Point estimates and confidence regions for sequential trials involving selection

A number of authors have proposed clinical trial designs involving the comparison of several experimental treatments with a control treatment in two or more stages. At the end of the first stage, the most promising experimental treatment is selected, and all other experimental treatments are dropped from the trial. Provided it is good enough, the selected experimental treatment is then compared with the control treatment in one or more subsequent stages. The analysis of data from such a trial is problematic because of the treatment selection and the possibility of stopping at interim analyses. These aspects lead to bias in the maximum-likelihood estimate of the advantage of the selected experimental treatment over the control and to inaccurate coverage for the associated confidence interval. In this paper, we evaluate the bias of the maximum-likelihood estimate and propose a bias-adjusted estimate. We also propose an approach to the construction of a confidence region for the vector of advantages of the experimental treatments over the control based on an ordering of the sample space. These regions are shown to have accurate coverage, although they are also shown to be necessarily unbounded. Confidence intervals for the advantage of the selected treatment are obtained from the confidence regions and are shown to have more accurate coverage than the standard confidence interval based upon the maximum-likelihood estimate and its asymptotic standard error.

Hitting emissions targets with (statistical) confidence in multi-instrument Emissions Trading Schemes

A means of assessing, monitoring and controlling aggregate emissions from multi-instrument Emissions Trading Schemes is proposed. The approach allows contributions from different instruments with different forms of emissions targets to be integrated. Where Emissions Trading Schemes are helping meet specific national targets, the approach allows the entry requirements of new participants to be calculated and set at a level that will achieve these targets. The approach is multi-levelled, and may be extended downwards to support pooling of participants within instruments, or upwards to embed Emissions Trading Schemes within a wider suite of policies and measures with hard and soft targets. Aggregate emissions from each instrument are treated stochastically. Emissions from the scheme as a whole are then the joint probability distribution formed by integrating the emissions from its instruments. Because a Bayesian approach is adopted, qualitative and semi-qualitative data from expert opinion can be used where quantitative data is not currently available, or is incomplete. This approach helps government retain sufficient control over emissions trading scheme targets to allow them to meet their emissions reduction obligations, while minimising the need for retrospectively adjusting existing participants’ conditions of entry. This maintains participant confidence, while providing the necessary policy levers for good governance.

Acute stroke predisposes to oral gram-negative bacilli - a cause of aspiration pneumonia?

Objectives: To assess any change in the oral flora in the mouths of stroke patients during the acute and rehabilitation phases and to determine whether this is related to episodes of aspiration pneumonia and clinical outcome. Materials and Methods: This observational study was carried out in hospital wards in a University teaching hospital. The subjects were patients immediately post-stroke and during the rehabilitation period, acute admissions and a group of healthy volunteers. An assessment of dentition and swallow in the presence or absence of oral aerobic gram-negative bacilli (AGNB) was correlated. Results: Of the acute stroke patients 52% had an unsafe swallow. AGNB carriage was documented in 34% of the acute stroke group. Of the 11 patients who died 55% had AGNB, 73% had an unsafe swallow and 36% had a combination of both. Conclusion: AGNB is a common finding in acute stroke patients. It is not a consequence of age or acute hospitalisation and is associated with an unsafe swallow and a higher mortality. Copyright (C) 2003 S. Karger AG, Basel.

Cost-benefit evaluation of routine influenza immunisation in people 65-74 years of age

OBJECTIVES: To determine the cost-effectiveness of influenza vaccination in people aged 65-74 years in the absence of co-morbidity. DESIGN: Primary research: randomised controlled trial. SETTING: Primary care. PARTICIPANTS: People without risk factors for influenza or contraindications to vaccination were identified from 20 general practitioner (GP) practices in Liverpool in September 1999 and invited to participate in the study. There were 5875/9727 (60.4%) people aged 65-74 years identified as potentially eligible and, of these, 729 (12%) were randomised. INTERVENTION: Participants were randomised to receive either influenza vaccine or placebo (ratio 3:1), with all individuals receiving pneumococcal vaccine unless administered in the previous 10 years. Of the 729 people randomised, 552 received vaccine and 177 received placebo; 726 individuals were administered pneumococcal vaccine. MAIN OUTCOME MEASURES AND METHODOLOGY OF ECONOMIC EVALUATION: GP attendance with influenza-like illness (ILI) or pneumonia (primary outcome measure); or any respiratory symptoms; hospitalisation with a respiratory illness; death; participant self-reported ILI; quality of life (QoL) measures at 2, 4 and 6 months post-study vaccination; adverse reactions 3 days after vaccination. A cost-effectiveness analysis was undertaken to identify the incremental cost associated with the avoidance of episodes of influenza in the vaccination population and an impact model was used to extrapolate the cost-effectiveness results obtained from the trial to assess their generalisability throughout the NHS. RESULTS: In England and Wales, weekly consultations for influenza and ILI remained at baseline levels (less than 50 per 100,000 population) until week 50/1999 and then increased rapidly, peaking during week 2/2000 with a rate of 231/100,000. This rate fell within the range of 'higher than expected seasonal activity' of 200-400/100,000. Rates then quickly declined, returning to baseline levels by week 5/2000. The predominant circulating strain during this period was influenza A (H3N2). Five (0.9%) people in the vaccine group were diagnosed by their GP with an ILI compared to two (1.1%) in the placebo group [relative risk (RR), 0.8; 95% confidence interval (CI) = 0.16 to 4.1]. No participants were diagnosed with pneumonia by their GP and there were no hospitalisations for respiratory illness in either group. Significantly fewer vaccinated individuals self-reported a single ILI (4.6% vs 8.9%, RR, 0.51; 95% CI for RR, 0.28 to 0.96). There was no significant difference in any of the QoL measurements over time between the two groups. Reported systemic side-effects showed no significant differences between groups. Local side-effects occurred with a significantly increased incidence in the vaccine group (11.3% vs 5.1%, p = 0.02). Each GP consultation avoided by vaccination was estimated from trial data to generate a net NHS cost of 174 pounds. CONCLUSIONS: No difference was seen between groups for the primary outcome measure, although the trial was underpowered to demonstrate a true difference. Vaccination had no significant effect on any of the QoL measures used, although vaccinated individuals were less likely to self-report ILI. The analysis did not suggest that influenza vaccination in healthy people aged 65-74 years would lead to lower NHS costs. Future research should look at ways to maximise vaccine uptake in people at greatest risk from influenza and also the level of vaccine protection afforded to people from different age and socio-economic populations.

Depression, confidence, and decision: evidence against depressive realism

This research examined how retrospective self-assessments of performance are affected by major depression. To test the validity of the depressive realism versus the selective processing hypotheses, aggregate posttest performance estimates (PTPEs) were obtained from clinically depressed patients and an age-matched comparison group across 4 decision tasks (object recognition, general knowledge, social judgment, and line-length judgment). As expected on the basis of previous findings, both groups were underconfident in their PTPEs, consistently underestimating the percentage of questions they had answered correctly. Contrary to depressive realism, and in partial support of the selective processing account, this underconfidence effect was not reduced but modestly exacerbated in the depressed patients. Further, whereas the PTPEs of the comparison group exceeded that expected on the basis of chance alone those of the depressed individuals did not. The results provide no support for the depressive realism account and suggest that negative biases contribute to metacognitive information processing in major depression.

A method for assessing confidence in requirements analysis

Context: During development managers, analysts and designers often need to know whether enough requirements analysis work has been done and whether or not it is safe to proceed to the design stage. Objective: This paper describes a new, simple and practical method for assessing our confidence in a set of requirements. Method: We identified 4 confidence factors and used a goal oriented framework with a simple ordinal scale to develop a method for assessing confidence. We illustrate the method and show how it has been applied to a real systems development project. Results: We show how assessing confidence in the requirements could have revealed problems in this project earlier and so saved both time and money. Conclusion: Our meta-level assessment of requirements provides a practical and pragmatic method that can prove useful to managers, analysts and designers who need to know when sufficient requirements analysis has been performed.

Dissociation between the impact of evidence on eye movement target choice and confidence judgements

It has been suggested that the evidence used to support a decision to move our eyes and the confidence we have in that decision are derived from a common source. Alternatively, confidence may be based on further post-decisional processes. In three experiments we examined this. In Experiment 1, participants chose between two targets on the basis of varying levels of evidence (i.e., the direction of motion coherence in a Random-Dot-Kinematogram). They indicated this choice by making a saccade to one of two targets and then indicated their confidence. Saccade trajectory deviation was taken as a measure of the inhibition of the non-selected target. We found that as evidence increased so did confidence and deviations of saccade trajectory away from the non-selected target. However, a correlational analysis suggested they were not related. In Experiment 2 an option to opt-out of the choice was offered on some trials if choice proved too difficult. In this way we isolated trials on which confidence in target selection was high (i.e., when the option to opt-out was available but not taken). Again saccade trajectory deviations were found not to differ in relation to confidence. In Experiment 3 we directly manipulated confidence, such that participants had high or low task confidence. They showed no differences in saccade trajectory deviations. These results support post-decisional accounts of confidence: evidence supporting the decision to move the eyes is reflected in saccade control, but the confidence that we have in that choice is subject to further post-decisional processes.