Validation and Reliability of the VF-14 Questionnaire in a German Population

Abstract

Purpose.: To evaluate the validity, reliability, and measurement characteristics of the Visual Function 14 (VF-14) in a German sample using Rasch analysis.

Methods.: This was a clinic-based, cross-sectional study with 184 patients with low vision recruited from an outpatient clinic at a German eye hospital. Participants underwent a clinical examination and completed the German VF-14 scale. The validity of the VF-14 scale was assessed using Rasch analysis. The main outcome measure was the overall functional score provided by the VF-14.

Results.: After collapsing two response categories for items 13 and 14, the VF-14 scale satisfied fundamental criteria to achieve fit to the Rasch model, namely, ordered thresholds, the ability to distinguish between different strata of participant ability, absence of misfitting items, no evidence of unidimensionality, and no significant differential item functioning for key sociodemographic covariates. The VF-14 is able to discriminate between participants with different levels of vision impairment and across different cultural groups.

Conclusions.: The VF-14 is a valid, reliable, and unidimensional questionnaire for use in a German population. These findings contribute to the growing evidence base for second generation patient reported outcome measures in ophthalmology, and support the use of the German VF-14 in tertiary eye clinics in Germany to capture the impact of visual impairment on visual function from the patient's perspective and to inform low vision rehabilitation and interventions.

Rigorously validated instruments to assess patient reported outcomes are needed to comprehensively assess the impact of vision loss. ^{1 –8} Over the last two decades, the use of patient-reported outcomes in ophthalmology has gained greater acceptance, ^9,10 as is evidenced by the availability of a large number of instruments to assess the impact of vision loss. ^{11 –14}  

One of the most commonly used vision-related functional questionnaires is the Visual Function 14 (VF-14). ¹⁴ Initially, it was designed to assess vision-related functioning in patients undergoing cataract surgery, but it has also been validated for use with other eye conditions such as glaucoma, ^{15 –17} retinal, ^18,19 and corneal diseases. ²⁰ The VF-14 has a format which makes it easy to administer and has a high rate of patient compliance. Although it has been validated in several languages other than English, ^21,22 its applicability in a German population has yet to be determined. 

Importantly, previous studies assessing the psychometric properties of the VF-14 have used the classical test theory (CTT). However, this method has several limitations. For example, the key measure of reliability in CTT is Cronbach's alpha but this value is not independent of the number of items in the scale and it can be artificially inflated by including additional items even if they are redundant. Traditional methods are also limited in assessing the unidimensionality of the scale; the targeting of the questions to the respondents; and whether bias exists for an item(s) among subgroups in the sample (i.e., sex, age groups, location). ²³ Additionally, traditional methods use Likert scoring which assumes, incorrectly, that response scale categories are of equal distance apart on a measurement scale, and that all questions are equally difficult. Rasch analysis, a form of item response theory, provides a means with which to address these limitations. In Rasch analysis, raw scores are converted to a linear interval scale allowing parametric statistical techniques to be confidently used. ²⁴ To date however, the psychometric validity of the VF-14 for use in a visually-impaired German sample has never been investigated using Rasch analysis. 

The aim of this study is therefore to evaluate the validity, reliability, and measurement characteristics of the VF-14 in a German sample using Rasch analysis. 

Methods

Study Population

Patients and fully-sighted controls were recruited from outpatient clinics at the University Eye Hospital in Bonn, Germany. Eligible patients were ≥18 years of age, able to converse, read, and write in German, presented with a visual acuity equal to or < 6/12 and had a duration of vision impairment equal to or greater than 2 weeks. Controls had visual acuity of at least 6/6 in both eyes, no history of ocular disease, and no current ocular complaints. A brief clinical assessment was done in all controls to confirm the absence of ocular pathology. The controls completed the questionnaires without assistance. 

All patients underwent a complete ophthalmic examination including best-corrected visual acuity, biomicroscopy, intraocular pressure measurements, and funduscopy. Further diagnostic tests (fluorescein angiography, optical coherence tomography, and electrophysiology) were performed as appropriate in each individual case. All participants completed the VF-14 and an additional questionnaire to collect sociodemographic data and information about the patients' past medical history. Questionnaires were either self- or interview-administered by trained interviewers as required. Data collection was completed between March and April 2009. 

Ethical approval was obtained from the ethics committee of the University of Bonn. Informed consent was obtained from every participant before the interview. The study adhered to the tenets of the Declaration of Helsinki. 

Characteristics of the VF-14

The VF-14 questionnaire is based on 14 vision-dependent activities performed in everyday life, such as: reading, recognizing people, seeing steps, doing fine handiwork, writing, playing games, doing sports, cooking/preparing meals, watching television, and driving (Table 1). The difficulty undertaking each activity is rated on a five-category Likert scale: (0) Not possible; (1) a lot of difficulty; (2) some difficulty; (3) a little difficulty; and (4) no difficulty at all, for all questions except for 13 and 14 which are rated on a four-category scale (categories: 1, a lot difficulty; 2, some difficulty; 3, a little difficulty; and 4, no difficulty at all). 

Table 1.

 

Item Content of the VF-14 Used in the German Population

Item Number	Item Description
1	Do you have difficulty reading small letters like in a telephone book or in package inserts for medications (with glasses for close vision)?
2	Do you have difficulty reading a newspaper or a book (with glasses for close vision)?
3	Do you have difficulty reading big headlines and the numbers on your telephone (with glasses for close vision)?
4	Do you have difficulty recognizing people who are standing right in front of you (with glasses)?
5	Do you have difficulty (with glasses) noticing steps, curbs, or stairs?
6	Do you have difficulty (with glasses) noticing traffic, information, or shop signs?
7	Do you difficulty (with glasses) sewing, knitting, crocheting, or doing handicrafts?
8	Do you have difficulty (with glasses) filling out official forms or cheques?
9	Do you have difficulty (with glasses) playing cards, chess, or dominos?
10	Do you have difficulty (with glasses) performing sport activities like tennis, golf, bowling, or boule?
11	Do you have difficulty (with glasses) cooking?
12	Do you have difficulty (with glasses) watching TV?
13	How much difficulty do you have driving during the daytime due to your vision problems?
14	How much difficulty do you have driving during the night due to your vision problems?

Details about the translation of the questionnaire and patient recruitment have been described previously. ^25,26 Briefly, the original VF-14 questionnaire was translated into German and back translated into English, and a pilot test was carried out to assess the acceptability of the new questionnaire. 

Rasch Analysis

Rasch analysis is a modern psychometric method that transforms raw ordinal scores into interval-level measurement (expressed in log of the odds units, or logits), permitting the use of parametric statistical techniques. It calculates item difficulty (item measure) in relation to person ability (person measure) by placing both on the same linear continuum. A high person measure (in logits) indicates that a person possesses a high level of the assessed latent trait (e.g., vision functioning). Rasch analysis also provides insight into the psychometric properties of a scale, such as how well items fit the underlying latent trait being measured, how well items discriminate between the respondents, how well item difficulty targets person ability, and the appropriateness of the response scale used. ²³  

Rasch analysis was applied to the responses of the 184 patients using the Andrich rating scale model ²⁷ with software (Winsteps version 3.68; Chicago, IL). ²⁸ The controls (n = 90) were not included in the analysis. Five fundamental indicators ^{29 –36} were used to assess the validity of the VF-14 in this German sample. They include: 

Response category threshold ordering assesses the appropriateness of the response options. Disordered thresholds indicate that participants have difficulty discriminating between the response options or that they perceive categories as interchangeable. Disordered thresholds may considerably affect model fit; however, this can be overcome by collapsing poorly used response categories with adjacent ones. 

Person separation indices provide an estimate of the ability of the VF-14 questionnaire to discriminate between strata or groups of participant ability. ^{33,37 –39} A person separation index (PSI) of > 2.0 and a person reliability (PR) score of > 0.8 are generally considered to be the minimum requirements for satisfactory discrimination of at least three strata of participants level of the trait being investigated (i.e., vision functioning). ^33,40,41  

Targeting of items to the sample (i.e., whether the items are of an appropriate level of difficulty for the ability of the sample) can be observed from the person-item map. Persons of higher ability and items of greater difficulty are located at the top of the map, and vice versa. Poor targeting occurs when participants generally have a higher or lower ability than the most or least item difficulty threshold, or when items are clustered at particular levels of difficulty leaving large gaps. The map can be used to isolate items for removal that contribute to suboptimal targeting. ³⁶ Item difficulty and person ability are measured in logits. Ideally, the mean person score should be approximately 0 logits. ³⁵ In general, a difference between the mean person and item score of > 1.0 logits indicates notable mistargeting. ³⁶  

Unidimensionality refers to the ability of a scale to measure a single underlying construct (in this case, visual function) in addition to how well each item measures the underlying trait. ³³ The two key indicators for unidimensionality are item ‘fit statistics’ and testing the assumption of local independence. Item fit determines how well each item ‘fits’ the underlying trait and items with an ‘infit’ mean square (MNSQ) value between 0.7 and 1.3 were considered acceptable in this study. ³⁴ Values below 0.7 may indicate redundancy and values over 1.3 indicate an unacceptable level of ‘noise’ in the responses. Such items may be removed from the scale to improve model fit. 

To test for local independence, principal component analysis (PCA) of the residuals is examined, whereby the variance explained by the measures for the empiric calculation should be comparable with that of the model (>50% for an acceptable model). Furthermore, the unexplained variance in the first contrast of the residuals can indicate if any patterns in the differences within the residuals exist large enough to suggest the existence of a secondary dimension. The first contrast of the residuals should be < 3.0 eigenvalue units (<5%) which is close to that seen with random data. An eigenvalue greater than 3.0 is problematic and indicates that there is another underlying trait being captured by the instrument. ²⁸  

Differential item functioning (DIF) indicates whether different groups within the sample are systematically responding differently despite having equal levels of the trait being assessed. Age, sex, and education level were included in the differential item functioning analysis because causes of visual impairment are known to be age- and sex-specific. Education was also considered because it could reflect the level of visual function reported. We used the following criteria for differential item functioning assessment: small or absent if the differential item functioning contrast in logits was <0.50 logits; minimal differential item functioning if the differential item functioning contrast was 0.5 to 1.0 logits; and notable differential item functioning if the differential item functioning contrast was >1.0 logits. If significant and meaningful differential item functioning is found, it may indicate that the interpretation of the scale differs by group and that it is influenced by confounding factor(s). 

Results

Sociodemographic and Clinical Characteristics of the Study Participants

In total, 184 people participated in the study. Overall, the participants' mean ± SD age was 68.9 ± 15.49 years and there were more male (60.7%) than female participants (Table 2). Using presenting vision in the better eye, 71 (38.6%) participants had no visual impairment (6/12 or better), 49 (26.6%) had mild visual impairment (>6/12 to 6/18) and 64 (34.8%) had moderate to severe visual impairment (>6/18). The main cause of vision loss was aged-related macular degeneration (ARMD), followed by ‘other retinal diseases.’ 

Table 2.

 

The Personal and Clinical Characteristics of the Participants and Visual Functioning Scores

Characteristics	n (%)	P
Participants mean ±SD	184 (67.2)
Sex*		0.431
Male	111 (60.7)
Female	72 (39.3)
Age, y		0.929
≤60	48 (26.1)
>60	136 (73.9)
Education*		0.027 †
None/primary school only	96 (52.7)
Some secondary, technical, or commercial schooling	49 (26.9)
Secondary completed	15 (8.2)
Trade, TAFE, or university	22 (12.1)
Work status*		0.027
Working (part- or full-time)	145 (81.5)
Not working (retired, home duties, disability pension, unemployed)	33 (18.5)
Non-ocular comorbid conditions
Yes	176 (95.7)	0.107
Hypertonia	109 (59.2)	0.545
Diabetes	55 (29.9)	0.395
Pulmonary diseases	36 (19.6)	0.047
Cardiovascular diseases	85 (46.2)	0.118
CNS diseases	21 (11.4)	0.062
Restricted mobility	66 (35.9)	0.370
Depression and other mental illness	16 (8.7)	0.387
Neoplastic diseases	28 (15.2)	0.474
Other	64 (34.8)	0.439
Eye conditions		0.480
ARMD	54 (23.9)
Diabetic retinopathy	27 (14.7)
Glaucoma	15 (8.2)
Other retinal disease	49 (26.6)
Corneal diseases	12 (6.5)
Other	27 (14.7)
Presenting unilateral visual impairment		0.000 ‡
None (>6/12)	71 (38.6)
Mild (6/12 to 6/18)	49 (26.6)
Moderate-severe (≤6/18)	64 (34.8)
Self-reported health rating*		0.114
Poor to fair	106 (58.6)
Good	68 (37.6)
Very good to excellent	7 (3.9)

 Bolded P values indicate statistical significance (P < 0.05). SD, standard deviation; CNS, central nervous system; ARMD, age-related macular degeneration.

*  Data incomplete.

†  Significance found between those with no/primary education versus those with secondary level qualifications.

‡  Significance found between all three groups of vision impairment.

Validation of the VF-14 Questionnaire in the German Sample

Threshold Order and Person Separation.

The pattern of item thresholds was first examined for disordering. Evidence of disordered thresholds was found for items 13 and 14. Examination of these items indicated that response category 2 (some difficulty) did not have a point along the ability continuum where it was the most likely response (Fig. 1, left). Consequently, these two items were recoded by collapsing categories 2 (some difficulty) and 3 (a little difficulty) (coded 4332) and ordered thresholds were evident (Fig. 1, right). 

Figure 1.

 

Category probability curves. Left, category probability curves showing disordered thresholds. The response category 2 does not have a range along the ability scale were it is most likely to be chosen. It appears to be interchangeable with categories 1 and 3. Right, category probability curves showing ordered thresholds after categories 2 and 3 were collapsed.

Person Separation Indices.

The PSI and PR values of 3.23 and 0.91, respectively, indicated good discriminant ability of the scale, namely that it was able to distinguish between at least three statistically significant strata (separation) of participant ability. 

Unidimensionality.

All items showed acceptable fit to the model expectation, with each having an ‘infit’ MNSQ value between 0.7 and 1.3 (Table 3). The hardest item was driving at night and the easiest item was recognizing people, with logit scores of 3.29 and −1.88, respectively. 

Table 3.

 

Item Infit Statistics and Item Difficulty Rating (Logits)

Items	Difficulty
	Item Measure (Logit)	Infit MNSQ
Driving at night	3.29	0.88
Reading small letters	1.51	1.20
Knitting	0.82	0.86
Reading newspaper	0.76	0.94
Driving during the day	0.12	1.14
Filling out forms	0.05	0.82
Performing sports activities	−0.11	1.28
Notice traffic/information	−0.33	0.82
Noticing steps	−0.48	0.96
Watching TV	−0.54	0.79
Playing card games	−0.62	0.83
Reading big headlines	−1.27	1.27
Cooking	−1.32	1.19
Recognizing people	−1.88	0.94

No evidence of multidimensionality in the VF-14 scale was observed in this sample. The PCA of the residuals explained 76.6% of the raw variance and the unexplained variance by the first contrast of the residuals was 2.1 eigenvalue units. 

Person-Item Map.

The person-item map in Figure 2 displays the participant scores on the Rasch calibrated scale and the relative difficulty of each of the VF-14 items. Participants with the highest level of vision functioning are located at the top of the figure while those with the lowest vision function are found at the bottom. As demonstrated in Figure 2, the spread of items across the range of participant ability was adequate, although there were few items of sufficient difficulty to target participants at the high end of the scale. Moreover, the difference in item and person mean was 1.16 logits, suggesting that overall participants had a higher level of ability than the average of the scale items (i.e., 0 logits). In other words, most items in the questionnaire were too easy to perform for this sample. 

Figure 2.

 

Person-item map for the Rasch-scaled VF-14 questionnaire. M, mean; S, 1 SD from the mean; T, 2 SDs from the mean.

To the left of the dashed line are the participants, represented by X, and on the right are the items, denoted by their content. Participants with higher ability and the most difficult items are near the top of the diagram, and participants with lower ability and items that can be performed with least difficulty are near the bottom (Fig. 2). 

Differential Item Functioning (DIF).

According to the Rasch model requirements, a scale should function consistently irrespective of the presence of particular subgroups in the sample under assessment. We explored whether particular subgroups in our German sample, such as age, sex, and education level responded in the same way to the VF-14 items. Notable differential item functioning was found for item 10—difficulty performing sport activities like tennis, golf, bowling, or boule (i.e., people aged >60 years were more able to carry out this task than those aged ≤60 years [1.36 logits]), and item 7—difficulty sewing, knitting, crocheting, or doing handicrafts (i.e., females were more able than males to do this task by 0.87 logits). Minimal differential item functioning was again present for item 10 except in the sex variable (i.e., females were more able than males [0.99 logits]), and item 11—difficulty with cooking (i.e., males had slightly more ability than females by 0.59). The rest all had differential item functioning contrasts; all <1.0 (Table 4). These results indicate that the vast majority of items functioned similarly for persons at the same level of ability, irrespective of other factors. 

Table 4.

 

Differential Item Functioning (DIF) by Age, Sex, and Education Level*

Items	DIF by Age				DIF by Sex				DIF by Education
	≤60 y	>60 y	Contrast	Welch's test P-value	Female	Male	Contrast	Welch's test P-value	Low	High	Contrast	Welch's test P-value
VF 1. Do you have difficulty reading small letters like in a telephone book or in package inserts for medications (with glasses for close vision)?
	−1.24	−1.74	0.50	0.0327	−1.54	−1.69	0.15	0.4492	−1.64	−1.40	−0.23	0.3185
VF 2. Do you have difficulty reading a newspaper or a book (with glasses for close vision)?
	−0.79	−1.27	0.48	0.0336	−1.15	−1.15	0.00	1.000	−1.21	−0.89	−1.37	0.1756
VF 3. Do you have difficulty reading big headlines and the numbers on your telephone (with glasses for close vision)?
	0.48	0.35	0.13	0.5807	0.50	0.20	0.30	0.1485	0.37	0.49	−0.12	0.6537
VF 4. Do you have difficulty (with glasses) recognizing people who are standing right in front of you?
	0.77	0.73	0.03	0.8891	0.90	0.47	0.43	0.0467	0.76	0.55	0.21	0.4536
VF 5. Do you have difficulty (with glasses) noticing steps, curbs, or stairs?
	−0.05	−0.31	0.26	0.2548	−0.25	−0.29	0.04	0.8423	−0.25	−0.28	0.03	0.8957
VF 6. Do you have difficulty (with glasses) noticing traffic, information or shop signs?
	−0.47	−0.24	−0.24	0.2940	−0.30	−0.25	−0.05	0.8020	−0.21	−0.61	0.21	0.0943
VF 7. Do you difficulty (with glasses) sewing, knitting, crocheting, or doing handicrafts?
	−0.18	−0.37	0.19	0.4183	−0.87	0.61	−1.47	<0.0001	−0.30	−0.55	0.25	0.3102
VF 8. Do you have difficulty (with glasses) filling out official forms or cheques?
	−0.43	−0.32	−0.11	0.6565	−0.30	−0.43	0.14	0.5236	−0.36	−0.40	0.04	0.8825
VF 9. Do you have difficulty (with glasses) playing cards, chess, or dominos?
	0.29	0.50	−0.21	0.3925	0.49	0.38	0.11	0.6255	0.43	0.52	−0.09	0.7464
VF 10. Do you have difficulty (with glasses) performing sport activities like tennis, golf, bowling, or boule?
	0.78	2.14	−1.36	<0.0001	2.13	1.14	0.99	0.0002	1.80	1.46	0.33	0.3152
VF 11. Do you have difficulty (with glasses) cooking?
	0.55	0.81	−0.26	0.2796	0.53	1.13	−0.59	0.0095	0.66	1.12	−0.46	0.1081
VF 12. Do you have difficulty (with glasses) watching TV?
	−0.20	−0.24	0.04	0.8654	−0.15	−0.38	0.24	0.2381	−0.23	−0.37	0.13	0.5915
VF 13. How much difficulty do you have driving during the daytime due to your vision problems?
	1.45	0.89	0.57	0.5279	1.27	0.99	0.29	0.7513	1.19	1.16	0.03	0.9755
VF 14. How much difficulty do you have driving during the night due to your vision problems?
	−1.22	−0.84	−0.38	0.5155	−1.39	−0.55	−0.84	0.1559	−1.12	−0.66	−0.46	−0.4688

*  DIF is bolded.

Cultural DIF.

To test for culture-specific differential item functioning between German subjects and people from other cultures, item measures were compared with previously published VF-14 item measures from patients with visual impairment in the United States (US). ⁴² Using the Bland and Altman method of assessing agreement, the mean difference between the two populations was 0.14 logits (95% confidence interval [CI], −0.34 to 0.63) (Fig. 3). The 95% limits of agreement were −1.54 to 1.83 logits (P < 0.0001). Figure 3 shows substantial agreement between the two sets of item measures, except for the item driving at night (r = 0.82 with that item included and r = 0.66 with it excluded). Despite this, the limits of agreement are small enough to conclude that the questionnaire can function similarly for different groups of people at the same level of ability regardless of cultural factors. 

Figure 3.

 

Bland and Altman plot; agreement in item measure calibration between German and US populations, including the item driving at night.

Criterion Validity.

The criterion validity of the Rasch-scaled VF-14 was tested by assessing its ability to discriminate between participants with normal (>6/12), mild (6/12 to 6/18), or moderate to severe (≥6/18) visual impairment. There was a statistical difference between the three groups (Kruskal Wallis; χ² = 13.19; P = 0.0014), with poorer visual acuity being associated with worse visual functioning (mean of 44.72, 51.05, and 60.65 logits for moderate/severe, mild, and no visual impairment, respectively). Collectively these results show that the VF-14 is a unidimensional, reliable, and valid scale to assess vision-specific functioning in this German sample. 

Discussion

We have used a modern psychometric validation technique—Rasch analysis—to examine the validity of a German-translated VF-14 questionnaire in a German sample population. Our study found that after collapsing two response categories for items 13 and 14, the VF-14 scale satisfied fundamental criteria to achieve fit to the Rasch model, namely ordered thresholds, the ability to distinguish between different strata of participant ability, absence of misfitting items, no evidence of unidimensionality, and no significant differential item functioning for key sociodemographic covariates. In addition, the results from this study demonstrate that the VF-14 is able to discriminate between participants with different levels of visual impairment and across different cultural groups. Therefore, the VF-14 is a valid, reliable, and unidimensional questionnaire for use in a German population. 

By using Rasch analysis, we were able to conduct a detailed evaluation of the German VF-14 scale. Threshold ordering was investigated using the rating scale model and evidence suggested that the response scale of the German VF-14 was initially suboptimal. Analyses indicated that the four-category response option for items 13 and 14 was problematic for participants who had difficulty discriminating between some difficulty and a little difficulty. After collapsing these two categories, further analyses showed that a three-category rating scale was effective for these two items. Reduction to a three-category response scale has been found to be optimal in other studies investigating response category utilization. ^40,43  

The German VF-14 displayed excellent validity and reliability. The person separation indices indicated that it could distinguish between several levels of person ability and evidence supported the ability of the scale to assess a single underlying trait, namely visual function. Moreover, all 14 items displayed fit to the Rasch model, indicating that items were neither redundant nor tapping into other indeterminate traits. Other studies, in contrast, have found it necessary to delete items to fit the Rasch model. ^43,44  

Our results also provide evidence of the considerable construct validity of the German VF-14, with very minimal differential item functioning for sex and age and no differential item functioning for education and culture. It is feasible that the measurement of a patient-reported outcome measure such as the VF-14 could vary across different study populations due to social, cultural, and clinical factors. Using the Bland Altman method of agreement, our findings support existing studies that have found the VF-14 to remain valid across different populations ^21,22,45,46 as well as with English-speaking participants. ^14,16,17,20 For example, Maharajah and associates ⁴⁶ found that the Bahasa Malaysian version of the modified VF-14 questionnaire was a valid tool to quantify visual functioning among glaucoma patients in a rural area of Malaysia with high rates of illiteracy. Similarly, Gresset and colleagues ²¹ found that the French VF-14 was a reliable and valid tool in measuring the visual function in patients with cataracts and corneal opacities attending tertiary eye clinics. Moreover, using Rasch analysis, Lamoureux and colleagues ³¹ found the modified VF-14 (VF-11) to be a unidimensional, reliable, and valid scale to assess visual-specific functioning in a population of Asian Malays in Singapore. Future research could aim to validate the VF-14 in other culturally and linguistically diverse areas such as China, as China comprises 17% of the world's population and Chinese is one of the three most commonly spoken languages is in the world. 

The criterion validity of the German VF-14 was illustrated by its ability to significantly distinguish between participants with none, mild, and moderate to severe visual impairment. Moreover, with patients with retinal-related eye conditions comprising just over 40% of our sample, our findings support previous studies that show the VF-14 to be valid for use in a general low vision sample or across a range of eye conditions other than cataracts. ^20,47 Similarly, although the VF-14 items focus primarily on functional activities that rely on central vision, previous work by our group has also found it to provide meaningful assessment of vision-related functioning in glaucoma participants. ^48,49 Thus, overall, our results suggest that the German VF-14 is applicable across all ocular conditions. 

In contrast, the person-item map of the German VF-14 shows limited targeting of the scale, with many participants possessing greater ability than required by the most difficult item of the scale (driving at night). This is most likely because over a third of the participants in our sample had no visual impairment although they were diagnosed with an eye condition. It is likely that the targeting would greatly improve with a more visually-impaired sample. This suggests that the VF-14 may not be fully equipped to assess the impact of milder levels of visual impairment on patients' visual functioning. This could be a potential issue for researchers or clinicians who wish to examine visual functioning in patients with early stage glaucoma or diabetic retinopathy where patients may suffer few visual symptoms. ¹⁷ Future research could therefore investigate whether the inclusion of items of greater difficulty to better cater to patients with mild visual impairment may improve the targeting of the scale. 

The item map also reveals that several items represent the same level of difficulty among the ability continuum which is suggestive of redundancy. However, given that the VF-14 is relatively short and easy to administer, removing items is not likely to noticeably reduce respondent burden. Furthermore, individual items provide specific information about which aspects of visual functioning are affected which would be lost if the number of items were reduced. For example, items pertaining to reading could guide interventions to improve this aspect of visual functioning while the item about noticing steps or curbs could aid orientation and mobility rehabilitation. Similarly, patient-specific information about the effectiveness of low-vision rehabilitation or interventions may be lost if items are deleted. 

Importantly, the person-item map also demonstrates that all items have inherent levels of difficulty and require disparate levels of participant ability, which is contrary to the assumption of the Likert system. The VF-14 items identified as most difficult by this German outpatient population are similar to those found in other studies. ^{44,50 –52} For instance, as found in our study, driving at night was also the most difficult reported item in a rural, vision-impaired population in Malaysia. ⁴⁶ Items pertaining to reading small print or doing near tasks were also found to be difficult in our study, which is again similar to other cultural groups. For example, in a study using the VF-14 in an East Timorese population, ⁵³ participants rated reading a newspaper, bible, or book as being the most difficult. Similarly, in a Western population, reading labels and instructions on medicine bottles/packets was found to be the most challenging item. ⁵⁴ This suggests that the difficulties faced by people with visual impairment are consistent across populations, irrespective of their cultural differences. This is in congruence with our cultural-specific differential item functioning findings. 

The limitations of our study include the sex imbalance in our sample (i.e., there were more males than females), and the large number of participants with no or only mild visual impairment, both of which could have impacted on our findings. 

In summary, our study found that the German version of the VF-14 is a valid and reliable scale to assess the vision-related functional ability of German people with visual impairment. This was confirmed through the application of Rasch analysis. These findings support the use of the German VF-14 in tertiary eye clinics in Germany to capture the impact of visual impairment on visual function from the patient's perspective and to inform low vision rehabilitation and interventions. Future research should aim to validate the VF-14 using modern psychometric methods in other culturally and linguistically diverse populations. 

Footnotes

 Disclosure: P.P.-C. Chiang, None; E. Fenwick, None; M. Marella, None; R. Finger, None; E. Lamoureux, None