High-deductible health plans, with or without some form of health savings account attached to them, have become a permanent and growing feature of the health insurance landscape. A recent estimate was that 80% of the plans offered on the exchanges are high-deductible designs. Research to determine their cost and quality effects is important. A new report published at the National Bureau of Economic Research looked at several years of experience at employer-based CDHPs. (NBER Paper) The researchers used three years of data after introduction of a CDHP at 54 large employers, comparing those who adopted CDHPs with those who did not. The employers had a variety of CDHP strategies and designs. Not all the employers had a sole replacement strategy with the CDHP and they used different approaches to concurrently offering a health savings or health reimbursement account. The fundamental question is whether when patients reduce utilization and spending in response to higher cost-sharing, they are making good choices that maintain their health status or are they avoiding needed care, and eventually their health may deteriorate to the point that higher costs are incurred in future years. This paper focused on the question of verifying that there is sustained spending reduction, and exploring whether it seemed to be linked to any particular plan design features, and did not really address the quality issue.
The basic finding was that CDHPs were associated with about a 5% reduction in total health care spending in each of the three years following introduction, compared to what was spent at firms without CDHPs. If the effect was attributed solely to those employees who enrolled in the CDHP, it would be almost 15% per year, which is an astounding compounded rate of savings. Most of the decrease in spending was for outpatient care and drug use, with little change in emergency room or inpatient use or spending. They did not find statistically significant differences related to plan design, but it did appear that the spending reduction was greater when HSA or HRA account contributions by employers were smaller. Since there did not appear to be increases in ER or inpatient use, at least in this relatively short follow-up period, it could be that the decrease in use in other categories was not leading to worsening health. But it would be useful especially to have a more in-depth analysis of what drug use was being reduced, to ascertain if it was often for medications with a well-demonstrated value for chronic diseases. And to the extent that actual biometric and lab data on enrollees could be gathered to see if there were changes associated with plan design would also be useful.