Implementing a system of quality-of-life diagnosis and therapy for breast cancer patients: results of an exploratory trial as a prerequisite for a subsequent RCT

Abstract

A system for quality-of-life diagnosis and therapy (QoL system) was implemented for breast cancer patients. The system fulfilled the criteria for complex interventions (Medical Research Council). Following theory and modeling, this study contains the exploratory trial as a next step before the randomised clinical trial (RCT) answering three questions: (1) Are there differences between implementation sample and general population? (2) Which amount and type of disagreement exist between patient and coordinating practitioners (CPs) in assessed global QoL? (3) Are there empirical reasons for a cutoff of 50 points discriminating between healthy and diseased QoL? Implementation was successful: 74% of CPs worked along the care pathway. However, CPs showed preferences for selecting patients with lower age and UICC prognostic staging. Patients and CPs disagreed considerably in values of global QoL, despite education in QoL assessment by outreach visits, opinion leaders and CME: Zero values of QoL were only expressed by patients. Finally, the cutoff of 50 points was supported by the relationship between QoL in single items and global QoL: no patients with values above 50 dropped global QoL below 50, but values below 50 and especially at 0 points in single items, induced a dramatic fall of global QoL down to below 50. The exploratory trial was important for defining the complex intervention in the definitive RCT: control for age and prognostic stage grading, support for a QoL unit combining patient's and CP's assessment of QoL and support for the 50-point cutoff criterion between healthy and diseased QoL.

Figures

Figure 1

QoL profile and experts' report produced from the EORTC questionnaire by the patient and health status questionnaire by the doctor in the QoL unit and sent to the coordinating practitioner of the patient implemented in QoL diagnosis and therapy. Example of the largest difference between patient's and clinician's assessment of global QoL in 170 patients. Female patient with primary breast cancer, no. 170 in the series, 1 month after BCT with axillary lymph adenectomy, 44 years, married, two children. Prognostic classification T1c, N0, M0, G2, ER pos/PR pos, HER2neu neg. Cutoff level: 50 points (grey bar). For further details of the QoL system see reference Klinkhammer-Schalke et al, 2008.

Figure 2

Age distribution of patients with breast cancer in the study region as documented by the tumour centre and that of patients selected by the coordinating practitioners during implementation. Histograms of the two groups (January 2003 until June 2004).

Figure 3

Prognostic stage (UICC) of the patients in the implementation sample and the regional population. Missing values in sample n=17, in population n=64, global test: χ2=14.689 (d.f. 4), P<0.005; single tests: UICC 0: χ2=3.913 (d.f. 1), P<0.05, UICC III: χ2=9.296 (d.f. 1) P<0.005, all other tests not significant.

Figure 4

Relationship between assessment of global QoL by the patient herself and by her CP. Comparison with the UICC grouping. For conditions of assessment see Patients and methods. For UICC grading a linear scale was assumed for the statistical model. (A) y=22+0.56x; Spearman ρ=0.43; P(2α)<0.01; n=158; ⊙ case with the strongest disagreement, for details of this patient see Figure 1. (B) y=64−0.48x; Spearman ρ=0.006; p(2α)=n.s.; n=148; case with the worst prognosis, but maximum global QoL.

Figure 5

Bland–Altman plot (Bland and Altman, 1986) for agreement analysis between the judgement of the patient and her CP about global QoL. Differences and mean values were calculated for each of the patients. Limits of agreement are the upper and the lower two s.d. values calculated for normal distribution of all differences in the sample (n=158).

Figure 6

Direction of relative negativity patient/doctor in global QoL depending on decrease of global QoL assessed by the patient.

Figure 7

Influence of drop of QoL in single symptom dimensions of QoL on global QoL. Sensitivity analysis with five subgroups: all nine values of the single dimensions >60, >50, >25, >0 or at least one with 0 points. Note that global QoL decreases (reacts) at single items

Figure 8

Histograms of global QoL of patients (A) either with no very bad value (0-value=worst breakdown) in one of the symptom and deficit scales or (B) with at least one very bad value (0-value) in one of the symptom and deficit scales. For reasons of simplicity, there was no differentiation between 0 in one dimension or the other. 50 points=cutoff between healthy and diseased QoL.