 0.97 (0.63, 1.51) 1.05 (0.66, 1.66) 1.31 (0.79, 2.17)
0.71 (0.14, 3.67) 1.18 (0.44, 3.12) 1.38 (0.40, 4.78) 0.88 (0.29, 2.62) 0.76 (0.34, 1.67) 1.56 (0.75, 3.24) 0.05 (0.00, 5.24) 0.30 (0.04, 2.26)
0.31 (0.08, 1.26) 0.96 (0.36, 2.54) 0.63 (0.16, 2.45) 0.66 (0.21, 2.11) 0.57 (0.24, 1.36) 0.99 (0.44, 2.23) 0.04 (0.00, 4.11) 0.28 (0.04, 2.16)
With DM: 0.09 (0.02, 0.36) Without DM: 1.86 (0.21, 16.37) 0.94 (0.30, 2.94) 0.43 (0.09, 2.03) 0.68 (0.18, 2.65) 0.51 (0.18, 1.42) 0.94 (0.35, 2.49) 0.03 (0.00, 3.68) 0.29 (0.04, 2.18)
Analyses were performed with log-transformed values.
Age, Gender, BMI, Type of Surgery, ASA, Operative Time and Diabetes Adjusted
442
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443
Actual preoperative and postoperative glucose values were not significantly associated with the PJI risk, but there was some evidence to suggest that the effect of 1-day preoperative blood glucose values on PJI risk differs among patients with and without a diagnosis of diabetes mellitus. We conclude that, once various clinical risk factors (such as BMI and surgical factors) are taken into account, actual glucose values do not add any additional information for the purpose of PJI risk stratification or prevention. Our findings build upon the current knowledge on the relationship between diabetes mellitus, perioperative glucose levels and SS and PJI outcomes in patients undergoing joint arthroplasty. Studies in surgical cohorts (mostly cardiovascular surgery) as well as arthroplasty cohorts reported significant associations between both preoperative and postoperative glucose levels and SSIs, but findings were inconsistent and evidence appeared less strong for preoperative levels than postoperative levels [4,8,10,13–20,24–39]. It has been suggested that hyperglycemia may be a stronger predictor of SSIs than diabetes mellitus itself. Yet, studies that found significant associations were typically small, underpowered and did not adequately adjust for potential confounders [15,25,29,35]. Similar to our findings, previous studies noted that the association between hyperglycemia and SSI risk was no longer significant when other important risk factors are considered in multivariable analyses [13]. At least 3 studies in joint arthroplasty reported significant association between perioperative glucose values and SSIs or PJIs [24,25,28] but all three studies were small and underpowered. For example, in a cohort of 7181 primary THA and TKA [24], the risk of PJIs increased with increasing preoperative and postoperative glucose levels among patients without diabetes mellitus, but this was based on 32 PJIs among 5805 patients. In the same study, the type of diabetes mellitus medication was not associated with the risk of PJIs. Other measures, such as glycemic variability [40], may be more important than the actual blood glucose values in predicting the risk of SSIs or PJIs in joint arthroplasty. Prospective studies with serial glucose measurements on all patients are needed to answer some of these questions. Regarding glycemic control, our findings are consistent with the majority of previous studies that did not demonstrate any association between HbA1c levels and SSI risk [11,12,14,29,41] but not all [28,42]. Important clinically relevant questions are whether glucose and HbA1c testing should be performed for all joint arthroplasty candidates, irrespective of a prior diagnosis of diabetes mellitus, whether surgery should be delayed for patients with abnormal values, what the target values should be and whether glucose monitoring should be routine during the postoperative period. The American Diabetes Association endorses glucose monitoring in all patients with diabetes mellitus and those receiving therapies associated with hyperglycemia, whereas the Endocrine Society guidelines recommend that all surgical patients be tested for blood glucose and HbA1c levels irrespective of a diabetes mellitus diagnosis [43,44]. This is based on compelling evidence that the prevalence of inpatient hyperglycemia is high and the hospitalization period provides an opportunity to diagnose unrecognized diabetes mellitus. Indeed, it is conceivable that part of the association between elevated glucose levels and SSIs observed in some studies is mediated by unrecognized diabetes mellitus which we were unable to assess in this study. As such, identifying unrecognized diabetes mellitus during the perioperative period could still be a rationale for blood glucose testing on all arthroplasty patients who are typically obese and at a higher risk of diabetes mellitus. Moreover, PJIs are only one of many potential complications of joint arthroplasty surgery. Hyperglycemia can negatively impact other infections (e.g. pneumonia, bacteremia, urinary tract infections), surgical complications and mortality [7,38,45]. Therefore, despite our findings, hyperglycemia may still be a potential target in all patients undergoing THA and TKA. Although a target glucose range depends upon the patient and comorbidities, considering the potential risks associated with glycemic overtreatment [46], the optimal perioperative glucose range for PJI risk is unclear. Unfortunately, results of interventions for strict glucose control are currently
inconclusive and further randomized controlled trials are needed in the arthroplasty population. Our study has several potential limitations. First, it is limited to procedures performed at a single hospital, reducing the ability of generalizing the findings to other institutions. Yet, we believe this is also a strength of our study because confounding by both unknown and known healthcare delivery factors, such as infection prevention practices, is minimized. Second, due to its retrospective design, blood glucose values were not available for all patients. The majority of blood glucose values were point-of-care measurements which are not as accurate as other methods of testing [47]. Only a quarter of the subjects had at least one preoperative and a quarter had one postoperative glucose value, reducing the study power. Furthermore, glucose testing was probably not random, i.e., physicians selectively requested glucose testing for patients with a high likelihood of glucose abnormalities and less frequently for patients with a low likelihood. Also, one abnormal glucose value would likely trigger serial follow-up glucose testing. Although we are unable to determine the extent of this testing bias, it is unlikely to affect our risk estimates because comparisons were made only among individuals with at least one corresponding glucose value. Third, we relied on electronically available administrative data to ascertain diabetes diagnosis. Therefore, we may have missed diabetes diagnoses not recognized around the time of index hospitalization. We also could not account for the severity of diabetes. Otherwise, the percentage of patients with diabetes in our cohort is similar to other large series. We also did not have information on the type of diabetes mellitus, and how blood glucose was monitored and controlled during perioperative period (hospital team, or diabetes service, order set or sliding scale). Despite these limitations, this is possibly one of largest orthopedic studies to date examining the role of hyperglycemia on the risk of PJIs in joint arthroplasty. Prospective studies with glucose values on all patients would be valuable to better document the impact of hyperglycemia and glucose control on PJI risk. Fourth, our analyses with medication use should be interpreted with caution as we could not distinguish ongoing diabetes medications from those used for acute glucose control during hospitalization. Of note, 18% of insulin use in our cohort was among patients without a clinical diagnosis of diabetes mellitus. Finally, confounding by indication limits interpretation of our results with medication use as insulin use is mostly likely a marker of higher risk rather than a risk factor itself. Unique strengths of our study are large sample size and complete ascertainment and chart-review based validation of all PJI outcomes. We were able to ascertain all PJIs during the 1-year time window, irrespective of when and where they occurred. In conclusion, diabetes mellitus, hyperglycemia and antidiabetic medication use are all associated with a higher risk of PJIs in THA and TKA patients but these associations are largely mediated by other risk factors for PJIs, particularly obesity, operative time and ASA scores. No significant association was found with HbA1c levels, suggesting that HbA1c is not a reliable measure for predicting the risk of PJIs in patients undergoing THA and TKA. Accounting for diabetes mellitus sufficiently identifies high risk patients without the need for using glucose, HbA1c values or medication use for risk stratification. Acknowledgments The authors thank Dr. Victor Montori for providing input on an earlier version of this manuscript. References 1. Kurtz SM, Lau E, Ong K, et al. Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030. Clin Orthop Relat Res 2009;467(10):2606. 2. Berbari EF, Osmon DR, Lahr B, et al. The Mayo prosthetic joint infection risk score: implication for surgical site infection reporting and risk stratification. Infect Control Hosp Epidemiol 2012;33(8):774.
H. Maradit Kremers et al. / The Journal of Arthroplasty 30 (2015) 439–443 3. Chapter 1: mitigation and education. In: Parvizi J, Gehrke T, editors. Proceedings of the International Consensus Meeting on Periprosthetic Joint Infection. Brooklandville MA: Data Trace Publishing Company; 2013. p. 1. 4. Golden SH, Peart-Vigilance C, Kao WHL, et al. Perioperative glycemic control and the risk of infectious complications in a cohort of adults with diabetes. Diabetes Care 1999;22(9):1408. 5. Moghissi ES, Korytkowski MT, DiNardo M, et al. American Association of Clinical Endocrinologists and American Diabetes Association consensus statement on inpatient glycemic control. Diabetes Care 2009;32(6):1119. 6. Kao LS, Phatak UR. Glycemic control and prevention of surgical site infection. Surg Infect (Larchmt) 2013;14(5):437. 7. Murad MH, Coburn JA, Coto-Yglesias F, et al. Glycemic control in non-critically ill hospitalized patients: a systematic review and meta-analysis. J Clin Endocrinol Metab 2012;97(1):49. 8. Kwon S, Thompson R, Dellinger P, et al. Importance of perioperative glycemic control in general surgery: a report from the Surgical Care and Outcomes Assessment Program. Ann Surg 2013;257(1):8. 9. Abdelmalak BB, Lansang MC. Revisiting tight glycemic control in perioperative and critically ill patients: when one size may not fit all. J Clin Anesth 2013;25(6):499. 10. Marchant Jr MH, Viens NA, Cook C, et al. The impact of glycemic control and diabetes mellitus on perioperative outcomes after total joint arthroplasty. J Bone Joint Surg Am 2009;91(7):1621. 11. Adams AL, Paxton EW, Wang JQ, et al. Surgical outcomes of total knee replacement according to diabetes status and glycemic control, 2001 to 2009. J Bone Joint Surg Am 2013;95(6):481. 12. Iorio R, Williams KM, Marcantonio AJ, et al. Diabetes mellitus, hemoglobin A1C, and the incidence of total joint arthroplasty infection. J Arthroplasty 2012;27(5):726 [e721]. 13. Jeon CY, Furuya EY, Berman MF, et al. The role of pre-operative and post-operative glucose control in surgical-site infections and mortality. Plos One 2012;7(9):e45616. 14. Latham R, Lancaster AD, Covington JF, et al. The association of diabetes and glucose control with surgical-site infections among cardiothoracic surgery patients. Infect Control Hosp Epidemiol 2001;22(10):607. 15. Ata A, Lee J, Bestle SL, et al. Postoperative hyperglycemia and surgical site infection in general surgery patients. Arch Surg 2010;145(9):858. 16. Ramos M, Khalpey Z, Lipsitz S, et al. Relationship of perioperative hyperglycemia and postoperative infections in patients who undergo general and vascular surgery. Ann Surg 2008;248(4):585. 17. Gandhi GY, Nuttall GA, Abel MD, et al. Intraoperative hyperglycemia and perioperative outcomes in cardiac surgery patients. Mayo Clin Proc 2005;80(7):862. 18. Richards JE, Kauffmann RM, Obremskey WT, et al. Stress-induced hyperglycemia as a risk factor for surgical-site infection in nondiabetic orthopedic trauma patients admitted to the intensive care unit. J Orthop Trauma 2013;27(1):16. 19. Gandhi GY, Nuttall GA, Abel MD, et al. Intensive intraoperative insulin therapy versus conventional glucose management during cardiac surgery - A randomized trial. Ann Intern Med 2007;146(4):233. 20. Kao LS, Meeks D, Moyer VA, et al. Peri-operative glycaemic control regimens for preventing surgical site infections in adults. Cochrane Database Syst Rev 2009;3: CD006806. 21. Buchleitner AM, Martinez-Alonso M, Hernandez M, et al. Perioperative glycaemic control for diabetic patients undergoing surgery. Cochrane Database Syst Rev 2012; 9:CD007315. 22. Pathak J, Murphy SP, Willaert BN, et al. Using RxNorm and NDF-RT to classify medication data extracted from electronic health records: experiences from the Rochester Epidemiology Project. AMIA Annu Symp Proc 2011;2011:1089. 23. Osmon DR, Berbari EF, Berendt AR, et al. Diagnosis and management of prosthetic joint infection: clinical practice guidelines by the Infectious Diseases Society of America. Clin Infect Dis 2013;56(1):e1. 24. Jamsen E, Nevalainen P, Eskelinen A, et al. Obesity, diabetes, and preoperative hyperglycemia as predictors of periprosthetic joint infection: a single-center analysis of 7181 primary hip and knee replacements for osteoarthritis. J Bone Joint Surg Am 2012;94(14):e101.
443
25. Jamsen E, Nevalainen P, Kalliovalkama J, et al. Preoperative hyperglycemia predicts infected total knee replacement. Eur J Intern Med 2010;21(3):196. 26. Olsen MA, Nepple JJ, Riew D, et al. Risk factors for surgical site infection following orthopaedic spinal operations. J Bone Joint Surg Am 2008;90A(1):62. 27. Richards JE, Kauffmann RM, Zuckerman SL, et al. Relationship of hyperglycemia and surgical-site infection in orthopaedic surgery. J Bone Joint Surg Am 2012;94(13): 1181. 28. Stryker LS, Abdel MP, Morrey ME, et al. Elevated postoperative blood glucose and preoperative hemoglobin A1C are associated with increased wound complications following total joint arthroplasty. J Bone Joint Surg Am 2013;95(9):808 [S801–802]. 29. King Jr JT, Goulet JL, Perkal MF, et al. Glycemic control and infections in patients with diabetes undergoing noncardiac surgery. Ann Surg 2011;253(1):158. 30. Guvener M, Pasaoglu I, Demircin M, et al. Perioperative hyperglycemia is a strong correlate of postoperative infection in type II diabetic patients after coronary artery bypass grafting. Endocr J 2002;49(5):531. 31. Jackson RS, Amdur RL, White JC, et al. Hyperglycemia is associated with increased risk of morbidity and mortality after colectomy for cancer. J Am Coll Surg 2012;214(1):68. 32. Estrada CA, Young JA, Nifong LW, et al. Outcomes and perioperative hyperglycemia in patients with or without diabetes mellitus undergoing coronary artery bypass grafting. Ann Thorac Surg 2003;75(5):1392. 33. Swenne CL, Lindholm C, Borowiec J, et al. Peri-operative glucose control and development of surgical wound infections in patients undergoing coronary artery bypass graft. J Hosp Infect 2005;61(3):201. 34. Trick WE, Scheckler WE, Tokars JI, et al. Modifiable risk factors associated with deep sternal site infection after coronary artery bypass grafting. J Thorac Cardiovasc Surg 2000;119(1):108. 35. Vriesendorp TM, DeVries JH, Hulscher JB, et al. Early postoperative hyperglycaemia is not a risk factor for infectious complications and prolonged in-hospital stay in patients undergoing oesophagectomy: a retrospective analysis of a prospective trial. Crit Care 2004;8(6):R437. 36. Vriesendorp TM, Morelis QJ, Devries JH, et al. Early post-operative glucose levels are an independent risk factor for infection after peripheral vascular surgery. A retrospective study. Eur J Vasc Endovasc Surg 2004;28(5):520. 37. Zerr KJ, Furnary AP, Grunkemeier GL, et al. Glucose control lowers the risk of wound infection in diabetics after open heart operations. Ann Thorac Surg 1997;63(2):356. 38. Frisch A, Chandra P, Smiley D, et al. Prevalence and clinical outcome of hyperglycemia in the perioperative period in noncardiac surgery. Diabetes Care 2010;33 (8):1783. 39. Sehgal R, Berg A, Figueroa R, et al. Risk factors for surgical site infections after colorectal resection in diabetic patients. J Am Coll Surg 2011;212(1):29. 40. Mendez CE, Mok KT, Ata A, et al. Increased glycemic variability is independently associated with length of stay and mortality in noncritically ill hospitalized patients. Diabetes Care 2013;36(12):4091. 41. Harris AH, Bowe TR, Gupta S, et al. Hemoglobin A1C as a marker for surgical risk in diabetic patients undergoing total joint arthroplasty. J Arthroplasty 2013;28(8 Suppl.):25. 42. Dronge AS, Perkal MF, Kancir S, et al. Long-term glycemic control and postoperative infectious complications. Arch Surg 2006;141(4):375 [discussion 380]. 43. Standards of medical care in diabetes—2011. Diabetes Care 2011;34(Suppl. 1):S11. 44. Umpierrez GE, Hellman R, Korytkowski MT, et al. Management of hyperglycemia in hospitalized patients in non-critical care setting: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2012;97(1):16. 45. Ascione R, Rogers CA, Rajakaruna C, et al. Inadequate blood glucose control is associated with in-hospital mortality and morbidity in diabetic and nondiabetic patients undergoing cardiac surgery. Circulation 2008;118(2):113. 46. Berkowitz SA, Aragon K, Hines J, et al. Do clinical standards for diabetes care address excess risk for hypoglycemia in vulnerable patients? A systematic review. Health Serv Res 2013;48(4):1299. 47. Desachy A, Vuagnat AC, Ghazali AD, et al. Accuracy of bedside glucometry in critically ill patients: influence of clinical characteristics and perfusion index. Mayo Clin Proc 2008;83(4):400.
