Variability of patient and surgical risk factors for infection in a single, urban, academic total joint replacement center

Abstract

Background

We describe surgeon-specific patient and procedure variability in a single center to determine how much variability exists between surgeons.

Methods

Data was analyzed from 2009 to 2013 at a single center. The total number of primary and revision hip and knee arthroplasty surgeries were quantified for each surgeon.

Results

Surgeon caseload varied significantly, with the largest differences observed in primary TKA caseload. The largest patient differences were in regards to percentage of patients with diabetes mellitus amongst primary TKA patients

Conclusion

Significant differences in patient characteristics that could significantly impact outcomes after total joint arthroplasty were found amongst surgeons.

1. Introduction

The US Department of Health & Human Services provides three fundamental criteria that combine to define quality. These are: “getting the right care to the right patient at the right time - every time".¹ The process of medical care involves having the right people and facilities but most importantly doing the right thing at the right time. Outcomes can measure rates of improvement, complications, and quality of life. There is currently increasing attention addressed towards the evaluation and public reporting of the individual practitioner's outcomes as a means of measuring the quality of healthcare they deliver.^10,11 Differences in patient- and case-characteristics and their associated risk are correlated with post-operative morbidity, including infection.12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26 Therefore, it is logical that a surgeon who performs more cases on high-risk patients may expect different results than a surgeon doing more routine cases on healthy patients. Infection rates are one common quality outcome measure used to evaluate total joint replacement. Quality comparisons at the physician level should factor in the potential variability in patient populations between individual surgeons and their risk profile. The purpose of this study is to examine surgeon-specific patient populations and types of joint replacement procedures from a hospital infection control database in a single, urban, academic medical center to determine how much variability in surgeon practices exists.

2. Methods

Data was obtained from a hospital infection control administrative database. A total of 5727 primary total knee arthroplasty (TKA) surgeries and 947 revision TKA procedures were performed by 17 and 4 surgeons, respectively, at a single joint replacement center over a 5-year time-frame (2009–2013). Over the same period, 6039 primary total hip arthroplasty (THA) and 947 revision THA procedures were performed by 15 and 5 surgeons, respectively. Variability between surgeons was assessed in terms of surgeon, patient, and peri-operative factors (Table 1) using descriptive statistics (mean, range, standard deviation, minimum, maximum).

Table 1.

Surgeon & Patient Factors: different colored boxes will indicate whether a variation is attributed to surgeon factors, patient factors, or peri-operative factors.

3. Results

3.1. Knee arthroplasty

3.1.1. Surgeon variability

The mean annual surgeon caseload ranged from 21 to 110 cases (mean ± SD, 64 ± 29) for primary TKA (Table 2) versus 16 to 30.6 cases (21.25 ± 6.8) for revision TKA (Table 3). The mean annual percentage of bilateral TKA procedures ranged from 0 to 21.6% (7 ± 6.5) for primary TKA versus 0.4–3.6% (1.7 ± 1.4) for revision TKA.

Table 2.

Primary Knee Arthroplasty: Variability observed in primary TKA surgery associated with surgeon factors, patient factors, and peri-operative factors.

Table 3.

Revision Knee Arthroplasty: Variability observed in revision TKA surgery associated with surgeon factors, patient factors, and peri-operative factors.

3.1.2. Patient variability

The mean annual age ranged from 59.2 to 70.2 years (64 ± 2.4) for primary TKA versus 59.2–64.6 years (62 ± 2.2) for revision TKA. The mean annual percentage of patients with commercial insurance ranged from 10 to 21.4% (17 ± 3.2) for primary TKA and 21.4–26.8% (25 ± 2.5) for revision TKA. The mean annual percentage of patients with Medicaid ranged from 0 to 2.4% (0.3 ± 0.6) for primary TKA and 0–0.4% (0.1 ± 0.2) for revision TKA. The mean annual percentage of patients with Medicare ranged from 5.2 to 18.4% (11 ± 3.9) for primary TKA and from 17 to 26.4% (20.2 ± 4.3) for revision TKA. The mean annual percentage of patients with diabetes mellitus (DM) ranged from 8 to 21.4% (14 ± 3.6) for primary TKA and from 12.4 to 15.4% (13.75 ± 1.25) for revision TKA. The mean annual percentage of patients with BMI>30 ranged from 6.8 to 17.4% (12 ± 3.6) for primary TKA and from 6.8 to 22.8 (15.5 ± 7) for revision TKA. The mean annual percentage of patients who used tobacco ranged from 3.4 to 11.4% (7 ± 2.4) for primary TKA and from 10.4 to 13.4% (11.7 ± 1.2) for revision TKA. The mean annual percentage of patients with ASA>3 ranged from 21 to 52.2% (31 ± 7.2) for primary TKA and from 38.6 to 51.2% (45 ± 5.2) for revision TKA.

3.1.3. Peri-operative variability

The mean annual average operative time ranged from 74.8 to 188 min (138 ± 32.6) for primary TKA and from 114.6 to 140.8 min (132 ± 11.8) for revision TKA. The mean annual percentage of surgeries in which transfusions of packed red blood cells (pRBC) were used ranged from 0.6 to 19% (10 ± 4.1) for primary TKA and from 10.6 to 14.6% (12.7 ± 1.6) for revision TKA. The mean number of units of pRBCs used (when given) ranged from 0.4 to 2 units (1.3 ± 0.4) for revision TKA.

3.2. Hip arthroplasty

3.2.1. Surgeon variability

The mean annual caseload for each surgeon performing primary THA (Table 4) ranged from 20 to 172.2 cases (mean ± SD, 80 ± 49.4) versus 16.6 to 34.4 cases (23.32 ± 6.75) for revision THA (Table 5).

Table 4.

Primary Hip Arthroplasty: Variability observed in primary THA surgery associated with surgeon factors, patient factors, and peri-operative factors.

Table 5.

Revision Hip Arthroplasty: Variability observed in revision THA surgery associated with surgeon factors, patient factors, and peri-operative factors.

3.2.2. Patient variability

The mean annual age of the patients over the 5-year time frame ranged from 58 to 80 years (63 ± 5.5) for primary THA versus 60.8 to 65.6 (62.76 ± 1.95) for revision THA. The mean annual percentage of patients who had commercial insurance ranged from 9.9% to 28.4% (19.56 ± 4.29) for primary THA and from 26.6% to 30.2% (28.32 ± 1.61) for revision THA. The mean annual percentage of patients who had Medicaid insurance ranged from 0 to 2.8% (0.53 ± 0.89) for primary THA (no revision THA patients had Medicaid insurance). The mean annual percentage of patients who had Medicare insurance ranged from 4% to 22% (10.73 ± 4.63%) for primary THA and from 17.6% to 23.8% (21 ± 2.47) for revision THA. The mean annual percentage of patients with DM ranged from 5.4% to 14% (9.1 ± 3.1%) for primary THA and from 3.2% to 11.2% (7.84 ± 3.42) for revision THA. The mean annual percentage of patients with BMI>30 ranged from 1% to 10.4% (4.88 ± 2.77%) for primary THA and from 2% to 7.2% (4.52 ± 1.96) for revision THA. The mean annual percentage of patients who used tobacco ranged from 5.2% to 16.6% (10.67 ± 3.45) for primary THA and from 6.8% to 16.4% (12.36 ± 3.55) for revision THA. The mean annual percentage of patients with ASA>3 ranged from 18.6% to 65.2% (32 ± 11.87) for primary THA and from 28.2% to 44.4% (40 ± 6.75) for revision THA.

3.2.3. Peri-operative variability

The mean annual operative time ranged from 108.2 to 187.6 min (144.28 ± 25.95) for primary THA and from 166.2 to 218.4 min (186 ± 26.9) for revision THA. The mean annual percentage of surgeries in which pRBCs were used ranged from 6.8% to 27.2% (16.27 ± 6.19) for primary THA and from 21.2% to 29.8% (25.28 ± 3.19) for revision THA. The mean number of units of pRBCs (when given) ranged from 1.2 to 2.8 units (1.93 ± 0.46) for primary THA versus 2–3.2 units (2.52 ± 0.52) for revision THA.

4. Discussion

Improving quality of care related to total joint replacement is an ongoing national effort. More than 330,000 hip and 710,000 knee replacements are performed annually in the US,²⁷ and the demand for joint arthroplasty is expected to increase in the near future.^28,29 Arthroplasty surgery is generally very successful for the majority of patients, with outcomes studies reporting excellent functional and clinical results.^30,31 However, there are complications, which place a large burden not only on the patient and their family, but also on society due to the significant number of these procedures performed each year. This has led to the desire to report quality data so that patient consumers may evaluate institutions and individual providers from whom they seek care.

In this study, we quantified differences in the examined parameters from a hospital infection control database of the patient populations for surgeons in a single academic medical center that could impact current quality reporting outcomes. Several significant differences were noted. The number of cases performed per year per surgeon varied widely across the primary and revision hip and knee arthroplasty categories. Hospital caseload has been shown to affect postoperative function as well as infection rates.33, 34, 35 Geubbels et al. proposes this trend may be secondary to the increased technical skill and experience associated with a higher caseload.³⁴ If generalized from the hospital setting to individual surgeon, it is possible that the surgeon who operates on a lower number or percentage of patients, especially those who are high-risk, may be predisposed to a higher morbidity secondary to less experience.

The patient-specific patient factors, such as diabetes, BMI>30, tobacco usage, and ASA score >3, also differed widely across individual surgeons between each category when examining patient risk factors for infection. Several surgeons operated on two to three times the percent of patients with DM and BMI>30, when compared to other surgeons in the same practice location. For primary hip arthroplasty cases, at least one surgeon operated on 5–10 times the percent of diabetic patients as at least two other surgeons in the groups. Additional inter-surgeon variability was observed in the rates of tobacco usage; for instance, one surgeon's percentage of primary hip arthroplasties performed on current smokers was almost twice as high as another's. DM, elevated BMI, and tobacco usage are well established in the literature as predisposing risk factors for developing periprosthetic joint infections (PJI's) following total joint arthroplasty.^{13,20,21,25,36,37} Therefore, it is logical that surgeons in this study who operated on significantly higher percentages of obese, diabetic patients accrued significantly more risk of infection in their patient populations. In addition, some surgeons operated on significantly more ASA>3 patients than other surgeons. A higher percentage of ASA>3 patients was observed in the revision cases compared to primary arthroplasty cases. Overall, any increased rate of infection may reflect preoperative patient risk factors and types of cases rather than a direct surrogate for the quality of the surgeon's care.

When examining peri-operative variability in this cohort, significant differences were noticed between surgeons in terms of the operative time, the percentage of cases in which pRBC's were transfused, and the number of units of pRBC's transfused per case, when indicated. All of these factors have been shown to affect the risk of periprosthetic joint infection.41, 42, 43, 44, 45, 46, 47 Transfusion of allogenic pRBC's has been shown across the general orthopaedic and specific arthroplasty literature to be associated with a risk of infection, whether that be surgical site infections or deeper infections.41, 42, 43 Therefore, it is a logical extrapolation that surgeons who have a higher percentage of cases requiring allogenic pRBC transfusion and those cases that require higher numbers of allogenic units transfused are at increased risk for developing PJI's. Additionally, longer operative time has been found to be an independent risk factor for adverse effects following primary arthroplasty.44, 45, 46 A recent large retrospective database analysis of more than 100,000 primary hip and knee arthroplasty patients found increased operative time to be independently associated with anemia requiring transfusion, wound dehiscence, renal insufficiency, surgical site infection, and urinary tract infection.⁴⁶ There is evidence in the literature as well that operative time greater than 2 h is an independent risk factor for 30-day surgical site infections following revision hip and knee arthroplasty surgery.⁴⁷ It can be therefore safely assumed that surgeons who tend towards longer operating times, whether due to patient factors, technical difficulty of cases, or surgeon's experience, are at increased risk of their patients developing PJI.

As mentioned briefly above, there can be numerous explanations for the observed variations in surgeons' rate of transfusions, amount of blood transfused, and longer operative times. These can be explained by simple differences such as the variation between different surgeons' experience/years in practice and their average arthroplasty case load. On the opposite side of the spectrum are explanations based on factors that are more difficult to quantify, such as some surgeons’ patient populations representing more complex primary and revision surgeries that are inherently associated more blood loss and extended operative times.

There were several limitations inherent to this study. This was a retrospective analysis of surgeon, patient, insurance, and peri-operative factors specific to individual surgeons from an infection control database at one total joint replacement institution. We were not able to specifically measure the impact of differences in surgeon caseload and patient characteristics or surgical factors to specific quality outcomes in this data set and the analysis was limited to factors recorded in the infection control database.

5. Conclusion

As the economic structure of our medical system increasingly evaluates the individual orthopaedic surgeon, quality of care must be defined using a balance of outcomes measures as well as the accumulated risk from different types of patients and cases. In this study, we demonstrate that there is a substantial variability in patient and case factors that can be related to infection risk between individual surgeons within a single, urban, academic medical center. The degree to which this can be controlled is unclear. Further research is needed to determine precisely the factors that lead to patient variability among individual surgeons and its impact on quality measuring and reporting. Efforts to evaluate and report the quality of care of individual surgeons should account for the variability of patient and case factors that impact the quality outcome measures to allow for equal comparison.

Declaration of competing interest

One of the authors has financial interest in Genovel and Proventus health, in the form of stock options, not relevant to this submitted work. One of the authors is a paid speaker for Pacira, receives fellow support from Smith & Nephew and research support from Biomet, not relevant to this submitted work.

All of the authors assert that none of the declared financial interests or relationships represent any conflict of interest with this submitted work.

For full information regarding the conflicts of interest please refer to our available conflict of interest declarations.