Optimizing Medication Use for Cognitively Impaired Older Adults

Visiting Nurse Service of New York (VNSNY)

Project Description:

This program addressed the challenge of managing complex medication regimens for a particularly vulnerable population – community dwelling, cognitively impaired (CI) older adults with chronic illness. The program implemented and evaluated a multi-faceted, information technology (IT) based intervention designed to better support nurses, as well as cognitively impaired patients and their caregivers, in the challenging process of managing multiple medications in the context of multiple co-morbidities. VNSNY extended to the CI population a medication complexity technology intervention that has been developed for chronically ill older adults receiving health services at home. Specifically, the intervention uses an algorithm to alert the home health nurse at the point of service of patients who are at dual risk due to the complexity of his/her medication regimen and the patient’s cognitive impairment. Intervention clinicians were provided with decision support including 1) complex medication management recommendations that were integrated into the clinician’s visit documentation system and the electronic patient health record; and 2) education materials for the patient/caregiver to improve awareness and self-management practices. The medication complexity management intervention was assessed relative to usual care. Randomization occurred at the nurse level. The underlying premise of the program was: better medication management and adherence should lead to reduced ED visits and hospitalizations. Goals of this project were to:

  • Extend an IT intervention to improve management for CI patients in home health care with complex medication regimens
  • Examine the effects of the intervention on medication management practices of intervention home health care nurses compared to usual care
  • Examine the effects of the intervention on patient outcomes and service use.


To assess the effectiveness of the medication complexity management intervention, VNSNY used a randomized design whereby nurses of eligible CI patients were randomly assigned to intervention or control group. A total of 406 nurses who served 1,308 eligible patients were randomized. Data sources included: 1) the Outcomes Assessment and Information Set (OASIS) – a standardized home care data collection instrument used as part of usual care; 2) VNSNY’s electronic medications database; 3) patients’ electronic clinical records; 4) a caregiver survey conducted at a fixed point (~60 days) post-assignment to the program; and 5) administrative and claims data routinely collected by the agency. Uptake of the medication complexity intervention was assessed by tabulating the frequency that intervention nurses documented processes of care in the electronic decision support tool that was made available to them directly in the patient’s electronic clinical record. Outcome measures included: changes in the MRCI (medication complexity risk score) and two outcomes-based quality improvement (OBQI) measures (hospitalization and emergency department {ED} use) that are used for public reporting. Outcome measures for caregivers included: caregiver report of patient regimen adherence, caregiver report of processes of care, and caregiver report of medication management burden.


VNSNY evaluated intervention-nurse use of the medication complexity decision support tool and patient outcomes utilizing two approaches: 1) an intent-to-treat approach where data on the designated nurse and patient were included in the analysis without specification of the number of visits a nurse had with a patient, length of stay, or any other nurse or patient characteristic that might have affected exposure to or actual uptake of the intervention and 2) a sub-analysis of intervention nurses that actually used the medication complexity decision support tool.

  • Intervention Was Used Less than Expected. There was only moderate use of the medication complexity decision support tool with 60% of the intervention nurses using a component of it with at least one of their targeted patients – overall, affecting 40% of the patients. There were no reductions in the complexity of patients’ medication regimens for either randomized group.
  • Intent-To-Treat Analysis Indicated: Intervention Was Not More Effective than Usual Care. The unadjusted results demonstrated no differences between the usual care and intervention group in regard to ED visits or hospitalizations. Additional data was collected from 156 family caregivers of the targeted CI patients. No differences were found between the usual care and intervention targeted caregivers in regard to home care nurse teaching, caregivers’ report of adherence to the medication regimen, physician contacts to simplify regimen, or caregivers’ reported management burden.
  • Sub-analysis of Nurses that Actually Used the Intervention Indicated: Intervention was Associated with Reduced ER Use and Fewer Hospitalizations. Patient outcomes were compared between the 189 nurses that utilized the medication complexity management intervention and the 283 nurses that did not utilize the intervention. Patient outcomes were statistically different (at p < 0.05) between the two groups. Use of the intervention was associated with a lower rate of ER use (13% vs. 21%) and a lower hospitalization rate (18% vs. 27%).

Lessons Learned:

Need to Identify High Intervention Users. Program outcomes can be significantly impacted due to limited program implementation or unmotivated program staff. VNSNY found improvements in patient ER use and hospitalization rates among nursing staff that were most engaged in using the medication decision support tool. It should be noted that when an intervention is implemented and tested through a randomized controlled trial design (with an intent-to-treat analysis), it is often difficult to demonstrate statistically significant differences between intervention and control groups. VNSNY was able to show intervention effects only when the non-intent-to-treat analysis was utilized.

Need to Train Nurses in Use of the Intervention. An effective intervention has to be used in order to favorably impact patient outcomes. It is likely that additional training on the medication complexity management intervention and additional promotion among nurses would have increased the nurse usage rate of the intervention.