Interventions used in disease management programmes for patients with chronic illnesswhich ones work? Meta-analysis of published reports
BMJ 2002; 325 doi: https://doi.org/10.1136/bmj.325.7370.925 (Published 26 October 2002) Cite this as: BMJ 2002;325:925
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.
Dear Editor,
On 28 October 2002 Dr. P. Badrinath commented on the possible
language bias (1) in the paper of Weingarten et al. (2). All sorts of
biases are being described when conducting meta-analyses (3), but database
bias seems to be overlooked. Weingarten et al. (2) searched Medline,
HealthStar, and Cochrane databases and Badrinath (1) searched Medline
alone.
The CINAHL database appears to be more appropriate to search for
subject headings like ‘primary nursing care’.
1)Badrinath P. Does the LANGUAGE bias deserve a mention?
bmj.com Rapid Responses for Weingarten et al., 325 (7370) 925.
2)Weingarten SR, Henning JM, Badamgarav E, Knight K, Hasselblad V, et
al. Interventions used in disease management programmes for patients with
chronic illness-which ones work? Meta-analysis of published reports. BMJ
2002;325:925.
3)Jadad AR. Randomised Controled Trials. A user’s guide. BMJ Books
1998
Competing interests: No competing interests
Dear Editor,
The authors of the recent meta-analysis(1)on the management of
chronic illness need to be commended for their efforts. They have
generated evidence on the effectiveness of different disease management
programmes, which no doubt will be of interest to patients, and providers
of care. However, I was surprised to find that language bias has not been
addressed by the authors.
It has been shown(2) that authors are more likely to publish
randomised controlled trials (RCTs) in an English-language journal if the
results are significant. Hence, English language bias may be introduced in
reviews and meta-analyses if they include only trials reported in English.
There is mixed evidence on the impact of performing meta-analysis leaving
out papers published in languages other than English. Gregoire et al
(3)found that for 1 in 36 meta-analyses results and conclusions differed
when non-English papers were included in reanalysis. According to Juni et
al (4) excluding trials published in languages other than English
generally has little effect on summary treatment effect estimates but it
would be hard to predict the impact on individual systematic reviews.
I performed a quick search of Medline selecting all the non-English
languages and used the key words utilised by the authors in their search.
There were over 100 hits in the six different languages although it is not
known how many of these might be relevant. BMJ, which has published papers
on investigating (5)bias in meta-analysis, should have paid more attention
to the language bias in the discussion section of this publication.
References
1)Weingarten SR, Henning JM, Badamgarav E, Knight K, Hasselblad V, et
al. Interventions used in disease management programmes for patients with
chronic illness-which ones work? Meta-analysis of published reports. BMJ
2002;325:925.
2)Egger M, Zellweger-Zahner T, Schneider M, Junker C, Lengeler C,
Antes G. Language bias in randomised controlled trials published in
English and German. Lancet 1997;350(9074):326-9.
3)Gregoire G, Derderian F, Le Lorier J. Selecting the language of the
publications included in a meta-analysis: is there a Tower of Babel bias?
J Clin Epidemiol 1995;48(1):159-63.
4)Juni P, Holenstein F, Sterne J, Bartlett C, Egger M. Direction and
impact of language bias in meta-analyses of controlled trials: empirical
study. Int J Epidemiol 2002;31(1):115-23.
5)Sterne JA, Egger M, Smith GD. Systematic reviews in health care:
Investigating and dealing with publication and other biases in meta-
analysis. BMJ 2001;323:101-5
Dr.P.Badrinath
Specialist Registrar in Public Health Medicine &
Honorary Clinical Lecturer, Suffolk Public Health Network &
University of Cambridge, Ipswich IP3 8LS.
Competing interests: No competing interests
Unit of analysis errors in systematic reviews
Dear Editor
We congratulate Weingarten and collegues(1) on their generally clear
presentation of a comprehensive study in what is a complex area of
research. However we were unclear whether any of the included primary
studies had “unit of analysis errors” and how the authors dealt with such
studies within the meta analysis.
Unit of analysis errors(2) occur within cluster randomised trials
when individual patient data are analysed as if there was no clustering
within the provider, practice or units randomised to the intervention
groups (patient data are analysed as independent observations). Standard
statistical methods that do not account for cluster effects within cluster
randomised trial data can result in the overestimation of the significance
of an intervention (artificially extreme p-values and over narrow
confidence intervals)(2). Correspondingly, the inclusion of studies with
unit of analysis errors in a meta analysis will give greater weight to the
results of such studies(3).
The table of included studies reported by Weingarten and colleagues
indicated that the unit of analysis differed from the unit of
randomisation in 22 cluster randomised trials but it was not clear from
the report how often unit of analysis errors occurred in these studies or
how the authors dealt with studies with unit of analysis errors in the
meta-analysis. There are methods for re-analysing studies with unit of
analysis errors. However, we recently completed a systematic review of
guideline dissemination and implementation strategies, 51 out of 110
cluster randomised trials had unit of analysis errors and re-analysis was
only possible in one study. Poor reporting of cluster randomised trials
has led to a proposed extension to the CONSORT statement which is
currently under discussion.(4) Systematic reviews of studies with unit of
analysis errors should clearly state how they handled such studies within
a review.
1 Weingarten SR, Henning JM, Badamgarav E, Knight K, Hasselblad V, et
al. Interventions used in disease management programmes for patients with
chronic illness-which ones work? Meta-analysis of published reports. BMJ
2002;325:925.
2 Whiting-O'Keefe QE, Henke C, Simborg DW. Choosing the correct unit of
analysis in medical care experiments. Medical Care 1984;22:1101-14.
3 Donner A, Piaggio G, Villar J. Statistical methods for the meta-
analysis of cluster randomised trials. Statistical Methods in Medical
Research 2001;10:235-338.
4 Elbourne DR, Campbell MK. Extending the CONSORT statement to cluster
randomised trials: for discussion. Statistics in Medicine 2001;20:489-496.
Competing interests:
None declared
Competing interests: No competing interests