|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2000 American Society for Clinical Oncology
Health-Related Quality of Life in Survivors of Wilms’ Tumor and Advanced Neuroblastoma: A Cross-Sectional StudyFrom the McMaster University, Hamilton; Health Utilities Inc, Dundas; University of Toronto, Toronto; and University of Alberta and Institute of Health-Economics, Edmonton, Ontario, Canada; and University of Aberdeen, Aberdeen, Scotland. Address reprint requests to Ronald D. Barr, MD, Room 3N27B, McMaster University, Health Sciences Centre, 1200 Main St W, Hamilton, Ontario, Canada L8S 4J9; email rbarr{at}fhs.csu.mcmaster.ca
PURPOSE: In pediatric oncology, Wilms’ tumor and advanced neuroblastoma represent opposite ends of the spectra of survival probability and therapeutic intensity. Consequently, it was envisaged that survivors of Wilms’ tumor would enjoy better health status and health-related quality of life (HRQL) than survivors of advanced neuroblastoma.
PATIENTS AND METHODS: Health status questionnaires were sent to the parents of all eligible children and to the children themselves if they were RESULTS: A greater burden of morbidity was identified in the survivors of advanced neuroblastoma than in survivors of Wilms’ tumor based on the assessments of the parents of these children. In particular, survivors of advanced neuroblastoma exhibited deficits in hearing and speech. It is possible that this morbidity burden reflects the prevalent use of platinum compounds (causing ototoxicity) in this group. Within parent-child dyads there was a high level of percentage agreement on responses in all attributes except cognition. CONCLUSION: Extension of this study to a larger sample size of patients will provide clarification of these observations.
SOME 10 YEARS AGO, we embarked on an economic evaluation of treatments for cancer in childhood1 using health-related quality of life (HRQL) utility scores for comprehensive outcome health states.2 Three common forms of malignant disease were selected to represent different prospects for cure, inversely related to the intensity of therapy:3-5 high probability of cure, more than 80% (Wilms’ tumor); intermediate probability of cure (high-risk acute lymphoblastic leukemia [ALL]); and low probability of cure, less than 20% (advanced neuroblastoma). In earlier publications, we reported the results of HRQL studies in survivors of high-risk ALL who exhibited a greater burden of global morbidity than was observed in children with standard-risk disease who, in turn, were barely distinguishable from children in the general population.6,7 Now intent on examining the extremes of the original, selected spectrum of prognosis and therapeutic intensity, ie, Wilms’ tumor and advanced neuroblastoma, we speculated that survivors of Wilms’ tumor would enjoy better health status and HRQL than would survivors of advanced neuroblastoma. (Our previous review of the literature8 revealed no reports of the comprehensive health status of survivors of Wilms’ tumor or neuroblastoma, and we have found none since.) In this report we describe the results of the studies undertaken to address this speculation, which is based on relative chemotherapeutic intensity in the comparative groups. A related, but secondary, objective was to assess the inter-rater agreement between the health status reported by the survivors themselves and that provided by their parents using the subsample of subjects for whom we had assessments from both perspectives. As in our earlier studies of survivors of ALL, we used the Health Utilities Index (HUI), a generic system for measuring health status and HRQL that has been shown to be a useful tool for assessing the overall health status of groups not well studied previously and for economic evaluations of treatment programs.
All survivors of Wilms’ tumor (stages 2 to 53 ) and advanced neuroblastoma (stage 4 and inoperable stage 39 ) who were treated in the Children’s Hospital at Chedoke-McMaster in Hamilton (now part of the Hamilton Health Sciences Corporation) and the Hospital for Sick Children in Toronto over a decade (1985 through 1995) were considered suitable for study. Children with stage 1 Wilms’ tumor and those with stages 1, 2, or 4S neuroblastoma were excluded because they received little or no adjuvant treatment (radiotherapy and/or chemotherapy). Additional exclusion criteria were continuing treatment and relapsed disease. Demographic, disease-, and treatment-related data were obtained from hospital charts and used for defining subgroups of patients (see Results) for preparing descriptive and comparative results. The HUI is a family of multiattribute, preference-based systems for measuring comprehensive health status and HRQL that have been developed during the past 20 years. Each HUI system consists of two basic components: a health status classification system and a preference-based valuation system. HUI1 has, for most purposes, been superseded by HUI2 and HUI3. Both HUI2 and HUI3 have strengths and weaknesses and, therefore, these two systems are considered complementary. The HUI is applied by collecting sufficient information about the health status of each study subject to classify health status according to at least one of the health status classification systems. The classification systems consist of a set of attributes (ie, dimensions), and each attribute has three to six levels of functional capacity. A comprehensive HUI health state is defined by a vector consisting of one level from each of the attributes defined by the classification system. Utility scores of morbidity (ie, valuation of single-dimension disability) and single-summary utility scores of overall HRQL (ie, valuation of overall disability reported by each subject) are determined from published utility functions. The HUI utility functions (preference-based mathematic scoring models) provide an efficient method for determining a utility score for the comprehensive health state experienced by a person at a point in time. The HUI has been included, or proposed to be included, in studies being undertaken by approximately 200 investigative teams. These teams are based in major centers around the world: North and South America, Europe, Australia, and Asia. HUI2/3 was chosen, in part, because it has been used widely in pediatric oncology and so comparative data are available. Furthermore, HUI3 has been included in Statistics Canada National Longitudinal Survey of Children and Youth and National Population Health Surveys, providing an abundance of normative data. Each family was sent by mail one or two well-validated questionnaires (HUI23SU.15Q and HUI23PU.15Q). One parent of each child was asked to complete a questionnaire (HUI23SU.15Q), and children who had achieved at least a grade 3 education were asked to complete a similar questionnaire (HUI23PU.15Q). A cover letter stated that "The questionnaires should be completed at home, by each of you independently, and returned as soon as possible in the enclosed envelope bearing prepaid postage." The questionnaires were accompanied by a letter of explanation from a staff oncologist in the respective institution together with a statement that the study had been approved by the respective institutional review board. Those families who had not returned the questionnaires promptly were then contacted by telephone (on a maximum of two occasions at intervals of 1 week). Packages that were returned undelivered prompted contact with the family physician of most recent record to obtain an updated address. A well-established algorithm was used to convert responses to the questions to attribute levels of the HUI2 and HUI3 systems. The resulting seven element vectors for HUI2 (with all children assigned a level of 1.00 for fertility) and the eight element vectors for HUI3 were used then to generate single-attribute and overall utility scores, all as described previously.10-13 The multiattribute utility functions were used to determine utility scores of single-attribute morbidity burden and overall HRQL from a community perspective. Utility scores are defined on a scale of 0.00 to 1.00. For the overall HRQL scale, death has a utility score of 0.00 and perfect health has a utility score of 1.00. For each of the single-attribute morbidity scales, the worst level on each attribute has a utility score of 0.00 and the level describing no morbidity has a utility score of 1.00 All data were entered twice into computerized databases and discrepancies were reconciled with information from the original questionnaires. SAS (SAS/STAT User’s Guide, SAS Institute, Cary, NC ) was used for statistical analysis. Selection of the appropriate test for statistically significant differences was based on the homogeneity of variance among and between study groups. Homogeneity of variance was tested using the two-tailed Folded Form Statistic in SAS. Where such homogeneity was unequal or could not be assessed, a Wilcoxon test (based on median scores) was used. Otherwise, the Students’ t test (based on mean scores) was used. Agreement between attribute levels derived from responses of children and their parents is measured in terms of percentage agreement.
Of the entire group of patients treated for Wilms’ tumor or neuroblastoma, 73% (n = 240) were believed to be alive, off therapy, and free of disease at the initiation of this study. Subsequently, almost 13% (n = 31) of this remainder were lost to follow-up. Close to one half of the residual number of patients were excluded on the basis of stage of disease, and an additional small group was discovered either to be still on active therapy or to have relapsed. The response rate from the final study population (n = 93) was approximately 84% overall; the respective proportions in Wilms’ tumor and neuroblastoma being 52 (87%) of 60 patients and 26 (79%) of 33 patients. Of the children who were eligible to complete questionnaires independently, the corresponding proportions for those surviving Wilms’ tumor and neuroblastoma were 31 (86%) of 36 and 11 (73%) of 15 patients. These several proportions are not statistically significantly different. The ages at diagnosis, completion of therapy, and time of study; the duration from end of therapy to study; the sex distribution; and the distribution for stages of disease are listed in Table 1. In large measure, these reflect the typical combination of features at diagnosis modified by survival. The exception is the preponderance of girls in the neuroblastoma group. The rates for types of surgery, the protocols of chemotherapy used, the fields of irradiation, and whether bone marrow transplantation (BMT) was undertaken are also listed in Table 1. Almost all patients with Wilms’ tumor underwent complete resection of the primary tumor, in most as the first therapeutic event. Fewer children with advancedneuroblastoma underwent complete resections, and many of these were undertaken as delayed interventions after systemic chemotherapy (with or without irradiation). All but one child received multiagent chemotherapy. Patients with neuroblastoma were treated using a wide variety of protocols (but almost 90% of the children received cisplatin), whereas those with Wilms’ tumor were treated according to protocols of the National Wilms’ Tumor Study or with in-house regimens at the Hospital for Sick Children. Autologous BMT was performed on only 20% of the patients with advanced neuroblastoma, but all received a platinum compound as part of the preparative regimen (which was devoid of total-body irradiation).
More children surviving Wilms’ tumor, in comparison with those surviving advanced neuroblastoma, were assessed by their parents as having no disabilities (Table 2), and the reverse is true for multiple disabilities (  two attributes affected). Many more children surviving advanced neuroblastoma than those surviving Wilms’ tumor had sensory deficits (HUI2), and this was all in the attributes of hearing and speech (HUI3; Tables 3 and 4). With the exception of one child who had had Wilms’ tumor and one who had survived advanced neuroblastoma, both of whom were assessed by their parents as affected at level 4 in the attribute of self-care, no very severe deficits (levels 4 or 5) were revealed by HUI2 in either population of survivors (Table 3). Using HUI3, with the exception of one survivor of Wilms’ tumor with level 4 dexterity, the only assessments at level 4 were in hearing (n = 1) and speech (n = 2) among survivors of neuroblastoma and in cognition for five children who had had Wilms’ tumor and four who had had neuroblastoma (Table 4).
The differences in the frequency distributions of levels in the attributes of emotion and pain between HUI2 and HUI3 (Tables 3 and 4) reflect the different constructs of these attributes in the two HUI systems. In HUI2, emotion focuses on worry and anxiety, whereas in HUI3, the focus is on happiness versus depression. In HUI2, pain focuses on severity and analgesia, whereas in HUI3, the focus is on severity and disruption of activity. Frequency distributions of global health states are displayed in the Appendix. Using both the HUI2 and HUI3 systems, the survivors of Wilms’ tumor exhibited 16 unique health states (0.3 health states per patient). The survivors of advanced neuroblastoma exhibited 13 unique health states using HUI2 and 14 using HUI3 (approximately 0.5 health states per patient). The difference between the mean overall HRQL utility scores of survivors of Wilms’ tumor and that of survivors of advanced neuroblastoma did not reach formal statistical significance (Table 5). However, it should be noted that the power to detect the observed differences in mean utility of 0.05 for HUI2 scores and 0.06 for HUI3 scores was only 64% and 41%, respectively. Furthermore, statistically significant differences were observed between mean single-attribute utility scores (Table 5), with the burden of morbidity in sensation (HUI2) being greater in survivors of advanced neuroblastoma, this being explained by higher rates of deficits in hearing and speech (HUI3). Additionally, on examining the frequency distribution for HUI utility scores, notable differences were observed for proportions of subjects having multiple or severe deficits: 23% of neuroblastoma patients had HUI2 utility scores of less than 0.80 compared with 6% of patients with Wilms’ tumor; 31% of neuroblastoma patients had HUI3 utility scores of less than 0.80 compared with 15% of patients with Wilms’ tumor. Moreover, there was greater variance in overall HRQL utility scores in the neuroblastoma group than in the Wilms’ group (HUI2, P = .0009; HUI 3, P = .0065)
The levels of agreement between the self-reporting children and their parents, in relation to the respective assessments of the children’s health status, are listed in Table 6. We reviewed the two-way contingency tables for pairs of child and parent attribute level results to assess informally the extent of agreement between parental and child reports. Formal analysis of agreement between pairs of child and parent responses, using kappa statistics, is not reported because of limitations in the underlying data: (1) the sample sizes are small (in part because only children with at least a grade 3 education were eligible for self-report), (2) for most attributes there are relatively few pairs of observations representing disagreement between the responses of parents and children, and (3) for most attributes there were few reports of any disability (ie, most subjects were reported by both child and parent as being normal [level 1] on all attributes). Furthermore, intraclass correlation coefficients were not calculated on these health status measures because they are not continuous variables. The data in Table 6 reveal a high or very high level of percentage agreement ( 84%) in all attributes except cognition. Of the 520 paired observations, only 12 were discordant by more than one level. Eight of these were in Wilms’ tumor patients, in the attribute of cognition in the HUI3 system, and in five of these, the child reported more disability than did the parent.
Children with Wilms’ tumor receive much less aggressive treatment (surgery, irradiation, and chemotherapy) than is given to most children with advanced neuroblastoma.14 Consequently, it may be anticipated that survivors of the former disease would experience fewer and less severe sequelae and so would enjoy better health status and HRQL. Conversely, the age distribution of Wilms’ tumor patients is skewed toward older children, relative to the neuroblastoma distribution, and so these children may have greater awareness of their disease, treatment, and prognosis with adverse effects on their HRQL. In this study, parents were chosen as the primary assessors (respondents) to collect data about the maximum number of eligible subjects because there were more responses from parents than from children. The characteristics of the children from whose parents responses were received (n = 78) were very similar to the characteristics of the entire group of eligible children (n = 93; data not shown). The health status and HRQL of the survivors of Wilms’ tumor were comparable to those of survivors of standard-risk ALL7 and of children in the general population.7,15 By comparison, the burden of morbidity was greater in the survivors of advanced neuroblastoma, although the differences in mean overall HRQL utility scores were attenuated and did not reach conventional levels of statistical significance. Survivors of advanced neuroblastoma seem to experience HRQL similar to that found among survivors of high-risk ALL.7 This burden compares favorably to the HRQL observed in a cohort of survivors of brain tumors in childhood (mean global utility score of 0.85).16 There was notable sensory loss among survivors of advanced neuroblastoma in the components of hearing and speech. It is plausible that this reflects the ototoxicity of platinum compounds,17,18 which were received by the great majority of these patients. Whether this observation is indicative of parental awareness of such a risk is being addressed in a continuing audiometric evaluation of these patients. The cognitive morbidity burden observed in both groups of survivors remains to be explained. This burden was revealed only by HUI3 in which the construct for cognition is based on memory function and problem-solving. In HUI2 the construct is based on learning and remembering. Whether the cognitive morbidity antedates the cancer diagnosis, reflects the global cancer experience, or represents a relatively specific toxicity is worthy of further study. These observations support our speculation, namely that survivors of Wilms’ tumor will enjoy better health status and HRQL than will survivors of advanced neuroblastoma. This conclusion derives further support from the number of unique health states per patient (see Appendix), which reflects the overall morbidity burden in a population.19 However, the heterogeneity of treatment regimens for advanced neuroblastoma (a feature even of current therapeutic strategies) precludes simple extrapolation of these findings to other cohorts of patients. The relatively modest burden of morbidity, with limited variability, in the survivors of Wilms’ tumor is consistent with the efforts of modern therapy to sustain high cure rates in this disease at ever-diminishing cost. In survivors of neuroblastoma, the burden of morbidity is greater and more variable, reflecting the variety of earlier efforts to improve survival with intensification of therapy. This leads us to anticipate even more morbidity from current treatment strategies that involve myeloablative therapy and autologous stem-cell transplantation. Furthermore, estimates of this variability will inform estimates of sample sizes for future studies. Overall, the level of agreement was quite high between the responses of the children self-reporting on their health status and the proxy responses of their parents. Of course, agreement could reflect a lack of independence between parent and child assessments, despite the clear instructions. However, the lower level of agreement for cognition suggests that, for the most part, the assessments were independent. Disagreements about performance in school or cognitive capacity are not uncommon. Further, there is evidence of systematic differences in HUI assessments by viewpoint or perspective.20 A variety of viewpoints are credible, viewpoints should not be considered interchangeable, and each viewpoint should be respected for what it contributes to an understanding of the health status and HRQL of patients treated for cancer in childhood. With the exception of the preponderance of girls in the sample of survivors of neuroblastoma, the distributions of characteristics for patients in this study are similar to those of large samples, such as were reported by Breslow et al21 and Green22 for Wilms’ tumor. Nevertheless, a larger sample size would provide greater confidence with respect to the generalizability of these results and to testing formal hypotheses based on them. We hope that these objectives will be addressed soon and we encourage others to collaborate in such studies.
Frequencies of Unique Comprehensive Health States: HUI2 
Frequencies of Unique Comprehensive Health States: HUI3
Supported in part by the Children’s Cancer and Leukemia Research Fund and the Summer Career Placement Program of Human Resources Development–Canada. We thank Humaira Khan, Kim Nagel, and Saba Safdar for their valuable facilitation of this study.
This study was presented in part at the annual meeting of the International Society of Pediatric Oncology in Istanbul, Turkey, September 23-27, 1997, and at the Fifth International Conference on the Long-Term Complications of Treatment of Children and Adolescents for Cancer in Niagara-on-the-Lake, Ontario, Canada, June 11-13, 1998.
1. Feeny D, Furlong W, Barr RD, et al: A comprehensive multi-attribute system for classifying the health status of survivors of childhood cancer. J Clin Oncol 10: 923-928, 1992[Abstract] 2. Torrance GW, Feeny D: Utilities and quality-adjusted life years. Int J Tech Assess Health Care 5: 559-575, 1989 3. D’Angio GJ, Breslow N, Beckwith JB, et al: Treatment of Wilms’ tumor: Results of the Third National Wilms’ Tumor Study. Cancer 64: 349-360, 1989[Medline] 4. Barr RD, de Veber LL, Pai KM, et al: Management of children with acute lymphoblastic leukemia by the Dana-Farber Cancer Institute protocols: An update of the Ontario experience. Am J Pediatr Hematol Oncol 14: 136-139, 1992[Medline] 5. Bernstein ML, Leclerc JM, Bunin G, et al: A population-based study of neuroblastoma incidence, survival and mortality in North America. J Clin Oncol 10: 323-329, 1992[Abstract] 6. Feeny D, Leiper A, Barr RD, et al: The comprehensive assessment of health status in survivors of childhood cancer: Application to high risk acute lymphoblastic leukemia. Br J Cancer 67: 1047-1052, 1993[Medline] 7. Barr RD, Furlong W, Dawson S, et al: An assessment of global health status in survivors of leukemia in childhood. Am J Pediatr Hematol Oncol 15: 284-290, 1993[Medline] 8. Barr RD, Furlong W, Feeny D, et al: Evaluating treatments for childhood cancer: A process for critical appraisal of the literature and a summary of the results. Int J Technol Assess Health Care 11: 1-10, 1995[Medline]
9.
Brodeur GM, Pritchard J, Berthold F, et al: Revisions of the international criteria for neuroblastoma diagnosis, staging and response to treatment. J Clin Oncol 11: 1466-1477, 1993 10. Feeny D, Furlong W, Boyle M, et al: Multi-attribute health status classification systems: Health Utilities Index. PharmacoEconomics 7: 490-502, 1995[Medline] 11. Torrance GW, Furlong W, Feeny D: Multi-attribute preference functions: Health Utilities Index. PharmacoEconomics 7: 503-520, 1995[Medline] 12. Torrance GW, Feeny DH, Furlong WJ, et al: Multi-attribute preference functions for a comprehensive health status classification system: Health Utilities Index Mark 2. Med Care 34: 702-722, 1996[Medline] 13. Furlong W, Feeny D, Torrance GW, et al: Multiplicative, multi-attribute utility function for the Health Utilities Index Mark 3 (HU13) system: A technical report. McMaster University Centre for Health Economics and Policy Analysis Working Paper No 98-11, 1998 14. Pizzo PA, Poplack DG (eds): Principles and Practice of Pediatric Oncology (ed 3). Philadelphia, PA, Lippincott-Raven, 1997 15. Saigal S, Feeny D, Furlong W, et al: Comparison of the health-related quality of life of extremely low birthweight children and a reference group of children at age eight years. J Pediatr 125: 418-425, 1994[Medline] 16. Barr RD, Simpson T, Whitton A, et al: Health-related quality of life in survivors of tumors of the central nervous system in childhood: A preference-based approach to measurement in a cross-sectional study. Eur J Cancer 35: 248-255, 1999 17. Piel IJ, Meyer D, Perlia CP, et al: Effects of cis-diammine dichloroplatinum (NSC-119875) on hearing function in man. Cancer Chemother Rep 58: 871-875, 1974[Medline] 18. Skinner R, Pearson ADJ, Amineddine HA, et al: Ototoxicity of cisplatinum in children and adolescents. Br J Cancer 61: 927-931, 1990[Medline] 19. Feeny D, Furlong W, Barr RD: Multi-attribute approach to the assessment of health-related quality of life: Health Utilities Index. Med Pediatr Oncol 1: 54-59, 1998 (suppl) 20. Grootendorst P, Feeny D, Furlong W: Does it matter whom and how you ask? An investigation into inter- and intra-rater agreement in the 1990 Ontario Health Survey. J Clin Epidemiol 50: 127-135, 1997[Medline] 21. Breslow N, Olshan A, Beckwith JS, et al: Epidemiology of Wilms’ tumor. Med Pediatr Oncol 21: 172-181, 1993[Medline] 22. Green DM: Wilms’ tumor. Eur J Cancer 33: 409-418, 1997 Submitted January 11, 1999; accepted May 5, 2000.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2000 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
