➢ Diabetes has negative effects on the outcomes of total joint arthroplasty, including increased numbers of complications, decreased function, early revision, and higher costs.
➢ The prevalence of diabetes is increasing rapidly worldwide; therefore, the orthopaedic surgeon should have an understanding of how diabetes affects surgical outcomes.
Diabetes mellitus (DM) has clinically relevant effects on musculoskeletal tissues and their treatments1. Clinical treatments that are affected include total joint arthroplasty. We reviewed the current orthopaedic literature, searching PubMed and Google Scholar for original articles published between 1994 and 2015 and focusing on diabetes mellitus and arthroplasty. The search terms used were “diabetes,” “postoperative hyperglycemia,” “arthroplasty,” “osteoarthritis,” and “human studies.” All articles identified were written in English and were full-text, peer-reviewed papers. Two recent studies illustrated that diabetes increases the risk of undergoing total joint arthroplasty2,3. In the first study, the Veterans Health Administration (VHA) administrative database was queried during fiscal year 20002. Having diabetes was defined as having two diabetes-related inpatient or outpatient visits or having a prescription for antidiabetic medication. Patients with diabetes were found to be at greater risk for primary total knee arthroplasty (odds ratio [OR] = 2.8, 2.5, and 1.2 for patients who were forty-six to fifty-five, fifty-six to sixty-five, and sixty-six years of age or older, respectively) and revision arthroplasty (OR = 2.9 for patients forty-six to fifty-five years of age and 1.5 for those sixty-six years of age or older). Only the youngest members of the diabetes cohort (forty-six to fifty-five years of age) had a significantly greater risk of having total hip arthroplasty, with an odds ratio of 1.7 for primary total hip arthroplasty and 3.4 for revision total hip arthroplasty. It should be noted that the VHA population in that study was likely mostly male.
However, similar results were found in a different study that had the same percentage of male and female patients with type-2 diabetes3. The hazard ratio for future arthroplasty of either the hip or the knee was about 2 and remained significant in statistical models that controlled for age, sex, and body mass index (BMI). It should be noted that the population in that study was small (<1000) and was from a single location (Bruneck, Italy).
Considering those findings along with projections for future increases in diabetes4, osteoarthritis5, and arthroplasty6, this is an opportune time to review the impact of diabetes on key outcomes of arthroplasty, including mortality, infection, function (or disability), the return to home and/or work status, and the need for revision arthroplasty.
Soohoo et al., in a review of a California-wide registry of discharge data, reported that the odds ratio for mortality following total hip arthroplasty was 1.45 (95% confidence interval [CI] = 1.18 to 1.77) for patients with uncomplicated diabetes and 2.65 (95% CI = 1.67 to 4.22) for those with complicated diabetes (defined as diabetes with end-organ damage)7. In a study of shoulder arthroplasty using the Nationwide Inpatient Sample (NIS), a database of postoperative discharge data from participating hospitals in the United States, the mortality rates (adjusted for covariates) for the year 2011 were higher for patients with diabetes8. In contrast, a study of NIS data from 1988 to 2003 demonstrated that patients with diabetes had no significant increase in postoperative mortality rates following total knee and total hip arthroplasty9. There was also no association between diabetes and mortality following total knee arthroplasty or total hip arthroplasty in studies examining the United States Medicare administrative database10,11. Whether this difference in findings was due to the age difference between the populations is not clear. A study that specifically compared patients who had diabetes and were taking insulin (diabetes type 1 or 2) with those who had diabetes but were not taking insulin demonstrated that the odds ratio for thirty-day mortality after total hip or total knee arthroplasty was 3.74 for patients with diabetes who were taking insulin in comparison with those who did not have diabetes12. That study demonstrated that variations in the negative effects on morbidity due to diabetes may influence perioperative outcomes.
As a key goal of arthroplasty is to improve the function and mobility of joints with severe osteoarthritis, restoration of function and reduction of disability are important outcomes. Poor functional outcomes in patients with diabetes can lead to limitations in the activities of daily living13.
A five-year prospective study of total knee arthroplasty in the practice of a single surgeon demonstrated a significant association between diabetes and stiffness at the time of the one-year follow-up (OR = 2.8)14. In another study, in which patients were followed for as long as seven years after total knee arthroplasty, every year demonstrated significantly poorer average Knee Society function scores for patients with insulin-dependent and non-insulin-dependent diabetes15. Other, smaller studies with four to eight years of follow-up demonstrated similar results, with poorer Knee Society scores for patients with diabetes as compared with the matched control group16,17. The results of studies in which the risk of manipulation with the patient under anesthesia was used as a surrogate for decreased range of motion were mixed, with some demonstrating an effect of diabetes on range of motion18,19 but one demonstrating no risk20.
The most commonly reported diabetes-associated surgical complication in the medical literature today is the increased rate of infection. Because infection is a relatively infrequent event21, studies involving <1000 arthroplasties are not included here.
Queries of the NIS database have shown mixed results with regard to the effects of diabetes on infection. In a study in which patients with uncontrolled diabetes were compared with a control group of patients without diabetes, the adjusted odds ratio for operative (non-urinary tract) infections was 2.31 (95% CI = 1.42 to 3.75)22; in a separate study, however, there was no increased risk of operative infections9. In a population of patients undergoing primary total knee arthroplasty at a single Finnish hospital, diabetes (defined as a preoperative glucose level of >126 mg/dL) was associated with a more-than-fourfold increase in the risk of periprosthetic joint infection within one year of follow-up (adjusted hazard ratio = 4.41, 95% CI = 1.31 to 14.83)23. A similar study from the same group demonstrated that periprosthetic joint infection following either total knee or total hip arthroplasty was about twice as likely for patients in whom diabetes was diagnosed prior to surgery24. However, unlike in the first study, preoperative hyperglycemia (without a diagnosis of diabetes) was not associated with periprosthetic infection but a diagnosis of diabetes was associated with periprosthetic infection (OR = 2.31, 95% CI = 1.12 to 4.72)24. A study of the Kaiser Permanente Registry in the United States demonstrated that having diabetes resulted in an adjusted hazard ratio of 1.28 (95% CI = 1.03 to 1.60) for deep surgical infection following total knee arthroplasty25. Two studies of the United States Medicare database by the same group showed that diabetes was a significant risk for periprosthetic joint infection after total knee arthroplasty (adjusted hazard ratio = 1.19, 95% CI = 1.06 to 1.34)10 but was not a significant risk following total hip arthroplasty11. Other population studies that did not involve adjusted hazard or odds ratio statistical tests also showed an increased risk of deep infection following arthroplasty in patients with diabetes and/or hyperglycemia15,26. In a single-site retrospective study, diabetes was an independent risk factor for surgical site infection, with an adjusted odds ratio of 1.8327.
A specific, important outcome measure following any inpatient procedure is the length of hospital stay. Two studies examined the effect of diabetes on the length of stay following total hip arthroplasty. The American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database (2005 to 2008) was used to identify an increase of 0.282 day (standard error, 0.130 day) from total hip arthroplasty to hospital discharge in patients with diabetes28. The NIS database (2005 to 2007) was used to compare elective (scheduled) and non-elective (not scheduled before admission) revision total hip arthroplasty cases29. For elective revision procedures, the median length of stay was one day longer for patients with complicated diabetes as compared with that for the entire population, whereas for non-elective revision procedures, it was two days longer for patients with complicated diabetes. In another study, the length of stay was found to be longer for patients with insulin-dependent diabetes as compared with those with non-insulin-dependent diabetes12.
A broad but important quality-of-life outcome for arthroplasty is whether the patient is able to return to his or her previous work activity or even whether the patient is being discharged to home rather than to an extended-care or rehabilitation facility. One of the earliest confounder-controlled studies used the NIS database (1988 to 2000) to determine that diabetes (independent of obesity) had a risk ratio of 1.3 (95% CI = 1.27 to 1.32) for non-home discharge status30. A second study involving the NIS database (1988 to 2003) demonstrated a significantly higher rate of non-home discharge following primary and revision total knee arthroplasty and total hip arthroplasty in patients with diabetes9. In addition to potentially decreasing quality of life, non-home discharge increases the costs of care9,22.
Rates of Revision Arthroplasty
The need for revision arthroplasty adds considerable cost and increases the risk of complications. For the present review, we considered studies that were reported to include at least 100 subjects with diabetes (as revision rates can be as low as 2% in the first eighteen months31) and that specifically assessed the arthroplasty revision rate. Some of those studies counted revisions within three to twelve months7,32-34, and others used longer time periods (up to eleven years)15,35,36.
A registry-based study of all orthopaedic practices in Denmark (1996 to 2005) demonstrated an adjusted hazard risk of revision due to deep infection of 1.49 (95% CI = 1.02 to 2.19) within two years after primary total hip arthroplasty in patients with diabetes; the hazard risk was adjusted for the Charlson Comorbidity Index and socioeconomic status but not for BMI36. Another study of total hip arthroplasty revision rates, conducted using the statewide databases of California and New York (1997 to 2005), identified a hazard risk ratio of 1.44 (95% CI = 1.18 to 1.76) for septic revision at five years in patients with diabetes35. A study of total knee arthroplasty in a single practice in the United States (mid-1987 to mid-1999), with an average duration of follow-up of fifty-two months (diabetes group) or fifty-eight months (reference group), demonstrated significantly higher rates of revision due to aseptic loosening in patients with diabetes (3.6% for patients with diabetes, compared with 0.4% for patients without diabetes), but the investigators did not calculate adjusted risk ratios15. Conversely, other studies demonstrated no significant risk of revision due to diabetes within one year after total hip arthroplasty7,34 or total knee arthroplasty32,33; however, these differences in findings are potentially related to the shorter length of follow-up.
Price of Care
Arthroplasty in patients with diabetes can generate higher costs and charges. An examination of records from a single large hospital demonstrated that the hospital’s direct cost for primary total knee arthroplasty was ∼US$1000 to US$2000 higher per patient with diabetes37. A study of the NIS database demonstrated average hospital charges to be US$1000 more per case for patients with controlled diabetes and US$1500 more per case for those with uncontrolled diabetes22. A retrospective study demonstrated significantly higher inflation-adjusted hospital charges for primary total knee arthroplasty (∼US$1200), primary total hip arthroplasty (∼US$1700), and revision total hip arthroplasty (∼US$2600), but not revision total knee arthroplasty, for patients with diabetes9. Another study, which reviewed only revision total hip arthroplasty data (2005 to 2007), also demonstrated increased average charges for patients with diabetes29. Thus, diabetes adds to charges, and even a small increase in the revision rate is clinically relevant as a result of the increased hospital charges associated with the comorbidity of diabetes.
Uncontrolled and Controlled Diabetes
In a study of data from the NIS database (1988 to 2005), the risks of death, infection, and longer hospital stay following total joint arthroplasty (hip and knee, primary and revision) were greater for patients with uncontrolled diabetes than for those with controlled diabetes, regardless of diabetes type (type 1 or type 2). The adjusted OR for death was 3.23 (95% CI = 1.87 to 5.57)22. In a Kaiser Permanente Registry study sampling the years 2001 to 2009, no such hazard of uncontrolled diabetes for negative arthroplasty outcomes was found32. Although both studies used the glycolated hemoglobin (HbA1c) level to define uncontrolled diabetes, the difference may have been due to improvement of preoperative and perioperative clinical care between 1988 and 200938. A careful analysis of data from the VA Surgical Quality Improvement Program (VASQUIP) demonstrated that patients with an HbA1c level of >7.0% (a commonly used threshold for uncontrolled hyperglycemia) had a greater risk of perioperative and thirty-day postoperative complications (adjusted OR = 1.22)39. The investigators further analyzed their data and noted a risk zone of 6.5% to 7.5% HbA1c in which the increased risk for complications is linear, but they did not recommend a specific threshold for delay of surgery39. A retrospective study of 3468 primary and revision hip and knee arthroplasties that were performed over five years at a single institution in the United States demonstrated that HbA1c levels were not predictive of postoperative infection40. A recent prospective study of 462 patients who were managed with primary total knee arthroplasty by the same surgeon with use of a consistent approach demonstrated that a cutoff of either an HbA1c level of >8% or a preoperative fasting blood glucose level of >200 mg/dL would be a risk factor for surgical site infections41. A small but case-controlled retrospective study of hip, knee, and shoulder total joint arthroplasties demonstrated that a preoperative HbA1c level of >6.7% and a postoperative blood glucose level of >200 mg/dL were both associated with a significantly greater risk of wound complications within thirty days after total joint arthroplasty42.
Postoperative Blood Glucose Level
A small prospective randomized controlled study of primary total hip arthroplasty demonstrated increased postoperative blood glucose in association with the use of general anesthesia but not spinal anesthesia43. Blood glucose levels were higher immediately and one hour after surgery compared with preoperative levels; the postoperative levels did not correlate with the duration of surgery, and these effects were seen in subjects both with and without diabetes. However, in a larger prospective observational study of primary total hip and knee arthroplasty, spinal anesthesia was used and postoperative hyperglycemia was observed in 40% of all subjects, regardless of which joint was involved or whether the patient had diabetes 44. Logistic regression analysis identified several factors as being important, including previously diagnosed diabetes, an American Society of Anesthesiologists (ASA) score of II or III, the preoperative HbA1c level, and blood glucose at the time of induction of anesthesia. The authors also found that the effect of the ASA score on postoperative hyperglycemia was greater in patients who had been previously diagnosed with diabetes and in those with a preoperative HbA1c level of >6.5%. The operation-related variables that were tested were not associated with postoperative hyperglycemia.
Increased postoperative glucose is relevant to the outcomes of total joint arthroplasty. In a case-controlled retrospective study, a postoperative blood glucose level of >200 mg/dL was associated with a significantly increased risk of wound complications within thirty days after hip, knee, and shoulder arthroplasty42. In a separate retrospective study, the risk of periprosthetic joint infections (hip and knee) was more than doubled in patients with a postoperative blood glucose level of >200 mg/dL26. In a third study involving post hoc analysis of data collected in a phase-III trial, the risk of venous thromboembolism (VTE) was higher after total hip arthroplasty, but not after total knee arthroplasty, in patients with a postoperative blood glucose level of >7.9 mM (>142.2 mg/dL)45. A pilot study of nine total hip arthroplasties demonstrated that coagulation factors (relevant to VTE) increased after the perioperative increase in glucose46. In a retrospective study of total hip arthroplasty at a single institution, a postoperative blood glucose level of >126 mg/dL (but not previously diagnosed diabetes) was associated with early mortality risk47.
The effects of diabetes on the outcomes of arthroplasty are substantial and may include VTE, higher mortality rates, increased infection rates, delayed and/or increased non-home discharge rates, decreased function, increased revision rates, and higher hospital charges. Patients who are dependent on insulin may have even higher risks. The literature has large variations in terms of how data were managed, whether variables were continuous or categorical, and whether analyses controlled for additional influential factors (e.g., age, additional comorbidities). These variations complicate the interpretation of the current literature. Nonetheless, given the research studies discussed above and the increasing demand for joint replacement procedures, coupled with the increasing rates of diabetes, a preoperative assessment protocol is crucial. An assessment would include preoperative laboratory analysis of average glycemic control through the HbA1c test (even for patients without a previous diagnosis of diabetes) and postoperative blood glucose measurement. Jämsen et al. found that nearly all patients with diabetes who had a preoperative HbA1c level of >6.5% had postoperative hyperglycemia44, which can have clinically important adverse effects, including infection, thrombosis, and death26,42,45,47.
We recommend that HbA1c be used as the metric for glucose control when determining the suitability of total joint arthroplasty because it is routinely available and can be consistently applied. However, various HbA1c threshold levels for patient selection have been proposed, ranging from 6.5% to 8.0%41,44. Harris et al. concluded that choosing a specific threshold may not be easy as the risk of complications rises linearly—but not at an even rate—with HbA1c; as they stated, each choice “has its own balance of access, benefits, and risk.”39
In addition, we recommend that future clinical studies use HbA1c as a continuous variable and also control for confounding factors in data analysis. Such studies could settle the current discrepancies on whether diabetes increases the risks of mortality, infection, and revision and also could better define the effects of controlled and uncontrolled hyperglycemia for total joint arthroplasty. Future studies are also needed to determine the acceptable range of postoperative hyperglycemia for patients with and without additional comorbidities. Finally, we also recommend that the heterogeneity of diabetes pathology be strongly considered when designing research studies.
Increasing safety and positive outcomes for arthroplasty in patients with diabetes is a concern and a responsibility of both the orthopaedic surgeon and the patient. How this responsibility will be met by both parties is unclear at this time48. Certainly patients seeking arthroplasty may find lowering the glucose level challenging49. However, the published data suggest that using this criterion would minimize postoperative infection rates, decrease hospital charges, decrease revision rates, and potentially increase positive outcomes.
Source of Funding: This research was supported in part by the Orthopaedic Research Education Foundation (OREF) Goldberg Osteoarthritis Research Grant. The contents of this review do not represent the views OREF. Drs. Bucknell and King receive standard salary support from the Department of Veterans Affairs. The contents do not represent the views of the U.S. Department of Veterans Affairs of the United States Government.
Investigation performed at the University of Colorado School of Medicine, Aurora, and the Veterans Affairs Eastern Colorado Health Care System, Denver, Colorado
Disclosure: One or more of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of an aspect of this work. In addition, one or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2016 by The Journal of Bone and Joint Surgery, Incorporated