METHODS

Study sample

The study was conducted in Rome and was approved by the ethics committee of La Sapienza University. Patients were recruited from the multiple sclerosis centre of the neurological department of La Sapienza University, or from other institutional centres for multiple sclerosis in Rome. To be eligible, all patients screened for study participation had to be affected by clinically definite multiple sclerosis¹⁷ and live in the Rome service area. Eligible patients were assessed for suitability by the coordinator of the home based care team (LP). Written informed consent was obtained for entry into the trial from the patients and, when appropriate, from their carers.

Randomisation

Of the 297 consecutive patients screened, 201 were willing to participate in the study. Ninety six patients declined to participate. The majority of refusers (n = 62) were young, fully ambulant, and employed. They were uncertain that home based management could be adopted in their situations. A minority of refusers (n = 34) did not consent to take part to the study because they were referred to the multiple sclerosis centres only for a second opinion but were regularly followed by their own neurologists.

The patients were randomised to home (intervention) or hospital care (control) in a ratio of 2:1, in order to ensure that the home care population was sufficiently represented. Randomisation (in blocks of six) was stratified by age and expanded disability status score (EDSS),¹⁸ because both health outcomes and costs differ according to disease disability and age^5,19 and a valid conclusion can be drawn only by comparing similar patients. Randomisation was done using a computer generated algorithm.

Baseline interview

Baseline data for both groups (that is, demographic information, neurological and psychological assessment, cognitive and functional abilities, and health related quality of life measures) were collected after randomisation, using standardised instruments. Neurological impairment and cognitive abilities were evaluated by the EDSS and the mini-mental state examination (MMSE).²⁰ Disability and fatigue were measured using the functional independence measure (FIM)²¹ and the fatigue severity scale (FSS).²² Mood measures included the state trait anger expression inventory (STAXI),²³ the state trait anxiety inventory (STAI),²⁴ and the clinical depression questionnaire (CDQ).²⁵ Quality of life was measured by means of the 36 item short form health survey questionnaire (SF-36).^26,27

Treatment schedules

Based on individual needs, patients randomised to home based care were followed through home visits and telephone follow up. A dedicated phone number was available five days a week from 9 am to 5 pm, where an operator addressed questions or concerns from patients and caregivers and contacted the different specialists according to the specific requirement. After 5 pm and during the weekend an answering machine was on, and recorded messages were regularly listened to by the operator. The home based care multidisciplinary team included two neurologists, a urologist, a rehabilitation physician, a psychologist, a physical therapist, a nurse, a social worker, and a coordinator. These were provided with cell phones and could easily be reached for advice (telephone intervention) or direct face to face intervention when required.

The multidisciplinary team collaborated with the patient, physician, and caregiver in designing individualised clinical care and in coordinating home services as appropriate for the individual patient. The type of care was more than is normally available in the community through National Health Service care. It consisted of observation, administration of intravenous drugs, nursing care, rehabilitation of the patients in their home, patient and caregiver education, psychological support, and the services of the social secretariat.

Patients randomised to routine hospital care were followed as usual in their multiple sclerosis referral centres. A brief monitoring phone call once a month was used to obtain information about the patient's medical visits and hospital admissions in the previous month.

At the end of a one year follow up period, all patients underwent the same comprehensive interview as at baseline.

Statistical analysis

Sample size considerations

With a two tailed 5% significance level, a total sample size of 201 (with a 2:1 randomisation ratio and taking into account a drop out rate of 10%) would yield about 90% power to detect a standardised difference of 0.5 standard deviations on the SF-36 general health perception domain. The figure was obtained assuming a standard deviation of 10 for the pre-post differences in SF-36 general health perception domain and a five point difference between the two groups, which is the smallest change in score considered clinically relevant.²⁷

Statistical calculations

Baseline data for intervention and control groups were compared using a χ² test for categorical variables, a t test for normally distributed continuous variables, and the Wilcoxon rank-sum test for abnormally distributed variables.

For each scale of the SF-36 and separately for the case and control group, we adopted a regression model in which score changes at the end of the follow up period were adjusted for the baseline values. Changes in scores were approximated by a normal distribution and so the primary analysis after adjustment for baseline assessment was by two sample t test.

Economic evaluation

We adopted a cost minimisation analysis to test the hypothesis that home based care and traditional hospital care delivered in a multiple sclerosis centre are associated with an equivalent consumption of resources. The economic analysis was conducted from a third party payer perspective; accordingly, only direct health care costs were considered.²⁸ Indirect costs (lack of productivity for the patient and for caregivers) and non-medical costs were not included. The resources considered were the cost of inpatient, outpatient, and home care services, as well as the cost of the home based coordination programme. We did not included costs of pharmaceuticals and aids for daily life activities.

The costs of inpatient care (ordinary, rehabilitation, day hospital, and diagnostic tests) were estimated using Italian third party payer reimbursement for diagnosis related groups (DRGs) as a proxy for the real cost. Outpatients and community resources were calculated through the outpatient prices developed by the National Health Service.²⁹ An additional fee for service reimbursement established by a national contract was used to cost domiciliary visits. Telephone intervention was estimated directly in the trial as half of the resources of an outpatient visit; this estimate was tested in the sensitivity analysis.

For the intervention group, we estimated the costs to cover the local management and administration of the home based management scheme (for the operator, coordinator, telephone, and overheads). The costs used in the analysis were calculated on the basis of 1999 Italian prices converted to euros. The one year time horizon did not require discounting in the analysis.²⁸

Sensitivity analysis was used to test the robustness of the economic results. We employed a multivariate sensitivity analysis, altering each variable simultaneously to take the most optimistic/pessimistic value in order to generate a best/worst case scenario.

RESULTS

Between January 1997 and January 1998, 201 consecutive patients were randomised (133 to the intervention group and 68 to the control group). Data on the follow up assessment at 12 months were available in 188 patients (123 in the intervention group and 65 in the control group) (fig 1).

Baseline characteristics

Table 1 summarises the baseline sociodemographic and clinical characteristic of the patients. Many of those in the study were unemployed or retired people and were in the progressive phase of the disease.

The two groups were similar in all baseline variables measured, including functional and cognitive abilities. However, baseline differences between the two groups emerged in four of the eight SF-36 domains (role, physical: p = 0.01; bodily pain: p = 0.01; general health: p = 0.05; role, emotional: p = 0.03), with the intervention group reporting a lower quality of life than the control group. To correct for these observed differences, all analyses of SF-36 were adjusted for the relevant baseline assessment.

Functional outcomes

Functional outcomes were calculated as the mean change of scores from baseline to the final follow up at 12 months. No significant differences between intervention and control groups were detected for outcome measures, including EDSS, FIM, MMSE, CDQ, FSS, STAI, and STAXI. There was a trend in favour of the intervention group for changes in depression as measured by the CDQ score. A decrease in CDQ score was seen in the intervention group (−7.8%) while it was slightly increased (+0.7%) in the control group (p = 0.11).

Figure 2 shows the percentage changes in the SF-36 scores at one year. In the intervention group we observed an improvement in eight SF-36 scales. In the control group, an increase in the score of four SF-36 scales was detected; however, the improvement was less consistent than in the intervention group. Table 2 shows the changes in SF-36 scales at one year, with the positive differences being in favour of the intervention group. The intervention group had a significant improvement in bodily pain (p = 0.0001), general health (p = 0.0001), social functioning (p = 0.001), and role, emotional (p = 0.0001) when compared with the control group. In addition, scores on the eight SF-36 dimensions were reduced to two summary scores—a physical component (PCS) and a mental component (MCS)—by means of component analyses.²⁷ There was a significant difference between the two arms of the trial in favour of the intervention group for both PCS (p = 0.0001) and MCS (p = 0.0001).

Resource use and costs

Table 3 shows the resource consumption and costs in each arm of the trial. Patients included in the home based care scheme had less inpatient stay than the control group (p = 0.001). The needs of patients receiving home based care increased substantially for problems requiring both medical care and nursing, social, and psychological support (p = 0.0002 and 0.0067, respectively). Nevertheless, home based care presented a saving of 822 euros per patient compared with the controls.

In the best case scenario (an increase of 10% in reimbursement for admissions, a decrease of 10% for home based management costs, and the upper limit of the 95% confidence interval (CI)), each home based care patient presented a saving of 2086 euros. In the worst case scenario (a decrease of 10% in reimbursement for admissions, an increase of 10% for home based management costs, and the lower limit of the 95% CI), there was an incremental cost of 234 euros per patient.

DISCUSSION

Home based medical care is a popular alternative to standard hospital care, but there is uncertainty about its cost-effectiveness. There are two basic assumptions implicit in establishing home based health care—that is preferred by patients, and that is economically advantageous.³⁰ However, comparisons of the two modes of care, both in terms of patient outcomes and in terms of the cost to the health service, often produce conflicting results.³¹ The types of patient admitted to home based care may influence its cost-effectiveness significantly: patients suffering from chronic conditions such as multiple sclerosis are likely to require long term provision of care, with fewer occurrences of “emergencies” and more requests for social support and help for caregivers.

This study showed that a comprehensive home based follow up intervention implemented by an interdisciplinary team and designed specifically for patients with multiple sclerosis improved some aspects of their quality of life without increasing the cost of care. Thus the intervention suggested great economic potential, with improved outcomes being achieved at similar cost mainly by reducing hospital admissions and length of stay.

The results of the trial depend on local service provision, which may have influenced recruitment to the study, as well as on the characteristics of patients recruited. In the first place, the multidisciplinary team was well trained in order to ensure that the members had adequate case management skills. Secondly, patients with a strong preference for a hospital centre may have declined to enter the study, and patients could be admitted to the scheme only if they agreed to be randomised. Although it was randomised, our study could not be performed within the rigid criteria of a clinical trial. Because of the nature of the intervention all the professionals concerned were aware of the assignment of patients to either group. Also, as a result of the need for informed consent, patients and physicians were aware of the ongoing project. By chance, there was an imbalance between the groups in some of the quality of life dimensions at baseline, but this was reflected in the fact that randomisation to intervention occurred before the baseline interview. Thus it was not possible to fully blind patients or interviewers to the intervention during the baseline interview. To minimise the absence of blinding in the follow up interview, the assessors had no access to initial scores.

The quality of standard care in the control group must be also considered in the interpretation of the results. It could be argued that the study compared a rapid, responsive, and flexible type of care (home based management) with less responsive standard hospital care, which involving prebooked clinic visits and hospital access only in cases of significant medical deterioration. However, our control group was regularly followed in multiple sclerosis centres, where no more than one or two weeks are generally required to obtain an outpatient appointment, and hospital admissions are rapid in cases of need. The difference between conventional and home based care seems more related to the characteristics of home based intervention, implemented by a multidisciplinary team, than to the speed and flexibility of the service.

The results of our study indicated that home based management did not influence the most common measures of neurological impairment/disability (EDSS, FIM) but led to an improvement of SF-36 dimensions related to daily living and basic social routines. On the other hand, changes in the quality of life measures that are highly correlated with EDSS (that is, physical functioning and physical role limitation)¹⁹ were also similar in the intervention group and the control group. Physical disability may not be the main determinant of overall health related quality of life in multiple sclerosis. While physicians are usually more concerned than patients about the physical manifestations of disease, the patients identified role limitations caused by emotional problems as the most important determinant of their overall quality of life.^32,33 It is worth noting that the greatest difference between patients receiving home based care and the control group was observed within this dimension.

We assessed quality of life by means of SF-36, which is generally considered to be the gold standard generic measure of health status and has been validated cross culturally in multiple sclerosis.³⁴ New scales have recently been developed for quality of life in multiple sclerosis by adding specific items to SF-36.^35,36 However, modifying existing measures by simply adding clinically chosen items may not be as useful as anticipated in improving the measurements properties of an instrument.³⁷

SF-36, reflecting the patients' perspective, provides a broad measure of the disease impact in multiple sclerosis because all eight quality of life dimensions are reduced compared with the general population.¹⁹ Problems in measuring quality of life in multiple sclerosis, however, may occur in longitudinal studies and are related to the poor responsiveness of SF-36 and to the “response shift.” Response shift refers to the fact that the physical disability of the patients can change over time, thus influencing assessment of the quality of life.³⁸ We stress, however, that in the present study no significant differences between intervention and control groups occurred for scales measuring neurological impairment or physical disability such as EDSS and FIM.

The poor responsiveness of the SF-36 is particularly marked when patient selection is limited to those with moderate to severe disability, and it is related to the large floor effect in some of the SF-36 dimensions.³⁹ We investigated a wide range of patients with multiple sclerosis including those with RR disease. Thus our sample is representative of the entire multiple sclerosis population and it is not restricted to moderately or severely disabled patients. Two of the four health dimensions (emotional role limitation and bodily pain) which were significantly improved by home based management are influenced by floor and ceiling effects.³⁹ These effects, however, did not occur in the other two SF-36 dimensions (general health and social functioning) or in MCS and PCS, which were also favourably influenced by home based care.

Our economic analysis suggests that care in multiple sclerosis can be provided in patients' homes using a model of home based management at the same or lower cost than an equivalent admission to hospital. At 12 months, the intervention generated a mean saving of 822 euros per patient. The higher costs in the control group were entirely related to the greater proportion of patients being admitted to hospital. A significant increase in both medical and non-medical resources was observed in the intervention group. This is likely to reflect the nature of our scheme based on a multidisciplinary approach and a comprehensive model of care.

Our study suggests a lower annual direct cost of multiple sclerosis (2265 euros in the control group) compared with other studies performed in European countries.^5,40 Our lower figures may reflect a different approach to multiple sclerosis in Italy, but also the exclusion of drugs and disability aids from the analysis. Further limits of the economic analysis could be the use of reimbursement values to approximate costs²⁴ and the use of third party payer perspective, not including the indirect costs. Nevertheless, despite the important role of the indirect costs in the multiple sclerosis burden,^3,4,40 there is still considerable variability in the measurements depending on the methods used to collect resource.⁴¹

Conclusions

This is the first controlled study investigating the effectiveness of home based management in patients with multiple sclerosis. Our results show that the scheme is effective at improving the quality of life in people with multiple sclerosis and may provide a viable alternative to a hospital centre. One of the best features of the home based care is that it allows family members to receive instruction and help from formal caregivers. Moreover, the family can remain involved and assist in the care of their loved ones.

Home based care is an appropriate model of care for severely disabled patients who are still living at home but usually spend long periods in hospital. Given the current attention to new models of patient care in multiple sclerosis, we consider the study findings especially important for decision makers in the definition of priorities for multiple sclerosis patients. Home based care could also be run as a complement to hospital care, and play a role in managing the demand for hospital admission. More research is required to define the exact size of home based schemes in multiple sclerosis and its relation to hospital and community services.