Cost effectiveness of day and inpatient psychiatric treatment: results of a randomised controlled trialBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7091.1381 (Published 10 May 1997) Cite this as: BMJ 1997;314:1381
- Francis Creed, professor of community psychiatrya,
- Patrick Mbaya, research registrara,
- Stuart Lancashire, social research workera,
- Barbara Tomenson, statisticiana,
- Bill Williamsa, research registrar,
- Sarah Holme, health economista
- a University Department of Psychiatry, Rawnsley Building, Manchester Royal Infirmary, Manchester M13 9WL
- Correspondence to: Professor Creed
- Accepted 3 February 1997
Objective: To compare direct and indirect costs of day and inpatient treatment of acute psychiatric illness.
Design: Randomised controlled trial with outcome and costs assessed over 12 months after the date of admission.
Setting: Teaching hospital in an inner city area.
Subjects: 179 patients with acute psychiatric illness referred for admission who were suitable for random allocation to day hospital or inpatient treatment. 77 (43%) patients had schizophrenia.
Interventions: Routine inpatient or day hospital treatment.
Main outcome measures: Direct and indirect costs over 12 months; clinical symptoms, social functioning, and burden on relatives over the follow up period.
Results: Clinical and social outcomes were similar at 12 months, except that inpatients improved significantly faster than day patients and burden on relatives was significantly less in the day hospital group at one year. Median direct costs to the hospital were £1923 (95% confidence interval £750 to £3174) per patient less for day hospital treatment than inpatient treatment. Indirect costs were greater for day patients; when these were included, overall day hospital treatment was £2165 cheaper than inpatient treatment (95% confidence interval of median difference £737 to £3593). Including costs to informants when appropriate meant that day hospital treatment was £1994 per patient cheaper (95% confidence interval £600 to £3543).
Conclusions: Day patient treatment is cheaper for the 30-40% of potential admissions that can be treated in this way. Carers of day hospital patients may bear additional costs. Carers of all patients with acute psychiatric illness are often themselves severely distressed at the time of admission, but day hospital treatment leads to less burden on carers in the long term.
When inpatient treatment is avoidable day hospital treatment is cheaper for acutely ill psychiatric patients
Carers of acutely ill psychiatric patients experience severe distress and warrant help in their own right
Inpatient treatment relieves symptoms more rapidly than day hospital treatment but may lead to increased burden on carers one year later
The Manchester Royal Infirmary's psychiatric day hospital has treated acutely ill patients as an alternative to inpatient treatment for over a decade.1 2 3 Roughly 40% of potential admissions can be treated in this way. Other workers have reported similar findings.4 5 6 Day patient and inpatient treatment lead to similar social and clinical outcomes.
This paper compares the costs of day patient and inpatient treatment. Other studies have either included only neurotic and personality disorders,7 or shown no difference because inpatient beds were held open for day patients,8 9 or shown day hospitals to be an unwarranted, expensive alternative to outpatient treatment for milder illness.10 11 In our previous study some day patients were so ill that they were transferred to the inpatient unit.2 Such “failures” of day hospital treatment were included in this study in the intention to treat analysis of costs. These costs were assessed over one year because we find that inpatients with a brief admission have a higher rate of readmission over the subsequent year than day patients.
Day hospital treatment for acutely ill patients may place an excessive burden on the carers. Hence costs to these family members were included both as monetary variables (for example, loss of income and travel costs) and non-monetary variables (for example, stress symptoms in the carer).
Patients and methods
The study was conducted over three years. Randomisation was as in our previous study, by randomly assorted cards in sealed envelopes opened by an independent administrator.2 After randomisation clinicians managed the patients as usual, determining discharge dates and readmissions independent of the researchers.
All patients aged 18-65 years presenting to the service for admission were considered for the study. Exclusions were compulsory admissions or patients too ill for day treatment, patients discharged in under five days, admissions solely for detoxification of drugs and alcohol, and patients with a diagnosis of organic brain disease, personality disorder, or mania. Psychiatric diagnosis was assessed at admission (by PM and BW) with the present state examination,12 and the severity of psychiatric symptoms was measured with the comprehensive psychopathological rating scale13 at admission and after two weeks and one, two, three, six, and 12 months after admission. Demographic data, previous psychiatric admissions, and mode of referral were recorded at the first interview.
The social behaviour assessment schedule14 was administered by the social research worker (SL) to an informant at admission and at one, two, three, six, and 12 months. This instrument assesses the patient's social performance, abnormal behaviours, and burden on relatives. Disturbed behaviour after admission was assessed with the modified social behaviour schedule,15 completed by the patient's key nurse. Distress in carers was also assessed by the informant completing the general health questionnaire.16
Use of resources
Direct costs to Central Manchester Health Care Trust–The duration of the first and any subsequent inpatient or day hospital admissions, number and length of interviews with medical staff and community psychiatric nurses, and investigations at the hospital were costed at local rates. Costs of the mental health service in central Manchester, including “hotel” and staffing costs, were identified down to unit of service. These were then linked to appropriate clinical activity to give the unit costs for the study. The cost of drugs was based on British National Formulary figures, adjusted to take account of Central Manchester Health Care Trust overheads. Day hospital admissions were costed (staffing and overheads) by using the number of days that patients actually attended the day hospital.2
Direct costs to other agencies–Information on direct costs to other agencies was collected from the records of general practitioners and social workers. Costs of visits were based on national unit costs of community care,17 as detailed local cost information was not available for these services.
Indirect costs to patients and carers–Indirect costs to patients and carers were estimated from interviews with the main carer. The carer recalled travelling costs (number of journeys, mode of transport, and return mileage) related to the patient's illness, and an estimate was also made of any increased household expenditure and reduction in the patient's or carer's income (that is, as a result of time lost from work due to the illness). The time that the patient and carer spent travelling and in outpatient and other departments during appointments with medical, nursing, and social work staff was estimated but could not be costed. The average time per day that the carer spent in direct care of the patient (while ill) was also estimated and expressed as hours per day.
All the above monetary costs were adjusted to 1994-5 prices by using the relevant price index.
During the second half of the study additional support was provided to help the treatment of acutely ill day patients. A community psychiatric nurse worked specifically to aid their attendance at the day hospital and a nurse was available on call during evenings and weekends. This increased contacts with the community psychiatric nurses but did not significantly affect overall costs.
Ethics–The study was approved by central Manchester's ethical committee.
Data analysis and statistics
An intention to treat analysis compared inpatient and day hospital groups. Clinical and social outcome and costs were compared over 12 months after the first admission. The comprehensive psychopathological rating scale measures psychiatric symptoms, and the social behaviour assessment schedule measures social role performance, abnormal behaviours observed by the informant, and burden on carer. The general health questionnaire completed by the informant measures distress experienced by the carer. These separate outcome measures reflect separate dimensions.18
The statistical package for the social sciences was used. Baseline measures were compared by Χ2 and t tests as appropriate. Scores at all follow up assessments were compared by analysis of covariance to control for any baseline differences. In order to check differences between the groups the area above and below the curve (see fig 1) was calculated for each patient and compared by t test.
Costs had a highly positively skewed distribution, so the Mann-Whitney U test was used for analysis. Table 3 gives the resulting P values and also shows the median figures (and 95% confidence intervals) for each cost category for inpatients and day patients and the difference. The data were also analysed by log transformation, the overall pattern of results being identical with those shown (data available on request).
Ninety three inpatients and 94 day patients were randomised. Eight were excluded because of diagnosis or early discharge, leaving 89 inpatients and 90 day patients. Five randomised inpatients were transferred to the day hospital because of lack of beds, and 11 randomised day patients were transferred to the inpatient unit because they were too ill for the day hospital. Only 103 patients (52 inpatients, 51 day patients) had a resident carer who was available for repeated interviews. Patients with and without a carer had similar comprehensive psychopathological rating scale and present state examination scores (severity of psychiatric symptoms).
At admission inpatients and day patients showed no significant differences in sex (49 (55%) and 53 (59%) male, respectively); mean age (37 (SD 12) and 39 (14) years); ethnic origin (69 (78%) and 78 (87%) white); employment status (38 (43%) and 39 (43%) unemployed); marital status (26 (29%) and 36 (40%) married or cohabiting); diagnostic categories (schizophrenia 41 (46%) and 36 (40%), depression 28 (32%) and 32 (36%), neurosis 20 (23%) and 22 (25%); Χ2 = 0.68, df = 2, P = 0.71); and severity of illness (mean comprehensive psychopathological rating scale scores 23.02 (SD 8.58) and 25.3 (11.72); mean disturbed behaviour social behaviour assessment schedule scores 11.64 (SD 7.13) and 12.66 (7.30)).
Clinical and social outcome
Scores for psychiatric symptoms, social behaviour, and role performance were all significantly (P<0.001) reduced six and 12 months after admission and there was no significant difference between day patients and inpatients at these times (table 1; fig 1). At two and four weeks inpatients showed fewer psychiatric symptoms and fewer abnormal behaviours, indicating more rapid recovery than day patients. Social behaviour assessment schedule burden scores showed no difference in the early months of treatment, but day hospital patients were less of a burden to carers at one year (table 2). General health questionnaire scores of carers were not significantly different at any time (table 2). Calculating the area above and below the curve for each patient as a summary statistic confirmed that there was no significant difference on any clinical or social outcome measure (data available on request).
Resource use and costs
Direct costs–Randomised inpatients accumulated a mean of 62 inpatient days and seven day hospital days over the 12 months. For day patients the figures were 32 day hospital days and 21 inpatient days. Table 3 gives the costs. The duration of interviews with medical staff (11 hours v 10 hours; P = 0.44) was similar but day patients spent more time with community psychiatric nurses (P < 0.05). Costs of hospital investigations and drugs were similar. The median overall difference in costs to the Central Manchester Health Care Trust was £1923 (95% confidence interval £750 to £3174) less for day patients (table 3). Direct costs to other agencies showed no significant differences (table 3). Loss of patients' income through illness absences from work or unemployment was similar. Not surprisingly, patients' travel costs were greater for the day hospital group (P < 0.05). Total costs given monetary value showed day hospital treatment to be significantly cheaper (median difference £2165 (£737 to £3593); P = 0.001).
Costs to carers–There was a significantly greater loss of income among the carers of day hospital patients, primarily as a result of two carers becoming unemployed. This loss of employment could not be attributed directly to the patient's illness or its treatment, but the costs were included in the cost analysis. Travel costs were significantly greater for the carers of inpatients (table 3). When all direct and indirect costs to patients and carers were considered the median costs for day hospital treatment were £1994 less than for inpatient treatment (95% confidence interval £600 to £3543).
Costs not given a monetary value
Time spent travelling was significantly greater for day patients, but carers' travelling time and outpatient and consultation times with the general practitioner were all significantly greater for inpatients (table 3). A considerable amount of time was spent by carers on the care of their mentally ill relatives but there was no significant difference between inpatient and day hospital groups.
This analysis does not include accommodation costs. Sixty seven (35%) patients were living with a spouse or cohabitee, 51 (27%) were living with a family member, 42 (22%) were in lodgings or a hostel, and 29 (15%) lived alone. There was no significant difference between the inpatient and day hospital groups in this respect.
So far as we know this is the first cost effectiveness study of day hospital treatment as an alternative to inpatient psychiatric treatment. There were two principal methodological problems. The first was the attrition rate of the sample over 12 months; this was similar for day patients and inpatients and comparable with other studies in an inner city area.2 19 20 Secondly, only 103 (55%) patients had a resident carer to interview; this reflects the small social network of such patients and compares favourably with the 28-32% in other studies.19 21 22
This study shows that day hospital treatment is cheaper than inpatient treatment. This remained true despite increased costs associated with a few day patients being transferred to inpatient care, additional travelling costs and community psychiatric nurse time for day patients, and a few carers of day patients who incurred considerable loss of income (though in this series it was not clear that this was related to the psychiatric treatment). Carers of day patients recorded lower burden scores at 12 months, possibly because day hospital care leads to better coping with the illness by patients and relatives.23 Early in treatment, however, most carers had high general health questionnaire scores–64% (66/103) were above the threshold for probable “cases” of psychiatric disorder in their own right. The more rapid improvement of inpatients is a new finding. This may reflect their removal from a stressful environment, the increased support from staff, or the increased use of drug treatment.24
This study included patients with acute symptoms presenting to an inner city psychiatric service. Most presented as emergencies and nearly half (77; 43%) had schizophrenia. These patients represented only 30-40% of potential admissions2 and were therefore not as ill as patients included in other intensive community treatment programmes in Britain.19 25 They were more comparable to patients in early intervention studies22 26 27 and cannot be generalised to all acutely ill patients.
These patients were more ill than those in our previous study2 (data not shown), reflecting increased confidence of our day hospital staff to treat acute illness, the proximity of the inpatient unit for rapid transfer when necessary, and considerable input from medical and community psychiatric nursing staff. These factors probably helped to prevent “burn out” of staff, which is a feature of some alternatives to hospitalisation for psychiatric patients.28 29 30 31
This study confirms that day hospital treatment is feasible and cheaper to the health service than inpatient treatment for some acute psychiatric illnesses. Relatives of patients with acute mental illness warrant further support to reduce their distress and enhance their caring role.
We thank Dr Max Marshall for helpful comments on an earlier draft of this paper and Mrs Joan Bond for invaluable secretarial work. The results of log transformation and calculation of the area above and below the curve for each patient as a summary statistic for differences in clinical or social outcome measures may be obtained by writing direct to FC.
This study was funded by the Department of Health, the North Western Regional Health Authority, and the Mental Health Foundation.
Conflict of interest: None.