Serum cholesterol reduction and risk of ischaemic heart disease

The randomised controlled trials of serum cholesterol reduction have been the subject of a number of meta-analyses1419 20 and much controversy. In conjunction with the review of the 10 prospective studies just described, the results of 28 randomised trials available in 1994 were summarised 14 this omits the results of trials of serum cholesterol reduction, notably those using statins, that have become available more recently. The aim was to quantify the effect of serum cholesterol reduction on...

The Steering Group

The official membership of the Collaboration consists of all registered entities. Each entity, in turn, determines its own membership, who is eligible to vote for candidates to represent that type of entity on the Collaboration's Steering Group. The Steering Group has fourteen elected members, who meet twice a year, once during the annual Cochrane Colloquia and on one other occasion. The minutes of its meetings are on the Collaboration's web sites (Box 25.2). The Steering Group has overall...

Meta Win version 20 httpwwwmetawinsoftcom

Data is entered in a spreadsheet, and Excel files may be imported directly. For each study summary statistics are calculated from the numbers of patients with and without disease, or the mean and standard deviation of the response in each group. Fixed and random effects models are available to combine these summary statistics, and additionally nonparametric resampling tests and confidence intervals can be computed.2 Cumulative meta-analysis and radial plots...

The dissemination of research findings

The dissemination of research findings is not a dichotomous event but a continuum ranging from the sharing of draft papers among colleagues, presentations at meetings, published abstracts to papers in journals that are indexed in the major bibliographic databases.31 It has long been recognised that only a proportion of research projects ultimately reach publication in an indexed journal and thus become easily identifiable for systematic reviews.32 Scherer et al.33 showed that only about half of...

Does smoking cause suicide

A large number of cohort studies have shown a positive association between smoking and suicide, with a dose-response relationship being evident between the amount smoked and the probability of committing suicide.13-18 Figure 12.1 illustrates this for four prospective studies of middle-aged men, including the massive cohort of men screened for the Multiple Risk Factor Intervention Trial (MRFIT).19 Based on over 390 000 men and almost five million years of follow-up, a meta-analysis of these...

Should unpublished data be included in systematic reviews

Publication bias clearly is a major threat to the validity of any type of review, but particularly of unsystematic, narrative reviews. Obtaining and including data from unpublished trials appears to be the obvious way of avoiding this problem. However, the inclusion of data from unpublished studies can itself introduce bias. The trials that can be located may be an unrepresentative sample of all unpublished studies. Unpublished trials may be of lower methodological quality than published trials...

Who is responsible for publication bias authors reviewers or editors

Trials with negative results could remain unpublished because authors fail to write them up and submit to journals, because such trials are reviewed less favourably, or because editors simply don't want to publish negative results. The peer review process is notoriously unreliable and susceptible to subjectivity, bias and conflict of interest.63,64 Experimental studies in which test manuscripts were submitted to reviewers or journals showed that reviewers are more likely to referee favourably...

Example 1 intravenous streptokinase in myocardial infarction

The following table gives data from 22 randomised controlled trials of streptokinase in the prevention of death following myocardial infarction.1-3 Trial Trial name Publication Intervention group Control group Trial Trial name Publication Intervention group Control group These data were saved in Stata dataset strepto.dta which is available from the book's website (http www.systematicreviews.com). We can list the variables contained in the dataset, with their descriptions (variable labels) by...

Identifying ongoing andor unpublished studies

Whilst it is difficult to identify published studies, identifying ongoing and or unpublished studies presents even more of a challenge. If a review is to minimise bias, it is important that unpublished studies are identified (for a discussion of publication bias and related biases see Chapter 11). The least biased way to identify studies is through the registration of ongoing research. Registers of ongoing trials have existed in a number of areas for some years. A Register of Registers was set...

Acknowledgements

We are grateful to Iain Chalmers for helpful comments on an earlier draft of this chapter. We thank Jim Neaton (MRFIT Research Group), Juha Pekkanen and Erkki Vartiainen (North Karelia and Kuopio Cohorts) and Martin Shipley (Whitehall Study) for providing additional data on suicides. This chapter draws on material published earlier in the BMJ.77 1 Surgeon General. Smoking and Health. Washington US Government Printing Office, 1964. 2 Feinstein AR. Scientific standards in epidemiological studies...

General statistical software which includes facilities for metaanalysis or for which metaanalysis routines have been

For readers who are already familiar with and who have access to a commercial statistical software package, using its facilities for meta-analysis is likely to be the most convenient way to perform meta-analysis. Details of general packages are given in Box 17.1. The packages marked with an asterix do not provide ready-made meta-analysis routines but programs for meta-analysis have been written by users and made freely available. While the addresses of some useful websites are given in Box 17.1...

Examples of how the collection of IPD can make a difference to the findings of a review

A number of projects comparing the results of reviews which used IPD with the results that would have been found with reviews based only on data abstracted from trial publications have been reported over the last few years. These are now the subject of a systematic review,15 but some examples are discussed below. An IPD review of 1329 patients in 11 trials of platinum-based combination chemotherapy versus single non-platinum drugs for ovarian cancer found little difference between the...

Case study Cathepsin D and diseasefree survival in nodenegative breast cancer

Ferrandina et al.33 described a meta-analysis of 11 studies which examined the relation between cathepsin D (a proteolytic enzyme) and disease-free survival (time to relapse) in node-negative breast cancer. Their inclusion criteria included the presentation of outcome separately for groups with high and low cathepsin D values. Studies were included if they had used a cytosol assay with a cutpoint in the range 20-78 pmol mg or a semiquantitative method based on histochemical assay. It is not...

A n RR

The standard error of the log risk ratio being The risk difference6 for each study is given by For the Peto odds ratio method7 (see below) the individual odds ratios are given by where E aJ nXi (a. + c) N. (the expected number of events in the intervention group under the null hypothesis of no treatment effect) and Individual study estimates of treatment effect continuous outcomes If the outcome is a continuous measure, we require the number of participants, the mean response and its standard...

Rare insight The protective effect of betacarotene that wasnt

Observational studies have consistently shown that people eating more fruits and vegetables, which are rich in beta-carotene, and people having higher serum beta-carotene concentrations have lower rates of cardiovascular disease and cancer.36 Beta-carotene has antioxidant properties and could thus plausibly be expected to prevent carcinogenesis and athero-genesis by reducing oxidative damage to DNA and lipoproteins. Unlike many other associations found in observational studies, this hypothesis...

Problems in risk stratification

When there have been many trials conducted in a particular field, it is possible to perform risk stratification at the level of individual trials. This was carried out in the case of cholesterol lowering,17 using the CHD mortality rate in the control arm of the trials as the stratification variable. This stratification is of clinical use, since this is the risk of CHD death of patients without treatment, i.e. the risk level which the clinician would want to use for deciding whether or not...

Example 3 trials of BCG vaccine against tuberculosis

The following table is based on a meta-analysis by Colditz et al.20 which examined the efficacy of BCG vaccine against tuberculosis. Trial Trial name Authors Start Latitude* Intervention Control Trial Trial name Authors Start Latitude* Intervention Control * Expressed in degrees from equator. * Expressed in degrees from equator. The data were saved in Stata dataset bcgtrial.dta. Scientists had been aware of discordance between the results of these trials since the 1950s. The clear heterogeneity...

Exploring sources of heterogeneity

Some observers suggest that meta-analysis of observational studies should be abandoned altogether.64 We disagree, but think that the statistical combination of studies should not, in general, be a prominent component of reviews of observational studies. The thorough consideration of possible sources of heterogeneity between observational study results will provide more insights than the mechanistic calculation of an overall measure of effect, which may often be biased. We re-analysed a number...

Determining the benefit risk ratio in an individual patient

If the results of a systematic review are judged applicable and feasible, the clinician must then evaluate the likely benefits and harms from the intervention. This involves two steps deriving clinically useful estimates of the overall results and extrapolating from the overall results to derive estimates for the individual patient. To illustrate this process, an example is outlined in Box 19.1 (adapted from Glasziou et al.4) and will be referred to below. Deriving clinically useful estimates...

Methods of metaanalysis of prognostic factor studies

Even if it is felt that a set of published studies is of good enough quality to attempt a formal meta-analysis, there are many potential barriers to success. In essence, we wish to compare the outcome for groups with different values of the prognostic variable. The method of analysis suitable for pooling values across several studies will depend on whether the prognostic variable is binary, categorical, or continuous. In principle, it should be relatively easy to combine data from studies which...

Funnel plots and tests for funnel plot asymmetry

The metabias command16,17 performs the tests for funnel-plot asymmetry proposed by Begg and Mazumdar18 and by Egger et al.11 (see Chapter 11). If the graph option is specified the command will produce either a plot of standardized effect against precision11 (graph(egger)) or a funnel plot (graph(begg)). For the magnesium data there is clear evidence of funnel plot asymmetry if the ISIS-4 trial is included. It is of more interest to know if there was evidence of bias before the results of the...

Why do we need systematic reviews A patient with myocardial infarction in 1981

A likely scenario in the early 1980s, when discussing the discharge of a patient who had suffered an uncomplicated myocardial infarction, is as follows a keen junior doctor asks whether the patient should receive a beta-blocker for secondary prevention of a future cardiac event. After a moment of silence the consultant states that this was a question which should be discussed in detail at the Journal Club on Thursday. The junior doctor (who now regrets that she asked the question) is told to...

Heterogeneity between study results

The thoughtful consideration of heterogeneity between study results is an important aspect of systematic reviews.30,31 As mentioned above, this should start when writing the review protocol, by defining potential sources of heterogeneity and planning appropriate subgroup analyses. Once the data have been assembled, simple inspection of the forest plot is informative. The results from the beta-blocker trials are fairly homogeneous, clustering between a relative risk of 0-5 and 1-0, with widely...

Language bias

Reviews are often exclusively based on trials published in English. For example, among 36 meta-analyses reported in leading English-language general medicine journals from 1991 to 1993, 26 (72 ) had restricted their search to studies reported in English.91 Investigators working in a non-English speaking country will, however, publish some of their work in local journals.92 It is conceivable that authors are more likely to report in an Box 3.5 Cholesterol lowering after myocardial infarction...

Performance and detection bias

Performance bias occurs if additional therapeutic interventions are provided preferentially to one of the comparison groups. Blinding of Box 5.2 Randomisation consists of two parts Adequate if resulting sequences are Adequate if patients and enrolling Inadequate if resulting sequences patients and care providers prevents this type of bias, and, in addition, safeguards against differences in placebo responses between comparison groups.18,24 Detection bias arises if the knowledge of patient...

Biased inclusion criteria

Once studies have been located and data obtained, there is still potential for bias in setting the inclusion criteria for a meta-analysis. If, as is usual, the inclusion criteria are developed by an investigator familiar with the area under study, they can be influenced by knowledge of the results of the set of potential studies. Manipulating the inclusion criteria could lead to selective inclusion of positive studies and exclusion of negative studies. For example, some meta-analyses of trials...

Easy MA version 99 httpwwwspcunivlyon1frmcueasyma

EasyMA is a DOS-based package which was developed by Michel Cucherat from the University of Lyon. All menu headings are written in English but contextual help is available only in French, however a paper-based manual translated into English and a paper4 describing the program are available. EasyMA was developed for meta-analysis of clinical trials with one or several dichotomous outcomes. It is menu driven and offers fixed effects (e.g. Mantel-Haenszel, Yusuf-Peto) and random effects models for...

What was the evidence in 1981 Cumulative metaanalysis

Myocardial Infarction Blocker

What conclusions would our junior doctor have reached if she had had access to a meta-analysis Numerous meta-analyses of trials examining the effect of beta-antagonists have been published since 1981.17,45-48 Figure 1.4 shows the results from the most recent analysis that included 33 randomised comparisons of beta-blockers versus placebo or alternative treatment in patients who had had a myocardial infarction.48 These trials were published between 1967 and 1997. The combined relative risk...

Case study 1 eradication of H pylori in nonulcer dyspepsia

Pylori Eradication

H. pylori is a bacterium that inhabits the stomach and has been linked to the development of peptic ulcer eradication of the bacterium with antibiotics is an effective cure for most ulcer disease. H. pylori is also considered to have a possible causal role in the development of non-ulcer dyspepsia. A meta-analysis of the five relevant trials reported a small reduction in dyspepsia rates 12 months after eradication, which was just statistically significant.19 The effect measure used in the...

Use of stratified analyses for investigating sources of heterogeneity

In a stratified analysis the trials are grouped according to a particular feature or characteristic and a separate meta-analysis carried out of the trials within each subgroup. The overall summaries calculated within each subgroup can then be inspected for evidence of variation in the effect of the intervention, which would suggest that the stratifying characteristic is an important source of heterogeneity and may moderate treatment efficacy. Stratified analysis can be used when the trials can...

Assessing trial quality the component approach

The analysis of individual components of study quality overcomes many of the shortcomings of composite scores. This component approach takes into account that the importance of individual quality domains, and the direction of potential biases associated with these domains, will vary between the contexts in which trials are performed. In the heparin example Figure 5.1 8,9 regression analyses showed that none of the 25 composite scales yielded a statistically significant difference in effect...