If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Surgeon and hospital volumes may affect outcomes of various orthopedic procedures. The purpose of this study is to characterize the volume dependence of both facilities and surgeons on morbidity and mortality after total knee arthroplasty.
Methods
Adults who underwent total knee arthroplasty for osteoarthritis from 2011 to 2015 were identified using International Classification of Diseases-9 Clinical Modification diagnostic and procedural codes in the New York Statewide Planning and Research Cooperative System database. Readmission, in-hospital mortality, and other adverse events were compared across surgeon and facility volumes using multivariable Cox proportional hazards regression, while controlling for patient demographic and clinical factors. Surgeon and facility volumes were compared between the lowest and highest 20%.
Results
Of 113,784 identified patients, 71,827 were treated at a high- or low-volume facility or by low- or high-volume surgeon. Low-volume facilities had higher 1-month, 3-month, and 12-month rates of readmission, urinary tract infection, cardiorespiratory arrest, surgical site infection, and wound complications; higher 3- and 12-month rates of pneumonia, cellulitis, and in-facility mortality; and higher 12-month rates of acute renal failure and revision. Low-volume surgeons had higher 1-, 3-, and 12-month rates of readmission, urinary tract infection, acute renal failure, pneumonia, surgical site infection, deep vein thrombosis, pulmonary embolism, cellulitis, and wound complications; higher 3- and 12-month rates of cardiorespiratory arrest; and higher 12-month rate of in-facility mortality.
Conclusions
These results suggest volume shifting toward higher volume facilities and/or surgeons could improve patient outcomes and have potential cost savings. Furthermore, these results can inform healthcare policy, for example, designating institutions as centers of excellence.
The national healthcare expenditure in the United States is projected to increase to $5.4 trillion by 2024, which will account for 19.6% of the gross domestic product [
]. Total knee arthroplasty (TKA) is a target of healthcare reform given the high annual volume and overall cost burden on the healthcare system. According to projection models based on primary TKAs from 2000 to 2014, the estimated annual TKA volume will be approximately 935,000 procedures by 2030 [
]. Furthermore, other models estimate an overall 143% growth in volume by 2050, consequently predicting that TKA will be performed for 725 of every 100,000 people [
A 2014 review of Medicare beneficiaries receiving primary or revision total joint arthroplasties (TJAs) showed that the average cost ranged greatly: primary TJAs for patients without comorbidities had an average cost of $25,568, and revision TJAs for those with major comorbidities or complications had an average cost of $50,648 [
]. Postdischarge care accounted for 35% of total cost, the biggest contributors being the 49% of patients who were transferred to post–acute care facilities (70% of postdischarge costs) and the 10% of patients who were readmitted for complications related to their TJA (11% of postdischarge costs) [
] of the Nationwide Readmissions Database from the Healthcare Cost and Utilization Project showed that the overall annual total cost for 90-day readmissions after TKA was $629 million with 239,700 days of hospital stays and $417 million covered by Medicare. Considering the significant national economic burden, alongside both the aging United States population and the increased life expectancy [
], it is critical to explore the delivery of TKA and to promote safe pathways to cost-effective care.
Both surgeon and hospital volumes are well-known characteristics that affect the outcomes of various orthopedic procedures. For example, a 2011 analysis of the Pennsylvania Health Care Cost Containment Council database reported that 1-year mortality was significantly higher among patients aged 65 years and older who received elective primary TKA at a lower volume hospital [
]. While numerous studies have explored the relationship between provider and hospital volume and TKA results, they have consistently demonstrated increased risks of postsurgery complications after procedures by low-volume providers or in low-volume hospitals [
]. The purpose of the current study is to characterize the volume dependence of both facilities and surgeons on post-TKA morbidity and mortality. This study also explores a wider range of complications than similar articles and simultaneously examines the effect of patient demographics such as comorbidities and social deprivation. We hypothesize that patients who receive their treatment from high-volume hospitals and high-volume surgeons will have reduced rates of mortality and complications compared with patients of low-volume hospitals and surgeons.
Material and methods
Patients ≥40 years old were identified in the New York Statewide Planning and Research Cooperative System (SPARCS) database from 2011 to 2015. The SPARCS is a comprehensive all-payer database collecting all inpatient and outpatient (emergency department, ambulatory surgery, and hospital-based clinic visits) claims in New York State. This includes International Classification of Diseases (ICD) diagnosis codes and ICD/Current Procedural Terminology (CPT) procedure codes associated with all visits. Inpatient claims were first identified using the ICD-9 Clinical Modification (CM) knee osteoarthritis diagnosis codes (715.16, 715.26, 715.36, and 715.96). Claims were then filtered by ICD-9-CM procedure codes to isolate patients who went on to receive a TKA (ICD-9 CM: 81.54). Only a patient’s first operation was considered eligible for follow-up. Nonresidents of New York were not included in our analysis. Given ICD-9 coding was discontinued after the third quarter of 2015, only the first 3 quarters of 2015 were used as these statistics are still likely to be indicative of the low to high volume comparison.
Unique surgeon and facility identifiers were used to calculate the total number of procedures per surgeon and facility per year. Based on the total volume per year, surgeons and facilities were subject to the lowest 20% of volume, middle 60% of volume, or highest 20% of volume. The boundaries for the lowest and highest 20% deviated slightly by year but were selected to minimize the difference from the 20% volume mark.
Patients were followed up to a maximum of 1 year postoperatively in the inpatient and outpatient setting. The 1-month, 3-month, and 12-month risks of interest were as follows: readmission, urinary tract infection, acute renal failure, cardiorespiratory arrest, pneumonia, acute stroke, surgical site infection, deep vein thrombosis, acute respiratory failure, pulmonary embolism, cellulitis, wound complications, in-facility mortality, and revision surgery (see Supplemental Table 1 for codes used). SPARCS claim dates are listed as the first day of the month in which the service occurred owing to SPARCS deidentification policy. Therefore, if a complication occurred within the same month as the primary procedure, the time to complication was defined as 0.5 months [
Patient demographics were compared separately across facility volume and surgeon volume using chi-squared analysis. T-tests were used for comparing sample means, and Mann-Whitney U tests were used when appropriate when continuous data were found to be not normally distributed.
Multivariable Cox proportional hazards regression was used for the analysis of risk likelihood across the volume groups. Each complication was modeled separately while controlling for patient age, sex, race, ethnicity, Charlson Comorbidity Index (CCI), primary insurance type, and social deprivation index (SDI). Other race excludes White, Asian, and African American but does include multiracial patients. The regression models assess the risk difference across surgeon and facility groups simultaneously by controlling for both in the same model.
The CCI was calculated using the method described by Deyo et al [
]. The CCI was dichotomized to a score of 0 vs a score of ≥1. The SDI as described by Butler et al. was linked to each patient based on ZIP code. The SDI provides a robust measure of social determinants of health not traditionally captured by healthcare administrative databases by converting the following categories to an index from 1-100: percent living in poverty, percent with less than 12 years of education, percent single parent household, percent living in rented housing unit, percent living in overcrowded housing unit, percent of households without a car, and percent nonemployed adults younger than 65 years. A higher SDI score equates to increased social deprivation. SDI data in this study were based on 2015 statistics [
A P-value <.05 was considered significant across all statistical analyses. All analyses were performed using SAS, version 9.4 (SAS Inc, Cary, NC).
Results
Of the 113,784 patients identified, 71,827 patients were treated at a high- or low-volume facility or by a high- or low-volume surgeon. Yearly facility volume ranged from 1 to 3442 (mean: 156, median: 78) procedures. Yearly surgeon volume ranged from 1 to 495 (mean: 34, median: 18) procedures. The number of procedures per year in New York increased slightly from 24,313 in 2011 to 25,536 in 2014 (19,626 through 3 quarters of 2015). The range for the number of procedures used as the upper boundary for the lowest 20% of volume by facility was 128-149 (115 through 3 quarters of 2015), and the range for the lower boundary for the highest 20% was 766-897 (645 through 3 quarters of 2015). Low-volume facilities accounted for 22,561 procedures, and high-volume facilities accounted for 23,291 procedures. The range of the number of procedures used as the upper boundary for the lowest 20% of volume by surgeon was 34-37 (29 through 3 quarters of 2015), and the range for the lower boundary for the highest 20% was 149-161 (126 through 3 quarters of 2015). Low-volume surgeons accounted for 23,232 procedures, and high-volume surgeons accounted for 22,865 procedures (Table 1, Table 2).
Table 1Patient demographics and characteristics, by facility volume.
Several demographic differences were noted to be statistically significant. Low-volume facilities and surgeons had patient age distributed toward younger ages relative to high volume and higher social deprivation relative to high volume (Table 1, Table 2). Low-volume facilities had increased incidence of female sex, Hispanic ethnicity, African American race, other race, federal insurance, self-pay, and having ≥1 Charlson comorbidity (Table 1). Low-volume surgeons had increased incidence of Hispanic ethnicity, Asian race, African American race, other race, and having ≥1 Charlson comorbidity (Table 2).
Compared with high-volume facilities, low-volume facilities had higher 1-month, 3-month, and 12-month rates of readmission, urinary tract infection, cardiorespiratory arrest, surgical site infection, and wound complications; higher 3 and 12-month rates of pneumonia, cellulitis, and in-facility mortality; and higher 12-month rates of acute renal failure and revision. Low-volume facilities had lower 1-, 3-, and 12-month rates of pulmonary embolism and lower 1-month rate of acute stroke (Table 3). Compared with high-volume surgeons, low-volume surgeons had higher 1-, 3-, and 12-month rates of readmission, urinary tract infection, acute renal failure, pneumonia, surgical site infection, deep vein thrombosis, pulmonary embolism, cellulitis, and wound complications; higher 3- and 12-month rates of cardiorespiratory arrest; and higher 12-month rate of in-facility mortality (Table 4).
Table 3Risk of complication after knee arthroplasty, by facility volume.
Complication
Low volume, n = 22,561
High volume, n = 23,291
Hazard ratio (95% CI)
P-value
Readmission
1 month
1280 (5.7)
952 (4.1)
1.192 (1.091-1.303)
.0001
3 month
1966 (8.7)
1420 (6.1)
1.244 (1.158-1.338)
<.0001
12 month
4277 (19)
3469 (14.9)
1.176 (1.122-1.233)
<.0001
Urinary tract infection
1 month
780 (3.5)
593 (2.6)
1.196 (1.067-1.34)
.002
3 month
920 (4.1)
659 (2.8)
1.277 (1.148-1.42)
<.0001
12 month
1388 (6.2)
993 (4.3)
1.287 (1.18-1.404)
<.0001
Acute renal failure
1 month
536 (2.4)
402 (1.7)
1.104 (0.962-1.266)
.1583
3 month
595 (2.6)
442 (1.9)
1.121 (0.984-1.277)
.0871
12 month
858 (3.8)
610 (2.6)
1.191 (1.067-1.329)
.0018
Cardiorespiratory arrest
1 month
24 (0.1)
7 (0)
2.611 (1.081-6.308)
.033
3 month
37 (0.2)
8 (0)
3.533 (1.59-7.852)
.0019
12 month
57 (0.3)
27 (0.1)
1.791 (1.105-2.905)
.0181
Pneumonia
1 month
206 (0.9)
154 (0.7)
1.149 (0.92-1.435)
.2209
3 month
259 (1.2)
182 (0.8)
1.252 (1.024-1.531)
.0285
12 month
495 (2.2)
319 (1.4)
1.357 (1.169-1.575)
<.0001
Acute stroke
1 month
192 (0.9)
242 (1)
0.789 (0.644-0.965)
.0214
3 month
230 (1)
267 (1.2)
0.864 (0.715-1.043)
.128
12 month
391 (1.7)
382 (1.6)
1.007 (0.866-1.17)
.9323
Surgical site infection
1 month
410 (1.8)
276 (1.2)
1.224 (1.041-1.44)
.0146
3 month
495 (2.2)
331 (1.4)
1.232 (1.063-1.428)
.0057
12 month
671 (3)
480 (2.1)
1.173 (1.035-1.329)
.0121
Deep vein thrombosis
1 month
435 (1.9)
350 (1.5)
1.053 (0.907-1.222)
.4973
3 month
519 (2.3)
415 (1.8)
1.067 (0.93-1.223)
.3563
12 month
615 (2.7)
519 (2.2)
1.034 (0.914-1.171)
.595
Acute respiratory failure
1 month
78 (0.4)
67 (0.3)
1.034 (0.73-1.465)
.8501
3 month
91 (0.4)
74 (0.3)
1.137 (0.82-1.577)
.4412
12 month
152 (0.7)
111 (0.5)
1.277 (0.984-1.658)
.0656
Pulmonary embolism
1 month
171 (0.8)
307 (1.3)
0.5 (0.411-0.61)
<.0001
3 month
203 (0.9)
325 (1.4)
0.561 (0.466-0.675)
<.0001
12 month
268 (1.2)
356 (1.5)
0.672 (0.568-0.794)
<.0001
Cellulitis
1 month
476 (2.1)
370 (1.6)
1.128 (0.976-1.304)
.104
3 month
548 (2.4)
409 (1.8)
1.18 (1.03-1.353)
.0173
12 month
723 (3.2)
515 (2.2)
1.264 (1.121-1.425)
.0001
Wound complications
1 month
474 (2.1)
163 (0.7)
2.641 (2.188-3.188)
<.0001
3 month
524 (2.3)
196 (0.8)
2.405 (2.021-2.862)
<.0001
12 month
637 (2.8)
271 (1.2)
2.141 (1.841-2.489)
<.0001
In-facility mortality
1 month
35 (0.2)
17 (0.1)
1.707 (0.924-3.153)
.0879
3 month
47 (0.2)
21 (0.1)
1.936 (1.123-3.339)
.0175
12 month
109 (0.5)
48 (0.2)
1.851 (1.296-2.642)
.0007
Revision
1 month
7 (0)
2 (0)
2.44 (0.478-12.452)
.2835
3 month
15 (0.1)
4 (0)
2.393 (0.759-7.543)
.1362
12 month
43 (0.2)
9 (0)
3.951 (1.87-8.35)
.0003
CI, confidence interval.
Bolded values are for P < .05.
Hazard ratios are adjusted for surgeon volume, age, sex, race, ethnicity, primary insurance type, CCI, and SDI.
Figure 1 illustrates how the SDI varies across New York ZIP codes, with darker areas representing higher social deprivation. Figure 2 illustrates the rate of 3-month complications among patients by ZIP code stratified by facility and surgeon volume. Higher rates of complications can be appreciated in northern and western New York in Figure 2. These areas are also associated with higher social deprivation in Figure 1. Figure 3 illustrates the density of low- and high-volume facilities by county code. High-volume facilities are scarcer and tend to be concentrated in metropolitan areas. There is also a disproportionate amount of low-volume facilities in areas with the highest SDI scores: western New York, northern New York, and western Long Island. Figure 4 shows the density of patients with 1 or more Charlson comorbidity. Western Long Island has both high SDI scores as well as a high density of patients with a Charlson comorbidity (Figure 1, Figure 4).
Figure 1SDI by New York ZIP code. Gray ZIP codes had no TKA cases during the study period.
Figure 2Three-month complication rates by facility and surgeon volume by ZIP codes. Gray ZIP codes had either no complications or no TKA cases during the study period.
Figure 4Density of patients with 1 or more Charlson comorbidities in New York by ZIP code. Gray ZIP codes had either no TKA cases or no patients with a Charlson comorbidity.
This study supplemented the current literature concerning the relationship between hospital and surgeon volume and postoperative TKA morbidity and mortality by examining a wide range of complications, patient demographic and socioeconomic factors, and varying postoperative time periods. Additionally, this study evaluated the regionalization of complication rates and its relationship to socioeconomic status. The data showed an overall association between facility and surgeon volume with complications after TKA, thus coinciding with findings by other authors. For example, low-volume hospitals had significantly higher rates of, among other complications, readmission, wound complication, pneumonia, and cardiorespiratory failure. Likewise, low-volume surgeons had higher rates of acute renal failure, surgical site infection, deep vein thrombosis, etc. The literature on post-TJA results similarly states increased rates of complications, readmissions, reoperations, and mortality with low-volume centers and providers [
]. Additionally, our study found increased risk of revisions after 12 months for low-volume hospitals, a result that not only parallels other publications [
] but also reflects projection models that estimate increasing incidence of revision TKAs and consequently encourage institutions to generate revision-specific protocols to promote effective care [
The study also found an exception in the association between volume and outcome: low-volume facilities had lower 1-, 3-, and 12-month rates of pulmonary embolism and a lower 1-month rate of acute stroke. As stated previously, such findings have not been similarly shown in other TKA studies, as the literature tends to report increased rates of complications with decreased volume. A study of the American College of Surgeons National Surgical Quality Improvement Program database from 2008 to 2016 reported that overweight and obese patients had an increased risk of pulmonary embolism after primary TJA and the risk was elevated despite aggressive pharmacologic anticoagulation regimens [
]. Additionally, Anis et al. recently found that patients with a body mass index >40 were more likely to be treated at high-volume centers, thus suggesting a possible reason as to why high-volume facilities have increased risks of pulmonary embolism [
The current study also showed that, compared with patients with a CCI of 0, those with a CCI of 1 or greater were more likely to be treated at low-volume facilities. In contrast, a recent study has reported that increased CCI scores are associated with treatment at high-volume centers [
]; however, our findings suggest a counterintuitive association where patients with more comorbidities are treated at low-volume facilities and thus have an increased likelihood of postsurgery morbidity and mortality. Despite our retrospective study controlling for varying demographic features in its analysis of complication rates, there is a chance our results are due to a reversed causal effect and that patients treated at low-volume hospitals have more complications owing to having a higher CCI.
Our study has additionally found that more vulnerable demographics are suffering increased risk of post-TKA complications: in general, Hispanic, non-White patients, and those without private insurance were significantly more likely to be treated at low-volume hospitals and by low-volume surgeons. Additionally, areas with higher SDI scores tended to have an increased rate of patients with complications. Such disparities in access to health have been shown previously in the New York metropolitan area, as a study of adults undergoing surgery for cancer, cardiovascular disease, and orthopedic conditions showed that African American, Asian, and Hispanic patients were significantly less likely to be operated on by a high-volume surgeon or at a high-volume hospital [
]. Possible explanations for these trends include geographic location of providers, patients, and hospitals, as well as financial incentives where high-volume providers may be able to attract patients with better-paying insurance, a majority of whom may be White [
]. Thus, it is critical to consider racial and ethnic disparities in provision of care and consequent complications in an increasingly common orthopedic procedure. Furthermore, our study controlled for demographic factors such as race, SDI, and comorbidities in the analysis of risk for complications and still found significant effects of surgeon and facility volume. This highlights that it is critical that both high-volume care become more accessible and the gaps in the treatment between high- and low-volume care be identified and resolved.
The increased risk of postoperative morbidity and mortality at low-volume hospitals and surgeons affects not only the patient but also the healthcare system as a whole. Kurtz et al. showed that post-TKA complications had an annual economic burden of $64 million for infections, $52 million for acute cardiac events, $23 million for acute vascular and thrombotic events, $42 million for localized osteoarthrosis, etc. [
] Naturally, because high-volume hospitals have a greater capacity for care and are not limited to specialty care facilities, specialist medical teams, physiotherapy, and other resources, they may consequently be better equipped to proactively identify and resolve issues before they escalate and adversely influence patient outcomes [
A national analysis of the relationship between hospital volume, academic center status, and surgical outcomes for abdominal hysterectomy done for leiomyoma.
Finally, it is important to consider the fact that although a majority of related literature shares the consensus that lower volume yields worse outcomes in TKA patients, the definitions of “low” and “high” can vary. For example, Singh et al. defined a high-volume hospital as one that performs 101-200 procedures annually, whereas Anis et al. determined >500 as high volume [
]. This inconsistency is a consequent caveat to generalizing the results of different studies that analyze outcomes as a function of volume. We sought to apply volume percentiles as a way to improve the generalizability of this current study.
This study exhibits several limitations. The use of a large database inherently requires accurate coding. Because this study evaluated outcomes for the same procedure across the database, any differences in reporting should be global and the large sample size should help minimize substantial changes to the observed outcomes. Moreover, there are several significant demographic differences between the cohort included in this study (Table 1, Table 2), although we did attempt to control for these during our statistical analysis. Our study involved patients within the confined geographic zone of SPARCS database. Therefore, national and global trends cannot be directly considered, possibly limiting appropriate extrapolation to other areas. However, New York is a large state composed of a highly variable population of patients, hospitals, and surgeons with a great degree of demographic variability and therefore may be generalizable to larger populations [
The importance of case volume in TKA is relevant for both facilities and providers. Both low-volume facilities and surgeons performing primary TKA have higher rates of readmission, urinary tract infection, acute renal failure, cardiorespiratory arrest, pneumonia, surgical site infection, cellulitis, wound complications, and in-facility mortality. These results suggest volume shifting toward higher volume facilities and/or surgeons could improve patient outcomes and have potential cost savings. Furthermore, these results can inform healthcare policy, for example, designating institutions as centers of excellence.
Conflicts of interest
The authors declare that there are no conflicts of interest.
A national analysis of the relationship between hospital volume, academic center status, and surgical outcomes for abdominal hysterectomy done for leiomyoma.
00233-8&id=gr1.jpg){kind=link}
00233-8&id=gr2.jpg){kind=link}
00233-8&id=gr3.jpg){kind=link}
00233-8&id=gr4.jpg){kind=link}