Trials in palliative careBMJ 1995; 310 doi: https://doi.org/10.1136/bmj.310.6979.598c (Published 04 March 1995) Cite this as: BMJ 1995;310:598
- Gerdy Rinck,
- Jos Kleijnen,
- Trudi G A M Van Den Bos,
- Hanneke J C J M De Haes,
- Egbert Schade,
- Cees H N Veenhof
- Junior clinical epidemiologist Clinical epidemiologist Department of Clinical Epidemiology and Biostatistics, Academic Medical Centre, University Hospital of Amsterdam, PO Box 22700, 1100 DE Amsterdam, Netherlands
- Professor of social medicine and chronic diseases Institute of Social Medicine, Academic Medical Centre
- Associate professor of medical psychology Department of Medical Psychology, Academic Medical Centre
- Professor of family medicine Department of General/Family Medicine, Academic Medical Centre
- Associate professor of medical oncology Division of Medical Oncology, Department of Internal Medicine, Academic Medical Centre
EDITOR,—In their discussion of the problems and pitfalls of evaluating a palliative care service McWhinney and colleagues mentioned three published trials in palliative care services.1 Apparently only trials with additional care services were selected, and trials on palliative treatment were excluded. In the field of additional care services we found one other trial.2 Many trials in palliative cancer care report problems with patient selection, accrual, attrition, contamination of exposure, and selection of adequate outcome measures. We also conducted a randomised trial on palliative care services, for patients with disseminated colorectal cancer. Our trial compared fluorouracil as single agent by continuous infusion at home versus combination therapy with leucovorin at an outpatient clinic. Main outcome measures were quality of life, use of care services, and satisfaction with care. Quality of life was assessed with a global assessment with a linear analogue scale, the medical outcomes study's 24 item short form, the affect balance scale, the Rotterdam symptom checklist, the instrumental help for daily living scale, the activities of daily living scale, and the social experiences scale. Satisfaction was assessed with the Rand corporation's patient satisfaction questionnaire.
Our trial failed to attain the intended sample size too and we stopped the intake of patients. McWhinney and colleagues reported problems with patient recruitment, attrition, and the exposure of the control group to services equivalent to the home support team. In our case the major problem was patient accrual. As it happened, the number of eligible patients was nearly half that of previous years. To improve patient intake we invited on several occasions 11 associated centres to refer their patients. During two years only 22 patients entered the trial, of whom nine came from associated centres. Further-more, 22 eligible patients refused to participate in the trial. This was suspected early on in the trial, and the reason for refusal and the characteristics of these patients were registered. Eight patients did not want chemotherapy, 10 refused either continuous infusion or randomisation, and four were not asked by the physician because of medical reasons. In contrast with McWhinney and colleagues' experiences, attrition did not have a major role—four patients died before the first evaluation at two months, and all other patients completed the questionnaires. A possible explanation for this difference could be that the patients in McWhinney and colleagues' trial had a more advanced disease than our patients.
Despite inclusion criteria allowing only patients with disseminated colorectal cancer, without metastasis to the brain, and with normal results of haematology tests, with normal liver and kidney function, World Health Organisation performance status of 2 or more, and no serious comorbidity to participate, our study population turned out to be rather heterogeneous. Intersubject variability of the measures of quality of life was great. Individual plots of quality of life scores over time seemed stable when the clinical situation was stable and seemed to decline around the time of disease progression. No apparent relations between quality of life and satisfaction were found. Small sample size and great intersubject variability do not allow for further statistical analysis.
In conclusion, randomised trials in palliative cancer care are difficult, but not impractical. Our trial failed because of difficulties in patient accrual, but other common problems, such as attrition, contamination of exposure, or selection of adequate outcome measures did not have a major role. Future randomised trials in palliative cancer care should consider multicentre participation and the possibility of patient entry without limitations to type of primary cancer. Moreover, the care intervention should be well defined, and adequate outcome variables should be selected.