Medication errors are common in nursing homes around the country and associated with significant morbidity and mortality. As a group, the residents of VA managed nursing homes (henceforth referred to as Community Living Centers or CLCs) represent some of the frailest members of the VA health care system and are at particularly high risk for medication errors and adverse drug events. Conservatively, adverse drug events result in 3.5 billion dollars in extra health care costs to the VHA, Medicare, and other health funding agencies. This figure does not take into account the tremendous financial burden associated with lost patient wages and legal costs to health care institutions.
The overall aim of this career development proposal was to study medication errors in VA CLCs and test the feasibility needed to develop a future measurement toolkit of 8-10 quality indicators related to medication errors. It is anticipated that this toolkit might one day be used by facilities to benchmark their performance related to their peers for commonly encountered medication errors. In addition, the toolkit would identify correlates of high and low performing facilities and provide guidance for future quality improvement initiatives using Continuous Quality Improvement (CQI) methodologies. The proposed aims of this CDA-II application included:
1)To determine the frequency of common drug errors within VA CLCs and identify any patient, facility, and organizational factors correlated with them. Specifically, VA retrospective data will be used to:
a.Assess the rates of prescribing within CLCs for the "Always Avoid" and "Rarely Appropriate" 2006 Healthcare Effectiveness Data and Information Set (HEDIS) list of inappropriate medications.
b.To determine the degree of laboratory monitoring that occurs in VA CLCs for Veteran's following Angiotension (ACE) Inhibitor initiation.
c.To evaluate the frequency of na ve initiation of long acting opioids in VA CLCs.
2)To conduct a series of case studies with high and low performers consisting of qualitative interviews with key stakeholders and structured observation of staff as they dispense medications.
VHA administrative data was obtained for calendar years 2003-2010 for all Veterans who resided in a VA CLC. NH level data was obtained from the VA Minimum Data Set. Relevant Pharmacy and Laboratory Data was obtained from the Decision Support System Database (DSS). All data were combined using a VA version of the Residential History File methodology which establishes a longitudinal daily history of a Veterans' health services location. The RHF was validated against the California Death Statistical Master File on a cohort of Veterans who used VA NHs and died in California resulting in >90% concordance.
In terms of qualitative methodology for Aim 2, a panel of 3 interviewers identified at least 6 NH facilities in each tercile of performance based on the utilization of inappropriate HEDIS medications. A structured interview was conducted highlighting policies and procedures related to medication errors that might be associated with high and low performing facilities.
(Aim1A) Between 2004 and 2009, 28,970 of 176,168 (mean 16.4 9.5%) veterans admitted to a CLC received at least one HEDIS high-risk medication while in the facility. Over time, the number of veterans receiving high risk medications decreased. In 2004, the average rate of elderly VA CLC residents receiving at least one HEDIS high-risk medication was 23.9 10.0%; by 2009, the average rate had decreased to 10.0 6.6%. Despite the lower rate, significant facility variability remained. In 2009, rates of use ranged from a minimum of 0% at one facility to a maximum of 44.4%; the 2004 rates of HEDIS criteria medication use ranged from a low of 6.7% to a high of 63.3%. (Aim 1C) Between 2004 and 2009 the number of episodes of na ve opioid initiation within CLCs decreased from 1,784 to 1,376 for a total of 9,348 episodes over the 6 year period among 8,818 patients. The percent (n) of episodes that began with a LOA was 10.6% (1,010), however that rate decreased over time from 15.2% (271) in 2004 to 6.1% (133) in 2009. Over this same time period, reported daily pain among the cohort decreased from 66.1% (1,179) in 2004 to 64.2% (874) in 2009, with an overall rate of 66.2% (6,185).
This study successfully tested the feasibility of the development of medication quality indicators in VA CLCs. We identified high and low performing facilities for the utilization of Potentially Inappropriate HEDIS medications and tracked performance over a period of time from 2004 to 2009. We also identified high and low performers in terms of ACEI monitoring and identified the rate of potentially dangerous na ve long acting opioid initiation. Qualitative interviews with high and low performers will shed additional light on practices and policies in place at VA CLCs that might portend one or the other outcome. This study also partially supported the development, validation, and utilization of the Residential History File within the VA, an important data infrastructure that will continue to provide relevance to the Providence VAMC HSR&D Center of Innovation on Long-Term Services and Supports and the VACO supported Geriatrics and Extended Care Data Analysis Center.
External Links for this Project
None at this time.