Influence of proxy respondents and mode of administration on health status assessment following central nervous system tumours in childhood

Transcription

1 . Quality of Life Research, 6, pp Influence of proxy respondents and mode of administration on health status assessment following central nervous system tumours in childhood A. W. Glaser,* K. Davies, D. Walker and D. Brazier Department of Child Health (A. W. Glaser, D. Walker) and Department of Paediatric Physiotherapy (K. Davies, D. Brazier), Queen s Medical Centre, Nottingham NG7 2UH, UK Quality of Life Research Vol Central nervous system (CNS) tumours account for 20% of childhood cancers. Survivors often experience severe physical, neuropsychological and social sequelae of the disease and its treatment. Health status assessment in these individuals is an essential clinical outcome measure, yet little consensus exists regarding the optimum methodology. The influence of proxy respondents (parents, physiotherapists and doctors) and mode of administration (home and clinic) in which assessment is performed has been evaluated in a cohort of 37 survivors of childhood CNS tumours. A health-related quality of life (HRQOL) questionnaire, incorporating the Mark II and III Health Utilities Indices, was completed at home and in clinic by patients and parents. Doctors and physiotherapists completed this questionnaire plus Lansky Play-Performance and Karnofsky Performance scores. No significant differences between raters for single attribute scores occurred either at home or in clinic, although a wide range of agreement (κ = , percentage agreement %) between observers was revealed. Most agreement occurred between parents and patients: this was greatest on home completion (κ = , percentage agreement %). Doctors and physiotherapists agreed less on subjective attributes (emotion, cognition and pain). Better agreement between all observers was demonstrated if responses were classified as normal or abnormal. Inter-observer agreement was greater for the HRQOL questionnaire than for Karnofsky and Lansky scores. Home completion of questionnaires provides a reliable, acceptable and convenient method of assessing health status. This project was funded by grants from the University of Nottingham Medical School Trust Fund, the Nottingham Brain Tumour Research Fund and the Rank Foundation (AWG). Valuable comments from Quality of Life Research reviewers were received. * To whom correspondence should be addressed at Department of Haematology/Oncology, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada. Key words: Central nervous system neoplasms; health status; proxy respondents. Introduction A dramatic increase in survival from childhood cancer has occurred over the last twenty years. By the year 2000 it is estimated that 1 in 900 young adults will be survivors of malignancy. Survival is often complicated by physical, psychological and social side-effects of both the disease and its treatment. Central nervous system (CNS) tumours account for 20% of these neoplasms. 1 The location of the primary pathology and CNS-directed therapy (surgery, radiotherapy and chemotherapy) all increase the chance of disability in this group compared to survivors of other forms of cancer. 2 It is widely accepted that mortality statistics taken in isolation are inadequate as a measure of outcome in children s cancer studies. Yet there is no consensus regarding optimal methodology for the evaluation of health status, or health-related quality of life (HRQOL), in survivors. 3,4 The wide range of instruments available for the assessment of HRQOL in adults are not appropriate for use with children due to issues of growth, development and dependence on parents and carers. Issues of comprehension and interpretation mean that proxy respondents are needed for most generic instruments that are currently available. Few self-completed instruments exist for use with paediatric populations and consequently little data is available to define the effect of this on health status assessment. However, evidence exists that the more subjective dimensions of health status are less adequately measured by proxy respondents than the more objective domains which include functional capacity. 5,6 Proxies under-estimated morbidity for the subjective domains of pain and emotion 1997 Rapid Science Publishers Quality of Life Research. Vol

2 A. W. Glaser et al. in the 1990 Ontario Health Survey using the Mark III Health Utilities Index. 7,8 The goal of this study was to clarify some of the methodological issues involved in the determination of HRQOL in childhood survivors of CNS tumours. The relationship between responses of patients, their parents, doctors and physiotherapists to a HRQOL questionnaire has been examined. Additionally the effect of completion of questionnaires at home and in clinic has been investigated. Clarification of these issues is essential if HRQOL data is to be collected in a reliable standardized fashion as part of randomized clinical trials in the treatment of children s cancer. This paper focuses on survivors of CNS tumours yet aims to answer methodological issues relevant to all HRQOL assessments in paediatric populations independent of the instrument used or population studied. The methodology described would be equally applicable and relevant to the follow-up of other illnesses in childhood. Materials and methods Subjects Thirty-seven patients were eligible for inclusion. Selection criteria were that: (1) patients were attending the neuro-oncology follow-up clinic at Queen s Medical Centre, Nottingham; (2) the person had been treated at this centre for a brain or spinal tumour of any histological or primary site; (3) the diagnosis had been made when the child was under 17 years of age; (4) they were over 5 years of age and had not received treatment for at least one year at the time of the study and (5) they were not in the terminal stages of their disease (five patients were excluded for this reason). Procedure Local ethical committee approval was obtained. Written consent was requested by post at the beginning of the study from all parents and patients aged over nine years. Two weeks prior to their routine neuro-oncology clinic appointments, parents and patients aged 10 years or more were sent an HRQOL questionnaire which they were instructed to complete and return (in provided stamped, addressed envelope) prior to clinic review. The lag time between completion of questionnaires at home and in clinic was between seven and twelve days. Patients aged under 10 years were excluded from home completion as their grammar and comprehension skills were thought to be inadequate for self-completion of questionnaires. In clinic, identical questionnaires were filled in prior to any consultation; additionally respondents were asked to describe any event that had happened since completion of the previous questionnaire that may influence their responses. Patients aged between five and nine years inclusive completed an HRQOL questionnaire with the neuro-oncology play-specialist. This interview was conducted in the absence of parents or other family members. Following the medical consultation the doctor (consultant oncologist, radiotherapist or neurosurgeon) completed the HRQOL questionnaire and Lansky and Karnofsky assessments. Patients were then reviewed by a paediatric physiotherapist who provided Lansky, Karnofsky and HRQOL questionnaire ratings. Questionnaires HRQOL questionnaire. This contained fifteen questions providing information sufficient to classify health status according to the Mark II and III Health Utilities Indices. These were augmented by two additional questions covering the domains of self-esteem and confidence for the future. The Mark II and III Health Utilities Indices are generic assessments of HRQOL. The Mark II covers six attributes (mobility, self-care, senses, cognition, emotion and pain) whilst the Mark III has been expanded to eight (ambulation, dexterity, vision, hearing, speech, cognition, emotion and pain) To enhance comprehension in the United Kingdom, wording of questions was adapted as previously described. 12 Two additional questions were added to cover the essential areas of self-esteem and confidence for the future. These had a similar format to the Health Utilities Index questions and were formulated following extensive discussion with cancer patients, their parents, social workers, psychologists and other members of the multi-disciplinary oncology team (Appendix 1). Lansky Play-Performance Scale. A proxy-rated instrument based on the child s level of play and activity. 13 It is a decile score offering ten responses ranging from 100 (normal functioning) through 0 (death). Karnofsky Performance Index. The prototypic decile quality of life measure for assessing physical function. It takes no account of the patient s psychosocial status Quality of Life Research. Vol

3 Health status assessment in children Statistical analyses Analyses were performed using the Statistical Package for the Social Sciences (SPSS PC+). Mark III Health Utilities Attribute scores were calculated according to the developmental characteristics of the instrument. 15 The language modification of Bilson and Walker alters the most distressed (extreme) response options to the Health Utilities Index emotion attributes: following a mapping exercise using 40 English physiotherapists, the responses D and E to questions 7 and 14 are interpreted as Mark III and Mark II emotion level 4 respectively. Inter-observer agreement was evaluated using percentage agreement and Cohen s kappa, a chance corrected measure of agreement for categorical responses. 16,17 No value of kappa (κ) is regarded as indicating good agreement although, following the guidelines of Altman, κ > 0.60 was interpreted as good agreement and κ = as moderate agreement. 18,19 T-tests for independent groups have been used to determine differences in selected attribute scores across and within groups over a period of time (for example, home completion of questionnaires compared with clinic scores). T-tests for single attributes defined by the Mark II Health Utilities Index can only be completed as numerical preference data is not yet available for Mark III scores. Hence vision, speech and hearing are condensed into sensation and ambulation and dexterity into mobility, whilst emotion, cognition and pain remain unchanged. 15 Results Response rates and patient characteristics All eligible families provided written consent; however, one family did not return any questionnaires and failed to attend repeated clinic appointments. Thirty-three (89%) parents and 19 (90%) of the 21 patients aged 10 years or more completed questionnaires at home. Seven (19%) patients failed to attend clinic during the study period despite repeat appointments being made (five of these returned their postal questionnaires). All parents and 28 (93%) patients attending clinic completed a questionnaire. The play- Table 1. Characteristics of subjects and treatment summary Parents completing questionnaire at home Parents completing questionnaire in clinic Patients completing questionnaire at home* Patients completing questionnaire in clinic (n = 33) (n = 30) (n = 19) (n = 28) Sex Male Female Age at diagnosis 1 13 yrs 1 13 yrs 3 13 yrs 1 13 yrs Mean Standard deviation CI Age at assessment 5 16 yrs 6 16 yrs yrs 6 16 yrs Mean Standard deviation CI Time from diagnosis to assessment 1 10 yrs 1 10 yrs 1 10 yrs 1 10 yrs Mean Standard deviation CI Chemotherapy 11 (33%) 10 (33%) 5 (26%) 10 (36%) Radiotherapy 11 (33%) 11 (37%) 7 (37%) 11 (39%) Surgery 33 (100%) 30 (100%) 19 (100%) 28 (100%) * Fewer patients completed questionnaires at home than in clinic as those under 10 years of age were excluded from home completion. Those patients aged between 5 9 years inclusive completed questionnaires with the play-specialist in clinic. ** CI = 95% confidence interval. Quality of Life Research. Vol

4 A. W. Glaser et al. specialist was unable to see the two patients who did not comply in clinic. A doctor completed questionnaires on 27 (90%) patients and physiotherapy assessments were completed on all 30 (100%) clinic attenders. Demographic and treatment details of all patients and parents completing questionnaires at home and in clinic are shown in Table 1. No significant differences existed between groups although patients participating at home were older than the other groups. This is due to our exclusion of patients aged under 10 years who were considered unlikely to have adequate language and grammar skills for correct understanding and interpretation of the questionnaire. Home vs. clinic Table 2 presents the responses to individual questions from the patients and parents who completed the assessment both at home and in clinic. Percentage agreement and κ-values for reliability of the 17 questions are shown in Table 3. The changes in number of each group responding with normal scores is shown in Figure 1. No events were reported by parents or patients that may have changed the responses to Figure 1. Home vs. clinic assessment: parents and patients questions. Despite this, considerable variation between responses given at home and in clinic was seen with percentage agreements of between % and κ- values ranging from Considerable variation in responses assumed to represent abnormal functional capacity was found and this is reflected in the poor percentage agreement and κ-values, whereas responses signifying normal levels of function for each attribute (questions 1 15 answer a, questions 16 and 17 answer a or b ) showed little difference across time. Higher κ-estimates of test retest agreement by attribute for both home and clinic completion were found when responses were grouped as normal or abnormal (Table 4). All κ-values were greater than or equal to 0.6, with the exception of pain and self-esteem, and percentage agreements were greater than 75% (excluding pain). Inter-observer agreement Patients and parents at home and in clinic. Table 5 shows the κ-estimates of agreement between patients and their parents by Mark III attribute at home and in clinic. Higher κ-values were obtained for all attributes following home completion with the most marked differences being for speech, hearing and emotion. With the exception of cognition, confidence for the future and self-esteem all κ-values at home were greater than No significant differences existed for Mark II attribute scores between parents and patients completing at home vs. in clinic, or between parents and patients when answering questionnaires either at home or in clinic. Clinic all observers. Figures 2a and 2b demonstrate inter-observer agreement by Mark III attribute. No significant difference existed between the Mark II or III attribute scores as assessed by the patients, parents, doctors and physiotherapists. Table 6 gives the κ-estimates and percentage agreement between observers in clinic for Mark III attributes in addition to Karnofsky and Lansky scores between doctors and physiotherapists. Greater than 80% agreement between the physiotherapists and the doctors assessments existed with the exception of emotion (58%), cognition (62%) and pain (76%). Similar patterns of agreement between the doctors ratings and those of parents and patients were found. Less agreement between physiotherapists and both parents and patients was documented for ambulation and dexterity compared to that between the doctors and these groups. 46 Quality of Life Research. Vol

5 Table 2. Responses of parents and patients at home and clinic, to individual questions Attribute Question Parent home Parent clinic Patient home Patient clinic n % n % n % n % Vision 1 A 22 85% 22 88% 14 88% 12 75% B 4 15% 3 12% 2 12% 4 25% Vision 2 A 23 88% 22 85% 14 88% 13 86% B 3 12% 3 12% 2 12% 1 7% C 0 1 4% 0 1 7% Hearing 3 A 25 96% 24 92% 15 94% 14 93% B 0 1 4% 1 6% 1 7% C 1 4% 1 4% 0 0 Hearing 4 A 24 96% 19 76% 15 94% 15 94% B 1 4% 6 24% 1 6% 1 6% Speech 5 A 18 72% 19 76% 13 81% 13 81% B 7 28% 6 24% 3 19% 3 19% Speech 6 A 20 77% 22 85% 11 69% 14 93% B 6 23% 4 15% 5 31% 1 7% Emotion 7 A 17 63% 16 62% 7 44% 9 56% B 8 30% 7 27% 7 44% 5 31% C 1 3% 3 11% 2 12% 2 13% D 1 3% Pain 8 A 12 48% 14 56% 6 40% 10 71% B 6 24% 7 28% 4 27% 3 21% C 7 28% 4 16% 5 33% 1 7% Ambulation 9 A 16 64% 17 65% 8 53% 10 67% B 5 20% 5 19% 5 33% 4 27% C 2 8% 2 8% 1 7% 0 D 2 8% 1 4% 1 7% 0 E 0 1 4% 0 1 6% Dexterity 10 A 16 62% 18 69% 10 63% 11 68% B 5 19% 4 15% 2 12% 1 6% C 1 4% 1 4% 0 1 6% D 3 11% 2 8% 2 12% 1 6% E % 1 6% F 1 4% 1 4% 1 6% 1 6% Cognition 11 A 16 62% 18 69% 9 56% 9 56% B 8 31% 7 27% 6 37% 6 37% C 1 3% % D 1 3% 1 4% 1 6% 0 Cognition 12 A 13 52% 15 55% 7 44% 8 50% B 6 24% 9 33% 4 25% 6 38% C 4 16% 2 7% 4 25% 1 6% D 2 8% 1 4% 1 6% 1 6% Self-care 13 A 15 60% 16 64% 11 73% 11 69% B 7 28% 8 32% 3 20% 4 25% C 1 4% D 2 8% 1 4% 1 7% 1 6% Emotion 14 A 11 44% 13 52% 7 44% 7 44% B 10 40% 10 40% 6 37% 7 44% C 3 12% 2 8% 2 12% 1 6% D 1 4% 0 1 6% 1 6% Pain 15 A 10 38% 13 48% 6 37% 8 53% B 15 58% 12 44% 9 56% 5 33% C 1 4% 2 8% 1 6% 2 13% Confidence for future Health status assessment in children 16 A 11 46% 13 52% 3 20% 7 47% B 10 42% 9 36% 9 60% 6 40% C 3 12% 3 12% 3 20% 2 13% Self-esteem 17 A 2 8% 7 27% 5 31% 6 40% B 16 62% 11 42% 4 25% 2 13% C 5 19% 6 23% 5 31% 5 33% D 2 8% 2 7% 2 13% 2 13% E 1 4% Quality of Life Research. Vol

6 A. W. Glaser et al. Table 3. HRQOL questionnaire test retest reliability: home vs. clinic completion Attribute Ques. no. Parent Patient %* κ %* κ Vision 1 96% % 1.0 Vision 2 96% % 1.0 Hearing 3 96% % 0.34 Hearing 4 48% % 0.17 Speech 5 52% % 0.17 Speech 6 64% % 0.18 Emotion 7 52% % 0.08 Pain 8 52% % 0.14 Ambulation 9 12% % 0.10 Dexterity 10 85% % 0.67 Cognition 11 84% % 0.45 Cognition 12 58% % 0.24 Self-care 13 62% % 0.34 Emotion 14 68% % 0.15 Pain 15 64% % 0.45 Confidence 16 46% % 0.47 for the future Self-esteem 17 48% % 0.35 Table 4. Kappa values for agreement by Mark III attribute between home and clinic, assuming either normal or abnormal responses Attribute Patient Parent κ %* k %* Vision % % Speech % % Hearing % % Ambulation % % Dexterity % % Emotion % % Pain % % Cognition % % Confidence % % for future Self-esteem % % * Percentage agreement * Percentage agreement Figure 2a. Inter-observer agreement by attribute Figure 2b. Inter-observer agreement by attribute 48 Quality of Life Research. Vol

7 Health status assessment in children Assuming a Lansky and Karnofsky score of 90 or 100 to be normal, less agreement was found between Lansky ratings (κ = 0.14, percentage agreement = 90%) than Karnofsky scores (κ = 0.62, percentage agreement = 76%). Comparisons of Karnofsky and Lansky score ranges between raters are shown in Figure 3. This is the first reported study to examine the effect of mode of administration and inter-observer agreement on responses to the Mark III Health Utility Index, a generic health status assessment, when applied to children. Additionally questions relating to self-esteem and confidence for the future have been used and compared with the more traditional tools used in clinical trials as outcome measures. Discussion Mode of administration Completion of the HRQOL questionnaire at home and clinic produced results which did not statistically Table 5. Kappa estimates and percentage agreement between patients and parents at home and in clinic, assuming either abnormal or normal results Figure 3. Lansky and Karnofsky scores, assessed by doctor and physiotherapist. Error bars represent standard error Attribute Home Clinic κ %* k %* Vision % % Speech % % Hearing % % Ambulation % % Dexterity % % Emotion % % Pain % % Cognition % % Confidence % % for future Self-esteem % % * percentage agreement Table 6. Kappa and percentage agreement to demonstrate level of agreement between observers in clinic, assuming either abnormal or normal results for Mark III attributes, Lansky, Karnofsky, Self-esteem and Confidence for the Future Attribute Doctor/patient Doctor/parent Doctor/physio Physio/patient Physio/parent Patient/parent κ %* k %* k %* k %* k %* k %* Vision % % % % % % Speech % % % % % % Hearing % % % % % % Ambulation % % % % % % Dexterity % % % % % % Emotion % % % % % % Pain % % % % % % Cognition % % % % % % Confidence % % % % % % for future Self-esteem % % % % % % Karnofsky % score Lansky score % * percentage agreement; Physio = Physiotherapist Quality of Life Research. Vol

8 A. W. Glaser et al. differ. However a wide range of agreement was demonstrated in the absence of any attributable life-events. This may be indicative of poor test retest reliability of the instrument or the effect of the setting on the individual s response patterns. The reliability of the Mark III Health Utility Index has been assessed following its use in the 1991 Canadian General Social Survey. 20 Substantial reliability for vision, ambulation and emotion were reported with moderate reliability for hearing, cognition plus pain and poor reliability for speech and dexterity. 16 However the sample and methodology for this study differed from ours, it being composed of Canadians aged 15+ years and the information being derived by telephone interviews as opposed to home and clinic self-completion. These results demonstrate good agreement (κ = ) for all attributes, when assessed at home and in clinic, between patients and their parents except for vision and pain (patients) and self-esteem (parents) when moderate agreement (κ = ) was found. Additionally, percentage agreement was greater than 75% for all attributes except pain (patients 71% and parents 61%). No significant differences between single attribute scores generated at home and in clinic were demonstrated. Despite no statistical differences between single attribute scores of parents compared to those of the patients either at home or in clinic, higher levels of percentage agreement and κ-values were obtained when questionnaires were completed at home. This may reflect the exclusion of children under ten years of age from completing questionnaires at home. In clinic, the patients and their parents were taken to separate areas of the weighting area to complete questionnaires. No such separation could be enforced during home completion and thus increased collusion between respondents at home may simply explain this finding. In clinic, parents and patients were asked about difficulties experienced during questionnaire completion. None were reported during home or clinic completion and no participant phoned the 24-hour telephone enquiry service. This assessment was made by one investigator (AWG) and is therefore subjective. However, the high completion rates at home, agreement of all participants to complete second questionnaires in clinic and the higher participation in questionnaire completion than clinic attendance would all support the conclusion that they were acceptable and easy to use. Clinic attendance may be stressful and intimidating for patients and their families for a multitude of reasons not least returning to the hospital which is associated with previously distressing experiences. It appears that the HRQOL questionnaire is acceptable for home use with higher compliance obtained in this study for postal questionnaires than clinic attendance. Completion prior to clinic reviews may help to focus both the families and the physicians minds on problem areas during the consultation. Additionally malignancies in childhood are frequently treated at specialist centres geographically distant from the families residence. Completion of this questionnaire as a screening instrument between clinic visits may enable clinic appointments to be less frequent. Inter-observer agreement Proxy respondents are necessary when assessing HRQOL in childhood as no instrument exists that is simple, reliable, valid and comprehensible for young children. 21 Doubt remains as to the ability of year olds to correctly interpret and complete the questionnaires used in this study although the high level of agreement demonstrated between this group and their parents would suggest that this is unfounded. Our results agree with previous findings that proxies who have the greatest contact with patients (the parents) respond most comparably with the patients, yet when they do disagree, this group overestimate patient disability. 22 Other studies have shown children to report more problems related to emotional distress whilst their parents report more associated with observable, behavioural reactions. 23 No differences in the mean single attribute scores between observers was found although there was a wide range of κ-estimates of agreement. Doctors and physiotherapists identified the same or fewer deficits in single attributes than did patients and parents although this did not reach statistical significance. This is in keeping with our previous experience with the Mark II Health Utilities Index. 8 Poor κ-estimates of agreement were demonstrated between doctors and physiotherapists assessment of the more subjective attributes: emotion, pain, cognition and confidence for the future. However good agreement for the more objective domains (vision, hearing, speech, ambulation and dexterity) was seen. Interestingly higher κ-estimates for the Lansky score were found than for the Karnofsky, yet a higher percentage agreement for Karnofsky scores were seen. These results may well reflect the increased sensitivity of the Lansky score which has been designed to be specific to a child s play-performance as opposed to the Karnofsky scale which is an assessment of global functional status in adults. 50 Quality of Life Research. Vol

9 Health status assessment in children Study population Five terminally ill patients were excluded as it was felt unlikely that they would be well enough to attend the appropriate clinics (to enhance quality of life we attempt to perform as much terminal care as possible at home rather than in hospital). All five died during the study period. Their exclusion may have introduced a bias such that morbidity has been underestimated in this population, although the aim of this project was to assess methodological issues rather than actually quantify health status and levels of morbidity. The latter would require a larger sample and reflects one of the difficulties of single centre studies investigating rare conditions. The overall health status and morbidity in survivors of CNS tumours in childhood will therefore be assessed in collaboration with two other centres and this will be reported separately. However, for studies comparing inter-observer agreement, multi-centre trials have their own difficulties including the varied training and background of observers. Conclusion The HRQOL questionnaire incorporating the Mark II and III Health Utilities Index was both easy to use and acceptable to parents, patients, doctors and physiotherapists. A wide range of agreement between home and clinic responses was found. This may reflect poor reliability of the questionnaire or the effect of the mode of administration. Clinics are frequently noisy, stressful environments which are associated with previous bad news and severe stress. Completion of questionnaires at home in familiar surroundings may provide a more accurate reflection of the individual s actual day-to-day level of performance and functioning. The variation in responses seems to be confined to those individuals with scores reflecting abnormal or impaired function whilst the scores for those with normal function remained constant across time. This may be indicative of the difficulties in assessing individuals with impairment compared to those functioning normally. The instrument is of value in defining those individuals functioning normally, yet its ability to monitor changes in abnormal function, as would be necessary in a longitudinal study, needs to be further examined. The widely used Lansky and Karnofsky scores were less sensitive than the HRQOL questionnaire and provided less information about the patients health status. In view of these results and as information regarding reliability and other methodological issues become available, the more comprehensive assessments of health status should be incorporated into clinical trials in place of the more traditional instruments. References 1. Stiller C, Bunch K. Brain and spinal tumours in children under 2 years: incidence and survival in Britain, Br J Cancer 1992; 66(Suppl 18): Lashford L, Campbell R, Gattamaneni HR, Robinson K, Walker D, Bailey C. An intensive multiagent chemotherapy regimen for brain tumours occurring in very young children. Arch Dis Child 1996; 74: Glaser A, Walker D. Quality of life. Lancet 1995; 346: Jenny MEM, Kane RL, Lune N. Developing a measure of health outcomes in survivors of childhood cancer: A review of the issues. Med Pediatric Oncol 1995; 24: Clarridge BR, Massagli MP. The use of female spouse proxies in common symptom reporting. Med Care 1989; 27: Herjanic B, Reich W. Development of a structured psychiatric interview for children: agreement between child and parent on individual symptoms. J Abnormal Psychol 1982; 10: Ontario Ministry of Health. Ontario Health Survey, 1990, User s Guide. Vol. 1: Documentation. Toronto, Canada: Ontario Ministry of Health Grootendorst P, Feeny D, Furlong W. Does it matter whom and how you ask? A technical report on inter- and intrarater agreement in the Ontario Health Survey. Centre for Health Economics and Policy Analysis Working Paper Hamilton, Ontario: McMaster University, Barr RD, Pai MR, Weitzman S, et al. A multi-attribute approach to health status measurement and clinical management illustrated by an application to brain tumors in childhood. Int J Oncol 1994; 4: Kanabar DJ, Attard-Montalto S, Saha V, Kingston JE, Malpas JE, Eden OB. Quality of life in survivors of childhood cancer after megatherapy with autologous bone marrow rescue. Pediatric Hematol Oncol 1995; 12: Torrance GW, Furlong W, Feeny D, Boyle M. Multiattribute Preference Health Utility Index. Pharmacoeconomics 1995; 7: Bilson A, Walker DA. Assessment of health status in survivors of cancer. Arch Dis Child 1994; 70: Lansky SB, List MA, Lansky LL, Ritter-Sterr C, Miller DR. The measurement of performance in childhood cancer patients. Cancer 1987; 60: Karnofsky DA, Burchenal JH. The clinical evaluation of chemotherapeutics in cancer. In: MacLoed CM, ed. Evaluation of chemotherapeutic agents. New York, NY: Columbia University Press, 1949: Torrance GW, Zhang Y, Feeny D. Multi-attribute Preference Functions for a Comprehensive Health Status Classification System. Hamilton, Ontario: McMaster University, Centre for Health Economics and Policy Analysis, Boyle M, Furlong W, Feeny D, Torrance G, Hatcher J. Reliability of the Health Utilities Index-mark III used in Quality of Life Research. Vol

10 A. W. Glaser et al. the 1991 cycle 6 Canadian General Social Survey Health Questionnaire. Qual Life Res 1995; 4: Cohen J. A co-efficient of agreement for nominal scales. Educational and Psychological Measurement 1960; 20: Altman D. Some common problems in medical research. In: Practical Statistics for Medical Research. London: Chapman and Hall, 1991: Landis RJ, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: Statistics Canada. General Social Survey: Health, 1991: Public Use Microdata File Documentation and User s Guide Goodwin DAJ, Boggs SR, Graham-Pole J. Development and validation of the pediatric oncology quality of life scale. Psychological Assessment 1994; 6: Magaziner J, Simonsick E, Kashner M, Hebel J. Patient proxy response comparability on measures of patient health and functional status. J Clin Epidemiol 1988; 41: Stone WL, Lemanek KL. Developmental issues in children s self-reports. In: La Greca AM, ed. Through the Eyes of the Child: Obtaining Self-reports from Children and Adolescents. Boston: Allyn and Bacon, (Received 20 April 1996; accepted 16 August 1996) 52 Quality of Life Research. Vol

11 Health status assessment in children Appendix Questions to assess self-esteem and confidence for the future Self-esteem Which one of the following describes how you see yourself? a. Excellent, as if on top of the world b. Very good, but a few problems c. All right d. Not good, lots of problems e. Disastrous, could not be worse Confidence for the future Which one of the following best describes the way you see your future? a. Extremely confident and positive b. Confident but occasional doubts c. Gloomy with occasional hope d. Very gloomy and dark with no hope These questions follow a similar format to the Mark III Health Utility Index. They were designed following extensive discussion with patients, parents, Malcolm Sargent social workers, paediatric nurses and oncologists. Quality of Life Research. Vol