Sexual risk behaviours following medical male circumcision: a matched pseudo-cohort analysis using population-based survey data

Using a matched pseudo-cohort analysis of the 2024 Zambia Demographic and Health Survey, this study found that medical male circumcision was not associated with increased sexual risk behaviors, thereby supporting the behavioral safety of the intervention and demonstrating the value of adapting cross-sectional data to address temporal challenges.

The Gist

Imagine you have a new, highly effective shield designed to protect soldiers from a specific enemy (in this case, the HIV virus). This shield is Medical Male Circumcision (MMC). It's a proven medical procedure that works great.

However, there's a nagging worry among the generals (public health officials): "If these soldiers feel safer with their new shields, will they start running into more danger? Will they take bigger risks because they think they're invincible?" This is called "risk compensation."

For years, scientists have tried to answer this question, but the data has been messy. It's like trying to figure out if a driver started speeding after they got a new car, but you only have a snapshot of them driving today. You don't know if they were speeding before they got the car, or if the car made them speed up.

The New Approach: Rewinding the Tape

This study, conducted in Zambia, decided to solve that "snapshot" problem by using a clever trick. They took a massive survey of nearly 10,000 men and used a method called a "matched pseudo-cohort."

Think of it like this:
Instead of just comparing two different groups of people (circumcised vs. uncircumcised), the researchers acted like time-traveling detectives. They looked at the circumcised men and asked, "What were you doing before you got your shield?" Then, they compared that to what uncircumcised men were doing at the exact same age.

They also used a statistical tool called "overlap weighting." Imagine you have two groups of people: one group is very fit, and the other is less fit. To compare them fairly, you don't just look at the extremes; you focus on the "middle ground" where the two groups look most similar. This ensures the comparison is fair, like comparing apples to apples rather than apples to oranges.

The Findings: No "Superhero" Effect

So, did the men with the new shields start acting recklessly? No.

Here is what the study found, translated into plain English:

1. The "Seatbelt" Effect: Men who got circumcised were actually more likely to use condoms during their last sexual encounter. It's as if getting the shield made them feel responsible enough to wear a seatbelt, not reckless enough to drive without one.
2. No More Parties: There was no evidence that these men started having sex with more partners. They didn't throw more parties just because they felt safer.
3. No More Infections: They didn't report more symptoms of sexually transmitted infections (STIs).
4. Overall Safety: When you look at the big picture of all risky behaviors combined, the men with the procedure were actually slightly less likely to engage in risky behavior than those without it.

The Bottom Line

The study concludes that giving men this medical protection does not make them act like they have a "get out of jail free" card. They don't throw caution to the wind.

Instead, the data suggests that when you give people a tool to stay safe, they tend to use it responsibly. The researchers also showed that even with "snapshot" surveys (which usually can't tell time), you can use smart math to reconstruct the timeline and get a clear answer.

In short: The shield works, and the men wearing it aren't running into danger; they're actually being a bit more careful than before.

Technical Summary

Problem Statement

Medical Male Circumcision (MMC) is a proven biomedical intervention for HIV prevention. However, a persistent concern in public health is risk compensation: the hypothesis that circumcised men may engage in higher-risk sexual behaviors post-procedure, potentially negating the protective benefits of the surgery. Previous observational studies have yielded inconsistent results, largely due to a methodological limitation: many analyses fail to adequately distinguish between sexual behaviors occurring before versus after the circumcision event. This temporal ambiguity makes it difficult to isolate the causal effect of MMC on subsequent behavior using standard cross-sectional data.

Methodology

The study employed a sophisticated analytical framework to overcome the limitations of cross-sectional survey data:

• Data Source: The analysis utilized nationally representative data from the 2024 Zambia Demographic and Health Survey (ZDHS), focusing on 9,609 men aged 15–59 years.
• Study Design: The researchers constructed a matched pseudo-cohort design. This approach reconstructed temporal ordering by utilizing the reported "age at circumcision" to differentiate behaviors occurring before and after the procedure, effectively simulating a longitudinal cohort within a cross-sectional dataset.
• Statistical Approach:
  • Propensity Score Overlap Weighting: Applied to enhance comparability between the treated group (medically circumcised men) and the control group (uncircumcised men), ensuring the analysis focused on the "overlap population" where causal inference is most robust.
  • Complex Survey Design: Logistic regression models incorporated overlap weights and accounted for the survey's complex design (clustering and stratification) to generate accurate estimates.
• Outcome Measures: The study assessed sexual behaviors occurring post-circumcision, including:
  • Condom non-use at last sexual intercourse.
  • Multiple sexual partners in the past 12 months.
  • Self-reported Sexually Transmitted Infection (STI) symptoms.
  • Composite measures of overall sexual risk behavior.

Key Contributions

1. Methodological Innovation: The paper demonstrates a novel adaptation of population-based cross-sectional survey data to address the temporal challenge inherent in behavioral studies. By reconstructing the timeline of events based on age at circumcision, the study provides a more rigorous test of risk compensation than previous observational studies.
2. Rigorous Causal Inference: The application of propensity score overlap weighting within a matched pseudo-cohort framework significantly reduces selection bias, offering a clearer picture of the association between MMC and subsequent behavior.

Results

Among the 9,609 participants, 33.3% were medically circumcised. The analysis yielded the following adjusted odds ratios (aOR):

• Condom Use: MMC was associated with lower odds of condom non-use at last sexual intercourse (aOR = 0.75; 95% CI: 0.67–0.85).
• Composite Risk: MMC was associated with lower odds of reporting any sexual risk behavior (aOR = 0.83; 95% CI: 0.72–0.95).
• Null Findings: No statistically meaningful associations were found between MMC and:
  • Reporting multiple sexual partners.
  • Self-reported STI symptoms.
  • Higher levels of composite sexual risk behavior.

Significance

The study provides robust evidence that, under routine program conditions in Zambia, MMC is not associated with sexual risk compensation. Instead, the data suggests that medically circumcised men may actually exhibit safer sexual behaviors (specifically higher condom use) compared to their uncircumcised counterparts.

These findings support the behavioral safety of MMC as an HIV prevention strategy. Furthermore, the study validates the utility of explicitly addressing temporality when analyzing behavioral outcomes in cross-sectional surveys, offering a methodological blueprint for future research in similar contexts where longitudinal data is unavailable.