DETERMINANTS OF HOME HEALTH VISIT LENGTH - RESULTS OF A MULTISITE PROSPECTIVE-STUDY

OBJECTIVES. The authors (1) compare visit length across four categorie s of skilled nursing home health visits which reflect recent changes i n home health casemix-AIDS-related, hospice/terminal (HT), intravenous (IV) therapy, and maternal and child health (MCH)-with general adult medical/surgical (MS) visits and (2) identify factors influencing visi t length. METHODS. The study sites were 12 nonproprietary Massachusett s home health agencies (HHAs). Staff nurses collected data concurrentl y on a sample of visits they provided between December 1, 1992 and Nov ember 30, 1993. The visits were stratified by agency, time of year, an d visit category. The authors used analysis of variance to test for si gnificant differences across visit categories in Home Length of Visit (the number of minutes between when the nurse entered and left the hom e) (HLOV). The authors used multivariate regression analysis to develo p models identifying determinants of HLOV and adjusted R-2 to measure the explanatory power of partial models. RESULTS. In univariate analys is, the categories differed significantly from each other in length (P < 0.0001). HT visits were the longest (median visit length = 60, 80, and 59 minutes for HT Only visits, visits in both the HT and AIDS cate gories (HT/AIDS), and HT/IV visits, respectively). MS visits were the shortest (median = 30 minutes). The remaining categories were intermed iate in length (medians = 37 to 50 minutes). Almost half the variabili ty in HLOV was explained by the full multivariate regression model, wh ich includes all independent variables (adjusted R-2 = .4486; P < 0.00 01). Visit characteristics alone in a partial model explained 18% of t he variability in HLOV. Three other variable sub-groups-agency, client characteristics, and nursing workload-each explained about 15% of the variability in HLOV. Nursing activities performed during the visit ex plained 11%; several of these related to teaching, education, or asses sment. CONCLUSIONS. Accurate reimbursement reflecting casemix differen ces is important to protect the teaching, education, and assessment fu nctions of nurses; measure nurse productivity and allocate caseloads; maintain access to services for clients with greater needs; and avoid creating economic disincentives to the agencies that serve them. Payer s formulating prospective payment systems can adjust per visit reimbur sement rates to reflect differences in visit length by category and in corporate functional limitations, clinical instability, and case coord ination as classification variables. Developers of home health casemix systems can use factor analysis to improve the robustness of multivar iate models and include nursing workload in predicting visit length. H ome health agencies measuring productivity and caseload across complex client populations can classify visits into three groups-MS; HT; and AIDS, IV, and MCH-or use the regression results to develop more refine d predictors of visit length and nursing caseload.