|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology Self-Rated Health as a Predictor of Survival Among Patients With Advanced CancerByFrom the Clinical Epidemiology and Health Outcomes Centre and the Medical Oncology Unit, Canberra Hospital, Canberra, Australia. Address reprint requests to Paul Craft, MD, Medical Oncology Unit, Canberra Hospital, PO Box 11, Woden ACT, 2606, Australia; email: Paul.Craft{at}act.gov.au
PURPOSE: Evidence is emerging about the strong predictive relationship between self-rated health (SRH) and survival, although there is little evidence on palliative populations where an accurate prediction of survival is valuable. Thus, the relative importance of SRH in predicting the survival of ambulatory patients with advanced cancer was examined. SRH was compared to clinical assessments of performance status, as well as to quality-of-life measures. PATIENTS AND METHODS: By use of a prospective cohort design, 181 patients (76% response rate) with advanced cancer were recruited into the study, resurveyed at 18 weeks, and observed to record deaths. RESULTS: The average age of patients was 62 years (SD = 12). The median survival time was 10 months. SRH was the strongest predictor of survival from baseline. Also, a Cox regression comparing changes in SRH over time yielded hazard ratios suggesting the relative risk (RR) of dying was greater for fair ratings at 18 weeks (approximately 3 times) compared with consistent good or better ratings; the RR was even greater (4.2 and 6.2 times) for poor ratings, especially when ratings were poor at baseline and 18 weeks (31 times). Improvement in SRH over time yielded the lowest RR. CONCLUSION: SRH is valid, reliable, and responsive to change as a predictor of survival of advanced cancer. These qualities suggest that SRH should be considered as an additional tool by oncologists to assess patients. Similarly, health managers could use SRH as an indicator of disease severity in palliative care case mix. Finally, SRH could provide a key to help us understand the human side of disease and its relationship with medicine.
INCREASING EVIDENCE IS emerging about the strong predictive relationship between people’s self ratings of their health and length of survival.1-7 One of the most studied ratings is the question, "In general, would you say your health is excellent, very good, good, fair, or poor?" Responses to this question, typically called self-rated health (SRH), significantly contribute to the explanation of survival duration over and above more objective measures of health.8-11 Similarly, SRH maintains a strong relationship with survival after considering a wide range of clinical prognostic factors.3,12-14 In the present study, we focused on patients with advanced cancer receiving care at a community oncology center. There is little research examining the relationship between SRH and survival among this population. Evidence is growing, however, about the inaccuracies of physician-based estimates of palliative care on patients’ length of survival.15-18 For instance, Christakis and Lamont17 conclude after examining the survival estimates of 343 doctors for 468 terminally ill patients before hospice referral that doctors are systematically optimistic. They suggest that this phenomenon may be adversely affecting decision making for patients near the end of life. We considered that given the predictive value of SRH in community populations, SRH may provide a valuable addition to the tools oncologists use to clinically assess patients with advanced cancer living in the community. Furthermore, we thought that SRH may be an indicator that health managers can use to assess disease severity for outpatient palliative care, forming a part of palliative care case mix. Finally, we believe it is useful to examine the predictiveness of SRH for this population because SRH may be a valuable indicator for clinical trials: as a surrogate end point, or as an exclusion, inclusion, or stratifying indicator at enrollment, possibly complementing standard performance status (PS) measures (clinically assessed physical health status).
Sample The study was designed to capture a broad population of patients being managed with palliative intent. The sample was randomly selected from patients by means of outpatient and acute inpatient oncology services at the Canberra Hospital in the Australian Capital Territory between March 1996 and September 1996. To be eligible, patients had to have a diagnosis of incurable malignant disease; had to be older than 18 years; and had to have or be likely to develop within 6 months symptoms attributable to the disease. Definitional criteria were agreed on between medical and radiation oncology clinicians. Patients with incurable malignant disease of a relatively indolent type, such as low-grade lymphoma with minimal extent of disease, for whom there was no reasonable likelihood of any symptoms (or treatment) within the 6 months study period were excluded. Of the 237 patients identified as eligible, 56 declined to take part, yielding a response rate of 76% (n = 181). The study was approved by the Australian Capital Territory Health and Community Care Ethics Committee. All patients provided written informed consent before participating in the study.
Study Design and Measures Clinical information used included the diagnostic group, treatment given within 1 month of enrollment, and the extent of the disease. The study was not designed collect blood and other biologic samples.
Data Analysis
Baseline Predictors of Survival The average age of participants was 62 years (SD = 12; range, 22 to 85 years). There were slightly more women (51%) than men (49%). The clinical characteristics of the patients are listed in Table 1. From the date of recruitment, the median survival time was 10 months, with a total of 163 deaths. At baseline, 5% of participants were totally bedridden (ECOG 4), 18% were in bed more than 50% of the time and only capable of limited self-care (ECOG 3), 33% were in bed less than 50% of the time and capable of self-care (ECOG 2), and 44% were ambulatory and capable of light work (ECOG 1). In terms of SRH at this time, 9% of patients rated their health as excellent, 14% as very good, 35% as good, 28% as fair, and 14% as poor. SRH and PS (ECOG) were weakly to moderately correlated (Spearman’s rho = 0.29; Table 2).
In relation to the other HRQoL measures, SRH at baseline was moderately to strongly related to the SF-36’s General Health Perceptions scale (Spearman’s rho = -0.63) and the QLQ-C30’s Quality of Life scale (Spearman’s rho = -0.50). SRH was mostly moderately correlated with the other SF-36 and comparable QLQ-C30 scales. The strongest correlations were with physical functioning, pain, vitality, and social functioning. The symbols assessed via QLQ-C30 mostly had weak correlations with SRH. The SF-36’s General Health Perceptions scale (it includes the SRH question) tended to have weaker correlations with the measures than SRH; the QLQ-C30’s Quality of Life scale and ECOG’s PS generally correlated more strongly with the measures than SRH (Table 2).
Each of the indicators was examined separately for its association with survival (Table 2). The physical health and global measures were all significant, with better health at baseline associated with a greater probability of longer survival. The mental health measures were not significant, apart from the SF-36’s Social Functioning scale, with better functioning associated with a greater likelihood of longer survival. Five out of eight of the QLQ-C30’s symptoms were significant, and the more severe the symptom, the greater the probability of dying. On the basis of the likelihood ratio statistic, the strongest indicator associated with survival was SRH ( Figure 1 demonstrates Kaplan-Meier cumulative survival curves by the SRH categories regrouped as excellent, very good, and good into one group, fair as another group, and poor as another. The probability of survival significantly decreased with worse SRH (log-rank = 23.5, df = 2, P < .00001). The positive ratings (excellent, very good, and good) were grouped together because their survival probabilities were similar (median survival times ranged 1.27 to 0.96 years); this is common practice with SRH.1,5,22
SRH was the strongest predictor of survival compared with selected clinical indicators, PS, SF-36, and QLQ-C30 when a stepwise method of entry into Cox regression models was used (Table 3 lists SRH v SF-36 and Table 4 demonstrates SRH v QLQ-C30). This ranking remained whether a forward or backward stepwise inclusion criteria was used, or whether SRH was considered for entry into the regression after entering the clinical indicators, PS, SF-36, and QLQ-C30 (Table 5), with these indicators having little effect on SRH’s effect size (in fact, the effect size was enhanced). Patients who rated their health as fair were approximately two times more likely to die (median survival time, 8.5 months) than those who rated their health as at least good (median survival time, 1 year), whereas those who assessed their health as poor were approximately four times (seven times in the model in Table 5) more likely to die (median survival time, 2.9 months).
Other patient experience predictors found to be significant in the multivariate models were appetite loss, emotional functioning, and fatigue. For both appetite loss and fatigue, the more severe the assessment, the greater the risk of dying. On the other hand, better emotional functioning was associated with a greater risk of dying; this relationship, however, only existed after SRH and appetite loss were entered into the model. Significant clinical indicators included diagnosis type and treatment. Lung cancer patients tended to have had a greater risk of dying than those who were diagnosed with lymphoma, myeloma, or leukemia. Radiotherapy and hormonal manipulation tended to have been associated with longer survival than chemotherapy (including chemotherapy + radiotherapy) and supportive care only.
Changes in SRH and Survival
A Cox regression comparing changes in SRH over time yielded hazard ratios that suggest the relative risk (RR) of dying was greater for participants who rated their health as fair at 18 weeks (approximately three times) compared with those who assessed their health as consistently good or better, whereas the RR was even greater (4.2 and 6.2 times) for those who rated their health poor at 18 weeks, especially when patients rated their health poor at baseline and 18 weeks (31 times). The RR of dying among patients who improved their ratings between baseline and 18 weeks later was not significant; this group’s median survival time was 2.0 years, compared with those who consistently had at least good ratings (2.4 years; Tables 7 and 8).
We found that SRH is by far the best predictor of survival in comparison with well-studied indicators, including selected clinical indicators, PS, appetite loss, fatigue, and popular HRQoL measures. This finding indicates that SRH may be an accurate measure of disease severity and that patients have a tremendous ability to use external and internal information to assess their own health.1-6,13 We found that the measures of HRQoL (SF-36 and QLQ-C30) are significantly less important in the prediction of survival. Thus, HRQoL is not linearly associated with the process leading to death—that is, the poorest functioning is not necessarily one step away from death. The unusual relationship between survival and emotional functioning, however, needs further investigation (Table 4). The finding may suggest that after adjusting for variations in disease severity through SRH, people prepare emotionally as they get closer to death. As for other population groups,1,3-7,13,22-25 our findings demonstrate that SRH is valid and reliable as a measure in advanced cancer. Regardless of where patients are in the course of their disease, SRH consistently predicted the associated RRs of death and survival times. Furthermore, SRH is responsive to change, with poorer ratings associated with shorter survival times and a greater risk of death, and conversely, better ratings yielded relatively longer survival times and a lower RR of death. These qualities suggest that SRH should be considered a valuable tool that could be added to the methods used by oncologists to assess patients (along with tumor response, PS, and symptom assessment). The use of SRH in clinical practice may help clinicians improve their evaluations of patients’ health and their predictions of duration of survival, and thereby improve decision making and quality of care.17 Also, more research identifying the factors influencing persons rating may provide opportunities for intervention and perhaps even improve survival times. Similarly, health managers could use SRH as an indicator of disease severity in palliative care case mix. Differentiating palliative care patients according to their SRH responses could increase the accuracy of palliative care case mix. In addition, SRH could be used as a surrogate end point in clinical cancer trials or as an inclusion/exclusion indicator, with our findings suggesting that SRH would be a better indicator than PS, which is commonly used in this way in clinical trials in advanced disease. In conclusion, we envisage that as more evidence unfolds about the importance of SRH in measuring the disease process, SRH could provide a key to help us understand the human side of disease and its relationship to medicine.
Supported in part by a research grant from the Palliative Care Project of the Commonwealth Department of Health and Family Services. We acknowledge the contribution of research nurses MaryAnne Bustead, Annette Dahler, and Cate Farrell. Maurene Bourne provided invaluable assistance with data management.
1. Lesser GT: Social and productive activities in elderly people. Self rated health is important predictor of mortality. BMJ 320: 185, 2000
2.
Batty D: Re: Self-perceived health and 5-year mortality risks among the elderly in Shanghai. China. Am J Epidemiol 150: 219, 1999 3. Bosworth HB, Siegler IC, Brummett BH, et al: The association between self-rated health and mortality in a well-characterized sample of coronary artery disease patients. Med Care 37: 1226-1236, 1999[CrossRef][Medline]
4.
Idler EL, Kasl SV, Lemke JH: Self-evaluated health and mortality among the elderly in New Haven, Connecticut, and Iowa and Washington counties, Iowa, 1982-1986. Am J Epidemiol 131: 91-103, 1990
5.
Idler EL, Angel RJ: Self-rated health and mortality in the NHANES-I Epidemiologic Follow-up Study. Am J Public Health 80: 446-452, 1990 6. Idler EL, Benyamini Y: Self-rated health and mortality: A review of twenty-seven community studies. J Health Soc Behav 38: 21-37, 1997[CrossRef][Medline] 7. Idler EL, Kasl S: Health perceptions and survival: Do global evaluations of health status really predict mortality? J Gerontol 46: S55-S65, 1991[Abstract] 8. Gaston-Johansson F, Fall-Dickson JM, Bakos AB, et al: Fatigue, pain, and depression in pre-autotransplant breast cancer patients. Cancer Pract 7: 240-247, 1999[CrossRef][Medline]
9.
Lee Y: The predictive value of self assessed general, physical, and mental health on functional decline and mortality in older adults. J Epidemiol Community Health 54: 123-129, 2000
10.
Blakely TA, Kennedy BP, Glass R, et al: What is the lag time between income inequality and health status? J Epidemiol Community Health 54: 318-319, 2000 11. Mulsant BH, Ganguli M, Seaberg EC: The relationship between self-rated health and depressive symptoms in an epidemiological sample of community-dwelling older adults. J Am Geriatr Soc 45: 954-958, 1997[Medline]
12.
Idler EL, Russell LB, Davis D: Survival, functional limitations, and self-rated health in the NHANES I Epidemiologic Follow-up Study, 1992: First National Health and Nutrition Examination Survey. Am J Epidemiol 152: 874-883, 2000 13. Idler EL, Kasl SV: Self-ratings of health: Do they also predict change in functional ability? J Gerontol B Psychol Sci Soc Sci 50: S344-S353, 1995[Abstract] 14. Idler EL: Survival, functional limitations, and self-rated health in the NHANES I Epidemiologic Follow-up Study 1992. Am J Epidemiol 152: 874-883, 2000
15.
Morita T, Tsunoda J, Inoue S, et al: Survival prediction of terminally ill cancer patients by clinical symptoms: Development of a simple indicator. Jpn J Clin Oncol 29: 156-159, 1999 16. Kutner JS, Steiner JF, Corbett KK, et al: Information needs in terminal illness. Soc Sci Med 48: 1341-1352, 1999
17.
Christakis NA, Lamont EB: Extent and determinants of error in doctors’ prognoses in terminally ill patients: Prospective cohort study. BMJ 320: 469-472, 2000 18. Lamont EB, Christakis NA: Some elements of prognosis in terminal cancer. Oncology (Huntington) 13: 1165-1170, 1999[Medline] 19. Oken MM, Creech RH, Tormey DC, et al: Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 5: 649-655, 1982[Medline] 20. Ware JEJ, Sherbourne CD: The MOS 36-item short-form health survey (SF-36): I—Conceptual framework and item selection. Med Care 30: 473-483, 1992[Medline]
21.
Aaronson NK, Ahmedzai S, Bergman B, et al: The European Organization for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 85: 365-376, 1993
22.
Dasbach EJ, Klein R, Klein BE, et al: Self-rated health and mortality in people with diabetes. Am J Public Health 84: 1775-1779, 1994
23.
McCallum J, Shadbolt B, Wang D: Self-rated health and survival: A 7-year follow-up study of Australian elderly. Am J Public Health 84: 1100-1105, 1994 24. Gold M, Franks P, Erickson P: Assessing the health of the nation. The predictive validity of a preference-based measure and self-rated health. Med Care 34: 163-177, 1996[CrossRef][Medline] 25. Engstrom G, Hedblad B, Janzon L: Subjective well-being associated with improved survival in smoking and hypertensive men. J Cardiovasc Risk 6: 257-261, 1999[Medline] Submitted August 7, 2001; accepted February 12, 2002.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 = 19.5, df = 1, P < .0001). The next strongest was fatigue (
