Elsevier

European Journal of Cancer

Volume 37, Issue 14, September 2001, Pages 1729-1735
European Journal of Cancer

Response shift in the perception of health for utility evaluation: an explorative investigation

https://doi.org/10.1016/S0959-8049(01)00196-4Get rights and content

Abstract

We previously showed that patients with newly diagnosed colon cancer change the internal standards on which they base their quality of life estimation. In the present investigation, we explored whether this response shift similarly affects the perception of health for utility evaluation in cancer clinical trials. After radical resection of adenocarcinoma of the colon (pT1-4 pN>0 M0 and pT3-4 pN0 M0) and perioperative chemotherapy, patients were randomised to three treatment arms: observation only (A), 5-fluorouracil (5-FU) 450 mg/m2 plus levamisol (B), or 5-FU 600 mg/m2 (C). Subjective health was assessed by a linear analogue self-assessment (LASA) scale anchored at ‘perfect health–worst health’ developed for serial assessment of utility values (Hürny C, van Wegberg B, Bacchi M, et al. Subjective health estimations (SHE) in patients with advanced breast cancer: an adapted utility concept for clinical trials. Br J Cancer 1998, 77, 985–991). Patients estimated their pre-surgery health among various quality of life indicators both before surgery and retrospectively thereafter, and their pre-adjuvant health both at the beginning of randomly assigned chemotherapy or observation and retrospectively approximately 2 months later. Thereafter, current subjective health was assessed. Paired t-tests were used to test the hypotheses of no change. Patients' estimated pre-surgery health was worse after surgery than before (n=127, mean change=−6.7, standard deviation (S.D.)=30, P=0.01), and their estimated pre-adjuvant health was worse under treatment or observation than at the beginning (n=132, mean change=−7.1, S.D.=23.8, P=0.001), in agreement with the quality of life indicators. Chemotherapy had no impact on these changes attributed to a response shift. Conventionally assessed changes between the beginning of adjuvant treatment or observation and 2 months later indicated no change in subjective health. Change scores relative to patients' retrospective estimation revealed an improvement (n=122, mean change=6.6, S.D.=24.8, P=0.004) in this period. Patients with colon cancer substantially reframe their internal standard of health as they do for quality of life. This explorative finding questions the assumption, generally made in decision models, that health estimates for utility evaluation are independent of time. Given that patients may change their standards, comparisons of health estimates across different populations and clinical situations are to be interpreted with caution.

Introduction

In evaluating patient utilities under cancer treatment, a patient's state of health within the periods of treatment or follow-up is often assumed to be stable at least until recurrence. This assumption is the basis for cross-sectional comparisons of different interventions. Regarding health-related quality of life (QL), however, there is increasing evidence that patients with a chronic disease may change the internal standards on which they base their QL estimation in the process of becoming and remaining ill. This phenomenon has been described as ‘response shift’ 1, 2.

We previously showed that patients with newly diagnosed colon cancer changed their internal standards of QL 1, 3. This observation was made with regard to radical resection with perioperative chemotherapy and to randomly assigned post-operative adjuvant chemotherapy or observation within a clinical trial of the Swiss Group for Clinical Cancer Research (SAKK 40/93). We described these changes as ‘reframing’, as used for similar phenomena in psychology. ‘Reframing’ signifies that patients do not assess their health against a fixed reference point (i.e. a ‘true’ baseline) but against a frame of reference which shifts in the light of experience.

Whether this phenomenon similarly affects patient-estimated health status is not known. This question is relevant for cost-utility evaluations in oncology. As an alternative to formal utility assessment, health status has been assessed in many studies by visual analogue scales, rated either by the patients or by clinicians [4]. In both cases, it is assumed that these estimates reflect a ‘true’ value for a given health status which may be compared among different populations and clinical situations. This amounts to an assumption of stable internal reference points which allow an assessment of health status at any given point in time.

Such a visual analogue scale was included in the trial cited above. It was developed for serial assessment of health status for utility evaluation in cancer clinical trials [5]. Patients were asked to imagine they would have to live the rest of their life in their current condition and to indicate how they would rate a life in this condition between perfect health and worst health.

In this report, we explore the question of whether health status estimates by patients are similarly affected by a response shift as has been shown for QL indicators, and whether this effect changes the interpretation of health status estimates.

Section snippets

The trial

The trial (SAKK 40/93) was open for all patients with radically-resected and histologically-proven adenocarcinoma of the colon with pathologically confirmed stages pT1-4 pN>0 M0 and pT3-4 pN0 M0. The patients had to have a potentially curative resection (R0-resection) and no additional rectal carcinoma. The perioperative intraportal chemotherapy was a 7-day infusion of 5-fluorouracil (5-FU) starting immediately after surgery and interrupted by a 2-h infusion of Mitomycin-C after the first 24 h.

Sample description

As described in the original report [3], 215 patients were randomised in German speaking centres. For this investigation, 187 patients with at least one pair of corresponding questionnaires for either pre/then-test comparison were selected (87%). At the time of surgery, the main reasons for missing QL data were administrative problems at the local centres. In the adjuvant phase, QL data were mainly missing due to patients' failure to send back the questionnaire.

Biomedical and sociodemographic

Discussion

In this report, we explored the question of response shift in perception of health for utility evaluation and contrasted the findings with those of physical well-being and global QL, which were previously described [3].

Baseline correlations among the three indicators were moderate, related to the fact that QL and subjective health are different although overlapping constructs. However, we found similar effects of reframing. After surgery, patients' retrospective estimates of their pre-surgery

Acknowledgements

To study reframing of QL and health perception within a clinical trial needed considerable persuasiveness and logistic efforts. We would like to thank the patients who took the time to complete the questionnaires, the data managers, nurses and surgeons who took care of this study in the centres, Heidi Gusset for central data management, and Brigitte van Wegberg for trial coordination at the beginning.

References (16)

There are more references available in the full text version of this article.

Cited by (38)

  • Response shift in self-rated health after serious health events in old age

    2017, Social Science and Medicine
    Citation Excerpt :

    Recalibration response shift refers to a change in internal standards, which means that stable SRH during the aging process results in lowered standards for good health. SRH is not only rather stable during the aging process despite health declines (Jylhä et al., 2001), but there also is often no change in SRH pre and post a health event (Bernhard et al., 2001; Hillen et al., 2003; Yardley and Dibb, 2007). Several studies suggest that people tend to retrospectively overestimate their previous health assessment, which is noteworthy as a number of clinical studies used this question to assess SRH before a health event.

  • An exploratory study of response shift in health-related quality of life and utility assessment among patients with osteoarthritis undergoing total knee replacement surgery in a tertiary hospital in Singapore

    2012, Value in Health
    Citation Excerpt :

    In other words, there may be situations where true change measured by HRQOL instruments may not be simply calculated as the difference between respective pre- and postintervention test scores [10,11]. In surgical interventions including total knee replacement (TKR), pre- and postintervention comparisons of HRQOL have been used as a standard method to evaluate patients' improvement in both generic and disease-specific health status, and consequently to determine the cost-effectiveness of the treatment [12,13]. Because of response shift's possible impact on such studies, its exploration has become an emerging area in HRQOL research of surgical interventions [14–16].

  • Understanding inconsistencies in patient-reported outcomes after spine treatment: response shift phenomena

    2009, Spine Journal
    Citation Excerpt :

    Clinicians began to recognize that response shift could obfuscate important treatment-related changes and indeed might even be the subtext or the desired effect in rehabilitation and psychosocial interventions [5–7]. Response shift has now been studied and recognized to affect adaptation to a wide degree of health conditions, including multiple sclerosis [8–11], cancer [12–24], stroke [25–27], diabetes [28,29], geriatrics [30–32], palliative care [33–36], dental disorders [37], and, most recently, orthopedics [38]. A meta-analysis done on response shift reported that response shift findings ranged from moderate to small effect sizes [39].

  • Response shift in self-reported functional scores after knee microfracture for full thickness cartilage lesions

    2009, Osteoarthritis and Cartilage
    Citation Excerpt :

    All four scores that we compared (VAS pain, Lysholm, IKDC1 and IKDC2) had a significant shift. Such a positive RS is also found in other studies following surgery11,16,28. A very likely explanation for this is adaptation to the illness28,29.

View all citing articles on Scopus
View full text