Findings from this study suggest that small, but statistically significant reductions in exposure to HRME overall defined by the HEDIS criteria occurred between 2004 and 2006 for older VA patients. While the reductions in overall exposure were relatively small during this time period (13.1% in FY04 to 12.3% in FY06), exposure is markedly reduced compared to our previous assessment of 19.6% in FY00. This is quite similar to the rate of 12.9% reported by Albert and colleagues from a previous national sample of retirees from one company in 2003-2005.26
The rates of exposure in this current study were considerably lower than those reported by the National Committee on Quality Assurance for 2006 among Medicare enrollees, where 23.1% of individuals meeting criteria for the denominator in 2006 had one or more HEDIS HRME exposure.15
VA's leadership in geriatric care, the active role of pharmacists in VA, and VA formulary management may have contributed to the reduction. VA created its One-National Formulary in 2002 by freezing formularies at the facility and regional level, then creating a uniform formulary which resulted in excluding drugs that had previously been on a number of regional formularies such as propoxyphene.
Examination of change over time by individual drugs/classes revealed some reduction in most drug classes, stability in some others, and increases in nitrofurantoin. Reductions in use of skeletal muscle relaxants, psychotropic drugs, and opioid pain relievers is notable as recent studies have demonstrated that the use of these medications increase the risk of falls and fractures in older adults.27-29
One possible explanation for the increase in nitrofurantoin use is related to increasing resistance to common urinary tract pathogens such as e-coli with common antibiotics (e.g., ciprofloxacin trimethoprim/sulfamethoxazole). One invitro study found that nitrofurantoin was effective in killing e-coli isolates in 98.1% of those with trimethoprim/sulfamethoxazole resistance and 89.6% of those with ciprofloxacin resistance.30
Unfortunately, invitro testing does not translate to nitrofurantoin being effective in older adults. The use nitrofurantoin, a primarily renally cleared medication, should be avoided in those with estimated creatinine clearances less than 60ml/min as insufficient concentrations reach the bladder to be capable to kill bacteria such as e-coli.31,32
A recent study of veterans residing in a VA community living center found that this agent was in the top four suboptimally prescribed medications.33
Of concern is an increased risk for serious adverse drug events with nitrofurantoin that include chronic, sub-acute, or acute pulmonary hypersensitivity reactions and peripheral neuropathy.
This study also adds to our understanding of risk factors for potentially inappropriate prescribing. Prior studies have either examined a single drug such as propoxyphene or potentially inappropriate drugs in the Beers or HEDIS criteria as a single entity.7,14-18,34,35
Neither the aforementioned Albert nor the NCQA studies examined risk factors for the use of high risk medications.15,26
Consistent with our prior report and other studies examining exposure to potentially inappropriate medications, we found that Whites, women, and those with more medications were more likely to be exposed. Examination of the four most commonly prescribed HEDIS HRME groups suggests that findings from studies of HEDIS HRME as a whole provide insufficient insight into this problem. In particular, the effects of race and psychiatric comorbidity and primary care utilization depend upon the type of potentially inappropriate medication.
With regard to race, African Americans were less likely to have exposure to suboptimal opioid and psychotropic medications than were whites, and Hispanics were also significantly less likely than whites to have exposure to opioid medications. The finding for African Americans is consistent with literature finding lower use of psychotropic medications and analgesics in blacks compared to whites.36-38
The finding for Hispanics is less clear, in part because many previous studies have not distinguished between blacks and Hispanics, but rather evaluated them as “nonwhites”.
It is interesting to note that those with multiple psychiatric comorbidities had a lower risk of being prescribed high risk opioids (e.g. propoxyphene). This finding may result from clinicians being less likely to prescribe opioids for pain in patients receiving psychotropic medications for psychiatric comorbidities due to concern of that total CNS medication burden (e.g., opioids, benzodiazepines, antidepressants, antipsychotics) increase the risk of falls in older adults.39
There are a number of potential limitations that should be noted
First, restriction of our assessment to individuals who received VA care between FY03-06 may lead to selection bias. Although this was necessary in order to understand change in a consistent population of patients, this may bias the results since individuals who are sicker and die during this 3 year period are not represented in the findings. However, examination of the entire population revealed similar rates of exposure and trends overall and for HEDIS HRME drug groups (range 14.1% in FY04 to 12.6% in FY06) and predictors of exposure. Second, our assessment was restricted to medications received within the VA. It is possible that some HEDIS HRME were purchased outside the VA and thus our assessment may be conservative. One potential problem is with medications that can be purchased over the counter (OTC) such as diphenhydramine. Substantial variation in the relationship of copayment status on HEDIS HRME exposure for antihistamines would support the idea that reduced risk for exposure among those with required copayments was differentially affected by OTC medications. While small variation did exist, the direction and magnitude of the copayment variable was similar across drug classes. A second issue regarding use of medications received from the VA is that our findings may be affected by implementation of Medicare Part D in January, 2006.40 Because we did not see a sudden, marked decrease in the average number of prescriptions per patient or marked decreases in exposure between 2005-2006 (), it is unlikely that our findings for FY06 were affected substantially. Moreover, our assessment occurred using data from a timeframe before the implementation of the HEDIS HRME measure. Because the Beers criteria from which the HEDIS HRME measure was derived has been in existence in various forms for nearly a decade, the time period examined for this assessment is not unreasonable. Moreover, this study provides a foundation for subsequent study of change in exposure by chronicling the years up to and including the first year of HEDIS HRME implementation. We cannot rule out residual confounding due to potential important factors for which information was not available (e.g., smoking) and HEDIS HRME. Because this study provides information on change in prescribing on a national sample of older primarily male VA patients, it does not reflect prescribing in non-VA settings. Finally, since propoxyphene was removed from the market, we expect that rates of HEDIS HRME exposure will be significantly reduced nationally. This would follow a consistent trend in the VA where rates were reduced considerably due to formulary restrictions. However, since opioids rank third among the most common high risk drug classes, high risk medication exposure continues to be of concern.