Health related quality of life in a nationally representative sample of haematological patients

Objectives: Knowledge of health related quality of life of haematological patients is limited. This study aimed at investigating the prevalence and predictors of symptoms and problems in a representative sample of haematological patients in Denmark. Methods: A random sample of patients with leukaemia, multiple myeloma and advanced lymphoma (n = 732) received the European Organisation for Research and Treatment of Cancer quality-of-life questionnaire (EORTC QLQ-C30). Mean scores were calculated. In addition, scores were dichotomised using two thresholds: patients reporting at least ‘a little’ of each EORTC QLQ-C30 symptom/problem were classified as having a ‘symptom/problem’, and patients reporting at least ‘quite a bit’ were classified as having a ‘severe symptom/problem’. Multiple logistic regression was used to identify predictors. Results: In total, 470 (64%) patients participated. The most frequent symptoms/problems were fatigue (55%; severe 20%), reduced role function (49%; severe 23%), insomnia (46%; severe 15%), and pain (37%; severe 15%). Older patients and patients in active antineoplastic treatment had more symptoms and problems. There was only little evidence of social inequalities. Conclusion: This is probably the first nationally representative study of symptoms and problems in haematological patients. These patients have symptoms/problems that deserve attention. Health related quality of life is an important issue in haematological malignancies.

Most studies of HRQOL in haematological patients have either focused on long-term survivors or have been part of clinical trials investigating different treatment regimens. To our knowledge, there exist no study assessing HRQOL in the entire group of haematological patients seen in hospitals. Such a study would provide clinically valuable information. It would specifically answer the two simple questions: to what extents are haematological patients burdened by symptoms and problems that deserve attention? And which groups of patients suffer from which symptoms and problems?
This study was part of a larger study assessing symptoms and problems in patients with advanced cancer defined as patients with TNM stages 3 or 4 (TNM: system for classifying cancers; T = local tumour growth, N = regional lymph nodes, M = metastasis) (the results concerning patients with solid tumours will be reported elsewhere). However, the TNM system is not fully applicable to haematological malignancies. Thus, the TNM system was employed for lymphoma but not for patients with leukaemia and multiple myeloma.
In the present study we therefore investigated HRQOL of adult patients with lymphoma in stage 3 or 4, and patients with multiple myeloma or leukaemia in any stage. The study was conducted as a nationally representative, cross-sectional study with the following aims: (i) to measure symptoms and problems; and (ii) to identify predictors of symptoms and problems.

Patients
All clinical hospital departments (except for departments of psychiatry and paediatrics) from three out of 14 regions across Denmark (total population 5.5 millions) were invited to participate (n = 81). The three regions were selected to represent Denmark concerning geographic and demographic characteristics. In each participating department all cancer patients born from the 1st to the 22nd in the month were identified from the patient register. An exception was made for four departments where a smaller proportion of patients were selected (fewer birth days were included). This leads to a slight under-representation of patients from these departments, but since only few patients with haematological malignancies were identified from these departments this has no practical implications. Only patients who had been in contact with the hospital department within the past year, and lived in one of the three regions were retrieved from the registers.
Medical students reviewed the medical records of all patients in the retrievals. Patients were included if they had lymphoma stage 3 or 4 or leukaemia or multiple myeloma. Patients were excluded if they did not speak Danish, had a cognitive impairment or had a psychiatric co-morbidity.
Included patients received an information letter, a consent form, and a questionnaire by mail. A reminder was sent after 2 weeks. The study was approved by the local ethics committee (01-116 ⁄ 03 and 11-143 ⁄ 03) and took place from October 2004 to January 2006.

Clinical data and assessments
The following clinical data were extracted from the medical records: gender, age, diagnosis, stage of disease for lymphomas [using the American Joint Committee on Cancer (AJCC) manual (15)], time of (first) diagnosis, contact type (latest contact with the hospital: out-patient or hospitalised), and treatment status (in active antineoplastic treatment or not).
Participants received the EORTC QLQ-C30 (16, 17) questionnaire assessing health-related quality of life. It consists of five multi-item function scales: physical, role, emotional, cognitive, and social function; a global health status ⁄ quality of life (QOL) scale; three multi-item symptom scales: fatigue, nausea and vomiting, and pain; and six single-item symptom scales: dyspnoea, insomnia, appetite loss, constipation, diarrhoea and financial difficulties. For the five function scales and global health status ⁄ QOL higher scores represent better functioning. Conversely, for the nine symptom scales higher scores represent more symptoms.

Statistics
The analyses were performed using sas statistical software version 9 (SAS Institute Inc., Cary, NC, USA) (18). Participants and non-participants were compared using Wilcoxon (age, time since diagnosis), chi-square (region, type of departments, diagnosis) and Fisher's exact tests (gender, treatment status, contact type).
The responses to the EORTC QLQ-C30 were converted into 0-100 scales according to the scoring manual, and mean scores were calculated (17). For the EORTC QLQ-C30 there are no predefined thresholds for when a symptom or function score should be interpreted as a case. However, to ease clinical interpretation of the data, and in accordance with previous procedures (19), we dichotomised the scores using two thresholds: (i) we calculated frequencies of 'symptoms ⁄ problems' and defined that a patient had a 'symptom ⁄ problem' if the scale score corresponded to at least 'a little' (symptom scale ‡ 33, function scale £ 67); and (ii) we calculated frequencies of 'severe symptoms ⁄ problems' and defined that a patient had a 'severe symptom ⁄ problem' if the score corresponded to at least 'quite a bit' (symptom scale ‡ 66, function scale £ 34) (see Fig. 1). This was done for all scales except global health status ⁄ QOL. Finally, we calculated the number of 'symptoms ⁄ problems' and 'severe symptoms ⁄ problems' for each person. These numbers include both symptom scales and function scales, but exclude the global health status ⁄ QOL scale, and thus range from 0-14.
We used ordinal logistic regression to identify the predictors of symptoms and loss of function. We used a stepwise procedure with inclusion ⁄ exclusion criteria of P = 0.01. The following clinical and sociodemographic variables were tested: gender, age, diagnosis, time since diagnosis, contact type, treatment status, type of department, region, formal education, children and marital status. Figure 2 shows the inclusion process. In total, 56 (69%) of the invited hospital departments participated (27 departments of surgery, 23 departments of medicine, four departments of oncology and two departments of haematology), and from their patient registers we were provided with lists of 8217 patients. From 19 of the 25 non-participating departments we were provided with retrievals containing 686 additional patients. Hence, we could have reviewed 8% more medical records if they had participated too. None of the non-participating departments were departments of haematology, oncology, or internal medicine. We reviewed the medical records of 7661 of the 8217 patients. The remaining 556 (7%) medical records were not reviewed because they were impossible to get a hold on after several attempts (e.g. the medical record was misplaced or stored in an achieve another place).

Patients
Of the 7661 medical records reviewed, 1225 patients were identified as having a haematological malignancy. Of these, we excluded 61 because they had not had contact to the department within the previous year; 269 because they had lymphoma stage 1 or 2; 60 because stage of lymphoma was inaccessible; and 34 because they had a cognitive impairment, had a psychiatric illness, or did not speak Danish. Of the 801 patients included, 54 did not receive a questionnaire because of administrative failure, and 15 died before receiving a questionnaire. Thus, questionnaires were sent to 732 patients representing 19 hospital departments. In total, 470 (64%) patients returned a completed questionnaire.
Characteristics of participants and non-participants can be seen in Table 1. Compared to non-participants, participants were younger (mean age 63 years vs. 66 years), and were less likely to have a diagnosis of CLL and non-Hodgkin's lymphoma. Table 2 shows the mean scores for the symptom and function scales, and the percentages of patients having a 'symptom' or a 'severe symptom' for the total sample and for each diagnosis. The symptom with the highest mean score (indicating most symptoms) was fatigue, followed by pain and insomnia. The function scale with the lowest mean score (indicating the most severe impairment) was role function, followed by physical function.

Symptoms and problems
Using the cut-off values, 55% of the patients had fatigue (20% severely), 49% reduced role function (23% severely), 46% insomnia (15% severely), and 37% pain (15% severely). The mean number of 'symptoms ⁄ prob-lems' was 4.3, and the mean number of 'severe symptoms ⁄ problems' was 1.5. In total, 82% of the patients had at least one 'symptom ⁄ problem', and 45% had at least one 'severe symptom ⁄ problem'. There were significant differences in the levels of physical function, role function, social function, pain, and constipation across the diagnoses (P-values not shown in Table 2). Generally, patients with multiple myeloma had most symptoms and problems. The mean number of 'symptoms ⁄ problems' ranged from 3.7 acute myeloid  1 Patients with ALL, myelofibrosis or unclassified leukaemia. 2 Patients from departments of surgery and oncology. 3 These variables were assessed in participants only. leukaemia (AML) to 5.6 (multiple myeloma); 'severe symptoms ⁄ problems' ranged from 0.9 (AML) to 2.3 (multiple myeloma).

Predictors
Results of the multivariate logistic regressions can be seen in Table 3. Older patients were more reduced in physical and role function, had lower quality of life, and more constipation, appetite loss and pain than younger patients, but less financial difficulties. Women had lower physical function, and more insomnia than men.
Recently hospitalised patients had more nausea and appetite loss than out-patients. Patients in active treatment had more reduced physical function, lower quality of life, and more appetite loss and fatigue than patients not in active treatment. Patients from departments of medicine had lower quality of life and more fatigue, than patients from haematological departments. Diagnosis was only a significant predictor of one scale in the multivariate analysis: patients with multiple myeloma had more reduced role function. Finally, patients with shorter education (as opposed to longer education) reported more reduced cognitive function (see Table 3).

Discussion
This nationally representative, cross-sectional study assessed a random sample of patients with multiple myeloma, leukaemia or lymphoma (stage 3 or 4) seen at Danish hospitals, mainly departments of haematology at university hospitals, within the previous year. The study showed that most of the symptoms and problems assessed were frequent. Using the cut-off values the most prevalent symptoms and problems were fatigue (55%; severe 20%), reduced role function (49%; severe 23%), insomnia (46%; severe 15%), pain (37%; severe 15%), and dyspnoea (36%; severe 15%). This indicates that efforts to improve the management should be given high priority.
The figures above do not take into consideration the potential co-morbidity of the patients. As such, the figures show the symptoms and problems experienced by this population regardless of their aetiology, and thus, of course, not everything can be attributed to a haematological disease.
No previous studies have included a patient sample similar to ours, and therefore comparisons are difficult. A study of CLL patients who were not receiving current anti cancer treatment (12) showed similar levels of symptoms and problems as the CLL patients in our study, al though the women of that study generally had higher levels of symptoms and problems whereas the males generally had lower levels.
In a study of AML patients (13) the reported levels of symptoms and problems at the end of in-patient treatment were similar to those reported by the AML patients in our study. However, our patients had less appetite loss, nausea and dyspnoea, and better social function. This may reflect that some of the patients included in our study had been without treatment for a long time.
A Norwegian study of Hodgkin's lymphomas survivors (20) found levels of symptoms and problems that were generally lower than found for the Hodgkin group in our study. This probably reflects that our study only included patients with advanced lymphomas. In addition, our study included patients in all disease phases and thus, some of the patients' experienced treatment-related symptoms. However, the level of dyspnoea and constipation was higher in the Norwegian study.
A Nordic study of patients with multiple myeloma (14) generally showed higher levels of symptoms and problems before patients began treatment than showed for the myeloma patients in the present study (14). The level of symptoms and problems after 6 months of treatment reported in the study by Wisloffs et al. (14) were similar to our findings.
It is difficult to interpret our results by comparing them with the few, relevant other studies conducted, since there are substantial differences in the sampling of patients. The most important way to interpret our results, from a clinical perspective, is probably done by a careful examination of the prevalences of symptoms and problems in the various subgroups.

Predictors of symptoms and problems
Age was a strong predictor of several symptom and function scales with older patients having more symptoms and problems. This is consistent with findings from general populations, where older age is associated with more symptoms and problems (21)(22)(23), and older age has also been associated with lower quality of life in two studies of CLL patients (6,12). In contrast, studies of advanced cancer patients have generally found that younger patients are more burdened by symptoms and problems (24)(25)(26)(27)(28), and younger patients reported more depression and worse emotional and social function in a study of CLL patients (11). Thus, results of the present study draw attention to the symptom-burden of the older patients, but the social and emotional strain of disease of younger patients should not be forgotten.
Patients in active treatment had more symptoms and problems. This is consistent with previous findings in haematology (29), and in cancer in general (25,30).
Recently hospitalised patients had more nausea and appetite loss than out-patients, which is an important finding. It indicates that identification and treatment of these symptoms for especially hospitalised patients is needed.
Patients with multiple myeloma generally had the highest level of symptoms and problems. This may reflect that multiple myeloma is characterised by distress due to bone pain, pathological bone fractures and recurrent infections (14). However, in the multivariate analysis, patients with multiple myeloma only were significantly more reduced in role function. A possible reason may be, that patients with multiple myeloma generally were older than patients with all other diagnoses (except CLL), and older age showed to be a strong predictor. Patients diagnosed with AML showed the lowest level of symptoms and problems. This is consistent with the findings from a study that compared patients with acute leukaemia to patients with advanced Hodgkin's lymphoma 5 years after treatment, and found that Hodgkin's patients had significantly greater psychological distress and greater fatigue (31).
Women reported more reduced physical function, and had more insomnia than men. This is consistent with findings from general populations (21)(22)(23), and has also been found in CLL patients (12,29), and in adult leukaemia survivors (32). However, a study of Hodgkin's lymphomas survivors found that women had less fatigue and better quality of life (20).
Previous studies have shown that highly educated patients have fewer symptoms and problems (24,25,30,33), and in a study of adult leukaemia survivors patients with less education reported worse overall adjustment (32). Surprisingly, education was related to cognitive function only in this study. Hence, the study only partially supports the notion that there are social inequalities in the prevalence of symptoms and problems. It has previously been found that unmarried patients with advanced cancer have more symptoms and problems (25,26), but this was neither supported in the present study nor in a study of adult leukaemia survivors (32).

Strengths and limitations
A major strength of this study is that we included an almost representative, national sample of patients with lymphoma (stage 3 or 4), leukaemia and multiple myeloma who have been in contact with a hospital department within the previous year. The participation of hospital departments was relatively high: all haematological departments in the selected regions participated, and we reviewed the majority of the medical records from the patients identified. However, there is an under-representation of patients coming from the rural region 'Ringkoebing' because some patients from this region are referred to a specialised department of haematology in a region not included in the study.
Given the representativeness of the patients included, the study gives a picture of the burden of symptoms and problems in haematological patients in contact with a hospital department. The important message is that these patients do have symptoms and problems that do deserve attention and resources in the health care system.
There are three primary weaknesses of the study. First, not all patients participated. A response rate of 64% is not optimal. However, it is within the range of what is often found in cross-sectional surveys with mailed selfassessment questionnaires (11,(34)(35)(36) and it is impossible to avoid some non-participation in such studies. If patients had been approached in person and ⁄ or we had used structured interviews we might have achieved a better response rate, but this was not feasible with our design. The oldest patients participated least in the study and together with the fact that the most burdened patients are the ones least likely to participate, this means that the study probably underestimates symptoms and problems. However, this does not jeopardise the primary conclusion of the paper: that there is a substantial prevalence of symptoms and problems in this group of patients.
Second, the fact that we included patients irrespective of diagnoses, treatment, etc. made it more difficult to subdivide patients in clinically well characterised subgroups than if a highly selected subgroup of patients was included. For example, we found it impossible to devise classifications of treatments that were applicable across the diverse spectrum of diagnoses and treatment phases. Knowledge about the effect of specific treatments on quality of life must be obtained from studies designed for this.
Third, due to space limitations in the questionnaire (we were afraid the response burden was already significant) we did not collect data on co-morbidity, which would have been beneficial, and it is a limitation of the study that we are not able to control for this potential confounder. On the other hand, for the primary purpose of this study, co-morbidity did not seem crucial. Patients need attention to and care for their experienced symptoms and problems regardless of aetiology.

Conclusion
This is probably the first nationally representative study of symptoms, problems and quality of life in haematological patients. These patients have symptoms and problems that deserve attention, and older patients and patients undergoing treatment are especially burdened. The study found little evidence of social inequalities in HRQOL. The study indicates that haematological malignancies affect HRQOL broadly and points to the challenge of alleviating this impact on HRQOL.