Neil Craig

Evaluating the health effects of social interventions

Is no evidence better than any evidence when controlled studies are unethical? Rigorous evidence on the health effects of social interventions is scarce1 2 despite calls for more evidence from randomised studies.3 One reason for the lack of such experimental research on social interventions may be the perception among researchers, policymakers, and others that randomised designs belong to the biomedical world and that their application to social interventions is both unethical and simplistic.4 Applying experimental designs to social interventions may be problematic but is not always impossible and is a desirable alternative to uncontrolled experimentation.3 However, even when randomised designs have been used to evaluate social interventions, opportunities to incorporate health measures have often been missed.5 For example, income supplementation is thought to be a key part of reducing health inequalities,6 but rigorous evidence to support this is lacking because most randomised controlled trials of income supplementation have not included health measures.5 Current moves to increase uptake of benefits offer new opportunities to establish the effects of income supplements on health. In attempting to design such a study, however, we found that randomised or other controlled trials were difficult to justify ethically, and our eventual design was rejected by funders. #### Box 1 Attendance allowance A pilot study carried out by one of us (RH) showed substantial health gains among elderly people after receipt of attendance allowance. We therefore decided to pursue a full scale study of the health effects of income supplementation. …

Can we evaluate population screening strategies in UK general practice? A pilot randomised controlled trial comparing postal and opportunistic screening for genital chlamydial infection

Study objective: To assess whether opportunistic and postal screening strategies for Chlamydia trachomatis can be compared with usual care in a randomised trial in general practice. Design: Feasibility study for a randomised controlled trial. Setting: Three West of Scotland general medical practices: one rural, one urban/deprived, and one urban/affluent. Participants: 600 women aged 16–30 years, 200 from each of three participating practices selected at random from a sample of West of Scotland practices that had expressed interest in the study. The women could opt out of the study. Those who did not were randomly assigned to one of three groups: postal screening, opportunistic screening, or usual care. Results: 38% (85 of 221) of the approached practices expressed interest in the study. Data were collected successfully from the three participating practices. There were considerable workload implications for staff. Altogether 124 of the 600 women opted out of the study. During the four month study period, 55% (81 of 146) of the control group attended their practice but none was offered screening. Some 59% (80 of 136) women in the opportunistic group attended their practice of whom 55% (44 of 80) were offered screening. Of those, 64% (28 of 44) accepted, representing 21% of the opportunistic group. Forty eight per cent (59 of 124) of the postal group returned samples. Conclusion: A randomised controlled trial comparing postal and opportunistic screening for chlamydial infection in general practice is feasible, although resource intensive. There may be problems with generalising from screening trials in which patients may opt out from the offer of screening.

Less fog on the Tyne? Programme budgeting in Newcastle and North Tyneside

Programme Budgeting (PB) has been widely promoted as a model for the better conduct of the work of Health Authorities in the National Health Service in the United Kingdom. This paper reports on a project which looked at the development of PB in Newcastle and North Tyneside Health Authority (NNTHA), concentrating on the construction of a computerised tool for the compilation and analysis of programme budgets. The main activities carried out were a survey of user requirements for PB, a survey of data availability, the collection of data to construct programme budgets, and development of a relational database for storing and manipulating PB information. The main source of data was the Contract Minimum Data Set, which was supplemented by data from a number of other sources to give comprehensive information on spending in NNTHA. Costed activity data were produced, which could be aggregated in a large number of ways, such as by care setting (inpatient, outpatient, community, general practice, etc.), disease group (ICD9 chapter headings), case mix (Healthcare Resource Groups) and socio-demographic variables (age/sex, locality of GPs practice).

Does health care improve health

Do we place insufficient emphasis on the role of health care interventions in improving population health? While it is widely accepted that the population’s health is a product of the complex interaction of multiple factors, there is still debate about the relative contribution that different determinants have made and that different interventions and policies make to improving the health of the population. This debate affects the priority given to different policies. Much of the public health contribution to this debate reflects two widely held beliefs, attributable in part to the work of Thomas McKeown. The first is that health services have contributed little to the improvement in population health seen over the last 150 years and that the future role for health care in improving population health is limited. The second is that health services have largely responded to illness rather than attempted to maintain and promote people’s health. A number of studies suggest that a reappraisal of this orthodoxy is warranted. In addition, while the impact of health care on population health has still to be accurately quantified, this is also true of the other main determinants of health and of public health interventions to improve health. In the light of these uncertainties, two questions arise. First, why is there such a strong belief among public health practitioners in the capacity to improve health by tackling non-health care determinants through policies such as housing and community development? Second, is this belief a useful guide to policy and, if not, what if anything should be done about it? The evidence that health is determined by social, environmental and economic influences throughout a person’s life is not at issue. What is lacking is secure evidence that many broad public health interventions are effective. Priority must be given to addressing this lack of evidence. In the meantime, instead of polarized positions, an appropriate balance needs to be struck between the contrasting strategies of developing health services and intervening outside the health system.

Computer simulation to estimate the effectiveness of carotid endarterectomy.

Objectives: To investigate the benefit of carotid endarterectomy relative to medical treatment, by comparing the outcome for different groups of patients following transient ischaemic attacks. Methods: A Markov model was used to describe the survival and quality of life of patients treated for a transient ischaemic attack. The benefit is measured in terms of quality adjusted life months (QALMs). The outcome was estimated using a computer simulation with parameters based on published studies on the probability of events following treatment. The benefit of carotid endarterectomy was explored using a baseline set of parameters and a sensitivity analysis. Results: The baseline scenario of a 65-year-old male patient with the model factors set at an intermediate level showed a benefit for surgery of 3 QALMs compared with medical treatment alone. The sensitivity analysis showed that the most favourable combination of factors had a benefit of 13.4 QALMs and the least favourable a loss of 2 QALMs. Of all 128 factor combinations, 79.9% showed a benefit for surgery, 5.5% showed equal benefit, and 15.6% showed a benefit for medical treatment. Conclusions: Computer simulations have the potential for deriving estimates of benefit for different patient groups from the results of clinical trials. Combined with reliable information on costs, the technique could also demonstrate variations in cost-effectiveness for these groups. For patients following a transient ischaemic attack, the results from this simulation and limited cost information suggest that carotid endarterectomy is unlikely to be a cost-effective intervention in the UK for many patient groups despite a reduction in the risk of stroke.

Marginal analysis in practice

EDITOR,—We congratulate David Cohen on two counts: firstly, on achieving formal acceptance of a framework for assessing purchasing priorities and seeing this through to changes in contracts; and, secondly, on achieving a disinvestment list of 10 areas.1 Many purchasing authorities seem to have no framework (some even seem unaware that they need one), and clinicians can be reluctant to consider disinvestments.2 Cohen emphasises that the composition of the working group in the first stage needs careful consideration because the process is value laden. Despite this he gives no explanation …