Relative risk; absolute risk

Standard

Howard White of 3ie discusses some of the problems we often see in how people handle data in this post, “Using the Causal Chain to Make Sense of the Numbers“.  The essay makes many excellent points which are relevant both to how programmes are designed and how they are evaluated.

However, I have to take issue with one section:

And different ways of presenting regression models can give a misleading sense of impact. A large reduction in relative risk – a ‘good odds ratio’ – can reflect quite a small change in absolute risk. Three randomised controlled trials have found circumcision reduces the risk of transmission during unprotected sex by around 50 percent. The reduction in risk was from around 3.5 percent to 1.5 percent. Just a 2 percentage point absolute reduction, so 50 men need to be circumcised to avoid one new case of HIV/AIDS.

This is a bit misleading. In assessing effects we are interested both in relative and absolute effects, yes.  But White fails to acknowledge here that the absolute risk at the outset (3.5% in the case of the pooled results of the three trials) is a characteristic of the people being researched.  And indeed the absolute risk in the populations in the three studies (Kenya, South Africa, Uganda) was different. The 3.5% and 1.5% figures come from pooling the results of the three trials.  If study subjects had come from a population where the pre-existing HIV prevalence was higher, and risk factors (including unprotected sex) were higher, then the baseline absolute risk would have been more than 3.5%.  If the risk factors had been lower, the baseline risk would have been lower. White’s estimate that 50 men need to be circumcised to avert one infection is not universally valid. In some places it will be many more; in others it will be fewer. This is one of the reasons male circumcision is primarily promoted in higher HIV prevalence settings.

Having said that, “just” a 2% absolute reduction is actually pretty good when compared to other HIV prevention interventions. Especially when you consider that once a man is circumcised, he stays circumcised, so the risk reduction is permanent. Look at it another way: if the intervention being tested led to a 100% risk reduction, then (according to White’s post), that would be “only” a 3.5% reduction in absolute terms. Still doesn’t look very impressive, does it? Except in this case there would be no new infections whatsoever.

The reason the results of these trials (and any trials) are reported as relative risks is because if you want to estimate what the effects of the intervention might be in another population, you have to apply the relative risk reduction to the absolute risk in each and every population. Reporting it any other way is misleading.  White is of course correct that absolute risk reduction is what matters when looking at the overall effect of an intervention or policy, but absolute risk reduction is a function not just of the relative risk reduction of that intervention, but of all the other relevant factors.

Advertisements

HIV and multiple (concurrent) partnerships

Standard

There’s a big to-and-fro in the latest issue of the Lancet about role of “multiple concurrent partnerships” versus “multiple partnerships” in the spread of HIV: four comments on an article published last week which argues (to simplify horribly) that there is no evidence that *concurrent* multiple sexual partnerships play more of a role than *sequential* multiple ones.  The four comments and the response from the authors of the original study response are here (scroll down).

It’s all quite interesting methodologically and epidemiologically speaking… in fact it is a debate that has been going on for some time.  But what strikes me most about the debate is the conclusions that a lot of the protagonists draw about what their findings mean for programmes.  Again, to simplify rather a lot, some of them (the ones that believe concurrency is a major factor) are saying “therefore we need messaging on limiting numbers of partners including concurrent partners”, whereas the others (the ones who dismiss concurrency) are saying “therefore we need messaging on limiting numbers of partners”.

And here’s the issue: there’s actually not a whole lot of evidence for the effects of either of these interventions. There’s certainly no evidence to show that either intervention is particularly more effective or more harmful or more costly than the other.  So the whole debate seems somewhat theoretical.

Where does policy end and practice start? Review of a paper on HIV programmes with sex workers.

Standard

“Risk” groups for HIV prevention

A few years ago, colleagues in an NGO asked me to comment on their HIV prevention strategy.  The strategy had a list of priority population groups that the programme would aim to target – these groups were selected based on actual or perceived risk of HIV infection.  One of the groups on the list was “wives of unfaithful men”.

Targeting health programmes towards people who are most affected or most at risk makes sense.  Not only is it more economical, but it also more practical since different types of people have different circumstances and needs.  But an important aspect of targeting is to make sure that the categories being defined are “operational” from the perspective of the people providing the services.  Was the idea that programmes would specifically target women with unfaithful husbands?  How might an outreach worker identify the women in the marketplace whose husbands are unfaithful, so as to focus on them?  Were sexual health clinicians being asked to factor in the likelihood of spousal infidelity when deciding what tests or treatments to prescribe to a married woman who comes to see them with symptoms?  And if “wives of unfaithful men” really were a high-risk group, were programmes also identifying the men and reaching them with sexual health services and advice?

Redefining risk categories – a new study from Karnataka, India

I was reminded of this story when reading a recent research paper, “Devising a female sex work typology using data from Karnataka, India”*.  The authors set out to define a typology for female sex workers that more accurately predicts HIV risk than the typology currently used in India’s National AIDS Control Organisation guidelines.  A more robust typology, that can act as a “predictor” of risk for HIV, would help programmes to discriminate between sex workers at higher and lower risk and prioritise programmes reaching the former, according to the article.

The existing typology classifies female sex workers according to where they solicit sex – because this is seen as the most practical basis for delivering services or outreach.   But the authors of this new study argue that this typology masks differences in risk for HIV, specifically measured by the actual HIV prevalence in each category.  The new classification proposed by the authors is based on a combination of the place of solicitation of sex with the place where sex actually takes place.  Categories include “Brothel to brothel”, “home to home”, “street to home”, “street to rented room “street to lodge” “street to street” and “other” (in each case, the first term represents the place of solicitation and the second the place of sex). 

The authors successfully demonstrate that the newly proposed set of seven categories gives a more complex picture of how risk for HIV – measured by the current level of HIV prevalence – is distributed, when compared to the other typologies in use (which had 4 and 6 categories respectively).  The analysis shows that the category most affected by HIV is the “brothel to brothel” one (34.0%), the lowest is other (11.1%).  The authors also argue that further research might enable the development of even more sensitive categorisations can be developed based on.

How can the findings be used?

While it is true that the study shows clearly that there are bigger variations in HIV prevalence between different categories of sex worker under the new proposed typology, might the comparative statistics might be getting in the way of the bigger picture?  While the odds of HIV infection among the “street to lodge” category may well be nearly 3 times higher than it is for the “other” category, sex workers in the other category are still over 20 times more likely to be HIV positive than women in the general population.  Levels of other STIs in this group are also not negligible, as are levels of condom breakage.  Members of the “other” group were the most likely to have been raped in the past year.  Levels of unwanted pregnancy or of other health problems are not reported.

It isn’t clear that there is a strong justification for focussing on some of the new categories rather than old ones.  Are categories such as “street to home” operational when compared to “street to lodge”?  Can outreach workers or clinicians differentiate between the street-soliciting sex workers based on whether they go home or to a lodge to have sex, and should they?  And are the types of services required by members of each of these new categories demonstrably different from each other? 

On the one hand, while the commitment to understanding context and not providing a one-size-fits-all approach is commendable, it isn’t clear what to do with the findings.  Given how affected all sex workers in the study were by HIV, STIs, and violence, it hardly seems appropriate to focus only on those most at risk.  On the other hand, if the study findings demonstrate that some of the new categories are clearly “underserved” by existing programmes, the findings might help encourage programmers to ensure that this is redressed.  In India there is still a way to go in reaching sex workers with HIV and AIDS programmes: according to India’s most recent HIV report to the UN, coverage of sex workers by essential HIV programmes is only just over 50%.

When does policy stop and practice start?

Epidemiological studies on HIV risk aim to identify who the “most at risk” groups are.  They look at a number of different variables – including actual levels of HIV infection, other STIs, and behaviours that may contribute to risk.  But studies of this kind are inevitably generalisations because in any given population group, there will be subcategories, and eventually individuals, who are at higher or lower risk.  While there may be a temptation to stratify population groups further and further so as to increase the precision of risk categories, might there also be a point at which the increasing numbers of categories make it harder, not easier, to deliver programmes?  This might also be the point at which the analysis is no longer be the job of the researcher, policy maker or programme designer, but of the health care provider or outreach worker – because front line providers should be responding based on what individual clients tell them, not based on what the epidemiological categories are telling them to assume. 

 

* Buzdugan R, Copas A, et al “ (2010) Devising a female sex work typology using data from Karnataka, India”, International Journal of Epidemiology 39 (439-448)

Epi…dream…iology. How HIV prevention programmes are sometimes based on made up stuff.

Standard

In my view, one of the strengths of many HIV and AIDS programmes is that they emphasise community participation, involving the people affected in analysing the issues they face and developing strategies to overcome them.  It can work really well, when it is well facilitated and based in reality.  But in the absence of good data on HIV vulnerability and risk – which is the case in many low income countries – everyone becomes an expert, and there can be a fair amount of people sitting around trying to conjure up the most worst scenarios to respond to.  And as a result, HIV prevention strategies often reflect prejudices and moral panic.  You often end up with long lists of stuff that could, conceivably happen, but no sense of whether it actually does.

I recently came across some conclusions from a study carried out by a development NGO on risks of HIV in the NGO workplace, in a sub-saharan African country.  It is described as “baseline research” but the methodology is not clear.  I do know that some of the recommendations were developed at a workshop where the participants “decided that xxxxx was a risk factor”, which suggests that the conclusions were based on guesswork.

Here are a few excerpts from the study, relating to risk and vulnerability factors (translated from French, hence the oddness of some language).  As far as I can tell, the ultimate conclusion is that AIDS programmes cause AIDS.  Note the blind spot on same-sex relationships, and the fact that condoms aren’t mentioned anywhere. 

“The factors that negatively affect the fight against aids in institutional settings are as follows:

  • The planning of unnecessary overtime by employers and employees of different sexes
  • Placing managers’ offices in isolated or overly discrete (hidden?) places
  • Sending colleagues of different sexes together on missions lasting more than a day
  • Work with beneficiary communities can be tempting for employees especially girl beneficiaries (orphans) and women beneficiaries (widows)

In terms of employees, factors identified were:

  • long field visits away from partner
  • unnecessary overtime by managers and juniors of different sexes
  • the authority of the boss which can lead to more or less forced sex
  • seeking protection at work which can lead to sexual provocation (temptation?)
  • spending a long time in each other’s company (sharing the same office every day) which can lead to over-familiarity and can end in sex
  • participating in workshops lasting more than one day by colleagues of different sexes
  • work in the field that can lead to intimate relationships between employees and programme beneficiaries
  • Infidelity of the partners of employees because they are away so often.

Practices identifed at the level of beneficiary communities are field missions including participation in trainings and improving the lives of beneficaries for whom the new socioeconomic status opens the way to friends who let them satisfy their sexual needs… eg child headed househoulds and vulnerable women who got enough money to get boyfriends and took on other risk behaviours like polygamy, unmarried relationships and alcooholism.”