There’s likely not going to be a RCT looking at abortions as an outcome, due to ethical reasons.
Could you explain what those reasons are? It’s fine to have death, cancer progression, miscarriage and other bad things as endpoints, so I don’t see why abortion would not be an ethically permitted endpoint? In my experience most of the ethical concerns have been around the intervention you are permitted to vary, not the endpoint. Gathering data might be hard, but that seems like a practical limitation, not an ethical one.
In most African countries at least, most forms of abortion are illegal and carry a lot of social stigma. It’s not about the outcome just being bad life your other examples.
The only way to gather this data would be to ask people in in a survey, which (I could be wrong) I doubt you would get through a local ethics board both because of the stigma and the illegality.
And yes practicality would also be very difficult as people are very unlikely to admit to having had abortions anyway due to above reasons.
I’m a little bit locally focused though, there may well be parts of the world where this is possible and has been done, but I’m sure it wouldn’t be easy both ethically and practically anywhere in the world.
There are ways to survey about illegal/taboo topics. Researchers studying GiveDirectly use them when asking if people spent the money on drugs, for example.
One common technique is list randomization, which allows surveyors to collect statistically meaningful results without respondents having to admit to anything in particular.
There are lots of survey-based studies on abortion in countries where abortion is illegal—the problem is not so much getting it through an ethics board as the reliability of the results. You could alternatively measure using hospitalisations for incomplete abortion as a proxy—you won’t be able to identify the exact magnitude of the problem or the change since an unknown proportion of these are from miscarriages (some have tried to estimate the ‘natural’ miscarriage presentation rate, but I think these estimates are obviously unreliable), but you could see if there is a change and whether it is a big or small change, since the miscarriage rate should remain relatively constant. It would need a big enough sample size though.
Hey Callum practically wise on the ground here I would be like 70 percent sure that it would be impossible to meaningfully assess abortion reduction and a study endpoint
An anonymous survey based study is a lot easier ethics wise than a big RCT, I think it would be a struggle to get through ethics approval in Uganda here—again I have a bit of experience with ethics board but could be wrong. For better or worse (I think worse) ethics boards are understandably often tighter on RCTs than other study forms.
Could you explain what those reasons are? It’s fine to have death, cancer progression, miscarriage and other bad things as endpoints, so I don’t see why abortion would not be an ethically permitted endpoint? In my experience most of the ethical concerns have been around the intervention you are permitted to vary, not the endpoint. Gathering data might be hard, but that seems like a practical limitation, not an ethical one.
In most African countries at least, most forms of abortion are illegal and carry a lot of social stigma. It’s not about the outcome just being bad life your other examples.
The only way to gather this data would be to ask people in in a survey, which (I could be wrong) I doubt you would get through a local ethics board both because of the stigma and the illegality.
And yes practicality would also be very difficult as people are very unlikely to admit to having had abortions anyway due to above reasons.
I’m a little bit locally focused though, there may well be parts of the world where this is possible and has been done, but I’m sure it wouldn’t be easy both ethically and practically anywhere in the world.
There are ways to survey about illegal/taboo topics. Researchers studying GiveDirectly use them when asking if people spent the money on drugs, for example.
One common technique is list randomization, which allows surveyors to collect statistically meaningful results without respondents having to admit to anything in particular.
There are lots of survey-based studies on abortion in countries where abortion is illegal—the problem is not so much getting it through an ethics board as the reliability of the results. You could alternatively measure using hospitalisations for incomplete abortion as a proxy—you won’t be able to identify the exact magnitude of the problem or the change since an unknown proportion of these are from miscarriages (some have tried to estimate the ‘natural’ miscarriage presentation rate, but I think these estimates are obviously unreliable), but you could see if there is a change and whether it is a big or small change, since the miscarriage rate should remain relatively constant. It would need a big enough sample size though.
Hey Callum practically wise on the ground here I would be like 70 percent sure that it would be impossible to meaningfully assess abortion reduction and a study endpoint
An anonymous survey based study is a lot easier ethics wise than a big RCT, I think it would be a struggle to get through ethics approval in Uganda here—again I have a bit of experience with ethics board but could be wrong. For better or worse (I think worse) ethics boards are understandably often tighter on RCTs than other study forms.
Thanks Nick. I have come across quite a lot of abortion surveys in countries with restrictive laws. Here is one from Uganda: https://journals.plos.org/globalpublichealth/article?id=10.1371/journal.pgph.0002340#sec007 I agree entirely that the results are likely not to be very reliable, but I think that is the bigger problem; less so the lack of ethics approval.