- Research article
- Open Access
- Open Peer Review
Subgroup effects despite homogeneous heterogeneity test results
© Groenwold et al; licensee BioMed Central Ltd. 2010
- Received: 28 January 2010
- Accepted: 17 May 2010
- Published: 17 May 2010
Statistical tests of heterogeneity are very popular in meta-analyses, as heterogeneity might indicate subgroup effects. Lack of demonstrable statistical heterogeneity, however, might obscure clinical heterogeneity, meaning clinically relevant subgroup effects.
A qualitative, visual method to explore the potential for subgroup effects was provided by a modification of the forest plot, i.e., adding a vertical axis indicating the proportion of a subgroup variable in the individual trials. Such a plot was used to assess the potential for clinically relevant subgroup effects and was illustrated by a clinical example on the effects of antibiotics in children with acute otitis media.
Statistical tests did not indicate heterogeneity in the meta-analysis on the effects of amoxicillin on acute otitis media (Q = 3.29, p = 0.51; I2 = 0%; T2 = 0). Nevertheless, in a modified forest plot, in which the individual trials were ordered by the proportion of children with bilateral otitis, a clear relation between bilaterality and treatment effects was observed (which was also found in an individual patient data meta-analysis of the included trials: p-value for interaction 0.021).
A modification of the forest plot, by including an additional (vertical) axis indicating the proportion of a certain subgroup variable, is a qualitative, visual, and easy-to-interpret method to explore potential subgroup effects in studies included in meta-analyses.
- Included Study
- Random Effect Model
- Aggregate Data
- Acute Otitis Medium
Practice guidelines increasingly rely on systematic reviews and meta-analyses. The ultimate purpose of a meta-analysis is to produce an overall estimate of the effect of an intervention by quantitatively combining study results. However, several issues arise in the process of integrating evidence. One of the main issues concerns heterogeneity, i.e. the extent to which different studies give similar or different results. Statistical tests are routinely available to evaluate the presence of statistical heterogeneity (between-study heterogeneity) in meta-analysis [1–3]. Strictly speaking, however, one is not really interested in statistical heterogeneity. What one is interested in is clinical heterogeneity, i.e., specific causes that underlie heterogeneity across studies, especially since the direction and magnitude of the effect in the meta-analysis is often used to guide decisions about clinical practice for a wide range of patients. Yet, relevant subgroup effects may not be revealed by a test for (statistical) heterogeneity. In meta-regression analysis the relation between a certain subgroup characteristic and the size of the treatment effect can in fact be quantified, but such analyses might be difficult to conduct or interpret, and rely on several assumptions. Furthermore, the observed treatment effect and subgroup variables are actually estimates, rather than true values. Ordinary meta-regression analysis (weighted least squares) does not take measurement errors in treatment and subgroup variables adequately into account and may consequently give a biased estimate of the slope of the regression line . We will show that clinically relevant subgroup effects can be explored in a simple manner by modifying the forest plot.
Tests for heterogeneity
Several tests have been developed to assess heterogeneity. The so-called Cochrane's Q (or Cochrane's χ2 test) weights the observed variation in treatment effects by the inverse of the variation in each study . A large value of Q indicates large differences between studies, and hence, the effects from the included studies can be considered heterogeneous . A modification of Cochrane's Q is the measure I2, which is the ratio of variation that exceeds chance variation and the total variation in the treatment effects. Possible values for I2 range from zero to one, with a high value for I2 indicating much heterogeneity. Both Q and I2 are standardized measures, meaning that they don't depend on the metric of the effect size. A third measure of heterogeneity, indicating the variance of the true effect sizes is T2, where (similar to Q and I2) large values of T2 indicate heterogeneity. This method of estimating the variance between studies (T2) is also known as the method of moments, or the DerSimonian and Laird method . A fourth measure is the prediction interval, which indicates the distribution of true effect sizes and is based on T2 . Cochrane's Q is sensitive to the number of studies and especially when the number of studies included in a meta-analysis is small, Cochrane's Q too often leads to false-positive conclusions (too large type I error) . The modification I2 takes account of the number of included studies and has a correct probability of a type I error . The measure T2 is insensitive to the number of studies as well, but sensitive to the metric of the effect size .
Currently, I2 appears to be used routinely in most published meta-analyses. Interestingly, the observed amount of heterogeneity depends on the effect measure that is considered in a meta-analysis: little heterogeneity when considering odds ratios implies large heterogeneity when considering risk differences and vice versa . The reason for this is analogous to effect measure modification in a single study: if odds ratios are the same between strata (e.g., age categories) of a single study, risk differences are likely to differ between strata.
Consequences of heterogeneity
Tests for heterogeneity indicate whether the variation in observed effects is either large or small. When heterogeneity is low (non-significant) for the chosen effect measure, variation between effects from different studies is (relatively) small. Thus, a fixed effects model can be used to synthesize the data, since the assumption underlying a fixed effects model is that the treatment effect is the same in each study, and variation between studies is due to sampling (i.e., chance) [3, 7]. If variation in the effects found in the different studies is (relatively) large they could be considered as sampled from a distribution of effects, i.e., the true treatment effect that is estimated in the different studies is not a single value, but rather a distribution of effects. In that case, a random effects model has been recommended [3, 7]. It has also been suggested that heterogeneity is inevitable in meta-analysis , and random effects models are therefore obligatory. If, however, heterogeneity is (very) large, one could even consider not pooling results from different studies at all, since studies are likely to be (very) different . Furthermore, if there is a cause for heterogeneity, for example a subgroup effect, neither fixed nor random effects models take such relations between the effect size and subgroups into account. Another explanation for heterogeneity (other than differential treatment effects) could be a systematic error in the included studies. For example, systematic error that is related to e.g., the proportion of women, or differences in methodology (e.g., differences in outcome ascertainment) of the included studies .
Relevant subgroup effects
What is important is that a regular forest plot (Figure 1) only contains a horizontal axis (indicating the effect size), whereas the modified forest plot (Figure 2) contains two axes. Both in the regular forest plot and in the modified forest plot the horizontal axis indicates the effect size. The additional vertical axis in the modified forest plot indicates the proportion of a certain subgroup variable in the included studies. Importantly, the vertical axis does not simply indicate the order of the subgrouping variable, but also scales this variable.
An often used quantitative approach to investigate the association between a certain subgroup characteristic and the size of the treatment effect is by applying meta-regression analysis . Such analyses however rely on several assumptions, e.g., linearity of the association, and might be hard to interpret for their quantitative nature. Furthermore, in ordinary meta-regression analysis the treatment effects from the included studies are handled as if they are true values rather than estimates, which can result in bias when using least squares regression . In addition, aggregated data meta-(regression) analyses are inappropriate to estimate unbiased treatment effects in patient subgroups, since such comparisons are observational by nature. As a result, the observed subgroup effect may be attributable to other variables than the subgrouping variable . Furthermore, as indicated before, neither fixed nor random effects models address the cause for heterogeneity. Individual patient data meta-analysis can be a valid alternative to study subgroup effects . In conclusion, the modified forest plot is a qualitative, visual alternative to assess the potential for a clinical relevant subgroup effects.
Neither the absence nor the presence of heterogeneity (as indicated by the result of heterogeneity tests) in aggregated data meta-analyses appears to be indicative for subgroup effects. Irrespective of statistical test results, subgroup effects can be present in the data. A modified forest plot, including an additional (vertical) axis indicating the proportion of the subgroup variable (e.g., the proportion of bilateral otitis), may be helpful to identify clinically relevant subgroup effects. Unfortunately, patterns in a modified forest plot can only indicate (qualitative) subgroup effects. To quantify a subgroup effect, meta-regression analysis can be applied, but validity of results is then subject to several assumptions, and individual patient data meta-analysis might be needed . The modified forest plot is a qualitative, visual, and easy-to-interpret alternative for exploring potential clinically relevant subgroup effects in studies included in aggregated data meta-analyses and should be considered when exploring such effects.
- Ioannidis JP: Interpretation of tests of heterogeneity and bias in meta-analysis. J Eval Clin Pract. 2008, 14: 951-7.View ArticlePubMedGoogle Scholar
- Borenstein M, Hedges LV, Higgins JPT, Rothstein HR: Identifying and quantifying heterogeneity. Introduction to Meta-analysis. 2009, Chichester, UK: John Wiley&Sons, 107-26. full_text. 1View ArticleGoogle Scholar
- Higgins JPT, Thompson SG: Quantifying heterogeneity in a meta-analysis. Stat Med. 2002, 21: 1539-58. 10.1002/sim.1186.View ArticlePubMedGoogle Scholar
- Arends LR, Hoes AW, Lubsen J, Grobbee DE, Stijnen T: Baseline risk as a predictor of treatment benefit: three clinical meta-re-analyses. Stat Med. 2000, 19: 3497-3518. 10.1002/1097-0258(20001230)19:24<3497::AID-SIM830>3.0.CO;2-H.View ArticlePubMedGoogle Scholar
- Cochran WG: Some methods for strengthening the common χ2 tests. Biometrics. 1954, 10: 417-51. 10.2307/3001616.View ArticleGoogle Scholar
- DerSimonian R, Laird N: Meta-analysis in clinical trials. Control Clin Trials. 1986, 7: 177-88. 10.1016/0197-2456(86)90046-2.View ArticlePubMedGoogle Scholar
- Higgins JPT, Thompson SG, Deeks JJ, Altman DG: Measuring inconsistency in meta-analysis. BMJ. 2003, 327: 557-60. 10.1136/bmj.327.7414.557.View ArticlePubMedPubMed CentralGoogle Scholar
- Engels EA, Schmid CH, Terrin N, Olkin I, Lau J: Heterogeneity and statistical significance in meta-analysis: an empirical study of 125 meta-analyses. Stat Med. 2000, 19: 1707-28. 10.1002/1097-0258(20000715)19:13<1707::AID-SIM491>3.0.CO;2-P.View ArticlePubMedGoogle Scholar
- Higgins JP: Commentary: heterogeneity in meta-analysis should be expected and appropriately quantified. Int J Epidemiol. 2008, 37: 1158-60. 10.1093/ije/dyn204.View ArticlePubMedGoogle Scholar
- Sterne JA, Juni P, Schulz KF, Altman DG, Bartlett C, Egger M: Statistical methods for assessing the influence of study characteristics on treatment effects in 'meta-epidemiological' research. Stat Med. 2002, 21: 1513-24. 10.1002/sim.1184.View ArticlePubMedGoogle Scholar
- Thompson SG, Sharp SJ: Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med. 1999, 18: 2693-708. 10.1002/(SICI)1097-0258(19991030)18:20<2693::AID-SIM235>3.0.CO;2-V.View ArticlePubMedGoogle Scholar
- Groenwold RH, Donders AR, Heijden van der GJ, Hoes AW, Rovers MM: Confounding of subgroup analyses in randomized trials. Arch Intern Med. 2009, 169: 1532-4. 10.1001/archinternmed.2009.250.View ArticlePubMedGoogle Scholar
- Rovers MM, Glasziou P, Appelman CL, Burke P, McCormick DP, Damoiseaux RA, Gaboury I, Little P, Hoes AW: Antibiotics for acute otitis media: a meta-analysis with individual patient data. Lancet. 2006, 368: 1429-35. 10.1016/S0140-6736(06)69606-2.View ArticlePubMedGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2288/10/43/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.