Comparison of 2 quality-of-life questionnaires in women treated for breast cancer: the RAND 36-item health survey and the functional living index-cancer.

Breast carcinoma is the most frequently occurring cancer in women. Enhanced early detection techniques and improved treatment options have increased 5-year survival rates to 86% or higher, depending on age, ethnicity, and disease stage at diagnosis. (1) As length of breast cancer survival increases, so does the prevalence of treatment-related side effects, many of which can adversely affect health and quality of life. For this reason, the use of health-related quality-of-life (HRQoL) measures has become increasingly common in clinical trials and rehabilitation program assessments.

Health-related quality of life has been described as the effect that a medical condition or its treatment has on a person. (2) Although the concept and measurement of HRQoL remains a controversial subject, most questionnaires operationally define quality of life according to the World Health Organization definition of health (3) by assessing at least 3 domains of well-being: physical, emotional, and social. (2,4) Health-related quality-of-life assessment tools typically take the form of questionnaires completed by patients and may be classified as either generic or condition-specific. Measures of condition-specific HRQoL are intended for use with people diagnosed with particular diseases, such as diabetes or cancer. Generic questionnaires are designed to be used by anyone, allowing health status comparisons across groups of patients as well as people without disease or impairment. (4,5) Measures designed for specific populations are believed to be more relevant and practical to patients because they address prevalent physical findings and symptoms. (2,5) Generic health status measures, however, are believed to discriminate among patient groups on important dimensions of overall health and functioning, making them potentially useful as dependent measures in randomized clinical trials. (2,5) Because quality of life is a multifaceted phenomenon, HRQoL instruments often have a subscale structure, allowing outcomes to be presented as a profile. These profiles identify specific domains in which patients perform well or fairly well and others in which they perform less well. Most instruments also include summary items that permit the assessment of overall HRQoL. (2)

There are good reasons to compare quality-of-life scores obtained from women with breast cancer with and without secondary lymphedema using different HRQoL questionnaires. Previous studies (6-18) have examined the consequences of lymphedema and other treatment-related side effects in this population, with conflicting, and sometimes contradictory, results. Comparison of outcomes across different studies is hampered by the fact that a variety of health status questionnaires have been used to assess quality of life. No single instrument has emerged as the criterion measure of quality of life for studies involving patients with breast cancer, and the ability of available HRQoL questionnaires to detect quality-of-life deficits in women with lymphedema secondary to breast cancer is largely unknown. (19-21) Moreover, some authors have used only summary scores to report overall quality-of-life outcomes for women with breast cancer. (13,15) Careful interpretation of the results of these studies is warranted, because it is unclear whether contradictory results are the result of underlying differences between populations, differences in intervention efficacy, or differences in the content and psychometric properties of the quality-of-life instruments used. (22)

In this study, we compared 2 HRQoL questionnaires, the RAND 36-Item Health Survey (SF-36) (4,23) and the Functional Living Index-Cancer (FLIC), (24) in a sample of women treated for breast cancer. These questionnaires were selected because they are 2 of the most widely used generic (SF-36) and cancer-specific (FLIC) HRQoL instruments available to clinicians and researchers. Both questionnaires have been used as outcome measures in previous rehabilitative oncology studies, (13,14,16) and both are available at no cost in the public domain. The primary goals of this article are to help investigators select outcome measures for use in studies of women with breast cancer and to help clinicians interpret results from these studies.

Specific questions addressed by our study were:

1. Are SF-36 and FLIC summary scores interchangeable as indicators of overall HRQoL in women treated surgically for breast cancer (construct overlap)?

2. Do similarly named subscales of the SF-36 and the FLIC measure the same dimensions of HRQoL (subscale convergence)?

3. To what extent are the SF-36 and the FLIC sensitive to HRQoL differences between women with breast cancer with and without secondary lymphedema (discriminative validity)?

Method

Subjects

Women treated surgically for breast cancer at a National Cancer Institute-designated Comprehensive Cancer Center were recruited to participate in our study. They were required to be 18 to 65 years of age and capable of reading and understanding English at an 8th grade level. To help control for the effects of normal postoperative edema, candidates who were less than 3 months removed from breast cancer surgery were excluded from participation. Potentially eligible patients identified from outpatient appointment lists were approached by a member of the research team at the time of their surgical follow-up appointments and invited to participate in the study. Eighty-eight (53.7%) of 164 eligible women approached in this manner provided written consent to participate and returned completed questionnaires in person or by mail. Ten of these participants responded "yes" to the question "Has your doctor ever told you that you have lymphedema?" Later, a second recruitment phase was added to include additional women who had lymphedema secondary to breast cancer. In this phase, 30 consecutive patients who had breast cancer and were referred by their oncologists for treatment of ipsilateral upper-extremity lymphedema and who met the inclusion criteria were invited to participate. Twenty-five patients (phase 2 acceptance rate=83.3%) approached at the time of their initial physical therapy appointment provided written consent and returned completed questionnaires. When both recruitment phases were considered, 113 of 194 eligible women participated in this study (overall acceptance=58.2%). The most frequent reasons given for not participating were disinterest in questionnaire studies and lack of time to complete the questionnaires.

Quality-of-Life Questionnaires

RAND 36-Item Health Survey (SF-36). The SF-36 is a generic, 36-item questionnaire that measures 8 health-related domains--physical functioning, role-physical (role limitations due to physical impairments), bodily pain, role-emotional (role limitations due to personal or emotional problems), mental health, social functioning, vitality/fatigue, and general health--with higher scores representing a more favorable health status. We scored the questionnaire using the RAND method, which uses the same items as the Medical Outcomes Study (MOS) SF-36, (25) but with simplified algorithms for scoring the bodily pain and general health subscales. Validation studies conducted with data from the MOS have shown that the RAND and MOS scoring methods produce equivalent forms of these 2 subscales (r =.99). (4,23) The test manual distributed by the Medical Outcomes Trust also includes instructions for computing 2 health summary scores, the Mental Component Summary score and the Physical Component Summary score, by summing weighted subscale scores. The value of the SF-36 is that it allows comparisons of quality of life experienced by people with breast cancer and adults who are healthy. To do this, raw scores are converted to standardized T-scores (X= 50, SD= 10) using US general population norms reported in the test manual. (26) Accordingly, a score of 50 represents the national average for adults who are healthy, whereas a score of 40 is 1 standard deviation below the national norm.

Functional Living Index-Cancer (FLIC). Also known as the Manitoba Functional Living Cancer Questionnaire, the FLIC is one of the most frequently used quality-of-life instruments designed for use by people diagnosed with cancer. (12,22,27-30) The questionnaire contains 22 items with 7-point Likert-type linear analog scales. Researchers in previous studies of women with breast cancer have typically reported results as a total score obtained by summing the 22 unweighted item responses (range=22-154), with higher scores indicating better health status. However, several psychometric studies have shown that the FLIC measures 5 dimensions of health related to quality of life: physical functioning, mental functioning, social functioning, general health/well-being, and gastrointestinal symptoms. (22,29,30) Subscale scores for the FLIC are computed by summing individual items, and maximum scores vary, ranging from 14 to 35 points. Four items included in the total score do not appear to correlate with any of the 5 subscales. (30) Because this 5-factor structure has demonstrated stability, both when applied to different populations of people with cancer and when compared with other HRQoL measures that have well-defined subscales, we used it to determine whether similarly named FLIC and SF-36 subscales measure the same quality-of-life dimensions.

Data Analysis

Before the main analysis, descriptive statistics were computed and the results were visually examined for missing cases and normality. Three cases with missing data were deleted from the data set, leaving a total of 110 completed records. Because most subscale scores were negatively skewed and quality-of-life scores generally have not been shown to meet the criteria for interval-level data, nonparametric statistics were selected for the data analyses. Ordinal-level correlation coefficients (Kendall tau-b) were used to assess construct overlap between overall quality-of-life scores (our first research question). Large coefficients between SF-36 and FLIC summary scores would indicate substantial congruence of quality-of-life content between the questionnaires, whereas smaller values would indicate dissimilarities of construct coverage. Differences between groups based on respondent age and length of time since diagnosis of breast cancer were explored using independent t tests (2-tailed, [alpha]=.05).

A multi-trait-multi-method correlation (tau-b) matrix was constructed to determine whether similarly named subscales of the 2 questionnaires (methods) assess the same underlying quality-of-life dimensions (traits). (31) According to this approach, similar HRQoL dimensions measured with different questionnaires (eg, the physical functioning subscales of the SF-36 and the FLIC) would be expected to correlate highly with each other; that is, they should converge. However, different health dimensions measured with different questionnaires (eg, the physical functioning subscale of the SF-36 and the social functioning subscale of the FLIC) and different dimensions of health measured within the same questionnaire (eg, SF-36 physical functioning and social functioning subscales) would be expected to show low (divergent) correlations with each other.

In this study, convergent relationships were expected within each of the 3 major quality-of-life domains: physical well-being (FLIC physical functioning subscale; SF-36 physical functioning, role-physical, and bodily pain subscales), emotional well-being (FLIC mental functioning subscale, SF-36 mental health and role--emotional subscales), and social well-being (FLIC social functioning and SF-36 social functioning subscales). We also assessed convergence between the general health subscales of the FLIC and SF-36. Chi-square statistics were used to test the proportion of convergent correlations actually exceeding divergent correlations. For example, the convergent correlation between the physical functioning subscales of the FLIC and SF-36 was compared with all other correlations involving these 2 subscales. This procedure was repeated for the 2 remaining convergent correlations within the physical domain (ie, FLIC physical functioning and SF-36 role-physical correlation; FLIC physical functioning and SF-36 bodily pain correlation). These 3 convergent correlations were then compared with the remaining 66 divergent correlations involving these subscales. Under the null hypothesis, the proportion of pair-wise comparisons in which convergent correlations exceed divergent correlations is equal to the proportion of comparisons in which divergent correlations exceed convergent correlations. If subscales within a domain show convergent validity, the proportion of convergent correlations exceeding divergent correlations should be high ([chi square]<0.05). (22)

A contrasted groups approach was used to assess discriminative properties of the SF-36 and FLIC. An instrument is said to possess discriminative validity if a patient group expected to have lower quality of life actually produces lower scores than the group of comparison subjects. (32) For this part of the analysis, women treated for breast cancer were assigned to 1 of 2 subgroups based on condition. Participants reporting a previous diagnosis of upper-extremity lymphedema were assigned to the lymphedema group (n=32), and those without lymphedema were assigned to the comparison group (n=78). We expected that quality-of-life scores obtained from women with lymphedema secondary to breast cancer would be lower than scores obtained from women who had breast cancer and did not have lymphedema. Mann-Whitney U tests and effect sizes (difference of group means/standard deviation of the entire sample) were used to evaluate and compare the discriminative powers of both questionnaires across multiple dimensions of health. (33) Scores on the FLIC were tested separately from scores on the SF-36 to simulate studies in which either questionnaire was used separately. (29) Bonferroni corrections were applied so that the overall false-positive rate for each questionnaire would not exceed 5%. All tests and analyses were performed using SPSS, version 12.1 for Windows. *

Results

Descriptive Statistics

Characteristics of the entire sample and both subgroups (ie, women with lymphedema and breast cancer and women with breast cancer without lymphedema) are summarized in Table 1. No differences were noted between the lymphedema group and the comparison group with regard to age or length of survival.

Construct Overlap of HRQoL Summary Scores

Bivariate correlations between the SF-36 Physical Component Summary scores, SF-36 Mental Component Summary scores, and FLIC total scores are displayed in Table 2.

Convergent Validity of Subscale Scores (Multi-trait--Multi-method Analysis)

Kendall tau-b correlations between subscales of the FLIC and SF-36 are displayed in Table 3. Expected convergent correlations (monotrait-multi-method) among subscales representing physical, emotional, social, and general dimensions of health are outlined on the diagonal of the correlation matrix.

Within the physical domain of HRQoL, the 3 convergent correlations exceeded the corresponding divergent correlations in 59 of the possible 66 pair-wise comparisons ([chi square=20.48, P<.001). This finding suggests that the developers of the SF-36 and the FLIC viewed physical well-being in similar ways. The relationship between the FLIC physical functioning and the SF-36 role-physical subscales was particularly noteworthy, exceeding divergent correlations in all 22 pair-wise comparisons ([chi square]=11.0, P<.001) and indicating that these subscales are potentially interchangeable in women with lymphedema secondary to breast cancer.

Similarly, convergent correlations within the domain of mental well-being exceeded divergent correlations in 35 of the 44 possible comparisons ([chi square]=7.68, P<.01). Strong evidence was found to support the assumption that the FLIC mental functioning subscale is interchangeable with the SF-36 mental health subscale ([chi square]=9.09, P<.005) but not the SF-36 role-emotional subscale ([chi square]=0.82, P<.50). Evidence of convergence also was found between the FLIC and SF-36 subscales representing social functioning ([chi square]=4.45, P<.05). Tests for convergence within the general health dimension, however, were not significant ([chi square]=3.27, P<.10), indicating that these similarly named subscales measured somewhat different quality-of-life dimensions in this sample.

Discriminative Validity (Contrasted Groups Approach)

Quality-of-life scores obtained from the total sample and both subgroups of subjects are summarized in Table 4 (FLIC) and Table 5 (SE-36).

Discussion and Conclusions

In our study, we compared convergent and discriminative properties of the FLIC and SF-36 in a sample of women with breast cancer. We found that SF-36 Physical Component Summary and Mental Component Summary scores were only weakly related (tau-b=.247), confirming that these summary measures assessed separate and distinct domains of HRQoL in this sample (Tab. 2). Correlations between the FLIC total score and the SF-36 Physical Component Summary score (tau-b=.556) and Mental Component Summary score (tau-b=.490) were moderate, indicating only a modest degree of construct overlap between the FLIC and the SF-36 (Tab. 2). A multi-trait-multi-method correlation analysis of subscale scores confirmed similar patterns of construct delineation within the physical, social, and emotional dimensions of HRQoL but not within the general health dimension. These results suggest that the developers viewed the domains of physical, mental, and social well-being in ways that were similar but that the SF-36 and FLIC questionnaires measure global HRQoL somewhat differently.

Several factors may be responsible for these findings, including different concepts of overall health, different scoring methods, and different ways of framing general health questions. The SF-36 general health questions address the patient's health status directly or in comparison with that of other people, whereas the FLIC general health questions ask patients to rate their feelings about their health status. For example, the SF-36 asks respondents to agree or disagree with general health status statements (eg, "My health is excellent," "I am as healthy as anyone I know"), whereas the FLIC asks about health-related moods on the day the questionnaire is completed (eg, "How do you feel today?," "How well do you appear today?"). Individuals who may not feel or look particularly well on a given day but who believe that they are healthy on most days might give themselves low marks when completing the FLIC but might rate their overall health as "excellent" when completing the SF-36.

Findings from our study support a growing body of knowledge demonstrating that the problem of construct divergence between the FLIC and the SF-36 exists across populations with cancer and that this problem extends to other HRQoL inventories as well. (22,29,33,34) Kuenstner et al (22) compared FLIC and SF-36 subscale scores gathered from a sample of 234 patients with a variety of solid tumors and hematological malignancies and observed patterns of convergence and divergence similar to those found in our study. Specifically, convergent correlations exceeded divergent correlations among subscales assessing physical well-being (41 of 44 comparisons) and mental well-being (44 of 44 comparisons) but not general health status (12 of 24 comparisons). The authors concluded that patients with cancer may associate general health questions from the SF-36 with physical well-being, whereas items included in the general health subscale of the FLIC seem to be associated with subjective feelings related to mental health. (22)

Controlled assessments of discriminative validity are rare in the field of rehabilitation oncology. To our knowledge, our study is the first to compare generic and condition-specific quality-of-life scores obtained from women with breast cancer with and without secondary lymphedema. Discriminative validity is an essential property for HRQoL questionnaires because failure to discern deficits along important dimensions of well-being may contribute to ceiling effects when these questionnaires are used to assess and compare clinical populations. As expected, FLIC total scores for women with lymphedema were lower than similar scores obtained from the comparison group. Substantial differences in physical well-being also were found when SF-36 Physical Component Summary scores from both groups were compared. However, no differences were found between groups when SF-36 Mental Component Summary scores were examined. When health status profiles developed from each questionnaire were analyzed in detail, it was evident that the discriminative powers of both the SF-36 and the FLIC varied by subscale. The greatest differences between groups (effect sizes=0.5-0.75) were found within the domain of physical well-being for both instruments. In contrast, no differences were found between groups using the mental health subscale of the SF-36 (effect size=0.35), the FLIC social functioning subscale (effect size=0.44), or the FLIC gastrointestinal symptom subscale (effect size=-0.1).

When relatively large samples are studied, it is possible for small differences in HRQoL scores to be statistically significant yet clinically unimportant. The concept of minimally clinically important difference (MCID) indicates the smallest difference in a score that is believed to be important. Although absolute thresholds for MCID are difficult to define, the literature on the SF-36 generally shows effect sizes for MCID lying in the range of 0.3 to 0.5. (4) Statistically significant differences in SF-36 scores collected from women with breast cancer with and without lymphedema during our study consistently fell within these limits, suggesting that those HRQoL differences also may be clinically meaningful. To the best of our knowledge, MCIDs have not been published for the FLIC.

Our findings support previous studies that have shown that women with lymphedema appear to experience greater physical challenges than other women with breast cancer of similar age and length of survival. Kwan et al (11) reported that women with lymphedema scored lower than other women with breast cancer on the SF-36 physical functioning subscale but found no differences with regard to social functioning or mental health. Using another multidimensional quality-of-life questionnaire, the Functional Assessment of Cancer Therapy, Beaulac et al (6) observed lower overall quality of life and physical functioning but not mental or social well-being in patients with lymphedema. These findings support the belief that quality-of-life questionnaires tend to be most sensitive to differences between patient groups along the physical dimension of health and functioning. (2,5)

Results from our study also suggest that the SF-36 possesses relatively weak discriminative power with regard to emotional well-being, failing to demonstrate mental health differences between women with lymphedema secondary to breast cancer and women with breast cancer without lymphedema that were detected using the FLIC. This finding is consistent with the results of Broeckel and colleagues, (7) who compared SF-36 scores from patients with breast cancer treated with adjuvant chemotherapy with those from a group of age-matched comparison subjects with no history of cancer. In that study, depressive symptomatology reported by the patients who received chemotherapy using the Center for Epidemiologic Studies Depression Scale was greater than that reported by the comparison subjects. As in our study, large differences in physical functioning were observed between groups, but no group differences were found when SF-36 Mental Component Summary or mental health subscale scores were compared. The authors noted that the SF-36 contains only 2 items addressing anxiety and 3 items assessing depression and suggested that this instrument's coverage of mental health symptoms may be too limited to be able to detect the types of emotional problems experienced by patients with breast cancer. These findings imply that condition-specific measures such as the FLIC may be more sensitive to emotional issues that people with breast cancer and lymphedema consider important than generic HRQoL measures.

This view is supported by previous reports of concurrent validity between the FLIC and commonly used indexes of depression, anxiety, and other cancer-related symptoms. (30) Other investigators (7) have suggested that HRQoL assessment of patients with cancer may be further improved by using additional measures assessing specific symptoms or emotional states (eg, depression, anxiety) to augment HRQoL questionnaires. Alternatively, researchers may choose to use a combination of generic and condition-specific health status measures to assess the impact of breast cancer and treatment-related side effects on quality of life. An advantage of these comprehensive outcome assessment strategies is that they have the potential to capitalize on the strengths of various instruments while minimizing their limitations. (5) Incorporating and comparing several measures of HRQoL also may contribute to the accurate measurement and understanding of the effects of cancer-related morbidities on perceived health status and life satisfaction. (35)

A secondary finding of our study was that women with secondary lymphedema averaged 1 standard deviation below the norms for the US population with regard to physical well-being. (36) This classification is similar to people diagnosed with chronic lung disease, arthritis, diabetes, or symptomatic human immunodeficiency virus. (26,36) These results are consistent with 1 of 2 previous studies comparing SF-36 subscale scores from people with lymphedema secondary to breast cancer with national norms. As in our study, Pain and colleagues (14) reported that a substantial portion of subjects in their sample (n=48) were well below US general population values for physical functioning but not for mental health. In contrast, Velanovich and Szymanski (16) found that a small sample of women with secondary lymphedema (N=11) scored well below national norms for SF-36 subscales measuring mental health but not those measuring physical functioning.

Reasons for these somewhat contradictory results are unclear, but they may be related to differences in subject characteristics, selection procedures, or psychometric properties of the SF-36. The SF-36 has demonstrated an interesting age bias with regard to emotional well-being in women with breast cancer, with older women reporting better mental health outcomes than younger ones. (7,12) The patients with lymphedema in the study by Velanovich and Szymanski, (16) however, were somewhat older than those in our sample (59.1 versus 50.6 years), so the direction of any instrumental age bias should have resulted in reports of emotional well-being that were better rather than worse for adults who were healthy. A more likely explanation for these conflicting results may be related to longitudinal improvements in emotional well-being following treatment for breast cancer. In our study, length of survival for women with lymphedema averaged about 2.6 years, ranging from 3 months to 9 years. Health-related quality-of-life scores reported from the study by Velanovich and Szymanski were obtained 6 to 48 months following surgery, but group statistics regarding average length of survival were not provided. Because many of our subjects appear to have had longer periods of time to cope with and adapt to the psychosocial consequences of lymphedema, it seems reasonable to expect that their emotional well-being scores would be higher. A similar coping/adaptation phenomenon also might have contributed to the positive age bias observed in SF-36 mental health scores in previous studies of women with breast cancer. (35)

In contrast, participants in the comparison group in our study reported remarkably good health when compared with adults who were healthy participating in the Medical Outcomes Study. (36) This finding is consistent with results from several short-term (2-5 years) follow-up studies of people with breast cancer. (9,12,28,35) Ganz and associates (9) studied 864 women surveyed 1 to 5 years following diagnosis of early-stage breast cancer. Profiles on the SF-36 were at or above age-matched population norms for women who were healthy and substantially higher than those of outpatients with other chronic medical conditions. Similar findings have been reported by other investigators comparing quality of life in people with breast cancer with population norms. (12,28,35) Total scores on the FLIC obtained from our study also are within the range (124-136) reported from previous studies of women with breast cancer. (12,28,29)

Because the main objective of our study was to compare the measurement properties of the SF-36 and the FLIC, sociodemographic variables (eg, income, ethnicity/cultural preference, marital status) and clinical data (eg, disease stage, surgical treatment, types of adjuvant therapy) that might influence HRQoL were not collected. Consequently, we cannot comment on the effect that these potentially confounding factors may have had on the physical, psychological, or social dimensions of health reported by the participants in this study. We realize that this omission limits the generalizability of the findings. This weakness, however, should not detract from our findings concerning the measurement properties of the SF-36 and FLIC questionnaires.

For users of HRQoL questionnaires, particularly the FLIC and SF-36, the following implications can be drawn from our findings. When selecting an HRQoL instrument, investigators should not rely simply on the names of the questionnaires and their subscales or domains but also should take into account the content and structure of the individual items. Although subscales of the SF-36 and the FLIC addressing physical, mental, and social well-being appear to measure similar dimensions of health, these subscales and overall HRQoL scores (ie, FLIC total score, SF-36 Physical Component Summary score, and SF-36 Mental Component Summary score) are not interchangeable. In addition, although both the SF-36 and the FLIC appear to be useful for measuring physical functioning in patients with breast cancer, the condition-specific FLIC may be more sensitive to psychological factors influencing the health and well-being of people with breast cancer and secondary lymphedema than the generic SF-36.

Clinicians and researchers should note that the same caveat applies to the interpretation and comparison of study results. When a particular research question has been investigated in different studies using either the SF-36 or the FLIC, readers should recognize that inconsistent findings may be the result of construct divergence between the 2 assessment instruments rather than of true differences in treatment effectiveness or actual quality-of-life outcomes.

In our study, we compared quality-of-life scores obtained using 2 questionnaires: the SF-36 and the FLIC. When only 2 instruments are compared, it is unclear whether low correspondence between similarly labeled subscales, as in the case of the general health subscales of the FLIC and SF-36, is the result of poor statistical properties of one instrument, whether the 2 questionnaires stress different aspects of one domain, or whether a quality-of-life dimension is more difficult to assess than other dimensions. (22) Future studies using a third questionnaire may provide additional insight into the likely causes of weak convergent validity observed in our study.

Finally, we should note that instrument validity is contextual and that its evaluation is an iterative process. Individual studies of convergent and discriminative properties provide important evidence but not absolute proof of construct validity. (37) That evidence is most relevant when gathered from the populations and settings in which the instrument is intended to be used. Further research comparing various HRQoL instruments is needed to identify meaningful differences among questionnaires and to evaluate their usefulness and reliability within specific subpopulations of people with cancer.

References

(1) Edwards BK, Howe HL, Ries LA, et al. Annual report to the nation on the status of cancer, 1973-1999, featuring implications of age and aging on the US cancer burden. Cancer. 2002;94:2766-2792.

(2) Cella D, Nowinski CJ. Measuring quality of life in chronic illness: the Functional Assessment of Chronic Illness Therapy measurement system. Arch Phys Med Rehabil. 2002;83(12 suppl 2):S10-S17.

(3) World Health Organization. The Constitution of the World Health Organization. Geneva, Switzerland: WHO Basic Documents; 1978.

(4) Hays RD, Morales LS. The RAND-36 Measure of Health-Related Quality of Life. Ann Med. 2001;33:350-357.

(5) Patrick DL, Deyo RA. Generic and disease-specific measures in assessing health status and quality of life. Med Care. 1989;27(3 suppl): S217-S232.

(6) Beaulac SM, McNair LA, Scott TE, et al. Lymphedema and quality of life in survivors of early-stage breast cancer. Arch Surg. 2002;137: 1253-1257.

(7) Broeckel JA, Jacobsen PB, Balducci L, et al. Quality of life after adjuvant chemotherapy for breast cancer. Breast Cancer Res Treat. 2000;62:141-150.

(8) Coster S, Poole K, Fallowfield LJ. The validation of a quality of life scale to assess the impact of arm morbidity in breast cancer patients post-operatively. Breast Cancer Res Treat. 2001;68:273-282.

(9) Ganz PA, Rowland JH, Desmond K, et al. Life after breast cancer: understanding women's health-related quality of life and sexual functioning. J Clin Oncol. 1998;16:501-514.

(10) Hack TF, Cohen L, Katz J, et al. Physical and psychological morbidity after axillary lymph node dissection for breast cancer. J Clin Oncol. 1999;17:143-149.

(11) Kwan W, Jackson J, Weir LM, et al. Chronic arm morbidity after curative breast cancer treatment: prevalence and impact on quality of life. J Clin Oncol. 2002;20:4242-4248.

(12) Lindley C, Vasa S, Sawyer WT, Winer EP. Quality of life and preferences for treatment following systemic adjuvant therapy for early-stage breast cancer. J Clin Oncol. 1998;16:1380-1387.

(13) Mirolo BR, Bunce IH, Chapman M, et al. Psychosocial benefits of postmastectomy lymphedema therapy. Cancer Nurs. 1995;18:197-205.

(14) Pain SJ, Vowler SL, Purushotham AD. Is physical function a more appropriate measure than volume excess in the assessment of breast cancer-related lymphoedema (BCRL)? Eur J Cancer. 2003;39: 2168-2172.

(15) Sitzia J, Sobrido L. Measurement of health-related quality of life of patients receiving conservative treatment for limb lymphoedema using the Nottingham Health Profile. Qual Life Res. 1997;6:373-384.

(16) Velanovich V, Szymanski W. Quality of life of breast cancer patients with lymphedema. Am J Surg. 1999;177:184-188.

(17) Voogd AC, Ververs JM, Vingerhoets AJ, et al. Lymphoedema and reduced shoulder function as indicators of quality of life after axillary lymph node dissection for invasive breast cancer. Br J Surg. 2003;90: 76-81.

(18) Williams AF, Vadgama A, Franks PJ, Mortimer PS. A randomized controlled crossover study of manual lymphatic drainage therapy in women with breast cancer-related lymphoedema. Eur J Cancer Care (Engl). 2002; 11:254-261.

(19) Gerber LH. A review of measures of lymphedema. Cancer. 1998;83: 2803-2804.

(20) Megens A, Harris SR. Physical therapist management of lymphedema following treatment for breast cancer: a critical review of its effectiveness. Phys Ther. 1998;78:1302-1311.

(21) Sitzia J, Stanton AW, Badger C. A review of outcome indicators in the treatment of chronic limb oedema. Clin Rehabil. 1997;11:181-191.

(22) Kuenstner S, Langelotz C, Budach V, et al. The comparability of quality of life scores: a multitrait multimethod analysis of the EORTC QLQ-C30, SF-36 and FLIC questionnaires. Eur J Cancer. 2002;38: 339-348.

(23) Hays RD, Sherbourne CD, Mazel RM. The RAND 36-Item Health Survey 1.0. Health Econ. 1993;2:217-227.

(24) Schipper H, Clinch J, McMurray A, Levitt M. Measuring quality of life of cancer patients: the Functional Living Index-Cancer, development and validation. J Clin Oncol. 1984;2:472-483.

(25) McHorney CA, Ware JE Jr, Raczek AE. The MOS 36-Item Short-Form Health Survey (SF-36), II: psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care. 1993;31:247-263.

(26) Ware JE Jr, Kosinski M, Keller SD. SF-36 Physical and Mental Health Summary Scales: A User's Manual. Boston, Mass: The Health Institute; 1994.

(27) Clohisy DR, Le CT, Umen AJ. Measuring health status in patients with skeletal metastases treated by surgery. Am J Clin Oncol. 1997;20: 424-428.

(28) Ganz PA, Coscarelli A, Fred C, et al. Breast cancer survivors: psychosocial concerns and quality of life. Breast Cancer Res Treat. 1996;38:183-199.

(29) King MT, Dobson AJ, Harnett PR. A comparison of two quality-of-life questionnaires for cancer clinical trials: the Functional Living Index-Cancer (FLIC) and the Quality of Life Questionnaire core module (QLQ-C30).j Clin Epidemiol. 1996;49:21-29.

(30) Morrow GR, Lindke J, Black P. Measurement of quality of life in patients: psychometric analysis of the Functional Living Index-Cancer (FLIC). Qual Life Res. 1992;1:287-296.1

(31) Campbell DT, Fiske DW. Convergent and discriminant validation by the multitrait-multimethod matrix. Psychol Bull. 1959;56:81-105.

(32) Kirshner B, Guyatt G. A methodological framework for assessing health indices. J Chron Dis. 1985;38:27-36.

(33) Holzner B, Kemmler G, Sperner-Unterweger B, et al. Quality of life measurement in oncology: a matter of the assessment instrument? Eur J Cancer. 2001;37:2349-2356.

(34) Kemmler G, Holzner B, Kopp M, et al. Comparison of two quality-of-life instruments for cancer patients: the Functional Assessment of Cancer Therapy-General and the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30. J Clin Oncol. 1999;17:2932-2940.

(35) Matthews BA, Baker F, Hann DM, et al. Health status and life satisfaction among breast cancer survivor peer support volunteers. Psychooncology. 2002;11:199-211.

(36) Hays RD, Hahn H, Marshall G. Use of the SF-36 and other health-related quality of life measures to assess persons with disabilities. Arch Phys Med Rehabil. 2002;83(12 suppl 2):S4-S9.

(37) McLachlan SA, Devins GM, Goodwin PJ. Validation of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (QLQ-C30) as a measure of psychosocial function in breast cancer patients. Eur J Cancer. 1998;34:510-517.

* SPSS Inc, 233 S Wacker Dr, Chicago, IL 60606-6307.

RW Wilson, PT, PhD, is Assistant Professor of Physical Therapy, University of South Florida College of Medicine, and Researcher, Health Outcomes and Behavior Program, H. Lee Moffitt Cancer Center and Research Institute. Address all correspondence to Dr Wilson at School of Physical Therapy, MDC 77, University of South Florida College of Medicine, 12901 Bruce B. Downs Blvd, Tampa, FL 33612-4766 (USA) ([e-mail]).

LM Hutson, MS, ARNP, is Patient Care Coordinator, Comprehensive Breast Program, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Fla.

D VanStry, PT, BS, is Physical Therapy and Occupational Therapy Coordinator, H. Lee Moffitt Cancer Center and Research Institute.

All authors provided research concept/idea development and consultation (including manuscript review before submission). Dr Wilson provided research design, fund procurement, project management, data analysis, and writing. Ms Hutson provided institutional liaisons, and Ms VanStry provided facilities/equipment. The authors are grateful to Joan Foster, OTR, Rochelle Harris, MA, Rebecca Reid, PT, and Julie Ruopp, PT, for their enthusiastic assistance with participant recruitment and data collection.

This study was approved by the H. Lee Moffitt Cancer Center and Research Institute's Scientific Review Committee and the Institutional Review Board (Biomedical) of the University of South Florida.

This study was funded in part by the American Cancer Society (#93-032-07 to Dr Wilson) through the University of South Florida/H. Lee Moffitt Cancer Center Institutional Research Grant program.

This article was received July 12, 2004, and was accepted March 9, 2005.

Table 1.

Characteristics of Participants by Group

                                  Lymphedema         Comparison
                                  Group (n=32)       Group (n=78)

                                  [bar.X]    SD      [bar.X]    SD

Age (y)                           50.6       10.1    52.8        9.1
Years since diagnosis              2.6        2.1     2.1        1.7
RAND-36 subscales (raw scores)
  Physical functioning            53.9       26.4    81.2       21.2
  Role--physical                  42.3       28.8    75.4       29.2
  Role--emotional                 62.0       33.4    81.7       26.2
  Vitality/fatigue                44.4       21.6    62.9       23.3
  Mental health                   66.2       22.3    74.5       21.6
  Social functioning              60.9       28.6    80.0       27.1
  Bodily pain                     60.3       28.1    78.3       23.6
  General health                  54.6       21.6    73.1       24.3

                                  Total Sample
                                  (N=110)

                                  [bar.X]    SD

Age (y)                           51.8        9.1
Years since diagnosis              2.2        1.9
RAND-36 subscales (raw scores)
  Physical functioning            73.4       25.4
  Role--physical                  65.1       31.3
  Role--emotional                 76.3       28.5
  Vitality/fatigue                56.4       23.7
  Mental health                   71.3       21.4
  Social functioning              73.1       29.5
  Bodily pain                     71.8       25.8
  General health                  66.4       25.7

Table 2.

Construct Overlap Among Health-Related Quality of Life (HRQoL)
Summary Scores (a)

                          SF-36 Physical    SF-36 Mental
                          Component         Component

SF-36 Mental Component    .247
FLIC total                .556              .490

(a) SF-36=RAND 36-Item Health Survey, FLIC=Functional Living
Index-Cancer.

Table 3.

Multi-trait--Multi-method Correlation Matrix (Tau-b Coefficients) (a)

            Physical                            Emotional

            SF-36 PF    SF-36 RP    SF-36 BP    SF-36 MH    SF-36 RE

SL-36 RP    .628
SF-36 BP    .485        .552
SF-36 MH    .403        .433        .446
SF-36 RE    .481        .613        .438        .528
SF-36 SF    .557        .650        .526        .521        .497
SF-36 GH    .459        .450        .440        .443        .331
SF-36 VT    .576        .572        .541        .598        .530
FLIC PF     .616#       .691        .564        .461        .443
FLIC MF     .371        .426        .410        .586#       .455
FLIC SF     .380        .477        .443        .444        .451
FLIC GH     .440        .477        .419        .549        .420
FLIC GI     .257        .273        .227        .224        .242

            Social      General

            SF-36 SF    SF-36 GH    SF-36 VT    FLIC PF

SL-36 RP
SF-36 BP
SF-36 MH
SF-36 RE
SF-36 SF
SF-36 GH    .484
SF-36 VT    .598        .514
FLIC PF     .594        .412        .560
FLIC MF     .463        .456        .423        .398
FLIC SF     .526#       .366        .463        .499
FLIC GH     .421        .503#       .536        .541
FLIC GI     .304        .169        .246        .372

Note: Correlations between similarly named subscales are boldfaced.

(a) High correlations indicating convergence are expected between
outlined pairs of subscales within the same quality-of-life domains
(ie, similar dimensions assessed with different questionnaires). The
remaining cells are expected to show low correlations representing
construct divergence. (22) SF-36=RAND 36-Item Health Survey (subscales:
PF=physical functioning, RP=role-physical, BP=bodily pain, MH=mental
health, RE,=role-emotional, SF-social functioning, GH=general health,
VF=xitality/fatigue); FLIC:=Functional Living Index-Cancer (subscales:
PF=physical functioning, MF=mental functioning, SF=social functioning,
GH=general health/well-being, GI=gastrointestinal symptoms).
Correlations between similarly named subscales are boldfaced.

Table 4.

Means and Standard Deviations for Functional Living Index-Cancer
(FLIC) Total Scores and Subscales

                               Total Sample       Comparison
                               (N=110)            Group (n=78)

                               [bar.X]    SD      [bar.X]    SD

FLIC total score                 125.1    18.4      129.1    16.8
FLIC Subscales
  Physical functioning            27.5     7.4       29.6     6.4
  Mental functioning              26.4     6.1       27.3     5.6
  Social functioning              12.5     2.5       12.8     2.3
  General health                  23.9     4.2       24.6     3.9
  Gastrointestinal symptoms       12.8     2.1       12.8     2.5

                               Lymphedema
                               Group (n=32)       Mann-
                                                  Whitney        Effect
                               [bar.X]    SD      U Test (a)     Size

FLIC total score                 115.4    18.8     674.5          0.74
FLIC Subscales
  Physical functioning            22.6     7.2     564.0          0.95
  Mental functioning              24.0     6.7     846.5          0.54
  Social functioning              11.7     3.1    1035.5 (NS)     0.44
  General health                  22.2     4.4     856.0          0.57
  Gastrointestinal symptoms       13.0     1.7    1245.5 (NS)    -0.10

(a) All compariscms were statistically significant with Bonferroni
correction (P<.008) unless noted otherwise (NS).

Table 5.

Means and Standard Deviations for the RAND 36-Item Health Survey
(SF-36) Component Summary Scores and Subscales

                                Total Sample       Comparison
                                (N=110)            Group (n=78)

                                [bar.X]    SD      [bar.X]    SD
SF-36 summary scores
  Physical Component Summary       46.1    10.0       49.3     8.1
  Mental Component Summary         49.6     9.6       50.1     9.1
SF-36 subscales
  Physical functioning             45.3    11.1       48.9     8.8
  Role--physical                   45.3     9.3       48.1     8.2
  Role--emotional                  48.4     8.6       50.2     7.4
  Vitality/fatigue                 48.0    11.4       50.4    10.9
  Mental health                    48.2    11.9       49.4    11.6
  Social functioning               45.3    13.2       47.6    12.8
  Bodily pain                      48.9    10.6       51.1     9.3
  General health                   47.4    12.7       49.9    12.7

                                Lymphedema
                                Group (n=32)       Mann-
                                                   Whitney       Effect
                                [bar.X]    SD      U Test (a)    Size
SF-36 summary scores
  Physical Component Summary       38.1     9.9    448.0         1.20
  Mental Component Summary         48.3    10.7    976.0 (NS)    0.19
SF-36 subscales
  Physical functioning             36.6    11.4    444.0         1.11
  Role--physical                   38.6     8.4    524.0         1.02
  Role--emotional                  44.0    10.0    811.0         0.72
  Vitality/fatigue                 42.0    10.2    667.5         0.74
  Mental health                    45.2    12.2    968.5 (NS)    0.35
  Social functioning               39.7    12.6    783.5         0.60
  Bodily pain                      43.5    11.7    754.0         0.72
  General health                   41.2    10.5    682.0         0.69

(a) All comparisons were statistically significant with Bonferroni
correction (P<.005) unless noted otherwise (NS).