Article Text

Download PDFPDF

Evaluation of a new model of short-term palliative care for people severely affected with multiple sclerosis: a randomised fast-track trial to test timing of referral and how long the effect is maintained
  1. Irene J Higginson1,
  2. Massimo Costantini2,
  3. Eli Silber3,
  4. Rachel Burman4,
  5. Polly Edmonds4
  1. 1Department of Palliative Care, Policy and Rehabilitation, King's College London, Cicely Saunders Institute, London, UK
  2. 2National Cancer Research Institute, Genoa, Italy
  3. 3King's College Hospital NHS Foundation Trust, London, UK
  4. 4Cicely Saunders Institute, King's College Hospital NHS Foundation Trust, London, UK
  1. Correspondence to Irene J Higginson, Department of Palliative Care, Policy and Rehabilitation, King's College London, Cicely Saunders Institute, Bessemer Road, London SE5 9PJ, UK; irene.higginson{at}kcl.ac.uk

Abstract

Aims In this randomised fast-track phase II trial, the authors examined (1) whether the timing of referral to short-term palliative care (PC) affected selected outcomes, and (2) the potential staff-modifying effect of the short-term PC intervention (whether the effects were sustained over time after PC was withdrawn).

Methods PC comprised a multiprofessional PC team that provided, on average, three visits, with all care completed by 6 weeks. Recruitment commenced in August 2004 and continued for 1 year. Follow-up was performed for 6 months in both groups. Outcomes were a composite measure of five key symptoms (pain, nausea, vomiting, mouth problems and sleeping difficulty) using the Palliative care Outcome Scale–MS Symptom Scale, and care giver burden was measured using the Zarit (Care Giver) Burden Interview (ZBI).

Results 52 patients severely affected by multiple sclerosis were randomised to receive PC either immediately (fast-track group) or after 12 weeks (control group). Patients had a high level of disability (mean Expanded Disability Status Scale: 7.7; median: 8; SD: 1). Following PC, there was an improvement in Palliative care Outcome Scale–MS Symptom Scale score and ZBI score. A higher rate of improvement in ZBI score was seen in the fast-track group. After withdrawal of PC, effects were maintained at 12 weeks, but not at 24 weeks.

Conclusions Receiving PC earlier has a similar effect on reducing symptoms but greater effects on reducing care giver burden, compared to later referral. In this phase II trial, the authors lacked the power to detect small differences. The effect of PC is maintained for 6 weeks after withdrawal but then appears to wane.

Trial Registration Number National Institutes of Health, USA, http://www.Clinicaltrials.gov, NCT00364936.

  • Palliative care
  • randomised
  • multiple sclerosis
  • end of life care
  • wait-list trial
  • palliative care
  • epidemiology
  • oncology
  • adult palliative care
  • thoracic medicine
  • statistics and research methods
  • education and training (see medical education and training)
  • medical education and training
  • internal medicine
  • primary care
  • pain management
View Full Text

Statistics from Altmetric.com

Introduction

Patients with considerable disability as a result of multiple sclerosis (MS) (either primary progressive or secondary progressive) compose a group whose need for palliative care (PC) is well established.1–6 Although some patients with MS have little disability, 15% of patients present with a primary progressive course, and half of those with relapsing–remitting disease develop secondary progression after 10–15 years. These patients may have both marked physical and psychological impairments and may develop severe symptoms, including pain, spasticity, fatigue, depression, bedsores and incontinence, which are as prevalent as among cancer patients receiving PC.4 7 Patients may need acute hospital care because of relapses and progression or as a result of complications such as pressure sores and infections. Many are cared for at home or in nursing homes by care givers and family members. Most symptoms are well managed by MS teams, including nurses when available; however, in some patients, these may be difficult to control. The long trajectory of the illness suggests that a model of PC different from that of cancer, with input at the final stages of illness, is needed.

A report from the Royal College of Physicians UK in 2007 concluded that ‘referrals (to palliative care services) often are made late, when patients have multiple problems and significant distress. Clinicians must recognise that end-of-life care does not mean that someone is going to die imminently but that the approach to their care should change. A proactive approach should be adopted to ensure that patients and their families have timely and appropriate care. It is important to identify triggers to ensure that palliative care treatments or referrals are initiated early and appropriately and to use systems of communication that facilitate continuity of care between doctors working in shift patterns.’8 Patients with non-cancer conditions (such as chronic obstructive pulmonary disease), heart failure and neurological conditions (such as MS) experience symptoms for a long period of time; experience profound losses and have psychological, social and spiritual needs that necessitate support, often well before the period closely before death.3 5 9–12

The WHO and others have recommended that PC be made available earlier in the course of illness13 14 and be provided on the basis of patient and family need and complexity rather than diagnosis or prognosis. Such an approach would seem logical for many non-cancer conditions such as MS, where the trajectory of illness is more variable and prognosis is uncertain. However, this model has critics, who believe that PC should focus on the end of life and that early referral to PC may not be beneficial. PC services also fear being overwhelmed by patients with non-cancer conditions, who have a long and variable disease trajectory. Currently, PC is mainly offered to people with cancer or, occasionally, motor neuron disease. To test this model of care, we developed a short-term PC intervention for people severely affected by MS with, on average, three visits to the patient. We have already reported that, at 12 weeks, the intervention appeared to have benefits in terms of reduction in care giver burden, symptoms (pain, nausea, vomiting, mouth problems and sleeping difficulty) and costs compared to a waiting-list control group.15 16 In this analysis, we wished to determine (1) whether the timing of referral to the short-term PC team (3 months earlier) had an impact on longer-term outcomes, and (2) the potential staff-modifying effect of the short-term PC intervention (ie, whether the effects were sustained over time after the PC was withdrawn or whether the effects waned).

Methods

This is a secondary analysis of a randomised phase II controlled trial, whose main results and methodology have been previously published (see box 1).15–18

Box 1

Summary of methods and main results of the primary study

Method

  • This study used a randomised controlled fast-track phase II trial where 52 people severely affected by MS were referred for short-term PC either immediately or after 12 weeks.

  • Primary outcomes were measured at 12 weeks, with interim measurement performed at 6 weeks.

  • The intervention comprised up to three visits from the PC team.

  • Recruitment commenced in August 2004 and ended in August 2005. Patients were followed up for 6 months until after both groups had received the intervention.

  • Trial registration: National Institutes of Health, USA, clinical trials.gov, NCT00364936.

Results

  • The group that received PC at 12 weeks had:

    • improvement in five key symptoms (mean change: −1.0), compared to deterioration in the control group (mean change: 1.1; F=4.75, p=0.035);

    • improvement in care giver burden, as assessed with the ZBI, compared to deterioration in the control group (F=7.6, p=0.013);

    • lower costs (£1789; bootstrapped 95% CI −£5224 to £1902) compared to the control group.

Study design

Following service modelling, we developed and tested a new PC service for people severely affected by MS using a randomised fast-track controlled phase II trial.17 Patients were randomised to either fast-track or control using the minimisation method, following consent and baseline interview. Ethical approval of the study was granted by the King's College Hospital NHS Trust Local Research Ethics Committee (protocol number: 01-04-018). The study was performed in two periods, each lasting 12 weeks.

In the first period, patients randomised to the fast-track arm were referred to PC within 48 h, while patients randomised to the control arm were followed by the best standard care for 12 weeks. In the second period, after 12 weeks, patients randomised to the delayed-start arm were referred to PC. Patients referred to PC in both phases were discharged from PC about 6 weeks from the initial referral. The study design is described in figure 1.

Figure 1

Study design and period of PC in the fast-track group and the control (delayed-start) group. PC, palliative care.

Intervention and subjects

PC was provided by a multidisciplinary team comprising a specialist in palliative medicine, a PC nurse, a psychosocial worker (part time) and a coordinator (part time). Individuals from the team, selected according to patient needs, visited patients at home or wherever required; liaised with and advised other professionals, such as neurologists and community services, about their care; managed symptoms; provided psychosocial support and set plans in place for future care or established advance directives. Overall, patients received around three visits from the team.

Subjects were men and women in South East London who were severely affected by MS and referred mainly by MS specialists or MS nurses. Referral criteria were problems or unresolved issues with symptom control, psychosocial needs, end of life, advanced planning (directive and competency or consent), planning needs for nutrition or difficulty with nutrition and/or hydration problems or a complex combination of problems.15 17

Research interviews

Face-to-face interviews with patients were conducted by trained interviewers usually in the patients' homes following a standardised schedule. After the baseline interview, interviews were repeated for patients in both groups at 6 weeks (time 2), 12 weeks (time 3) and 24 weeks (final interview). In addition, at 18 weeks, the control group was interviewed (aiming to have an interview 6 weeks after receipt of PC). At the same time, care givers self-completed short questionnaires and either gave these back to the interviewer at the time of patient interview or returned them by freepost.

Measures

For these analyses, the following outcome measures were used:

  • The Palliative care Outcome Scale–MS Symptom Scale (POS-MS-5), which self-assesses potential MS symptoms4 using a scale from 0 (no problem) to 4 (overwhelming problem). For this analysis, we used the composite scores of five common and problematic symptoms (POS-MS-5): pain, nausea, vomiting, mouth problems and sleeping difficulty, all of which PC aimed to ameliorate (range: 0 (best)–20 (worst)).

  • The self-reported care giver burden, as measured by the Zarit (Care Giver) Burden Interview (ZBI)—12-item (range: 0–48 (highest burden)).19

Diagnostic, demographic, clinical and functional information was collected at interview. We used the United Kingdom Neurological Disability Scale, comprising 12 sections to assess disability, and the Expanded Disability Status Scale (EDSS), a single-item 10-point interviewer assessed scale, at the first interview.

Statistical analysis

We examined two hierarchic research questions:

  1. whether the timing of referral to short-term PC (3 months earlier) had an impact on selected outcomes;

  2. the potential staff-modifying effect of short-term PC on selected outcomes.

Figure 2 shows a schematic illustration of the potential results of the study. Figure 2A,B shows possible scenarios when there is staff-modifying effect and the benefit persists. Figure 2C,D shows possible scenarios when the benefit is not sustained.

Figure 2

Schematic representation of four potential outcome trajectories: (1) the effect of PC wanes after it is withdrawn (C, D), or the effect of PC is maintained once achieved (A, B); and (2) the effect of PC is equal regardless of whether it is given earlier or later (A, C), or the effect is greater when it is given earlier (B, D). PC, palliative care.

The two research questions were analysed as follows:

  1. For the first research question, we compared the estimated changes between T1 (baseline) and T3 (12 weeks) in the fast-track group with the estimated changes between T3 (12 weeks) and T5 (24 weeks) in the control group. The null hypothesis claimed that there was no difference between the two groups. This analysis determines whether the rate of change was influenced by the timing of referral of the intervention (early vs 3 months delayed).

  2. For the second research question, we performed two analyses with different end points:

    • a. In the first analysis, we compared the estimated changes in outcomes between T2 (4–6 weeks) and T3 (10–12 weeks) between the fast-track group and the control group. The null hypothesis claimed that there was no difference in the T2T3 estimated changes between the two groups. This analysis determines whether there was a difference in the rates of change of outcomes between the two groups in the 6 weeks following the removal of the direct effect of PC on the fast-track group (T1T2). A staff-modifying effect of PC is expected to slow the rate of worsening of outcomes in the fast-track group as compared to the control group, including the time when the direct effect of PC had ended.

    • b. In the second analysis, we compared the estimated changes in outcomes between T1 (baseline) and T5 (22–24 weeks) between the fast-track group and the control group. The null hypothesis claimed that there was no difference in the T1T5 estimated changes between the two groups. This analysis determines whether the improvement in outcomes observed at T3 was still present at the end of the study (T5) when all groups received the experimental intervention. If there is a staff-modifying effect of PC, the early benefits observed in the fast-track group are expected to persist until the end of the study.

Statistical analyses were performed using intention-to-treat analysis, including all subjects who underwent evaluations at baseline (T1), 4–6 weeks after baseline (T2), 10–12 weeks after baseline (T3), 16–18 weeks (T4) and 22–24 weeks (T5). As previously reported, there was little missing data in the study,18 so imputations were not conducted. We compared the mean differences between groups to estimate the size of difference and 95% CIs; t test was used to test for significance, and p<0.05 was considered significant.

The original study was a phase II trial where we wished to establish whether the short-term PC should be studied further. We estimated that a sample of more than 25 patients in each arm would enable us to detect differences of >2 on POS-8 at p<0.05 with a power of 80% (SD: 2.25) at 12 weeks. Because of the phase II nature of the trial, we also determined CIs and which outcomes appeared close to significance, following the guidance of Ratain and Sargent,20 as these are valuable in determining whether to progress to a phase III trial and in refining the intervention and trial methods.

Results

Fifty-two patients were randomised to the fast-track group or the control group. Patients had a high level of disability (mean EDSS: 7.7). There were no differences between the two groups at baseline (table 1). Data on 26/26 (fast-track) and 24/26 (control) patients at 12 weeks (time 3) and on 25/26 (fast-track) and 21/26 (control) patients at 24 weeks (final time) were available for analysis18 (Consolidated Standards of Reporting Trials diagram; figure 3). The main reasons for loss to follow-up were death or illness. There was one death in the fast-track group and three deaths in the control group. Two hundred and seventeen of 225 possible questionnaires were completed.18 Care giver data were missing when patients did not have care givers and when care givers did not return the questionnaires.

Table 1

Characteristics of patients and care givers in the two randomised groups

Figure 3

Consolidated Standards of Reporting Trials diagram showing the flow of patients through the study. Numbers listed refer to patients remaining in the study (interviews actually completed at this time point). A small number of interviews were missed because the patient was unavailable (away/in hospital) and because of timing error.

The PC team covered a large geographical area of South East England, with travelling times of up to 2.5 h each way. Of the 52 patients accepted, 46 patients received three visits or less from the PC team. After three visits, six patients were referred on to local established specialist PC services in their local area. Of those patients with three visits or less, their main issues were as follows: 25 had symptom control issues, 12 discussed advance planning, four were seen specifically about issues regarding percutaneous endoscopic gastrostomy insertion/feeding and three were seen specifically for discussions about resuscitation and ventilatory support. One person was seen to facilitate the choice of a preferred place for end-of-life care, and one had multiple concerns. Some people had several concerns.

First research question

The first research question concerns the role of timing of referral to the PC intervention (table 2, figure 4).

Table 2

Comparison of changes in scores in answers to research questions between the fast-start group and the delayed-start group

Figure 4

Changes in outcomes during the trial: mean POS-MS-5 symptom scores (A) and ZBI scores (B) in groups at baseline and over time. POS-MS-5, Palliative care Outcome Scale–MS Symptom Scale; ZBI, Zarit (Care Giver) Burden Interview.

In both groups, during the receipt of care from the PC team, improvement in palliative symptoms (according to POS-MS-5 score) and care giver burden (according to ZBI score) was observed. A comparable rate of improvement was observed for both groups, but a higher reduction in care giver burden was observed for the fast-track (ie, early referral) group.

Second research question

The second research question concerns the potential staff-modifying effect (table 2, figure 4).

Six weeks after baseline, after the removal of the direct effect of PC, and for the next 6 weeks (up to week 12), slightly better symptom control was observed for the fast-track patients compared to those who received the best usual care (POS-MS-5 score mean difference: −0.7; 95% CI −2.2 to 0.7). This difference is even larger in terms of care giver burden (ZBI score mean difference: −4.6; 95% CI −1.6 to −7.6; p=0.004).

At the end of the study, the mean differences between the two groups in terms of symptom control were minimal (POS-MS-5 score mean difference: −0.4; 95% CI −2.3 to 1.6), larger and against the research hypothesis for care giver burden, but with very wide CIs (ZBI score mean difference: +1.8; 95% CI −1.8 to 5.4).

Thus, the data appear to be following the model of figure 2C—the effect of PC continues for around 6 weeks after it is withdrawn (up to week 12) and then wanes in the following 12 weeks, and the difference between groups diminishes over time. Unfortunately, we do not know the exact point between weeks 12 and 24 that the effect begins to wane, as these data were not collected.

Discussion

In this study, PC was developed to offer short-term support and planning to lever change among existing services and then to withdraw existing services. We found that referral to the short-term PC had benefits whether the patient was referred earlier or later. There was significantly greater improvement in care giver burden and no other significant differences, although there was a trend towards better symptom control in the fast-track group. The effects waned after PC was withdrawn, although the exact time is not clear—it was at some point after 12 weeks, with PC usually ending some 6–8 weeks earlier. Therefore, at 24 weeks, those who received PC in the fast-track group (immediate) or in the control group (after 12 weeks) had similar levels of eventual symptoms, care giver burden and costs. Obviously, an earlier reduction of symptoms and care giver burden was preferable to a later reduction, as early reduction leads to a longer period of lesser symptoms or burden. During the 24 weeks of the study, the area under the curve, used for the analysis of global symptoms, would have been lower in the fast-track group. The finding that earlier referral to PC may be helpful for care givers in reducing their burden may be valuable in reducing subsequent morbidity in bereavement. We had hoped that the effects of the short-term PC might be longer lived than 12 weeks, but this did not appear to be the case, and patients began to deteriorate after 12 weeks, although we did not collect sufficient information to determine when. This would need to be the subject of further research, ideally in the phase III trial, testing the intervention. This suggests that, in clinical practice, further PC contact is needed to prevent the reduction in the effect of PC over time. Alternatively, education of other staff may achieve this. The PC team conducted education at the introduction of the service, but not at discharge, and this may be a further opportunity. As the PC team moves into diseases other than cancer, alternative models with short-term interventions and repeated pulses of contact may be needed.

The effect of an earlier referral to PC is poorly researched. Only four published studies have considered this, with contradicting results: none considered care giver outcomes. The US National Hospice Study used a quasi-experimental design to compare hospice care and conventional care. In a secondary analysis, Kidder21 22 suggested that there were cost savings in the last 6 months of life but that earlier referral did not give this cost reduction. In the follow-up of a randomised trial of a US palliative home care team, McCusker and Stoddard23 found that extension of the team, which included accepting patients earlier in the course of illness, was beneficial in terms of costs and outcomes. In the UK, Higginson et al24 found that 227 patients referred earlier to a home and hospital PC team often had lesser symptoms at referral and did not present with the subsequent high levels of symptom distress found in patients referred shortly before their death. A recent randomised trial of 151 patients with newly diagnosed metastatic lung cancer referred early to PC found that the patients who had PC had better quality of life (Functional Assessment of Cancer Therapy–Lung scale: 98 vs 91.5, p=0.03) and that fewer patients had depressive symptoms (16% vs 38%, p=0.01) at 12 weeks compared to the control group. Interestingly, the PC group also had improved survival (median survival: 11.6 months vs 8.9 months, p=0.02).25

Strengths and limitations

This is the first study to rigorously evaluate short-term PC among those seriously affected by MS, a disease that can be a model for other long-term neurological and chronic conditions. It is also the first to evaluate a new model of PC that is consistent with the relatively scarce resources available, as it does not require long-term input. The fast-track design allowed us to explore whether PC had a longer-term effect after it was withdrawn and gave us good recruitment.16 Our limitations include the use of a phase II trial, which, although appropriate at this stage, was based on one centre and had smaller numbers than a phase III trial. Our numbers for care givers were especially small because not all care givers completed the questionnaires. This may be especially important in this secondary analysis, where we are testing not for a response but for differences in responses between early receipt and later receipt of PC. The small sample size may mean that small longer-term effects of PC (positive or negative) were not detected, and our study was not powered to test such differences. Ratain and Sargent20 advocated that a type I error rate at the 0.05 level in phase II trials is not necessary, and they proposed a one-sided test of the null hypothesis with a type I error rate of 0.20. In our study, we found a small effect (−0.7) for symptoms and a larger one for care giver burden (−4.6), both with wide CIs. A further problem in testing for earlier effects versus delayed effects may be that the period of time (12 weeks) for the delay before receiving PC was not long enough. It may be that a longer delay may find effects larger than those found in this study. Patients were referred to the team for different reasons. For example, 25 of those referred had symptom control issues, and 12 needed advanced care planning. Although it is likely that most patients had some symptom control issues, it was probably only a major feature for around half of the patients, with other patients having mixed problems. Therefore, it may be that POS-MS-5 may have only detected changes for those patients who had symptom problems, further reducing our power. Equally, those patients who had difficulties with advanced planning may have been unlikely to experience long-term effects from the intervention. However, our sample was too small for subgroup analysis.

Conclusions

Short-term PC for people severely affected by MS improves symptoms and reduces care giver burden, whether given early or later. This service should be considered for people severely affected by MS and similar neurological conditions. Early receipt of PC appears to have a greater effect on care giver burden. Unfortunately, the effect of short-term PC appears to wane after 12 weeks; thus, we need to consider whether additional short-term PC support should be offered at later points. Alternatively, the short-term intervention may need to be modified to make its effects more sustainable, perhaps with more attention to providing advice and training at the point of discharge or throughout the intervention. The fast-track design is feasible and useful for studying the effects of short-term PC interventions among patients with longer-term conditions.

Main messages

  • For patients severely affected by MS, an earlier referral to a PC team for a short-term intervention had a greater effect on reducing care giver burden compared to a later referral but had a similar effect on reducing symptoms. The trial was a phase II trial and lacked power to detect small differences.

  • The PC team provided, on average, three visits, with all care completed by 6 weeks. Its effects were maintained for 6 weeks after withdrawal but then appeared to wane and were not maintained at 24 weeks.

Questions arising from the study

  • Because the study is a phase II trial, these results would need testing in a fully powered phase III trial.

  • The point at which the effect of PC begins to wane should be determined.

  • New models of short-term PC that have a more sustained effect, either through pulsed contact or greater emphasis on education, should be developed and evaluated.

Current research questions

  • Because the study is a phase II trial, these results would need testing in a fully powered phase III trial.

  • The point at which the effect of PC begins to wane should be determined.

  • New models of short-term PC that have a more sustained effect, either through pulsed contact or greater emphasis on education, should be developed and evaluated.

Key references

▶ Royal College of Physicians, Report of Working Party. Palliative care services: meeting the needs of patients. London: Royal College of Physicians; 2007.

▶ Edmonds P, Hart S, Wei G et al. Palliative care for people severely affected by multiple sclerosis: evaluation of a novel palliative care service. Mult Scler 2010;16:627–36.

▶ Higginson IJ, McCrone P, Hart SR et al. Is short-term palliative care cost-effective in multiple sclerosis? A randomized phase II trial. J Pain Symptom Manage 2009;38:816–26.

▶ Davies E, Higginson IJ. Palliative Care: The Solid Facts. Denmark: World Health Organisation; 2004.

▶ Temel JS, Greer JA, Muzikansky A et al. Early palliative care for patients with metastatic non-small-cell lung cancer. N Engl J Med 2010;363:733–42.

Acknowledgments

We thank the MS Society whose funding made this project possible; the staff involved in data collection, interviews and other aspects of the study, including Sam Hart, Tariq Saleem, Bella Vivat, Troy Cartwright, Gay Foxwell, Barbara Gomes, Farida Malik and Michael Walton; and the Project Advisory Committee (Keith Andrews, Fiona Barnes, Cynthia Benz, Colin Campbell, Sharon Haffenden, Martin Hunt, Robin Luff, David Oliver, Michael Ritchie, Sally Plumb and Carolin Seitz) that oversaw the development of methods and trial conduct. The MS team members at King's College Hospital, as well as at Lambeth, Southwark and Lewisham PC, who referred and collaborated, are greatly acknowledged. Most importantly, we thank the patients and families for sharing their thoughts, experiences and concerns with us.

References

View Abstract

Footnotes

  • Funding MS Society, 676/01.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval King's College Hospital. Ethical approval for the study was granted by the King's College Hospital NHS Trust Local Research Ethics Committee (Protocol Number: 01-04-018).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.