A systematic review of costs and benefits of population screening for dementia

Abstract

BackgroundDespite over a decade of efforts to improve timely detection, as many as 50% of people with dementia do not receive a formal diagnosis or receive it too late to be clinically useful. Population screening (routine testing of a defined population even when no symptoms or signs have been reported) might provide part of the solution. However, the costs and benefits of this large-scale intervention are unclear. We undertook a systematic review of population screening to clarify the issue. We report on the evidence pertaining to clinical, psychosocial, and economic outcomes. MethodsEligible studies included those assessing population screening either as a standalone or component of a dementia intervention compared with a routine pattern of care in the general population or non-selected general practice or community care attendees. Non-comparative studies and models were also eligible. Relevant manuscripts published by May, 2012, for clinical and psychosocial outcomes (cognition, behaviours, physical and mental health, psychological and emotional wellbeing, stress, stigma, suffering, independence) and by April, 2013, for economic outcomes (cost, cost-effectiveness, cost benefit from any perspectives) were identified by unlimited searches tailored to outcome categories and databases (including Medline, Embase, CINAHL, DARE). Guided by the AMSTAR instrument, we used state-of-the-art methodology, including duplicate title scanning and extraction; no study was excluded on the basis of quality. FindingsAfter screening 12 054 titles and examining the full text of 143 manuscripts, we found no studies that assessed the effect of screening on patient outcomes, such as cognitive, physical, mental or emotional health, social function, or planning. No studies were identified that assessed the potential harms caused by the screening results, such as risk of depression, anxiety, stigma, or loss of independence. For economic outcomes, we screened 5210 titles and examined the full text of 30 manuscripts. We identified three primary studies (one US study, mean age 74·6 years [SD 6·9] for positive screen and 71·1 years [5·5] for negative screen; two related Spanish studies, mean age 72·5 years [11·3]) reporting on the direct cost per patient diagnosed. Four modelling studies compared population screening scenarios with other models of dementia care—one model building on primary data (USA) and two on hypothetical cohorts (UK). Another model (USA) examined the costs of a screening programme for elderly drivers. Combined, these studies suggest that the substantial resources needed to screen for dementia are determined by the age of the screened population, the properties of the screening instrument (in terms of false positive and false negative), the extent to which GPs are involved in follow-up assessments, and the rates of participation at various stages of the diagnostic process. InterpretationThe published work is sparse on the potential effect of population screening on individuals and the health-care system. We found no evidence that, if rolled out, population screening would lead to improved clinical or psychosocial outcomes, no evidence furthering our understanding of the risks it entails, and no suggestion of its added value compared with present practice. Learning from present research and scrutiny of available data is needed to assess the benefits and harms of the process of screening in representative patient groups and to identify those most likely to benefit in a population setting. Rigorous, up-to-date, and context-specific models are also needed to assess a range of possible scenarios in which population screening might have added value. Meanwhile, policy makers should be cautious about the introduction of resource-intensive screening programmes for dementia without any evidence of benefits, risks, or cost-effectiveness. Careful assessment of the benefits, risks, and cost-effectiveness of population screening for dementia is needed before introduction of this costly large-scale intervention. FundingThis article presents independent research funded by the Alzheimer's Society (project grant 129) and supported by the NIHR CLAHRC for Cambridgeshire and Peterborough. The views expressed are those of the authors and not necessarily those of the Alzheimer's Society, NHS, NIHR, or the Department of Health.