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


Of 70 patients invited to participate, 56 were enrolled and 42 (60%) completed the study. Five individuals declined to participate because they did not feel well enough and nine did not want to be bothered. When statistical tests were conducted to determine demographic differences between those who enrolled and those who refused, sex was the only factor that was significantly different, with more women than men refusing (p=0.0054).

Fourteen patients (25%/56) were withdrawn before the data-collection period was completed. Of these patients, 10 lost or discarded their MEMS V lids, 3 left the health care system, and 1 died. We compared differences between patients who finished the study and those who were withdrawn. Mental status[11] was significantly different between those groups, with patients who finished the study having significantly higher mental status scores than those who did not ( 2 = 2.24, p=0.0295).

Study participants' ages ranged from 29-77 years, with a mean ± SD age of 57.2 ± 11.6 years. Approximately 60% were men, and 60% were African-American. Approximately 75% had NYHA class II or III. Mean mental status score was 8.7 ± 1.1. The average number of years of education was 10.4 ± 3 years. Table 1 shows the demographic variables of the participants.

The diuretic most commonly prescribed was furosemide (88%). In addition to diuretics, most patients were taking at least one additional agent, such as angiotensin-converting enzyme inhibitors, digoxin, and -adrenergic antagonists. On average, a patient had 43 ± 22 prescriptions filled or refilled over 6 months, indicating that each patient had approximately seven prescriptions filled each month.

Table 2 shows patient adherence as measured by MEMS V lids. Patients had an average taking adherence of 71.9%, but the standard deviation of 30% suggests wide variability. Indeed, taking adherence ranged from 4-103%. As a result, low taking adherence outliers considerably reduced the mean adherence level. For this reason, the median, 85%, may better represent this variable. Scheduled adherence also was widely dispersed. Daily and scheduled adherences were moderately correlated (r = 0.65, p<0.0001). Multiple linear regression analysis indicated that level of education was positively associated with daily adherence (p=0.0069) but was not significantly associated with scheduling adherence.

Each participant was hospitalized at least once during the 25-week study. Three participants had more than 10 hospitalizations, with one admitted 36 times. Visits to emergency departments were also frequent, with five patients having seven or more visits and one having nine. Most patients visited the emergency department once, although not uniformly for a cardiac- or heart failure-related diagnosis ( Table 3 ).

Association between adherence and adherence measures is reported in Table 4. Separate log-linear models were fitted to examine the effects of adherence measures on utilization. Poor scheduling adherence was a significant predictor for increased hospitalizations for both cardiovascular- and heart failure-related causes (p=0.0006 and 0.0444, respectively). Other significant factors were NYHA classification (p<0.0001 both cardiovascular and heart failure models) and household income (p<0.0001 cardiovascular hospitalization model, p=0.0007 heart failure hospitalization model), both associated with more hospitalizations. Taking adherence measures were used in model selection but were eliminated because of the stronger effects of scheduling adherence on response variables.

Similar log-linear regression analyses were used to model emergency department visits. In model selection, both adherence measures were eliminated early, suggesting their considerably weaker effects on various types of emergency department visits. The NYHA classification, however, showed a much stronger effect on cardiovascular-related visits (p=0.001) and heart failure-related visits (p=0.07).

Other demographic and clinical factors, such as other heart failure-related drugs and total number of drugs, were not significant predictors of adherence or health care utilization.