Assessing the payback from health R & D: From ad hoc studies to regular monitoring

Abstract

Chapter 1 : Introduction • The increasing demands for the benefits of payback from publicly funded R&D to be assessed are based partly on the need to justify or account for expenditure on R&D, and partly on the desire for information to assist resource allocation and the better management of R&D funds. The former consideration is particularly strong in relation to the R&D expenditure that comes out of the wider NHS budget. • In this report a range of categories of payback will be identified along with a variety of methods for assessing them. • The aim of the report is to make recommendations as to how the outcomes from health research might best be monitored on a regular basis. The specific context of the report is the NHS R&D Programme but many of the issues will be relevant for a wide range of funders of health R&D. • The introduction sets out not only a plan of the report but also suggests that readers familiar with the general arguments and existing literature may choose to jump to Chapter 6.

Chapter 2 : Review of Existing Approaches to Assessing the Payback from Research • Existing work describes various approaches to valuing research. Some are ex ante and attempt to predict the outcomes of research being considered, others are ex post or retrospective. • The five categories of benefit or payback from health R&D that have been identified involve contributions: to knowledge; to research capacity and future research; to improved information for decision making; to the efficiency, efficacy and equity of health care services; and to the nation’s economic performance. These are shown in Table 1 of the report • The process by which R&D generates final outcomes can be modelled as a sequence. This includes primary outputs such as publications; secondary outputs in the form of policy or administrative decisions; and final outcomes which comprise the health and economic benefits. Feedback loops are also introduced and mitigate the limitations of a linear approach. • Qualitative and quantitative approaches can be used but there are immense problems with time lags and attributing outcomes, and sometimes even outputs, to specific items of research funding. • Four common methods of measuring payback can be used. Expert review, by peers or, sometimes, users is the traditional way of assessing the quality of research. Bibliometric techniques can involve not only counting publications but also using datasets such as the Science Citation Index and Wellcome’s Research Outputs Database (ROD). The various methods of economic analysis of payback are difficult to undertake given the costs and problems of acquiring relevant information and estimating benefits. Social science methods include case studies, which can provide useful information but are resource intensive, and questionnaires to researchers and potential research users.

Chapter 3 : Characteristics of a Routine Monitoring System • In moving from ad hoc or research studies of payback towards a more regular monitoring it is noted that whereas there has always been a tradition of evaluation of research, in the public services in general there is now a greater emphasis on audit and performance measurement and indicators. A review of these various systems suggests we should be looking to develop a system of outcomes monitoring that incorporates performance indicators (PIs) and measurement rather than an audit system that is trying to monitor activities against predetermined targets. • Standard characteristics of performance measurement systems do not necessarily apply to research where, for example, there are non-standard outputs. Difficulties have arisen in the USA in attempting to apply the Government Performance and Results Act to research funding agencies. It is shown that because the findings of basic research, in particular, enter a knowledge pool in which people and ideas interact, it is difficult to use a PIs’ approach to track eventual outcomes. However, for some types of health research it has proved more feasible to trace the flow between research outputs and outcomes. • An outcomes monitoring system could be useful if it met the following criteria: relevant to, with as comprehensive coverage as possible of, the funders objectives; relevant to the funder’s decision making processes; encourages accurate compliance; minimises unintended consequences; and has acceptable costs.

Chapter 4 : Differences Between Research Types • The range of differences between types of research can be relevant for the design of a routine monitoring system. The OECD distinguishes between basic research, applied research and experimental development. Most DH/NHS research is applied. There might be more of a tradition of publication of findings in applied research in health than in other fields. Nevertheless, the publication and incentives patterns operating in basic research mean that it would be inappropriate to use bibliometric indicators in a simple way across all fields even in health research. • Despite having some differences from health research in publication patterns and in the detailed categories of payback, the broad approach proposed in Chapter 6 could be applied to social care research. • Research that is commissioned, especially by the government, has some of the minimum conditions built into it that are associated with outcomes being generated, in particular because the funder has identified that a contribution in this area will be valuable.

Chapter 5 : What Units of Research? • The term programme has various meanings including being used to describe a collection of projects on a common theme and to describe a block of funding for a research unit. • Three main streams or modes of funding can be identified: projects, which are administratively grouped into programmes including a responsive programme; institutions/centres/units; individual researchers. These 3 streams are displayed in Figure 1. It is probable that the regular data-gathering for a monitoring system would operate at the basic level of each stream or mode. • Previous work demonstrates that the full range of benefits can sometimes be applied at the level of projects, either in the responsive mode or in programmes, through the use of questionnaires to researchers. Expert and user review and user surveys have also been applied. • Institutions and centres increasingly have experience not only of traditional periodic expert review but also of producing annual reports, although there are debates about what dimensions to include in such reviews and reports. • Individuals in receipt of research development awards have completed questionnaires during and after the awards. These concentrate on the development of research capacity but can go wider.

Chapter 6 : A Possible Comprehensive Outcomes Monitoring System • The proposed system is intended for DH/NHS to monitor the outcomes from its R&D in order to justify the R&D expenditure and assist with managing the portfolio. More detailed information is required for the latter purpose. • We propose a multidimensional approach be adopted to cover all the dimensions of payback and that information be gathered from three sets of sources and Table 3 shows which methods would cover which output/outcome categories. • Firstly, possibly annually, a questionnaire (possibly electronic) covering most payback categories should gather data from the basic level of each funding stream ie. from lead researchers of projects, from research institutions/centres, and from individual award holders. • Secondly, supplementary information should be gathered from external databases (including the citation indices and Wellcome’s ROD). • Thirdly, a range of approaches ie. user surveys, reviews by experts and peers, case studies including economic evaluations, and analysis of sources used in policy documents such as NICE guidelines, would be undertaken on a sample basis. They would provide not only supplementary information but, as with the external databases, would also verify the data collected directly from researchers. • These proposals can be evaluated against the criteria set out in Chapter 3: • The system is relevant to DH’s objectives of generating payback in a range of categories. • Various problems have to be overcome before the system could be fully decision relevant. Firstly it might be necessary to ask researchers to apportion the contribution made to specific outputs from various funding streams. Second, to be decision relevant the information would have to be analysed and presented in a manner consistent with funders’ decision making processes. This would involve a) showing how for each outcome and output, for example publications, data from one project or stream could be compared with those from another and b) demonstrating how different outputs and outcomes could be aggregated. • The questions of accuracy of data, minimisation of unintended consequences and the acceptability of the net costs are also addressed.

Chapter 7 : Research and Monitoring • Whilst this report is primarily concerned with moving from ad hoc studies towards a routine monitoring system there are issues that need further research. • Before embarking on full implementation the feasibility needs to be tested of items such as on-line recording of data and asking researchers to attribute proportions of research outputs to separate funding agencies. • Once the system is implemented the value of some items can be better assessed, for example the additional value provided by self reporting of publications beyond that gained from relying on external databases. • The data provided by the system would provide opportunities for further payback research on, for example, links between publications and other categories of payback. • Some items such as network analysis could potentially be added to the monitoring system after further examination of them. • Finally the benefit from the monitoring system itself should be assessed.