H3IT: Home Healthcare, Hospice, and Information Technology Conference Washington – D.C., 2014

Developing a Tool to Support Decisions on Patient

Prioritization at Admission to Home Health Care

Maxim Topaz

1, 2

, Kathryn H. Bowles

1,3

illions of Americans are discharged from hospitals to home health every year and about third of them return

to hospitals. A signiﬁcant number of rehospitalizations (up to 60%) happen within the ﬁrst two weeks of

services. Early targeted allocation of services for patients who need them the most, have the potential to

decrease readmissions. Unfortunately, there is only fragmented evidence on factors that should be used to

identify high-risk patients in home health. This study aimed to (1) identify factors associated with priority for the ﬁrst home

health nursing visit and (2) to construct and validate a decision support tool for patient prioritization.

Methods: We recruited a geographically diverse convenience sample of nurses with expertise in care transitions and care

coordination to identify factors supporting home health care prioritization. This was a predictive study of home health visit

priority decisions made by 20 nurses for 519 older adults referred to home health. Variables included socio-demographics,

diagnosis, comorbid conditions, adverse events, medications, hospitalization in last 6 months, length of stay, learning ability,

self-rated health, depression, functional status, living arrangement, caregiver availability and ability and ﬁrst home health

visit priority decision. A combination of data mining and logistic regression models was used to construct and validate the

ﬁnal model.

Results: The model identiﬁed ﬁve factors associated with ﬁrst home health visit priority. A cut point for decisions on

low/medium versus high priority was derived with a sensitivity of 80% and speciﬁcity of 57.9%, area under receiver operator

curve (ROC) 75.9%. Nurses were more likely to prioritize patients who had wounds (odds ratio [OR]=1.88), comorbid con-

dition of depression (OR=1.73), limitation in current toileting status (OR= 2.02), higher numbers of medications (increase

in OR for each medication =1.04) and comorbid conditions (increase in OR for each condition =1.04).

Discussion: This study developed one of the ﬁrst clinical decision support tools for home health, the PREVENT- Priority

for Home Health Visit Tool. Further work is needed to improve the speciﬁcity and generalizability of the tool, implement an

electronic version and test its eﬀects on patient outcomes.

University of Pennsylvania, Scho ol of Nursing

Brigham and WomenÕs health hospital

Visiting Nurse Service of New York