Reduced health-related quality of life among Hodgkin's disease survivors: A comparative study with general population norms

Transcription

1 Annals of Oncology 10: 7177, Kluwer Academic Publishers. Primed in the Netherlands. Original article Reduced healthrelated quality of life among Hodgkin's disease survivors: A comparative study with general population norms J. H. Loge, 1 ' 4 A. Foss Abrahamsen, 2 0. Ekeberg 1 & S. Kaasa 3 ' 4 1 Department of Behavioural Sciences in Medicine, 2 The Norwegian Radium Hospital, University of Oslo; 3 Palliative Medicine Unit, Department of Oncology and Radiotherapy, Trondheim University Hospital; *Unit for Applied Clinical Research, The Norwegian University for Science and Technology, Trondheim, Norway Summary Background: Late complications after curative treatment of Hodgkin's disease are of special relevance because most of the cured are young adults. The aims of the present study were: (1) to compare healthrelated quality of life (HRQOL) in Hodgkin's disease (HD) ' urvivors with normative data from the general Norwegian population and (2) to examine the relations between disease/treatment characteristics and HRQOL in the HD survivors. Patients and methods: 459 HD survivors aged 1974 years (mean 44.0, SD 11.8) treated at the Norwegian Radium Hospital were approached in 1994 and compared to norms from 2214 subjects approached in The norms are representative of the general Norwegian population. HRQOL was assessed by the Short Form 36 (SF36), which measures HRQOL in eight separate scales (0 =.worst health state, 100 = best health state). Results: The HD survivors had lower scores than the normal controls on all scales after adjustment for age, gender and educational levels. Statistically significant differences (JP < 0.01) were found in general health (10.4), physical functioning (6.1), role limitations (physical, 9.3), social functioning (3.6) and in vitality (4.7). Patients with disease stage IBIIB had the lowest scores on all scales. The differences in relation to stage/substage reached statistical significance (P < 0.01) in physical functioning and in role limitations (physical). Time since diagnosis, types of primary treatment or having relapsed were not associated with statistically significant differences in HRQOL. Conclusion: Longterm HD survivors have poorer HRQOL, primarily in physical health, than the general Norwegian population. Key words: health related quality of life, Hodgkin's disease, normative data, questionnaires (SF36) Introduction Hodgkin's disease (HD) was a nearly fatal disease until successful treatment with irradiation was established in the 1960s. Throughout the 1960s and 1970s, the treatment was further improved by the introduction of chemotherapy as primary treatment or in combination with irradiation. The cure rate is now as high as 90%95% among subjects with localised disease [1]. In addition to improved diagnostic techniques and refined treatment, the improved survival rate has partly been achieved through more aggressive treatment. The concerns for the consequences of cure have therefore increased. Most of the affected are young adults [2], and the costs of survival in terms of health problems, family and labour adjustment are therefore of special relevance. Good documentation upon late effects is necessary for adequate rehabilitation programmes and may also have implications for the choice of primary treatment. Late medical complications among HD survivors such as secondary cancers, chronic pericarditis, lung fibrosis and hypothyroidism among subjects treated with irradiation in mantle fields are well documented [3, 4]. Less is known about social consequences and the survivors' healthrelated quality of life (HRQOL) after curative treatment. Reports indicate that most HD survivors resume work [5, 6]. However, many HD survivors report work as exhausting with reduced energy left for leisure activities [7]. The survivors' social functioning in general may therefore be affected. The disease, the diagnostic procedures, the toxic treatment regimens or the late medical complications may have effects upon different dimensions of the HRQOL of the survivors. For example, reduced heart or lung function after irradiation or chemotherapy [4, 8] may have implications for physical functioning or role performance. We have previously reported high levels of anxiety and fatigue in a Norwegian cohort of HD survivors [9, 10]. Previous studies of HRQOL among HD survivors have demonstrated that they suffer from fatigue and lacking return of energy [5,11], reduced cognitive functioning [11], increased levels of psychological distress [12], reduced physical functioning [11, 13] and restrictions in role functioning [11, 13]. However, different measures of HRQOL were employed in the different studies, render Downloaded from

2 72 ing direct comparisons of the results difficult. Further, in some of the studies the survivors had been recruited for clinical trials [5] or been treated for advanced disease [12], possibly limiting the generalisability of the findings. Two of the samples were also relatively small (< 100 subjects) [11, 13], and clinically relevant differences in HRQOL may be undetected since HRQOL measures have substantial variance. Comparison groups were only included in two of the studies [11, 13]. One of these used hospital visitors [13], and the other used matched (age, gender and city of residence) controls from the same region of France [11]. The selection of the comparison groups may therefore have affected the generalisability of the findings [11]. The interpretation of HRQOL scores may be problematic, and implies answering to questions such as: What is a high/low score? What is the clinical relevance of a score? Normbased interpretation is one method to decide whether an observed score is typical or not. Norms are calculated based on a sample of the general population, and normbased interpretations imply calculating departures from the norm [14]. This also allows for interpretation of the clinical meaningfulness of an observed score [14]. The aims of the present study were: (1) to compare HRQOL in an unselected and relatively large sample of HD survivors with norms from the general Norwegian population, and (2) to examine if characteristics of the disease or the treatment were related to differences in HRQOL among the HD survivors. Patients and methods Study samples The study included two samples, HD survivors treated at the Norwegian Radium Hospital and general population controls. Both samples were mailed selfreport questionnaires. Subjects who had not responded after the first mailing received one written reminder. HD survivors In the 1970s, Norwegian adult HD patients, except some of the oldest with poor prognosis were treated at the Norwegian Radium Hospital [2]. Since the 1980s, the Norwegian Radium Hospital has treated about 40% of Norwegian adult HD patients, all coming from a defined region of Norway [2]. Subjects were included in the present study according to the following criteria: Admittance to the Norwegian Radium Hospital for HD in the period of 1971 to 1991, alive by the end of 1993, age between 15 and 61 years at time of diagnosis and 74 years or less by the end of In total, 557 HD survivors were approached in May 1994, and 459 (82%) replied. More females (86%) than males (80%) responded {P 0.05). General population controls The Norwegian Government Computer Centre performed a random draw from The National Register of Norway including all Norwegian inhabitants aged 1980 years with the same last digit in their social security number. The draw should reflect the general Norwegian population in terms of gender and agedistributions. In total, 3500 subjects were drawn out and approached in May 1996, and 2323 (67%) responded [15]. All respondents 74 years or younger (n = 2214) were included in the present study. Measures Short Form 36 (SF36) The SF36 is a widely used generic (i.e. not specific to any disease or population) HRQOLmeasure [16]. The Norwegian translation of the standard version was used in the present study [17]. The 36 items in the SF36 are grouped into eight scales: bodily pain, general health perceptions (= general health), mental health, physical functioning, role limitations due to emotional problems (= role limitations, emotional), role limitations due to physical problems (= role limitations, physical), social functioning and vitality. An additional item reports health transition over the last year (= health transition). The response choices in the rolefunctioning scales (physical and emotional) are dichotomous ('yes/no') The other items have three to six response choices. After coding the raw scores, 10 items were recalibrated following standard guidelines [18]. The items were then summed and transformed to the eight 0100 scales (0 = worst health state, 100 = best health state) [18]. Missing values in half or less of the items within a scale were substituted with personspecific values according to the SF36 algorithm [18]. Diseases and health problems Information on date of diagnosis, stage, histology, treatment and current disease status (relapsed or not) among the HD survivors was based on the hospital records. The controls were asked about their health condition. They answered to questions about five common diseases and nine common health problems. They were asked if they suffered or had suffered from the following diseases: hypertension, myocardial infarction, heart failure, cancer or diabetes. Further, they were asked if they at present were experiencing any of the following health problems: chronic allergy, arthritis, back pain, visual impairments, chronic skin problems, chronic lung problems, deafness or hearing problems, functional impairment in leg or arm or other chronic health problems. Based on the answers to these questions, the controls were divided into four groups (1: no disease or health problem, 2: disease but no present health problem, 3: no disease but present health problem, 4: disease and present health problem) [15]. Sociodemographic items The data files from NRH and The National Register supported information on age and gender for the HD survivors and the controls respectively. Educational and marital status were assessed by the questionnaires. Analysis Bivariate analyses were performed by /tests (continuous variables), Oneway ANOVAs with Scheffe's correction for multiple comparisons (continuous variables) and chisquare statistics (categorical variables). Multivariate analyses were performed by ANOVAs with multiple classification analysis (MCA). Fn the analysis adjusting for age, age was entered as a continuous covariate before the main effects. The scalescores were standardised in the graphical presentation In each group, the mean scores were computed by ANOVAs with MCA and adjusted for age, gender and educational level. A deviation score was calculated by subtracting the scores for the HD survivors and the three groups reporting diseases/health problems from the scores of controls not reporting health problems. Finally, the deviations were divided by each scale's SD in the total sample. Due to the multiple comparisons with increasing risk of type I errors, the level of statistical significance was set at The data were analysed using the SPSS for Windows v 6.1 software (SPSS Inc., IL, USA). Downloaded from

3 73 Table 1. Subject characteristics. Age at time of study (years, mean (SD)) Age groups, n (%) 1929 years 3039 years 4049 years 5059 years 6074 years Gender, n (%) Male Female Educational status, n (%) 10 years or less 11 years or more, not university University Stage/substage b, n (%) Stage IAIIA Stage IBIIB Stage IHAIVA Stage IIIBIVB Primary treatment 0, n (%) Irradiation Chemotherapy Irradiation + chemotherapy Health condition, n (%) No disease/present health problem Past or present disease only Present health problem only Disease and present health problem Hodgkin disease survivors (n = 459) 44(11.8) 50(11%) 117(26%) 169(37%) 67(15%) 56(12%) 255 (56%) 204 (44%) 168(37%) 173(38%) 117(26%) 224 (49%) 54(12%) 88 (19%) 93 (20%) 174(38%) 66(14%) 217(47%) General P* population controls (n = 2214) 45(16.5) NS 510(23%) 487 (22%) 446 (20%) 363 (16%) 408 (18%) (49%) 1127(51%) 558 (25%) 1005(46%) 635 (29%) 870 (40%) 165 (8%) 835 (38%) 335(15%) Abbreviation: NS not statistically significant differences between HDS and controls. a Significance for difference in proportions by Chisquare statistics or in means by Mest. b At time of diagnosis according to Ann Arbor staging classification. c Two patients had only surgery without radiotherapy or chemotherapy. Results Characteristics of the samples The characteristics of the respondents in the two samples are presented in Table 1. The gender and agedistributions and the educational status differed between the HD survivors and the controls. More of the HD survivors were males, aged 40 to 49 years and had education at the lowest level. Marital status did not differ between the samples after adjusting for age and gender. Mean time since diagnosis among the HD survivors was 12.2 years (SD = 5.5, range 323 years). Patients with disease stage I and II without risk factors for relapse had been treated by irradiation. Since 1980, patients with risk factors for relapse have also received four cycles of chemotherapy before irradiation. Stage III patients received total nodal irradiation before 1980 and eight cycles of chemotherapy after Stage IV patients received eight cycles of chemotherapy. After 1980, stage III and IV patients treated by chemotherapy also received involved field irradiation to bulky or residual mass. Among the 391 subjects treated by irradiation, 76% had received mantle field or mantle field and inverted Yfield. MOPPlike chemotherapy regimens, MVPP (mustine, vinblastine, procarbazine and prednisone) and LVPP (chlorambucil, vinblastine, procarbazine and prednisone) had been administered to 54% of the 283 subjects receiving chemotherapy. Chemotherapy regimens containing bleomycin and anthracyclines, ABOD (doxorubicin, bleomycin, vincristine and dacarbazine) administered solely or alternating with LVPP and EBVP (epirubicin, bleomycin, vinblastine and prednisone) had been administered to 131 of subjects receiving chemotherapy. Fiftyseven (12%) of the HD survivors had experienced a relapse, seven within the two years preceding the present study. No statistically significant relations between relapse and stage of the disease were found. Data quality More controls (79%) than HD survivors (72%) had complete SF36 questionnaires (P = 0.002). After substituting the missing values, the percentage of complete SF36 scales ranged from 92% (role limitations, emotional) to 98% (social functioning) (HD survivors) and from 95% (role limitations, emotional) to 99.5% (social functioning, controls). HRQOL among the HD survivors and the general population controls In both samples there were differences in scale scores in relation to age, gender and educational level. For the controls, this has been described in the norm study [15]. Among the HD survivors, the women had statistically significantly lower scores in physical functioning (difference = 7.8), in role limitations (emotional, difference = 10.1) and in role limitations (physical, difference = 10.8) than the males. On all the 'physical scales' (bodily pain, physical functioning, general health and role limitations, physical) and in health transition there were statistically significant differences in relation to age. Subjects sixty years or older had the lowest scores on all these scales. After adjustment for age and gender, the differences in relation to educational level reached statistical significance in the physical functioning and in the role limitations (physical) scales. The best educated had the highest scores and the poorest educated the lowest scores (physical functioning: 83.9 versus 77.7, role limitations (physical): 76.6 versus 62.4). The HD survivors had lower scores than the controls on all scales. After adjustment for age, gender and educational levels, the differences between the HD survivors and the normal controls reached statistical significance in general health, physical functioning, role limitations (physical), social functioning and in vitality scales (Table 2). Downloaded from

4 74 Table 2 SF36 scalescores for the HD survivors and the normal controls (crude mean scores (SD) and scores adjusted for age, gender and educational status). Scales Subjects status P value Effect size HD survivors Normal controls Bodily pain General health Health transition Mental health Physical functioning Role limitations (emotional) Role limitations (physical) Social functioning Vitality Crude (SD) 73.5 (27.8) 66.1 (24.8) 58.5 (21.6) 78.1 (17.0) 81.7 (20.9) 78.5 (31.0) 69.6 (38.0) 82.3 (25.0) 55.5 (21 3) Crude (SD) 75.7 (25.6) 77.4 (21.8) 60.2 (16.9) 78.8 (16.4) 88.4 (17.4) 82.5 (31.8) 79.7 (34 4) 86 1 (21.8) 60.3 (20.7) NS NS NS NS < Abbreviation: NS not statistically significant differences between the HD survivors and the controls. * /"value for differences between the HD survivors and the controls by ANOVAs b Mean differences divided by the standard deviations of controls. The magnitudes of the differences were 10.4 (general health), 9.3 (role limitations.physical), 6.1 (physical functioning), 4.7 (vitality) and 3.6 (social functioning) respectively. The controls included subjects with different health conditions, and the HD survivors were also compared to the controls not reporting diseases or present health problems (n = 870). The comparisons were conducted after adjustment for age, gender and educational level. The HD survivors had statistically significant lower scores on all scales compared to the controls not reporting diseases or health problems. The differences between the HD survivors and these controls ranged from 20.9 (role limitations (physical)) to 3.1 (health transition). Other differences exceeding 10 points were found in general health (20.4), physical functioning (11.6) and in vitality (10.5). The smallest differences were found in health transition (3.1) and in mental health (3.9). In Figure 1 the scores of the subjects not reporting health problems were set at 0. The standardised scores (sscores) of the three other groups of the normal controls, i.e. subjects reporting past or present disease (PD), subjects reporting present health problems (PP) and subjects reporting past or present disease and present health problems (PD/PP) represent these groups' health profiles. For example, 40% of the PP reported back pain as their present health problem. This is reflected in the group's low BPscore. Except for the BPscale, the scores of the HD survivors were lower than the scores of the subjects reporting present health problems, but Figure 1. Standardised scores for the HD survivors (HDS) and the normal controls reporting past or present disease (PD), present health problem (PP) and past/present disease and present health problem (PD/PP) (0 = controls not reporting diseases or health problems) (adjusted for age, gender and educational level). Abreviations: BP bodily pain; GH general health; HT health transition; MH mental health; PF physical functioning; RE role limitations (emotional); RP role limitations (physical); SF social functioning; Vt vitality. higher than the scores of the subjects in poorest health (PD/PP). On all scales except for the MHscale, the HD survivors had lower scores than the subjects only reporting past or present disease. The lowest sscore among the HD survivors was found in general health (0.88). Other sscores lower than 0.50 were found in physical functioning (0.63), in role limitations (physical) (0.58) and in vitality (0.52). Impact of disease and treatment factors There were statistically significant differences in physical functioning and in role limitations (physical) in relation to stage/substage after adjustment for age, gender and educational level (Table 3). Subjects with stage IBIIB had the lowest adjusted scores on both scales (physical functioning: 70.9, role limitations (physical): 55.8). Subjects with disease stage/substage IBIIB also had the lowest scores in bodily pain and in vitality (62.5 and 47.6 respectively) (0.05 > P > 0.01). In relation to type of primary treatment (irradiation, chemotherapy or combined treatment), subjects treated with chemotherapy had the highest scores on all scales, but the differences in relation to type of primary treatment did not reach statistical significance after adjustment for age, gender and educational level. There was a close to significant difference in vitality (P = 0.035). Subjects treated with irradiation had the lowest adjusted score (53.6), and subjects treated with chemotherapy had the highest (61.7). The adjusted score of the subjects treated by the combined modality was The types of chemotherapy regimens were divided into MOPPlike regimens (MVPP and LVPP) and regimens containing anthracyclines and bleomycin (ABOD, ABOD alternating with LVPP and EBVP). After adjustment for age, gender and educational level, there were no statistically significant differences in any of the SF36 scales in relation to type of chemotherapy regimen or in relation to number of chemotherapy cycles. The combined treatment with irradiation and chemotherapy containing anthracyclines and bleomycin was not associated with reduced HRQOL compared to the combined Downloaded from

5 75 Table 3. SF36 scalescores for the HD survivors in relation to stage/substage (crude mean scores (SD) and scores adjusted for age, gender and educational status). Scales Stage/substage IAIIA (n = 223) IBIIB («= 54) IIIAIVA (n = 87) IIIBIVB(n = 93) Crude (SD) Crude (SD) Crude (SD) Crude (SD) BP GH HT MH PF RE RP SF Vt 75.4 (27.0) 67.5(25.1) 57.3 (20.7) 77.9(17.0) 84.4 (18.9) 79.0(31.1) 73.5(36.5) 82.7 (26.3) 56.1 (21.2) a a (31.5) 60.9 (26.9) 57.2(21.7) 76.3(18.7) 69.9 (26.0) 75.0 (30.9) 54.1 (41.8) 76.4 (27.6) 47.8(19.2) a a (25.9) 68.2 (23.2) 61.9(22.7) 81.1 (15.6) 84.2(18.8) 77.4 (32.0) 74.5 (33.0) 82.8 (23.8) 58.3 (22.2) " " (27.9) 63.2(24.1) 59.0 (22.9) 76.3(17.1) 79.9(21.5) 80.4 (30.3) 64.0(41.6) 84.2(21.2) 55.9(21.6) a " Abbreviations: BP bodily pain; GH general health; HT health transition; MH mental health; PF physical functioning; RE role limitations (emotional); RP role limitations (physical); SF social functioning; Vt vitality. ' P ^ 0.01 for differences between groups (stage/substage) by ANOVAs. treatment with irradiation and MOPPlike chemotherapy regimens. Having experienced a relapse or the time since diagnosis were not associated with statistically significant differences in any of the SF36 scales after adjustment for age, gender and educational level. Discussion This study demonstrates reduced HRQOL 323 years after curative treatment of Hodgkin's disease, primarily in physical health (i.e., physical functioning, role limitations due to physical problems and general health perceptions). The present sample of HD survivors is large, unselected and from one hospital. The compliance rate was excellent. SF36 is a welldocumented instrument translated into Norwegian according to an internationally accepted protocol [19]. This procedure allows for comparisons with studies from other countries, supported by the insignificant differences in SF36 scores across nations [20]. As shown in the norm study, the two role functioning scales functioned suboptimally, due to the significant roof and ceiling effects (i.e., many subjects achieving the highest or lowest possible scores), the most missing values and the lowest reliability estimates [15]. The results from these scales should therefore be interpreted most cautiously. The norms were collected two years after the HD survivors were surveyed, but at the same time of the year. If the health state of the population or the willingness to report are changed during such a period, the comparisons will be affected. The health state of the Norwegian population has been stable between 1994 and Since the SF36 does not include sensitive areas such as sexuality, we do not believe that willingness to report affects the responses in the surveys. The compliance rate in the norm study was high and at the same level as in other recent general population studies [21, 22]. The lowest compliance rates were found among women 70 years or older (53%) and among men 29 years younger (61%) [15]. The norms among the oldest (> 60 years) may have been slightly higher than the 'true' population values in the scales mostly effected by age (i.e. the physical scales) [15]. Since only 12% of the HD survivors were 60 years or older, possible differences between the norms and the 'true' population values are, in our opinion, negligible. Physical health Only one of the published studies of HRQOL among HD survivors has employed the SF36 [13], but the samples were considerably smaller (81 HD survivors and 114 controls). This study also demonstrated reduced physical functioning, increased role limitations due to physical problems and reduced general health. It is therefore noteworthy that the HD survivors in the present sample had close to similar scores in both physical functioning and in role limitations (physical). Both these scales are highly affected by age, and the mean ages of the HD survivors in the two studies were similar. In the general health scale the mean score of the present sample was 4.2 lower than that reported by van Tulder et al. [13]. Generally, different measures and different samples have limited the possibility to conduct comparisons of HRQOL data across studies. This is therefore the first opportunity to directly compare HRQOL measured by the same instrument in two different samples of HD survivors from different countries. Since both studies found HRQOL was affected on the same scales and to the same extent, we think this is a very strong indication of longterm reduced HRQOL after curative treatment for HD. The study of Joly et al. [11] employed the EORTC QLQ C30 questionnaire [23] and also demonstrated Downloaded from

6 76 reduced physical functioning and role performance compared to normal controls. The content of the physical functioning scales in the two instruments is quite comparable. The role performance scale in the EORTC QLQ C30 does not differentiate between limitations due to physical or emotional problems, and direct comparison is therefore restricted. The largest difference (10 points) between the HD survivors and the controls was found in general health, and the same difference was reported by Van Tulder et al. [13]. Joly et al. found no difference in global health status between the HD survivors and controls [11]. The general health scale in the SF36 differs from the global health status in the EORTC QLQ C30 (5 vs. 2 items), so direct comparison between the three studies is hindered. Since the general health scale primarily measures physical health [16], the findings imply that the HD survivors overall consider their physical health to be poorer than the controls. Our findings may of course reflect the experience of having had a potentially fatal disease, but the controls only reporting past or present diseases had a higher score on this scale. Physical functioning and role limitations (physical) were significantly lower in subjects with stages IBIIB (respectively 17.1 and 23.2 points lower than the controls). The IBIIB patients received the most intensive treatment, including four cycles of chemotherapy (containing bleomycinanthracyclines) prior to full irradiation. In a subcohort of the present sample, Lund et al. found this combination to predict lung function impairment [24]. These data were not available when the present study was conducted, but the combination of irradiation and bleomycinanthracyclines was not associated with reduced physical health. Generally, type of primary treatment was not associated with statistically significant differences in HRQOL. We have previously demonstrated that stage IBIIB is associated with increased levels of fatigue [10], and fatigue might explain the decreased physical health in subjects with stage IBIIB. Still, future research should include late medical complications as predictors of HRQOL in HD survivors. Mental health and fatigue Mental health among the HD survivors did not differ compared to the controls. Joly et al. [11] and van Tulder et al. [13] found the same. Kornblith et al. found psychological distress to be elevated one standard deviation (SD) above that of healthy subjects [12]. However, their HD survivors were recruited from subjects included in clinical trials for advanced HD. Compared to the healthy subjects in the present study, the HD survivors had slightly reduced mental health (0.25 SD). We have previously reported high levels of anxiety measured by the Hospital Anxiety and Depression Scale [9] in this cohort but without the use of a comparison group. Whether the mental health scale of the SF36 has sufficient sensitivity to detect minor differences in mental health is uncertain. However, the scale is sensitive to the presence and severity of depressive symptoms [25]. The possibility of higher levels of anxiety among the HD survivors can therefore not be excluded in spite of the similar scores of the HD survivors and the controls on the mental health scale. The HD survivors had decreased vitality (i.e. more fatigue) than the controls. Both the study of Fobair et al. [5], our study on fatigue in the same cohort [10] and Joly et al. [11] reported increased levels of fatigue/decreased vitality. Van Tulder et al. also found decreased vitality at the same level as the present study [13], but the difference failed to reach statistical significance and was probably related to the sample sizes. Although fatigue may be a symptom of depression and as such an indicator of mental health, the vitality scale in the SF36 also correlates substantially with physical health [16]. Further, in chronic diseases fatigue has been found to correlate very weakly with psychological distress [26]. Reduced vitality among the HD survivors may therefore not be an aspect of mental health but may reflect somatic aspects of the disease or the treatment. Clinical significance Statistically significant differences in HRQOL between the HD survivors and the controls were found in five of the SF36 scales (range 3.6 (social functioning) to 10.4 (general health)). As in medical research in general, how to assess the clinical impact of statistical significant differences, especially in large cohorts, poses a challenge. In HRQOL research, one strategy has been to regard differences ^ 10 on 0100 scales as clinically significant [27]. There are some strong objections to such a strategy. Firstly, the scales' means vary considerably. Secondly, the standard deviations across scales also vary considerably. Thirdly, the scales in the SF36 do not relate linearly to external criteria. This implies that, for example, a 10 point difference in the upper part of the scale must be interpreted differently from a 10point difference in the lower [28]. Additionally, much smaller differences than 10 have been proposed as clinically meaningful [29]. Normbased comparisons are another strategy, since the HRQOL among the patients of interest, i.e., the HD survivors, then can be anchored to the population as a whole [14]. For example, a 10point decrease in general health equalises the difference in general health between men aged 4049 years and men aged 6069 years in the general population. Similarly, the 6.1 difference in physical functioning between the HD survivors and the controls equalises the difference between men aged 3039 years and 5059 years in the general population. The difference in role functioning (physical) is 1.2 points less than the difference between men aged 29 years or younger and 5059 years in the general population. By standardising the scores, the scales become comparable, and this allows us to assess which aspect of HRQOL is mostly affected. For example, the numerically biggest difference between the HD survivors and Downloaded from

7 77 the normal controls not reporting any diseases or health problems was found in role limitations (physical, 20.5). After standardising the scores, the biggest deviation was found in general health (0.88), and the sscore in the physical functioning scale (0.63) also exceeded the deviation in role limitations (physical, 0.58). Figure 1 anchors HRQOL among the HD survivors to different health states in the general population. As illustrated, HRQOL among the HD survivors is somewhat better than what is found among the general population reporting most health problems (i.e. poorest HRQOL). Conclusion HRQOL among HD survivors is affected many years after curative treatment. Future studies should address the effects of the late medical complications upon HRQOL in HD survivors. Acknowledgements The Regional Committee for Ethics in Medical Research approved the study. The study was financially supported by grant no /003 from the Norwegian Cancer Society. References 1. Abrahamsen AF, Hanmsdal E, Nome O et al. Clinical stage I and II Hodgkin's disease: Longterm results of therapy without laparatomy. Ann Oncol 1996; 7: Abrahamsen AF, Egeland T, Hansen S et al. Hodgkin's disease in a national and hospital population: Trends over 20 years. Eur J Cancer 1997; 33: Urba WJ, Longo DL. Hodgkin's disease. N Engl J Med 1992; 326: HenryAmar H, Joly F. Late complications after Hodgkin's disease. Ann Oncol 1996; 7: SI Fobair P, Hoppe RT, Bloom J et al. Psychosocial problems among survivors of Hodgkins's disease. J Clin Oncol 1986; 4: Abrahamsen AF, Loge JH, Hannisdal E et al. Sociomedical situation for longterm survivors of Hodgkin's disease: A survey of 459 patients treated at one institution. Eur J Cancer 1998 (in press). 7. Bloom JR, Gorsky RD, Fobair P et al. Physical performance at work and at leisure: Validation of a measure of biological energy in survivors of Hodgkin's disease. J Psychosoc Oncol 1990; 8: Lund MB, Kongerud J, Boe J et al. Cardiopulmonary sequelae after treatment for Hodgkin's disease: Increased risk in females? Ann Oncol 1996; 7: Loge JH, Abrahamsen AF, Ekeberg 0 et al. Psychological distress after cancer cure. A survey of 459 Hodgkin's disease survivors. Br J Cancer 1997; 76: Loge JH, Abrahamsen AF, Ekeberg 0 et al. Hodgkin's disease survivors more fatigued than the general population. J Clin Oncol 1998 (in press). 11. Joly F, HenryAmar M, Arveux P et al. Late psychosocial sequelae in Hodgkin's disease survivors: A French populationbased casecontrol study. J Clin Oncol 1996; 14: Kornblith AB, Anderson J, Cella DF et al. Hodgkin disease survivors at increased risk for problems in psychosocial adaption. Cancer 1992; 70: van Tulder MW, Aaronson NK, Bruning PF. The quality of life of longterm survivors of Hodgkin's disease. Ann Oncol 1994; 5: Lydick EG, Epstein RS. Clinical significance of quality of life data. In Spilker B (ed): Quality of Life and Pharmaeconomics in Clinical Trials. Philadelphia: LippincottRaven 1996; Loge JH, Kaasa S. Short Form 36 (SF36) health survey: Normative data from the general Norwegian population. Scan J Soc Med 1998 (in press). 16. Ware JE. The SF36 health survey. In Spilker B (ed): Quality of Life and Pharmaeconomics in Clinical Trials. Philadelphia: LippincottRaven 1996; Loge JH, Kaasa S, Hjermstad MJ et al. Translation and performance of the Norwegian SF36 health survey in patients with rheumatoid arthritis. I. Data quality, scaling assumptions, reliability and construct validity. J Clin Epidemiol 1998; 51: How to score the SF36 health survey. Boston: Medical Outcomes Trust Aaronsen NK, Acquadro C, Alonso J et al. International Quality of life assessment (IQOLA) project. Qual Life Res 1992; 1: Ware JE, Keller SD, Gandek B et al. Evaluating translations of health status questionnaires Methods from the IQOLA project. Int J Technol Assess Health Care 1995; 11: Brazier JE, Harper R, Jones NMB et al. Validating the SF36 health survey questionnaire: New outcome measure for primary care. BMJ 1992; 305: Sullivan M, Karlsson J, Ware JE. The Swedish SF36 Health survey. I. Evaluation of data quality, scaling assumptions, reliability and construct validity across general populations in Sweden. Soc Sci Med 1995; 41: Aaronsen NK, Cull AM, Kaasa S et al. The European Organization for Research and Treatment of Cancer (EORTC) modular approach to quality of life assessment in oncology: An update. In Spilker B (ed): Quality of Life and Pharmaeconomics in Clinical Trials. Philadelphia: LippincottRaven 1996; Lund MB, Kongerud J, Nome O et al. Lung function impairment in longterm survivors of Hodgkin's disease. Ann Oncol 1995; 6: McHorney CA, Ware JE, Raczek AE. The MOS 36item shortform health survey (the SF36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 1993; 31: Krupp LB, Larocca NG, MuirNash J et al. The fatigue severity scale. Application to patients with multiple sclerosis and systemic lupus erythematosus. Arch Neurol 1989; 46: Hjermstad MJ, Fayers PM, Bjordal K, Kaasa S. Healthrelated quality of life in the general Norwegian population assessed by the European Organization for Research and Treatment of Cancer Core QualityofLife Questionnaire: The QLQC30 (+3). J Clin Oncol 1998; 16: Ware JE, Keller SD. Interpreting general health measures. In Spilker B (ed): Quality of life and pharmaeconomics in clinical trials. Philadelphia: LippincottRaven 1996; Testa MA, Simonson DC. Assessment of qualityoflife outcomes. N Engl J Med 1996; 334: Received 17 August 1998; accepted 25 November Correspondence to: J. Havard Loge, MD Department of Behavioural Sciences in Medicine University of Oslo P.O. Box 1111 Blindern N0317 Oslo Norway j.h.loge@medisin.uio.no Downloaded from

8 Downloaded from