What is recall bias?
BMJ 2012; 344 doi: https://doi.org/10.1136/bmj.e3519 (Published 23 May 2012) Cite this as: BMJ 2012;344:e3519
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Only gestational age of the pregnancy, between the cases and controls, are matched and congenital anomalies are excluded. Other confounders (obstetric, e.g. Antepartum haemorrhage and medical e.g. anaemia) are neither matched nor excluded.
Moreover, in this endgame example the researchers reported that to reduce recall bias, participants were not told of the study hypotheses being tested and therefore would not have been aware of the potential link between stillbirth and sleep practices.
This is ethically wrong as the participants of the study have to be informed about the study before taking ethical consent to be interviewed or examined.
Only way to remove recall bias which would have been foolproof, would be a prospective case-cohort study (from cause/exposure to effect- e.g. the sleeping practice reported by the patient to the effect on foetal heart e.g. cardiotocography instead of taking frank stillbirth).
Competing interests: No competing interests
You say: "However, women who slept on their back or
on their right side on the previous night (before
stillbirth or interview) had more than twice the risk
of a late stillbirth compared with women who slept on
their left side." But Stacey et al, the authors you
quote, give as their results:
"...(adjusted odds ratio for back sleeping 2.54 (95%
CI 1.04 to 6.18), and for right side sleeping 1.74
(0.98 to 3.01))."
By convention, if the odds ratio includes 1.0 then
the result is not significant. So right side sleeping
had no effect in this study.
J.R. Johnstone PhD
7 Bruce St,
Nedlands 6009,
Western Australia
08 9386 7080
0408 990 936
ray@iinet.com.au
www.iinet.com.au/~ray
Competing interests: No competing interests
Re: What is recall bias?
I thank Gupta[1] and Johnstone[2] for their rapid responses to my endgame[3], giving me the opportunity to discuss the topic of recall bias further.
Gupta’s rapid response included several comments. Firstly, they commented on the matching variables and potential confounders investigated by the authors of the article referenced in the endgame[4]. The relevance to this endgame is not obvious. Would it be more appropriate if these comments were directed to the authors of the original article?
It is not clear why Gupta perceived it to be unethical that participants in the study were not fully told of the aims before being interviewed. The health of the mother and foetus (for controls with ongoing pregnancy) were not placed at risk by the women simply reporting their past sleep practices. Presumably all women were subsequently informed of the results of this retrospective study. The authors of the article did indicate that their study received full ethical approval in New Zealand.
Gupta finally comments that the only way recall bias could have been removed would have been to use a case-cohort study design. This is debatable since the actual study design that was used minimised recall bias and possibly even eliminated it. If a case-cohort study design were used it may eliminate recall bias; however, it might possibly introduce another bias - expectant mothers may actually alter their sleep practices as a response to being fully aware of the study aims.
Johnstone comments on the results presented - that is, “….women who slept on their back or on their right side on the previous night (before stillbirth or interview) had more than twice the risk of a late stillbirth compared with women who slept on their left side." The presented odds ratios our best estimates of the population risks. However, just because the result was not statistically significant (as Johnstone correctly indicates), Johnstone was incorrect to suggest that such sleep practices have no effect or pose no risk. Statistical significance does not equate to clinical significance, and vice-versa.
1. Gupta N. The example given is not, statistically and ethically, without faults. 28th May 2012.
2. Johnstone JR. Re: What is recall bias? BMJ 27th May 2012.
3. Sedgwick P. What is recall bias? BMJ 2012;344:e3519.
4. Stacey T, Thompson JMD, Mitchell EA, Ekeroma AJ, Zuccollo JM, McCowan LME. Association between maternal sleep practices and risk of late stillbirth: a case-control study. BMJ 2011;342:d3403.
Competing interests: No competing interests