Fax: (011) 598-2-4873182
Determinants of quality of life in patients with cancer
A South American study
Article first published online: 21 JAN 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 5, pages 1072–1081, 1 March 2005
How to Cite
Dapueto, J. J., Servente, L., Francolino, C. and Hahn, E. A. (2005), Determinants of quality of life in patients with cancer. Cancer, 103: 1072–1081. doi: 10.1002/cncr.20870
- Issue published online: 17 FEB 2005
- Article first published online: 21 JAN 2005
- Manuscript Accepted: 5 NOV 2004
- Manuscript Revised: 29 OCT 2004
- Manuscript Received: 31 MAY 2004
- Comisión Honoraria de Lucha Contra el Cáncer, Uruguay. Grant Number: 117
- determinants of quality of life;
- Functional Assessment of Cancer Therapy-General questionnaire;
- Functional Assessment of Chronic Illness Therapy-Spiritual Well-Being scale;
- South America;
Because health-related quality of life (QOL) is an important outcome in cancer management, the authors sought to better understand its determinants. To address this subject, they analyzed QOL, as measured with the Functional Assessment of Cancer Therapy-General questionnaire (FACT-G), Spanish Version 4, and depicted the complex relations among physical, psychological, social, and cultural factors, including spirituality.
A cross-sectional study design was used with a sample of 309 patients with cancer. The influence of several possible determinants was first studied by univariate regression analysis. Variables showing an association were included in a forward stepwise multivariate regression model.
Five regression models were studied, for the FACT-G total score and its four subscales. Five variables explained 32.1% of the variance of the FACT-G total score: tumor stage, spiritual well-being, income, mood disorders, and mode of questionnaire administration. The type and relevance of the explanatory variables differed among the various dimensions of QOL.
The authors underlined the entwining of biologic, psychosocial, and spiritual factors as determinants of the QOL of patients with cancer, thus supporting the multidimensional definition and modeling of the construct. Cancer 2005. © 2005 American Cancer Society.
Health-related quality of life (QOL) has emerged as a significant medical outcome measure, and its enhancement is an important goal in oncology and other clinical disciplines. Hence, there is a need to understand better the relative contributions of several factors affecting patients' perceptions and reports of well-being and satisfaction with life.
These determinants behave differently in terms of their effects on QOL. Although deficiencies in some factors may lead to poor QOL, abundance may not necessarily lead to very high QOL. For example, when health and income have reached a certain level, very good health and very high income may not result in further QOL improvements. Other determinants, when present, may promote high QOL, but their absence may not cause significant deterioration, e.g., leisure, sports, and art and cultural activities. Therefore, one key strategy in QOL interventions in cancer treatment is to introduce as many desirable or positive determinants as possible and to remove undesirable determinants.1
Several studies have evaluated the impact of clinical and sociocultural factors on the QOL assessments of patients with cancer.2–4 However, there are no published reports on this issue in South American Spanish-speaking populations that could shed light on the particular perceptions of patients with cancer living in underdeveloped economies.
The importance of including spirituality as a relevant determinant of QOL has been recommended.5, 6 Spirituality has become distinguished from religiousness or the practice of religious behavior. Whereas religion is seen as participation in activities of a certain faith group,7 spirituality has been conceptualized as a personal search for meaning and purpose in life, and as beliefs in a transcendent dimension of existence.8 Several studies have evaluated the relation between spirituality and health, and several instruments are now available to measure this dimension in people with chronic and life-threatening illnesses.9, 10
The purpose of the current study was to depict the relations among physical, psychological, and sociocultural factors, including spirituality, and patients with cancer assessments of health-related QOL.
MATERIALS AND METHODS
A cross-sectional study design was used with a convenience sample of Spanish-speaking Uruguayan patients with cancer who ranged in age from 18 to 75 years old. To ensure heterogeneity of socioeconomic status, patients from one private (Centro de Asistencia del Sindicato Médico del Uruguay) and three public hospitals (Hospital de Clínicas de la Universidad de la República, Instituto de Oncología del Ministerio de Salud Pública, and Servicio de Oncología Radioterápica del Hospital Pereira Rossell) in the city of Montevideo were recruited for the study. Potential participants were identified from the daily record of office visits, treatment visits, and inpatient hospitalizations. Clinical eligibility criteria were broad to cover various tumor sites, stages of disease, and treatments. To ensure sufficient experience with treatment-related side effects, patients must have completed a minimum of 2 cycles of chemotherapy (ChT) and/or 10 radiotherapy (RT) sessions, or 1 month of hormone therapy. There must have been ≥ 1 month since receipt of last surgery. Patients with ostensible cognitive deficits or serious psychiatric dysfunctions were excluded. The study was approved by the corresponding ethics committees of the institutions involved. Informed consent was obtained from all patients.
A total of 361 patients were approached and asked to participate. Among them, 36 patients (10%) either refused to participate, were too sick to complete the battery of questionnaires, or a family member prevented the patient from participating. Data collection for 16 other participants (4%) was not completed because of other reasons. The study analyses were based on the remaining 309 patients.
Independent variables were grouped into three categories: biologic/clinical, psychosocial, and mode of questionnaire administration. Biologic/clinical variables were tumor stage, patient location at the time of assessment (outpatient vs. inpatient), performance status, current treatment status, time since diagnosis in months, and type of treatment (e.g., ChT, RT, hormone therapy, or combined treatment). Tumor stage was assessed by the treating oncologist who used the criteria of the TNM classification of malignant tumors.11 The patient's current Eastern Cooperative Oncology Group performance status rating (ECOG-PSR), a 5-point scale12 ranging from 0 (asymptomatic and fully ambulatory) to 4 (not being able to leave bed), was provided by the oncologist or the treating physician. The information regarding type and stage of disease, time since diagnosis, and current and previous treatments was obtained from the oncologist or treating physician and verified by the research assistants using the participants' medical record.
The psychological impact of cancer was assessed using the Profile of Mood States, Short Form version (POMS-SF).13 The POMS-SF is a widely used 30-item rating scale that measures subjective mood states, such as anxiety, tension, vigor, depression, fatigue, and confusion. It is a valid measure of self-reported affective states and psychological adjustment of patients with cancer, and is available in Spanish. A total mood disturbance score (POMS TMD; range: −20 to 100) may be obtained by summing the five scores of Tension, Depression, Anxiety, Fatigue, and Confusion subscales and subtracting the Vigor score from these scores. Only the POMS TMD score was introduced as an independent variable in the analyses.
The psychosocial variables included in the analyses were age, gender, estimated family income, perception of income, years of formal education, living arrangements (living with other(s) vs. living alone), employment status, health coverage (private vs. public hospital), religious affiliation, spirituality, and mood. Spiritual well-being was assessed using the Functional Assessment of Chronic Illness Therapy-Spiritual Well-Being scale (FACIT-Sp). This is a 12-item instrument with 2 subscales—one measures a sense of meaning and peace, and the other assesses the role of faith in illness. A total score for spiritual well-being is also produced (range, 0–48). The FACIT-Sp is a psychometrically sound measure of spiritual well-being for people with cancer and other chronic illnesses that has been validated in both English and Spanish.10 The total FACIT-Sp score was used as a possible explanatory variable of patients' assessments of QOL.
The mode of administration of the questionnaires (e.g., self-administered vs. interviewer administered) was not randomized in our study. Although the Functional Assessment of Cancer Therapy-General questionnaire (FACT-G), the FACIT-Sp, and POMS-SF were designed to be completed by self-administration, patients in our study were free to request the interviewers to read the questionnaires to them. In every case, the patients' preferences and the mode of questionnaire administration were documented.
QOL outcomes were evaluated using the FACT-G Spanish Version 4.14, 15 It comprises 27 questions and assesses 4 dimensions of QOL: physical well-being (PWB; 7 items; range, 0–28), social and family well-being (SFWB; 7 items; range, 0–28), emotional well-being (EWB; 6 items; range, 0–24), and functional well-being (FWB; 7 items; range, 0–28). The FACT-G uses 5-point Likert-type response categories ranging from 0 (“not at all” [“nada”]) to 4 (“very much” [“muchísimo”]). The total FACT-G score is the summation of the 4 subscale scores and ranges from 0 to 108. The FACT-G is a widely used QOL instrument that has been adapted for use with Spanish-speaking patients with cancer, and it has shown content, semantic, and technical equivalence across cultures.16–18 It is written at about the sixth-grade reading level.
Five separate outcome measures were evaluated, which corresponded to the FACT-G total score and the four subscale scores. Univariate regression analyses were first performed for each independent variable. Variables shown to have significant associations in the univariate analysis (P < 0.05) were included in a forward stepwise multivariate regression model. Various methods have been proposed for building multiple regression models. Stepwise methods consist of variations on two basic ideas (forward selection and backward elimination). Forward selection starts with no variables in the model and adds one variable (or one subset of variables) at a time until a stopping criterion is satisfied. In backward elimination, the model starts with all variables and eliminates one variable (or subset) at a time. Stepwise multiple regression is especially useful when many variables are involved to help determine which are independent predictors of the outcome.
For the analyses, nominal variables were added as indicator or dummy variables. The tumor stage categories in situ, Stage I, II, and III tumors were combined as locoregional disease for comparison to Stage IV metastatic disease. ECOG-PSR scores were included as a five-level, Likert-type scale. Treatment was included as ChT versus other, RT versus other, and combined ChT and RT versus other. Hormone therapy was excluded because of the few patients who received this type of treatment. Dummy variables also were created for several dichotomous variables: patient location (outpatient vs. inpatient), receipt of treatment (< 3 months since last treatment vs. ≥ 3 months), gender (male vs. female), educational level (higher than a 6th-grade of primary school vs. lower than a 6th grade level of education), living arrangement (lives with other[s] vs. alone), employment status (employed vs. other), health coverage (private hospital vs. other), religious affiliation (yes vs. no), and mode of questionnaire administration (self-administered vs. interviewer administered). Perception of income was dichotomized as high satisfaction with income (“I get enough” and “I can save”) versus lower satisfaction (“I am in need,” “I don't get enough for my needs, ” and “I get just fair for my needs”).
Collinearity diagnostics were performed by means of the variance inflation factors (VIF) for each independent variable entered in the five regression equations. A rule of thumb for evaluating VIFs is to be concerned with any value > 10.0.19
Cronbach's coefficient alpha was calculated for the FACT-G and FACIT-Sp scales as a measure of internal consistency reliability. Values > 0.70 are generally acceptable for group data.
Patient sociodemographic and clinical characteristics are summarized in Table 1. Although most patients had advanced-stage disease (80% of patients had regional or metastatic disease), only 22% reported needing bed rest (ECOG-PSR 2–4). The sample was somewhat socioeconomically disadvantaged—with 21% were currently unemployed, 27% had not reached the 6th year of formal education, a median monthly income of only $300, and 57% reported inadequate income for their needs. As many as 41.7% of patients in the study sample reported no current religious affiliation, which is a typical characteristic of Uruguayan culture, yet quite dissimilar to other South American countries. The mean spirituality (FACIT-Sp) score was 32.4, which is slightly lower than the mean score of 38.5 reported for English-speaking patients with cancer.10 The FACIT-Sp demonstrated good evidence of internal consistency and reliability (Cronbach's alpha = 0.75).
|Characteristics||No. of patients (%)||Characteristics||No. of patients (%)||Characteristics||No. of patients (%)|
|Tumor stage||Treatment||Marital status|
|In situ||2 (0.7)||ChT only||95 (30.7)||Married||164 (53.1)|
|Local||59 (19.3)||RT only||98 (31.7)||Unmarried couple||20 (6.5)|
|Regional||129 (42.3)||Hormone therapy only||11 (3.6)||Divorced||27 (8.7)|
|Metastatic||115 (37.7)||ChT plus RT||100 (32.4)||Separated||16 (5.2)|
|Other combinations||5 (1.6)||Widow||37 (12.0)|
|Tumor site||Single||45 (14.6)|
|Lung||44 (14.2)||Mean age||56.7 (SD 12.9)||Living arrangement|
|Colorectal||35 (11.3)||Lives alone||36 (11.7)|
|Head and neck||24 (7.8)||Gender||Lives with family||259 (83.8)|
|Uterus and cervix||20 (6.5)||Female||170 (55.0)||Lives with others||11 (3.6)|
|Prostate||18 (5.8)||Lives in institution||3 (1.0)|
|Gastrointestinal||12 (3.9)||Family monthly income (US $)||Employment|
|Urothelial||12 (3.9)||Mean||512||Full-time||43 (14.1)|
|Ovary||8 (2.6)||Median||300||Part-time||17 (5.6)|
|Lymphoma||7 (2.3)||Range||0–5000||Retired||107 (35.1)|
|Primitive unknown||7 (2.3)||Interquartile range||192–571||Unemployed||64 (21.0)|
|Soft tissue||4 (1.3)||Medical leave||22 (7.2)|
|Melanoma||3 (1.0)||Perception of income||Housekeeping||40 (13.1)|
|Brain||1 (0.3)||“ I am in need”||38 (12.4)||Others||12 (3.9)|
|Patient location||“I don't get enough for my needs”||136 (44.3)|
|Inpatient||86 (26.5)||“I get just fair for my needs”||100 (32.6)||Health coverage|
|Outpatient||223 (68.6)||“I get enough, no money problems”||23 (7.5)||Private hospital||97 (31.4)|
|“I can save”||10 (3.3)|
|0||133 (45.4)||Primary school < 3 yrs||36 (11.7)||Yes||180 (58.3)|
|1||94 (32.1)||Primary school 3–5 yrs||46 (14.9)|
|2||37 (12.6)||Primary school 6 yrs||101 (32.8)||Mean Spirituality||32.4 (SD 8.2)|
|3||20 (6.8)||High school 1–3 yrs||51 (16.6)||(FACIT-Sp)||10.9 (SD 17.4)|
|4||9 (3.1)||High school 4–6 yrs||32 (10.4)||Mean POMS TMD|
|Receiving treatment||183 (59.2)||Mode of questionnaire administration|
|Interview administered||205 (67.7)|
|Mean time since diagnosis (mos)||29 (SD 39.5)||Self-administrated||98 (32.3)|
Table 2 contains the descriptive statistics for the FACT-G total and its subscales. All scales demonstrated good evidence of internal consistency and reliability. The mean PWB, EWB, and FWB scores were a few points lower (poorer) than those reported by the initial Spanish language translation sample.15
|Subscale||FACT-G Spanish version 4|
|Mean (SD)||Alpha coefficients|
|PWB (7 items)||15.7 (8.9)||0.91|
|SFWB (7 items)||18.6 (5.9)||0.81|
|EWB (6 items)||13.5 (6.3)||0.78|
|FWB (6 items)||16.0 (6.0)||0.81|
|FACT-G total||63.7 (19.0)||0.89|
Table 3 shows the simple regression model results for the FACT-G total score and the four dimension scales. Most of the independent variables were associated significantly with at least one of the five QOL outcome measures. No significant differences in scale scores were observed for time since diagnosis, gender, living arrangement, and religious affiliation.
|Biologic variables||Physical well-being||Social family well-being||Emotional well-being||Functional well-being||Total FACT|
|β (SE)||P value||β (SE)||P value||β (SE)||P value||β (SE)||P value||β (SE)||P value|
|Locoregional||3.82 (1.02)||< 0.000||−2.27 (0.69)||0.74||1.96 (0.74)||0.008||1.74 (0.70)||0.01||7.29 (2.16)||0.001|
|Outpatient||1.16 (1.12)||0.30||2.85 (0.71)||< 0.000||0.99 (0.80)||0.22||4.43 (0.71)||< 0.000||9.43 (2.30)||< 0.000|
|ECOG-PSR||−1.04 (.48)||0.03||−0.85 (0.32)||0.008||−0.73 (0.35)||0.03||−2.42 (0.30)||< 0.000||−5.04 (0.99)||< 0.000|
|Receiving treatment||−0.69 (1.03)||0.50||0.05 (0.67)||0.94||−1.06 (0.73)||0.14||1.87 (0.68)||0.007||0.16 (2.15)||0.94|
|Month from diagnosis||−0.008 (0.01)||0.52||0.01 (0.008)||0.23||−0.01 (0.01)||0.12||0.01 (0.01)||0.10||0.001 (0.03)||0.95|
|ChT||−2.61 (1.04)||0.01||1.61 (0.68)||0.02||−2.23 (0.74)||0.003||0.90 (0.70)||0.20||−2.33 (2.19)||0.29|
|RT||1.12 (1.06)||0.29||−1.69 (0.68)||0.01||0.80 (0.75)||0.28||−1.18 (0.71)||0.01||−0.94 (2.22)||0.67|
|Combined ChT - RT||−1.05 (1.08)||0.32||−1.20 (0.70)||0.78||−0.97 (0.77)||0.20||−0.62 (0.72)||0.39||−2.85 (2.26)||0.20|
|Psychosocial variables||Statistics||P value||Statistics||P value||Statistics||P value||Statistics||P value||Statistics||P value|
|Age||0.05 (0.04)||0.15||0.06 (0.02)||0.02||0.04 (0.03)||0.13||−0.01 (0.03)||0.58||0.02 (0.08)||0.79|
|Gender||1.94 (1.01)||0.06||−0.57 (0.66)||0.39||0.81 (0.72)||0.26||0.98 (0.68)||0.15||3.16 (2.12)||0.14|
|Family income||0.0002 (0.000)||0.001||0.0001 (0.000)||0.002||0.0001 (0.000)||0.02||0.0002 (0.000)||< 0.000||0.0006 (0.000)||< 0.000|
|Perception of income “ I get enough” and “I can save”||2.25 (1.63)||0.17||2.45 (1.05)||0.02||0.89 (1.17)||0.44||4.86 (1.06)||< 0.000||10.45 (3.39)||0.002|
|Education||0.88 (1.14)||0.44||2.24 (0.73)||0.002||0.63 (0.81)||0.44||1.68 (0.76)||0.03||5.42 (2.38)||0.02|
|Living arrangement||0.65 (1.57)||0.68||1.64 (1.02)||0.11||−1.10 (1.12)||0.33||−0.45 (1.06)||0.67||0.74 (3.30)||0.82|
|Employment status||2.56 (1.20)||0.03||2.09 (0.78)||0.008||1.75 (0.86)||0.04||3.11 (0.80)||<0.000||9.51 (2.48)||< 0.000|
|Employed Health coverage Private hospital||1.76 (1.09)||0.11||0.54 (0.71)||0.45||0.87 (0.78)||0.26||2.32 (0.72)||0.002||5.49 (2.27)||0.02|
|Religious affiliation||1.62 (1.02)||0.11||−0.92 (0.66)||0.17||1.10 (0.73)||0.13||−0.88 (0.69)||0.20||0.92 (2.15)||0.67|
|Spirituality||0.06 (.06)||0.35||0.32 (0.04)||<0.000||0.11 (0.04)||0.01||0.44 (0.03)||<0.000||0.93 (0.12)||< 0.000|
|POMS TMD||−0.05 (.03)||0.08||0.08 (0.02)||<0.000||0.06 (0.02)||0.003||−0.18 (0.02)||<0.000||−0.37 (0.06)||< 0.000|
|Mode of administration||Statistics||P value||Statistics||P value||Statistics||P value||Statistics||P value||Statistics||P value|
|Self-administered||1.17 (1.09)||0.28||3.90 (0.68)||< 0.000||0.04 (0.77)||0.96||3.17 (0.71)||< 0.000||8.28 (2.25)||< 0.000|
After evaluating the correlations among the independent variables, no multicollinearity problem was detected. In addition, the VIF for each independent variable entered in the five regression models indicated the absence of multicollinearity. These data (not shown) are available upon request from the author.
Table 4 shows the final multivariate regression models. Both unstandardized and standardized beta coefficients for each explanatory variable with the corresponding t statistics are displayed. The 5 models are each statistically significant at the P < 0.05 level.
|Physical well-being||Social family well-being||Emotional well-being||Functional well-being||Total FACT|
|Unstandardized coefficients β (SE)||Standardized coefficients β (t statistics)||Unstandardized coefficients β (SE)||Standardized coefficients β (t statistics)||Unstandardized coefficients β (SE)||Standardized coefficients β (t statistics)||Unstandardized coefficients β (SE)||Standardized coefficients β (t statistics)||Unstandardized coefficients β (SE)||Standardized coefficients β (t statistics)|
|Tumor stage Locoregional||4.03 (1.14)||0.226 (3.52)b||—||—||—||—||—||—||5.89 (2.22)||0.152 (2.66)b|
|ECOG-PSR||—||—||−1.27 (0.28)||−0.231 (−4.54)c|
|ChT||−2.34 (1.16)||−0.129 (−2.02)a||—||—||−2.07 (0.853)||−0.159 (−2.43)a||—||—||—||—|
|RT||—||—||−1.34 (0.68)||−0.111 (−1.98)a||—||—||—||—||—||—|
|Income||0.0002 (0.000)||0.218 (3.41)b||—||—||—||—||0.0001 (0.000)||0.163 (3.32)a||0.0005 (0.000)||0.234 (3.92)c|
|Spirituality||—||—||0.30 (0.04)||0.416 (7.44)c||—||—||0.31 (0.04)||0.413 (7.88)c|
|POMS TMD||—||—||—||—||−0.07 (0.02)||−0.185 (−2.82)b||−0.08 (0.02)||−0.238 (−4.37)c|
|Mode of administration|
|Self-administered||—||—||4.24 (0.67)||0.356 (6.33)c||—||—||1.56 (0.61)||0.125 (2.56)a|
|Model F||4.08; P < 0.05||3.94; P < 0.05||5.89; P < 0.05||6.57; P < 0.051|
Five explanatory variables explained 32.1% of the variance of the FACT-G total score regression model. These are (in order of decreasing R2 change values): spiritual well-being subscale (R2 change, 0.177), income (R2 change, 0.082), tumor stage (locoregional disease; R2 change, 0.033), POMS TMD (R2 change, 0.014), and mode of questionnaire administration (R2 change, 0.015). The regression model for the PWB subscale accounted for only 11.6% of the variance with 3 variables having significant coefficients: tumor stage (locoregional; R2 change, 0.054), income (R2 change, 0.045), and ChT (R2 change, 0.017). As for the SFWB subscale, 32.1% of the total variance was explained by 3 independent variables: spiritual well-being (R2 change, 0.175), mode of questionnaire administration (R2 change, 0.133), and RT (R2 change, 0.012). The regression model for the EWB subscale explained 6.0% of the total variance with significant beta coefficients for POMS TMD (R2 change, 0.035) and ChT (R2 change, 0.025). The explanatory model for the FWB subscales accounted for 55.6% of the variance. The important predictors were spiritual well-being (R2 change, 0.345), ECOG-PSR (R2 change, 0.126), income (R2 change, 0.028), POMS TMD (R2 change, 0.034), and mode of questionnaire administration (R2 change, 0.014).
To evaluate further the impact of the mode of questionnaire administration, we performed a post-hoc t-test analysis of the SFWB and FWB subscales. A statistically significant mean difference was found for every item, such that patients who answered by self-administration reported higher levels of well-being. The frequency distributions of responses to the rating categories of each item were skewed to the left, regardless of the mode of administration, but this was more evident for the self-administration group. Thus, even when both groups were more likely to give positive responses, interviewed patients tended to moderate their responses in the presence of the interviewer, resulting in lower use of the category very much (muchísimo). The relation between mode of questionnaire administration and other variables also was reported elsewhere.18 In our study, we found that patients who preferred the self-administered questionnaire were younger (P < 0.001), had a higher educational level (P < 0.001), and had a better ECOG-PSR (P < 0.001) compared with patients who preferred the interviewer-administered questionnaire. No differences in gender were observed.
There is an increasing need for a better understanding of the determinants underlying patient-reported QOL, their multiple and complex interactions, and strategies to assist clinicians in ways to enhance patient well-being.1 The current study evaluated several biologic, psychosocial, and cultural variables as possible determinants of QOL as measured by the FACT-G total and four subscale scores. The five multivariate regression models identified different sets of variables as significant determinants of patients' reports. Tumor stage was the only clinical variable that proved to be a relevant determinant of patients' ratings on the FACT-G overall measure of QOL. It was also an influencing clinical factor on PWB. Specifically, patients with locoregional diseases (vs. advanced-stage cancer) reported better QOL. Similar results were described elsewhere.2, 3 ECOG-PSR was a relevant determinant only for FWB, with higher (poorer) PSR associated with lower (poorer) FWB. As for treatment modality, ChT affected physical and EWB ratings, and RT had a modest but significant effect on SFWB. No significant effects were observed for ChT, RT, or combined treatments on patients' assessments of overall QOL and FWB. These findings are consistent with the results of a recent study that found that the type of treatment for breast carcinoma was not related to long-term physical or emotional role function.20 Conversely, it was the process of care, particularly having a choice of treatments, and the perceptions of care, which were associated with better long-term general physical and mental function, impact of the disease, and satisfaction with medical care.20
In our study, age was not significantly associated with four of the five QOL outcome measures. It was modestly associated with SFWB, but only before adjustment for other factors. Some other studies have found that older patients with cancer reported better PWB and overall QOL.2, 3 It has been suggested that younger patients may have higher expectations concerning their physical and functional status.21 If true, these previous findings are conceptually consistent with Calman's definition of QOL as the gap between personal expectations and the individual's actual experience.22
The amount of family income proved to be a relevant positive determinant of PWB, FWB, and overall QOL. Considering our sample of economically disadvantaged patients, the level of satisfaction with basic needs might have a very important effect on general well-being. In future studies of QOL in nonindustrialized countries, a more accurate assessment of the level of satisfaction with basic needs as well as higher needs of love and belongingness, self-esteem and self-actualization,23 and indicators of social risk should be included. Private hospital health coverage was not significantly associated with better QOL after adjustment for other factors.
Cancer often is associated with a great deal of psychological distress and, in turn, mood disturbance affects QOL. In our study, psychological distress exerted a significant negative effect on the EWB and FWB domains and overall QOL.
An important aspect of the current study refers to the inclusion of spiritual well-being as a potentially influencing factor on patients' QOL. Several studies have addressed the issue of the influence of personal attitudes, beliefs, and values on medical patients' QOL assessments.2, 7, 8 Spiritual well-being has been considered either as a manifestation of ego strength, as an adequate coping strategy leading to better QOL assessments, or as a consequence of better QOL, which enables a person to be more optimistic and strongly spiritually-oriented. We found spiritual well-being to be an important determinant of patients' assessments of overall QOL, SFWB, and FWB. These findings are especially relevant considering the sample characteristics regarding religious affiliation. Most studies that have assessed the role of religiosity or spirituality in medical patients have been carried out in the United States where the rates of religious affiliation and practices are very high, with ≤ 81 % of the adult population stating that they identified with one religion.24 It is noteworthy that the influence of spiritual beliefs and well-being remained relevant even when 41.7% of the studied patients stated they did not profess any religion, which is consistent with findings in other studies.25, 26 Spiritual characteristics might be considered as being permanently embedded in a person's repertoire of coping mechanisms, thus leading to better reports of QOL, or as being dependent on fluctuating mood and health status, with patients with more physical and emotional impairment also having worse spiritual well-being.7 We cannot make assumptions on the direction of causation based on these results.
In our study, the mode of questionnaire administration had a significant effect on patients' ratings on the FACT-G and the SFWB and FWB subscales. We found that interviewed patients were less likely to use the category very much (muchísimo) in the SFWB and FWB subscales. One reason may be related to the finding that all the items in these subscales are positively worded so that endorsing the higher scores corresponds to higher levels of well-being, whereas the other subscales contain mostly reversed items. This, together with the finding that these subscales contain items referring to very sensitive aspects such as general contentment and enjoyment with existence and family, partner and sex life, might explain why interviewed patients were less likely to give a positive response of very much (muchísimo). Other explanations may be entertained. In a previous validation study using the same sample of patients,18 we found that lower educated patients (vs. patients with more education) reported lower levels of QOL in univariate analyses. Concurrently, lower educated, older, and more functionally impaired patients more frequently demanded interviews to fill out the questionnaire. Thus, it is possible that this more vulnerable group of patients reporting lower QOL would request interviews as a means for contact and support. A more specific study design would be necessary to depict the complex relations among education, age, performance status, mode of questionnaire administration, and QOL assessments. Some research has been conducted to evaluate the effects of mode of questionnaire administration on QOL or health status scores. Research has been motivated primarily by the need to demonstrate that different methods of administration will not bias the measurement of patient self-report data. Several studies to date have focused on interview (in-person or telephone) versus self-administration (in the clinic or by mail) of a paper-and-pencil questionnaire.27–35 Most studies found high reliability for scales administered with different methods, but response effects varied and were not consistently in the same direction. As in our case, some studies found evidence of more favorable reports of well-being on the self-administered questionnaires,27, 34 whereas others found the opposite effect.31, 32 However, the majority of studies found no differences due to the mode of administration.28, 29, 35–40 Other studies in U.S. Spanish-speaking populations found no significant effect of the mode of administration of the FACT-G on the reporting of overall QOL.3
The current study identified multiple determinants of QOL in patients with cancer. These findings confirmed the multidimensional nature of the construct by showing the intertwining of clinical, psychosocial, and cultural factors.1, 41 Of special interest was the association between spiritual well-being and QOL, even in a population with high rates of agnosticism. Other possible influencing factors such as prevalent physical symptoms commonly associated with the disease and its treatment (e.g., anemia and fatigue) or symptom burden were not evaluated in the current study and should be included in future research.
The absence of published data on the determinants of QOL in South American patients with cancer prevents us from making assumptions on the generalizability of the results. A previous study carried out among U.S. Hispanic patients with cancer reported similar significant variables such as disease stage, performance status, socioeconomic status, and spiritual beliefs, and differed in others such as age, gender, and living arrangements.3 However, considering the cultural, ethnic, and socioeconomic similarities, we might expect similar findings in other countries of the Latin America Southern Cone (Argentina, Chile, and south of Brazil). This would be a relevant hypothesis for future transnational studies.
The authors thank Professors Enrique Barrios, M.D., Ricardo Bernardi, M.D., Ignacio Musé, M.D., and the staff of the participating institutions for their cooperation and support.
- 12002, 7: 4–5. Available from URL: http://www.isoqol.org [accessed April 30, 2004].. A system model of quality of life for rehabilitation and psychosocial care. ISOQOL Newsletter [serial online]
- 11SobinLH, WittekindCh, editors. TNM classification of malignant tumors, 5th ed. New York: John Wiley & Sons, 1997.
- 12Appraisal of methods for the study of chemotherapy of cancer in man: comparative therapeutic trial of nitrogen mustard and triethylene thiophosphoramide. J Chron Dis. 1960; 11: 7–33., , , et al.
- 13Manual of the Profile of Mood States. San Diego: Educational and Industrial Testing Service, 1992., , .
- 19Applied regression analysis and other multivariable methods. 2nd edition. Belmont, CA: Duxbury Press, 1988., , .
- 22Definitions and dimensions of quality of life. In: AaronsonNK, BeckmannJH, editors. The quality of life of cancer patients. New York: Raven Press, 1987: 1–11..
- 23El hombre autorealizado. Hacia una Psicología del ser [Toward a psychology of being]. Barcelona: Kairos, 1974..
- 24American religious identification survey 2001 [monograph online]. New York: The Graduate Center of the City University of New York, 2001. Available from URL: http://www.gc.cuny.edu/studies/aris.pdf [accessed April 30, 2004]., , .
- 25Creencias y religiones. La religiosidad de los montevideanos al fin del milenio. [Beliefs and religions of Montevideons at the end of the millennium]. Montevideo: Editorial Trilce, 1996., , .
- 26FACTUM. Encuesta FACTUM de Opinión Pública: entre devotos y no devotos [FACTUM public opinion poll: devotees vs. non-devotees]. [monograph online]. Montevideo: FACTUM Digital, 2000. Available from URL: http://www.factum.com.uy/encuesoc/finsem/2000/efs00003a.html [accessed November 20, 2003].
- 30European Organization for Research and Treatment of Cancer Study Group on Quality of Life. The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst. 1993; 85: 365–376., , , et al. for the
- 32Mode effects in a survey of Medicare prostate surgery patients. Proc Surv Methods Section Am Sociol Assoc, 1993: 730–735., , .
- 36A critical comparison of three strategies of collecting data from households. J Am StatAssoc. 1967; 62: 976–989..
- 38Survey errors and survey costs. New York: John Wiley & Sons, 1989..
- 39Survey research methods 2nd edition. Newbury Park, CA: Sage, 1993..
- 40Data collection methods. In: SpilkerB, editor. Quality of life and pharmacoeconomics in clinical trials. 2nd edition. Philadelphia: Lippincott-Raven Publishers, 1996: 381–386..