Accuracy of Medicare claims data for rheumatologic diagnoses in total hip replacement recipients

https://doi.org/10.1016/S0895-4356(03)00056-8Get rights and content

Abstract

This analysis was performed to examine whether Medicare claims accurately document underlying rheumatologic diagnoses in total hip replacement (THR) recipients. We obtained data on rheumatologic diagnoses including rheumatoid arthritis (RA), avascular necrosis (AVN), and osteoarthritis (OA) from medical records and from Medicare claims data. To examine the accuracy of claims data we calculated sensitivity and positive predictive value using medical records data as the “gold standard” and assessed bias due to misclassification of claims-based diagnoses. The sensitivities of claims-based diagnoses of RA, AVN, and OA were 0.65, 0.54, and 0.96, respectively; the positive predictive values were all in the 0.86–0.89 range. The sensitivities of RA and AVN varied substantially across hospital volume strata, but in different directions for the two diagnoses. We conclude that inaccuracies in claims coding of diagnoses are frequent, and are potential sources of bias. More studies are needed to examine the magnitude and direction of bias in health outcomes research due to inaccuracy of claims coding for specific diagnoses.

Introduction

Administrative data are often used in studies of population or community-based samples [1]. Typically, these data contain information on diagnoses, surgical procedures, and health care utilization for a large number of patients, and are less costly to obtain than data from medical records. However, the accuracy of administrative claims data varies across medical conditions [2], [3], [4]. Although procedure-based data are more accurate than those based solely on diagnoses, the accuracy of rheumatologic diagnoses derived from administrative data may not be ideal. For example, the predictive positive value for osteoarthritis (OA) obtained from HMO administrative data was reported to be 62% in one study [5], while the predictive positive value for OA obtained from Medicare data in another multisite study was 83% [6].

Total hip replacement is commonly used to reduce pain and improve functional status among elderly patients with advanced hip disease [7]. Although osteoarthritis is the most common underlying diagnosis in patients undergoing THR, rheumatoid arthritis, avascular necrosis, and other diagnoses also can lead to joint destruction necessitating total hip replacement. The underlying rheumatologic diagnosis may influence short-term perioperative complications and long-term functional outcomes of total hip replacement [3]; hence, it is important to capture the rheumatologic diagnoses in studies of THR outcomes.

The goal of this analysis was to determine the sensitivity and positive predictive value of specific rheumatologic diagnoses obtained from inpatient Medicare claims on a population-based cohort of THR recipients. We describe the association of hospital volume of THR and accuracy in the coding of specific rheumatologic diagnoses. To the best of our knowledge, no prior studies have examined the influence of hospital procedure volume on the coding accuracy of the underlying procedure-specific diagnoses. We illustrate our findings with two examples describing how inaccuracy in the diagnoses derived from administrative data may lead to biased estimates of the effect of specific diagnoses upon outcomes of elective total hip replacement.

Section snippets

Sample

This analysis was built upon a study of the association between hospital volume and outcomes of elective total hip replacement in the Medicare population [8], [9]. The participants in the parent study were selected through a two-stage process. Using inpatient Medicare claims data we first identified all Medicare beneficiaries who received elective primary total hip replacement in 1995. Subsequently, we selected a stratified random sample of THR recipients from Ohio, Pennsylvania, and Colorado.

Estimates of prevalence, sensitivity, and positive predictive value for hospital claims

The vast majority of the 922 study participants had osteoarthritis as the underlying indication for THR. The prevalence of osteoarthritis was 0.87 (95% CI: 0.85–0.89) as defined by medical records, compared with 0.94 (95% CI: 0.93–0.96) as defined by hospital claims based coding (Table 2). On the other hand, the prevalences of rheumatoid arthritis and of avascular necrosis were underreported in claims data. The prevalence of rheumatoid arthritis was 4% (95% CI: 0.03–0.05) in medical records vs.

Discussion

We found that Medicare claims tend to underreport some underlying rheumatologic diagnoses in patients undergoing total hip replacement. Although the accuracy of claims-derived diagnosis of osteoarthritis (the major underlying condition for total hip replacement) was high, and the positive predictive value for all three diagnoses was fairly high, the sensitivity for rheumatoid arthritis and avascular necrosis, other diagnoses leading to total hip replacement, was poor. We also found that

Acknowledgements

Funding for this article was supplied by: NIH P60 AR 47782, NIH K24 AR 02123, and a Clinical Science Grant from the Arthritis Foundation.

References (17)

There are more references available in the full text version of this article.

Cited by (95)

  • Association measures of claims-based algorithms for common chronic conditions were assessed using regularly collected data in Japan

    2018, Journal of Clinical Epidemiology
    Citation Excerpt :

    To deal with these concerns, many studies have assessed the validity of diagnostic and/or procedural codes of medical and pharmacy claims data and proposed algorithms to identify certain conditions including hypertension, diabetes, and dyslipidemia from such data [9,10,15–36]. These attempts, however, have only been applicable to limited populations because most of the conducted studies use medical charts as their gold standard [15,17,21,22,24,25,27,29,31,32,34]. Consequently, it remains unclear if the algorithms are applicable to a wide range of populations, and whether the claims data can accurately estimate the number of patients with corresponding diseases at population level despite the increasing number of studies using large-scale claims data.

View all citing articles on Scopus
View full text