Health Care Utilization and Cost Among Children With Asthma Who Were Enrolled in a Health Maintenance Organization

DISCUSSION

During 1992, the average child with asthma at GHC made approximately six nonurgent outpatient visits, filled 12 prescriptions, made 1.5 urgent care visits, and spent .23 days in the hospital (Table2). Health care for children with asthma at GHC cost 88% more per capita than for all children who were continuously enrolled and used services during that year. This represented approximately $2 million per year in this HMO of 380 000 enrollees. Asthma care accounted for 30 to 45% of all services received by children with asthma. The preponderance of the marginal cost of asthma was for services which were specifically for asthma. Children with asthma had 25% higher use of nonasthma services than the general population. Two-thirds of the cost of asthma care were attributable to nonurgent outpatient care and prescriptions; only one third was attributable to urgent care and hospitalizations. We believe that these results are generalizable to other similar HMO populations. Of course, health care administrators and policy makers should not extrapolate our findings to entire populations of enrolled children without taking into account the fraction of their population who are nonusers of health care.

The total cost of asthma-related health care in our population was comparable to the total charges reported by Stempel and colleagues for children with asthma in a recent study of four managed-care organizations ($389.57 in our study, vs $472.02 in theirs).²¹ The distribution of cost for asthma-related care was similar with the exception that asthma medications accounted for a larger proportion of costs in the study by Stempel and colleagues (21% vs 38%) probably reflecting that study's use of two or more asthma prescriptions as an inclusion criteria.²¹ The same study reported an increase in ED and inpatient care among adolescent girls (compared with males, young girls and adult women), and although we found similar trends in our data, they did not approach statistical significance (data not shown).

We believe our study has five methodologic advantages over earlier studies:

• It is population-based and measures utilization based on a defined denominator of children with asthma.

• We considered the entire spectrum of asthma care (nonurgent outpatient care, pharmacy services, urgent care, and hospitalizations) and we were able to measure the relative contribution of these categories of services to overall utilization and cost.

• We chose to examine all health care costs and utilization rather than just asthma-related costs. This is important because most health care utilization by children with asthma was for nonasthma services.

• Our estimates are based on true costs rather than charges. The problems with using charges as a proxy for costs have been well documented.²⁸ These concerns are heightened in a marketplace which places payers at a competitive advantage with respect to providers, resulting in cost estimates which reflect negotiated discounts that drive a wedge between charges and costs. GHC's cost accounting system allows us to study cost estimates that directly reflect the health care system's use of resources.

• Our subjects received care in an integrated delivery system, thus optimizing data capture and comparability across the study cohort.

There are four main limitations of this study. First, the demographics of our study population represents fewer nonwhite patients and a higher socioeconomic status than other metropolitan areas. However, it is relatively representative of the Puget Sound employed population.²⁷ Secondly, GHC is an established nonprofit HMO with a reputation for clinical excellence among HMOs. Although it could be argued that this makes the utilization patterns of its enrollees atypical, there were no coordinated changes in asthma management underway during the study period and thus we would argue that these results are generalizable to other similarly structured HMOs. The third limitation is that these data were from 1992, the first year in which diagnoses for ambulatory visits were recorded on Group Health's automated data. For this reason, we were concerned about possible misclassification of our subjects. In fact, chart review revealed that 92% of children with at least one an automated diagnosis of asthma had a clinical course highly consistent with asthma during the study period. Lastly, we were unable to measure out-of-pocket expenditures or control for differences in health care coverage, including copays. However, we were unable to find any substantial differences in asthma prevalence among study subjects in different employer groups and in groups of varying sizes. This suggests that it is unlikely that differences in coverage account for the observed differences in utilization between children with and without asthma.

The implications from our findings are threefold:

1. The impact of asthma on health care utilization was primarily on asthma-related services, although there was a 25% increase in nonasthma care among children with asthma.Furthermore, nonasthma care accounted for two-thirds of all health care received by children with asthma. For these reasons, we recommend that cost-effectiveness evaluations of asthma interventions take into account all health care utilization and cost, not only that which is coded as asthma. The fact that children with asthma have higher health care costs for nonasthma conditions is not surprising. Reasons for this may include higher use of ambulatory care and medications for conditions that can act synergistically with asthma (such as allergic rhinitis, sinusitis, otitis, respiratory infections; data not shown), higher health care use in general, and misclassification of asthma-related encounters as “non-asthma.”

2. Urgent care and hospitalization are less prominent in the cost distribution than might be suggested by national data. Our data differ substantially from national data on asthma-related health care costs which have shown ED and hospital costs to be the driving force behind national asthma expenditures. Note that, unlike national studies, we included urgent care and ED visits in the same category. In an analysis of 1985 data from a variety of national sources,²⁰ hospitalization and ED expenses represented 78% of all asthma costs excluding medications (age-specific medication data were not available in that study). In our population, the comparable figure would be 47% (ie, hospitalization and urgent care asthma-related costs as a fraction of all inpatient and outpatient asthma-related care excluding medications). Similarly, analysis of the 1987 National Medical Expenditure Survey suggests that hospitalization and ED expenses accounted for 61% of total costs for asthma (unpublished data), compared with 37% in our study (ie, hospitalizations and urgent care asthma-related costs as a fraction of all asthma-related costs). Although such comparisons are flawed by the somewhat different definitions of asthma and disparate data collection methods used, the predominance of ED and hospital charges in the national data is in striking contrast to the predominance of nonurgent outpatient and pharmacy costs in our staff-model HMO.

   • One explanation for the difference between our results and national data could be that GHC enrollees have better access to preventive asthma care than the patients in these national samples, resulting in fewer acute exacerbations requiring ED visits or hospitalizations. It is also important to bear in mind that African-Americans, who have been shown to bear a disproportionate burden of asthma morbidity, are under represented in the GHC (and western Washington) populations.

   • An alternative explanation could be that primary care providers in this HMO may be playing a large role in acute asthma care, perhaps as a result of an organizational emphasis on primary care as well as various triage mechanisms that discourage patients' use of EDs and urgent care clinics. (Acute asthma visits to primary care providers are difficult to distinguish from chronic asthma care using our automated data.) If this were true, then measuring urgent visits and hospitalizations for asthma would seriously underestimate the rate of acute asthma exacerbations in settings such as ours.

   • Furthermore, our findings suggest that although prevention of urgent care visits and hospitalizations for asthma remains an important goal, it may not necessarily lead to a large cost reduction for health care delivery systems in which primary care providers already provide such a large proportion of asthma care. In contrast, in populations with poor access to care and fragmented asthma services, there is likely to be a greater potential for cost savings through interventions which reduce the need for acute asthma care. Accurate measurement of acute asthma visits to primary care may be important for future studies of health outcomes and cost in managed-care settings such as ours. Asthma interventions aimed at reducing morbidity and cost in managed-care environments will benefit from realistic expectations about potential cost savings in asthma care. These expectations must be based on a setting-specific understanding of the patterns of health care utilization for children with asthma.

3. The marginal cost method has potential advantages over the unadjusted cost difference estimates. Both marginal cost and unadjusted cost methods include the cost attributed specifically to asthma services as well as the increase in costs of nonasthma care (relative to children without asthma) incurred by children with asthma. However, the marginal cost approach uses multivariate techniques which allow us to control for patient demographics and comorbidity. This approach may be particularly useful in clinical trials or observational studies where covariates may differ substantially between study groups.

In conclusion, children with asthma in this HMO had substantially greater health care cost and utilization than the general population of enrolled children, with the bulk of the difference being due to asthma care. Compared with national estimates, hospitalizations and urgent care visits represented a smaller portion of asthma-related costs. We are currently examining the populations of children with asthma at GHC in an attempt to characterize those children who contribute most to health care utilization and cost. Replicating this study in other settings would further enhance our understanding of health care utilization and cost for children with asthma, inform the improvement of asthma management and guide rational cost-reduction efforts. We recommend that future cost evaluations of asthma interventions measure total costs (not just asthma-related costs) as well as measuring all four categories of service discussed here, and that researchers consider employing a marginal cost approach.