Original Article
A multilevel item response theory model was investigated for longitudinal vision-related quality-of-life data

https://doi.org/10.1016/j.jclinepi.2009.06.012Get rights and content

Abstract

Objective

To investigate how a multilevel item response theory (IRT) model for longitudinal dependent data could provide average and individual quality-of-life outcomes of low-vision rehabilitation.

Study Design and Setting

In a nonrandomized longitudinal design, visually impaired older patients (n = 296) were referred to multidisciplinary rehabilitation or to an optometric service. The five-dimensional Low Vision Quality of Life Questionnaire was administered at four time points. The IRT model was characterized by the graded response model for rating scales. Covariates were added to the model, mainly to correct for missing data. The invariance assumption across time points was investigated.

Results

Average and individual rehabilitation effects were estimated. For multidisciplinary rehabilitation, significant average deterioration was seen on three dimensions after 4.4 years. Many individuals in the optometric service group significantly improved on the “reading small print” dimension (18.5%); in both groups, many individuals significantly deteriorated on “visual (motor) skills” (22.2–30.0%). Invariance across time points could be assumed for all dimensions, except for “adjustment.” Gender, education, visual acuity, and health status were significantly associated with the outcome.

Conclusion

We present how a multilevel IRT model can be applied to describe longitudinal dependent vision-related quality-of-life data, while focusing on average and individual effects.

Introduction

Many eye conditions in the elderly, such as age-related macular degeneration or diabetic retinopathy, are irreversible, progressive, and may lead to visual disability. In several countries, low-vision rehabilitation programs have been set up to improve visual ability by teaching patients skills, for example, in the areas of reading, mobility, and daily activities [1]. Moreover, by improving ability, it is expected that quality of life with respect to vision-related issues will also be enhanced. Previously, it has been investigated with different research strategies whether low-vision rehabilitation services met these expectations, that is, improving vision-related quality of life (VRQOL): first, most research has focused on the statistical significance of average rehabilitation outcomes of patient groups. In addition, overall effects are important for low-vision rehabilitation services to be able to determine or adjust their policies. However, even a small advantage of a low-vision rehabilitation program, when multiplied by large numbers of potential patients, could translate into a benefit for many people [2]. Moreover, in daily practice, rehabilitation workers might be more interested in which individual patients improved or deteriorated and less in an overall rehabilitation effect. Second, researchers have chosen different follow-up periods to evaluate low-vision rehabilitation in terms of VRQOL in elderly populations, but have generally not exceeded 1 year postrehabilitation [3], [4]. Although we expect that loss to follow-up in a longitudinal study in an older population will affect the outcome, we consider it important to know whether we can expect older patients to still experience some longer benefit after rehabilitation. Therefore, loss to follow-up should be taken into account [5], [6].

Furthermore, the concept of VRQOL has often been evaluated with classical test theory models [7]. In the field of low vision, a paradigm shift toward the use of Rasch analysis has been reported [8], because many researchers discovered the benefits of describing their patient-reported outcomes with these models. Rasch models are considered to be a special case of item response theory (IRT) models, and provide a more sophisticated way of analyzing data compared with classical test theory models. IRT models are statistical models of the relationship between a person's latent score on the construct being measured and their probability of choosing a response category on each item measuring that construct [9]. These general IRT models may also have the possibility to describe longitudinal patient-reported outcomes [9], [10], [11].

We recently investigated a multilevel IRT model (i.e., the graded response model for rating scales) to describe our longitudinal dependent data [12], [13], [14]. The main aim was to model change over time, and an advantage was that individual change could be directly estimated from the model. The model was described for an observational study in which visually impaired older patients (mean age: 78 years at baseline) were referred to two types of low-vision rehabilitation services [3], [12].

In the present study, we focus on the long-term outcome of our follow-up study, that is, to establish whether the VRQOL of our visually impaired older patients had changed 4.4 years after rehabilitation, either as a group or individually.

Section snippets

Design and patients

VRQOL was measured in a nonrandomized long-term follow-up study in two different low-vision rehabilitation services in The Netherlands [3], [12]. Consecutive patients with a need for low-vision rehabilitation were recruited from four ophthalmology departments in The Netherlands between July 2000 and January 2003. The criteria for inclusion in the study were referral to low-vision rehabilitation services by an ophthalmologist, age over 50 years, no previous contact with low-vision rehabilitation

Nonresponse and loss to follow-up

A total of 357 persons were eligible for inclusion in the study, but 17.1% did not participate. Fifty patients who completed the baseline measurements (n = 296) were lost to follow-up at t2 (16.9%), 31 (10.5%) at t3, and another 106 (35.8%) at t4. Reasons for loss to follow-up between t1 and t4 were mortality (24.7%), physical or mental inability to participate (16.9%), refusal by patient (9.8%), moved and not traced (4.7%), or unknown (7.1%). This means that 109 patients (36.8%) still

Discussion

In this study, we investigated the long-term VRQOL outcome of low-vision rehabilitation in visually impaired older patients. First, we discuss the multilevel IRT model which we developed to describe our longitudinal dependent data. The model was characterized by the graded response model [23], [24], [25] for rating scales [26]. It was useful to be able to estimate individual change directly from the model [12]. These random effects were presented as significant improvement or deterioration

Acknowledgments

Financial support for this study was provided by “ZonMw Inzicht” (The Netherlands Organisation for Health Research and Development—Insight Society, Grant no. 943-03-017), The Hague, the “Stichting Oogfonds Nederland”, Utrecht and, “Stichting Blindenhulp,” The Hague, The Netherlands.

References (36)

  • M.O. Edelen et al.

    Applying item response theory (IRT) modeling to questionnaire development, evaluation, and refinement

    Qual Life Res

    (2007)
  • B.B. Reeve et al.

    Psychometric evaluation and calibration of health-related quality of life item banks. Plans for the Patient-Reported Outcomes Measurement Information System (PROMIS)

    Med Care

    (2007)
  • J.B. Bjorner et al.

    Developing tailored instruments: item banking and computerized adaptive assessment

    Qual Life Res

    (2007)
  • R.M.A. van Nispen et al.

    Applying multilevel item response theory to vision-related quality of life in Dutch visually impaired elderly

    Optom Vis Sci

    (2007)
  • D.A. Pastor et al.

    Longitudinal Rasch modeling in the context of psychotherapy outcomes assessment

    Appl Psychol Meas

    (2006)
  • J.M. te Marvelde et al.

    Application of multidimensional item response theory models to longitudinal data

    Educ Psychol Meas

    (2006)
  • M.R. de Boer et al.

    Evidence-based guidelines on the referral of visually impaired persons to low vision services

    Eur J Ophthalmol

    (2005)
  • N.A. Frost et al.

    Development of a questionnaire for measurement of vision-related quality of life

    Ophthalmic Epidemiol

    (1998)
  • View full text