STATS ARTICLES 2010
If you take Viagra, will you get an STD?
Rebecca Goldin Ph.D and Jing Peng, August 2, 2010
Media coverage reveals a classic confusion between causation and correlation, as they implied that Viagra results in greater risk for older men. But if anything, the study suggested just the opposite: Men who are interested in Viagra have riskier sex lives..
To judge from recent headlines, the risk seems clear:“Sex Diseases Tripled in Men 40 or Older Taking Viagra, Cialis, Study Says” reports Bloomberg; “Older Viagra Users More Likely to Get STDs” says the Chicago Sun Times, presumably comparing older Viagra users with older non-Viagra users. And HealthDay was even more explict, saying “Drugs Like Viagra Linked to Higher Rates of STDs.” The immediate logic behind the claim seems persuasively obvious: if men who take Viagra are having more sex, then they have surely increased their risk of catching a sexually transmitted disease (STD). But as is this really the case?
The study compared men who took erectile dysfunction (ED) drugs with those who did not. Researchers found that ED drugs such as Viagra are linked to higher rates of STDs among older men, but not older women, especially after the introduction of Viagra in 1998. A study published in the July 6 issue of the Annals of Internal Medicine was the first to examine the relationship between ED drugs and STDs. But its findings turn out to be far different than media accounts would have you believe.
The main problem in the coverage is the direct suggestion that taking Viagra is associated with STDs, as opposed to being the sort of person who takes Viagra.
This is a critical distinction. To see what it means, let’s go backward in time and compare STD rates among people who plan to take Viagra but haven’t and people who are not planning on taking Viagra. This clever comparison is what the authors of the research article did; they combed medical records for people who filled prescriptions for ED drugs, and compared STD rates for these people prior to filling their prescription to the STD rates of people who did not subsequently fill a prescription for Viagra. It turns out that the rate of STDs is higher among people who intend to take it. In other words, the drug is absent, but those people who will take Viagra within a year are already at higher risk of STDs.
In fact, compared to those who don’t take ED drugs, those who plan to take Viagra had a slightly higher rate for STDs (an odds ratio (OR) of 2.80; 95% confidence interval (CI), 2.10 to 3.75) than those who actually take it (OR 2.65, CI, 1.84 to 3.81), though the difference was not significant. This means that the drug had no discernable effect on STD rates for this group of men.
Hats off to Reuters who quoted an author, saying that "These users have a different sexual risk profile than non-users." AOL Health also quoted the author, noting that “men who request ED drugs from them are at higher risk of already having an STD or getting an STD in the future.” But in both cases, the editors couldn’t resist making the wrong implication in the headline. For Reuters, the headline was “Viagra-Popping Seniors Lead the Pack for STDs” and for AOL Health, it notes, “Men Who Use Viagra More Likely to Get STDs.” A better headline might be “Men Who Want Viagra More Likely to Get STDs” or even “Requests for Viagra Predictive of Risky Sexual Behavior.”
These headlines demonstrated the classic confusion between causation and correlation, as they implied that Viagra results in greater risk for older men. But if anything, the study suggested just the opposite: Men who are interested in Viagra have riskier sex lives.
The authors did assess what the media associations implied, mainly whether the use of Viagra among the group of men inclined to use it, was associated with an increase in STD exposure. They termed this the difference-in-difference odds ratio. First, they broke the men into two groups: those who use ED drugs and those who don’t (the control group). Then, for each of these groups, they broke the data into two sets: for the men who filled an ED prescription, the STD rates the year before filling them and the STD rates after filling them; and for the control group, the STD rate in the same time period (before the other group of men filled their ED prescription) and the STD rate after the other group of men filled their ED prescription. Then they looked at whether STD rates went up or down more among men who intended/did take ED drugs compared to men who did not.
The result: There was no association with STDs at all
As the authors of the study noted (but HealthDay did not), this does not negate the possibility that the introduction of such drugs has influenced increased STDs. There can be many factors, such as how sexual behavior has changed since the introduction into society of such drugs. But if Viagra has had such an effect, this study would not be able to measure it.
How the numbers are contorted to obtain an exaggerated effect: Relative risk rather than absolute risk
The media reported the odds ratio (which ranged from 2.65 after taking ED drugs, to 2.8 before taking them) as if it were a relative risk, and then rounded the number up. Given that the percentages of STDs were fairly small in the population (far less than one percent), the odds ratio may not be far from the relative risk. The raw (unadjusted) numbers showed a relative risk of approximately two rather than three, but the adjusted relative risk was not calculated in this study due to methodological constraints (as we explain below).
Unfortunately, ignoring absolute risk has become standard operating procedure in journalism. It offers the pleasing simplicity of a straightforward calculation – you are twice as likely or three times as likely for x or y – which helps to turn complex studies into dramatic headlines. While a tripling of risk in this case sounds impressive, the actual numbers were far less so. Among people who took an ED drug, about two in a thousand got an STD. Among people who did not, about one in a thousand got an STD.
The Media Misses: Study’s limitations and the study’s take-home message
A few media sources noted that the practical take-home message is that men interested in Viagra are at higher risk for STDs than men not interested. But even for these astute reporters, there were still important limitations of the study that went unreported in the press.
HealthDay's coverage, for example, emphasized the patients-are-different explanation for the data, without noting that we cannot assess whether there is a direct effect of ED drug availability on STD rates. Since STD rates among likely users were not evaluated before the introduction of Viagra to the public, there is no way of knowing whether the introduction itself led to increased STDs.
HealthDay quoted the lead author of the study speculating that HIV/AIDS was the most frequently reported STD because “ the symptoms that are associated with a primary HIV infection are the kinds of things that make men more likely to show up to a doctor, rather than go to a free clinic where they know they can get tested for an STD anonymously.” However, according to the original study results, HIV is substantially more prevalent than other diseases (since HIV is not curable, each person has it for a long period of time compared to something like Chlamydia). For this reason, HIV may be the most likely to transmit due to its prevalence rather than any other reason.
The results of the study illustrate the difficulties with an observational study rather than a randomized controlled trial. The fact that the type of person who takes Viagra is different than the type who does not would not have been observed had the researchers only looked at what happened among Viagra users (rather than including those who intended to use). But it also points to its own limitation: We cannot tease out the reasons for the observed differences in users and nonusers.
Another limitation was that the researchers only identified persons who had an insurance plan (and excluded persons outside of the insurance claims). Finally, the last limitation was that the prevalence of STDs is very low and precludes a highly powered analysis at the individual STD level.
These misconceptions led to differing medical advice as well: researchers suggested that screening would be most effective if targeted toward those at highest risk. HealthDay, instead, emphasized that physicians take greater responsibility in prescribing ED medications. However, both of these messages are more in line with the research results than the incorrect conclusion that stopping Viagra will reduce STDs.
The Technical Details
Researchers collected a large data set of pharmacy and medical claims to examine STD patterns because STDs are still rare in the general population and even more among middle-aged and older adults. Their final data included 1,410,806 male beneficiaries continuously enrolled for 24 months: 12 months before and 12 months after the first ED drug prescription was filled. All men were over the age of 40 with private, employer-based insurance from 44 large companies.
The pharmacy claims included all the information in the drug claims – the type of drug, drug name, national drug code, dosage, and days supplied. The medical claims included the date of service, diagnosis, and procedure code. Researchers observed a person in one of four quarters of the year. They classified a person as using an ED drug if they filled one or more prescriptions for either sildenafil (Viagra), radalafil, or vardenafil in that quarter. People were flagged who had at least one claim for one of the following STDs: HIV/AIDS, chlamydia, gonorrhea, herpes, syphilis, or other (Haemophilus ducreyi infection, or lymphogranuloma venereum).The STD indicators in the medical claims were identified according to International Classification of Diseases, Ninth Revision, diagnoses.
Researchers also considered the possibility that the data would be influenced by other health problems, related to the existence of erectile dysfunction and the likelihood of an STD. These are called comorbid conditions. The following disease indicators were identified for comorbid conditions: anxiety, asthma, cancer, cardiac, disease, congestive heart failure, chronic obstructive pulmonary disease, depression, diabetes, hypercholesterolemia, hypertension, stroke, and vascular disease. A beneficiary was determined to have one of these chronic conditions if their medical claims included two or more office visits with the corresponding International Classification of Disease, Ninth Revision, Code.
The researchers not only wanted to know how users of Viagra compared to nonusers, but also whether the rates of STDs went up after filling a prescription. The hypothesis was that users of ED drugs would have higher rates of STDs than nonusers both before and after initiating ED drug therapy. In addition, among users of Viagra, the study compared STD rates before the first ED drug prescription was filled to the rate after it was filled (for control reasons, they also did this within the group of nonusers).
As the authors note, “The model included terms to define both differences between groups and changes within groups over time. This difference-in-difference approach estimates the difference in STD trajectories between users and nonusers of ED drugs before and after filling an ED drug…. The difference-in-difference OR would be greater than 1 if, relative to nonusers, users of ED drugs had greater increases in STDs in the year after filling their first ED drug prescription compared with the year before.”
The results of the study showed a clear association between ED drugs use and higher rates of STDs. Men who were not on ED drugs, but filled a prescription for one within a year later, had an adjusted OR for an STD of 2.80 compared with nonusers in the year before the first ED drug prescription was filled. This was mainly driven by HIV (OR, 3.32; P<0.001) and Chlamydia (OR, 2.25; P<0.09). In the year after the first ED drug prescription was filled, users continued to have a higher adjusted OR of an STD than nonusers (OR, 2.65; P<0.001), this time driven only by HIV (OR, 3.19; P<0.001).
For both users and nonusers, there was essentially no change in STD rates from the year before to the year after the first ED drug prescription was filled. The results demonstrated that users of ED drugs had higher rates of HIV, chlamydia, gonorrhea, and syphilis in the 12 months both before and after filling their first ED drug prescription, although only HIV and chlamydia were statistically significant. Researchers also grouped all STDs together and excluded HIV, they found users of ED drugs continued to have statistically significant higher cumulative rates of STDs compared with nonusers in the 12 months both before and after initiating ED drug therapy. However, this difference between groups did not change across the 2 periods, suggesting that the observed association between ED drug use and STDs may have more to do with the types of patients using ED drugs rather than a direct effect of ED drug availability on STD rates.