SEARCH · Search PubMed
Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
The number of meta-analyses or overviews appearing in the literature is rapidly increasing. Because of the problems inherent in non-randomised studies, which are not diminished by using meta-analysis techniques, this paper only refers to meta-analyses of randomised trials. Examples are given of the value of meta-analyses of randomised trials in many different circumstances and the different methods that have been used are discussed. The most reliable type of meta-analysis uses individual patient data and includes all trials fitting the pre-defined criteria, published and unpublished. As with a single randomised trial, large numbers of randomised patients are needed to give reliable results. Well-conducted meta-analyses or overviews are the best method of summarising all available unbiased evidence on the relative effects of treatments.
BACKGROUND: Colloid solutions are widely used in fluid resuscitation of critically ill patients. There are several choices of colloid and there is ongoing debate about the relative effectiveness of colloids compared to crystalloid fluids. OBJECTIVES: To assess the effects on mortality of colloids compared to crystalloids for fluid resuscitation in critically ill patients. SEARCH STRATEGY: We searched the Injuries Group specialised register, Cochrane Controlled Trials Register, MEDLINE, EMBASE and BIDS Index to Scientific and Technical Proceedings, and checked reference lists of trials and review articles. SELECTION CRITERIA: All randomised and quasi-randomised trials of colloids compared to crystalloids, in patients requiring volume replacement. Cross-over trials and trials in pregnant women and neonates were excluded. DATA COLLECTION AND ANALYSIS: Two reviewers independently extracted data and rated quality of allocation concealment. Trials with a 'double-intervention', such as those comparing colloid in hypertonic crystalloid to isotonic crystalloid, were analysed separately. The analysis was stratified according to colloid type and quality of allocation concealment. MAIN RESULTS: Colloids compared to crystalloidsAlbumin or plasma protein fraction. Nineteen trials reported data on mortality, including a total of 7576 patients. The pooled relative risk (RR) from these trials was 1.02 (95% confidence interval [95% CI] 0.93 to 1.11). When the trial with poor quality allocation concealment was excluded, pooled RR was 1.01 (95% CI 0.92 to 1.10). Hydroxyethyl starch. Ten trials compared hydroxyethyl starch with crystalloids, including a total of 374 randomised participants. The pooled RR was 1.16 (95% CI 0.68 to 1.96). Modified gelatin. Seven trials compared modified gelatin with crystalloid, including a total of 346 randomised participants. The pooled RR was 0.54 (95% CI 0.16 to 1.85). Dextran. Nine trials compared dextran with a crystalloid, including a total of 834 randomised participants. The pooled relative risk was RR 1.24 (95% CI 0.94 to 1.65). Colloids in hypertonic crystalloid compared to isotonic crystalloidEight trials compared dextran in hypertonic crystalloid with isotonic crystalloid, including 1283 randomised participants. Pooled RR was 0.88 (95% CI 0.74 to 1.05). REVIEWERS' CONCLUSIONS: There is no evidence from randomised controlled trials that resuscitation with colloids reduces the risk of death, compared to resuscitation with crystalloids, in patients with trauma, burns or following surgery. As colloids are not associated with an improvement in survival, and as they are more expensive than crystalloids, it is hard to see how their continued use in these patients can be justified outside the context of randomised controlled trials.
BACKGROUND: Lambert-Eaton myasthenic syndrome is an autoimmune presynaptic disorder of neuromuscular transmission. Treatments have attempted to overcome the harmful autoimmune process, or to improve residual neuromuscular transmission, in order to reverse the principal neurological symptom of muscle weakness. OBJECTIVES: The objective was to examine the efficacy of all forms of treatment in Lambert-Eaton myasthenic syndrome. SEARCH STRATEGY: We searched the Cochrane Neuromuscular Disease Group specialised trials register (September 2002), MEDLINE (January 1966 to November 2002) and EMBASE (January 1980 to November 2002). We checked the bibliographies in reports of the randomised trials and contacted authors to identify additional published or unpublished data. SELECTION CRITERIA: Types of studies: all randomised or quasi-randomised trials. TYPES OF PARTICIPANTS: all adults and children with a diagnosis of Lambert-Eaton myasthenic syndrome, with or without small-cell lung cancer. Types of interventions: any form of medical (pharmacological or physical) treatment. Types of outcome measures: Primary: change in the muscle strength scale score (Quantitative Myasthenia Gravis score), or limb muscle strength measured by myometry. Secondary: improvement in the mean amplitude of the resting compound muscle action potentials. The mean amplitude used was the mean of all muscles tested. DATA COLLECTION AND ANALYSIS: We identified three randomised controlled trials. Individual patient data were only available for one trial. MAIN RESULTS: The three eligible trials included two controlled trials of the effects of 3,4-diaminopyridine compared with placebo in a total of 38 patients with Lambert-Eaton myasthenic syndrome, one of which was of crossover design. A third crossover trial compared intravenous immunoglobulin treatment to placebo in nine patients with Lambert-Eaton myasthenic syndrome. A meta-analysis of the primary endpoint results of these trials was not possible because of differences in comparisons and endpoints and, in two trials, lack of individual patient data. EFFECTS OF 3,4-DIAMINOPYRIDINE: Two trials of 3,4-diaminopyridine reported a significant improvement in the primary endpoint of muscle strength score, or myometric limb measurement following treatment. Both trials also reported a significant improvement in the secondary endpoint of resting compound muscle action potential amplitude following 3,4-diaminopyridine, compared with placebo. A meta-analysis of the primary endpoint results was not possible because of marked differences in these two trials regarding primary outcome measures. However, a meta-analysis of the secondary endpoint (improvement in the amplitude of the mean resting compound muscle action potential) was possible. It was necessary to assume a known correlation (similarity) of the paired responses for each individual in the two treatment periods in order to properly allow for the crossover design of one of the two trials (the correlation coefficient was assumed to be 0.5 in calculations). Using this approach, meta-analysis revealed a significant overall benefit in compound muscle action potential amplitude after 3,4-diaminopyridine treatment. The overall weighted mean difference was 1.80 mV (95% confidence interval 0.82 to 2.78), favouring treatment. These results were not sensitive to the assumption made because the overall benefit estimated was still significant when the correlation was assumed to be less than 0.1. EFFECTS OF INTRAVENOUS IMMUNOGLOBULIN: A crossover trial reported a significant improvement in the primary outcome measure of myometric limb strength when patients received intravenous immunoglobulin compared to placebo infusions. This trial also demonstrated an improvement in the secondary outcome measure of change in the mean resting compound muscle action potential amplitude following intravenous immunoglobulin, but this improvement did not reach significance. Clinical improvement lasted for up to eight weeks. REVIEWER'S CONCLUSIONS: Limited evidence from randomised controlled trials showed that either 3,4-diaminopyridine or intravenous immunoglobulin improved muscle strength scores and compound muscle action potential amplitudes in patients with Lambert-Eaton myasthenic syndrome. There are insufficient data at present to quantify this treatment effect. Other possible treatments, such as plasma exchange, steroids and immunosuppressive agents have not been tested in randomised controlled trials.
BACKGROUND: Pelvic organ prolapse is common, and some degree of prolapse is seen in 50% of parous women. Women with prolapse can experience a variety of pelvic floor symptoms. Treatments include surgery, mechanical devices and conservative management. Conservative management approaches, such as giving lifestyle advice and delivering pelvic floor muscle training, are often used in cases of mild to moderate prolapse. OBJECTIVES: To determine the effects of conservative management (physical interventions and lifestyle interventions) for women with pelvic organ prolapse in comparison with no treatment or other treatment options (such as mechanical devices or surgery). SEARCH STRATEGY: We searched the Cochrane Incontinence Group Specialised Trials Register (searched on 19 September 2005), MEDLINE (January 1966 to August 2005), MEDLINE In Process & Other Citations (15 September 2005), EMBASE (January 1996 to Week 43 2005), CINAHL (January 1982 to October 2005), PEDro (September 2005), the UK National Research Register (Issue 3, 2005), the US National Institute of Health clinical trial register (5 October 2005), Current Controlled Trials register (5 October 2005), Controlled Clinical Trials (September 2005) and ZETOC (September 2005). We searched the reference lists of relevant articles. SELECTION CRITERIA: Randomised and quasi-randomised trials in women with pelvic organ prolapse that included a physical or lifestyle intervention in at least one arm of the trial. DATA COLLECTION AND ANALYSIS: Two reviewers assessed all trials for inclusion/exclusion and methodological quality. Data were extracted by the lead reviewer onto a standard form and cross checked by another. Disagreements were resolved by discussion. Data were processed as described in the Cochrane Handbook for Systematic Reviews of Interventions. MAIN RESULTS: Three trials of relevance to this review were identified. The largest of these, of pelvic floor muscle training in preventing anterior prolapse from worsening, had significant limitations which affect the generalisability and rigor of the findings. A small feasibility study (which is to be followed up with a larger trial) randomised 47 women to pelvic floor muscle training or control and found suggestions of better outcomes (better self-reported improvement, decreased severity) in the intervention group. The third trial evaluated peri-operative physiotherapy for women undergoing surgery for prolapse and/or incontinence. The authors report that urinary symptoms, pelvic floor muscle function and quality of life were improved more in the treatment group than the control group, but data were not provided to allow this to be assessed. The trial was small and no prolapse-specific outcome measures were used. It was not possible to combine data from the three trials. AUTHORS' CONCLUSIONS: Despite there now being reports of three eligible trials in this update, the evidence available is not significant to guide practice. There is some encouragement from a feasibility study that pelvic floor muscle training, delivered by a physiotherapist to symptomatic women in an outpatient setting, may reduce severity of prolapse. Further evidence from larger, better quality randomised control trials is however still necessary.
BACKGROUND: Pelvic organ prolapse may occur in up to 50% of parous women. A variety of urinary, bowel and sexual symptoms may be associated with prolapse. OBJECTIVES: To determine the effects of surgery in the management of pelvic organ prolapse. SEARCH STRATEGY: We searched the Cochrane Incontinence Group trials register (8 June 2004) and reference lists of relevant articles. We also contacted researchers in the field. SELECTION CRITERIA: Randomised or quasi-randomised controlled trials that included surgical operations for pelvic organ prolapse. DATA COLLECTION AND ANALYSIS: Trials were assessed and data extracted independently by at least two reviewers. Four investigators were contacted for additional information with two responding. MAIN RESULTS: Fourteen randomised controlled trials were identified evaluating 1004 women. Abdominal sacral colpopexy was better than vaginal sacrospinous colpopexy in terms of a lower rate of recurrent vault prolapse (RR 0.23, 95% CI 0.07 to 0.77) and less dyspareunia (RR 0.39, 95% CI 0.18 to 0.86), but the trend towards a lower re-operation rate for prolapse following abdominal sacrocolpopexy was not statistically significant (RR 0.46, 95% CI 0.19 to 1.11). However, the vaginal sacrospinous colpopexy was quicker and cheaper to perform and women had an earlier return to activities of daily living. The data were to evaluate other clinical outcomes and adverse events. For the anterior vaginal wall prolapse, standard anterior repair was associated with more recurrent cystoceles than when supplemented by Vicryl mesh overlay (RR 1.39, 95% CI 1.02 to 1.90) but data on morbidity and other clinical outcomes were too few for reliable comparisons. For posterior vaginal wall prolapse, the vaginal approach was associated with a lower rate of recurrent rectocele and/or enterocele than the transanal approach (RR 0.24, 95% CI 0.09 to 0.64), although there was a higher blood loss and postoperative narcotic use. However, data on the effect of surgery on bowel symptoms and the use of polyglactin mesh overlay on the risk of recurrent rectocele were insufficient for meta-analysis.Meta-analysis on the impact of pelvic organ prolapse surgery on continence issues was limited and inconclusive, although about 10% of women developed new symptoms after surgery. However, more women with occult stress urinary incontinence developed postoperative stress urinary incontinence after endopelvic fascia plication alone than after endopelvic fascia plication and tension-free vaginal tape (RR 5.5, 95% CI 1.36 to 22.32). REVIEWERS' CONCLUSIONS: Abdominal sacrocolpopexy is associated with a lower rate of recurrent vault prolapse and dyspareunia than the vaginal sacrospinous colpopexy. These benefits must be balanced against a longer operating time, longer time to return to activities of daily living and increased cost of the abdominal approach. The use of a polyglactin mesh overlay at the time of anterior vaginal wall repair may reduce the risk of recurrent cystocele. Posterior vaginal wall repair may be better than transanal repair in the management of rectoceles in terms of recurrence of prolapse. Adequately powered randomised controlled clinical trials are urgently needed.
The objective was to investigate the safety and efficacy of almotriptan in patients aged 11-17 years old with acute migraine. Fifteen patients aged 11-17 with a history of migraine with or without aura were treated with almotriptan. Reduction in headache severity, disability and adverse effects were studied. Almotriptan in doses ranging from 6.25 to 12.5 mg was well tolerated. There were virtually no adverse effects except for one case of transient mild stiffness. Of the 15 patients, only 2 demonstrated no efficacy without adverse effects. In the other 13 patients, not only was almotriptan effective, but again, no significant adverse effects were reported. Almotriptan is probably safe and effective in patients aged 11-17. This small open-label pilot study should support the feasibility of a large randomised controlled study to demonstrate tolerability and efficacy of almotriptan in children and adolescents with episodic migraine.
AIMS/BACKGROUND: Chronic hepatitis B is a serious health problem worldwide. Chinese medicinal herbs are widely used for treatment of chronic hepatitis B in China and many clinical trials have been conducted. This systematic review is to assess the efficacy and safety of Chinese medicinal herbs for chronic hepatitis B. METHODS: Randomised clinical trials comparing Chinese medicinal herbs versus placebo, no intervention, nonspecific treatment, or interferon treatment for chronic hepatitis B with > or = 3 months follow-up were included. No language and blinding limitations were applied. The electronic databases were searched, combined with handsearches on Chinese literature. Data were extracted independently by two reviewers. The methodological quality of trials was assessed by the Jadad-scale plus allocation concealment. RESULTS: Nine randomised trials (n=936) were included, with only one being of high quality. There was a funnel plot asymmetry (intercept 3.37, p=0.047). Compared to nonspecific treatment or placebo, the herbal compound Fuzheng Jiedu Tang showed an effect on clearance of serum HBsAg (relative risk 5.19, 95% CI 1.24-21.79), HBeAg (10.85, 3.56-33.06), and HBV DNA (8.50, 1.23-58.85). Polyporus umbellatus polysaccharide showed an effect on serum HBeAg (3.06, 1.13-8.29) and HBV DNA (4.14, 1.0-17.19); Phyllanthus amarus showed an effect on serum HBeAg (3.35, 1.49-7.56). Phyllanthus compound and kurorinone showed no significant difference on clearance of serum HBeAg and HBV DNA and on alanine aminotransferase normalisation compared to interferon. No serious adverse event was observed. CONCLUSIONS: Chinese medicinal herbs are not recommended for chronic hepatitis B because of the publication bias and low quality of the trials. Rigorously designed, randomised, double-blind, placebo-controlled trials are needed.
Large, simple trials can provide reliable evidence on the balance of risk and benefit of widely practicable treatments that have moderate effects on major clinical outcomes. Treatments for stroke will potentially be used extensively world-wide. Each year about 8,000,000 people have an acute stroke, with about 5,000,000 being ischaemic. Of these, perhaps 1,000,000 receive medical attention within the first few hours of stroke onset. If reliable decisions about the treatment of such large numbers of patients are to be made, we need to be certain about the effects of treatment on death from any cause (and on other important outcomes) and in many different types of patient. Generally, such reliable evidence on whether to treat and on whom to treat, will only come from large, randomised trials recruiting many thousands of patients. Such trials are only feasible and affordable if they are simple.
Explore the source record for details and available documents.
BACKGROUND: Tardive dyskinesia is a disfiguring movement disorder of the orofacial region often caused by antipsychotic drugs. A wide range of strategies has been used to help manage tardive dyskinesia and, for people who are unable to have their antipsychotic medication stopped or substantially changed, the calcium-channel blocking group of drugs (diltiazem, nifedipine, nimodipine, verapamil) has been suggested as a useful adjunctive treatment. OBJECTIVES: To determine the effects of calcium-channel blocker drugs (diltiazem, nifedipine, nimodipine, verapamil) for treatment of neuroleptic-induced tardive dyskinesia in people with schizophrenia, schizoaffective disorder or other chronic mental illnesses. SEARCH STRATEGY: We updated previous searches of the Cochrane Schizophrenia Group Register (1982-2000), Cochrane Library (Issue 4, 2000), Cochrane Schizophrenia Group's register of trials (November 2000), EMBASE (1980-2000), LILACS (1982-2000), MEDLINE (1966-2000), PsycLIT (1974-2000), and SCISEARCH by searching the Cochrane Schizophrenia Group Register (September 2003). We searched references of all identified studies for further trial citations and contacted authors of trials. SELECTION CRITERIA: Randomised clinical trials comparing calcium-channel blockers to placebo or no intervention for people with both tardive dyskinesia and schizophrenia or serious mental illness. DATA COLLECTION AND ANALYSIS: Data were to have been independently extracted and analysed on an intention-to-treat basis. The relative risk (RR) and 95% confidence intervals (CI) of homogeneous dichotomous data were to have been calculated using a random effects model, and, where possible, the number needed to treat calculated. Weighted mean differences (WMD) were to have been calculated for continuous data. MAIN RESULTS: No trials were included. We excluded fourteen studies; eight were not randomised, one did not use calcium channel blockers and five small, randomised, studies reported no usable data. REVIEWER'S CONCLUSIONS: The effects of calcium-channel blockers for antipsychotic induced tardive dyskinesia are unknown. Their use is experimental and should only be given in the context of well designed randomised studies.
From a review of thirty published papers it is clear that the variety of options for treatment of hepatitis B have led to a large number of small exploratory trials. However, small groups lead to imprecise estimates of seroconversion rates; for example a result of 4/10 gives an estimate of 40% with 80% confidence limits of 20% and 60%. A number of possible prognostic factors have been proposed and, whilst their importance is generally unproven, their estimated effects are considerable. Comparison between studies can therefore be misleading; for example comparing two treatments A and B of identical efficacy in two common populations could give a 1 in 4 chance of declaring B to give double the seroconversion rate of A. Larger, randomised, controlled trials on homogeneous patient groups are needed to give conclusive evidence of the efficacy or otherwise of various treatment regimes.
The odds ratio is very useful; when used as a measure of association (or to describe the results of a randomised controlled treatment trial) it expresses both the strength and direction of the association (or size and direction of treatment effect). Its 95% confidence interval estimates the likely degree of sampling error and provides a test of significance at the 5% level. Authors, when reporting the results of descriptive studies, case-control studies and randomised controlled trials, should consider presenting their results this way rather than by simple significance testing with a chi squared test.
Psoriatic arthritis (PsA) is a destructive form of inflammatory arthritis that occurs in about one-third of patients with psoriasis. The pathogenesis of PsA includes genetic and immunological factors. A review of the currently available therapies reveals that traditional disease-modifying medications have provided only marginal relief from joint inflammation in patients with PsA, and have not been successful in controlling the disease and preventing joint damage. On the basis of current understanding of the pathogenesis of joint destruction in PsA, several new medications have been introduced, including anti-tumour necrosis factor (TNF) agents and agents that interfere with T-cell function. Most of these medications have been found to be effective in both psoriasis and PsA. Recent randomised controlled trials suggest that at least anti-TNF agents may help prevent progression of joint destruction.
OBJECTIVE: To determine if the H1-receptor antagonist dimethindene maleate (DMM), topically applied, is able to improve the quality of life (QoL) in patients with seasonal allergic rhinitis better than placebo. METHODS: The study was a multi-centre, randomised, placebo-controlled, double-blinded phase III trial including 157 patients. Two parallel groups received either DMM nasal spray or placebo for 2 weeks. Patients answered 5 QoL questionnaires (Rhinoconjunctivitis QoL Questionnaire (LQ-AR), Munich Life Dimension List (MLDL), Profile of Mood States (POMS), 2 Visual Analogue Scales (VAS-QoL, VAS-GES)) on days 1, 4, 8, and 15. RESULTS: QoL improved significantly in all scales and groups. Statistically significant differences between groups were achieved in the sub-scales LQ-AR Eye Symptoms and Daily Activities, MLDL Total, POMS Depression and Energy, VAS-QoL and VAS-General Health. All differences favoured the DMM group and were present initially but disappeared in visit 4, except MLDL Total and POMS Energy. Almost all other scores had a better tendency of DMM. CONCLUSION: A significant improvement of QoL was found in both groups. Only weak differences between groups (in favour of DMM) were found. There are several reasons explaining this outcome, mainly the rainy weather during the study period (mean weather index 10-20% less sunny than in previous seasons).
BACKGROUND: Numerous types of arthroplasties may be used in the surgical treatment of a hip fracture (proximal femoral fracture). The main differences between the implants are the design of the stems, whether the stem is fixed in place with or without cement, whether a second articulating joint is included within the prosthesis (bipolar prosthesis) or whether the whole hip joint is replaced. OBJECTIVES: To review all randomised trials that have compared different arthroplasties for the treatment of hip fractures in adults. SEARCH STRATEGY: We searched the Cochrane Musculoskeletal Injuries Group specialised register. Additional trials were identified by searching reference lists of relevant articles, conference proceedings, and contact with trialists. Date of most recent search: January 2001. SELECTION CRITERIA: All randomised and quasi-randomised trials comparing different arthroplasties (and or cement), for the treatment of hip fractures. DATA COLLECTION AND ANALYSIS: Two reviewers independently assessed trial quality, by use of a ten-item checklist and extracted data. MAIN RESULTS: Thirteen trials involving 1464 patients were included. One trial investigated two comparisons. Cemented prostheses, when compared with uncemented (four trials, 391 participants) were associated with a lower risk of failure to regain mobility (relative risk (RR) 0.60, 95% confidence interval (CI) 0.44, 0.82) and of post-operation pain at a year or later (RR 0.51, 95% CI 0.31, 0.81). For this comparison, there were no significant differences in any other outcome. Comparison of unipolar hemiarthroplasty with bipolar hemiarthroplasty (six trials, 742 participants) showed no significant differences between the two types of implant. Two trials of 269 patients compared different types of hemiarthroplasty with a total hip replacement and two trials of 151 patients compared either different types of prosthesis head or different bipolar prostheses. Because of the limited number of cases and the use of different prostheses, no definite conclusions could be made from these four studies. REVIEWER'S CONCLUSIONS: Cementing prostheses in place seems to reduce pain post-operatively and results in better mobility, but because of the under-reporting of outcomes and the small number of patients involved, no definite conclusions can be made. The role of bipolar prostheses and total hip replacement is uncertain. Further well-conducted randomised trials are required.
BACKGROUND: Oxidative stress may cause gastrointestinal cancers. The evidence on whether antioxidant supplements are effective in preventing gastrointestinal cancers is contradictory. OBJECTIVES: To assess the beneficial and harmful effects of antioxidant supplements in preventing gastrointestinal cancers. SEARCH STRATEGY: We identified trials through the trials registers of the four Cochrane Review Groups on gastrointestinal diseases, The Cochrane Central Register of Controlled Trials on The Cochrane Library (Issue 1, 2003), MEDLINE, EMBASE, LILACS, and SCI-EXPANDED from inception to February 2003, and The Chinese Biomedical Database (March 2003). We scanned reference lists and contacted pharmaceutical companies. SELECTION CRITERIA: Randomised trials comparing antioxidant supplements to placebo/no intervention examining the incidence of gastrointestinal cancers. DATA COLLECTION AND ANALYSIS: Two reviewers independently selected trials for inclusion and extracted data. The outcome measures were incidence of gastrointestinal cancers, overall mortality, and adverse events. Outcomes were reported as relative risks (RR) with 95% confidence interval (CI) based on fixed and random effects meta-analyses. MAIN RESULTS: We identified 14 randomised trials (170,525 participants), assessing beta-carotene (9 trials), vitamin A (4 trials), vitamin C (4 trials), vitamin E (5 trials), and selenium (6 trials). Trial quality was generally high. Heterogeneity was low to moderate. Neither the fixed effect (RR 0.96, 95% CI 0.88 to 1.04) nor random effects meta-analyses (RR 0.90, 95% CI 0.77 to 1.05) showed significant effects of supplementation with antioxidants on the incidences of gastrointestinal cancers. Among the seven high-quality trials reporting on mortality (131,727 participants), the fixed effect (RR 1.06, 95% CI 1.02 to 1.10) unlike the random effects meta-analysis (RR 1.06, 95% CI 0.98 to 1.15) showed that antioxidant supplements significantly increased mortality. Two low-quality trials (32,302 participants) found no significant effect of antioxidant supplementation on mortality. The difference between the mortality estimates in high- and low-quality trials was significant by test of interaction (z = 2.10, P = 0.04). Beta-carotene and vitamin A (RR 1.29, 95% CI 1.14 to 1.45) and beta-carotene and vitamin E (RR 1.10, 95% CI 1.01 to 1.20) significantly increased mortality, while beta-carotene alone only tended to do so (RR 1.05, 95% CI 0.99 to 1.11). Increased yellowing of the skin and belching were non-serious adverse effects of beta-carotene. In four trials (three with unclear/inadequate methodology), selenium showed significant beneficial effect on gastrointestinal cancer incidences. REVIEWERS' CONCLUSIONS: We could not find evidence that antioxidant supplements prevent gastrointestinal cancers. On the contrary, they seem to increase overall mortality. The potential cancer preventive effect of selenium should be studied in adequately conducted randomised trials.