The use of patient reported outcome measures in routine clinical practice: lack of impact or lack of theory?
Introduction
The purpose of this paper is to question why the use of patient-reported outcome measures (PROs) in clinical practice, in particular, those measuring health-related quality of life (HRQoL), has little influence on clinical decision making. To do this, the paper applies a theory-driven approach (Weiss, 1995; Connell & Kubisch, 1995; Pawson, 2002) to argue that systematic reviews and randomised controlled trials assessing the efficacy of this intervention have focused almost exclusively on determining whether the intervention works without adequate consideration of how the intervention might give rise to the expected outcomes. Consequently, this literature provides little insight into the reasons why the use of HRQoL measures in clinical practice does not impact upon patient care and also calls into question the appropriateness of the outcome criteria used to judge the intervention's effectiveness. To address this, the paper identifies a number of mechanisms that might give rise to the expected outcomes that are currently implicit within the design of the intervention and hypotheses specified within the trials evaluating the use of HRQoL measures in clinical practice. It then examines how far current clinical practice matches these mechanisms and in doing so, a number of possible explanations for the lack of impact of HRQoL on clinical decision making are reviewed. The paper concludes by specifying how the intervention might be modified in order to maximise its potential impact on clinical decision making.
Section snippets
Evidence for the impact of HRQoL measures on clinical decision making
PRO measures include standardised measures of health status, disability, handicap and HRQoL as well as instruments assessing more specific dimensions of patient experience such as depression and anxiety (Bowling, 1995). The use of such instruments in clinical practice, in particular, those measuring HRQoL, has been proposed as a means of facilitating communication, uncovering patients’ problems and monitoring response to treatment (Higginson & Carr, 2001). However, recent systematic reviews
A theory-driven approach to evaluation
To query the legitimacy of the reviews’ conclusions and examine why the use of HRQoL measures in clinical practice does not influence clinical decision making requires an understanding of what the intervention is and how it is expected to work. Numerous authors have recognised the need for a theory-driven approach to the evaluation of complex interventions. For example, Weiss (1995) and Connell and Kubisch (1995) proposed the ‘theory of change’ approach that involves specifying not only the
Application of the theory-driven approach to the evidence base for the use of HRQoL measures in clinical practice
A starting point in applying a theory-driven approach to the feedback of HRQoL information to clinicians is to review the hypotheses tested by the randomised controlled trials that have evaluated the effectiveness of this intervention. The aim here is to identify the extent to which these hypotheses provide a series of mechanisms to link the intervention with its expected outcomes. The model in Fig. 1, below, summarises these hypotheses (hypotheses used in the trials have been italicised).
- 1.
The
Developing the theory-driven approach further
An analysis of the design of the intervention can reveal the implicit assumptions regarding how the intervention gives rise to the intended outcomes. In all the randomised controlled trials, the intervention was designed such that patients completed a questionnaire measuring some aspect of their HRQoL or health status and this was then fed back to their clinician. In the majority of trials the information was fed back to medical practitioners on a single occasion. Clinicians were given either
Defining and measuring HRQoL and the patient’s perspective
The concept of HRQoL has been criticised for its lack of a theoretical basis and agreement regarding its definition (Hunt, 1997). This plurality has given rise to a wide range of instruments claiming to measure HRQoL that cover different dimensions of the construct, with users of such instruments rarely defining what they mean by the term (Gill & Feinstein, 1994). It has been strongly argued that any conceptualisation of HRQoL needs to prioritise the views of the individual and recognise the
Summary: the intervention
In summary, the ways in which the intervention has been designed within the trials may not have served to maximise the potential value of the feedback on HRQoL information. The instrument used needs to prioritise the views of the individual in order to adequately reflect their HRQoL. Clinicians other than medical practitioners also care for patients and may find HRQoL information more useful. Many of the decisions that clinicians make about patient management (and that HRQoL information aims to
Mechanisms implicit within the hypotheses
The mechanisms implicit within the trial hypotheses specify the conditions that must be met in order to achieve the outcomes B-F in the model (Fig. 1). These are that patients want to talk about their HRQoL clinicians, clinicians feel it is appropriate to discuss HRQoL issues with their patients and see the information from such instruments as sufficiently clinical important to warrant a change to the patient's treatment. The extent to which these conditions are, or can, be met also provides a
Implicit mechanisms: summary
These studies suggest that current clinical practice is some distance from the ideal set out in the mechanisms implicit within the randomised controlled trials evaluating the efficacy of this intervention. Patients themselves vary in their wishes to discuss HRQoL issues and there is often a mismatch between their desires and the clinicians’ willingness or ability to discuss specific HRQoL domains. Although clinicians report giving a high priority to HRQoL in decision making, in reality, HRQoL
Summary and conclusions
This paper has applied a theory-driven approach to the feedback of health status measures to clinicians to bring together evidence of both whether and how this intervention is effective, and has reviewed a number of explanations for the lack of impact of HRQoL measures on clinical decision making. Two main observations can be made.
Firstly, unless the mechanisms through which the expected outcomes will be achieved are explicitly specified, the appropriateness of the outcome criteria used to
Acknowledgements
The authors would like Elaine McColl, Jane Blazeby, Alison Brettle and Andrew Brown for their comments on earlier drafts of this paper. JG is funded by the North West Office of the Department of Health on a post-doctoral research training fellowship.
References (59)
- et al.
Quality of lifeA dynamic construct
Social Science & Medicine
(1997) Assessing the quality of life of patients in phase I and II anti-cancer drug trialsInterviews versus questionnaires
Social Science & Medicine
(2003)- et al.
Patient–nephrologist discussions about kidney transplantation as a treatment option
Advances in Renal Replacement Therapy
(2000) - et al.
From best evidence to best practiceEffective implementation of change in patient's care
Lancet
(2003) - et al.
Health status reports in the care of patient with rheumatoid arthritis
Journal of Clinical Epidemiology
(1990) - et al.
Measuring outcomes in children's rehabilitationA decision protocol
Archives of Physical Medicine and Rehabilitation
(1999) Decision-making in nephrologyShared decision making?
Patient Education and Counseling
(2000)Listening to respondentsA qualitative assessment of the short Form 36 health status questionniare
Social Science & Medicine
(2002)- et al.
Monitoring dialysis patients’ health status
American Journal of Kidney Diseases
(1994) - et al.
Patients’ assessment of quality of life instrumentsA randomised study of SIP, SF-36 and SEIQoL-DW in patients with amyotrophic lateral sclerosis
Journal of the Neurological Sciences
(2001)