Introduction

According to the National Center for Health Statistics, the prevalence of obesity (BMI ≥ 30 kg/m2) and morbid obesity (BMI ≥ 40 kg/m2) in the adult population in the United States is approximately 36% and 6%, respectively [9, 29] . In addition the prevalence of super obesity, defined as a BMI greater than 50 kg/m2, has increased 120% between 2000 and 2010 in the United States, with 15.5 million adult Americans or 6.6% of the population having an actual BMI greater than 40 kg/m2 [26, 27] . Because obesity may be an independent risk factor for hip osteoarthritis [5, 22], many patients presenting for elective THA will be obese. According to Bourne et al. [4], the Canadian Joint Replacement Registry showed an 8.56-fold increased risk for a patient with class III obesity (BMI > 40 kg/m2) to undergo a total hip replacement. There is controversy regarding whether obesity poses additional complications and costs with THA [7, 20, 28]. Some studies have reported that the risk of postoperative complications after joint arthroplasty in patients who are obese is comparable to the risk in patients who are nonobese as long as their BMI is less than 40 kg/m2 [1, 17]. However, morbid obesity is associated with a higher risk of postoperative complications [6, 21, 24]. In addition, super obesity (BMI ≥ 50 kg/m2) has been reported to pose an even more substantial risk compared with the risk in patients who are less obese (BMI < 50 kg/m2) [8, 15, 23]. However, few peer-reviewed studies have addressed the surgical dilemma regarding whether to perform THA on patients who are morbidly obese or super obese [2, 6, 16, 23].

Previous studies of complications in patients who are super obese were done in single institutions and focused on surgical complications rather than patient complications or costs involved [2, 23, 25]. However, to our knowledge, a nationwide study has not been performed on risks and costs when stratified by the degree of obesity, as measured by BMI. As methods of payment for health care change from fee-for-service to bundled care or episode-of-care models, the potential added cost of care for complications associated with elective surgery in patients who are morbidly or super obese requires further exploration.

We therefore asked: What are (1) the surgical risks and (2) costs associated with complications after THA in patients who are morbidly obese or super obese?

Patients and Methods

Data Sources

For this retrospective study, the Medicare inpatient claims data were used to identify patients who underwent primary THA between October 1, 2010 and December 31, 2014. The Medicare hospital data were accessed in a limited data set format, which represents the traditional fee-for-service claims submitted by hospitals to the Centers for Medicare & Medicaid Services (CMS) for reimbursement and does not include claims from beneficiaries enrolled in health maintenance organizations and other private health insurance plans. Each beneficiary in these limited data set files is identified by a synthetic but unique number, allowing tracking of the patient for any subsequent complications.

THA was identified using the ICD-9-CM procedure code 81.51 (Table 1). Patients had to be 65 years or older at the time of the arthroplasty and residing in the United States. In addition, each beneficiary must have been enrolled in Medicare for at least 1 year before admission for THA. This preoperative period was used to capture diagnoses and procedures performed before the THA for establishing the general health status of the patient at time of the surgery. Additionally, patients undergoing THA who received any knee arthroplasties during the study period were removed. This screening was taken to ensure that any complication, if observed, was not associated with any knee procedure performed on patients with hip disorders.

Table 1 ICD-9-CM codes used to identify postoperative complications
Full size table

Patients’ obesity level can be identified using the ICD-9-CM codes 278.0x or V85.xx (Table 2). The three levels of the 278.0x code correspond to obese unspecified (278.00), overweight (278.02), and morbidly obese (278.01). In a previous validity study, Lau et al. [14] determined that the sensitivity of the 278.xx for capturing obesity status was 67.7% with a positive predictive value of 50%. The more-specific V85.xx code corresponds to specific levels of the BMI. For example, V85.34 indicates someone with a BMI of 34 to 35 kg/m2, and V85.4 indicates someone with a BMI of 40 kg/m2 or greater. Lau et al. [14] determined that the sensitivity and positive predictive value of the V85.x code were 100%. On October 1, 2010, the V85.4 code was expanded to include a fifth digit that specifically captures patients with high BMI levels (eg, V85.43 = BMI of 50 to 60 kg/m2). The nonobese or “normal-weight” comparison group was defined as patients without any BMI-related codes on their list of diagnoses. Diabetes mellitus is strongly associated with obesity [19] and some complications after THA [3]. To control for the confounding effect of diabetes, the analysis was stratified based on the diabetic status of the patient (Table 3). A dose-response behavior was analyzed with increasing levels of BMI from the overweight (BMI of 25 to 29 kg/m2) to the super obese (BMI ≥ 50 kg/m2). The dose-response relationship between obesity and postoperative complications also was investigated with patients grouped in six levels of BMI: 25 to 29 kg/m2 (V85.2x), 30 to 34 kg/m2 (V85.30–V85.34), 35 to 39 kg/m2 (V85.35–V85.39), 40 to 44 kg/m2 (V85.41), 45 to 49 kg/m2 (V85.42), and 50 kg/m2 or greater (V85.43–V85.45) (Fig. 1). Hazard ratios (HR) for each BMI level, relative to the patients with normal weight, were estimated after adjusting for age, sex, race, and other covariates described in the previous analysis. A contrast was constructed to test for a linear dose-response trend in the HR associated with increasing levels of BMI.

Table 2 Patient characteristics
Full size table
Table 3 90-day complications and corresponding hazard ratios
Full size table
Fig. 1
figure 1

The trend analysis of hazard ratios for 90-day complications by BMI level (kg/m2) is presented. MI = myocardial infarction; DVT = deep vein thrombosis.

Full size image

The Medicare Limited Data Set hospital data identified 593,486 records of patients who had primary hip arthroplasty between October 1, 2010 through December 31, 2014. After applying the exclusions outlined in Materials and Methods, 432,841 remained, which included 376,682 patients of normal weight, 9907 who were morbidly obese, and 805 who were super obese. A small fraction of the patients (approximately 0.3%) had bilateral procedures, which were analyzed with unilateral THA records, with no special treatment (Table 2). The group of 45,447 patients not included in the numbers above are those with milder degrees of obesity.

They were neither normal-weight nor morbidly or super obese. Some were just listed as overweight.

Complication Outcomes and Other Evaluations

Twelve complications occurring during the 90 days after THA were examined (Fig. 2). These complications were identified from hospital claims records, suggesting that the complication occurred while the patient was still hospitalized after the THA or was the cause of a subsequent readmission. For complications identified on the same records with the THA, the presence-on-admission code (only available from 2011 and onward) also was checked to ensure that the condition was not present before the surgery. Subsequent admission for rehabilitation after THA was not considered a complication and was not counted under all-cause readmission. Examining the risk of these short-term complications minimizes confounding by other subsequent procedures or changes in the health status of the patient during longer times. In addition to clinical outcomes, the total hospital charges and CMS payments were tracked from the time of the primary procedure through any additional rehospitalization during a 90-day period (hospital charges and CMS payments were adjusted to June 2015 level by the Consumer Price Index for medical services).

Fig. 2
figure 2

The complication-free trends 90 days after THA for the patients who were super obese, morbidly obese, or normal weight are shown. The scale on the third row is adjusted to show the full survival decrease for readmission. MI = myocardial infarction; DVT = deep vein thrombosis.

Full size image

Postoperative complications and risks were evaluated using multivariate Cox models adjusting for the patient’s age, sex, race, resident census region, economic status using state Medicaid buy-in as a proxy, and overall health status as captured by the Charlson Comorbidity Index. Specific morbidity indicators for diabetes, heart failure, pulmonary disease, depression, and acute renal failure were included to capture the effects beyond the overall health status characterized by the Charlson Comorbidity Index. Institutional factors included were hospital ownership (eg, private), bed size, teaching or nonteaching status, and urban or rural location. In addition, two volume measures treated as proxy variables for experience were included, quantifying the approximate annual THAs performed at the facility and performed by the physician for Medicare patients. If a revision was performed during the 90-day period, the search for complications stopped at the time of revision. Except for death, other complications did not have an obvious competing risk relationship. Each complication was evaluated independent of the presence or absence of other complications.

The data and results reported in our study were processed and analyzed using SAS® 9.4 statistical software (SAS® Institute Inc, Cary, NC, USA).

Results

90-day Complications

Patients with morbid obesity were at increased risk for complications compared with patients with normal weight (Table 3). Complications with an HR greater than 2.0 include periprosthetic joint infection (HR, 3.71; 95% CI, 3.20–4.31; p < 0.001), revision (HR, 1.91; 95% CI, 1.69–2.16; p < 0.001), and wound dehiscence (HR, 3.91; 95% CI, 3.14–4.86; p < 0.001). With the numbers available, there were no differences between patients who were morbidly obese or normal weight for death, acute myocardial infarction, pneumonia, and hip dislocation. When stratified by the presence of diabetes at the time of the THA, the association of obesity and the risk of periprosthetic joint infection and wound dehiscence, independent of diabetes, was more pronounced than for other complications. In patients without diabetes, periprosthetic joint infection and wound dehiscence had HRs of 4.44 (95% CI, 3.67–5.36; p < 0.001) and 4.75 (95% CI, 3.65–6.17; p < 0.001) respectively.

Among patients who were morbidly obese, 15% (HR, 1.48; 95% CI, 1.40–1.56; p < 0.001) returned to hospitals within 90 days for additional treatment compared with 10% among the patients with normal weight (Fig. 1). Periprosthetic joint infection was the complication with the second largest difference (HR, 3.71; 95% CI, 3.20–4.31; p < 0.001) between patients who were morbidly obese and those who were normal weight. Other than readmission, acute renal failure (HR, 1.68; 95% CI, 1.56–1.80; p < 0.001), hip revision (HR, 1.91; 95% CI, 1.69–2.16; p < 0.001), and implant failure (HR, 1.48; 95% CI, 1.30–1.99; p < 0.001) were the three most common complications during the 90-day period after THA among patients who were morbidly obese.

The complication risk posed by patients who were super obese was examined by comparing these patients with patients who were normal weight, and with patients with a BMI of 40 to 49 kg/m2 (Table 4). The second comparison (BMI ≥ 50 kg/m2 versus BMI of 40 to 49 kg/m2) helped identify additional risk beyond obesity at the BMI 40- to 49-kg/m2 level. There were only 805 patients who were super obese during the period studied, and the HR estimates were not precise compared with those of the less-obese patient groups with substantially more subjects. Several complications in the super obese patients had elevated HRs compared with the complications in patients of normal weight, including periprosthetic joint infection (HR, 6.48; 95% CI, 4.54–9.25; p < 0.001), readmission (HR, 2.16; 95% CI, 1.84–2.54; p < 0.001), and wound dehiscence (HR, 9.81; 95% CI, 6.31–15.24; p < 0.001) (Table 4). These three complications also showed an increase in risk in patients who were super obese compared with patients with a BMI of 40 to 49 kg/m2 (periprosthetic joint infection: HR, 1.87; 95% CI, 1.28–2.74; p < 0.001); (readmission: HR, 1.53; 95% CI, 1.28–1.82; p < 0.001); (wound dehiscence: HR, 2.85; 95% CI, 1.77–4.59; p < 0.001). Acute renal failure showed a higher HR for patients who were super obese (HR, 1.93; 95% CI, 1.55–2.40; p < 0.001) compared with patients with normal weight, but showed little additional risk more than the BMI 40- to 49-kg/m2 group (HR, 1.17; 95% CI, 0.93–1.47; p < 0.186). The readmission rate was 20% among patients who were super obese compared with 15% among those with a BMI 40 to 49 kg/m2 and only 10% among the patients with normal weight (HR, 2.16; 95% CI, 1.84–2.54; p < 0.001) (Table 4). Super obesity did not seem to be associated with an increased risk of dislocation (HR, 0.63; 95% CI, 0.23–1.68; p < 0.335).

Table 4 Comparison of complications for patients who were super obese with those of normal weight
Full size table

We found a strong dose response relationship between increasing BMI and an increased likelihood of the patient having a periprosthetic joint infection develop, readmission, renal failure, and wound dehiscence, and a weaker relationship between increasing BMI and the risk of death and postoperative pneumonia. Risks increased steadily for periprosthetic joint infections (HR, 3.0; 95% CI, 4.9–6.5; p < 0.048), readmission (HR, 1.3; 95% CI, 1.8–2.2; p < 0.001), and wound dehiscence (HR, 3.2; 95% CI, 4.4–9.8; p < 0.001) corresponding to BMIs of 40,45, and 50+ kg/m2 respectively. The HRs for death increased for a BMI of 40 to 44 kg/m2 and 45 to 49 kg/m2 but then decreased at a BMI of 50+ kg/m2 (Fig. 2). The HR for renal failure increased for BMIs of 35, 40, and 45 kg/m2 but then decreased for a BMI of 50+ kg/m2 (HR, 1.2; 1.5; 2.1; and 1.9; p < 0.001). The HR for pneumonia remained at a level of 1.1 (p < 0.048) for BMI 40 to 49 kg/m2 and increased incrementally to 1.3 beyond a BMI of 50+ kg/m2.

Costs of Care

In general, the costs of care were higher in patients with morbid obesity and super obesity than in the control group (Fig. 3). Controlling for patient and institutional factors, each THA had a median total hospital charge of USD 88,419 (± USD 1313) among patients who were super obese compared with USD 73,827 (± USD 609) for the control group, a difference of USD 14,592. Medicare payment for the patients who were super obese also was higher, but only by USD 3631 (Table 5).

Fig. 3
figure 3

The total hospital charges for patients with normal BMI, BMI of 40 to 49 kg/m2, and BMI ≥ 50 kg/m2 by age group (years) from the initial THA through 90 days postoperative are shown as average charges with 95% CI.

Full size image
Table 5 Total average hospital charges and CMS payment
Full size table

Discussion

Patients with morbid obesity (BMI ≥ 40 kg/m2) and super obesity BMI ≥ 50 kg/m2) increasingly present for total hip replacement. There is disagreement regarding whether these individuals have greater surgical risks and costs for the episode of care [2, 4, 6, 10] and the magnitude of those risks and costs. Further, there is no established threshold for obesity as defined by BMI in defining increased complications and risks. We therefore asked what were (1) the surgical risks and (2) costs associated with complications after THA in patients who were morbidly obese and patients who were super obese.

A limitation of our study is the reliance on the hospital records for identification of subsequent complications. The complications we studied are a subset of all complications that occurred after THA that are serious enough to require hospitalization. Some complications such as revisions, acute myocardial infarction, and pulmonary embolism require hospitalization. It is possible that complications occurred and were treated in an outpatient setting, therefore they would not be included in the inpatient data set. Nonetheless, complications requiring hospitalization can be viewed as more serious in terms of their effect on patients’ health and the financial implications to care for such patients. Second there were only 805 patients who were super obese included in the database during the time of our study, which reduces precision of hazard risk estimates compared with the less obese or normal weight groups, which included substantially more patients. However, our power analysis indicates that our findings are robust. Finally, hospital charges often bear little resemblance to true costs, and often reflect proprietary or institution-based charge master listings. However, as the patient groups were studied contemporaneously, were covered by a single insurance payer (Medicare), and were treated at a wide range of hospitals, we believe the overall trends in costs are accurate.

We found that most complications were more common in patients with morbid obesity than in patients whose BMI was less than 40 kg/m2. Surgically significant complications, including readmission, wound dehiscence, and prosthetic joint infection were markedly increased in the presence of super obesity. McCalden et al. [16] studied 3290 patients with the primary diagnosis of osteoarthritis at a minimum followup of 2 years. They concluded that morbid obesity does not affect the postoperative outcome after THA with the possible exception of a marginally increased rate of infection. This conclusion is distinctly different from our findings. We attribute this difference to the degree of obesity and the limited number of patients who were super obese. Some prior studies on obesity did not report super obesity as a separate category [1, 17], and were based on a limited population (41 and 1617 respectively) or defined morbid obesity as a BMI greater than 35 kg/m2 [5, 18]. Michalka et al. [18] reported on operative time, blood loss, suboptimal implant placement, the surgeons’ perception of operative difficulty, and the six- minute walk test. Davis et al. [6] studied dislocation, revision, duration of surgery, and deep and superficial infections. Both groups concluded total hip replacement “should not be withheld” or that “obese patients gain similar benefit”. These subjective conclusions were offered despite the documented 4.42 times greater dislocation rate in the group with a BMI of 35 kg/m2 or greater. Our data do not support a continuous increase in renal failure as previously reported [11–13]. Our data do correspond to that of Hanly et al. [10] who reported a 12.8%, 30-day readmission rate. Finally, to our knowledge there has not been a previous report of a 20% readmission rate for patients with super obesity undergoing a THA at 90 days, based on a large administrative database.

There is a large financial burden that corresponds to a 20% readmission rate and attending to the prosthetic infections, wound dehiscence, and component failures. A root-cause analysis in improving the complications we have identified may help minimize these readmissions.

Morbid obesity and super obesity should be considered along with other potentially modifiable risk factors, and should be optimized before elective arthroplasty. Overreliance on BMI may be simplistic, and we recommend consideration of age, percentage of lean body mass, serum markers of nutrition, and anthropomorphic measurements. Because we believe that it is not feasible or ethical to deny access to arthroplasty for these patients, we propose that institutions and providers providing care for this population should be compensated for the additional resources required. Use of improved coding systems, such as the ICD-10, might allow researchers, providers, and payers better information regarding the risks and costs of surgery in patients who are obese and super obese.