This study demonstrated the validity and reliability of the Persian version of the Lymphedema Life Impact Scale (LLIS). As such, the Persian version of LLIS had a relatively good construct validity and reliability, and thus could be used for measuring quality of life in Iranian patients with lymphedema.
During the psychometric evaluation of the Persian version of the questionnaire, only a few points with minor ambiguity were pointed out by the experts and patients and were corrected accordingly. It seemed that the corrected version had a sufficient clarity in terms of items and content. For higher quantitative accuracy, the CVI and CVR were also measured and the calculated values were at a relatively acceptable level. Although some items had low CVR, they were considered acceptable due to their high mean scores (equal or higher than 2.8 out of 3.00). Weiss [12] assessed the content validity of the same tool by asking for the comments from four experts. Since the mean scores of importance and necessity of each item were greater than or equal to 3 out of 4, (1=not pertinent; 4=highly pertinent), all items were reported acceptable. Although the Iranian medical community has limited knowledge on lymphology, the present study surveyed nine people and obtained good results.
In order to measure the discriminant validity, two age-matched groups with and without lymphedema were evaluated and compared. The two groups had some differences in terms of characteristics and clinical risk factors of lymphedema. The mean of excised lymph nodes (p=0.032) and time interval from cancer diagnosis (p = 0.001) in lymphedema patients was higher than the comparison group. Lymphedema patients were less educated and more obese. The frequency of chemotherapy (P = 0.03) and underlying disease (p = 0,029) were higher in lymphedema patients, too. These findings were concordant with results of previous study about lymphedema risk factors in Iran [4]. The mean total scores of the LLIS in patients with and without lymphedema were 0.38 and 0.29, respectively. In other words, the group with lymphedema perceived 9% higher impairment (P < 0.0001). However, one should be aware that due to a relatively large sample size and low difference of psychosocial subscale between patients with and without lymphedema, the difference of the total LLIS score might be diluted. Differences were also present between the two groups when physical, psychosocial, and functional subscales were compared however, the difference in the psychosocial subscale was not significant. This can be justified by the various individual, family, and social problems faced by patients with breast cancer. So Hyun Lee et al. used the Short Form (SF-36) Health Survey to evaluate the quality of life in patients with cancer-related lymphedema who had surgery at least one year before the study. They found that patients with lymphedema had significant differences with the general population in all subscales of the SF-36 except for vitality and mental health [18]. These findings may indicate that lymphedema has a greater effect on physical and functional aspects of the patients’ quality of life. In fact, these factors need to be examined in future studies. To sum up it is important to note that although here we reported good discriminant validity for the LLIS as pre-hypothesized assumption, this might be challenged on several ground including the fact that we did not indicate how much difference or differences in impairments between patients with and without lymphedema would be satisfactory or acceptable.
The SF-36 was also completed after evaluating the convergent validity of the LLIS in all patients. The relationships between the SF-36 scores and the total scores of the LLIS as well as its physical, mental, and pain subscales were examined in the general population, the group with lymphedema, and the Comparison group. Spearman’s correlation coefficient showed significant inverse correlations in all of the above cases. As described in the Materials and Methods section, higher scores of the LLIS indicated higher impairment in the quality of life and higher scores of SF-36 reflected better quality of life. Therefore, the inverse relationships between the mentioned variables suggested the convergence of the above-mentioned questionnaires. The two groups had significant differences in the mean scores of the physical, somatic pain, and mental health subscales of the SF-36. However, the obtained correlation coefficients in the physical, psychological, and pain subscales of the LLIS in the group with lymphedema did not show very strong relationships. Therefore, it seems that applying SF-36 for the assessment of the quality of life in patients with lymphedema maybe a less than ideal choice. Likewise, Lee [18] and Ahmed [19] failed to find significant relationships between the two groups in most subscales. Rather than using the SF-36, Keeley [12] and Weiss [13] administered the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ-C30) and found correlation coefficients higher than 0.70 between the LLIS and most subscales of the EORTC-QLQ-C30. In this study the LLIS and EORTC-QLQ-C30 were completed by 46 patients with lymphedema caused by breast cancer and high correlations (> 0.7) were found between the total scores of the LLIS and functional subscales of the EORTC-QLQ-C30. Therefore, the EORTC QLQ C-30 seems to be a more relevant tool in patients with lymphedema than the SF 36, but further studies are required to clarify this finding.
In this study, both confirmatory factor analysis and exploratory factor analysis were used to evaluate factor validity. The acceptable values of the indicators measured in the confirmatory factor analysis, i.e. CFI and RSMA, confirmed the fitness of the 18-item translated questionnaire. However, since some indicators, such as RMSEA, indicated the marginal fit of the model, the researchers suspected that the questionnaire might have a different structure (rather than its original structure arranging items in the physical, psychosocial, and functional subscales) in the population under study in Iran. Exploratory factor analysis was, hence, performed to determine the appropriate arrangement of the items. But according to the results, the item related to infection history was placed in the functional subscale (instead of the physical subscale). Weiss reported the inclusion of an item about the episodes of infection as an advantage of the LLIS over other quality of life tools. The frequency of infections is a factor with undeniable effects on patients’ quality of life and should thus be measured. The LLIS investigates the history of an infection leading to hospitalization or receiving antibiotics in the past two years through a five-choice item scored as 1 to 5 (never, less than once a year, 1-3 times a year, 4-6 times a year, and 7-9 times a year, respectively). In the evaluation of the questionnaire’s reliability, it was observed that this item reduced the internal consistency of the physical subscale. It was, hence, separately scored in another part of the analysis and removed from the physical subscale. This was considered a limitation of the LLIS and would require the classification of the frequency of infection under a different subscale in the future versions of the scale [13]. Similarly, Bogan et al. reported the frequency of infection as a major confounding factor in increasing edema, decreasing mobility and self-esteem, and impairing social communications [20]. However we believe that it would be more suitable to move the infection question from the physical subscale to the functional subscale. The relevance of this argument should, nevertheless, be evaluated by the developers of the LLIS. In contrast one might argue that the overlap among items on factor 2 and 3 raises the question whether they do (partially) measure the same thing. However, continuing to use the current factors seems the most reasonable approach.
In an attempt to measure the reliability of the Persian version of the LLIS, its repeatability over a one-week period was evaluated through test-retest and an intraclass correlation coefficient (ICC) of 0.96 was obtained for the total scores of the scale. The ICCs of the physical, psychosocial, and functional subscales of the LLIS were 0.927, 0.855, and 0.980, respectively. Because of ethical issue for postponing the treatment and acceptable statistical levels of ICC, only 13 patients were recruited for test-retest reliability assessment. However, the original study by Jan Weiss reported the ICCs for the total scores of the LLIS and its subscales as follows: 0.99, 0.97, 0.978, and 0.965, respectively [13]. Thus it is possible to say that the current study showed relatively good stability overtime.
The internal consistency of the LLIS was assessed by calculating Cronbach’s alpha. The obtained values for the total score of the scale and its physical, psychosocial, and functional subscales were 0.873, 0.854, and 0.884, respectively. These values were all at an acceptable level and consistent with the values reported for the English version of the scale (0.85, 0.841, and 0.888, respectively)[13].
The present study focused on the psychometric properties of the LLIS only in patients with lymphedema due to breast cancer and yielded relatively good results. Since the original version of the scale was designed for patients with lymphedema of various etiologies, future studies are recommended to evaluate its validity in different populations of patients with lower limb or primary lymphedema. Use of the LLIS to determine the life impact of lymphedema, can pave the way for interventional studies of methods to improve the quality of life of those living with this disease.