Palliative and Supportive Care
Impact of Patient-Reported Outcomes in Oncology: A Longitudinal Analysis of Patient-Physician Communication
Regularly collecting patient-reported outcomes (PROs) of health-related quality of life with feedback to oncologists may assist in eliciting and monitoring patients' problems during cancer treatment. This study examined how PRO feedback had an impact on patient-physician communication over time to gain a better understanding of how it may influence patient care.
Exploratory analyses were performed on a data set from a previous study. Patients were randomly assigned to intervention (regular completion of European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 and Hospital Anxiety and Depression Scale with feedback to oncologists), attention-control (completion of same questionnaires without feedback), and control (standard care) arms. The content of consultation audio recordings between 28 oncologists and 198 patients over four consecutive visits (792 consultations) was analyzed. Mixed-effects models and multivariate regressions were used to examine the longitudinal impact of the intervention on patient-physician communication, dynamics of patient-physician interaction, and the association between PROs and the content of clinic discussion.
Patients in the intervention arm discussed more symptoms over time compared with patients in the attention-control (P = .008) and control (P = .04) arms. No study arm effect was observed for function discussions. Discussion topics were predominantly raised by patients/relatives, regardless of arm allocation. Clinic discussions were associated with severity of patient-reported symptoms but not with patient-reported functional concerns.
A positive longitudinal impact of the intervention on symptom discussion was observed, but not for function discussion, suggesting that potentially serious problems may remain unaddressed. Training oncologists in responding to patient-reported functional concerns may increase the impact of this intervention.
There is a growing interest in routinely collecting patient-reported outcomes (PROs) as part of care for individual patients with cancer. The purpose of PROs in clinical practice is to capture patients' perspectives on the impact of cancer and treatment on their health-related quality of life (HRQoL). HRQoL is a multidimensional concept that represents patient's physical, psychological, and social response to the disease and its treatment.1 Feedback of patient-reported HRQoL information to oncologists has the potential to help detect unmet needs/problems leading to better control and monitoring of such issues,2,3 prompt oncologists to discuss HRQoL issues, act as patients' voice, and facilitate patient-centered communication and care. It has been anticipated that such improvements in processes of care would ultimately lead to improvements in patient outcomes (better health status and satisfaction).4–6 Although there has been evidence of PROs having positive effects on processes of patient care (eg, patient-physician communication),7–10 evidence for PROs having an impact on patient outcomes has been less persuasive.5,6,11–15
Several systematic reviews have examined key components in the intervention that may increase the impact of PROs in clinical practice. Recommendations include identifying patient populations most likely to benefit from PRO intervention,5 consideration of the type of HRQoL instrument,12 frequency and methods of PRO collection,11,16 and training for both patients (eg, self-efficacy training)12,13,15 and clinicians (eg, interpretation and use of PROs).11,12,15 These reviews highlight the complex nature of the intervention. It requires not only changes in the behavior of individuals (eg, patients completing questionnaires, clinicians receiving and incorporating new information in their clinical decision making), but also organizational changes to facilitate the implementation of the intervention in routine practice (including technology for data collection and integration with patient records). In addition to these recommendations, there are calls for research that would increase our understanding of the mechanisms by which PRO intervention may have an impact on patient outcomes and may help define barriers that need to be overcome.16
In a previously conducted randomized controlled trial (RCT),8 we have shown that regular PRO feedback to oncologists during cancer treatment led to improved communication and patient well-being. However, because of time and resource restrictions, we reported the preplanned analysis of patient-physician communication at a single time point. We have since undertaken analysis of all the consultations in this RCT. We conducted exploratory analyses on the rich longitudinal data set and focused on whether the PRO feedback actually led to the discussion of patient-reported problems and whether oncologists were prompted to raise these issues with the patient. The focus was on communication, because the acknowledgment and exploration of problems is the essential first step in using PROs to inform shared clinical decision making.
The aims and hypotheses of these analyses were to investigate:
Whether repeated PRO intervention had a longitudinal impact on patient-physician communication. We hypothesized that PRO intervention would lead to increased discussion of patients' HRQoL issues and that repeated intervention would help maintain level of discussion at subsequent clinic consultations.
Whether PRO feedback encouraged oncologists to initiate discussion of HRQoL issues. We hypothesized that the intervention may change the dynamics of the communication and prompt oncologists to actively initiate discussions about patient-reported problems.
The association between severity of patient-reported problems and clinic discussion. We hypothesized that clinic discussion will be reflected by severity of patients' problems and PRO feedback to oncologists would prevent important issues from being missed.
The analyses in this article were performed on a data set from the previous RCT.8 Detailed description of the trial design and methodology has previously been reported.8 The study was approved by the institutional ethical committee. Written informed consent was obtained from patients and physicians.
Briefly, this was a prospective longitudinal RCT with repeated measures. In all, 286 patients with cancer who attended Medical Oncology Outpatient Clinics at St. James's University Hospital (Leeds, United Kingdom) and started a course of chemotherapy/biologic therapy participated in this study. Patients were randomly assigned to intervention, attention-control, or control arms in a 2:1:1 ratio. All patients had a baseline consultation (no questionnaire intervention) followed by three study consultations.
In the intervention arm, patients completed European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 version 3.0 (EORTC QLQ-C3017) and Hospital Anxiety and Depression Scale (HADS)18 on touch-screen computers before their clinic consultation. EORTC QLQ-C30 is a 30-item questionnaire that includes five functional scales (physical, emotional, cognitive, social, role), three symptom scales (fatigue, pain, nausea/vomiting), a global HRQoL scale, and six single items (anorexia, insomnia, dyspnea, diarrhea, constipation, financial difficulties). HADS is a 14-item questionnaire designed to measure anxiety and depression in physically ill patients. Scale scores were presented in a graphic format in real time to the oncologists before patient encounters. In the attention-control arm, patients completed the same questionnaires on touch-screen computers, without feedback to the oncologists. Patients in the control arm received standard care. All clinic consultations were audio recorded.
Individual meetings were held with participating physicians to explain the questionnaire scores and graphs. A manual containing information about the questionnaires was also provided. We asked the physicians to use the data in their consultations, without further instructions.
All audio-recorded consultations were subjected to content analysis. A study-specific checklist was used to note discussion of symptoms/psychosocial function topics covered in EORTC QLQ-C30 and HADS and any other symptoms or issues raised. The person (oncologist or patients/relatives) initiating the discussion for each topic was also noted.8
Content analysis was performed directly from the audio recordings by nine raters. Each consultation was coded by two raters. The inter-rater reliability was good, with intraclass correlation coefficients of 0.69 to 1.00 (median, 0.94). Weekly meetings were held to achieve consensus.
Symptoms included in the analyses were pain, fatigue, dyspnea, anorexia, insomnia, nausea/vomiting, bowel habits (diarrhea/constipation) and five functions (physical, emotional, cognitive, social, role) to reflect the topics covered in the questionnaires.
A summated score was calculated for the total number of symptoms (0 to 7) and functions (0 to 5) discussed at each consultation. Mixed-effects models were used to assess whether number of symptoms/functions discussed differed between study arms over time. Potential covariates (patients' sex, age, diagnosis, response at 3 months, performance status, extent of disease, time in study, months since diagnosis, and a measure of the extent to which patients had seen the same oncologist) were identified by univariate regression (the number of issues discussed at first consultation as the outcome variable and each covariate as the predictor, controlling for baseline). Covariates meeting the inclusion criterion (P < .1) were entered in multivariate mixed effects models.
The models' outcome variable was the number of symptoms/functions discussed. Fixed effects were the number of symptoms/functions discussed at baseline, study arm, time (consultation 1, 2, or 3), arm by time interaction (only retained in the final model if significant), and any identified covariates. Patients were entered as a random effect. A significance level of P < .05 was used for this primary hypothesis testing analysis.
Multivariate logistic regression was used to explore predictors for who initiated discussions of symptoms/functions (oncologist v patients/relatives). To identify covariates for inclusion, univariate logistic regression models for each symptom/function were fitted separately for each visit with the person initiating the discussion as the outcome and the potential covariate (patients' sex, age, diagnosis, performance status, extent of disease, oncologists' sex, and oncologists' grade [consultant or specialist registrar]) as single explanatory variable. In the multivariate regression model, outcome variable was the person initiating the discussion at each visit, and independent variables were study arm and significant covariates (P < .1). This was repeated for all symptoms and functions. A significance level was set at P < .01 for the multivariate analysis to adjust for multiple tests.
Subgroup analyses were performed for patients in the intervention and attention-control arms (n = 146) who completed EORTC QLQ-C30 and HADS. Multivariate logistic regression was used to investigate the association between severity of problems and clinic discussion separately for each visit, adjusting for arm. Potential covariates (patients' sex, age, diagnosis, performance status, extent of disease, discussion of respective symptom/function at baseline, oncologists' sex, and oncologists' grade) were identified by univariate regression, with a particular symptom/function discussed or not as the outcome variable. In the multivariate regression model, outcome variable was whether a symptom/function was discussed or not, and the independent variables were questionnaire score for relevant symptom/function, study arm, and significant covariates (P < .1). Analysis was repeated for all symptoms and functions at each consultation. A significance level of P < .01 was used in the multivariate analysis.
For illustrative purposes, the EORTC QLQ-C30 scores (higher scores represent better functioning but worse symptoms) were divided into four groups. Three were equally spaced score ranges (1 to 33.3, 33.4 to 66.7, 66.8 to 99.9) and were categorized into “Poor,” “Moderate,” and “Good” for functions and “Mild,” “Moderate,” and “Severe” for symptoms. Patients scoring zero for symptoms and 100 for functions were grouped separately as reporting “no symptom” and “excellent functioning,” respectively. Similarly, HADS scores were divided into four groups guided by cutoff scores suggested by HADS developers to group patients into anxiety/depression categories of “No anxiety/depression” (score 0), “Mild” (scores 1 to 7), “Moderate” (scores 8 to 10), and “Severe” (scores ≥ 11).18
Of the 286 patients who participated in the RCT, 222 patients completed four consecutive consultations (seven patients refused after baseline consultation; 57 patients discontinued). More male patients (P = .002) and those with poorer performance status (P = .003) failed to complete the study. Following content analysis of all four consultations, a further 24 patients were excluded because of poor quality audio recordings, resulting in 198 patients with complete data set. (Table 1).
|Intervention(n = 100) ||Attention-Control(n = 46) ||Control(n = 52)|
|Extent of disease|
|WHO performance status|
|No. of symptoms discussed at baseline|
|No. of functions discussed at baseline|
Abbreviation: SD, standard deviation.
All 28 physicians working in the Oncology Department at St. James's University Hospital (Leeds, United Kingdom) participated. There were 17 male and 11 female physicians with median age of 33.5 years (range, 26 to 51 years); six were consultants and 22 were specialist registrars, with varied oncology experience (range, 0 to 24 years).
Table 2 presents the results of the mixed-effects models for number of symptoms/functions discussed. A time by arm interaction was not significant for either model. Patients in the intervention arm discussed more symptoms during consultations than those in the attention-control (P = .008) and control arms (P = .040). There was also a significant effect of time with fewer symptoms being discussed between the first and third consultations (P = .004). Figure 1 graphically represents the change in number of issues discussed compared with baseline, at first, second, and third consultations. The increase in symptom discussions was largest the first time PROs were provided to the oncologists and was maintained over time. There were no differences between arms for the discussion of functions and no time effect was observed. Of the identified covariates, only diagnosis remained significant in the functions model. In particular, patients with melanoma discussed more and patients with bladder cancer discussed fewer functional issues than patients with other diagnoses. However, the numbers of patients with these cancers were small.
|Variable||Least Square Means||Estimate of Effect||SE||95% CI||P|
|No. of symptoms discussed|
|Discussion at baseline||0.26||0.048||0.16 to 0.35||< .001|
|Control v intervention||2.91 v 3.32||−0.41||0.197||−0.79 to −0.02||.040|
|Attention-control v intervention||2.77 v 3.32||−0.55||0.207||−0.96 to −0.14||.008|
|Consultation 1 v consultation 3||3.19 v 2.79||0.40||0.140||0.13 to 0.68||.004|
|Consultation 2 v consultation 3||3.01 v 2.79||0.22||0.140||−0.05 to 0.50||.113|
|Sex||−0.16||0.277||−0.71 to 0.38||.562|
|Response at 3 months*||.851|
|No. of functions discussed|
|Discussion at baseline||0.16||0.041||0.08 to 0.24||< .001|
|Control v intervention||1.36 v 1.57||−0.22||0.125||−0.46 to 0.03||.084|
|Attention-control vintervention||1.41 v 1.57||−0.16||0.130||−0.42 to 0.09||.210|
|Consultation 1 v consultation 3||1.44 v 1.39||0.05||0.106||−0.16 to 0.25||.670|
|Consultation 2 v consultation 3||1.51 v 1.39||0.12||0.106||−0.09 to 0.33||.276|
|Extent of disease||0.02||0.137||−0.25 to 0.29||.873|
|Time since diagnosis||−0.002||0.002||−0.01 to 0.001||.261|
|Time on study||0.002||0.001||−0.0006 to 0.004||.154|
*Only P value of overall F test is given for categorical covariates with more than two levels.
Discussion of symptoms and functions were predominantly initiated by patients/relatives with the exception of dyspnea and bowel habits (Table 3). Overall, symptom discussions (26% to 64%) were more prevalent than function discussions (4% to 41%). Discussions about pain (64%), fatigue (59%), and nausea/vomiting (55%) were common. None of the covariates met the inclusion criterion; therefore, only univariate logistic regression was applied. No study arm effect was found, that is, PRO feedback did not increase inquiry about patients' problems by the oncologists. These findings did not change at subsequent consultations.
|Variable||No. of Consultations in Which Issue Was Discussed (N = 198) ||Physician Initiating ||Patients/Relatives Initiating ||P (effect of study arm)*|
*No significant covariates were identified from univariate analyses of patients' sex, age, diagnosis, performance status, extent of disease, oncologists' sex, and grade.
Figures 2A to 2C illustrate the prevalence and severity of symptoms/functions reported by this patient subgroup at first consultation. Similar results were seen at subsequent consultations. A substantial proportion of patients (32% to 73%) denied the presence of symptoms listed; only a minority of patients reported severe symptoms (3% to 9%). Fatigue was an exception, with 21% of patients reporting severe fatigue (Fig 2A).
A large proportion of patients expressed poor role and social functioning (32% and 40%, respectively), whereas physical and cognitive functions were generally good (poor functioning reported in 12% and 4%, respectively). Nine percent of patients reported poor emotional functioning on EORTC QLQ-C30, but “severe” anxiety and depression were reported on HADS by 16% and 13% of the patients, respectively (Figs 2B and 2C).
In the multivariate regressions, symptom/function discussion was predicted by relevant questionnaire score, study arm, and significant covariates. Patients reporting severe symptoms were more likely to have a discussion about those symptoms (Fig 3A). Severity was predictive of discussion about dyspnea (P < .01; covariate: baseline discussion) and pain (P < .001; covariate: extent of disease) during all three clinic encounters. Severity of fatigue (covariates: baseline discussion, patients' sex, oncologists' sex), nausea/vomiting (covariate: baseline discussion), anorexia (covariates: baseline discussion, age), and insomnia (covariates: baseline discussion, EORTC QLQ-C30 emotional functioning score, HADS depression score) were significant predictors of discussion at two of the three clinic encounters. Constipation score was a significant predictor for discussion of bowel habits only at first consultation (P = .001; covariate: baseline discussion), but there was a positive trend in the second and third consultations (P = .015 and P = .030, respectively). There was no significant effect of arm on whether a specific symptom was discussed.
In contrast, there was no clear relationship between severity of functional impairment and clinic discussions (Fig 3B). HADS anxiety score was predictive of emotional function discussion at first consultation only (P < .01; covariate: baseline discussion). No study arm effect was observed. Frequency of discussion about cognitive function was too small for analysis.
Our findings have highlighted differences in the communication of symptoms and functions in oncology clinics. Regular feedback of patient-reported symptoms to oncologists had an overall effect on the range of symptoms being discussed, which was maintained over time. Discussion of symptoms was consistent with patients' self-reported severity. Although the intervention aimed to alert oncologists to patients' problems, symptom discussions were predominantly initiated by patients/relatives, suggesting that PRO feedback did not necessarily influence the standard medical interview structure19 in which patients are encouraged to report their own experience. It may be that the physical presence of the PRO in the consultation may have served as a prompt for both patients and oncologists, resulting in a more comprehensive symptom discussion. However, the mechanisms underlying how PRO feedback had an impact on symptom discussions remain speculative.
In contrast, PRO feedback had no impact on discussion of patients' functioning. Again, patients were more likely to initiate discussions but there was no association between their frequency and patient-reported severity, suggesting that patients did not necessarily perceive oncology consultations as an appropriate forum to raise these issues. Furthermore, PRO feedback did not persuade oncologists to explore these issues further. The frequency of discussions regarding social and role functions was particularly low with respect to the proportion of patients reporting poor functioning. These findings contrast with those of studies in which EORTC QLQ-C30 was used in similar clinical settings, in which inquiry of problems by clinicians increased, including functional problems.10,20
Our results illustrate that cancer and treatment have an impact on many aspects of patients' HRQoL. There is increasing recognition that communication should encompass patients' psychosocial functioning in addition to cancer and treatment concerns21 to address the high prevalence of unmet psychological, supportive, and information needs among patients with cancer.22,23 One of the aims of PRO intervention is to identify such unmet needs so that appropriate input may be provided.
Our results suggest that the wider impact of cancer and treatment on patients' psychosocial functioning was largely ignored by oncologists, despite PRO feedback. This emphasizes that PRO feedback alone does not overcome the existing barriers that prevent oncologists from discussing functional issues with their patients. These barriers include time constraints, oncologists' preferences,24 their ability to discuss psychosocial issues,25 minimizing or ignoring psychosocial concerns because of the perception that these are inevitable consequences of cancer diagnosis,26 and oncologists' notion that they are unable to offer suitable advice or solutions. In addition, some oncologists may be skeptical about using questionnaires to measure patients' HRQoL,27 which may, in part, be due to lack of familiarity with the measures.28
We agree with other investigators11,12 that training for oncologists may be crucial in addressing some of these barriers. Several studies29–31 have demonstrated the benefits of communication skills training in changing clinicians' attitudes and beliefs and, most importantly, clinicians' acquisition of new skills. The training should encompass not only the interpretation of measures28 but also how oncologists might respond to PROs, particularly patients' psychosocial problems. Provision of guideline-driven recommendations for managing various symptoms and functional issues together with information about locally available resources for patients should be integral to assist oncologists in guiding patients to the most appropriate supportive care services. Training can also provide an opportunity to engage with the oncologists about the potential benefits of PRO intervention.11,15
The limitations in our analyses include the study population who were predominantly female with good performance status recruited from a single center. Therefore, the impact of cancer and treatment on patients' HRQoL may be underestimated. This may partly explain the relatively low frequency of serious symptoms reported. In addition, the items in the EORTC QLQ-C30 refer to the “past week,” which may not have fully captured the adverse effects of patients' treatment. Furthermore, oncologists encountered patients in all three arms of the study which may have caused contamination of the results. We also acknowledge that the study was limited by the consultation analysis method we used, which failed to provide information on how PRO feedback may have influenced the quality of patient-physician communication or how the oncologists used the PRO data. Further qualitative analysis on a sample of consultations is being undertaken to examine this further. Nevertheless, the analyses have been performed on a relatively large (198 patients, 792 consultations) longitudinal sample of real-life consultations with patients, many of whom were receiving palliative therapy for advanced cancer.
In conclusion, our findings indicate that PRO feedback possibly supports communication about physical symptoms, but the intervention alone is inadequate to overcome the pre-existing barriers that prevent oncologists from exploring the psychosocial impact of cancer and treatment on patients. This highlights the need for clinician-targeted strategies as a component of this complex intervention. These findings will inform and contribute to the development of a training program for oncologists in how to interpret and respond to patient-reported HRQoL.
Supported by Grant No. C7775/A7424 (G.V.) from Cancer Research UK and a Bramall Research Fellowship (E.E.T.) from the University of Leeds.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
The author(s) indicated no potential conflicts of interest.
Conception and design: Elena E. Takeuchi, Ada Keding, Peter J. Selby, Julia M. Brown, Galina Velikova
Financial support: Peter J. Selby, Galina Velikova
Administrative support: Noha Awad
Provision of study materials or patients: Lyndsay J. Campbell
Collection and assembly of data: Noha Awad, Lyndsay J. Campbell, Galina Velikova
Data analysis and interpretation: Elena E. Takeuchi, Ada Keding, Noha Awad, Ursula Hofmann, Lyndsay J. Campbell, Peter J. Selby, Julia M. Brown, Galina Velikova
Manuscript writing: All authors
Final approval of manuscript: All authors
|1.||IB Wilson, PD Cleary: Linking clinical variables with health-related quality of life: A conceptual model of patient outcomes JAMA 273: 59– 65,1995 Crossref, Medline, Google Scholar|
|2.||IJ Higginson, AJ Carr: Measuring quality of life: Using quality of life measures in the clinical setting BMJ 322: 1297– 1300,2001 Crossref, Medline, Google Scholar|
|3.||SM Gilbody, AO House, TA Sheldon: Routinely administered questionnaires for depression and anxiety: Systematic review BMJ 322: 406– 409,2001 Crossref, Medline, Google Scholar|
|4.||KN Lohr: Applications of health status assessment measures in clinical practice. Overview of the third conference on advances in health status assessment Med Care 30: MS1– MS14,1992 suppl 5 Crossref, Medline, Google Scholar|
|5.||M Espallargues, JM Valderas, J Alonso: Provision of feedback on perceived health status to health care professionals: A systematic review of its impact Med Care 38: 175– 186,2000 Crossref, Medline, Google Scholar|
|6.||J Greenhalgh, K Meadows: The effectiveness of the use of patient-based measures of health in routine practice in improving the process and outcomes of patient care: A literature review J Eval Clin Pract 5: 401– 416,1999 Crossref, Medline, Google Scholar|
|7.||A Boyes, S Newell, A Girgis , etal: Does routine assessment and real-time feedback improve cancer patients' psychosocial well-being? Eur J Cancer Care (Engl) 15: 163– 171,2006 Crossref, Medline, Google Scholar|
|8.||G Velikova, L Booth, AB Smith , etal: Measuring quality of life in routine oncology practice improves communication and patient well-being: A randomized controlled trial J Clin Oncol 22: 714– 724,2004 Link, Google Scholar|
|9.||SB Detmar, MJ Muller, JH Schornagel , etal: Health-related quality-of-life assessments and patient-physician communication: A randomized controlled trial JAMA 288: 3027– 3034,2002 Crossref, Medline, Google Scholar|
|10.||SB Detmar, NK Aaronson: Quality of life assessment in daily clinical oncology practice: A feasibility study Eur J Cancer 34: 1181– 1186,1998 Crossref, Medline, Google Scholar|
|11.||T Luckett, PN Butow, MT King: Improving patient outcomes through the routine use of patient-reported data in cancer clinics: Future directions Psychooncology 18: 1129– 1138,2009 Crossref, Medline, Google Scholar|
|12.||S Marshall, K Haywood, R Fitzpatrick: Impact of patient-reported outcome measures on routine practice: A structured review J Eval Clin Pract 12: 559– 568,2006 Crossref, Medline, Google Scholar|
|13.||K Haywood, S Marshall, R Fitzpatrick: Patient participation in the consultation process: A structured review of intervention strategies Patient Educ Couns 63: 12– 23,2006 Crossref, Medline, Google Scholar|
|14.||JM Valderas, A Kotzeva, M Espallargues , etal: The impact of measuring patient-reported outcomes in clinical practice: A systematic review of the literature Qual Life Res 17: 179– 193,2008 Crossref, Medline, Google Scholar|
|15.||MH Frost, AE Bonomi, JC Cappelleri , etal: Applying quality-of-life data formally and systematically into clinical practice Mayo Clin Proc 82: 1214– 1228,2007 Crossref, Medline, Google Scholar|
|16.||J Greenhalgh, AF Long, R Flynn: The use of patient reported outcome measures in routine clinical practice: Lack of impact or lack of theory? Soc Sci Med 60: 833– 843,2005 Crossref, Medline, Google Scholar|
|17.||NK Aaronson, S Ahmedzai, B Bergman , etal: 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 85: 365– 376,1993 Crossref, Medline, Google Scholar|
|18.||AS Zigmond, RP Snaith: The hospital anxiety and depression scale Acta Psychiatr Scand 67: 361– 370,1983 Crossref, Medline, Google Scholar|
|19.||J Bird, SA Cohen-Cole: The three-function model of the medical interview: An educational device Adv Psychosom Med 20: 65– 88,1990 Crossref, Medline, Google Scholar|
|20.||P Taenzer, BD Bultz, LE Carlson , etal: Impact of computerized quality of life screening on physician behaviour and patient satisfaction in lung cancer outpatients Psychooncology 9: 203– 213,2000 Crossref, Medline, Google Scholar|
|21.||TF Hack, LF Degner, PA Parker , etal: The communication goals and needs of cancer patients: A review Psychooncology 14: 831– 845,2005 discussion 846-847 Crossref, Medline, Google Scholar|
|22.||S Newell, RW Sanson-Fisher, A Girgis , etal: The physical and psycho-social experiences of patients attending an outpatient medical oncology department: A cross-sectional study Eur J Cancer Care (Engl) 8: 73– 82,1999 Crossref, Medline, Google Scholar|
|23.||R Sanson-Fisher, A Girgis, A Boyes , etal: The unmet supportive care needs of patients with cancer: Supportive Care Review Group Cancer 88: 226– 237,2000 Crossref, Medline, Google Scholar|
|24.||SB Detmar, NK Aaronson, LD Wever , etal: How are you feeling? Who wants to know? Patients' and oncologists' preferences for discussing health-related quality-of-life issues J Clin Oncol 18: 3295– 3301,2000 Link, Google Scholar|
|25.||L Fallowfield, D Ratcliffe, V Jenkins , etal: Psychiatric morbidity and its recognition by doctors in patients with cancer Br J Cancer 84: 1011– 1015,2001 Crossref, Medline, Google Scholar|
|26.||SD Passik, W Dugan, MV McDonald , etal: Oncologists' recognition of depression in their patients with cancer J Clin Oncol 16: 1594– 1600,1998 Link, Google Scholar|
|27.||KM Taylor, KG Macdonald, A Bezjak , etal: Physicians' perspective on quality of life: An exploratory study of oncologists Qual Life Res 5: 5– 14,1996 Crossref, Medline, Google Scholar|
|28.||MY Halyard, MH Frost, A Dueck , etal: Is the use of QOL data really any different than other medical testing? Curr Probl Cancer 30: 261– 271,2006 Crossref, Medline, Google Scholar|
|29.||L Fallowfield, V Jenkins, V Farewell , etal: Efficacy of a Cancer Research UK communication skills training model for oncologists: A randomised controlled trial Lancet 359: 650– 656,2002 Crossref, Medline, Google Scholar|
|30.||D Razavi, I Merckaert, S Marchal , etal: How to optimize physicians' communication skills in cancer care: Results of a randomized study assessing the usefulness of posttraining consolidation workshops J Clin Oncol 21: 3141– 3149,2003 Link, Google Scholar|
|31.||L Fallowfield, V Jenkins, V Farewell , etal: Enduring impact of communication skills training: Results of a 12-month follow-up Br J Cancer 89: 1445– 1449,2003 Crossref, Medline, Google Scholar|