Age and Ageing 2000; 29: 137 142 2000, British Geriatrics Society How great a burden does early discharge to hospital-at-home impose on carers? A randomized controlled trial DAVID GUNNELL, JOANNA COAST, SUZANNE H. RICHARDS, TIM J. PETERS, JOHN C. POUNSFORD 1, MARY-ANNE DARLOW 2 Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, UK 1 Care of the Elderly Department, Frenchay Hospital, Frenchay Park Road, Bristol BS16 1LE, UK 2 Hospital at Home Team, Community Directorate, Frenchay Hospital, Frenchay Park Road, Bristol BS16 1LE, UK Address correspondence to: D. Gunnell. Fax: (+44) 117 9287236. Email: D.J.Gunnell@Bristol.ac.uk Abstract Objective: to assess the effects of an early discharge hospital-at-home scheme on self-reported carer strain and quality of life. Design: a randomized controlled trial Setting: Bristol, UK. Subjects: 133 carers of patients receiving either early discharge from hospital to hospital-at-home (n = 93) care or usual hospital care and discharge (n = 40). Outcome measures: modified 12-item Carer Strain Index, COOP-WONCA charts and EuroQol EQ-5D at 4 weeks and 3 months post-randomization. Results: the mean age of carers was 65 years; 56% were women. There were no marked differences between the groups in any of the outcomes used at either 4-week or 3-month follow-up. Conclusion: there was no evidence of increased self-reported burden imposed on carers of patients discharged early from hospital. Decisions on the implementation of hospital-at-home schemes should be influenced by considerations of cost and effectiveness rather than effects on carers. The effects on carers may, however, differ for other forms of home-based care. Keywords: carers, hospital-at-home, older people, quality of life, randomized controlled trial Introduction Several forms of early discharge and admission prevention schemes have been introduced to try and relieve the increasing pressures on hospital beds [1, 2]. With a few exceptions [3, 4], published evaluations of such services have focused on patient outcomes [5] and health service costs [6, 7], with little attention given to the effects on carers [2]. Carers often play an important role in supporting older patients in hospital and at home and should be included in any assessment of new types of service provision [8]. We have compared self-reported levels of strain and quality of life in carers of patients treated in an earlydischarge hospital-at-home scheme with those for carers of patients discharged from hospital at the usual time, assessed within the context of a randomized controlled trial. Methods Between July 1994 and October 1995, 241 consenting patients suitable for early discharge were randomized on a 2:1 basis to hospital-at-home care or usual in-hospital treatment [5]. The sample size for the trial was based on detecting differences in numbers of days from randomization to discharge from care. The mean age of patients was 77 years (range 46 94), and only 7% were under 65. The proportion of younger patients was similar in the two arms of the trial. The main diagnoses were fractures (52%), elective orthopaedic surgery (16%) and stroke (10%). With ethics committee approval, patients with dementia (n = 11) were included in the trial as long as their carer gave consent on their behalf. Median length of stay in hospital postrandomization was 2 days for hospital-at-home and 9 days for hospital patients. 137
D. Gunnell et al. Before randomization, patients were asked to identify their main helper, defined as someone (nextof-kin or a close friend) who would be in close contact with them and provide additional support (emotional or practical) should the need arise. The main carer was then interviewed to obtain information about their age, health, employment and social circumstances. They were also given two health status measures to complete (COOP-WONCA [9] and EuroQol EQ-5D [10, 11]). The EuroQol EQ-5D consists of a series of questions and a thermometer, a visual analogue health assessment scale with values from 0 (worst imaginable health state) to 100 (best imaginable health state). Carers were sent postal questionnaires 4 weeks and 3 months after randomization. Reminders were sent to non-responders after 1 week. These questionnaires included a shortened (12-item) version of the Carer Strain Index (CSI), a seven-item patient satisfaction questionnaire as well as further COOP-WONCA and EuroQol EQ-5D charts. The CSI was developed in the United States [12] and elicits a series of yes/no responses on many issues, including an item covering financial strain to the carer. As a full economic evaluation [6] was being carried out concurrently, this item was excluded from this study. Results for the CSI are presented as proportions responding yes to each item, the proportion of yes responses across all items (summary measure) and differences in proportions and their 95% confidence intervals. Statistical significance for individual items was determined either by x 2 or Fisher s exact tests. Differences between the two groups in the summary CSI were assessed using the Mann Whitney test. Most responses to the patient satisfaction survey were clustered across one or two out of the four or five response categories. For summary statistics, responses have been dichotomized. For testing null hypotheses of differences between groups, more sensitive tests (Mann Whitney U and Fisher s exact test) were used based upon the data in their original form. The COOP- WONCA and EuroQol EQ-5D (score and thermometer) were analysed using multivariate repeated measures analysis of variance (MANOVA) methods for the baseline, 4-week and 3-month measurements. Confidence intervals for differences through time were obtained from separate (univariate) analyses of covariance for the two follow-up assessments adjusted for baseline. All analyses were carried out on an intention-to-treat basis. Results Altogether, 241 patients took part in the trial. Main carers were identified (before randomization) by 133 patients (55%): 93 (58%) in the hospital-at-home and 40 (49%) in the hospital arm of the trial. The mean age of carers was 65 (range 24 89 years), 75 (56.4%) were women and 98 (73.7%) lived in the same household as the patient. Thirty-nine percent of carers reported a long-standing illness or disability that had troubled them for a period before the patients hospitalization. Table 1 shows the characteristics of carers and those they cared for in the two arms of the trial. Carers of subjects randomized to hospital-at-home were more often women and were slightly younger than carers of Table 1. Characteristics of the carers and those they cared for in the two arms of the trial a Hospital Hospital-at-home Characteristic (n = 40) (n = 93)... Carers No. (%) women 14 (35%) 61 (66%) Age Median (interquartile range) 72 (59 75) 64 (54 76) Mean 66 65 No. (%) retired 19 (48%) 48 (52%) Patients No. (%) women 27 (68%) 53 (57%) Age Median (interquartile range) 75 (72 82) 78 (71 85) Mean 76 77 Median Barthel score (interquartile range) 16 (14 17) 16 (12 17) Median Folstein score (interquartile range) 26 (25 28) 26 (21 27) a Full information on some variables was missing. 138
Table 2. Burden imposed on carers at 4 weeks and 3 months as assessed by the Carer Strain Index (CSI): all proportions are for those answering yes to the question, where yes implies some burden has been imposed on the main carer (positive differences favour hospital-at-home, negative differences favour hospital) 139 4 weeks 3 months...... % answering yes % answering yes...... CSI item Hospital Home Difference 95% CI P-value Hospital Home Difference 95% CI P-value... Sleep was disturbed 34.4% 43.8% 9.4% 29.1% to +10.3% 0.36 14.3% 27.4% 13.1% 29.6% to +3.4% 0.17 It was inconvenient 18.8% 26.3% 7.6% 24.3% to +9.2% 0.40 22.2% 26.4% 4.2% 22.9% to +14.5% 0.67 It was a physical strain 21.2% 26.0% 4.8% 21.8% to +12.3% 0.38 17.9% 23.3% 5.4% 22.6% to +11.8% 0.55 It was confining 53.1% 50.0% +3.1% 17.4% to +23.7% 0.77 38.5% 43.2% 4.8% 26.6% to +17.1% 0.67 There have been family adjustments 48.5% 36.3% +12.2% 7.8% to +32.3% 0.23 32.1% 36.5% 4.3% 24.8% to +16.1% 0.68 There have been changes in personal plans 36.4% 28.8% +7.6% 11.6% to +26.8% 0.43 35.7% 22.2% +13.5% 6.7% to +33.7% 0.17 There have been other demands on my time 21.9% 29.1% 7.2% 24.7% to +10.2% 0.44 7.1% 23.6% 16.5% 30.2% to 2.8% 0.06 There have been emotional adjustments 39.4% 18.5% +20.9% +2.2% to +39.6% 0.02 26.9% 19.2% +7.7% 11.5% to +27.0% 0.41 Some behaviour is upsetting 27.3% 30.0% 2.7% 20.9% to +15.5% 0.77 23.1% 30.1% 7.1% 26.4% to +12.3% 0.49 It is upsetting to find the patient has changed so much 30.3% 30.0% +0.3% 18.3% to +18.9% 0.98 25.0% 33.8% 8.8% 28.1% to +10.5% 0.39 There have been work adjustments 24.2% 9.0% +15.3% 0.7% to +31.2% 0.06 a 10.7% 10.7% +0.04% 13.4% to +13.5% 1.0 a Feeling completely overwhelmed 27.3% 29.1% 1.8% 20.0% to +16.4% 0.84 14.3% 25.3% 11.0% 27.3% to +5.2% 0.23 Median proportion across all categories 25.0% 33.3% 0% b 8.3% to +12.9% 0.69 b 17.4% 16.7% 0% 15.2% to +8.3% 0.61 b The number of responders to individual items of the CSI ranged from 76 to 82 for hospital-at-home carers and from 32 to 33 for hospital carers at 4 weeks. The ranges at 3 months were 72 75 and 26 28 respectively. CI, confidence interval. a Fisher s exact test used as 25% of cells had expected counts <5. b Mann Whitney U test used on proportions of yes responses made by each carer; the difference quoted is the median of the differences [14]. Burden of early discharge to hospital-at-home on carers
D. Gunnell et al. Table 3. Differences in CO-OP WONCA and EuroQol EQ-5D scores for carers in the two arms of the trial 4 weeks 3 months...... Difference 95% CI a Difference 95% CI a P-value b... COOP chart c Physical fitness +0.103 0.498 +0.422 0.446 0.31 Feelings +0.063 0.450 +0.131 0.520 0.92 Daily activities +0.102 0.507 +0.338 0.489 0.61 Social activities +0.293 0.534 +0.484 0.569 0.20 Change in health +0.245 0.378 0.040 0.277 0.84 Overall health +0.415 0.339 +0.077 0.394 0.22 Pain +0.285 0.522 +0.687 0.476 0.022 EQ-5D d Score 0.107 0.099 0.089 0.096 0.15 Thermometer 0.242 5.74 4.21 7.65 0.82 a 95% confidence interval from univariate analyses of covariance adjusting for baseline. b From multivariate repeated-measures ANOVA. c Positive differences favour carers of hospital-at-home patients. d Negative differences favour carers of hospital-at-home patients. hospital randomized patients. There were no important differences in the characteristics of those cared for in the two arms of the trial. Carers of 82 (88%) hospital-at-home and 33 (83%) hospital patients responded to the 4-week questionnaire, although some did not complete all items. Response rates to the 3-month questionnaire were 79% and 66% respectively. Differences between proportions of carers answering positively to the dimensions of the CSI at 4 weeks and 3 months are shown in Table 2. No consistent differences are apparent. Similarly, the COOP-WONCA and EQ-5D generally showed no significant differences between the two groups at either follow-up (Table 3), although the directions of the differences tended to favour hospital-at-home (that is positive differences in the COOP-WONCA and negative in the EuroQol EQ-5D). The only significant difference was in the pain item of the COOP-WONCA chart, the direction of which favoured carers of hospital-at-home patients. In addition, although the overall time pattern is not significant (P = 0.15), the confidence interval for EQ-5D score at 4 weeks does not cross 0, and this too suggests a difference in favour of hospital-at-home. Analysis of responses to the carer satisfaction questionnaire (Table 4) showed significantly greater carer satisfaction in the hospital-at-home arm of the trial in relation to their discussions with staff and information they received from staff regarding patient treatment. Despite randomization, one-third of the carers of hospital patients were female compared with two-thirds of the carers of hospital-at-home patients. To assess whether these differences in gender distribution influenced our main findings, all the analyses were repeated for male and female main carers separately (results not shown). There was no consistent evidence that the effects of hospital-at-home on carers differed depending on the carer s gender. Discussion There was no evidence of an increase in self-report burden imposed upon carers of patients discharged home early under the care of a hospital-at-home scheme. This was despite the fact that these carers looked after patients for a median of 7 additional days whilst they might otherwise have remained in hospital. The one statistically significant difference (emotional adjustments at 4 weeks favouring carers of hospital-athome patients) should be interpreted with caution in view of the number of hypotheses tested here. These findings accord with those of another recently published hospital-at-home trial [4], although in that study most patients were discharged home early after elective surgery, rather than after acute admission. We had anticipated that carers of patients who were discharged home early might have experienced increased care-related strain and reduction in quality of life. Such possible effects were apparent (although not quite reaching conventional levels of statistical significance) in a recent trial comparing community therapy with conventional rehabilitation for stroke patients [3]. There are several explanations for the null 140
Burden of early discharge to hospital-at-home on carers Table 4. Carer satisfaction with services in the 4 weeks following randomization Satisfaction (%)... Hospital Hospital-at-home Difference 95% CI P-value... Content with services received by patient 67.7 56.8 +11.0% 8.7% to +30.6% 0.27 a (all of the time) Involved in decisions about patient s treatment 50.0 61.3 11.3% 31.6 to +9.1% 0.31 a (all/most/some of the time) Helper s discussions with staff (excellent) 29.2 51.9 22.7% 44.0 to 1.5% 0.007 a Information to helper about patient s illness 53.6 69.3 15.8% 37.0% to +5.5% 0.38 b Information to helper about patient s treatment 48.3 76.0 27.7% 48.3% to 7.1% 0.010 b Practical help for helper from family and friends 80.7 70.7 +10.0% 7.3% to +27.3% 0.34 b Emotional help for helper from family and friends 86.7 80.3 +6.4% 8.7% to +21.5% 0.57 b CI, confidence interval. a Mann Whitney U-test. b Exact test. effect found in the analyses presented here and those of Shepperd et al. [4]. First, carers may have received more information about the patient s treatment and care and felt more involved in decision-making through their regular contact with hospital-at-home staff. Such an effect is supported by a carer satisfaction survey carried out in this trial (Table 4). The better quality of discussions with staff and information reported by hospital-at-home carers are not, however, necessarily exclusive to hospital-at-home care; they may equally indicate that improvements in these areas might be made in hospital. Secondly, the difficulties with visiting the patient in hospital will have been avoided. Lastly, the patient s own satisfaction with home care may have influenced their carer s perception of any additional strain caused. The results should, however, be viewed with caution as the sample size for this study was not based on assessing burden on carers. Because of the surprisingly low proportion of subjects identifying main carers, the study is underpowered to detect small but possibly important differences between the groups. There is no marked impact on carers of early discharge hospital-at-home schemes in terms of the measurements that we used. On the basis of this trial, decisions on the implementation of hospital-at-home schemes should be influenced by considerations of their cost and effectiveness rather than their potential adverse effects on carers [6, 13]. Evidence in these areas remains unclear, although this situation is likely to change with the publication of further evaluations [1]. More research in this area is required. The effects on carers may differ for other forms of home-based care [3]. Factors influencing carer strain include levels of patient dependency as well as the staffing levels of hospital-at-home teams and their workload; all these issues require consideration when considering new service development. Key points Hospital-at-home treatments are increasingly being considered and implemented, yet the effect of such schemes on carers has received little attention. The carers of a group of mainly older people who participated in a randomized trial comparing early hospital discharge with usual hospital care did not report increased strain or reduction in their quality of life at 4-week or 3-month follow-up. Carers were generally satisfied with most aspects of both hospital-at-home and hospital treatment, any differences tending to favour hospital-at-home care. These results suggest that decisions on the implementation of hospital-at-home schemes should be based more on cost and effectiveness than their potential effects on carers. Our findings should not necessarily be extrapolated to hospital-at-home schemes for other patient groups. Acknowledgements This research project was funded by a grant from the South and West NHS Research and Development 141
D. Gunnell et al. Directorate; funding for the hospital-at-home pilot scheme was provided by Avon Health. Joy Lane set up the hospital-at-home team. The project could not have been carried out without the interest and cooperation of the hospital-at-home team, hospital ward staff, hospital consultants and general practitioners, and of course the patients and main carers who took part in the study. Thanks to Maggie Somerset, Maggie Evans and Sara Brookes for data collection. Ethics approval for this study was obtained from local research ethics committees. References 1. Iliffe S. Hospital at home: from red to amber? Br Med J 1998; 316: 1761 2. 2. Marks L. Home and Hospital Care: redrawing the boundaries. Research report 9. London: Kings Fund Institute, 1991. 3. Rudd AG, Wolfe CDA, Tilling K, Beech R. Randomised controlled trial to evaluate an early discharge scheme for patients with stroke. Br Med J 1997; 315: 1039 44. 4. Shepperd S, Harwood D, Jenkinson C, Gray A, Vessey M, Morgan P. A randomised controlled trial comparing hospital at home with inpatient hospital care (1): three month follow-up of health outcomes. Br Med J 1998; 316: 1786 91. 5. Richards SH, Coast J, Gunnell DJ, Peters TJ, Pounsford J, Darlow M- A. A randomised controlled trial comparing the effectiveness and acceptability of an early discharge hospital at home scheme with acute hospital care. Br Med J 1998; 316: 1796 801. 6. Coast J, Richards SH, Peters TJ, Gunnell DJ, Pounsford J, Darlow M- A. Hospital-at-home or acute hospital care: a cost minimisation analysis. Br Med J 1998; 316: 1802 6. 7. Hollingsworth W, Todd C, Parker M, Roberts JA, Williams R. Cost analysis of early discharge after hip fracture. Br Med J 1993; 307: 903 6. 8. Department of Health. Carer Support in the Community. London: HMSO, 1991. 9. Scholten JHG, Van Weel C. Functional status assessment in family practice. In: The Dartmouth COOP Functional Health Assessment Charts. Den Hag, Netherlands: WONCA MediTekst, 1992. 10. Brooks R, EuroQol Group. EuroQol: the current state of play. Health Pol 1996; 37: 53 72. 11. Coast J, Peters TJ, Richards SH, Gunnell DJ. Use of the EuroQol among frail elderly patients. Qual Life Res 1998: 7: 1 10. 12. Robinson BC. Validation of a carer strain index. J Gerontol 1983; 38: 344 8. 13. Shepperd S, Harwood D, Jenkinson C, Gray A, Vessey M, Morgan P. A randomised controlled trial comparing hospital at home with inpatient hospital care (2): a cost minimisation analysis. Br Med J 1998; 316: 1791 6. 14. Gardner MJ, Altman DG. Statistics with Confidence. London: British Medical Journal Publications, 1990. Received 20 April 1998; accepted in revised form13 November 1998 142