Phosphodiesterase type 5 inhibitor (PDE5i) drugs have been widely used as a treatment for erectile dysfunction.1 Although generally regarded as effective and safe, these drugs have also been associated with increased rates of high-risk sexual behaviour and HIV transmission in some men.2 Whether PDE5i increase the odds of condom breakage has not been studied. Breakage is not uncommon and largely negates the protective value of condoms relative to the prevention of pregnancy or sexually transmitted infections.3^–5 Given the enhanced erections provided by PDE5i, it is plausible that condoms become more tight-fitting. Tight-fitting condoms have been associated with breakage.6^–9

METHODS

Study sample

A convenience sample (n = 440) was recruited via advertisements in newspapers (two urban and one small town) and a blog on the website of a condom sales company that directed men to the questionnaire that was posted on the website of The Kinsey Institute. Inclusion criteria were male, at least 18 years old, used condoms for penile-vaginal intercourse (PVI) in the past 3 months and the ability to read English.

Measures

Assessments focused on the last condom use event for PVI. Condom breakage, PDE5i use and three covariates—duration of PVI, degree of vaginal lubrication and alcohol use (previously shown to be related to condom breakage9)—were assessed by asking the following questions, respectively:

Thinking about the last time you used a condom for PVI

Q1. Did the condom break during PVI? (no, yes, I don’t know)

Q2. Did you use Viagra, Cialis or Levitra? (no, yes)

Q3. How long was your penis in the vagina? (less than a minute, 1–5, 6–10, 11–15, 16–20, 21 or more minutes)

Q4. How well lubricated was your partner’s vagina? (very well lubricated (moist and slippery), adequate lubrication (no dry feeling), not quite enough lubrication (a little dry), poorly lubricated (very dry))

Q5. Were you drinking alcoholic beverages? (no, 1–2, 3–5, more than 5 drinks; a drink equals 1 beer, glass of wine or shot of liquor)

Data analysis

Continuous covariates not meeting normality assumptions were dichotomised by a median split. Variables achieving a screening-level of bivariate significance (p<0.15) with condom breakage or PDE5i use were entered into a multiple logistic regression model to calculate adjusted odds ratios (AOR), their 95% confidence intervals (CIs) and respective p values.

RESULTS

Characteristics of the sample

Average age was 29.6 years (SD 10.1). Breakage was reported by 26 men (5.9%) and 42 men (9.5%) reported PDE 5i use. Table 1 shows descriptive information relative to the dichotomised measures.

Bivariate associations

PDE5i use, alcohol use and minority ethnicity achieved screening significance (table 1). Table 1 displays the number of men responding “yes vs no” for each variable as well as the percentage of that number who reported condom breakage. For example, 42 reported PDE5i use and 11.9% of these (n = 5) also reported breakage compared with 5.3% (n = 21) of the 398 men not using PDE5i. Men’s age and reported duration of PVI were preserved as continuous-level variables; therefore, bivariate associations were based on independent group t tests. Age did not vary between men reporting breakage (mean age = 28.1 years) and not reporting breakage (mean age = 29.7 years) (t = 0.78 (438); p = 0.43). However, duration of PVI score was significantly longer among those reporting breakage (mean score = 4.7) versus those not reporting breakage (mean score = 4.0; t = 2.45 (438); p = 0.015).

PDE5i users were significantly older (mean age 41.1 years) than non-users (28.4 years; t = 8.3 (438); p<0.0001). PDE5i use was significantly more common among married men (18.5%) than single men (5.6%; p<0.001).

Multivariate associations

The logistic regression model using age, minority status (no/yes), single (no/yes), duration (scale 1 (less than 1 minute) to 6 (21 or more minutes)), alcohol use (no/yes) and PDE5i use (no/yes) was significant (χ² with 6df  = 16.01; p = 0.01) and produced a satisfactory fit with the data (goodness of fit χ² with 8df  = 9.45; p = 0.31). Minority ethnicity (AOR 2.59; 95% CI 1.04 to 6.43; p = 0.04), duration of PVI (AOR 1.36; 95% CI 1.02 to 1.83; p = 0.04) and PDE5i use (AOR 4.04; 95% CI 1.06 to 15.41; p = 0.04) achieved multivariate significance.

DISCUSSION

In a controlled analysis, condom breakage was about four times more likely among men using PDE5i drugs. However, a previous study of an older sample found that some men still experience erection problems while trying to use condoms in spite of PDE 5i use.10 Such findings highlight the importance of further studies, including qualitative research, to examine implications for physicians prescribing PDE5i drugs.

Findings are limited by the validity of retrospective self-report and the use of a convenience sample. Although the use of event-specific analysis adds considerable strength to the study, this feature precludes the use of a repeated measures design that would allow for within-person comparisons as a function of PDE5i use. Given support from further investigations regarding PDE5i use and condom breakage, counselling could emphasise the following messages: 1) use condoms large enough to fit the enhanced erections that may be experienced and 2) understand that longer duration of intercourse may be associated with increased risk of breakage. Because not all PDE5i use is through prescription,2 these same implications may also be relevant in sexually transmitted infections clinics.