Identifying workers at risk for high health care expenditures: a short questionnaire

The objective of this study was to create and measure the predictive accuracy of a brief questionnaire for screening new workers to identify those at increased risk for generating high health care insurance expenditures during the following year. Such an instrument could help health plans and providers intervene to mitigate the health risks of identified high-risk workers. We mailed a 53-item questionnaire to members of a "derivation cohort" (adult food processing workers, n = 15496) and obtained records of the eligible respondents' health insurance expenditures during the following year. Using multiple linear regression, we identified eight of the questions that predicted future expenditures most accurately, and created a formula to predict total expenditures from answers to these questions. To validate the formula's predictive accuracy, we used the eight-item questionnaire to survey two "validation cohorts" (transportation workers, n = 7445; and their dependents, n = 5562), inserted responses into the scoring formula, classified respondents into high-risk (top 10%) or low-risk (lower 90%) groupings, and then compared health insurance expenditures generated by the high- and low-risk groups during the following year. In the derivation cohort, age, sex, regular use of medications, frequent visits to physicians, and having arthritis, diabetes, cancer, or high cholesterol predicted future health care expenditures. In the worker and dependent validation cohorts, the respondents classified by the formula as high-risk generated insurance expenditures during the following year that were 2.4 and 1.8 times greater than those generated by the members of the low-risk groups (p < 0.001). An eight-item questionnaire and its scoring formula can identify high-risk groups of workers that will generate high health care expenditures during the following year. Healthcare organizations could use this questionnaire to help target new workers for care management and disease management interventions.