Psychometric Properties of Scales for Assessing the Vision-related Quality of Life of People with Low Vision: A Systematic Review

ABSTRACT

Objective

To synthesize information on the psychometric properties of scales used to assess vision-related quality of life in people with low vision.

Methods

A systematic review was conducted. The Cochrane Library, Embase, PubMed, Bireme and Epistemonikos databases were consulted in July 2020. Eligibility assessment of abstracts and full texts was performed independently by two investigators.

A standardized template was used for data extraction regarding study design, scale and version, clinical condition of participants, and psychometric properties measured, using database-specific controlled vocabulary terms for low vision and keywords for vision-related quality of life and validity. Data was synthesized considering two approaches for scales validations, Classical Test Theory and Rasch Analysis.

Results

A total of 53 articles were included in our analysis. In total, 40 studies evaluated the NEI VFQ scale, four evaluated the IVI scale, two evaluated the VA LV VFQ instrument and seven validated the LVQOL scale. This review found that the VRQoL NEI VFQ, IVI, LVQOL and VA LV VFQ-48 scales have adequate psychometric properties, with good internal consistency, when assessed using the CTT approach. The NEI VFQ scale also showed adequate test-retest reliability and adequate construct and content validity. The NEI VFQ and LVQOL scales showed inadequate items and multidimensionality when Rasch analysis was used. The IVI scale showed potential for assessing change in HRQoL after providing interventions to patients with low vision

Conclusions

Many questionnaires exist to measure vision-related quality of life in people with low vision, but the psychometric properties of the questionnaires are variable.

KEYWORDS:

• Low vision
• vision-related quality of life
• validity
• reliability
• NEI VFQ
• IVI
• LVQOL
• VA LV VFQ-48

Introduction

Low vision is included in the functional spectrum of people with visual impairment. Low vision is considered a health condition that can compromise an individual’s functioning and is characterized by the presence of a potentially usable visual remnant.¹ According to the World Health Organization (WHO), classification of low vision should be made in the best-corrected ranges of visual acuity (VA) in the better or single eye after best treatment, management, or possible optical correction. Accordingly, low vision is understood as VA worse than 20/60 but equal to or better than 20/400, or a visual field loss corresponding to < 20°.²

According to the WHO World report on disability, 15% of the world’s population has a disability, which means almost 1 billion people. Nearly 200 million people experience considerable difficulties in their daily lives.³ It was estimated by 2020 that 43 million people are living with blindness, while 295 million people experience moderate to severe visual impairment (low vision), which is 3.74% of the world’s population, with 37.4 cases per 1,000 people.⁴

Low vision can compromise different dimensions of quality of life, performance in activities of daily living, autonomy, health, well-being, and self-experience with low vision. During the 1980s, VA was the main and only outcome of interest within the scope of ophthalmologic evaluations.¹ However, currently, it is concluded that ophthalmologic assessment and medical follow-up focused on VA alone cannot capture or recognize all aspects of visual functioning and quality of life from a person’s perspective.⁵ There is needed general scales that measure health-related quality of life -HRQoL (i.e., how disease affects well-being) and specific instruments focused on vision-related quality of life -VRQoL (i.e., how the degree of vision impairment affects well-being).⁶

The use of quality of life scales and instruments in people with low vision is essential because they provide valuable information for continuing, terminating or redirecting vision rehabilitation treatments and processes.^7,8 Although there are several studies that evaluate the performance of scales designed to measure vision-related quality of life in people with low vision, to this date and to our knowledge, there is no systematic review aimed to identify the psychometric properties and their specific uses. The aim was synthesizing information on the psychometric properties of these scales, to support clinicians and researchers in selecting the most appropriate scale according to both treatment and patient follow-up goals.

Materials and methods

Search strategy

The Cochrane Library, Embase, PubMed, Bireme, and Epistemonikos databases were consulted in July 2020, using database-specific controlled vocabulary terms for low-vision and keywords (with some proximity operators) for vision-related quality of life and validity. We also searched gray literature in Current Controlled Trials, TripDatabase, and Google Scholar to identify information related to low-vision scale validations in Latin American countries. (See the search strategy in the supplementary material)

Study selection

The review included studies that met the following criteria: 1) focused on assessing VRQoL in patients with low vision, 2) all types of primary research studies that assessed the validity or reliability of VRQoL tools, 3) no language restrictions. Given the diversity in VRQoL measurement tools, the following were excluded: 1) studies that addressed VRQoL measurement tools infrequently used (less than three studies reporting use of the instrument), as well as 2) studies focusing on VRQoL of blind persons. Eligibility assessment of abstracts was performed independently by two investigators from a team of three. One investigator solved any discrepancies and produced a final list of studies for full-text review.

Data extraction and synthesis of findings

A standardized template was used for data extraction regarding study design (i.e., classical test theory validation, item response theory validation, or both), scale name and version (e.g., number of items and language), clinical condition of participants, study sample size, and psychometric properties measured (i.e., validity and reliability). Studies were not excluded because of their quality; however, findings from lower quality studies should be interpreted with greater caution and with consideration of their limitations.

There was significant heterogeneity among the studies, mainly with respect to populations, interventions, and psychometric measures. A pooled statistical meta-analysis was not possible; therefore, the findings are presented and discussed in narrative format.

Instruments included

The NEI VFQ scale was designed in 1998 to assess the perception of quality of life related to visual function and well-being in people with chronic eye disease. It measures the influence of visual disability and symptoms on physical and mental health domains. The original version has 51 items; however, the 25-item version is widely validated in different languages and countries. The assessment of visual functioning is presented in a score from 0 to 100, where 0 is the lowest and 100 is the best possible VRQoL.⁹

The LVQOL scale was developed in 2000 to study the quality of life in people with low vision.¹⁰ This instrument helps measure the initial quality of life and the changes that occur with the progression of the health condition.⁷ This instrument helps to measure the initial quality of life and the changes that occur with the progression of the health condition. This scale has 25 items, divided into four dimensions: 1) farsightedness, mobility and illumination; 2) adaptation; 3) reading and precision work; and 4) activities of daily living. The final score ranges from 0 to 125 points, where 0 equals low quality of life (high visual impairment) and 125 is a high quality of life (low visual impairment).¹¹

The VA LV VFQ-48 instrument has 48 items, but a short version is available with 20 items. The original scale was designed to measure difficulty in performing activities of daily living before and after a vision rehabilitation process.⁷ A higher score indicates better ability or less difficulty to carry out daily life activities.¹²

Lastly, the IVI scale was developed to appraise needs for vision rehabilitation within the context of limitations and restrictions of participation secondary to altered or deficient vision.^13,14It is an instrument inspired by the WHO’s International Classification of Functioning, Disability, and Health (CIF-2001). The original version contains 32 items grouped into five domains: 1) leisure and work, 2) social and consumer interaction, 3) personal care, 4) mobility, and 5) reaction to the loss of vision. ¹⁴ A 28-items version is available, with items grouped into three subscales.¹⁵ The total score and the score by domain are provided as an arithmetic average classified among 0 (the best score) and 5 (the worst score).

Results

After removing seven duplicates, we screened 700 titles and abstracts and included 53 studies in our analysis (See Figure 1 with the Prisma chart). Articles dated from 1998 to 2019 and represented the following countries: Australia, Brazil, Canada, China, Colombia, Denmark, Egypt, France, Germany, Greece, Italy, Japan, the Netherlands, Norway, Serbia, Spain, Taiwan, Turkey, the United Kingdom, the United States, and Vietnam.

Figure 1. Flow chart of studies selection.

Among the 53 studies included, 42 provided information on the NEI VFQ scale, 32 evaluated the 25-item version;^16–47 six studies evaluated the 39-items version;^{21,31,48–51} we identified one study for each version of 7, 8, 9, 51, and 52 items;^52–55 and three studies evaluated more than one version of the NEI VFQ scale.^21,31,55 Four studies assessed the IVI scale,^13,56–59 three addressed the 32-items version,^13,56,58 and one evaluated the 28-items version.⁶⁰ Two studies assessed the VA LV VFQ scale;^60,61 and seven validated the LVQOL scale.^{10,11,62–66} Compared to the proportion of men, the proportion of women was consistently higher in the 40 study samples. The characteristics of the included studies are described in Supplementary Table 1.

NEI VFQ

Among the studies that validated the NEI VFQ scale, 15 validated the English version, four the Chinese, four the Spanish, three the Turkish, two studies for each versions in Portuguese and Italian, and one for each of the French, Japanese, Dutch, Greek, Danish, Arabic, Vietnamese, Serbian and German versions. The evaluation of the psychometric properties of NEI VFQ was conducted through classical test theory (CTT) (n = 26), using Rasch analysis (n = 11), and three studies used both approaches (Supplementary Table 2 shows details of the studies and their findings).

Studies that evaluated the psychometric properties of the NEI VFQ using CTT reported good internal consistency. Only 10 of the 26 validations via CTT had Cronbach’s alpha < 0.7;^{17,21,23,33,34,36,39,43,48,49} only two studies reported internal consistency < 0.6.^34,36 Test-retest reliability was adequate in all validation studies; both construct and content validity were appropriate in several studies.

The study of development and validation of NEI VFQ-51 in 1998, reported that the psychometric properties of the scale were not influenced by the type or severity of the underlying ocular disease, suggesting that the scale provides valid and reproducible measurements when used across multiple ocular conditions.⁵² Using data from the MARINA and ANCHOR clinical trials, a validation study reported a four-to-six-point change in NEI VFQ-25 scores, representing a clinically significant change, corresponding to a 15-letter change best-corrected visual acuity (BCVA).²⁷ In the study by Revicki et al., a secondary analysis of data from two randomized clinical trials of patients with sub-foveal choroidal neovascularization due to AMD, authors found high internal consistency in the baseline measurement (0.96 for the total score). The complete and subscale scores of the NEI VFQ-25 were significantly correlated with the SF-36 scores (p < .05). The mean of the total scores and the subscales of the NEI VFQ-25, except for eye pain and overall health, varied according to the BCVA group (p < .001), with higher disability scores in the lower visual acuity groups.³⁰

Studies that used a Rasch approach to assess the psychometric properties of the NEI VFQ reported item misfit and multidimensionality, with several studies reporting two dimensions, a visual functioning scale and a socio-emotional scale.^31,32,44 The study by Pesudovs et al. reported that NEI VFQ-39 and NEI VFQ-25 had good precision, but evidence of multidimensionality, item misfit, sub-optimal targeting of the item’s difficulty to the person’s ability, and dysfunctional subscales (8 NEI VFQ-39; 12 NEI VFQ-25).³¹ Langelaan et al. suggested modifications to the original scale structure.⁴⁶ In contrast, Marella et al. proposed dichotomizing some response categories and eliminating three misfit items to improve the fit to the Rasch model.³²

Regarding the NEI VFQ-25 scale structure, six of the 12 original subscales could not fit the Rasch model due to insufficient items (less than two items). The remaining six subscales showed poor fit characteristics, which indicates that the NEI VFQ-25 does not have a viable subscale structure.³¹

The principal component factor analysis supported a two-factor model with visual (10 items) and socio-emotional (9 items) functioning scales. Most of the goodness-of-fit statistics were within the recommended range of values. The factorial loadings of the items in their respective subscales were statistically significant (p < .001) and ranged between 0.59 and 0.84. The two scales were individually adjusted to the Rasch model and found to be unidimensional with adequate psychometric characteristics.³²

The items showing the highest rate of non-response and adjustment problems were related to the driving subscale, given that many participants did not drive. This problem was reported in the validations of the versions in Spanish ¹⁷⁴¹, French,⁴⁵ Japanese,¹⁹ Chinese,²⁶ Chinese-Taiwanese,²⁸ and English.²⁹

IVI

Four studies were identified that evaluated the IVI; two evaluated the psychometric properties of the English version, one of the Chinese and one of the Thai versions. Two studies used a TCT approach, two used a Rasch analysis (See Supplementary Table 2 for details of studies included and findings).

The IVI scale showed validity, reliability, and sensitivity to change. Ratanasukon et al. reported high consistency in all subscales with a Cronbach’s alpha ranging from 0.787 to 0.849; test-retest reliability was also high (intraclass correlation coefficient = 0.96).⁵⁶ A Rasch analysis of the Thai version revealed the validity of this version across the three subscales.⁵⁶ The composite score of the IVI was significantly higher in participants with visual impairment than in healthy participants. In addition, scores on the reading and information access and emotional well-being subscales were higher in participants with AMD. Mobility and independence subscale scores were higher among those with cataract or diabetic retinopathy.⁵⁶ Goldstein et al. reported good separation reliability (R2 = 0.99) of the 28-items version and absence of problems with differential item functioning (DIF).⁵⁷ Studies by Fenwick et al. and Weih et al. reported psychometric characteristics that support its use in assessing the visual rehabilitation needs of visually impaired persons.^13,58 Those studies also showed their stability over time, indicating that IVI has the potential to evaluate change in VRQoL after providing interventions for patients with low-vision.^13,58

LVQOL

Seven studies assessed the psychometric properties of the LVQOL scale. Two studies evaluated the English version, two evaluated the Dutch version, and we identified one study for each Spanish, Turkish, and Chinese versions. Four studies used a TCT approach, one study used Rasch analysis, one study was only the cross-cultural adaptation, and another was a cross-sectional validation (See Supplementary Table 3 for details of studies included and findings).

Overall, studies that validated the LVQOL scale concluded that is a rapid, reliable, and internally consistent method for measuring the HRQOL of people with visual impairment in a clinical setting.^10,62–66 The LVQOL scale could quantify the quality of life of people with low vision and differentiate it appropriately from people with normal vision; it was also useful in determining the effects of rehabilitation in patients with low vision.^10,62–66 In the Wolffsohn & Cochrane study, unreliable, internally inconsistent, redundant, or not relevant items were excluded, resulting in a 25-item LVQOL version.⁶² The LVQOL scale had high internal consistency (Cronbach’s alpha = 0.88) and good reliability (0.72).¹⁰ The mean LVQOL score for a population with low vision (60.9 ± 25.1) was significantly lower than the mean score for people with normal vision (100.3 ± 20.8).¹⁰ Rehabilitation improved the LVQOL score of people with low vision by an average of 6.8 ± 15.6 (17%).¹⁰ When validations were performed using the Rasch approach, misfit of some items was consistently identified (item 1 ‘overall vision’ in the dimension “adjustment” and item 24 ‘use of tools’ in the dimension “reading and precision work”); by eliminating these items from the scale, its functioning and adjustment improved.^64,65

VA LV VFQ-48

Only two studies validating the VA LV VFQ-48 were included in this systematic review; one assessed the English version and the other one the Italian version (see Supplementary Table 2 for details of studies included and findings). One study used a TCT approach, and the other used Rasch analysis. Stelmack et al. reported item measurement fit statistics, indicating that the responses to 19% of the items were confounded by factors other than visual ability. The separation reliabilities for the pooled data (0.94 for individuals and 0.98 for items) demonstrated that the measurements estimated discriminate well between individuals and items along the visual ability dimension.⁶¹ The Intraclass Correlation Coefficients for the test-retest reliability were 0.98 for the items and 0.84 for the individuals. The mild and moderate difficulty ratings were used interchangeably, suggesting that the instrument could be modified to a 4-point scale, including not difficult, moderately difficult, extremely difficult, and impossible.⁶¹ In the study by Di Maggio et al., the VA LV VFQ-48 scale showed high internal consistency (Cronbach’s alpha: 0.98) and a good item-test and item-total correlation (median: 0.73 and 0.71, respectively). The general and subscale scores (reading, visual motor, mobility and visual information) were significantly correlated with visual acuity, reading acuity, and speed. Reading speed achieved the best absolute correlation with VA LV VFQ-48 scores (Spearman’s r: 0.39–0.49).⁶⁰

Discussion

Overall, this review found that the VRQoL scales NEI VFQ, IVI, LVQOL, and VA LV VFQ-47 have adequate psychometric properties, with good internal consistency, when assessed through the CTT approach. The NEI VFQ scale also showed adequate test-retest reliability, and appropriate construct and content validity. When Rasch analysis was used, the NEI VFQ scale showed items misfit and multidimensionality, suggesting that the original structure of the scale needs to be modified. The LVQOL scale also showed misfit of some items and better functioning and adjustment after eliminating the misfit items.^64,65 The IVI scale showed potential to evaluate change in VRQoL after providing interventions for patients with low-vision.^13,58 The VA LV VFQ-48 scale showed that the mild and moderate difficulty ratings were used interchangeably, suggesting that the instrument could be modified to a 4-point scale, including not difficult, moderately difficult, extremely difficult, and impossible.⁶¹

In forty studies included, women participated more than men, which is somehow explained by the higher reported prevalence of visual impairment in this population group. However, this can be a problem as it is likely that the studies may not be representative of the entire population (lack of generalizability), especially as men may respond differently to quality-of-life questionnaires than women. It is pertinent to carry out further studies that allow greater participation of men. Participants’ conditions commonly reported in the studies were age-related macular degeneration (38 studies), glaucoma (30 studies), diabetic retinopathy (24 studies), and cataract (26 studies). Most of these conditions are ocular diseases directly related to ageing, reported by the World Health Organization² as leading causes of visual disability. This review identified the necessity of including other conditions that cause low vision, which might impact a person’s functioning differently.

The LVQOL scale has a significant correlation (directly proportional) with the NEI VFQ-25 scores. It is commonly used in the clinical setting to assess low vision rehabilitation strategy and management; it is rapid, reliable, and internally consistent. It can quantify the quality of life of people with low vision.¹⁰

A study conducted in Ankara, Turkey, compared quality of life using the LVQOL and the NEI VFQ-25 in 64 adult patients seen for the first time in a low vision rehabilitation department, and found a good correlation between the total scores of the two scales.⁶⁷ The LVQOL allows the identification of needs in daily life and how these can be met in visual rehabilitation processes, which is a strength of the scale.⁶⁷ However, the NEI VFQ-25 stands out for its ability to assess the impact on quality of life of a wide range of eye diseases (age-related macular degeneration, diabetic retinopathy, uveitis, optic neuritis, central retinal vein occlusion, cataract, keratoconus, among others,⁶⁸ which means that the NEI VFQ-25 can be considered the “reference standard”.⁷ Studies suggested that NEI VFQ-25 provides more detailed and valuable information on quality of life compared to the LVQOL, but it requires more time to calculate its scores.⁶⁷ In addition, the NEI VFQ has a neuro-ophthalmological supplement to assess vision-related quality of life in patients with neurologically caused visual impairments, such as multiple sclerosis, myasthenia gravis, and conditions causing diplopia,^68,69 thus expanding its applicability and usefulness in care and research processes. The strength of the VA LV VFQ-48 is that it can measure outcomes focused on patients with low vision,⁶⁰ and has been used in clinical trials assessing rehabilitation protocols in these patients.^70,71

The IVI-28 is a valid and reliable scale to be used in the measurement of limitations in participation related to activities of daily living of patients with moderate and severe low vision, therefore is a valuable tool for using in rehabilitation services.⁷² Additionally, to the IVI-28, Finger et al., developed in 2014 the 28-items scale Vision Impairment-Very Low Vision (IVI-VLV).⁷³ The original study included 603 legally blind people in Australia. This scale can differentiate among different levels of vision-related quality-of-life, with the advantage that the measurement is less affected by general or mental health issues.

A large number of health-related quality of life and vision-related quality of life questionnaires exist; this fact poses challenges for health care professionals and researchers when selecting a measuring tool for using in healthcare or research processes.¹ Although this challenge, quality-of-life questionnaires and structured interviews should be regularly integrated into vision healthcare delivery in different eye care settings.⁷⁴

The prevalence of disability will continue to increase, especially sensory disabilities, including low vision. Therefore, exploring, and documenting health-related quality of life and visual health is crucial to the clinical following up, direct visual rehabilitation care processes, community wellness programs, and public policy on disability. Current concepts of disability move away from a solely medical or bodily perspective because they are based on the context surrounding the person. Therefore, quality of life scales related to visual health show part of this dynamic interaction and the degree to which people perceive how they feel, how they function and how they perceive their context.

It would be important to develop studies with the VRQoL scales adjusted for sociodemographic and environmental characteristics, such as access to health care and vision rehabilitation programs, as the environment and context are important in facilitating or diminishing functioning and disability.

This review has limitations because most studies of vision-related quality of life in people with low vision have been conducted in adult populations, particularly the elderly and people with macular degeneration. However, the psychometric properties of the NEI VFQ are not influenced by the ocular disease causing the visual impairment. The review also highlights those definitions vary widely even though quality of life is routinely assessed in ophthalmologic studies. The Lancet Commission on Global Health calls for the development and adoption of standardized, culturally sensitive quality of life measures to better understand the effect of visual impairment and ophthalmic interventions from the patient’s perspective.⁶

In conclusion, this review identified many questionnaires for measuring health-related quality of life and vision-related quality of life, several with good psychometric properties. Considering the differences in the development, scope and aims of the scales, clinicians should have a clear idea of the purpose of the measurement, the context, and the patient’s characteristics before selecting the appropriate instrument.

Disclosure statement

None of the authors has any financial interest or benefit that has arisen from the direct applications of this research.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/09286586.2022.2093919.

Additional information

Funding

Funding for the study was partially provided by Universidad Santo Tomás, Bucaramanga (Colombia) and Instituto de las Desigualdades (Spain).Universidad Santo Tomás Bucaramanga [CEPO2020; Instituto de las Desigualdades BarcelonaInstituto de las Desigualdades Barcelona [CEPO2020].