A Scott, S Simoens, D Heaney*, C A O’Donnell+, H Thomson#, K J Moffat+, S Ross**, N Drummond**
Health Economics Research Unit, University of Aberdeen;
*Department of Community Health Sciences (General Practice) University of Edinburgh;
+Department of General Practice, University of Glasgow; #MRC Social and Public Health Sciences Unit, University of Glasgow;
**Sunnybrook and Women’s College Health Science Centre, University of Toronto.
SMJ 2003 49(1): 61-66
Abstract:
Background: The changes to out of hours care provided by General Practitioners have led to wide variation in the types and costs of out of hours care across the country. Aim: To examine the costs of different models of service delivery for GP out of hours organisations. Methods: This was a prospective cross-sectional survey of eight GP out of hours organisations and samples of their patients. A deputising service, rotas, and various types of GP co-operative across Scotland were surveyed. Information on the quantities of resources used by each organisation was combined with unit costs. Costs incurred by patients and other NHS costs subsequent to the out of hours contact were also calculated. Results: Annual costs incurred by the GP out of hours organisation per 1,000 population ranged from £2,916 to £12,120. There was no relationship between costs and type and size of organisation. There was a three-fold variation in total costs per out of hours contact (£15 to £51). Costs per phone contact were lowest (£6 to £11), followed by cost per centre contact (£10 to £16) and cost per home contact (£21 to £60). Total costs per episode ranged from £78 to £136 for centre contacts, from £130 to £303 for home contacts, and from £70 to £553 for telephone contacts. Home contacts had the highest average cost per episode (£212), followed by telephone contacts (£117) and centre contacts (£85). Conclusions: There are wide variations in the costs of operating GP out of hours services. It is likely that the context in which organisations were set up and local geography influence variations in costs, as well as the level of GP cover.
Key words: Costs, primary care, out of hours, organisation of care.
Introduction
Since early 1995 far-reaching changes have taken place in the provision of general medical services outside normal surgery hours. GP co-operatives have been the most popular innovation, although these arrangements have varied considerably in their composition, organisation, and patterns of service delivery.1,2,3 The advent of NHS24 in Scotland and NHS Direct in England has also meant changes to the call handling system and the integration of services. However, in rural and remote areas, changes have been more difficult to achieve and there is greater diversity in the type of service provided.4 More recently, the new GP contract to be introduced in 2004 will enable GPs to opt out of their 24-hour responsibility of care for patients, with such services having to be provided by the local primary care organisation. Research to date has focused on the process of providing out of hours care and on outcome measures, including access, quality of care, patient and doctor satisfaction5. Studies have rarely addressed more than one aspect of out of hours provision and have often been limited to a small geographical area. Furthermore, most literature since 1995 has focussed exclusively on GP co-operatives in urban areas. Most empirical comparisons have been with deputising services, rather than other traditional models of care. It has not been possible to judge whether GP co-operatives are an improvement over more traditional models of out of hours care.
There has been no adequate research on the comparative costs of different models of out of hours care provided by GPs.6 Lattimer et al(2000) compared the costs of nurse telephone consultation with usual GP care in one GP cooperative. 7 However, the study did not report average costs per patient for an out of hours visit, and there were few details of costing methods. A study by Brogan et al(1998) examined the costs of all out of ours services (including accident and emergency departments) in one health authority.8 However, there was no comparison with other models of GP out of hours care. Finally, Hallam and Henthorne(1999) examined some aspects of financing and sources of income for seven GP co-operatives.3 However, they did not examine the economic value of resources used, which is most appropriate when attempting to compare the efficiency of out of hours organisations. Furthermore, studies to date have only considered costs incurred by the NHS and not those borne by patients. This is important because GP co-operatives have shifted the location of visits, and therefore may have increased costs to patients. The aim of this study was to examine and compare the value of resources used by a range of GP out of hours organisations in Scotland. As well as calculating unit costs for each type of visit within each organisation, costs per out of hours episode (ie including prescribing, day time GP contacts, further GP out of hours contacts, secondary care contacts, patient time and travel costs) are estimated. The results of this study will be useful for policy makers wanting to alter the nature of out of hours care provision, and also for those wanting to use unit costs data in other evaluations of GP services.
Methods
This research was conducted as part of a larger study examining a range of issues across 15 out of hours organisations across Scotland.9 Each organisation was selected to reflect diversity in the type of organisation, the size of the patient population covered, geography, deprivation, services provided by the out of hours organisation (eg nurse triage, transport for doctors/ patients), and management structure. For the purposes of comparison we included small practices providing their own services, a deputising service, rotas and various types of GP co-operative.
The aim of the analysis was to calculate all costs incurred during an episode of out of hours primary care. This is defined as from the point of contact with the out of hours provider to one week after the contact. For the purposes of comparison across models, out of hours services were defined as general medical services provided after 6pm and before 8am on weekdays and at weekends from Saturday noon to Monday at 8am.
The total cost of an out of hours episode was examined with respect to three components. First, costs incurred by the out of hours organisation; second, other NHS costs related to the out of hours episode; and third, the time and travel costs of patients. Therefore, the total cost of an out of hours episode reflected a mix of organisation specific (top-down) costs and patient specific (bottom-up) costs. All resource use and cost data were collected throughout 1999. The study adopted a societal perspective and costs were expressed in 1998/1999 prices.
Costs of the out- of of--hours or organisation
A postal questionnaire was sent to the manager or lead GP of each out of hours organisation. Further information was obtained by a follow-up letter, and our calculations were fed back to each organisation for comment. Information on the resources used by the out of hours organisation included staff (managers, administration, nursing), GP time, patient and GP transport (cars, drivers, communication, medical equipment), communication with the centre, computers and medical equipment within the premises, premises (including power), and consumables. The total amount of GP time per year was calculated based on opening hours of centres and the number of GPs covering each shift. Annual expenditure for each item based on market prices was estimated.
The questionnaire asked for information on quantities of resources used (eg whole time equivalents of managers), unit costs (eg remuneration per hour or session), and total annual expenditure for each item (ie quantities multiplied by unit costs). Where data were missing, assumptions were made based on data from other organisations, adjusted for the size of the population they serve. Rather than conduct a sensitivity analysis, information about the assumptions used were fed back to sites and the feedback and comments received were used to refine the calculations. This ensured that assumptions were acceptable to each site. Descriptive information was cross-checked with information provided in the interviews that took place as part of the larger study.9 The annual total cost for each organisation was used tocalculate an average cost per contact, and an average cost per 1,000 population.
Costs of centre, home and telephone contacts were calculated by allocating the relevant organisational cost to each type of contact on the basis of the numbers of each type of contact as a proportion of total contacts. Thus, the costs of management, administration, telephone equipment and costs of telephone calls, computers, and consumables were allocated across all three types of contact. The costs of overheads (ie power and cleaning) were allocated across centre and telephone contacts only. Equipment kept at the centre was allocated to centre contacts only. Premises costs for consulting rooms, reception areas and rest rooms were allocated to centre contacts only, while premise costs for offices were allocated across all three types of contact. Costs of cars, car communication, car equipment and drivers were allocated to home contacts only.
Average consultation lengths for each type of contact and for each site were obtained from the patient questionnaires (see below) and used to calculate the cost of GPs’ time for each type of contact. The value of GPs’ time during the night was based on the hourly fees paid to GPs by some out of hours organisations. These vary across organisations and, hence, one unit cost was used across all organisations. In some organisations, these fees are set so that total payments to GPs are equal to total subscriptions. In this case, the fee is set administratively, and does not reflect a true market price. In these cases the value of GP time could not therefore be determined from local data. However, in one organisation that had difficulty recruiting GPs, the organisation increased fees paid until all shifts were covered. These fees therefore more closely reflected the market value of GP time, and an estimate of £42 per hour was used across all organisations.
The opportunity costs of capital and premises (ie cars, medical equipment, communication equipment, and information technology) were based on market prices and annual equivalent costs were calculated based on a discount rate of 6%.
Collection of other NHS and patients’ costs
A postal questionnaire was distributed to a consecutive sample of two groups of patients attending each of the selected out of hours organisations: patients over 65 years old and parents of children under the age of five years. These groups were chosen as they represent the largest proportion of workload and as the services they receive were felt to be of particular importance, encapsulating the main issues surrounding the debate about out of hours provision.10 Information about other NHS and patient costs related to the out of hours contact were gathered for one week after the initial out of hours contact. This included other visits to health care providers as well as patients’ out of pocket and time costs. This was to capture the more immediate resource use attributable to the out of hours visit. The questionnaire was also used to gather other information on experiences and satisfaction of out of hours care, not reported here.9
The questionnaire was also used to gather information about the average length of each type of contact. This was used to help calculate the unit costs of a contact in each organisation (see previous section). Other primary care 63 Scott, Simoens, Heaney, O'Donnell, Thomson et al The costs of GP out of hours care costs included drug costs, GP surgery visits and home visits during the day, and other out of hours contacts that were related to the initial contact. For dispensed drugs, the assumption was made that GPs dispensed one day’s worth of drugs, unless information to the contrary was available. The cost of a daytime GP consultation or the cost of a GP home visit was used.9
Secondary care costs included the GP arranging an ambulance and admissions to hospital. The cost of providing an ambulance service, based on staffing costs and variable operating costs during out of hours periods, was derived from a study by Brogan et al(1998).8 Information on the hospital and the ward to which patients were admitted was obtained from the patient questionnaire. Unit costs by specialty were obtained from the Scottish Health Service Costs11. It was felt that the decision to refer patients to hospital was part of the out of hours episode and, hence, a cost was attached to attending hospital for one day. However, data on any subsequent treatment initiated by hospital staff was not collected, as this is less likely to be related to GP-based provision of out of hours care.
Patients’ travel costs were based on information on the means of transport to the out of hours organisation. Travel costs were only calculated for centre visits. Time taken up by the out of hours contact covered both time spent on travelling to and from the out of hours organisation (if appropriate) and time of consultation. The opportunity cost of time was based on what the patient would otherwise have done with his/her time. Non-work time was valued at 43% of the average wage rate.12,13 Average wage rates were used if the patient was otherwise working. The costs to any adult companions to the visit were also calculated, including travel costs and the opportunity costs of time forgone.
Results
Fourteen organisations responded to the costing questionnaire, data were useable for only eight. In particular, it was not possible to calculate reliable costs for the five rural sites providing their own cover. These sites did not keep any routine records of out of hours contacts, and it was felt impractical to review medical notes. As the organisation of out of hours care is so different in these sites, it was felt to be inappropriate to apply assumptions from the other sites to these sites. The response rate to the patient questionnaire was 52.3% (2284/4361) across 13 sites (two single-handed sites felt it was not possible to collect enough patient contacts in the time available). The quality of data from the remaining out of hours organisations was reasonable. The most common items of missing data were for overheads and premises – these were estimated using mean costs across other sites.
Costs incurred by the out- of of--hours or organisation
Annual total costs for the organisation ranged from £54,506 for a semi-rural rota covering 15,000 patients, to £3.2 million for a GP co-operative covering 350,000 patients in a rural health board with one large urban area (Table I). Costs per 1,000 population varied from £2,916 for a commercial deputising service to £12,120 for a medium sized urban/rural GP co-operative. These two organisations also had the lowest and one of the highest rates of contacts per 1,000 population (ie demand), respectively. Three organisations with the lowest total and operating cost per 1000 population (between £2,900 and £3,600, and £950 and £1,070 respectively) included a large deputising service, a large urban co-op that subcontracted home visits to a deputising service, and a small semi-rural rota. In addition, a small rural/urban co-operative providing a basic service also had low operating costs per 1000 population (£1,044). There was a three-fold variation in total cost per contact, from £15 for a semi-rural rota to £51 for a GP co-operative covering 350,000 patients in a largely rural health board with a major urban area.
There was more variation in GP costs per population (standard deviation of £2733) than in operating costs per population (standard deviation of £614). There is little evidence that total costs per contact were dependent on the size of the organisation. Some organisations covering less than 100,000 people had higher average costs than those covering more than 200,000 people, although the reverse was also true for other organisations. These differences were less pronounced for operating costs per contact. Across the sites, between 65-84% of total costs were for GP time, 9-21% of total costs were for other staff, 1-11% were for premises, 1-11% for transport, 0.5-11% for communications, 0-2% for equipment and 0.2-1.1% for consumables.
The costs of centre contacts varied from £10 to £16, home contacts were between £21 and £60, and phone contacts varied between £6 and £11 (Table II). There seemed to be more variation in home contact costs, possibly due to geography. Home contact costs were higher than phone or centre contacts, mainly because of the time GPs spent travelling, but also because of the costs of transport and drivers.
Costs of an out of hours episode
For centre contacts, total costs per out of hours episode ranged from£82 to £136(both medium sized cooperatives, but with different geographical catchment areas), with no obvious patterns by type of organisation (Table III). There were also few patterns for other primary care costs (between 6% and 14% of total episode costs), secondary care costs (between 53% and 79% of total episode costs) or patients’ time and travel costs (between 6% and 15% of total episode costs). Costs for home contacts ranged from £166 to £303 per out of hours episode ( a large co-operative and a medium co-operative respectively). There were no obvious patterns by type of organisation (Table IV). Between 58% and 81% of these costs were secondary care costs and between 6% and 16% were other primary care costs. Total costs per out of hours episode for telephone contacts varied much more widely than for centre and home contacts, from £70 to £553 (small co-operative and a medium cooperative respectively) (Table V). Across all three types of contact, the same medium sized urban GP co-operative seemed to have the highest secondary care and therefore total costs per episode. Finally, the average costs per out of hours episode across sites are shown in Table 6. Home contacts had the highest cost per episode (£212), followed by telephone contacts (£117) and centre contacts (£85). Home contacts had the highest primary and secondary care costs. Centre contacts had the lowest primary care costs following an out of hours contact.
Discussion
There are wide variations in the costs of operating out of hours services. There was a four-fold difference in total costs per 1,000 population and in average costs per contact. The cost of a home contact was higher than a centre contact, which was higher than a telephone contact. It must be emphasized that the quest for a single cost of a health care service than can be applied nationally is not realistic, since costs are comprised of local variations in the quantities of resources used and local variations in the value of GP’s time and the prices of other resources. Variations are to be expected. The main issue is determining the causes of such variations as they may reflect an inefficient use of resources. Overall, there was little evidence of a relationship between cost per patient episiode or per 1000 population and the model of service delivery. The study by Brogan et al(1998) estimated GP co-operative costs to be £5,190 per 1000 population and £27.12 per contact, both figures falling in the middle range of the data we have presented in this paper. However, some patterns emerge from these data. Excluding the cost of GP time (which was up to 84% of total cost), the operating costs per 1000 population of some GP co-operatives were almost double that of rotas and deputising services. GP co-operatives generally had a more formal management and administrative structure, more formalized clinical governance activities, most operated their own call handling and provided transport and drivers for GPs.9 GP transport also led to higher costs of home visits for some organisations.
There is also less variation in operating costs than in GP costs, suggesting that the level of GP cover is an important driver in the use of resources. This is also likely to influence quality of care. In turn, variations in the level of GP cover is likely to be related geography and the nature of the population covered. The level of GP cover is therefore an important area for further research. Costs per out of hours episode also varied, although there were few obvious patterns to this variation. The costs per episode for home contacts were higher than for centre and telephone contacts. This may reflect the ability of organisations to differentiate between the severity of patients, so that patients with more severe health conditions receive home visits and therefore received more services subsequent to that visit. It may also reflect the likelihood that elderly patients receive more home visits and again receive more services subsequently.
It would be expected that differences in the types of visits across organisations will influence the time and travel costs borne by patients. However, there was no relationship between patients’ time and travel costs and the proportion of visits that were centre visits. The location of the organisations and their populations is more likely to determine patients’ time and travel costs. These results must be interpreted with the following caveats in mind. The organisations were selected as examples of each type of out of hours organisation. However, each organisation may not be representative of its type. The results should not be interpreted as evidence of inefficiency since there may be "legitimate" sources of cost variation, and since efficiency can only be judged in relation to benefits to patients. Data on patient outcomes following the out of hours contact are not routinely collected, so evidence of benefit is difficult to gauge. Legitimate sources of variation include the context of setting up the organisation, urban/ rural factors, deprivation and population characteristics. For instance, one organisation was instructed by the Health Board to cover as much of the Health Board areas as possible.9
A further issue is that assumptions were made where information was unavailable, using estimates from other sites. Rather than conduct sensitivity analysis, these estimates were fed back to each organisation for their comment and were used to further refine cost estimates. This ensured that the estimates and assumptions were perceived to be realistic and acceptable. The costs presented here will also be higher than the level of financing of these organisations, as the data include resources used by the organisations which do not have to be paid for by the organisation or are financed from elsewhere (eg premises). Secondary care costs comprise the largest proportion of total costs per episode. This is particularly the case for telephone contracts. Secondary care costs are determined partly by the GPs decisions about referral, but also by differences in case mix and whether patients follow the advice of the GP. Although this study was not able to determine statistically the reasons for the wide differences in costs, and there were no clear patterns by type of organisation, the context in which these services were set up combined with local geography seemed to play an important role. The determinants of cost variation need to be explored more fully, particularly the role of factors outside the control of the out of hours organization and the level of GP cover. The calculation of costs in economic evaluation is scientific in the theory and principles one uses, but pragmatic in the use of available data of variable quality. This is particularly the case in primary care, where data on the quantities of resources used and unit costs are not routinely available. The aim of this study was to compare the costs of different models of out care, but the lack of data has prevented the authors from calculating the costs of GPs providing their own cover in the more traditional model of GP out of hours care (five rural sites could not provide such information), and so are unable to examine how costs are likely to have changed following the introduction of GP co-operatives. This study is also unable to examine provision by GPs in rural areas, where co-operatives and rotas are not a feasible option. However, the results presented here are still of interest as the dominant model of out of hours care in urban areas is the co-operative model, and this will be further encouraged with the removal of the contractual requirement that GPs provide 24-hour care for their patients. The results of this study will therefore be useful for policy makers wanting to alter the nature of out of hours care provision, and also for those wanting to use unit costs data in other evaluations of GP services.
ACKNOWLEDGEMENTS: We would like to thank the out of hours organisations across Scotland who participated in the study. The project was funded by the Primary Care Development Fund of the Scottish Executive Health Department. The Health Economics Research Unit is funded by the Chief Scientist Office of the Scottish Executive Health Department.
REFERENCES
1 Cragg DK, Hallam L. Quality standards for deputising services. Br Med J 1994; 309: 1630.
2 Jessopp L, Beck I, Hollins L, Shipman C, Reynolds M, Dale J. Changing the pattern out-of-hours: a survey of general practice co-operatives. Br Med J 1997; 314: 199-200.
3 Hallam L, Henthorne K. Cooperatives and their primary care emergency centres: organisation and impact. Combined report on seven case studies. Health Technology Assessment 1999; 3: 1-92.
4 Gillies J, Ross SJ. Recruitment and training for rural general practice: problems and solutions. Glasgow: University of Glasgow, 1997.
5 O’Donnell CA, Drummond N, Ross SJ. Out of hours primary care: acritical overview of current knowledge. Health Bulletin 1999; 57: 276-284.
6 Department of Health. Raising standards for patients. New partnerships in out-of-hours care. London: HMSO, 2000.
7 Lattimer V, Sassi F, George S, Moore M, Turnbull J, Mullee M, et al. Cost analysis of nurse telephone consultation in out of hours primary care: evidence from a randomised controlled trial. Br Med J 2000; 320: 1053-1057.
8 Brogan C, Pickard D, Gray A, Fairman S, Hill A. The use of out of hours health services: a cross sectional survey. Br Med J 1998; 316: 524-527.
9 Heaney D, O’Donnell K, Scott A, Drummond N, Ross S, Moffat K. A comparison of models of delivery of out of hours general medical services in Scotland. Report to the Scottish Executive Management Executive. Edinburgh: University of Edinburgh, 2001.
10 Salisbury C, Trivella M, Bruster S. Demand for and supply of out of hours care from general practitioners in England and Scotland: observational study based on routinely collected data. Br Med J 2000; 320: 618 – 621.
11 Information and Statistics Division. Scottish Health Service Costs, year ended 31st March 1999. Edinburgh: National Health Service Scotland, 1999.
12 Netten A, Dennett J, Knight J. Unit costs of health and social care 1999. Canterbury: Personal Social Services Research Unit, 1999.
13 Office for National Statistics. New Earnings Survey, 1999.