Development of a novel, objective measure of health care-related financial burden for U.S. families with children

Abstract

Objective: To develop and validate a theoretically based and empirically driven objective measure of financial burden for U.S. families with children.

Data sources: The measure was developed using 149,021 families with children from the National Health Interview Survey, and it was validated using 18,488 families with children from the Medical Expenditure Panel Survey.

Study design: We estimated the marginal probability of unmet health care need due to cost using a bivariate tensor product spline for family income and out-of-pocket health care costs (OOPC; e.g., deductibles, copayments), while adjusting for confounders. Recursive partitioning was performed on these probabilities, as a function of income and OOPC, to establish thresholds demarcating levels of predicted risk.

Principal findings: We successfully generated a novel measure of financial burden with four categories that were associated with unmet need (vs. low burden: midlow OR: 1.93, 95 percent CI: 1.78-2.09; midhigh OR: 2.78, 95 percent CI: 2.49-3.10; high OR: 4.38, 95 percent CI: 3.99-4.80). The novel burden measure demonstrated significantly better model fit and less underestimation of financial burden compared to an existing measure (OOPC/income ≥ 10 percent).

Conclusion: The newly developed measure of financial burden establishes thresholds based on different combinations of family income and OOPC that can be applied in future studies of health care utilization and expenditures and in policy development and evaluation.

Keywords: Financial burden; access to health care; families with children; health care costs; measurement; unmet health care need.

Figure 1

Prediction of Unmet Need for Any Family Member Using a Bivariate Tensor Product Spline Notes. Visualization of the marginal effect generated with weighted regression utilizing a bivariate tensor product spline, adjusting for predisposing, need and enabling factors. The z-axis and shading on the surface component of this plot illustrates the gradation in the marginal probability of unmet need, with light gray representing a lower probability and black representing the highest probability. Measure of OOPC and income were adjusted for inflation to 2010 U.S. dollars based on the annual average Consumer Price Index (CPI, obtained from the U.S. Bureau of Labor Statistics).