Original Articles
Searching multiple clinical information systems for longer time periods found more prevalent cases of asthma

https://doi.org/10.1016/j.jclinepi.2003.08.014Get rights and content

Abstract

Objective

The development of a reliable asthma registry is an important first step for conducting population-based asthma disease management. This study developed a computerized algorithm for defining prevalent asthma, identified operational difficulties, and summarized data on asthma prevalence in the study population.

Study design and setting

As part of a study of the incidence of occupational asthma, we used the electronic databases of a large health maintenance organization to develop a computerized algorithm for defining prevalent asthma and validated it against chart review. The predictive values of eight health care utilization profiles were validated by chart review to establish the algorithm.

Results

The 1-year treated prevalence of asthma was 4.1% among members aged 15–55; the pharmacy database identified 61% of cases, and the outpatient care database 66%. Extending the outpatient care window from 1 year to 2 years increased estimated prevalence to 5.3%, with 81% now found in the outpatient care database.

Conclusion

This analysis illustrates the benefit of using multiple databases for more accurate enumeration of cases and the impact of extending the search in time. These results are useful for researchers who can use such databases in selecting algorithms to define and identify asthma for their own purposes.

Introduction

Interest in population-based asthma management, particularly on the part of managed care organizations in the United States, is growing. This interest has been kindled both by the increasing prevalence of asthma and by the recognition that asthma is a chronic condition that, for many, requires ongoing daily controller medication to maintain good disease control [1]. Despite the existence of national treatment guidelines [1], many patients with asthma continue to be undertreated, and are therefore at greater risk for acute exacerbations that result in reduced quality of life, missed work or school, and expensive health care utilization [2], [3], [4].

Central to the disease management process is the development of some form of asthma registry, ideally containing information on all members with asthma along with some indication of disease severity and current level of control [2]. The member information enables accurate tracking of the impact of disease management efforts, while the indication of severity and control enables medical providers to intervene most intensively with those most in need of help.

Implementation of such registries has historically been hampered by the lack of administrative databases containing diagnostic-specific encounter data. Medication databases are highly sensitive for asthma, but lack specificity, particularly if one is interested in developing a comprehensive registry [5]. Increasingly, however, large managed care organizations are adopting clinical information systems that provide diagnostic-specific outpatient utilization data. These databases offer great promise for the development of disease registries, although their structure is often complex and their proper use is not always clear. Such electronic databases are also of interest to researchers interested in characterizing the clinical epidemiology of asthma [6], [7], [8], [9], [10]. As part of a National Institute for Occupational Safety and Health (NIOSH)-funded study to explore the incidence of occupational asthma, we used the computerized databases of a large health maintenance organization to develop a computerized algorithm for defining prevalent asthma and validated it against chart review. This article summarizes our methodology, discusses operational difficulties, and also summarizes data on asthma prevalence for the health plan. We hope that our results will aid health plan managers, epidemiologists, and health services reseachers attempting to use similar databases in the future. As more individuals gain access to such databases, we believe the need for studies such as this one will increase. Also, for those with access to more limited electronic databases, such as pharmacy records, our analysis illustrates some of the potential limitations relative to more comprehensive systems.

Section snippets

Research setting and target population

Kaiser Permanente Northwest Division (KPNW) is a large, group-model health maintenance organization centered in Portland, OR. KPNW provides comprehensive, prepaid health care service, including access to inpatient, outpatient, and emergency department (ED) services, to its approximately 430,000 members. The demographic and socioeconomic characteristics of its members correspond closely to those of the area population as a whole [11], and no evidence exists of a systematic selection of healthy

Results

Table 2 summarizes the results of the chart review process. Our a priori most-probable profile (#1), accounting for 46% (4465/9723) of all possible asthma patients, provided excellent predictive value (PV), with 21 of 22 (95%) of the charts reviewed classified as probable asthma. The next most common profile (#2), a single nonurgent outpatient asthma visit in EPIC with no acute care contacts and at most three beta-agonist dispensings, had a 90% predictive value (17 of 19 charts classified as

Discussion

The increasing use of diagnostic-specific clinical information systems by large managed care organizations is expanding the opportunities for conducting true population-based asthma health services research. Traditionally, epidemiology researchers have had to rely on pharmacy-based algorithms or on questionnaire-based surveys (perhaps augmented by clinical evaluation). The former were limited by the lack of specificity of pharmacy-based algorithms for defining mild asthma, while the latter were

Acknowledgments

This work was supported by contract #U60/CCU916057 from the National Institue for Occupational Safety and Health (NIOSH).

References (21)

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

Cited by (19)

  • Identification of asthma cases in Japan using health insurance claims data: Positive and negative predictive values of proposed discrimination criteria: A single-center study

    2023, Allergology International
    Citation Excerpt :

    In contrast, studies by Vollmer et al. and Kawasumi et al. demonstrated increased PPVs for criteria defined by disease codes and claims data for asthma control medications.18,19 However, definitions or criteria based on both disease codes for asthma and the claims data for asthma control medication are still problematic because they may miss patients with asthma who utilize only emergency departments or urgent care services.18 Considering this possibility, we included claims data for the treatment of asthma exacerbations (corticosteroid drip or β2-agonist nebulizer) in our criteria.10,20

  • Comparison and validation of algorithms for asthma diagnosis in an electronic medical record system

    2022, Annals of Allergy, Asthma and Immunology
    Citation Excerpt :

    The Study of Asthma, Genes and the Environment cohort evaluated various rule-based algorithms in 723 children and validated with pediatric allergist diagnosis of asthma, methacholine challenge, and skin tests, with their highest performing algorithm having a sensitivity of 80% and specificity of 89%.17 Vollmer et al18 detailed their means of validation; however, they have classified asthma diagnosis as possible and probable, making direct comparisons to other algorithms difficult. Our finding that asthma diagnosis codes restricted to encounter had the most balanced performance is similar to the results of a study of a Canadian primary care EMR, which included 398 patients enhanced for the presence of other respiratory conditions.

  • Improved incidence estimates from linked vs. stand-alone electronic health records

    2016, Journal of Clinical Epidemiology
    Citation Excerpt :

    For example, linked primary and secondary care data provide more complete information on outcomes, enriched data on covariates such as patients' medical and therapeutic histories, and accurate timing of events such as hospitalizations. The value of linked over stand-alone data has been investigated for conditions such as cardiovascular events, asthma, diabetes, and upper gastrointestinal bleeding [1–4]. However, the potential benefits of linked data for examining the burden of important infectious diseases are unclear.

  • Biomedical Informatics Applications for Asthma Care: A Systematic Review

    2006, Journal of the American Medical Informatics Association
  • Incidence of work-related asthma in members of a health maintenance organization

    2005, Journal of Occupational and Environmental Medicine
View all citing articles on Scopus
View full text