Showing papers by "Peter C Gøtzsche published in 2005"

Comparison of descriptions of allocation concealment in trial protocols and the published reports: cohort study

Abstract: Objectives To compare how allocation concealment is described in publications of randomised clinical trials and corresponding protocols, and to estimate how often trial publications with unclear allocation concealment have adequate concealment according to the protocol. Design Cohort study of 102 sets of trial protocols and corresponding publications. Setting Protocols of randomised trials approved by the scientific and ethical committees for Copenhagen and Frederiksberg, 1994 and 1995. Main outcome measures Frequency of adequate, unclear, and inadequate allocation concealment and sequence generation in trial publications compared with protocols, and the proportion of protocols where methods were reported to be adequate but descriptions were unclear in the trial publications. Results 96 of the 102 trials had unclear allocation concealment according to the trial publication. According to the protocols, 15 of these 96 trials had adequate allocation concealment (16%, 95% confidence interval 9% to 24%), 80 had unclear concealment (83%, 74% to 90%), and one had inadequate concealment. When retrospectively defined loose criteria for concealment were applied, 83 of the 102 trial publications had unclear concealment. According to their protocol, 33 of these 83 trials had adequate allocation concealment (40%, 29% to 51%), 49 had unclear concealment (59%, 48% to 70%), and one had inadequate concealment. Conclusions Most randomised clinical trials have unclear allocation concealment on the basis of the trial publication alone. Most of these trials also have unclear allocation concealment according to their protocol.

Physical activity for secondary prevention of disease. Systematic reviews of randomised clinical trials.

Abstract: Background Physical activity is recommended for secondary prevention of several diseases but it is not always clear how reliable the evidence is. Methods We searched MEDLINE and The Cochrane Library for systematic reviews of randomised clinical trials published 1998-2004. Results We identified 30 eligible systematic reviews and excluded 13 that contained trials covered in larger reviews or were older than other reviews on the same subject. Physical activity decreased all-cause mortality in patients with coronary heart disease, odds ratio 0.73 (95% confidence interval 0.54 to 0.98), increased maximum walking time in patients with intermittent claudication by 6.5 min (4.4 to 8.7), and decreased pain in patients with osteoarthritis of the knee, standardised mean difference 0.34 (0.24 to 0.44). There were positive effects also in heart failure, chronic obstructive lung disease, type 2 diabetes and fibromyalgia, but they need confirmation in high-quality trials. Exercise improved quality of life in several conditions and generally led to improved physical performance. An effect was not shown in stroke, asthma, rheumatoid arthritis, acute or chronic low back pain, chronic fatigue syndrome, depression, cystic fibrosis or HIV/AIDS. The occurrence of harms was generally not reported. Conclusion Physical activity can have important, and even life-saving, effects as secondary prevention of disease, but more and better trials are needed to fully assess its benefits and harms, in particular trials that compare exercise with drugs.

Short-term low-dose corticosteroids vs placebo and nonsteroidal antiinflammatory drugs in rheumatoid arthritis.

Abstract: BACKGROUND The effect of low dose corticosteroids, equivalent to 15 mg prednisolone daily or less, in patients with rheumatoid arthritis has been questioned. We therefore performed a systematic review of trials which compared corticosteroids with placebo or non-steroidal, anti-inflammatory drugs. OBJECTIVES To determine whether short-term (i.e. as recorded within the first month of therapy), oral low-dose corticosteroids (corresponding to a maximum of 15 mg prednisolone daily) is superior to placebo and nonsteroidal, antiinflammatory drugs in patients with rheumatoid arthritis. SEARCH STRATEGY Medline Silverplatter, The Cochrane Controlled Trials Register, reference lists and a personal archive. SELECTION CRITERIA All randomised studies comparing an oral corticosteroid (not exceeding an equivalent of 15 mg prednisolone daily) with placebo or a nonsteroidal, antiinflammatory drug were eligible if they reported clinical outcomes within one month after start of therapy. DATA COLLECTION AND ANALYSIS Decisions on which trials to include were made independently by two observers based on the methods sections of the trials only. Standardised effect measures were used for the statistical analyses; the random effects model was used if P<0.10 for the test of heterogeneity. MAIN RESULTS Ten studies, involving 320 patients, were included in the meta-analysis. Prednisolone had a marked effect over placebo on joint tenderness (standardised effect size 1.31, 95% confidence interval 0.78 to 1.83), pain (standardised effect size 1.75, 0.87 to 2.64) and grip strength (standardised effect size 0.41, 0.13 to 0.69). Measured in the original units, the differences were 12 tender joints (6 to 18) and 22 mm Hg (5 to 40) for grip strength. Prednisolone also had a greater effect than nonsteroidal, antiinflammatory drugs on joint tenderness (standardised effect size 0.63, 0.11 to 1.16) and pain (standardised effect size 1.25, 0.26 to 2.24), whereas the difference in grip strength was not significant (standardised effect size 0.31, -0.02 to 0.64). Measured in the original units, the differences were 9 tender joints (5 to 12) and 12 mm Hg (-6 to 31). The risk of adverse effects, also during moderate- and long-term use, seemed acceptable. REVIEWER'S CONCLUSIONS Prednisolone in low doses (not exceeding 15 mg daily) may be used intermittently in patients with rheumatoid arthritis, particularly if the disease cannot be controlled by other means. Since prednisolone is highly effective, short-term placebo controlled trials studying the clinical effect of low-dose prednisolone or other oral corticosteroids are no longer necessary.

Registries and registration of clinical trials

Abstract: The arguments in favor of the registration of clinical trials are now familiar.1–4 Chief among these addresses the practice of selective reporting, whereby negative or detrimental studies are not brought into the public domain, which experts on the subject of clinical trials consider an important form of scientific misconduct.5 This practice, as illustrated in a number of high-profile examples, have increased the demand for the mandatory public registration of clinical trials. Registration of trials should improve the completeness, reliability, and quality of the interpretation of clinical research. This need for registration prompts the question of where trials should be . . .

Research integrity and pharmaceutical industry sponsorship.

Abstract: Trial registration, transparency and less reliance on industry trials are essential.

Reduction in mortality from breast cancer: Presentation of benefits and harms needs to be balanced

Abstract: Editor—Olsen et al found a 25% reduction in breast cancer mortality in Copenhagen compared with what they would expect to find in the absence of screening.1 They did an observational study, which is not considered to be a reliable method for evaluating mortality reductions with screening. The full mortality reduction was seen after only three years of follow-up, where it nearly reached significance, and it stayed at that level for the next seven years. However, both randomised trials and cohort studies have shown clearly that a positive effect of screening does not come that quickly. It therefore seems likely that the study has provided an exaggerated mortality benefit. Olsen et al did not provide data on harms but refer to another of their studies when they claim that the introduction of mammography screening in Copenhagen did not lead to an increase in the incidence of breast cancer apart from the expected prevalence peak. Their conclusion in that study is even stronger since they say that mammography screening can operate without over-diagnosis. However, their data do not support the conclusion,2 and much larger studies, both randomised and non-randomised, have shown that the level of over-diagnosis is about 30%, or even more.3-5 Proper evaluation of findings about any medical intervention, in particular cancer screening of healthy people, requires an honest, balanced, and unbiased presentation of major benefits and harms.

[Selective reporting of positive outcomes in randomised trials--secondary publication.. A comparison of protocols with published reports].

Abstract: Introduction Selective reporting of positive outcomes is associated with bias. Materials and methods We conducted a cohort study of 102 protocols of randomised trials approved by the Scientific Ethics Committees for Copenhagen and Frederiksberg in 1994-1995 and the corresponding 122 reports. Results The median proportion of incompletely reported outcomes per trial was 50%. Statistically significant outcomes were more often fully reported, odds ratio 2.4. In 62% of trials, at least one primary outcome was changed. Discussion Outcomes in randomised trials are often reported selectively.

Model of outcomes of screening mammography: Information needs to support informed choices.

Abstract: EDITOR—Barratt et al report that for every 1000 women screened from the age of 50 over 10 years, two fewer will die from breast cancer and 13 more will be diagnosed with breast cancer (corresponding to 63% more cancers, which mostly constitute overdiagnosis).1 Similarly, we found that for every woman who …

Response to Meissner: our preference of patient-reported to observer-reported outcomes has trivial influence on our updated systematic review of the effect of placebo

Abstract: Dear Sir, Meissner claims that our preference for patientreported outcomes substantially influences the result of our updated review of the effect of placebo [1]. This is not correct. Our main analysis of observer-reported continuous outcomes was based on 43 trials with 2254 patients and gave a pooled standardized mean difference of )0.10 (95% confidence interval )0.20 to 0.01) [2]. The lower confidence interval is close to zero, implying that the observed difference between placebo and no treatment is close to being statistically significant (P 1⁄4 0.06). We also identified 14 trials with corresponding patientand observer-reported outcomes, of which 13 had been included in the main analyses of patient-reported outcomes, according to our prepublished protocol [3]. When Meissner reversed this decision post-hoc, and pooled the additional trials with the trials from the primary analysis of observerreported outcomes, the overall standardized mean difference unsurprisingly reached statistically significance, )0.16 ()0.26 to )0.06). But more important than statistical significance is the size of the effect measure, which was quite similar )0.10 vs. )0.16. Thus, the overall point estimate is low, and fairly insensitive to whether or not the additional trials are included. However,Meissner’s inclusion of the 13 trials in the main analysis of observer-reported outcomes is inappropriate because it is a post-hoc analysis conducted after having observed the somewhat larger effect in the subgroup of trials with corresponding patientreported outcomes. We decided a priori to prefer patient-reported outcomes in the main analyses as such outcomes are often more relevant to patients, and this analysis should therefore be the most important when interpreting the result. Furthermore, Meissner claims that we have suggested that effects of placebo interventions on objective outcomes is a myth ; and refer to our main result as there is no objective placebo effect . We have never made such suggestions. To find no evidence of an effect is not the same as evidence of no effect, and we cautiously concluded that we did not find a statistically significant pooled effect of placebo on observer-reported outcomes. Finally, observing a small tendency for improvement in the placebo group, even assuming no effect of placebo on observer-reported outcomes, is to be expected. The comparison between placebo and no-treatment is not blinded, and thus susceptible to various forms of bias, for example observation bias [4]. It is possible that further updates will produce enough statistical power to rule out the observed minor improvement as a random event. However, a possible true minor effect of placebo on observer-reported outcomes has to be distinguished from the probable bias induced by lack of blinding.