Elsevier

Ophthalmology

Volume 119, Issue 11, November 2012, Pages 2351-2357
Ophthalmology

Original article
Rasch Analysis Reveals Problems with Multiplicative Scoring in the Macular Disease Quality of Life Questionnaire

https://doi.org/10.1016/j.ophtha.2012.05.031Get rights and content

Purpose

To evaluate validity and psychometric characteristics of the Macular Disease Quality of Life questionnaire (MacDQoL), a multiplicative rating scale designed to measure vision-related quality of life (VRQoL) in macular diseases and age-related macular degeneration (AMD).

Design

Cross-sectional study.

Participants

We included 108 patients with neovascular AMD at baseline before ranibizumab treatment.

Methods

The psychometric properties of the MacDQoL were assessed using Rasch analysis, exploring key indices such as response category functioning, instrument unidimensionality, discriminant ability, and targeting of item difficulty to patient ability.

Main Outcome Measures

Measurement characteristics of the MacDQoL.

Results

In the MacDQoL's native form, the majority of response categories were underutilized and thresholds disordered. This could not be remedied without eliminating the importance ratings owing to the ambiguous nature of the response categories. Scaling problems were resolved by using the impairment rating scale only and collapsing response categories to 4. However, the MacDQoL was multidimensional, necessitating the omission of a number of items and splitting it into an activity limitation and mobility and a socioemotional well-being scale. This improved the psychometric parameters of the revised MacDQoL, although no correlation with clinical measures such as visual acuity was found.

Conclusions

The multiplicative rating scale of the MacDQoL is flawed and does not provide scientific measurement of VRQoL. Measurement can be restored with a series of revisions to the instrument. This study reinforces the importance of considering rating scale design when choosing patient reported outcomes instruments for healthcare research.

Financial Disclosure(s)

The authors have no proprietary or commercial interest in any of the materials discussed in this article.

Section snippets

Patients and Methods

We recruited 108 patients as part of the noninterventional, prospective WAVE (Lucentis in Wet AMD: Evaluation of Visual Acuity and Quality of Life) trial, a postmarketing survey of ranibizumab for nvAMD conducted in Germany from March 2008 to March 2010.12 Ethics approval was obtained from the Institutional review boards of all participating centers. All patients gave signed informed consent for study participation before enrollment and start of treatment. The study adhered to the tenets of the

Clinical Characteristics

Of the 108 patients, most were female (n = 72; 67%), and the average age was 77 years, with two-thirds of the sample aged >75 years (Table 1). Twenty percent of patients had had prior treatment for nvAMD (photodynamic therapy, laser photocoagulation, intravitreal bevacizumab injections). Mean presenting visual acuity was LogMAR 0.58±0.33 (mean value ± standard deviation). Two thirds of patients were classified as vision impaired (LogMAR >0.05; Table 1).

Discussion

Using Rasch analysis, we have demonstrated that the MacDQoL in its current form fails to meet the criteria of a valid functioning scale. In its multiplicative form, several disordered thresholds beyond repair suggested that respondents were unable to discriminate between the large number of response options required by the MacDQoL. Omitting the importance rating and using the impairment scale only also revealed issues with the response categories. However, after collapsing 2 response categories

References (31)

  • J. Mitchell et al.

    Design of an individualised measure of the impact of macular disease on quality of life (the MacDQoL)

    Qual Life Res

    (2004)
  • F. Scheibler et al.

    [Patient-reported and patient-weighted outcomes in ophthalmology]

    Ophthalmologe

    (2010)
  • J. Mitchell et al.

    Psychometric evaluation of the MacDQoL individualised measure of the impact of macular degeneration on quality of life

    Health Qual Life Outcomes

    (2005)
  • V.K. Gothwal et al.

    Multiplicative rating scales do not enable measurement of vision-related quality of life

    Clin Exp Optom

    (2011)
  • F. Denny et al.

    Rasch analysis of the daily living tasks dependent on vision (DLTV)

    Invest Ophthalmol Vis Sci

    (2007)
  • Cited by (28)

    • Using Rasch analysis to evaluate the psychometric functioning of the other-directed, lighthearted, intellectual, and whimsical (OLIW) adult playfulness scale

      2021, International Journal of Educational Research Open
      Citation Excerpt :

      Targeting can be evaluated by looking at the difference between the average person measure and the average item measure of the dataset. Finger et al. (2012) suggested that a difference below 1.00 logits represents good targeting of items and persons. Floor and ceiling effects concern the presence of respondents who are at the highest and lowest ends of the scale.

    • Assessment of patient-reported outcomes in retinal diseases: a systematic review

      2017, Survey of Ophthalmology
      Citation Excerpt :

      The 7 PRO instruments (ALQ, FRQ, Low Luminance Questionnaire, Age-related Macular Degeneration Health Impact Questionnaire, Discomfort Anxiety Fear Questionnaire, and NVQ-10) performed poorly against our quality criteria. Only the MacDQoL and the DLTV were assessed with Rasch analysis.53,65 The original version of the MacDQoL is flawed owing to its complex multiplicative rating scale and multidimensionality; however, after several revisions, 2 of its subscales (activity limitation and mobility and socioemotional well-being) are recommended for use in AMD.65

    • The impact of anti-vascular endothelial growth factor treatment on quality of life in neovascular age-related macular degeneration

      2014, Ophthalmology
      Citation Excerpt :

      Several techniques are available to determine how well items fit the latent trait being measured, how well the items discriminate between the respondents, and how well item difficulty targets person ability.19 We used the following steps to assess the psychometric properties of the IVI, the process of which has been described in more detail by Finger et al.20 Because the IVI consists of 3 subscales, items were split and all analyses were conducted on the basis of the 3 subscales: Reading and Accessing Information, Mobility and Independence, and Emotional Well-being. Response category threshold ordering was examined in regard to overuse or underuse of response categories, as well as the ability of respondents to discriminate between response categories using item reliability and separation.

    View all citing articles on Scopus

    Manuscript no. 2011-1715.

    Financial Disclosure(s): The authors have no proprietary or commercial interest in any of the materials discussed in this article.

    Funded by the German Research Council grant (DFG FI 1540/5-1) to RPF. CERA receives Operational Infrastructure Support from the Victorian Government. This study was supported by Novartis Pharma GmbH, Germany.

    View full text