- Ann Bowling, reader in health services researcha
- a Centre for Health Informatics and Multiprofessional Education (CHIME), University College London Medical School, Whittington Hospital, London N19 5NF
- Accepted 9 January 1996
Objective: To elicit the views of a large nationally representative sample of adults on priorities for health services.
Design: An interview survey based on a random sample of people aged 16 and over in Great Britain taken by the Office of Population Censuses and Surveys.
Subjects: The response rate to the survey was 75%, and the total number of adults interviewed was 2005.
Main outcome measures: A priority ranking exercise of health services supplemented with attitude questions about priorities, who should set priorities, and budget allocation.
Results: The results of the main priority ranking exercise of 12 health services showed that the highest priority (rank 1) was accorded to “treatments for children with life threatening illness,” the next highest priority (rank 2) was accorded to “special care and pain relief for people who are dying.” The lowest priorities (11 and 12) were given to “treatment for infertility” and “treatment for people aged 75 and over with life threatening illness.” Most respondents thought that surveys like this one should be used in the planning of health services.
Conclusions: The public prioritise treatments specifically for younger rather than older people. There is some public support for people with self inflicted conditions (for example, through tobacco smoking) receiving lower priority for care, which raises ethical issues.
This study is the first exercise in health service priorities based on a random sample of the British population
The highest priority of the public was the treatment for children with life threatening illness followed by special care and pain relief for people who are dying
The lowest priorities were for treatment for infertility and treatment for people aged 75 and over with life threatening illness
Most respondents thought that surveys of the public's opinions, like this one, should be used in planning health services
Prioritisation or rationing of health services is on government agendas across the world; different countries have adopted different approaches, ranging from policies of rationing by exclusion of specified treatments to rationing by guidelines.1 2 3 4 The British approach to setting priorities or rationing in health care has been described as “rationing by muddling through”5 and takes place at different levels.6 The 1995 House of Commons health committee report on priority setting in the NHS suggested that a preferable alternative to rationing by a “policy of exclusion” is to base policy on the health needs of local populations and to focus purchasing of health care on effective treatments.7 The government's response restated its commitment to encouraging a knowledge based NHS and maintained that rationing is not an issue while scope remains for improved effectiveness,8 although a national framework for rationing NHS treatments will probably eventually emerge.
The committee's report explicitly stated that the NHS must “remain responsive to shifting public concerns and debate.”7 The criteria that should influence rationing have become a commonplace subject in the British press, as well as in other forms of media, particularly since the case of “child B,” who had leukaemia and was denied treatment (as opposed to palliative care) by Cambridge and Huntingdon Health Commission on the grounds of its high chance of failure.9 Nicholson recently called for a public debate on health priorities and for the establishment of a Royal Commission on Priorities, with the public represented along with politicians and the medical profession.10 A recent study of the five year purchasing plans of 66 district health authorities in England shows a considerable increase in the number of purchasers adopting explicit policies on rationing health care. For example, 11 of the 66 plans specified treatments that will not be purchased, in contrast with four in the team's previous survey, and many others conceded that similar rationing may be unavoidable.11 These developments highlight the need to measure public opinion on rationing.
Several local surveys of the priorities of the public and doctors have been conducted in the United Kingdom.12 13 14 15 Apart from one survey based on a quota sample drawn by a market research company,16 however, there have been no published nationally representative surveys of health care priorities anywhere in the world. The studies in the United Kingdom have consistently shown that acute interventions that are perceived to be life saving are prioritised very highly by the public compared with many preventive initiatives (such as family planning and health education and promotion) and care for people with chronic illnesses and disabilities (such as people with mental illnesses and older people). In the United States the public consultation exercises of the Oregon health commission also found that the highest ranking priorities were for treatments for life threatening conditions (particularly acute conditions), maternity care, preventive care (but only for children), and palliative care.1
Obtaining a representative response from the public can be difficult.12 Rapid appraisal techniques that may be useful at neighbourhood level17 are relatively resource intensive across a whole population and are not a substitute for the need for representative information which deals with specific questions. This study aimed to obtain the views on priorities for health services of a random sample of the British population and used the Office of Population Censuses and Surveys omnibus survey as the vehicle.
Subjects and methods
The study design was an interview survey that was based on a random sample of people aged 16 and over in Great Britain taken by the Office of Population Censuses and Surveys for their May-June omnibus survey. The sampling frame for the survey was the postcode address file of “small users,” which includes all private household addresses. It was stratified by region, housing tenure, and socioeconomic group. The postal sectors were selected with probability proportionate to size, and, within each sector, 30 addresses were selected randomly. If an address contained more than one household, the interviewer used a standard procedure to select just one household randomly. Within households with more than one adult member just one person aged 16 or over was selected with the use of random number tables. Because only one household member was interviewed, people in households that contained few adults had a better chance of selection than those in households with many. A weighting factor was applied to correct for this unequal probability, and the individual adult was the unit of analysis. The number of selected addresses was 3000; this contained 328 ineligible addresses (for example, non-domestic). At the remaining 2672 addresses, 376 people refused to take part (14%), 23 were incapable of interview (1%), 268 were non-contactable (10%), and 2005 were interviewed, giving a response rate of 75%. Interviewing was completed within two weeks.
The questionnaire was based on an earlier version developed and extensively piloted and tested for the City and Hackney survey12 and on questions from surveys in the United States.3 The original pilot survey tested respondents' understanding and acceptability of different forms of question wording and methods of prioritising on 326 members of community groups in Hackney. Full details of the process of refinement have been reported elsewhere.12 To make it suitable for use in a large national population survey, with up to 100 interviewers and around 2000 respondents, I worked with survey experts at the Office of Population Censuses and Surveys to simplify it for this national application. The number of services listed was reduced from 16 to 12, and services were listed without any examples in brackets (such examples were given in the City and Hackney survey) because examples may have caused bias. The service items were deliberately biasing—for example, treatments for life threatening illness were itemised separately for children, for people aged 75 and over, and with no age specification to assess age biases in relation to these treatments. Respondents were asked to look at a card displaying the 12 services and “choose the four services that you consider the most essential,” then they were asked to “choose the four services that you consider the next most important.” The bottom four services, by deduction, were the four remaining services, and respondents were asked to check the order—”may I just check, that leaves….” The services were coded in the order that they were selected by respondents. The frequency distributions for each service's priority codes were also converted into averages and the averages listed in order of priority to provide the ranking.
Respondents were asked about the extent of their agreement or disagreement (strongly disagree to strongly agree) with six statements about priorities (“I am now going to read out a series of statements about health priorities, please tell me whether you agree or disagree. Choose your answer from this card”). They were asked who they thought should set priorities and asked to select their preference from a precoded list of doctors at local level, the public at local level, local NHS managers, local health authorities, and politicians or government at national level (“If health services rationing is inevitable, who should have most say in setting priorities for health services?”). Finally, they were asked about how they themselves would allocate £100000 (“If you were in charge of a health authority with £100000 left to allocate for your health budgets, which of the following would you choose to do?”). These further sets of questions on priorities were developed for use in the studies in Hackney and the United States and, as with the main priority ranking exercise, subjected to the same process of wording refinement during piloting3 12 and discussions with experts in survey methodology at the Office of Population Censuses and Surveys.
The standard questions of the Office of Population Censuses and Surveys on self rated state of health (“excellent,” “very good,” “good,” “poor,” “very poor”) and on the prevalence of longstanding illness, disability, or infirmity were asked, clarified as “… anything that has troubled you over a period of time or that is likely to affect you over a period of time?” Sociodemographic data were also collected. Results were analysed by the sociodemographic characteristics of the sample and by state of health.
SOCIODEMOGRAPHIC CHARACTERISTICS OF SAMPLE
Comparisons between respondents' age and sex and mid-term estimates for the age and sex structure of the population derived from census data show that respondents were similar to the population as a whole.18 If anything, as with the general household survey,19 there was a slight under-representation of people aged under 30 years but by less than 1% in each band. Table 1 compares some of the sociodemographic and health characteristics of the sample with those of the sample members aged 16 and over from the most recently published British general household survey.19 This shows that the responders to each survey were similar in their sociodemographic characteristics, as would be expected given that sampling methods are the same and both the omnibus survey and the annual general household survey are conducted by the Office of Population Censuses and Surveys. The general household survey is able to link with census data for most of the households sampled, and analyses of the linked data show that the general household survey, and by comparison the omnibus survey, slightly under-represents people who live in London (by <1%) and people living in single occupancy households in comparison with households with two or more people (the non-contact rate was 5.3% in comparison with 2.6% for the latter). Otherwise, the only notable difference is with the proportion who reported a longstanding illness, which is lower in the omnibus survey. Apart from fluctuations in affirmative responses by month of questioning, the total proportion of people reporting a longstanding illness over time is fairly stable in surveys conducted by the Office of Population Censuses and Surveys. The omnibus staff suggested that the difference arose between these two surveys because the question was asked at the end of the omnibus questions rather than at the beginning and was not in the context of a battery of questions about respondents' own health. It was also asked after complex questioning on health service priorities, perhaps leading respondents with milder conditions to reconsider whether they should be labelled as a “longstanding illness, disability or infirmity.”
Table 2 shows the frequency distributions and the mean priority rankings for the 12 services and treatments. The table shows that the highest priority (rank 1) was accorded to “treatments for children with life threatening illness,” the next highest priority (rank 2) was accorded to “special care and pain relief for people who are dying.” “Preventive screening services and immunisations” were ranked next highest (3). “Psychiatric services” was given a middle ranking (6) as was “high technology surgery” (7); “health promotion” was given a middle to low ranking (8). The lowest priorities were assigned to “treatment for infertility” (11) and “treatment for people aged 75 and over with life threatening illness” (12).
A further question on attitudes was asked in which respondents were asked if they strongly disagreed to strongly agreed (on a five point scale) with each of six statements. Table 3 shows the responses to these statements. Most respondents agreed or strongly agreed that high cost technology should be available to all, regardless of age, which somewhat contradicts their bottom ranking of treatments for people aged 75 and over with life threatening illness (table 2), illustrating the complexity of prioritisation by age group. In agreement with earlier research,12 however, most respondents agreed or strongly agreed that the patient's quality of life should be considered. More consistent with the priority ranking exercise of the 12 services, half (979) of respondents agreed or strongly agreed that if resources are to be rationed then higher priority should be given to treating the young rather than elderly people.
Table 3 also shows that respondents were more evenly divided on whether people who contribute to their own illness (for example, through smoking, obesity, or drinking) should have lower priority for health care, although 42% (842) agreed or strongly agreed with this. Most respondents agreed or strongly agreed that the responsibility for rationing spending on health care should rest with doctors rather than managers, health authorities, or the government, echoing a similar direct question asked (see below). Consistent with this again most people disagreed or strongly disagreed with the statement that the government should issue guidelines to doctors about rationing lifesaving treatments.
Respondents were asked who should set priorities and shown a precoded list: 56% (1104) said “doctors at local level,” 19% (377) said “local health authorities,” 17% (336) said “the public at local level,” 5% (89) said “local NHS managers,” and 3% (61) said “politicians and the government at national level.” In reply to a separate question 88% (1739) said that they thought that “surveys of the general public's opinions, like this one, should be used in the planning of health services,” 7% (149) disagreed with this, and 5% (91) said that they did not know.
They were also asked how they themselves would allocate £100000. Seventy one per cent (1393) of respondents selected “a health screening and education programme which could prevent a large number of people needing lifesaving operations in the future (for example, screening for cancers)” and 26% (521) selected “12 extra immediate lifesaving operations this year (for example, heart bypass)”; (3%: 65 said they did not know).
There were few associations between health service priorities and sociodemographic characteristics, although those that were found did make theoretical sense. For example, 44% (426) of people aged between 16 and 45 prioritised as their first choice “treatments for children with life threatening illness,” in comparison with 26% (218) of people aged between 45 and 75 and 21% (30) of people aged 75 and over (χ2=79.29; df=2; P<0.001); and whereas just 1% (15) of people aged under 75 years prioritised as first “treatments for people aged 75 and over with life threatening illness,” 10% (14) of people aged 75 and over prioritised this as first (χ2=75.13; df=1; P<0.001). Forty six per cent (300) of people with one or more children aged under 16 and 28% (375) of those with no children ranked as first “treatments for children with life threatening illness” (χ2=64.77; df=1; P<0.001).
The study presented here is the first prioritisation exercise based on a random sample of a total (national) population. The methodology of ranking lists of treatments and services may be criticised as superficial in relation to the complexity of the decisions to be made about health service priorities, which necessitate consideration of the costs and effectiveness of treatments and care programmes rather than sole reliance on values that may include prejudices. The prioritisation exercise presented here mostly entailed the ranking of broader treatments and services for specific groups of people rather than ranking individual procedures and diagnostic groups as in the Oregon experiment.1 This focus was deliberate to measure the public's values in relation to specific groups of patients and age groups. In the context of a lack of adequate knowledge about the costs and effectiveness of much medical care it is important to be democratic and involve everyone in an open debate about rationing. One first step must be to measure baseline public opinions and values. If the public's values seem to conflict with firm medical evidence on effectiveness or to be prejudiced against certain groups then open debate and the provision of sound, unbiased information for public consumption and education is even more essential.
Probably the most important shortcoming of the setting of public priorities is that priorities chosen by the public do not necessarily offer the most equitable solutions in relation to the original aspiration of the NHS of equal treatment for equal need. Overall, this research confirms the results of earlier surveys which showed that the public's priorities are not value free—they are most likely to prioritise treatments specifically for younger rather than older people and particularly lifesaving treatments12 13 16; it also shows some public support (42%) for people with self inflicted conditions receiving lower priority for care, which raises ethical issues. It clearly shows that different groups of people—for example, age groups—hold different values that need to be reconciled in policies on rationing. The debate about how to weight different value systems in every decision on the allocation of resources is in its infancy as is the education of the public when prejudices are detected in their priority setting. These issues were not satisfactorily resolved in the largest exercise on rationing health care in Oregon.1
Finally, most people wanted to be involved in the planning of health services. Three quarters thought that the responsibility of rationing spending on health care should rest with doctors rather than managers, health authorities, or the government. Health authorities should listen to the public's views on health priorities to add legitimacy to their decision making, given their own position as democratically unaccountable bodies. They also need to be seen by the public to be working with and not against their clinical colleagues in prioritisation or rationing exercises to retain the trust of the public.
I thank the staff of the Office of Population Censuses and Surveys for carrying out the survey and producing the results, and in particular Jack Eldridge and his colleagues for their invaluable advice on questionnaire design. Those who carried out the original analysis and collection of the data hold no responsibility for the further analysis and interpretation of them. Material from the omnibus survey, made available through the Office of Population Censuses and Surveys, has been used with the permission of the comptroller of HMSO. The dataset is held on the data archive of the Economic and Social Research Council at the University of Essex.
Funding Internal funds.
Conflict of interest None.