To sample a broad range of personal experiences with impaired vision, participants in the study were recruited from the Vision Australia Foundation (VAF) a low-vision service provider, the Royal Victorian Eye and Ear Hospital (RVEEH), and independent self-help groups for people with impaired vision. Eligibility requirements for inclusion in the study were the ability to converse in English, visual acuity less than 6/12 and/or impaired visual fields, and age 18 years or older. All participants were informed of the nature and purpose of the study and gave witnessed consent to participate. Participants were recruited between March 1, 2000, and January 31, 2001. The study received ethics approval from the Royal Victorian Eye and Ear Hospital in accordance with the Declaration of Helsinki for research involving human subjects.
Participants recruited from the RVEEH and VAF had clinical eye examinations that included measurement of visual acuity and subjective refraction as a routine part of their appointments at these agencies. Best corrected visual acuity, near visual acuity, and main cause of vision loss were abstracted from clinic files for these participants. Distance visual acuity and main cause of vision loss for members of self-help groups was by self-report and was confirmed using a directional-E screening test. Distance visual acuity was grouped into three categories: less than 6/12 to 6/18, less than 6/18 to 6/60, and less than 6/60, reflecting vision at which functional deficits have been noted,
18 19 20 21 22 according to the World Health Organization’s definition of low vision and severe vision impairment.
17 Those with visual field loss and visual acuity of 6/12 or better were categorized separately. All participants with visual field loss alone attended clinics at RVEEH, and visual fields were measured on an automated visual field analyzer (Humphrey Systems, Dublin, CA) using a standard program (24-2 Swedish interactive test algorithm [SITA]; Humphrey Systems). Visual field loss was noted when the results for the hemifield test (Humphrey Systems) were outside normal limits. Near vision was categorized as N8, N20, or N48 or less. Near vision was not available for members of self-help groups.
Participants answered demographic questions regarding age, duration of vision impairment, hearing loss, other health conditions, and limitations to performance of daily activities attributed to other health conditions. Participants also completed the standard 4-week Short Form-12 (SF-12, version 1; Physical and Mental Health Summary Scales)
23 and completed the IVI questionnaire. The SF-12 is a short version of the SF-36, a general health-related quality-of-life instrument. Two summary components related to the physical (physical summary scale; PCS12) and mental (mental summary scale; MCS12) domains of life were calculated from the questions in the SF-12, according to the algorithm developed by Ware et al.
23
All questionnaires were administered by interviewer to the person with impaired vision. All interviewers were trained to administer the IVI, using a questionnaire protocol developed for the study. Proxy answers were not solicited from caregivers or relatives to avoid biasing the IVI responses to the perception of another person’s opinion of the participant’s ability. A subset of participants completed the IVI on two occasions, to evaluate the reliability of the IVI between two interviewers and interviewer and self-administration and over a short period (1–2 weeks), before receiving vision support or rehabilitation services. Only participants with visual acuity sufficient to read large print (18-point font) were asked to self-administer questionnaires.
The IVI questionnaire was developed in three stages. Initially, focus groups to identify important activities that result in restrictions of participation were conducted with people who have the most common causes of impaired vision in Australia.
24 Issues explored included activities in mobility, household and personal care, consumer and social interactions, and leisure and work life domains. As a result of unstructured issue identification with these groups, a fifth domain was identified: emotional reaction to vision loss.
In the second stage of development, issues identified in focus groups were operationalized into a bank of 76 items. Existing instruments, including the Activities of Daily Vision Scale, the Visual Function (VF)14 questionnaire, the National Eye Institute-Visual Function Questionnaire (NEI-VFQ), and the Bristol Vision-Related Quality of Life (VQOL), were reviewed for content and scaling relevant to the issues identified in focus groups. The initial item bank was drawn mainly from the 139 questions in the VQOL and included 69 items with content that related specifically to restriction of participation and reflected the issues identified in focus groups. The content of many of the VQOL questions tends to overlap that of the other questionnaires reviewed, but question phrasing of the VQOL tends to be more closely aligned with the concept of restriction of participation. The VQOL contains a core of 10 items, and this core is included in the IVI. An additional seven questions were derived in areas in which the VQOL did not provide adequate coverage of issues identified in focus groups. Items regarding ocular symptoms and a person’s limitation in activities—for example, in seeing small objects—were not included in the IVI.
Trials of the IVI were then conducted in two consecutive versions of the questionnaire before deriving the final 32-item version. Trial version 1, containing the original 76 items, was administered to 96 people with impaired vision.
25 Trial version 2 was derived from trial version 1, with the number of items reduced to 45 by eliminating questions with a high degree of correlation (
r > 0.70) and items with floor and ceiling effects. Trial version 2 was administered to 256 people with impaired vision who had not participated in testing trial version 1. Data from this administration were then used to further reduce the number of items by eliminating similar questions with a high degree of correlation (
r > 0.60) and eigenvalue loading of less than 0.45 in the first component in principal-components analysis.
26 In both stages of item reduction through elimination of questions with high correlation, decisions regarding which items with similar content to discard were made in consultation with experienced vision rehabilitation professionals. All versions of the IVI have retained the core 10 questions of the VQOL and all have retained questions in each of the five provisional domains.
All versions of the IVI have used the question format of the VQOL.
16 Participants were asked how much their eyesight deficiency had interfered with an activity “in the past month.” Responses to items were rated from “not at all” (0), “rarely” (1), “a little” (2), “a fair amount” (3), and “a lot” (4) to “all the time” (5) and for some items, “can’t do because of eyesight” (5) or “don’t do because of other reasons” (8). Total and domain scores are an arithmetic average of the rating for applicable items. That is, when a participant rated an item 8, it was not included in the final average score. Items with similar content were provisionally grouped into domains in which specific rehabilitation interventions are directed. For example, in the mobility domain, intervention might include stair marking and orientation training, and in the household and personal care domain it might include provision of adaptive devices to aid in meal preparation.
All data were doubly entered in Access (Microsoft Access 97; Microsoft Corporation, Redmond, WA) and entry errors were confirmed by review of the participant file. Statistical analyses was performed by computer (SPSS version 10; SPSS Science, Chicago, IL), and
P < 0.05 was considered statistically significant. Only data from administration of the final 32-item version of the IVI are included in the analysis. In cases in which the participant completed more than one IVI for the reliability part of the study, the results from the first administration were used. (A trained interviewer always performed the first administration.) Bivariate Spearman correlation was used to determine the correlation between IVI item responses, item responses and the total IVI score, and each IVI domain and the total score with demographic characteristics. Analyses using visual acuity used the distance and near visual acuity groups at initial examination that have been defined previously. Analyses including near vision exclude members of self-help groups. Analysis of variance was used to determine difference between means of continuous variables for characteristics with more than two levels of categorization and
t-tests were used to determine difference in means between variables with two levels of categorization. The χ
2 analysis was used to evaluate differences in proportions. Internal consistency of the IVI and its domains was evaluated with Cronbach α, and Guttman split-half correlation was used to evaluate reliability between different forms of administration. The SE of measurement for reliability analyses was calculated according to the method of Fleiss,
27 based on the intraclass correlation coefficient.
Principal-components analysis was used to examine the underlying structure of the IVI. Sampling adequacy for principal-components analysis was evaluated with the Kaiser-Meyer-Olkin statistic greater than 0.5, and the Bartlett test of sphericity
P < 0.05 was used to determine that the population correlation matrix is not an identity. An initial principal-component extraction was used to determine the number of components needed to represent the data, by using eigenvalues greater than 1.0 or examination of the scree plot to define the number of components in the data. To simplify and interpret the components, a principal-components extraction specifying the number of components identified in the initial extraction was conducted using varimax rotation to transform the initial matrix. Items with loading less than 0.45 were considered unrelated to a factor.
28