This longitudinal cohort study found that peripheral alpha blocker use was associated with more than a fourfold increased odds of UI. This finding was consistent with Marshall et al. (11
) and Ruby et al. (13
). What sets this study apart though is that we accounted for confounding by other antihypertensives, and the two previous studies did not. The strong association between peripheral alpha blocker use and UI was confirmed by point estimates well above 2 in our study. This is a biologically plausible relationship because physiologically alpha receptor blockade can relax the bladder neck and urethral smooth muscle, thus compromising bladder outlet resistance. Although alpha blocker use is relatively uncommon, it represents a powerful contributor to UI with a number need to harm of 24.
Given the proposed mechanism by which alpha blocker use might lead to UI, we hypothesized that this risk would be worse if alpha blockers were concomitantly taken with medications (eg, lithium and diuretics) or conditions (eg, diabetes) that can increase bladder volume via polyuria. Unfortunately, the number of lithium users was small (n = 2), and we were only able to assess diuretic use. We chose to conduct post hoc analyses looking at the three diuretic subclasses separately, as there is a clinical perception that certain types of diuretics (ie, loop) may have a more brisk response than others (ie, thiazide and potassium-sparing). As suspected, we found that the addition of loop diuretics nearly doubled the odds of UI seen with peripheral alpha blockers alone. It is interesting that no increased risk was seen with other types of diuretics in those taking alpha blockers. We also substantiated the possible mechanism of polyuria by demonstrating a trend toward an increased risk with diabetes.
What should clinicians take away from these post hoc analyses? First, it is important to recognize that the sample size was small and CIs were wide; so, further studies are needed to replicate our finding that alpha blockers in combination with loop diuretics are particularly risky for UI in community-dwelling women. Having said that, clinicians who prescribe loop diuretics for older women with certain comorbid diseases (eg, congestive heart failure) requiring additional antihypertensive therapy should consider adding on agents other than peripheral alpha blockers. This recommendation is clinically sensible because, in addition to UI, peripheral alpha blockers are associated with dizziness, fatigue, somnolence, and orthostatic hypertension. Moreover, in 2000, the doxazosin arm of the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial was halted after the drug was found to double the risk of heart failure compared with chlorthalidone in subjects aged 55 and older with hypertension and one or more cardiovascular disease risk factor (36
Another interesting finding from our study was the lack of association between any antihypertensive use, antihypertensive count, SDD, or individual drug classes except peripheral alpha blockers and UI. A priori, we expected to identify a relationship with diuretic use, regardless of type, and UI, but such a relationship was not seen in bivariable or multivariable analysis. Other antihypertensives with potential urologic activity include calcium channel blockers, which may decrease bladder contractility, and beta blockers, which may increase bladder contractility (3
), but these drug classes were not found to be associated with UI either. Furthermore, the side effect of cough associated with ACE inhibitors has the potential to affect stress UI in women who take a medication from this drug class (7
Strengths of this study include its longitudinal cohort design and the rigorous analytical approach used. The sample of older, generally well-functioning women included in this study was large, included a high proportion of black participants, and had a low dropout rate. Furthermore, medication use data collected from each woman were thorough and included medication name and strength, dosage form, frequency of use, and whether the medication was taken on a scheduled basis or only as needed. A potential limitation of this study is reliance on self-report to assess UI; however, self-report is commonly used to measure UI in epidemiological studies (3
). Moreover, self-reported UI has been shown to be consistent with urodynamic testing and physician diagnosis of UI, with the sensitivity (0.56–0.85) and specificity (0.66–0.96) of self-report varying by type of UI (ie, stress, urge, or mixed) (38
). Of note, type of UI, details of UI severity beyond its occurring at least weekly, and impact on quality of life were not included in our analyses. Additionally, due to sample size constraints, we were unable to restrict the analyses to only those who had hypertension. Therefore, antihypertensive drug classes could have been used for other indications; however, we did create control variables for some of the most common indications for which antihypertensive agents can be prescribed (ie, hypertension, congestive heart failure, and coronary heart disease). As with any study, unmeasured factors that could have confounded the relationship between medication use and UI are possible. The women in this study were well functioning at baseline and lived in or near two cities in the United States, which may limit generalizability.