Avoidable morbidity in infants. A classification based on diagnoses in administrative databases.

Abstract

OBJECTIVES: A hierarchical classification for avoidable morbidity in infants was developed based on a conceptual model for causes of morbidity. Experts rated the impact of risk factors and health services on diseases coded according to the International Classification of Diseases, 9th Revision, Classification Modification (ICD-9-CM). An etiologic framework was chosen for the classification because knowledge of etiology often suggests strategies for prevention. Causes of morbidity that cluster on the basis of similar risk factors might be avoided using similar strategies. METHODS: Diseases (346 different diagnoses) were rated by 16 general pediatricians; 12 attributes were considered, including the impact on disease occurrence and on severity of five risk factors, preventive health services, and medical treatment. Raters evaluated the impact of health services, constitutional risk factors, and environmental risk factors without regard for service site (eg, inpatient, emergency department, primary care office). Environmental risk factors categories, including family, social, and physical environments, were rated separately. The impact of health services was rated on prevention, treatment, and complications of care. RESULTS: Only ratings indicating that the impact of a risk factor category was substantial were used for the final classification of 275 diagnoses. Consistent with the multifactorial etiology of many diseases, many diagnoses had ratings indicating substantial impact of multiple risk factors. Five mutually exclusive clusters were derived from the 12 ratings based on factor analysis and recognized strategies for prevention. Ordered by level of avoidability, these clusters were termed vaccine-preventable, health-care quality indicators, environmental, environmental/constitutional, and constitutional. CONCLUSIONS: The usefulness of this classification for policy-oriented epidemiologic and health services research is grounded in the premise that prevention is the cardinal objective of child health policy. Cluster-specific hospitalization rates, ie, rates aggregated for all diagnoses falling in a cluster, might be used for allocating resources to interventions directed at environmental or health service determinants of morbidity. Widespread use of ICD-9-CM codes in hospital discharge and ambulatory databases suggests many potential applications for this classification of morbidity burden in population groups.