Method: Stated-preferences were assessed using conjoint analysis across five CBHI attributes: premium price per capita, premium payment frequency, coverage of hospital costs, coverage of travel and meal costs, and frequency of communication with the insurer. Preferences were elicited through forced-choice paired comparison choice tasks designed based on D-efficiency. Data were analyzed using a conditional logistic regression, with preference differences examined between small (1-4 family members) and large (more than 5 member) households. The model estimated combined effect sizes and examined differences by household size via the likelihood ratio test and the Swait and Louviere test for differences in scale.
Result: 160 households were surveyed across 10 randomly selected villages in Northwest Cambodia where a CBHI scheme operates. Increased insurance premium price was associated with disutility (OR 0.95, p<0.01), while an increase in utility was noted for higher hospital fee coverage (OR 10.58, p<0.01), greater coverage of travel and meal costs (OR 4.08, p<0.01), and more frequent communication with the insurer (OR 1.33, p<0.01). While the magnitude of the preference for hospital fee coverage appears larger for the large household group (OR 14.14) than the small household group (OR 8.58), Wald tests indicate that no statistically significant differences exist for the order or effect sizes of all attributes including hospital fee coverage (p=0.06). Differences in scale were observed (p<0.05), which would make the preference difference by household size even more negligible.
Conclusion: Differences in stated-preferences may be due to differences in scale rather than true variations in preference. Coverage of hospital fee, travel and meal costs are given significant weight in CBHI enrollment decisions regardless of household size. It is essential to understand how community members make decisions about health insurance as many low- and middle-income countries look to introduce or scale-up health insurance plans towards universal health coverage. Policy makers should learn from preference data to guide benefit package design that aligns with people’s decision making practice in order to increase access to healthcare and prevent poverty due to illness.