Search

Research into the psychosocial consequences of cancer and cancer treatment is a rapidly expanding field of inquiry. To gain an overview of the nature and scope of such research in Europe, a survey was undertaken of the membership of the European Organization for Research and Treatment of Cancer Study Group on Quality of Life. The questionnaire yielded information regarding (1) the type of research being undertaken (clinical trial, descriptive, evaluative, or methodological study), (2) the types of cancer being studied, (3) sample sizes and the sociodemographic characteristics of the samples, (4) the aspects of the quality of life assessed, (5) the nature of the data-collection techniques employed, and (6) the range of methodological and practical problems encountered in conducting research in psychosocial oncology. Implications of the findings for the conduct of psychosocial research in the cancer field are discussed.

This study examines the nature, pattern, and consequences of pain experienced by cancer patients, both during hospitalization and during the immediate posthospitalization period. Of the 240 hospitalized cancer patients screened, 45% were found to be in pain. Although 59% of these patients had found ways to reduce their pain, nearly one third reported being seriously limited in their daily activities. Sleep disorder and mood disturbance also were related to the experience of pain. Applied pain management strategies were rated by two physicians as inadequate in 54% of cases. The most frequent recommendation for improved pain management was an increase in the dosage or a change in the class of analgesics. The importance of periodic assessment of pain and its consequences, both physical and psychological, is discussed in light of increasing the level of information available to physicians and patients regarding appropriate approaches to pain management.

During a 3-month recruitment period, a smoking cessation program was provided to all pregnant women who attended an HMO prenatal clinic and reported either that they were currently smokers ( n = 35) or that they had quit at pregnancy onset (n = 23). Except for its in-clinic introduction, the behavioral smoking cessation program used a home-correspondence format with seven weekly mailings and a telephone answering system adjunct. At follow-up, all 23 women who identified themselves at baseline as self-initiated quitters reported continued abstinence from smoking. Urine thiocyanate assays confirmed these self-report data in 82% of the cases. Results of a multivariate analysis identified three factors—gravida, number of years a smoker, and number of friends who smoke—as significant predictors of self-initiated quitting. Of the 35 women who indicated that they were still smoking at baseline, 10 (28.5%) reported that they had stopped smoking at the time of the postprogram assessment. Available urine thiocyanate assays confirmed a nonsmoking status in all cases. The multivariate results indicated that four variables—gravida, severe nausea, baseline smoking rate, and spouse's smoking status—were significant predictors of quitting following exposure to the program. The majority of both self-initiated quitters and postprogram quitters reported that the program was useful in helping them to quit smoking and/or to remain abstinent over time. These results lend support to the viability of a home-correspondence, self-help program as a means for facilitating smoking cessation and continued abstinence from smoking during pregnancy.

Previous research has documented the fact that policy makers underutilize social science research literature in shaping their decisions. The two factors most often identified as accounting for this utilization gap are quality of research and organizational/political constraints. A third factor that has received relatively less attention is barriers to accessing relevant literature. To determine the importance ofaccessproblems among mental health professionals, a national sample (N = 367) of community mental health center (CMHC) executives was surveyed regarding attitudes and practices with respect to published social science research. Thefindings revealed that CMHC policy makers (1) consider published literature as important relative to other sources of information; (2) have difficulty accessing desired information via traditional means (e.g., personal and institutional librariesf, and (3) are generally unfamiliar with currently available information accessing technologies (i.e., computer-based bibliographic retrieval systems). Thesefindings suggest the need for greater emphasis on access barriers in a model predicting utilization of social science research in mental health policy formulation.

This article reviews procedural and methodological characteristics of 121 outcome studies published by community mental health centers during the decade 1969-1979. These studies were compared with 2906 studies conducted at other mental health treatment sites. Attention is focused on the adequacy of description of sample characteristics, mental health condition treated and treatment or intervention employed, sophistication of study design, and quality of outcome measures. The findings suggest that CMHC studies are less precise and less rigorous than studies at other sites, although there appears to be consistent improvement in research quality over time. The findings raise the question of whether CMHCs generally should conduct high-level outcome research, aside from essential internal program monitoring. A feasible alternative may befor such research to be carried out at a few centers of mental health research excellence and for such findings to be disseminated to the mental health community at large.

This paper reviews a range of issues related to the assessment of subjective response and quality of life in prostate cancer clinical research. With regard to subjective response criteria, the Karnofsky performance status scale and the subjective components of the World Health Organization acute and subacute toxicity scales appear to hold certain advantages over competing measurement systems. Nevertheless, the available evidence suggests that further developmental work is needed to improve the precision of these instruments. In the area of quality of life assessment, there does not appear to be a clear choice among the array of available measures. Although there are several promising instruments, none has undergone sufficient field testing to justify widespread adoption in clinical research settings. A number of suggestions are offered to facilitate further development in quality of life instrumentation and research implementation.

In a previous study it was found that mental health professionals working within community mental health centers (CMHCs) value published social science research as input into the policy decision-making process, yet frequently experience problems in accessing such literature in a timely and efficient manner. The current study was undertaken in order to determine whether access barriers could be reduced via the use of computer-based literature searching. Both short-term (six week) and long-term (six month)follow-up data revealed substantial use of the search materials for both conceptual and instrumental purposes. Further, the positive experiences reported with regard to the study-specific search were mirrored in a more general assessment of the potential usefulness of such information-retrieval systems in meeting future information needs.

Background The European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (QLQ-C30) is a widely used quality-of-life measure in oncology. The ability to translate QLQ-C30 responses into utility scores would further expand its use in medical decision-making. The aims of this study were to: 1) map QLQ-C30 responses onto patient time trade-off utility scores; and 2) compare a multiattribute approach to a global evaluation approach to modeling utility scores. Methods Two distinct approaches were applied to data from 1432 cancer patients. The multiattribute approach used psychometric analysis and expert input to select a subset of functioning and symptom scale items for modeling. The second approach focused on global health and quality-of-life items based on a conceptual model. Model selection criteria included parsimony, statistical significance and logical consistency of parameter estimates, predictive accuracy, number of states described, and scale range. Results The optimal multiattribute model included nine variables for five items from different scales, described 144 unique states, predicted values ranging from 0.63 to 1.00, but it had poor predictive accuracy (cross-validation pseudo- R 2 = 0.056). The best-fitting global approach-based model described 24 unique states using eight indicators for two items from one scale (plus a constant) and predicted values ranging from 0.17 to 1.00 (cross-validation pseudo- R 2 = 0.127). Conclusions Multiattribute models produced a greater number of unique predicted values, while global models exhibited more desirable statistical properties and a wider range of values. The recommended models will enable users to predict cancer patients' utilities from existing and future QLQ-C30 data sets.

Purpose This study investigated adherence to treatment guidelines on cancer-related anaemia and fatigue (CRA/CRF) and factors influencing the choice of intervention.Methods In this prospective, observational study, 136 cancer patients being treated with chemotherapy in a large community hospital completed a questionnaire at consecutiveoutpatient visits assessing fatigue (the Functional Assessment of Chronic Illness Therapy—Fatigue) and fatigue-related counselling and advice received. Data on administration of chemotherapy and use of epoetin or blood transfusions were abstracted from the medical records.Results Fifty-three percent of patients with severe anaemia (HbConclusions In daily clinical practice, guidelines concerning the use of epoetin or blood transfusion in severe CRA are adhered to in about half of the cases. In patients with less severe anaemia, the level of fatigue did not play a significant role in the use of epoetin. According to current guidelines, counselling on CRF should be directed primarily at activity enhancement. However, only a minority of patients receive such counselling.

Cancer clinical trials focus upon the evaluation of such biomedical outcomes as duration of survival, retardation of the disease process and control of major physical symptoms (Buyse et al. 1984; Friedman et al. 1985). In recent years, however, these clinical end points have been criticized for a lack of comprehensiveness (Greer 1984; Greer and Silberfarb 1982). Following the lead of the WHO in defining health as complete physical, mental and social well-being, the scope of investigation has been extended to include assessment of a range of psychosocial variables that can be subsumed under the heading “quality of life.”

The current interest in incorporating psychosocial outcomes into health care research raises a number of conceptual and methodological issues regarding the assessment of health-related quality of life. This paper offers an overview of some of the principal factors to consider in choosing or developing a quality of life measure, including: (a) the bandwidth-fidelity problem; (b) generic versus disease-specific focus; (C) level of data aggregation; (d) interview versus questionnaire; (e) question time-frame; (f) response scales; (g) psychometric properties; and (h) feasibility. While the discussion focuses on quantitative approaches to quality of life assessment, it is suggested that a mixture of quantitative and qualitative strategies may be most fruitful.

The Halsted-Meier approach has been the standard treatment of early breast cancer for over 100 years. The efficacy of any new therapy must be weighed carefully against this classical approach before it can be considered a viable treatment alternative. This typically means the carrying out of prospective, randomized clinical trials. Yet, even with the aid of a randomized trial, evaluation of the relative efficacy of new approaches to the treatment of early breast cancer may be difficult. While we are quick to note the methodological limitations inherent in research employing nonexperimental designs, it is less often recognized that the use of a randomized design does not solve all of our interpretive problems. Rather, the randomized clinical trial provides us only with a sound starting point for comparative clinical evaluations. To ensure the robustness of our results, we must consider carefully the numerous pitfalls to which even this most elegant of research models may be prone.

The traditional evaluation end points employed in cancer clinical research include such biomedical outcomes as frequency of local recurrence, development of metastatic disease, length of disease-free and overall survival, and control of major physical symptoms. In recent years, however, there has been a growing recognition of the need to document the extent to which medical interventions impact on the functional, psychological, social, and economic life of the individual. Consideration of these nonmedical parameters of cancer and its treatment has been alternately termed “psychosocial,” “sociomedical,” or, most commonly, “quality of life,” research.

In recent years the assessment of subjective response criteria and of quality of life (QoL) has become a matter of increasing interest in the clinical management of cancer patients (Aaronson 1986; Fayers and Jones 1983; Selby et al. 1984; Orr and Aisner 1986). The individual’s physical, functional, psychological and social well-being is recognised as a significant endpoint of cancer treatment. On the one hand, this is due to the awareness that only limited therapeutic improvement has been achieved in the most prevalent malignancies in terms of increased cure rate and prolonged survival. On the other hand, for the curable cancer diseases (i.e. childhood cancer, malignant Hodgkin’s disease, testicular cancer), the physician can today choose.between several more or less intensive treatment options, with seemingly equivalent therapeutic results but with differing QoL and toxicity effects.

Background: The National Cancer Institute’s Symptom Management and Health-Related Quality of Life Steering Committee held a clinical trials planning meeting (September 2011) to identify a core symptom set to be assessed across oncology trials for the purposes of better understanding treatment efficacy and toxicity and to facilitate cross-study comparisons. We report the results of an evidence-synthesis and consensus-building effort that culminated in recommendations for core symptoms to be measured in adult cancer clinical trials that include a patient-reported outcome (PRO).Methods: We used a data-driven, consensus-building process. A panel of experts, including patient representatives, conducted a systematic review of the literature (2001-2011) and analyzed six large datasets. Results were reviewed at a multistakeholder meeting, and a final set was derived emphasizing symptom prevalence across diverse cancer populations, impact on health outcomes and quality of life, and attribution to either disease or anticancer treatment.Results: We recommend that a core set of 12 symptoms—specifically fatigue, insomnia, pain, anorexia (appetite loss), dyspnea, cognitive problems, anxiety (includes worry), nausea, depression (includes sadness), sensory neuropathy, constipation, and diarrhea—be considered for inclusion in clinical trials where a PRO is measured. Inclusion of symptoms and other patient-reported endpoints should be well justified, hypothesis driven, and meaningful to patients.Conclusions: This core set will promote consistent assessment of common and clinically relevant disease- and treatment-related symptoms across cancer trials. As such, it provides a foundation to support data harmonization and continued efforts to enhance measurement of patient-centered outcomes in cancer clinical trials and observational studies.

PURPOSE: To evaluate the antitumor activity and toxicity of a new steroidal aromatase inactivator, exemestane, in postmenopausal women with metastatic breast cancer who had progressive disease (PD) after treatment with a nonsteroidal aromatase inhibitor. PATIENTS AND METHODS: In this phase II trial, eligible patients were treated with exemestane 25 mg daily (n = 241) followed, at the time PD was determined, by exemestane 100 mg daily (n = 58). RESULTS: On the basis of the intent-to-treat analysis by independent review, exemestane 25 mg produced objective responses in 6.6% of patients (95% confidence interval [CI], 3.8% to 10.6%) and overall success (complete response + partial response + no change for 24 weeks or longer) in 24.3% (95% CI, 19.0% to 30.2%). The median durations of objective response and overall success were 58.4 weeks (95% CI, 49.7 to 71.1 weeks) and 37.0 weeks (95% CI, 35.0 to 39.4 weeks), respectively. Increasing the dose of exemestane to 100 mg upon the development of PD produced one partial response (1.7%; 95% CI, 0.0% to 9.2%). Both dosages were well tolerated and were discontinued because of adverse events in only 1.7% of patients. CONCLUSION: Exemestane 25 mg once daily seems to be an attractive alternative to chemotherapy for the treatment of patients with metastatic breast cancer after multiple hormonal therapies have failed.

The aim of the study was to evaluate activity, toxicity and health-related quality of life (HRQL) with gemcitabine as second-line treatment after previous chemo- or radiotherapy in non-small-cell lung cancer (NSCLC). Patients with previously treated NSCLC were treated with gemcitabine (1000 mg/m(2)) on days 1, 8 and 15 in a 28-day cycle. Eighty patients were included; median age was 57 years (range 38-77). Prior treatment consisted of platinum-containing chemotherapy in 29 patients and high-dose thoracic radiotherapy in 51 patients. Median number of cycles was three (range 1-6). Granulocytopenia CTC grade 3 and 4 occurred in 9% and thrombocytopenia CTC grade 3 and 4 in 9% of cycles. Non-haematological toxicity was mild. Tumour response was achieved in 13% of the patients (95% CI 7-20), median survival time was 26 weeks and 1-year survival was 22%. Tumour response to second-line gemcitabine could not be predicted from response to first-line therapy, first-line treatment modality or treatment interval. In a subset of 35 patients HRQL was assessed with the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 and QLQ-LC13 questionnaires and showed improvement or control of symptoms and functioning in approximately 30% of patients. We conclude that gemcitabine in second-line treatment has modest anti-tumour activity, is well tolerated, and may control tumour-related symptoms and improve HRQL in a significant minority of patients.

We are grateful for the opportunity to respond to Cameron McIntosh, and we do so on behalf of all of the authors. Chad Gundy, the lead author on our manuscript, “Comparing higher order models for the EORTC QLQ-C30,” died this past August after a sustained battle with cancer. Chad continued to work on the paper throughout the period of his illness, responding to reviewers’ comments and revising the manuscript to its near final form. Chad thoroughly enjoyed the challenge of the relatively complex work reported in our paper, as well as the on-going debate that surrounds the methods that we used. He would have relished the opportunity to respond McIntosh’s commentary. We hope that we have captured both the spirit and the intent of Chad’s perspective in this response. McIntosh raises a number of very interesting and pertinent points, most of which we agree with. At the same time, as we argued in our exchanges with the reviewers and we would still argue here, there is a great deal of controversy surrounding the proper use of the chi-square statistic and AGFI. As we did not foresee that consensus on this matter would be achieved any time soon, we decided to report both parameters in our paper. Interestingly, the controversy about the appropriate goodness-of-fit measures was reflected clearly in the diverging viewpoints expressed by the reviewers of our manuscript. Whereas McIntosh argued in favour of using chi-square as an appropriate indicator of model fit, regardless of sample size, another reviewer wrote: ‘“I’m glad you report the df and Chi-square in Table 2, but please stop talking about it as a measure of fit. It is useless as such with the N that you have.” In our multidisciplinary group, from the very start, we had had some particularly heated debates between statisticians and psychometricians. Statisticians, in particular, have long recognized that the chi-square test is fundamentally different from many other statistical significance tests (e.g., Berkson [1]). Statisticians invariably start by defining the null hypothesis: It is all but impossible to explain or discuss statistical significance tests without reference to the concept of the null hypothesis. Many non-statisticians fail to appreciate the need to take such a basic approach. A statistical significance test aims to estimate the probability that such extreme data as have been observed could have arisen purely by chance, if the null hypothesis is true. In the case of the chi-square test, the null hypothesis is that the specified model will fully explain the patterns in the observed data. The problem is that in many situations, including the present one, it is futile to expect that any of the relatively simple conceptual structural models that we and others have proposed will provide a complete representation of complex psychological and biological mechanisms. We can only hope to obtain approximate fit to the data, and we know, in advance, that any claim for perfect fit is implausible. This is quite different than the null hypothesis of many other statistical significance tests, such as in a clinical trial comparing two or more treatments (where the null hypothesis is commonly “no difference” in treatment effect), or in a regression model in which we might test whether a coefficient differs from the null hypothesis of zero. In the case of the chi-square test, there is no coefficient or effect to estimate and test; we simply test the rather absurd hypothesis of perfect and exact fit [2]. Does it spell disaster if we need to acknowledge that all of our (relatively simple) structural models will violate the null hypothesis of perfect fit? Not at all. A model may fit well enough for practical and clinical purposes, and it may then suffice to provide a useful, albeit simplified, conceptual model for the principal structural relationships. So why does the use of chi-square present a problem? First, as argued above, it is pointless to test a null hypothesis that we know to be false. In such a situation, lack of evidence of misfit as indicated by a non-significant p-value simply means that the sample size was inadequate. As Nunnally commented, these hypotheses, called “point null hypotheses,” are almost invariably known to be false before any data are collected; if such hypotheses are not rejected, it is because the sample size is too small [3]. By increasing the sample size, we can increase the chi-square statistic and make the p-value as highly significant as we wish. The magnitudes of chi-square and the p-value are thus completely uninformative. As Berkson summarized in 1938, what is the point of applying a chi-square test to a moderate or small sample if we already know that a large sample would show p highly significant? [1]. The second problem, as Berkson also noted, is that the name “goodness-of-fit” is misleading, again because the power of the chi-square test to detect an underlying disagreement between theory and data is controlled largely by the size of the sample. A model may show statistically significant evidence of misfit, yet still be a useful and practical simplification of reality. The term goodness-of-fit implies a measure of adequacy of fit. If a model provides good (or poor) fit, the same measured level of fit should be found irrespective of the size of the sample. Thus, the relationship of chi-square to sample size means that it does not satisfy the basic requirement for a goodness-of-fit index. Instead, a number of other indexes have been proposed that are less sensitive to sample size. Unfortunately, as McIntosh rightly observes, the performance of these indexes is also being called increasingly into question. However, the absence of consensus on alternative indexes does not make chi-square any the more acceptable. McIntosh queries why we are more willing to rely on the chi-square statistic when comparing two models. Here, the null hypothesis is that the models fit equally well. As McCullagh and Nelder write, “Data will often point with almost equal emphasis on several possible models, and it is important that the statistician recognize and accept this” [4]. In other words, the null hypothesis of no difference is no longer implausible and is now one worthy of testing. In SEM, it is sometimes impossible to discriminate between two or more models. As before, however, chi-square does not inform whether there is necessarily a large enough difference between the two models to be of practical or clinical importance. It merely informs us whether there is any evidence that the data support one model as providing better fit than the other. In summary, we would argue that use of chi-square is not valid for evaluating goodness-of-fit. We agree with McIntosh that there are a number pitfalls surrounding model fit assessment and that there is a need for clearer guidelines when using confirmatory factor analysis/structural equation modelling. In the meantime, the debate continues, or as Dan Cook, the sports journalist, originally put it: “It ain’t over till the fat lady sings.”