Association Between Adherence to Diuretic Therapy and Health Care Utilization in Patients With Heart Failure

Michelle A. Chui, Pharm.D., Ph.D., Melissa Deer, B.S., Susan J. Bennett, D.N.S., Wanzhu Tu, Ph.D., Stacey Oury, B.S., D. Craig Brater, M.D., Michael D. Murray, Pharm.D., M.P.H.


Pharmacotherapy. 2003;23(3) 

In This Article


Daily diuretic adherence in this small sample from an inner-city population with heart failure was variable and was related to years of education. The finding that education predicts taking adherence is not surprising, as knowledge predicts adherence and cognitive function, and adherence with therapy is a deliberate cognitive decision. A patient with more education would have the cognitive ability to understand the importance of taking prescribed drugs. This is consistent with research indicating that persons with lower levels of education were likely to be noncompliant.[15]

Adherence to treatment instructions has a major impact on health care outcomes and the cost of health care.[7,16,17,18,19,20,21,22,23,24] Therefore, it would seem likely that taking adherence would be inversely associated with health care utilization; that is, it would lead to fewer visits to the hospital or emergency department for disease exacerbation. However, in our study neither correlations nor regressions with taking adherence to diuretics were significant. In contrast, scheduled adherence was significantly associated with cardiovascular hospitalizations, suggesting increased morbidity among patients whose timing of diuretic administration is erratic. In other words, patients taking a greater proportion of their diuretic drug on schedule would be at reduced risk of cardiovascular hospitalization (p=0.0006). Our data also showed that better scheduled adherence significantly reduced heart failure-related hospitalizations (p=0.0444).

The importance of scheduling diuretic administration might be explained by the pharmacokinetic and pharmacodynamic properties of furosemide, the diuretic taken by 88% of our patients. Response to furosemide and other loop diuretics is determined by the amount of diuretic delivered to its urinary site of action and by the sensitivity of nephrons to the drug.[25] The former is a function of the pharmacokinetics and the latter of pharmacodynamics. The pharmacodynamics of all loop diuretics are similar, with the same maximum effects but with different potencies accounting for the different dosages that are administered clinically. The pharmacodynamic relationship is characterized by a steep sigmoidal curve. Thus loop diuretics have a narrow therapeutic range wherein small changes in the amount of drug reaching the site of action can equate to times when response is inadequate. The converse does not occur; namely, when more than sufficient drug reaches the urine, there is not necessarily an incremental natriuresis. This is the case because once maximum response is reached, even more drug at the site of action does not increase response.

Overall, the pharmacodynamics of loop diuretics are such that the risk of inadequate delivery to the site of action outweighs the risk of too much delivery. In the clinical setting, this means that erratic administration results in insufficient delivery of diuretic to the urinary site of action. Short-term lapses and erratic timing can quickly produce fluid overload requiring hospital admission.[26,27,28]

The results of our study are limited by the small number of participants and potential biases from patient withdrawal. Only 42 patients (60%) completed the study. Those who withdrew most commonly did so because they lost or discarded their MEMS V lid. To reduce loss of important data contained in the lids, we recommend labeling lids with "Do Not Discard" or "Return for Refund." The median for taking adherence of 85% was higher than we expected. One explanation for this may be that the MEMS V lids increased participants' vigilance in taking drugs by the so-called Hawthorne effect.[29] Before enrollment, patients were educated as to the purpose and methods of the study and thus were aware that their adherence to therapy would be recorded. Second, because this was a convenience sample, volunteers may have self-selected so that the group was enriched with more adherent patients.

It would have been helpful to have more comprehensive data on the diuretic doses actually taken at each dosing interval. The MEMS V lids do not measure drug ingestion specifically, but only dates and times of lid openings. As a result, the association between lid openings and true adherence to therapy is unknown. Patients may have tampered with the lids in such a way as to skew the results of the study. They also may have taken more or less of the prescribed dose at an interval they choose. Furthermore, physicians may have given verbal approval for patients to take bursts of increased doses as necessary for fluid overload. Our study did not capture these data.

Another limitation was lack of measurements of adherence to other agents and other components of health care utilization. Only diuretic adherence was measured, although most patients took a number of other drugs. To obtain a complete picture of adherence, it would be necessary to assess adherence to all prescribed drugs, as well as cardiovascular compounds, to determine if a different drug or a combination may have influenced outcome. For this study, only numbers of hospitalizations and emergency department visits were quantified. Other measurements that in future studies may provide insight are diagnostic tests and radiology, laboratory, and clinic visits. Finally, participants were recruited from a city-county hospital that serves indigent patients. Therefore, the results may not be generalized to other health care systems or patients.

Despite these limitations, important information accrued from this study can be applied to future investigations in this area. Although the observational design and small sample preclude more definitive conclusions, we did find fewer cardiovascular hospitalizations among patients with high scheduled adherence, suggesting that interventions designed to increase adherence may improve patient care outcomes in a significant way. Furthermore, it would be appropriate to focus future investigations on the mechanisms by which high scheduled adherence results in lower health care utilization. Along this vein, large randomized trials would help to identify existing barriers and risk factors to improving adherence. For example, it is widely recognized that compliance is related to the number of drugs prescribed. In this study, due to design and sample size restrictions, we were not able to examine the effects of these factors on adherence and outcome. But our results clearly show a need for more detailed examination on these potentially important factors. Finally, future investigations should examine adherence to all pertinent heart failure drugs and incorporate all venues of health care utilization, thus providing a better picture of their exact relationship.