Original Article
Mind the MIC: large variation among populations and methods

https://doi.org/10.1016/j.jclinepi.2009.08.010Get rights and content

Abstract

Objective

There is no consensus on the best method to determine the minimal important change (MIC) of patient-reported outcomes. Recent publications recommend the use of multiple methods. Our aim was to assess whether different methods lead to consistent values for the MIC.

Study Design and Setting

We used two commonly used anchor-based methods and three commonly used distribution-based methods to determine the MIC of the subscales: pain and physical functioning of the Western Ontario and McMaster University Osteoarthritis Index questionnaire in five different studies involving patients with hip or knee complaints. We repeated the anchor-based methods using relative change scores, to adjust for baseline scores.

Results

We found large variation in MIC values by the same method across studies and across different methods within studies. We consider it unlikely that this variation can be explained by differences between disease groups, disease severity, or lengths of follow-up. The variation persisted when using relative change scores. It was not possible to conclude whether this variation is because of true differences in MIC values between populations or to conceptual and methodological problems of the MIC methods.

Conclusion

To better disentangle these two possible explanations, the MIC methodology should be improved and standardized. In the meantime, caution is needed when interpreting and using published MIC values.

Introduction

Patient-reported outcomes (PROs) have become important outcome measures, because they supplement what is known about the effectiveness of interventions based on the clinician's perspective or physiological measures. If we want PRO instruments to be accepted as primary outcomes in clinical studies, we need to know the extent to which changes in scores on the instrument reflect changes in health status that the patients would consider important.

One problem is the lack of consensus about the best methodology to determine this minimal important change (MIC). Broadly speaking, there are two main approaches: anchor-based methods, which use an external criterion or “anchor” to define an important change (often a patient-based judgment) and distribution-based methods, which use statistical measures as a value for MIC. Within these two methods is a range of approaches to the actual measurement of the MIC. Recent publications recommend to use multiple MIC methods followed by triangulation (an approach to synthesize data from multiple sources) by eyeballing into one value or a small range of values for the MIC [1], [2], [3], [4], [5].

The aim of this study was to assess whether different commonly used methods lead to more or less consistent values for the MIC when applying these methods to data from five different studies.

Section snippets

Application of different MIC methods to the same data

We used two commonly used anchor-based methods and three commonly used distribution-based methods to determine the MIC of the subscales pain and physical functioning of the Western Ontario and McMaster University Osteoarthritis Index (WOMAC). The WOMAC is one of the most commonly used, and most extensively validated, outcome measures for patients with osteoarthritis [6]. It consists of three subscales measuring pain, stiffness, and physical functioning. We used the version with five-point

Results

All the MIC values for the WOMAC subscales pain and physical functioning are presented in Table 2. We found extreme wide and unsystematic variation in MIC values, ranging from −4.6 to 70.8 points (with 95% of the values lying between −3.0 and 29.7) for the subscale pain (Table 2) and from −37.4 to 59.3 (with 95% of the values lying between −2.8 and 23.7) for the subscale physical functioning (Table 3) (hip and knee combined).

We found similar variability in MIC values when using relative change

Discussion

Our results show that instead of converging into a small range of values, our MIC values are extremely variable. We are not aware of any guidelines or statistical tests for how homogeneous MIC values have to be to enable triangulation. However, regardless of the method of triangulation, we think that the variation found in this study was so large, that it is difficult to recommend any MIC value for the WOMAC for any patient subgroup.

Conclusions and recommendations for further research

We found large variation and lack of convergence in MIC values by the same method across studies and across different methods within studies. It is not possible to conclude whether this variation is due to true differences in MIC values between populations and situations or to differences in conceptual and methodological problems of the MIC methods. The answer might be a compromise, that is, that both explanations are playing a part and contributing to the variations observed in our analyses.

References (43)

  • R.R. Coeytaux et al.

    Four methods of estimating the minimal important difference score were compared to establish a clinically significant change in Headache Impact Test

    J Clin Epidemiol

    (2006)
  • E.F. Juniper et al.

    Determining a minimal important change in a disease-specific quality of life questionnaire

    J Clin Epidemiol

    (1994)
  • R.W. Ostelo et al.

    24-item Roland-Morris Disability Questionnaire was preferred out of six functional status questionnaires for post-lumbar disc surgery

    J Clin Epidemiol

    (2004)
  • F. Salaffi et al.

    Minimal clinically important changes in chronic musculoskeletal pain intensity measured on a numerical rating scale

    Eur J Pain

    (2004)
  • F. Tubach et al.

    The variability in minimal clinically important difference and patient acceptable symptomatic state values did not have an impact on treatment effect estimates

    J Clin Epidemiol

    (2009)
  • D. Turner et al.

    Using the entire cohort in the receiver operating characteristic analysis maximizes precision of the minimal important difference

    J Clin Epidemiol

    (2009)
  • U.S. Department of Health and Human Services Food and Drug Administration, Center for Drug Evaluation and Research...
  • D.A. Revicki et al.

    Responsiveness and minimal important differences for patient reported outcomes

    Health Qual Life Outcomes

    (2006)
  • K.W. Wyrwich et al.

    Triangulating patient and clinician perspectives on clinically important differences in health-related quality of life among patients with heart disease

    Health Serv Res

    (2007)
  • N.K. Leidy et al.

    Bridging the gap: using triangulation methodology to estimate minimal clinically important differences (MCIDs)

    COPD

    (2005)
  • C. Veenhof et al.

    Psychometric evaluation of osteoarthritis questionnaires: a systematic review of the literature

    Arthritis Rheum

    (2006)
  • Cited by (247)

    View all citing articles on Scopus
    View full text