Are Histological Examinations of Arthroplasty Specimens Performed Consistently Across the Country? A Large Database Study

Abstract

Background

It is currently unknown to what extent routine histological examination of joint arthroplasty specimens occurs across hospitals nationwide. Although this practice is neither supported nor refuted by the available evidence, given the increasing demand for joint arthroplasties, it is crucial to study overall utilization as well as its main drivers.

Questions/purposes

Using national data on joint replacements, we aimed to evaluate: (1) What is the current use of routine histological examination of joint arthroplasty specimens? (2) Does the use vary by geographic location and hospital characteristics? (3) Has use changed over time?

Methods

From the Premier Healthcare database (2006-2016) we included claims data from 87,667 shoulder (595 hospitals, median age 70 years, 16% nonwhite), 564,577 hip (629 hospitals, median age 65 years, 21% nonwhite), and 1,131,323 (630 hospitals, median age 66 years, 24% nonwhite) knee arthroplasties (all elective). Our study group has extensive experience with this data set, which contains information on 20% to 25% of all US hospitalizations. Included hospitals are mainly concentrated in the South (approximately 40%) with equal distributions among the Northeast, West, and Midwest (approximately 20% each). Moreover, the Premier data set has detailed billing information, which allows for evaluations of real-world clinical practice. There was no missing information on the main variables of interest for this specific study. We assessed frequency of histology examination (defined by Current Procedural Terminology codes) overall as well as by hospital characteristics (urban/rural, bed size, teaching status, arthroplasty volume), geographic region (Northeast, South, Midwest, West), and year. Given the large sample size, instead of p values, standardized differences were applied in assessing group differences where a standardized difference of > 0.1 (or 10%) was assumed to represent a meaningful difference between groups. For significance of trends, p values were applied. Percentages provided represent proportions of individual procedures.

Results

In most hospitals, histology testing was either rare (1%-10%, used in 187 of 595, 189 of 629, and 254 of 630 hospitals) or ubiquitous (91%-100%, used in 121 of 595, 220 of 629, and 195 of 630 hospitals) for shoulder, hip, and knee arthroplasties, respectively. Overall, histology testing occurred more often in smaller hospitals (37%-53% compared with 26%-45% in larger hospitals) and those located in the Northeast (59%-68% compared with 22%-44% in other regions) and urban areas (32%-49% compared with 20%-31% in rural areas), all with standardized differences > 10%. Histologic examination is slowly decreasing over time: from 2006 to 2016, it decreased from 34% to 30% for shoulder arthroplasty, from 50% to 45% for THAs, and from 43% to 38% for TKAs (all p < 0.001).

Conclusions

Although overall use is decreasing, a substantial number of hospitals still routinely perform histology testing of arthroplasty specimens. Moreover, variation between regions and hospital types suggests that this practice is driven by a variety of factors. This is the first study addressing national utilization, which will be helpful for individual hospitals to assess how they compare with national utilization patterns. Moreover, the findings have clear implications for followup studies, which may be necessary given the exponentially growing demand for arthroplasties.

Level of Evidence

Level III, therapeutic study.

Introduction

Routine histological examination of arthroplasty specimens has been evaluated in several single-institutional studies during the past two decades with varying conclusions on the meaningful contribution of this practice to patient care [3, 4, 6, 8, 12, 17, 18, 22, 27, 33]. Although valuable, these studies may not be generalizable to the national level. To date, a national assessment on the utilization of routine histological examination of arthroplasty specimens is still absent. With ever more national data sets available, clinical researchers are now able to address this research gap.

Given the rapidly increasing demand for total hip, knee, and shoulder arthroplasties [11, 14, 15], data on utilization patterns would provide valuable information on a single aspect of healthcare delivery for a large group of patients. In addition, several drivers of routine histology testing may exist [2, 3, 12, 17, 21, 22, 24] and thus an evaluation of trends and factors associated with this practice may lead to further insight into utilization patterns. Here, theoretically, hospital-specific rather than patient-specific factors are likely drivers given the presumed routine nature of this practice.

Therefore, using a nationwide database, we aimed to examine current utilization patterns and hospital-specific trends of routine histology examination of surgical specimens in shoulder, hip, and knee arthroplasties. We asked: (1) What is the current use of routine histological examination of joint arthroplasty specimens? (2) Does the use vary by region of the country, hospital size, or teaching status? (3) Has the use changed over time?

Patients and Methods

Data Source and Study Design

Data collected from the Premier Healthcare Database (Premier Healthcare Solutions Inc, Charlotte, NC, USA) [19, 31] were used for this retrospective cohort study. This database contains administrative claims data on approximately 20% to 25% of US hospital discharges. Records include International Classification of Diseases, 9th Revision (ICD-9) codes, Current Procedural Terminology (CPT) codes, and complete inpatient billing items. Next to containing detailed billing information for evaluations of real-world practice, Premier was specifically chosen based on our study team’s extensive experience with this data set, resulting in several high-impact publications [23, 26, 30]. Hospitals participating in Premier submit their institutional data for benchmarking purposes and are mainly concentrated in the South (approximately 40%) with equal distributions among the Northeast, West, and Midwest (approximately 20% each) [19]. Additionally, information is provided on crude measures of patient case mix to facilitate comparisons (Table 1). Variables of interest were extracted from inpatient billing; we expect a high degree of reliability of these variables, specific for this study, because reimbursement may be dependent on their accuracy. There was no missing information on most (hospital-level) variables of interest for this specific study; variables containing missing values are reported as such using separate categories. Our hospital’s institutional review board approved use of the HIPAA-compliant anonymized data for this study (project #14-00647).

Table 1.

Baseline patient characteristics for each study cohort

Study Sample

We included patients who underwent a shoulder, hip, or knee arthroplasty between 2006 and 2016 by searching ICD-9 codes for total shoulder arthroplasty (81.80), reverse total shoulder arthroplasty (81.88), THA (81.51), and TKA (81.54). To create a homogenous cohort without hospitals being subject to fluctuations in histology utilization based on low volume, we excluded patients undergoing more than one different arthroplasty during the same hospitalization (n = 363; this did not apply to bilateral arthroplasties), those treated in low-volume hospitals that performed < 30 arthroplasties over the entire study period (n = 322), and those classified as an outpatient procedure (n = 8505) or a nonelective procedure (n = 118,158). The final study cohort included a total of 1,783,567 patients (see Figure, Supplemental Digital Content 1, http://links.lww.com/CORR/A131).

Study Variables

The main finding of interest was histologic examination of an operative specimen in each of the arthroplasties. This was identified by assessing billing data on CPT codes generally associated with histologic examinations in these procedures [18]. In line with previous research [18], we assumed that histologic examination of an operative specimen occurred when any of the following CPT codes were present: 88184, 88185, 88187, 88188, 88189, 88300, 88304, 88305, 88311, 88313, 88342, or 88343 (Table 2). Other study variables included year of procedure and hospital region (Northeast, Midwest, South, West), location (rural, urban), bed size (< 300, 300-499, ≥ 500 beds), teaching status (teaching, nonteaching), and hospital-specific annual volume of arthroplasties.

Table 2.

Overview of frequency of billing for CPT codes used to identify histology testing

Statistical Analyses

All analyses were performed separately for each arthroplasty group. We first assessed any differences in prevalence of histologic examinations by hospital characteristics. Given our large sample size, we used standardized differences instead of p values to assess between-group differences. A standardized difference of 0.1 (or 10%) has been suggested to indicate a meaningful difference in covariate distribution between groups [1, 37]. We then evaluated annual percentages of histology examination by producing a line graph to reveal the trends. Trend significance was assessed using the Cochran-Armitage trend test with p < 0.05 defined as statistically significant. The use of hospital-specific histology testing was further evaluated in bar charts with hospitals stratified by deciles of histology testing. Finally, we assessed frequencies of the separate CPT codes generally associated with histologic examinations in shoulder, hip, and knee arthroplasties based on previous research [18].

All analyses were performed using SAS 9.4 (Cary, NC, USA).

Results

Overall, billing for histological examination was present in 41% (737,739 of 1,783,567) of joint arthroplasties during the study period. This was 31% for shoulder arthroplasties (27,168 of 87,667), 47% (267,810 of 564,557) for hip arthroplasties, and 39% (442,743 of 1,131,323) for knee arthroplasties, representing 595, 629, and 630 hospitals, respectively. The most prevalent CPT codes reflective of histological examination were 88300, 88304, 88305, and 88311 (Table 2).

Use of histology for arthroplasty specimens varied based on geographic location and hospital characteristics (Table 3). Across all three arthroplasty cohorts, utilization was highest in the Northeast of the United States (59% [7397 of 12,534] to 68% [79,459 of 117,341]) compared with the South (22% [9184 of 41,181] to 42% [94,122 of 225,983]), Midwest (35% [6436 of 18,414] to 44% [47,165 of 107,346]), and West (27% [4169 of 15,538] to 41% [47,064 of 113,907]); these regional differences in histology utilization were meaningful as demonstrated by standardized differences of > 10% in all three arthroplasty cohorts. In addition, histology testing was performed more commonly in urban hospitals (32% [25,538 of 79,608] to 49% [252,405 of 515,365]) compared with rural hospitals (20% [1648 of 8059] to 31% [15,405 of 49,212]) and in hospitals with < 300 beds (37% [13,207 of 35,538] to 53% [115,316 of 218,545]) compared with those with 300 to 499 beds (27% [7712 of 28,303] to 44% [78,530 of 179,901]) or ≥ 500 beds (26% [6267 of 23,826] to 45% [73,964 of 166,131]); standardized differences between hospital types was > 10% across all arthroplasty cohorts. In comparison, minor or less consistent group differences were seen when comparing histology utilization between hospitals based on teaching status or annual arthroplasty volume. The distribution for shoulder arthroplasties demonstrates that the majority of hospitals performing arthroplasties either rarely perform histology testing (1%-10% of the time, n = 47,357 patients in 187 hospitals) or almost always perform histology testing (91%-100% of the time, n = 14,161 patients in 121 hospitals), with a much smaller subset of hospitals occasionally (11%-90% of the time) performing histology testing (Fig. 1A). Similar patterns were seen in hip and knee arthroplasties (Fig. 1 B-C).

Table 3.

Hospital characteristics by yes/no histology performed in shoulder, hip, and knee arthroplasties

Fig. 1 A-C.

(A) This graph shows the distribution of hospital-specific histology testing (in % of all procedures) for shoulder arthroplasties; data labels represent the number of patients and do not reflect bar sizes, which are based on number of hospitals in each utilization category. RSA/TSA = reverse/total shoulder arthroplasty. (B) This graph shows the distribution of hospital-specific histology testing (in % of all procedures) for THAs; data labels represent the number of patients and do not reflect bar sizes, which are based on number of hospitals in each utilization category. (C) This graph shows the distribution of hospital-specific histology testing (in % of all procedures) for TKAs; data labels represent the number of patients and do not reflect bar sizes, which are based on number of hospitals in each utilization category.

Use of histology examinations after arthroplasty decreased over the time period we surveyed (Fig. 2; p < 0.001); for shoulder arthroplasties, the percentage utilizing histology examination decreased from 34% (896 of 2673) in 2006 to 30% (4083 of 13,792) in 2016. Similar trends were seen in hip and knee arthroplasties, decreasing from 50% (16,128 of 32,395) to 45% (30,002 of 66,629) and from 43% (29,940 of 69,404) to 38% (46,761 of 124,587), respectively. Histology testing for shoulder arthroplasties peaked in 2009 at 35% (1628 of 4638); this was 51% (23,690 of 46,645) in 2010 for hip arthroplasties and 43% (29,940 of 69,404) in 2006 for knee arthroplasties.

Fig. 2.

This graph shows trends in histologic examinations of arthroplasty specimens by procedure group. RSA/TSA = reverse/total shoulder arthroplasty.

Discussion

Utilization of routine histology testing of arthroplasty specimens has mainly been assessed in single-institutional studies that lack generalizability. National figures may be particularly helpful for individual hospitals to assess how they compare with national utilization patterns. Furthermore, such data may also provide insight into drivers of histology testing, which may be particularly interesting for policymakers. This is the first study to provide national estimates on utilization patterns, how this is influenced by region and hospital-specific factors, and how its prevalence evolved over time. We found that histology testing of arthroplasty specimens occurs in 31%, 47%, and 39% of shoulder, hip, and knee arthroplasties, respectively. Histology testing was more common in smaller and urban hospitals and those located in the Northeast. When examining the distribution of utilization patterns, we found that, in general, hospitals either perform histology testing infrequently (1%-10%) or routinely (91%-100%). Interestingly, slow decreases were observed in trends of histology testing.

This study is burdened by several limitations. Importantly, our data source only provides information on utilization patterns and does not include clinical details such as the result of histology testing and reasons for testing (other than routine). In addition, we relied on a variety of CPT codes based on previous research to identify histology testing. This was done to account for variations in coding practices between hospitals. For example, CPT codes 88305 and 88304 technically make a distinction between fractures and nonfractures. However, the former CPT code is used in 283,454 of 1,131,323 (25%) knee arthroplasties, which are not likely to be all representative of fractures, but rather differing coding practices. Another limitation refers to the fact that the Premier database covers only a sample of 20% to 25% of US hospitals. Although all regions are represented, it appears that particularly data from hospitals in the South are included in the database (Table 3). Given that histology testing rates were relatively low in the South, this could have resulted in an underestimation of true utilization. Finally, we were not able to include reliable estimates of cost in our study. This has been the focus of several smaller studies, although it may not be generalizable. Given the ongoing discussion on payment reform, in particular lower extremity joint arthroplasties, there is a need for detailed cost data that are generalizable. Multiinstitutional data sets specifically geared toward this goal may provide important insights. Overall, despite our study’s limitations, the descriptive statistics on current utilization are crucial data that underline the scale of this practice that has been questioned in the past and may inform followup studies.

We found that the percentage of hip arthroplasties in which histological evaluation was performed varied between 31% and 47% for shoulder, hip, and knee arthroplasties. Although previous studies have mainly focused only on cohorts in which histology testing actually took place [3, 4, 12, 17, 18, 22, 27], this is the first study to provide estimates on utilization by adding procedures in which histology testing did not occur to the study denominator. Interestingly, testing rates appear to be lower in shoulder compared with hip and knee arthroplasties. Although we can only speculate on the exact mechanism behind these differences, several factors may play a role, including the cohort-specific rates of discrepant diagnoses: disagreement between clinically reported and pathologic diagnosis, typically not affecting patient management, found in previous studies. Indeed, the largest study to date found higher rates of discrepant diagnoses in hip (18.8% [n = 7968]) compared with knee arthroplasties (9.4% [n = 8619]) [4]; this was recently found to be 5.9% among shoulder arthroplasty specimens [8]. Importantly, this hierarchy was also reflected in discordant diagnoses: disagreement between clinically reported and pathologic diagnosis that does affect patient management [4, 8]. Given that the current study could not focus on exact drivers of histology testing, future studies could evaluate surgeons’ and pathologists’ attitudes toward routine histology testing of arthroplasty specimens. This will further elucidate the interplay between various factors in the decision-making toward routine testing.

Our study showed that histology testing was more common in smaller hospitals and those located in the Northeast while we also demonstrate that, in general, hospitals either perform histology testing infrequently (1%-10%) or routinely (91%-100%); this suggests that drivers of histology testing are mainly hospital-related rather than related to individual patient characteristics. Indeed, practice and care differences between hospital types have been extensively documented; for example, smaller volume hospitals have been shown to be less adherent to evidence-based procedures than larger hospitals in a variety of medical fields, including the management of complex cardiovascular diseases [36] and the treatment of sepsis [5]. Within orthopaedics, larger hospitals are associated with reduced mortality, pulmonary embolism, and infection while also positively associated with joint arthroplasty implant survivorship in total joint arthroplasties [7, 9, 10, 13, 16, 20, 25, 34, 35]. One possible explanation for the higher rate of histology testing in the Northeast may relate to the overall higher costs of joint arthroplasties in the Northeast [11]. Although there may be several drivers of these regional cost differences, increased resource use in some regions may be an additional factor. Another explanation for this higher rate of testing could be the perception that medicolegal exposure is greater in the Northeast; however, we are unaware of data in support of this contention, and studies have not found wrongful diagnosis or pathology findings to be common causes of malpractice suits [28, 29]. Further research is needed to fully understand the background of these regional differences.

We found that, although histology testing is slowly decreasing, it still occurs in a substantial number of procedures and as routine practice in a large number of hospitals. There may be a variety of reasons for this, including patient protection from clinically overlooked occult but significant diseases, guarding practitioners from failure-to-diagnose litigation, facility accreditation reasons, or the provision of more accurate data for public health tracking of the true incidence and prevalence of disease. Indeed, there appears to be a complex interplay among local hospital protocols, mandates from the Joint Commission on Accreditation of Healthcare Organizations, and the College of American Pathologists, which accredits hospital pathology laboratories [2, 21, 24]. Crucially, these mechanisms are all independent of the question of cost-effectiveness, which has been the focus of several previous studies [3, 6, 8, 12, 18]. Given that some have questioned the practice of routinely examining arthroplasty specimens [3, 4, 6, 12, 17, 22, 27, 33], it is tempting to label it as “unnecessary”; however, this grossly overstates cost analyses from relatively small sample sizes and understates other noncost-effective-related reasons for histology testing. Indeed, cost-effectiveness alone should not be the deciding factor in whether to perform histopathological examinations, as previously noted [32]. Moreover, crucially, histologic examination of surgical specimens does not harm the patient, which may also facilitate the continuation of this practice. Given the current emphasis on cost containment in health care, specifically in highly standardized procedures such as hip and knee arthroplasties, the discussion on cost will persist and with it the continued need to assess all determinants of costs, including relatively minor ones such as histology testing. From a population perspective, however, even services with relatively minor costs may result in substantial cost burdens on the population level. For example, when extrapolating the frequency of histology testing found in the current study and using median costs of included CPT codes (Table 2) as recorded in the Premier database, annual estimated costs of histology testing in shoulder, hip, and knee arthroplasties easily exceed USD 600,000, USD 7,000,000, and USD 10,000,000, respectively, on the national level. However, although these costs appear considerable, they account for only a fraction of the overall annual costs of arthroplasties. Important followup studies therefore could focus on precise cost estimates, the balance between cost and clinical utility from different perspectives [33], and strategies for selective histology testing.

In conclusion, in this first national study looking into histology testing rates in shoulder, hip, and knee arthroplasties, we found that, although slowly decreasing, this practice still occurs in many patients and is still routine in a large number of hospitals. Histology testing particularly occurred in smaller and urban hospitals and those located in the Northeast. Because several studies do not support routine histology testing, future multiinstitutional studies should focus on the relative importance of all drivers of this practice as well as the utility and value of histological testing of arthroplasty specimens in light of these various drivers. Although this study provides important information on the current wide variation in practice in this country, further studies focusing particularly on cost-effectiveness may provide additional insights into mechanisms behind utilization patterns.