Opinion polls and attitudinal surveys have been criticized

Transcription

1 Original paper 429 Establishing patient preferences for gastroenterology clinic reorganization using conjoint analysis Paul Moayyedi a, Mark Wardman b, Jeremy Toner b, Mandy Ryan c and Sara Duffett d Objective The Department of Health states that patients with suspected cancer should be seen within 2 weeks, and the Patients' Charter suggests that patients should not wait for more than 30 min in outpatients. Decisions such as these are often made with little assessment of patient preferences. We have elicited patient preferences for the optimal use of time in the outpatient clinic. Design Questionnaire survey eliciting preference between different clinic scenarios evaluated using discrete choice conjoint analysis. Setting and participants Patients attending a teaching hospital gastroenterology outpatient clinic. Main outcome measures The relative importance of time spent on the waiting list, time waiting in the clinic, time spent with the specialist, and time waiting for investigation was assessed using a logit model. Results Patients placed a similar value on waiting for investigation and time spent on the waiting list. A clinic that had a 2-month waiting list but offered immediate investigations would therefore be more popular than a clinic that had a 2-week waiting list but whose investigations were deferred for 3 months. Patients would be prepared to spend an extra 30 min in the waiting room if they spent 1 month less on the waiting list or waiting for investigation. Time spent with a specialist is valued, and patients would be prepared to spend an extra 3 min waiting in the clinic for every extra minute spent with the doctor. Conclusions The present Department of Health recommendations and the Patients' Charter are too simplistic and do not take into account patient preferences. Eur J Gastroenterol Hepatol 14:429±433 & 2002 Lippincott Williams & Wilkins European Journal of Gastroenterology & Hepatology 2002, 14:429±433 Keywords: patient preference, conjoint analysis, clinic re-organization a City Hospital NHS Trust, Birmingham, b Institute for Transport Studies, University of Leeds, Leeds, c Health Economics Research Unit, Department of Public Health, University of Aberdeen, Aberdeen, UK, and d Centre for Digestive Diseases, General In rmary at Leeds, Leeds, UK Correspondence to Paul Moayyedi, Professor of Gastroenterology Health Services Research, City Hospital NHS Trust, Dudley Road, Winson Green, Birmingham B18 7QH, UK Tel: ; fax: ; p.moayyedi@bham.ac.uk Received 3 September 2001 Revised 26 October 2001 Accepted 5 December 2001 Introduction The Department of Health has proposed that patients with suspected gastrointestinal malignancy should wait a maximum of 2 weeks before being seen by a specialist [1]. This will have implications on the organization of outpatient clinics, especially as a previous white paper indicated that patients should not wait more than 30 min in the clinic before seeing a specialist [2]. These targets can be met either by expanding one-stop/openaccess endoscopy services or by shortening the time to see a specialist but increasing the waiting time for investigation. Health service managers and healthcare professionals usually make these decisions without consulting patients. The need to elicit patients' views on the care that they receive has been recognized [3, 4], and surveys have shown that the public's priorities can differ from those of clinicians and managers [5]. Opinion polls and attitudinal surveys have been criticized for not encompassing the notion of sacri ce in the decision-making process [6]. It can be anticipated that patients would prefer to spend a minimal time on a waiting list, be seen immediately, spend a long time discussing their symptoms with the specialist, and have the relevant investigations without delay. This is an unrealistic expectation for a publicly funded health service, and in practice some desirable attributes of a clinic will be achieved at the expense of others. There is therefore a need to establish the relative importance to patients of different outpatient clinic attributes so what is offered can be tailored more closely to what patients desire. This is a problem that is not peculiar to healthcare but applies to any situation where a supplier wishes to differentiate the product offered to better satisfy the wants of a target market. Market researchers have developed methods of eliciting consumer preferences 0954±691X & 2002 Lippincott Williams & Wilkins

2 430 European Journal of Gastroenterology & Hepatology 2002, Vol 14 No 4 based on the notion of sacri ce. Discrete choice conjoint analysis is one of these techniques and is `the one most likely to yield fruitful results in the valuation of quality of delivery in the public services' according to an HM Treasury report [6]. The technique involves asking respondents to choose between two competing scenarios with different attributes so that a coef cient of preference can be established for each attribute. This is considered to provide the most reliable responses, since the other possible responses of rating or ranking alternatives do not feature in real-life decision making. The main advantage of conjoint analysis within the health sector is that there are rarely real markets where patients can make choices that reveal their preferences amongst different features of health service provision. Conjoint analysis has been used extensively in marketing research, transport economics and geography, and has recently been applied to the health service, particularly in relation to in-vitro fertilization, miscarriage management, and the value of time on a waiting list for elective surgery [7±13]. To our knowledge, this technique has not been applied to the detailed evaluation of the optimum use of time in the outpatient clinic. We assessed patients' preferences towards gastrointestinal clinic services using conjoint analysis. Methods Discrete choice conjoint analysis involves the distinct stages of selecting the attributes to characterize, assigning levels to each attribute, data collection, data analysis and interpretation. Establishing the attributes to investigate The need to detect gastrointestinal malignancy early was discussed among four consultant gastroenterologists in one teaching hospital followed by wider consultation with nursing and clerical outpatient staff. Four key attributes were identi ed that could be altered to improve outpatient services. These were waiting time from the GP's referral, waiting time in the clinic, consultation time with the specialist, and waiting time for investigation. Appropriate levels of the attributes Conjoint analysis provides a means of approximating laboratory-controlled conditions because it controls the `stimuli' offered to decision makers. It is therefore important that this advantage is exploited to the fullest possible extent. The attribute levels need to be realistic, contain trade-offs across attributes, and allow the choices that are made to yield the maximum information for the purpose of establishing patients' preferences. The permissible ranges for each attribute were chosen with reference to national and local gures. These were 1±6 months for time on a waiting list, 0±60 min for waiting in the clinic, 5±30 min for time spent with a doctor, and 0±6 months for waiting for investigations. We carried out a pilot survey of 34 patients attending gastrointestinal outpatient clinics to obtain an approximate impression of patient preferences. A state-of-theart optimization procedure [14] used these data to select the attribute levels offered to patients with the explicit aim of minimizing the variance of the parameter estimates. This generated a conjoint analysis questionnaire containing nine choices between two alternatives (Fig. 1). Data collection The local research ethics committee approved the study. Patients attending gastrointestinal outpatients in one teaching hospital between 15 May and 15 June 1999 were invited to complete the conjoint analysis questionnaire. Participants were recruited from one of three clinic types: a one-stop endoscopy clinic, a normal outpatient service, and a colitis clinic. The one-stop endoscopy clinic is for patients with dyspepsia who are seen by a gastroenterologist, have an upper-gastrointestinal endoscopy, and receive the result on the same day. The normal outpatient service is offered to any patient with upper- or lower-gastrointestinal symptoms; patients see a specialist, are referred for the relevant investigation some time in the future, and return to the clinic for their results. The colitis clinic is for patients with ulcerative colitis or Crohn's disease and consists largely of patients who are stable and undergoing longterm follow-up for their condition. The type of clinic the patient attended and whether this was a rst or follow-up appointment was recorded. Data were also collected on the patient's age, gender, socioeconomic category, income, educational achievement, and fear of serious underlying disease, since these factors may in uence choices between scenarios. Fig. 1 Please choose which clinic you would prefer in each of the following nine scenarios: Scenario 1 Waiting time for first appointment Waiting time at the clinic to see the doctor Time spent with the doctor Waiting time for tests/treatment 1. Which clinic would you prefer? (tick one box only) Clinic 1 1 month 60 minutes 20 minutes 6 months Example of the questions presented to patients. Clinic 2 6 months Immediate 10 minutes Immediate Prefer Clinic 1 Prefer Clinic 2

3 Patient preferences for gastroenterology clinic reorganization Moayyedi et al. 431 Concept and means of estimating the relative importance (valuation) of attributes The choices made by participants will depend on how highly they value each of the clinic attributes. This value can be quanti ed using a logit model. This is a statistical technique that calculates coef cient weights for each attribute to denote their relative importance in a manner analogous to multiple regression. A positive coef cient implies that patients desire that attribute, and a negative coef cient implies that the attribute is disliked. The size of the coef cient indicates the strength of preference. Thus, if an increase in minutes of time spent with the specialist has a coef cient of 1, and an increase in minutes waiting in the clinic has a coef cient of 0.5, then patients value time with a specialist twice as highly as time waiting in the clinic. The relative importance of each of the four attributes was estimated using this logit model, and the in uence that fear of underlying illness, socioeconomic status, and demographic variables had on these preferences was also evaluated. The model results were used to evaluate four hypothetical scenarios (Table 1). Scenario one is based on national and local averages. Scenario two represents a possible response to the Department of Health directive, whereby patients with suspected cancer are `fasttracked' to be seen within 2 weeks but wait 13 weeks for the relevant investigation. Scenario three characterizes a one-stop clinic and does not conform to either the recent Department of Health proposals or the Patients' Charter. Scenario four is a combination of the one-stop and fast-track approaches. The rst three scenarios could be achieved by reorganizing existing clinic resources, whereas it could be argued that scenario four would require additional funding. The monetary value of each scenario was not assessed directly but was estimated by assuming that waiting in a clinic has a similar disutility to waiting for public transport. Results Three hundred and eighty-nine patients were invited to complete the questionnaire, and 354 (91%) participated. The mean age was 47 years (range 17±81 years), 164 (46%) were male, and 181 (51%) were afraid of severe underlying illness. One hundred and ninety-four attended a normal outpatient service, 115 attended the colitis clinic, and 45 attended a one-stop endoscopy clinic. Patients experienced a median time on the waiting list of 2 months (range 0±6 months), waited a median of 25 min (range 0±90 min) in the waiting room, and spent a median of 10 min (range 2±30 min) with the gastroenterologist. The analysis is based on 2897 choices after missing values were excluded. The coef cients of the estimated logit model are reported in Table 2. The negative sign of the coef cients for time spent on the waiting list and waiting for investigations indicate, as expected, that patients are averse to waiting. Overall, these two attributes have very similar values. Incremental effects were detected for males and those aged > 50 years, who were more averse to time spent on a waiting list. Those attending a one-stop clinic valued waiting time for the investigation 72% more highly than others; these patients formed a `self-selected' sample, since those with higher values of waiting time for the investigation will tend to choose the one-stop clinic. Waiting time in the clinic and time spent with the doctor were both speci ed in minutes and, as expected, indicate that patients are averse to waiting time in the clinic but prefer more time with the doctor. The exception to the latter is those patients attending a colitis clinic, who have no value of additional time with the doctor. The values of the various attributes can be put on a common scale by expressing them in equivalent units of waiting time at the clinic. This involves dividing each coef cient by the coef cient for waiting time in the clinic. The resulting values are given in the nal column of Table 2. For example, patients would be prepared to incur an additional min waiting in the clinic in order to achieve a 1-month reduction in waiting for investigation. Patients' preferences are fairly homogeneous. We investigated the impact of a wide range of socioeconomic, situational and demographic variables on choices, but only the four contained in Table 2 merited retention. The results were used to evaluate the four scenarios listed in Table 1. Application of the model showed that Table 1 Outpatient clinic scenarios evaluated Scenario Wait to see specialist (months) Wait in clinic (min) Time with specialist (min) Wait for investigation (months) Per person bene t 1 `Normal' outpatients Base 2 Fast-track outpatients One-stop outpatients Fast-track one-stop

4 432 European Journal of Gastroenterology & Hepatology 2002, Vol 14 No 4 Table 2 Estimated conjoint model for patient preferences for time spent in the clinic a Attribute Coef cient 95% CI P value Valuation b Time on waiting list c to 0.219, Male to Age.50 years old to Time waiting in clinic d to 0.007, ± Time with specialist d to 0.023, Colitis clinic patients to 0.024, Time for investigation c to 0.214, One-stop patients to 0.183, a The coef cients in normal type are the baseline values and the coef cients in italics represent the extra utility gained by patient subgroups. For example, the coef cient for the baseline preference for time on a waiting list is This value is for women aged,50 years. For men the coef cient is 0.183±0.032 ˆ (i.e. a greater disutility for waiting in the clinic). For patients aged.50 years, the coef cient is 0.183± ˆ b Valuation is minutes willing to spend in the waiting room to gain a unit improvement of the other attribute. c Per month. d Per minute. patients preferred scenario four to three, preferred scenario three to two, and preferred scenario two to one. In relative terms, scenario four is 1.8 times as good as scenario three, three is 1.33 times as good as two, and two is 1.6 times as good as one. If scenarios one, two and three coexisted, we estimate that 45% of patients would prefer the one-stop clinic, 35% would choose the fast-track clinic, and only 20% would choose the current situation. The monetary value patients would be willing to pay to transfer from one scenario to another was estimated by assuming that waiting in the clinic has a similar disutility to waiting for public transport. The Department of the Environment, Transport and the Regions' recommended values updated to 1999 gures suggest subjects would be willing to pay 0.15 for each minute of wait avoided [15]. Using this value, patients would be prepared to pay 0.46 per minute with a specialist. This translates to a full-time salary of per annum for the specialist. Patients would also be prepared to pay to move from the current situation to a fast-track service, to move from the current to a one-stop clinic, and to move from the current to a combined fast-track one-stop service. These values are for people who attend a conventional clinic; those attending a one-stop clinic have an even stronger preference for that type of service delivery. Discussion The initiative to improve outpatient services for patients with suspected cancer has concentrated on time spent on the waiting list. This study indicates that focusing on one aspect of the clinic process does not re ect patient preference adequately. Our data suggest patients value waiting for investigations as highly as time spent on a waiting list, as a reduction in either will lead to a more rapid diagnosis. Similarly, patients would be willing to spend over 30 min waiting to be seen in the clinic if their time on the waiting list is reduced by 1 month. Patient preferences were remarkably homogeneous in this study, although those attending with a chronic illness for a routine appointment placed a much lower value on time with a specialist. Patient preferences are related to their expectations, and this is emphasized by the additional utility placed on immediate investigation by those patients already attending a one-stop service. Patient expectations have shaped the UK National Health Service, and results may be different in other healthcare systems [16,17]. This is emphasized by the scenario of waiting 6 months after seeing a specialist before receiving investigation and treatment. This length of wait may be totally unacceptable to patients in healthcare systems with much shorter waiting times. This study elicited preferences from patients attending gastroenterology clinics, and further research is required to evaluate whether these results will pertain to users of other specialties. It has been argued that the views of the community are the most relevant to a publicly funded health service, and preferences should be obtained from the general population rather than patients [18]. The problem with eliciting values from the community is that this will include people with no knowledge or experience of outpatient services. The hypothetical choices will be of limited meaning to this group. Furthermore, studies evaluating community preferences often have lower response rates than achieved in this and other studies assessing patient populations. Patients appear to be willing to sacri ce one desirable clinic attribute for another, therefore the optimal use of time in the outpatient setting will be achieved only if the components are considered as a whole. We have compared two scenarios that may be achieved by reorganizing existing services. A fast-track service that offers a specialist appointment within 2 weeks but does

5 Patient preferences for gastroenterology clinic reorganization Moayyedi et al. 433 not reduce waiting time for investigation is preferred over the existing outpatient service. A one-stop clinic that offers immediate investigation but does not conform to present Department of Health guidelines or the Patients' Charter was still preferred to the fast-track or existing clinical service. This again emphasizes the importance of eliciting patient preferences and considering all components of time spent in the outpatient service rather than targeting one aspect. The scenarios we presented are hypothetical, but the results can be used to evaluate patient preferences to other clinic schemes that re ect local needs. The optimal clinic would combine a fast-track and onestop approach so patients with suspected cancer are seen within 2 weeks and have their investigation immediately. This may require additional resources. We have presented a novel method of placing a monetary value on clinic reorganization. This avoids asking patients to directly value the service, and uses waiting time in the clinic as a surrogate marker. The advantage of this is that patients are used to waiting in clinics but are not familiar with directly paying for the service. Further research is needed to establish the validity of the assumption that the monetary value of the disutility of waiting in clinic is similar to waiting for public transport. The method has some face validity, however, as it predicts that the value patients place on doctors' time is similar to the salary of a junior consultant. These data suggest patients would be willing to pay over 32 for the fast-track/one-stop combination compared with a `normal' outpatient service. There would therefore be a net bene t if an additional were made available for every 1000 patients seen in this clinic scenario. These data may be a useful guide to healthcare planners allocating additional resources to improve clinic services. the application of conjoint analysis. Luton: Of ce of Health Economics, White Crescent Press; Ryan M. Using conjoint analysis to go beyond health outcomes: an application to in vitro fertilisation. Soc Sci Med 1999; 8:535± Ryan M, Hughes J. Using conjoint analysis to value surgical versus medical management of miscarriage. Health Econ 1997; 6:261± Propper C. The disutility of time spent on the United Kingdom's National Health Service Waiting Lists. J Hum Resour 1995; 30:677± Ryan M, Farrar S. Using conjoint analysis to elicit preferences for health care. BMJ 2000; 320:1530± Bryan S, Buxton M, Sheldon R, Grant A. Magnetic resonance imaging for the investigation of knee injuries: an investigation of preference. Health Econ 1998; 7:595± Van der Pol M, Cairns J. Establishing preferences for blood transfusion support: an application of conjoint analysis. J Health Serv Res Policy 1998; 3:70± Toner JP, Wardman M, Whelan G. Testing recent advances in stated preference design. In: Proceedings of European Transport Conference, Cambridge, September. London: PTRC Education and Research Services Ltd, Seminar F, Transportation Planning Methods; pp. 51± Department of the Environment, Transport and the Regions. Highways economics note 2, November In: Design Manual for Roads and Bridges, vol. 13. London: TSO; Porter M, McIntyre S. What is, must be best: a research note on conservative or deferential responses to antenatal care provision. Soc Sci Med 1984; 19:1197± Salkeld G, Ryan M, Short L. The veil of experience: do consumers prefer what they know? Health Econ 2000; 4:4±9. 18 Strull WM, Lo B, Charles G. Do patients want to participate in medical decision making? JAMA 1984; 252:2990±2994. Acknowledgements Paul Moayyedi is funded by an MRC Health Services Research Fellowship. Mandy Ryan is funded by an MRC Senior Fellowship. We are grateful to Professor A.T.R. Axon, Dr D.C. Chalmers and Dr S. O'Mahony for allowing us to interview their patients. References Cleary PD. The increasing importance of patient surveys. BMJ 1999; 319:720± Groves T. Public disagrees with professionals over NHS rationing. BMJ 1993; 306: Blendon RJ, Marttila J, Benson JM, Shelter MC, Connolly FJ, Kiley T. The beliefs and values shaping today's health reform debate. Health Aff 1994; 13:274± Cave M, Burningham D, Buxton M, Hanney S, Pollitt C, Scanlan M, Shurmer M. The Valuation of Changes in Quality in the Public Services. London: HMSO; Ryan M. Using Consumer Preferences in Health Care Decision Making: