Showing papers on "Subgroup analysis published in 2000"

Assessing the potential for bias in meta-analysis due to selective reporting of subgroup analyses within studies

Abstract: Subgroup analysis is frequently used to investigate heterogeneity in meta-analysis. Subgroup data are not always available, and researchers should record what data were available for each trial. If data were not available, and it is known that the subgroup data were collected, the potential for selective reporting should be considered. Bias due to selective publishing of reports is widely recognized in meta-analysis. In contrast, selective reporting within studies is little discussed but potentially important. We explored this problem by evaluating the effect of potential bias in subgroup analysis due to within-study selective reporting with an existing meta-analysis. The review addressed malaria chemoprophylaxis in pregnancy. The conclusion in the original review, that benefit is limited to primigravidae, was based on subgroup analysis using the three trials out of five which reported on subgroups. We developed a method of sensitivity analysis that imputes data for the missing subgroups to assess the robustness of the results and the conclusions drawn. In this particular example, our analysis indicates that the estimate of effect reported in the review is most likely to overestimate the true effect and the conclusion that benefit is limited to primigravidae may be false.

Statistics applied to clinical trials

Abstract: Preface Foreword 1 Hypotheses, Data, Stratification 2 The analysis of efficacy data of drug trials 3 The analysis of safety data of drug trials 4 Equivalence testing 5 Statistical power and sample size 6 Interim analysis 7 Multiple statistical inferences 8 Principles of linear regression 9 Subgroup analysis multiple linear regression: confounding, interaction, synergism 10 Curvilinear regression 11 Meta-analysis 12 Crossover studies with continuous variables: power analysis 13 Crossover studies with binary responses 14 Post-hoc analysis in clinical trials, a case for logistic regression analysis 15 quality-of-life assessment in clinical trials 16 Statistics for the analysis of genetic data 17 Relationship among statistical distributions 18 Statistics is not "bloodless" algebra Appendix Index

Is dosage of physiotherapy a critical factor in deciding patterns of recovery from stroke: a pragmatic randomized controlled trial.

Abstract: Background and Purpose The best treatment and management of stroke patients has been shown to be in stroke units by multidisciplinary rehabilitation teams. Since the composition of stroke units differs it is important to know the extent to which the different components contribute to this result. Physiotherapy is one component of most rehabilitation teams and recent systematic reviews have shown that patients with stroke receiving more physiotherapy achieve more recovery from disability. However, information about the actual amounts of physiotherapy needed to achieve this result is not known. Method A pragmatic, randomized, single-blind, controlled trial comparing recovery from disability in subjects receiving the current standard amount of 30 minutes' physiotherapy with those receiving double that amount (60 minutes). The study included measures of physical performance and function, psychological aspects of anxiety and depression, and perceived control over recovery. Results Some 114 subjects were recruited to the study; full six-week data are available for 104 subjects and six-month data for 93 subjects. Comparison of initial to six-week difference scores in the control and intervention groups of the whole sample did not show a significant difference. Scrutiny of the recovery curves of the whole sample showed that, in half the sample, three distinct patterns of recovery were demonstrated. Conclusion These results suggest that doubling the physiotherapy time available for patients in a stroke unit will not provide a measurable benefit for all patients. The subgroup analysis of patterns of recovery must be regarded as speculative, but provides the basis for hypotheses about those likely to respond well to more intensive therapy. Copyright © 2000 Whurr Publishers Ltd.

Anti-tuberculous therapy for maintaining remission of Crohn's disease.

Abstract: Objectives To evaluate the effects of anti-tuberculous therapy for the maintenance of remission in patients with Crohn's disease. Search strategy We searched the Inflammatory Bowel Disease Trials Register, the Cochrane Controlled Trials Register and MEDLINE from 1966 to 1998 (supplemented by a manual search of Index Medicus from 1966 to 1994). We also searched for abstracts in Gut, Gastroenterology, and The American Journal of Gastroenterology from 1990 to 1996. Date of most recent search: August 1998. Selection criteria Randomized trials of anti-tuberculous therapy in patients with Crohn's disease. Data collection and analysis Data on the number of patients maintaining remission for each treatment group were abstracted. These data were pooled to yield Mantel-Haenszel odds ratios and numbers needed to treat for maintenance of remission in treated versus control groups. Main results A total of seven randomized trials which included 355 patients were identified. Two trials used anti-tuberculous therapy (clofazimine or clofazimine, rafmpin, ethambutol, and dapsone) in combination with corticosteroids to induce remission. Maintenance therapy consisted of the anti-tuberculous agents without corticosteroids. Control patients received corticosteroids to induce remission but no anti-tuberculous therapy. The analysis of all seven trials yielded an odds ratio for maintenance of remission of 1.36 (95% CI 0.87-2. 13). Removing the two studies that were published as abstracts did not significantly affect this result: the pooled odds ratio was 1.14 (95% CI 0.71-1.83). The two trials reported as abstracts were excluded from subgroup analyses because they did not include any information on adjunct therapy. Subgroup analysis of the two trials which used steroids to induce remission yielded an odds ratio for maintenance of remission of 3.37 (95% CI 1.38-8.24). The number needed to treat was three. However, these two trials included only 89 patients, and the results should be interpreted with caution. The remaining three trials compared the combination of anti-tuberculous therapy and 'standard therapy' with 'standard therapy alone'. The pooled odds ratio was 0.70 (95% CI 0.39-1.25). Reviewer's conclusions Anti-tuberculous therapy may be effective in maintaining remission in patients with Crohn's disease when remission has been induced with corticosteroids combined with anti-tuberculous therapy. However, the results which support this conclusion come from a subgroup of only two trials with small numbers of patients and should be interpreted with caution. Use of this therapy cannot be recommended on the basis of this evidence.

Changes in energy during treatment of depression: an analysis of fluoxetine in double-blind, placebo-controlled trials.

Abstract: More than two thirds of patients with depression present with symptoms of fatigue, low energy, and listlessness Because daytime sedation may be a concern in such patients, a "nonsedating" antidepressant should be considered The authors examined the effects of fluoxetine on depression-related disturbances in energy Data from seven double-blind, placebo-controlled clinical trials in 2,075 patients with major depression were retrospectively analyzed The Hamilton Rating Scale for Depression (HAM-D) Retardation factor score (total of items 1, 7, 8, and 14) was used as the primary measure of energy improvement, whereas the HAM-D-17 total score was used to assess changes in overall depression Elderly patients (aged 60 years and older) were included in the overall group and were also analyzed separately In addition, a subgroup analysis was performed using the HAM-D Retardation factor score to categorize patients as having low (score or = 8) levels of retardation at baseline Beginning at week 3, fluoxetine-treated patients experienced statistically significant reductions in their HAM-D Retardation factor score compared with placebo-treated patients The reductions for the elderly subgroup were less than those for the overall population, but they were still statistically significant beginning at week 4 Patients in both the low and high baseline retardation groups improved significantly HAM-D-17 total scores for fluoxetine-treated patients in all groups (total, elderly, high retardation, and low retardation) improved significantly compared with placebo-treated patients These findings demonstrate that fluoxetine-treated patients experience an improvement in energy symptoms as their overall depression improves

Impact of quality items on study outcome. Treatments in acute lateral ankle sprains.

Abstract: Objective: This study investigates the influence of different aspects of methodologic quality on the conclusions of a systematic review concerning treatments of acute lateral ankle sprain. Method: A data set of a systematic review of 44 trials was used, of which 22 trials could be included in this study. Quality assessment of the individual studies was performed using the Delphi list. We calculated effect sizes of the main outcome measure in each study in order to evaluate the relationship between overall quality scores and outcome. Next, we investigated the impact of design attributes on pooled effect sizes by subgroup analysis. Results: The quality of most studies (82%) was low; only 4 of 22 trials were of high quality. Studies with proper randomization and blinding procedure produce a slightly higher (not statistically significant) effect estimate compared to the other studies. Conclusion: Previous research has suggested that methodologically poorly designed studies tend to over-estimate the effect estimate. Our study does not confirm these conclusions.

Aspirin for primary prevention. Treatment policy should be based on all trial evidence, not subgroup analysis.

Abstract: Editor—We have suggested that aspirin for primary prevention is safe and worthwhile when the estimated 10 year coronary risk is >15%, provided that any hypertension is controlled.1 This conclusion comes from conservative interpretation of a meta-analysis examining the balance of benefit and risk in four large randomised controlled trials of aspirin for primary prevention, and fully supports recommendations in the Joint British Societies and British Hypertension Society guidelines.2 One assumption central to this analysis, and to these guidelines, is that relative risk reduction by aspirin is constant, so that the magnitude of benefit from aspirin is determined by pretreatment coronary risk. Unfortunately, Meade et al did not examine this assumption in their subgroup analysis of the thrombosis prevention trial.3 Rather, they present subgroup analyses according to individual risk factors (systolic blood pressure, age, and cholesterol concentration). These analyses are not really apposite to the guidelines and may even be misleading. For example, their results suggest little benefit (6% reduction in coronary heart disease) or even harm (8% increase in all cardiovascular events) from aspirin when systolic blood pressure exceeds 145 mm Hg. In the physicians' study in the United States there was a substantial (35%) reduction in coronary events at systolic blood pressure >150 mm Hg.4 In the hypertension optimal treatment study, men with hypertension that was controlled from 168/106 mm Hg to an average of 140/83 mm Hg, which is still “high normal,” had a coronary reduction of 42% (P=0.001) and a 13% reduction in all cardiovascular events.5 The important point is that subgroup analysis of the thrombosis prevention trial is certainly not representative of all the trial evidence available. Similar discrepancies are present in the findings for age. At age 65 and over this subgroup analysis suggests a 29% increase in coronary heart disease, but a 41% reduction in stroke, with aspirin. As the authors note, this is totally inconsistent with the physicians' study, which showed coronary reductions of 44% at ages 60-69 and 41% at ages 70-84.4 In the hypertension optimal treatment study, treated hypertensive patients aged 65 and over had reductions in coronary heart disease of 38% and all cardiovascular events by 21%.5 The treatment policy for aspirin for primary prevention should be based on all the trial evidence and estimation of absolute risk of coronary heart disease,2 not on subgroup analysis of a single trial or on a single coronary risk factor.

Using a balancing procedure in multicenter clinical trials. Simulation of patient allocation based on a trial of ventilation tubes for otitis media with effusion in infants.

Abstract: Objective: A basic issue in randomized controlled trials (RCTs) is whether we can safely assume comparability between groups at baseline with respect to all potentially important prognostic factors. In other words, did randomization work sufficiently well? In small trials balanced allocation procedures are employed, whereas in large-scale trials simple randomization will do. The question is: When should balancing be considered? Methods: We performed a simulation study in which we varied the number of categories in the prognostic factors and the number of patients. Results: Simulation showed that, in all instances, a balancing procedure almost always led to perfect or almost perfect balance, while the imbalance with simple randomization was larger. To study the effect of balanced and random allocation on subgroup analyses in our OME trial, we compared the quotient of the width of the confidence intervals (CI). The widest CI in random allocation over the 13 hospitals was on average 13% wider than in balanced allocation. Conclusion: Investigators should always consider balanced allocation, especially in categories with a low number of patients and when subgroup analysis over many categories is requested.

Reducing the impact of the diabetic heart's increased vulnerability to cardiovascular disease

Abstract: NIDDM and its metabolic control predict coronary heart disease in elderly subjects J. Kuusisto and others Angiographic findings and outcome in diabetic patients treated with thrombolytic therapy for acute myocardial infarction: the GUSTO-I experience S.L. Woodfield and others Cholesterol lowering with simvastatin improves prognosis of diabetic patients with coronary heart disease. A subgroup analysis of the Scandinavian Simvastatin Survival Study (4S) K. Pyorala and others

Monitoring and management of bone status in patients on chronic glucocorticoid treatment--the Medscheme experience.

Abstract: Objective: Review of administrative databases to gain insight into the investigation, management and sequelae of bone disease in patients on long-term glucocorticoid treatment. Design: Retrospective analysis of 1998 pharmaceutical and clinical claims data for ± 2 million lives administered by Medscheme. Data were extracted for members registered with the chronic medication programme as eligible for chronic glucocorticoid treatment. Those identified were subjected to further review for evidence of osteoporosis and/or hip fracture. Subgroup analysis of peri- and postmenopausal women was carried out and compared against a control group. Main outcame measures: Osteoporosis investigation and treatment rates in males and females; frequency of hip fractures; prescribing profiles; role of underlying disease, glucocorticoid route, gender and age in development of osteoporosis. Results: A total of 1614 subjects (54% females) was registered for chronic glucocorticoid treatment. Osteoporosis was diagnosed in 14.1% of females and 5.9% of males across abroad age range. Hip fractures were recorded for one female and three males. The subgroup analysis showed that osteoporosis was ± 1.5 times more common in women receiving glucocorticoids than in peri- and postmenopausal controls, and that there was greater use of vitamin D and calcium supplementation and bisphosphonates in those exposed to glucocorticoids. Multivariate analysis showed overall that female gender, increasing age and oral glucocorticoids were significantly related to osteoporosis. Conclusion: Reference to UK and US data suggests that while local practitioners are aware of the effect of glucocorticoids on bone, the level of awareness is probably suboptimal, especially with regard to male patients. S Afr Med J 20OO 90: 1125-1129

Clinical trials in developing countries: Discussions at the '9th International Symposium on Long Term Clinical Trials', London, UK, 19-20 June 2000.

Abstract: This symposium provided a useful forum for the discussion of issues relating to the design and conduct of clinical trials. There is a need for greater awareness of the complexity of modern day trials, in which a host of statistical, logistical, regulatory and ethical issues are involved. Issues discussed ranged from the effect of sample size on the outcome, and subgroup analysis, to defining and maintaining discrete endpoints. Some useful debate centred on the use of meta-analysis and the current limitations of combining information from different data sets. This brought up the subjects of trial registries and raw data repositories for all clinical trials. Progress and relevance of the Cochrane collaboration were reviewed. The economics of clinical trials was another important topic. Regulatory issues such as the role of data and safety monitoring boards (DSMB) and the guidelines in place for effective data monitoring and progress analysis were discussed. Representatives of government organisations and industry gave both European and American perspectives. This report however focuses specifically on the section devoted to the subject of clinical trials in developing countries.

Debate: A subversive view of subsets - a dissident clinician's opinion.

Abstract: Clinical trialists and statisticians are very wary of subgroup analysis, for good reasons. Clinicians have to deal with situations in which subgroups of patients differ widely from one another in their prognosis and response to treatment. Few trials are large enough to demonstrate convincingly these differences in outcome, but often provide suggestive evidence. Should we ignore this and treat all patients as the same, or should we allow dubious statistical evidence to buttress biological plausibility in making clinical decisions?

Commentary Debate: A subversive view of subsets — a dissident clinician's opinion

Abstract: Clinical trialists and statisticians are very wary of subgroup analysis, for good reasons. Clinicians have to deal with situations in which subgroups of patients differ widely from one another in their prognosis and response to treatment. Few trials are large enough to demonstrate convincingly these differences in outcome, but often provide suggestive evidence. Should we ignore this and treat all patients as the same, or should we allow dubious statistical evidence to buttress biological plausibility in making clinical decisions?

Optimisation versus certainty : developing the use of economic evaluation for decision making.

Abstract: This thesis assesses the methods used in economic evaluation, the relationship of economic evaluation to decision-making and investigates the possible limitations of economic evaluation as it is currently used to support policies aimed at maximising population health gain. It then evaluates alternative methods of analysing data from economic evaluations to better inform policy decisions. The hypothesis of this thesis is that a greater use of subgroup analysis in policy decisions could potentially improve the efficiency of allocating scarce health care resources. This study aims to investigate the impact on population health gain and service cost- effectiveness of using subgroup analysis within defined parameters to derive and evaluate estimates of effect, and compare it to the more traditional methods of statistical inference. Data from existing large trials are used to calculate cost-effectiveness ratios for the total study population and for subgroups. Total and subgroup estimates of cost-effectiveness are applied to patient populations through simulation, and outcomes predicted on the assumption that treatment decisions are guided by estimates derived from the trial. The distribution of cost-effectiveness ratios based on different rules for `allowing' the use of subgroup analysis results is compared with the distribution of cost-effectiveness ratios based on aggregate analyses. Results show that pre-selected subgroups can provide a stronger likelihood of maximising overall health gain. This thesis argues for optimisation in the use and interpretation of results rather than an over reliance on certainty and the resulting restriction on the use of available data. It concludes that under the scrutiny of a health care system for which the primary goal is health gain maximisation within resource constraints, policy decisions made using the results of subgroup analysis could result in a more efficient allocation of resources.