Original Article
Veteran's affairs hospital discharge databases coded serious bacterial infections accurately

https://doi.org/10.1016/j.jclinepi.2006.07.011Get rights and content

Abstract

Objectives

We sought to test the ability of large health care utilization databases to accurately identify serious bacterial infections and opportunistic infections leading to hospital admission.

Study Design and Setting

We conducted a cross-sectional validation study using patients admitted to hospitals in the administrative database of the Department of Veterans Affairs, VISN 1, between 2001 and 2004. Detailed hospital chart abstraction protocols were developed to define a gold-standard diagnosis of serious bacterial infections and opportunistic infections. Hospital acquired infections were not considered.

Results

A total of 158 patients who were hospitalized for selected bacterial infections and 69 patients for opportunistic infections were identified using ICD-9 discharge diagnoses. The positive predictive values (PPV) of identifying specific bacterial infections that lead to hospital admissions varied between 100% and 66%. All conditions combined yielded a PPV of 80%. Once the gold-standard definition of bacterial conditions was broadened to hospital admissions due to any acute infectious condition, the PPV increased to 90%. Excluding systemic candidiasis, the average PPV for the selected opportunistic infections was 76%.

Conclusion

Our findings suggest that ICD-9 codes of selected serious infections from hospital discharge files can be used as substitutes for chart-based diagnoses.

Introduction

The use of immunomodulatory drugs for inflammatory and neoplastic conditions has increased and will continue to do so. For many conditions, such medications provide important new therapeutic advances, but this must be balanced against the potential for adverse effects, including serious infections. While common side effects may be detected in the setting of randomized controlled trials, less frequent events that may only occur in subsets of patients will not be reliably detected in premarketing trials [1], [2], [3], [4], [5]. The recently reported cases of tuberculosis in patients using TNF-α antagonists demonstrate the importance of thorough postmarketing surveillance [6].

Health care utilization databases represent an important tool for postmarketing surveillance. Such databases reflect routine care, are available with little delay, and often cover large populations for an efficient surveillance of rare outcomes [7]. In addition, patients not included in randomized trials but frequent users of immunomodulatory drugs, particularly elderly patients, are represented in such databases.

Biologic disease modifying antirheumatic drugs (DMARDs) for rheumatoid arthritis are now among the most widely selling biologic immunomodulating drugs [8], [9]. The use of TNF-α blocking agents increased sharply in the past 5 years and is expected to continue to expand. As noted above, there are a number of case reports that suggest that TNF-α blocking agents may increase the risk of serious bacterial and opportunistic infections [10], [11], [12], [13], [14], [15], [16], [17]. These infections come with a substantial burden of disease and are of considerable public health relevance even if the absolute number of cases may be small.

Since the expected absolute number of serious bacterial or opportunistic infections in subgroups of DMARD categories is too small to be reliably studied in randomized trials, the safety of these drugs is most efficiently studied in large health care utilization databases. We sought to test the accuracy of health care utilization databases to identify serious bacterial infections and opportunistic infections that lead to hospital admissions.

Section snippets

Data sources

The Department of Veterans Affairs (VA) administrative database of the New England region (VISN I) was used to identify patients with serious bacterial infections and opportunistic infections using ICD-9-CM diagnostic codes from hospital discharge files. The database contains administrative information on all hospitalizations within the New England VA system for administrative purposes. Data completeness for hospital admissions is thought to be close to 100% and coding of discharge diagnoses

Results

A total of 158 patients who were hospitalized for the selected bacterial infections were identified in the database and 69 patients for opportunistic infections. The average age was 69.6 years and 98.6% of patients were male. Table 1 lists the number of patients identified for each infectious condition.

The PPV of the identified ICD-9-CM codes in predicting our gold-standard definition of specific bacterial infections that lead to hospital admissions varied between 100% and 66%. All conditions

Discussion

Our study evaluated the PPVs of ICD-9-CM codes for serious bacterial and opportunistic infections leading to hospital admissions using health care utilization data. The condition-specific PPVs were in a range comparable to many other outcomes frequently used in database studies. If the gold-standard definitions were relaxed to any infectious condition that led to the index hospitalization, the PPV for bacterial infections increased to 90%, which is considered high and compares to many other

Acknowledgments

This study was funded by the Engalitcheff Arthritis Outcomes Initiative through the Arthritis Foundation.

References (20)

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

Cited by (138)

View all citing articles on Scopus
View full text