The effects of predetermined payment rates for Medicare home healthcare

Health Services Research, Oct, 1997 by Randall Brown, Barbara Phillips, Christine Bishop, Craig Thornton, Grant Ritter, Amy Klein, Peter Schochet, Kathleen Skwara

Data on control variables were also drawn from several sources. Area characteristics (such as number of physicians per 1,000 area residents and nursing home beds per 1,000 elderly people) were obtained from the Area Resource File. Agency characteristics, such as for-profit status, auspice (whether private freestanding, visiting nurses association, or hospital-based), and location were obtained from the demonstration implementation contractor, Abt Associates, Inc. The patient-level analyses controlled for both these variables and for patient characteristics at admission that might affect outcomes. These patient characteristics were drawn from four sources: (1) a demonstration patient intake form that provided information on patients' functioning, diagnoses, type of care needed, referral source, prognosis at the time of admission to home health, and availability of informal care; (2) HCFA's plan of treatment forms (485/486s), which provided data on treatments planned at the time of admission; (3) prior Medicare claims, which provided data on patients' use of Medicare-covered services in the 12 months preceding admission; and (4) Medicare's master beneficiary file, which contained basic demographic characteristics (e.g., age, sex, and race). Each episode of home health was treated as an independent observation.

Because the patient-level data were available for essentially all episodes of care that agencies delivered during the three-year demonstration period, sample sizes were very large for the claims-based analyses. Over 88,000 home health episodes were included in the analysis.

DEMONSTRATION EFFECTS

Evidence of No Cost Savings. Our fixed-effects models of costs per visit find that the demonstration had no discernible effect on cost for any visit type. The results, displayed in Table 1, indicate that cost per skilled nursing visit for treatment group agencies increased from an average of about $81 to about $92 between the predemonstration and demonstration periods, or about 14.5 percent (an average annual increase of 4.6 percent between the two periods). Control group agencies' average costs were somewhat lower than treatment group agencies' costs prior to the demonstration but grew by a similar amount. The estimated treatment impact, $1.67, was positive, small, and not significantly different from zero. Similarly, estimated effects on the cost of home health aide visits and therapy visits were small and not [TABULAR DATA FOR TABLE 1 OMITTED] significantly different from zero. The estimated effect on the costs of medical social worker visits was large and in the expected direction, but statistically insignificant. The small number of such visits supplied by most agencies leads to wide variation in the cost of these visits. See Bishop, Ritter, Skwara, et al. (1995) for further results and discussion.

No Effects on Agency Volume. One concern about the demonstration was that, regardless of the effect on costs per visit, overall costs to HCFA might rise under prospective rate setting because the opportunity to earn profits might lead agencies to increase the number of visits they provided by more than they would have under cost-based reimbursement. Growth in agency volume that resulted from demonstration agencies attracting patients away from other agencies would not necessarily have increased overall costs to HCFA. If the increased visits resulted from agencies increasing the number of visits per episode beyond what they would have provided had compensation been on a traditional cost basis, however, costs to HCFA might have risen.