|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Quality of Life in Palliative Cancer Care: Results From a Cluster Randomized TrialByFrom the Palliative Medicine Unit, Department of Oncology and Radiotherapy, University Hospital of Trondheim, and Unit of Applied Clinical Research, Norwegian University of Science and Technology, Trondheim, Norway. Address reprint requests to Marit S. Jordhøy, MD, Unit of Applied Clinical Research, Norwegian University of Science and Technology, 5etg Kreftbygget, University Hospital of Trondheim, N-7006 Trondheim, Norway; email: mjordhoy{at}online.no
PURPOSE: To assess the impact of comprehensive palliative care on patients’ quality of life. The intervention was based on cooperation between a palliative medicine unit and the community service and was compared with conventional care. PATIENTS AND METHODS: A cluster randomized trial was carried out, with community health care districts defined as the clusters. Patients from these districts who had malignant disease and survival expectancy between 2 to 9 months were entered onto the trial. The main quality-of-life end points were physical and emotional functioning, pain, and psychologic distress assessed monthly by using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30 (EORTC QLQ-C30) questionnaire and Impact of Event scale (IES). In total, 235 intervention patients and 199 controls were included. RESULTS: During the initial 4 months of follow-up, the compliance was good (72%) and comparable among treatment groups. No significant differences on any of the quality-of-life scores were found. At later assessments and for scores that were made within 3 months before death, there was also no consistent tendency in favor of any treatment group on the main outcomes or other EORTC QLQ-C30 scales/items. CONCLUSION: A general program of palliative care may be important to ensure flexibility and to meet the needs of terminally ill patients. However, to achieve improvements on a group level of the various dimensions of quality of life, specific interventions directed toward specific symptoms or problems may have to be defined, evaluated, and included in the program.
DESPITE MODERN treatment advances, approximately 50% of all cancer patients die from their disease,1 meaning that for every second patient, the focus eventually has to shift from cure or life prolongation to palliation. Numerous studies have shown that health services provided at this stage need optimization.2-4 Palliative cancer care aims at improving a patient’s subjective well-being. To meet the multiple and varying needs, it is generally believed that the care should be holistic, multidisciplinary, and family- as well as patient-centered. Several models of organizing the care have been developed.5-7 However, limited impact on patients’ health-related quality of life (HRQL), including physical and psychosocial dimensions, has been demonstrated.8,9 Few randomized trials have been reported,10-14 and these, as well as most nonrandomized studies, have been criticized for methodological reasons.8,9,15 Further, a large, nonconsistent number of subjective outcome measures have been used. In the randomized trials,10-14 a majority of the instruments were not developed, or previously validated, among cancer patients in general or among cancer patients within palliative care. Lack of well-defined primary end points among a wide range of outcomes may also make it difficult to interpret the trial results, especially when they point in different directions.9,12 Hence, a need for further research on the HRQL achievements of comprehensive palliative care has been expressed.8,9 The Palliative Medicine Unit (PMU) at the University Hospital of Trondheim in Trondheim, Norway, conducted a cluster randomized trial to compare its service with conventional care (control). It was postulated that the PMU program would have a positive impact on patients’ HRQL; that is, result in improved pain control, better physical and emotional functioning, and less psychologic distress, hence these dimensions were defined as the main end points. The assessments were made monthly from enrollment to death using the European Organization for Research and Treatment of Cancer Quality of Life-C30 (EORTC QLQ-C30) questionnaire16,17 and the Impact of Event Scale (IES).18,19 The present article covers the HRQL results, while additional outcomes, place of death and hospital utilization, have been reported separately.20
Setting Health care within the trial area is entirely public. The community service is organized into independent, geographically defined districts, all of which include general practitioners, nursing homes, and home care, and are comparable in terms of available resources.20,21 The University Hospital of Trondheim (958 beds) provides all hospital service. Conventional care for advanced cancer patients is shared among the hospital departments and the community according to diagnosis and medical needs. No well-defined follow-up routines exist (Table 1). Poor communication between level of services has been addressed as a general problem.1 Apart from the PMU, no specialist palliative care service is available.
The Palliative Care Program The PMU was opened in November 1994 and includes outpatient and inpatient clinics as well as a multidisciplinary consultant team working daytime hours (Table 1). Besides enabling the patients to spend more time at home and eventually die at home, the PMU aimed at optimizing the patients’ HRQL. Our hypothesis was that this could be achieved by providing hospital care from a palliative care unit, training of community professionals, optimizing the cooperation between services, and applying a multidisciplinary approach to the patients’ problems. More specifically, we assumed that increased palliative care competence on both hospital and community level and a well-planned and coordinated follow-up would improve symptom control and hence lead to better physical functioning. The same aspects, including systematic assessment of psychologic, social, and spiritual needs, were intended to reduce psychologic distress and improve emotional functioning. A predefined intervention program including detailed guidelines for the interaction between the PMU and the community was initiated by the opening of the trial (Table 1). The patients’ general practitioner (GP) and a community nurse were defined as the main professional caregivers. When entering the program, the patients were referred to the PMU. Individual treatment plans were set up in a joint meeting of the patient, the informal caregiver, the GP, the community nurse, and a consultant nurse or physician from the PMU (Table 1). Follow-up consultations by the GP and the community nurse were arranged according to the patients’ needs and predefined minimum standards. Hospital service was offered on request and always at the PMU, that is, unless otherwise required for medical reasons (eg, surgery). The PMU consultant team participated in the inpatient care, handled the PMU outpatient clinic, coordinated the follow-up, and was available to the community staff for supervision and advice and to join visits in the patient’s home. An educational program for the community professionals included bedside training and 6 to 12 hours of lectures every 6 months.
Trial Design The recruitment procedure23 was based on cooperation with all health care professional within the trial area. Information about trial design and purpose was given through staff meetings or personal letters, and during the recruitment period, written reminders were distributed at regular intervals. Two research assistants made weekly screening visits at relevant hospital departments. They also handled the referrals, informed the patients, and completed the enrollment.23 All eligible patients were invited to participate in a study of quality of life and health care needs among advanced cancer patients. Those residing in a cluster (district) that was allocated to the intervention were asked for transferal to the PMU.23 The study was approved by the Regional Ethical Review Committee.
Outcome Measures The EORTC QLQ-C30 includes a total of 30 items and is composed of scales that evaluate physical (five items), emotional (four items), role (two items), cognitive (two items), and social (two items) functioning, as well as global health status (two items). Higher mean scores on these scales represent better functioning. There are also three symptom scales measuring nausea and vomiting (two items), fatigue (three items), and pain (two items), and six single items assessing financial impact and various physical symptoms. Higher mean values on the symptom scales/items mean more symptomatology. Before statistical analyses are made, the raw EORTC QLQ-C30 scores are linearly transformed to 0 to 100 scales.27 A mean change in scores of five to 10 has been found to represent "a little" subjective change to the patients, whereas a change of 10 to 20 represents a moderate change28; thus differences of 10 points or more may be regarded as clinically significant. The IES is a 15-item, self-reported scale that assesses how patients react to stressful events such as having incurable cancer. Several scoring systems have been used for the questionnaire.29 In the present study, using a Norwegian translation, each item was scored on a six-point categorical response scale, ranging from 0 (not at all) to 5 (very much).19 The IES items’ scores are summarized into two subscales: intrusion and avoidance. The intrusion subscale (range 0 to 35) describes how thoughts and impressions related to the disease reappear. The avoidance subscale (range 0 to 40) assesses behavior characterized by denying the meaning and consequences of the disease. Higher scores indicate more distress. Missing items were imputed for the EORTC QLQ-C30 and the IES multi-item scales, using the method advocated by the EORTC Quality-of-Life Study Group.27 If at least half of the items from a scale were completed, the values of missing ones were imputed as the mean value of the completed items. For the IES, which had a higher number of missing items, the analyses were made both with and without using imputation; imputation had a minor impact on the group means and did not alter the results concerning the comparisons between treatment groups. All questionnaires, except the baseline forms, were distributed by mail.23 Patients who did not respond within 2 weeks received a written reminder. If still no answer was given, the patients received no further questionnaires and were referred to as drop-outs.
Sample Size
Statistics The hypotheses tests were conducted in consistency with the cluster randomized design,20,33 and predictive factors were taken into account. First, the factors that previously have been reported to be of importance to the HRQL outcomes34 and/or those showing imbalance between the intervention and control group were considered for adjustment.20 These factors were screened using multiple linear regression against AUC/SAUC with a backward stepwise approach to exclude those not contributing significantly (P > .05). Thereafter, differences in AUC and SAUC between treatment groups were tested by bootstrap estimation35 to fit regression models allowing for the clustering36 and those factors that were found to be predictive to the outcome. Individual patients were resampled with replacement (500 times). To avoid overlooking contradicting results within time frames that were not covered by the hypothesis tests, scores from the fifth and sixth month of follow-up were explored by simple descriptive statistics. Comparisons beyond this point were not considered meaningful because of a small number of respondents. Finally, scores were regrouped according to time from assessment to death, and differences between treatment groups on main outcome scores that were made 1, 2, and 3 months before death were tested using bootstrap estimation to fit regression models allowing for the clustering.36 The P value for significance was set at P = .01 to provide some protection from multiple testing. Survival curves were estimated according to the Kaplan-Meier method, and log-rank tests adjusted for diagnosis used for comparison of treatment groups. The analyses were performed using SPSS version 9.0 for Windows (SPSS Inc, Chicago, IL)37 and STATA Release 5 (STATA Corp, College Station, TX).36
Patient Characteristics Between March 1995 and November 1997, 434 patients were entered onto the trial,23 including 235 intervention patients and 199 controls. At baseline, the groups differed for housing, access to informal help, home care nursing, and, slightly, for living situation (Table 2). It was suspected that the enrollment to the intervention and control group was unequally affected by factors related to the departments that traditionally handled the various cancer diagnoses (eg, workload and bed capacity). To explore this, diagnoses were classified according to traditional sharing of treatment responsibility within the University Hospital of Trondheim (Groups A, B, and C, Table 2). The distribution of patients to these groups differed significantly between the treatment groups. There was also a difference in time from diagnosis to inclusion (nonsignificant). Performance status and tumor burden were comparable (Table 2); 33 (14%) and 33 (17%) had lung metastasis in the intervention and control groups, respectively, whereas 53 (23%) and 43 (22%), respectively, had bone metastasis. In both groups, 49% had disabilities (including hearing loss and vision impairments) from causes other than cancer.
Attrition and Compliance The follow-up was closed in October 1999. In total, 219 intervention patients (93%) and 176 controls (88%) died. More intervention patients died at home: 25% compared with 15% of the controls.20 Ten intervention patients (4%) and 13 controls (7%) survived and completed a follow-up of 2 years, whereas 6 (3%) and 10 (5%) withdrew from the intervention and control groups, respectively. Median survival was 99 days (95% confidence interval [CI], 79 to 119 days) in the intervention group and 127 days (95% CI, 88 to 166 days) in the control group (P = .1). The number of patients who were still alive and under observation (at risk) at subsequent HRQL assessment points is listed in Table 3.
The mean proportion of days under observation that was spent in hospital was 31% and was similar in the two groups.20 Among the intervention patients, 200 (85%) were admitted to the PMU inpatient unit once or more, 152 (65%) visited the outpatient clinic, and 174 (74%) participated in a joint meeting between the PMU and the community at trial entry. For 11 patients, neither of these arrangements was recorded. Overall, 79% of the hospital days in the intervention group was spent at the PMU, and the mean number of PMU outpatient consultations was 3.6. Twelve control patients (6%) were seen at the PMU (outpatient clinic and/or inpatient unit); among these, three inpatient visits and no outpatient visit were recorded during the first 6 months of follow-up. On average, the control patients spent 2.9% of the overall inpatient time at the PMU. The compliance in completing quality-of-life questionnaires was calculated as the number of questionnaires completed in proportions of patients who were alive and under observation (at risk). For the assessments at 1 to 6 months, the compliance was 73% and 70% in the intervention and control groups, respectively. Up to 4 months after trial entry, the compliance at individual assessment points was comparable among treatment groups (Table 3). At later assessments, however, there was a tendency toward a higher response rate among the intervention patients. Excluding the baseline questionnaires, the compliance was also estimated backwards from death. A larger proportion of the intervention patients at risk completed a questionnaire at 1 and 2 months before death (36% and 62% in the intervention group compared with 25% and 50% in the control group, respectively; Table 3). The proportion of missing items in the EORTC part of the questionnaire was low in both groups, whereas for the IES, the mean percentage was 10% in the intervention group and 8% in the control group (Table 3). Overall, the proportion of missing items was highest in the forms that were filled in 1 month before death (Table 3).
HRQL and Psychologic Distress
For the AUC estimates, no statistically significant differences between the intervention and control groups were found, neither for psychologic distress, pain, physical and emotional functioning (P > .1), or for any of the other EORTC QLQ-C30 scores. Results of analyses on SAUC estimates were entirely consistent and are listed in Table 5. Of the factors that were considered for adjustment (age, sex, living with a spouse, housing, home care nursing at trial entry, access to informal help, diagnoses, and time from diagnosis to inclusion), none was found to contribute significantly to any of the AUCs of the main outcomes or for the SAUC of the intrusion and emotional and physical functioning scores. Having access to informal help (regression coefficient, 10.35; P = .02) and sex (regression coefficient, -2.97; P = .01) contributed to the SAUC of the pain and avoidance scores, respectively, and were allowed for in the bootstrap analysis. For those patients who died, the analyses were repeated allowing for time from inclusion to death. Consistent results were achieved.
To visualize the lack of difference between treatment groups on the main outcomes, the changes from baseline on which the SAUC and AUC calculations were based are shown in Fig 1. The scores for the fifth and sixth assessment were added to the figure to show that there was also no difference on the HRQL scores after the fourth month of follow-up.
For the HRQL scores that were obtained within 3, 2, and 1 months before death, no difference in mean scores was clinically significant (10 or more), neither for pain or emotional and physical functioning (data not shown). In both groups, there was a marked decline in physical functioning from 3 months to 1 month before death. The avoidance and intrusion scores were close to identical. Comparisons were made for scores that were obtained after an intervention period of at least 1 month, that is, the baseline assessments were excluded. No difference of either mean scores or changes from baseline was statistically significant.
Neither on pain, emotional and physical functioning, and psychologic distress, nor on any other HRQL dimension measured by the EORTC QLQ-C30, did the scores of patients entered into the intervention program show improvement in comparison with controls. Hence, the results are in accordance with the findings of earlier RCTs, which have evaluated the impact on patients’ HRQL of comprehensive palliative care programs. On a wide range of HRQL-related outcomes, patients’ satisfaction with care excluded, either no10,11,14 or very few achievements12,13 have been reported. Compared with most RCTs within the area,10-14 the study was large. A reasonable sample size and good and comparable compliance were maintained for 4 months, and although a few control patients had PMU visits, any serious contamination between treatment groups was avoided by cluster randomization. However, the number of clusters was restricted and this resulted in an imbalance on some sociodemographic characteristics, which reflected the variation of the general cluster populations. A difference in distribution of diagnostic groups was probably related to lack of concealment of individual patient allocation,20,38 because the treatment assignment of individual patients could be identified from their address. The imbalance seemed, however, to be based on practical considerations about the departments that traditionally handled these groups and not on patient-related factors. No other evidence of selection bias was revealed, and as the treatment groups were comparable on a wide range of baseline data, including HRQL scores, we found no reason to believe that the effect of the experimental intervention had been obscured.20 The design and the revealed imbalances were taken into account in the analyses. Assessing outcomes in palliative care is difficult.39 We carefully selected instruments that reflect clinically relevant issues in palliative care, and have been widely applied, rigorously developed, and extensively tested. The validity and reliability of the EORTC QLQ-C30 as well as its responsiveness to change and its ability to discriminate between treatment groups have been established among advanced cancer patients.17,40,41 The IES has also been demonstrated to perform well in similar settings.19 However, neither was developed for end-of-life care, and poor sensitivity during the final decline cannot be entirely ruled out. More relevant is that those patients to whom this problem most likely applied seemed unable to fill in the questionnaire. Few patients responded during their last month of life. The compliance data also indicated a tendency toward scores being made closer to death in the intervention group compared with the control. Allowing for time from assessment to death made, however, no difference to the results of the main analyses. Hence, we have found our findings to be reliable, and conclude that within four times the intervention period that was considered necessary to have an effect, there was no evidence of any impact on the patients’ HRQL. For the very last weeks of life, no firm conclusions can be drawn because of poor and incomparable compliance. There are several ways to explain our findings. First, they should be interpreted in light of the context in which they were obtained. The Norwegian Health System is known to perform well with fairness in distribution to the inhabitants.42 Locally, some hospital departments outside the PMU were rather focused on palliative care, awareness of the trial itself may have further increased this, and a pain clinic was also available on a consultative basis. The course of advanced cancer according to subjective outcome measures is poorly described, as is the extent to which it can be modified. Therefore, it is difficult to decide about the quality of the care in both trial arms, and it may be postulated that the conventional care was generally good, leaving room for only marginal improvements. Second, the timing of the study needs consideration. For ethical reasons, the trial had to be carried out before the PMU service was established as an integrated offer to all inhabitants, and so it was initiated just after the opening of the unit. By this time, neither the PMU staff nor the community workers had any experience with the overall concept and the new routines that were to be implemented. Third, the intervention was strongly based on the existing community service with limited palliative care competence, although staff was trained during the trial period. Hence, we cannot exclude more positive findings in another setting, or if the trial had been carried out at a later point in time. However, our results were consistent with those from other RCTs, and we find it pertinent to question whether it is possible to achieve significant quality-of-life improvements, measured on a group level, by such broadly defined and targeted interventions. Like most palliative care programs, our intervention included at heterogeneous groups of patients with varying symptoms. According to reviews on psychologic intervention trials, interventions targeted at cases, or patients at risk, had strong clinical effects, whereas the effect of those more broadly applied were weak.43 This may also apply to palliative care strategies. Further, our program represented an organizational change involving a broad range of health care offers. The care of individual patients depended on the actions and cooperation of several health care workers, whose knowledge and experience were bound to vary. Hence, we suggest that further research on the HRQL achievements of palliative cancer care should concentrate on evaluating the effect of more specific strategies in relevant subgroups of patients, whereas comprehensive models may be improved by including clearly defined guidelines for symptom management, to which all involved personnel is obliged to comply. The results of this study add important information about the impact of comprehensive palliative cancer care and its assessment, but they do not allow us to conclude that such programs are not worthwhile. The present intervention enabled more patients to stay at home to die,20 and according to published reports on patients’ wishes, this is a favorable outcome.44,45 Several others have reported improved family and patient satisfaction.8,10,11,14,46,47 Informal care givers, general practitioners, and nurses have rated palliative care superior to standard care in patients’ last days of life,48 for which this trial was not specifically designed and provided minor information. Programs including home care may also be justifiable for their cost-effectiveness.14,49,50 Thus, if these benefits are confirmed, there are strong arguments for continuing the development and implementation of such models.
Supported by grants from the Norwegian Cancer Society (grant nos. 95147 and 98004), the Swedish Cancer Society (grant no. 3650-B95-01XAC), and the Norwegian Medical Association Fund for Quality Improvement (grant nos. 4488/93).
We thank all the patients who participated in the trial and willingly filled in the quality-of-life questionnaires. We also thank Anne Lise Nessæther, Torgeir Fjermestad, Bjørg Petersvik, Nina Hassel, and Frøydis Høyem Koteng for their indispensable participation in trial planning and organizing. The invaluable work of the research assistants Torbjørn
1. Norges offentlige utredninger: Omsorg og kunnskap: Norsk kreftplan. NOU 20:88-89, 1997. Statens forvaltningstjeneste, Statens trykning, Oslo, Norway, 1997
2.
Addington-Hall J, McCarthy M: Dying from cancer: Results of a national population-based investigation. Palliat Med 9: 295-305, 1995 3. Houts PS, Yasko JM, Harvey HA, et al: Unmet needs of persons with cancer in Pennsylvania during the period of terminal care. Cancer 62: 627-634, 1988[Medline] 4. The SUPPORT Principal Investigators: A controlled trial to improve care of seriously ill hospitalized patients: The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). JAMA 274: 1591-1598, 1995[Abstract] 5. Stajduhar KI, Davies B: Death at home: Challenges for the family and directions for the future. J Palliat Care 14: 8-14, 1998[Medline] 6. Greer DS, Mor V, Morris JN, et al: An alternative in terminal care: Results of the National Hospice Study. J Chronic Dis 39: 9-26, 1986[Medline]
7.
Higginson I: Evidence based palliative care. BMJ 319: 462-463, 1999 (editorial)
8.
Smeenk FWJM, van Haastregt JCM, de Witte LP, et al: Effectiveness of home care programmes for patients with incurable cancer on their quality of life and time spent in hospital: A systematic review. BMJ 316: 1939-1944, 1998
9.
Salisbury C, Bosanquet N, Wilkinson EK, et al: The impact of different models of specialist palliative care on patients’ quality of life: A systematic literature review. Palliat Med 13: 3-17, 1999 10. Kane RL, Bernstein L, Wales J, et al: A randomised controlled trial of hospice care. Lancet 1: 890-894, 1984[Medline]
11.
Zimmer JG, Groth-Junker A, McCusker J: A randomized controlled study of a home health care team. Am J Public Health 75: 134-141, 1985 12. Addington-Hall JM, MacDonald LD, Anderson HR, et al: A randomised controlled trial of effects of coordinating care for the terminally ill cancer patients. BMJ 305: 1317-1322, 1992 13. McCorkle R, Benoliel JQ, Donaldson G, et al: A randomized clinical trial of home nursing care for lung cancer patients. Cancer 64: 1375-1382, 1989[Medline] 14. Hughes SL, Cummings J, Weaver F, et al: A randomized trial of the cost effectiveness of VA hospital-based home care for the terminally ill. Health Serv Res 26: 801-817, 1992[Medline] 15. Rinck GC, van den Bos GAM, de Haes HJCJM, et al: Methodological issues in effectiveness research on palliative cancer care: A systematic review. J Clin Oncol 15: 1697-1707, 1997[Abstract]
16.
Aaronson NK, Ahmedzai SA, Bergman B: The EORTC QLQ-C30: A quality of life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 85: 365-376, 1993 17. Kaasa S, Bjordal K, Aaronson N, et al: The EORTC Core Quality of Life Questionnaire (QLQ-C30): Validity and reliability when analysed with patients treated with palliative radiotherapy. Eur J Cancer 31A: 2260-2263, 1995
18.
Horowitz MJ, Wilner N, Alvarez W: Impact of Event Scale: A measure of subjective stress. Psychosom Med 41: 209-218, 1979 19. Kaasa S, Malt U, Hagen S, et al: Psychological distress in cancer patients with advanced disease. Radiother Oncol 27: 193-197, 1993[Medline] 20. Jordhoy M, Fayers P, Saltnes T, et al: A palliative care intervention and death at home: A cluster randomised trial. Lancet 356: 888-893, 2000[Medline] 21. Styrings og informasjonssystemet for helse-og sosialtjenesten i kommunene: Sammenligningstall for kommunene 1998. Nord-Trøndelag, Sør-Trøndelag, Møre og Romsdal. Hefte I-0920/2B. Oslo, Norway, Sosial og helsedepartementet i samarbeid med Statistisk sentralbyrå og Statens helsetilsyn, 1998 22. Murray DM: Design and analysis of group-randomized trials. New York, NY, Oxford University Press, 1998
23.
Jordhoy MS, Kaasa S, Fayers P, et al: Challenges in palliative care research, recruitment, attrition and compliance: Experience and recommendations from a prospective, randomized trial. Palliat Med 13: 299-310, 1999
24.
MacAdam DB: An initial assessment of suffering in terminal illness. Palliat Med 1: 37-47, 1987
25.
Fowlie M, Berkeley J, Dingwall-Fordyce I: Quality of life in advanced cancer: The benefits of asking the patient. Palliat Med 3: 55-59, 1989 26. Holmen J, Midthjell K, Bjartveit K, et al: The North-Troendelag Health Survey 1984-1986: Participation, non-participation and frequency distributions. Oslo, Norway, National Institute of Public Health, Unit of Health Service Research; Verdal, Norway, National Institute of Public Health, Community Medicine Research Centre, Report 4, 1990 27. Fayers P, Aaronson N, Bjordal K, et al, on behalf of the EORTC study group: EORTC QLQ-C30 Scoring Manual. Brussels, Belgium, EORTC Study Group on Quality of Life, 1995
28.
Osoba D, Rodrigues G, Myles J, et al: Interpreting the significance of changes in health related quality of life scores. J Clin Oncol 16: 139-144, 1998 29. Briere J, Elliott DM: Clinical utility of the Impact of Event Scale: Psychometrics in the General Population. Assessment 5: 171-180, 1998[Abstract] 30. Machin D, Campbell MJ, Fayers PM, et al: Sample Size Tables for Clinical Studies (ed 2). Oxford, United Kingdom, Blackwell Science, 1997, pp 24 31. Donner A, Klar N: Statistical considerations in the design and analysis of community intervention trial. J Clin Epidemiol 49: 435-439, 1996[Medline] 32. Fairclough DL: Methods of analysis for longitudinal studies of health related quality of life, in Staquet MJ, Hays RD, Fayers P (eds): Quality of life assessments in clinical trials. Oxford, United Kingdom, Oxford University Press, 1998
33.
Ukoumunne OC, Gulliford MC, Chinn S, et al: Methods in health service research: Evaluation of health interventions at area and organisation level. BMJ 319: 376-379, 1999 34. Hjermstad MJ, Fayers PM, Bjordal K, et al: Health-related quality of life in the general Norwegian population assessed by the European Organization for Research And Treatment of Cancer Core Quality-of-Life Questionnaire: The QLQ = C30 (+3). J Clin Oncol 16: 1188-1196, 1998[Abstract] 35. Effron B, Tibshirani RJ: An Introduction to the Bootstrap. New York, NY, Chapman & Hall, 1993 36. STATA Corporation: Statistics Data Management Graphics, Reference Manuals: STATA Statistical Software—Release 5.0. College Station, TX, STATA Press, 1997, P-Z, pp 118-138 37. SPSS Inc: SPSS (Statistical Package for the Social Sciences) for Windows: Base System User’s Guide, release 9.0.1. Chicago, IL, 1999 38. Schulz KF: Subverting randomization in controlled trials. JAMA 274: 1456-1458, 1995[Abstract]
39.
Hearn J, Higginson IJ: Outcome measures in palliative care for advanced cancer patients: A review. J Public Health Med 19: 193-199, 1997 40. Cunningham D, Pyrhonen S, James RD, et al: Randomised trial of irinotecan plus supportive care versus supportive care alone after fluorouracil failure for patients with metastatic colorectal cancer. Lancet 352: 1413-1418, 1998[Medline] 41. Fayers PM, Weeden S, Curran D: EORTC QLQ-C30 Reference Values. Brussels, Belgium, European Organization for Research and Treatment of Cancer, 1998 42. World Health Organization: The World Health Report 2000: Health systems—Improving performance. World Health Organization, WHO 2000, http://www.who.int/whr/ 43. Sheard T, Maguire P: The effect of psychological interventions on anxiety and depression in cancer patients: Results of two meta-analyses. Br J Cancer 80: 1770-1780, 1999[Medline] 44. Dunlop R, Davies R, Hockley J: Preferred versus actual place of death: A hospital palliative care support team experience. Palliat Med 3: 197-201, 1989 45. Townsend J, Frank AO, Fermont D, et al: Terminal cancer care and patients’ preference for place of death: A prospective study. BMJ 301: 415-17, 1990
46.
Hearn J, Higginson I: Do specialist palliative care teams improve outcomes for cancer patients? A systematic literature review. Palliat Med 12: 317-332, 1998 47. Bosanquet N, Salisbury C: Providing a Palliative Care Service: Towards an Evidence Base. Oxford, United Kingdom, Oxford University Press, 1999, pp 5-6, pp 97-130
48.
Grande GE, Todd CJ, Barclay SIG, et al: A randomized controlled trial of a hospital at home service for the terminally ill. Palliat Med 14: 375-385, 2000
49.
Raftery JP, Addington Hall JM, MacDonald LD, et al: A randomized controlled trial of the cost-effectiveness of a district co-ordinating service for terminally ill cancer patients. Palliat Med 10: 151-161, 1996 50. Hughes SL, Ulasevich A, Weaver FM, et al: Impact of home care on hospital days: A meta analysis. Health Serv Res 32: 415-432, 1997[Medline] Submitted September 26, 2000; accepted June 6, 2001. This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
, intervention group; 
, control group.
vreness, Gro Underland, and Bjørn Fougner and data managers Anders Skjeggestad and Turi Saltnes is highly appreciated. 
