Psychometric Properties of the Persian Translation of the 12-Item Utah Photophobia Symptom Impact Scale Questionnaire

ABSTRACT

To translate the the Utah Photophobia Symptom Impact Scale-12 questionnaire into Persian and assess the psychometric aspects to check its validity and reliability based on the Rasch modelling method. Translation and cultural adjustment of the English language UPSIS-12 questionnaire to Persian was undertaken. A total of 61 patients with complaints of photophobia participated in evaluating validity and reliability aspects. All the participants were asked to complete the Persian translation of the UPSIS-12 questionnaire. Rasch analyses of the survey items were conducted using WINSTEPS. All items fit the Rasch model. Point-measure correlation values varied from 0.41 to 0.77, providing a preliminary indication of adequate construct validity. All factor loadings were found more than 0.4. All items obtained infit and outfit mean square (MnSq) values of < 2.0. All participants except 5 had normal outfit values. Patients’ abilities relative to the items’ difficulty were analysed. Item difficulty was estimated and item characteristic curves were included. Sufficient unidimensionality, hierarchical order, and equal interval scoring were obtained. In conclusion, the Persian UPSIS-10 questionnaire has excellent psychometric properties and it will be valuable in both clinical practice and research. It will help Persian practitioners to assess their patients’ photophobia.

Introduction

Photophobia is a sensory disorder that is triggered by light. The word photophobia consists of two words photo and phobia which together mean fear of light.¹ Photophobia is an abnormally increased sensitivity to light. Photophobic patients may report photosensitivity in situations where most people do not. Some patients will find out that they are specifically sensitive to artificial indoor lighting. Another common cause of discomfort for photosensitive patients is computer use.²

Photophobia is present in many neurological and ophthalmic conditions including migraine, meningitis, intracranial tumours, traumatic brain injury, subarachnoid haemorrhage, uveitis, keratitis, retinitis pigmentosa, cone-rod dystrophy, corneal damage, blepharospasm, and dry eye.^3,4 In normal people, photophobia is a protective mechanism to avoid intense light damage, but in pathological conditions, photophobia increases abnormally.⁵ The main proposed aetiologies for photophobia are orbital and visual pathway pathology, drug-induced photophobia, neurological disorders, and psychiatric disorders.⁶

Quantitative measures of visual photosensitivity are important especially when visual photosensitivity is a prominent feature. These measures are important for characterizing the stage or severity of the disease, evaluating the efficacy of a treatment, and monitoring disease progression.⁵

The Utah Photophobia Symptom Impact Scale-17 (UPSIS-17) is a 17-item photophobia questionnaire that was developed at the University of Utah. This questionnaire utilizes a 0 to 5 Likert scale and yes/no response items with a total score range from 0 to 80. Then the Rasch analysis was used to determine whether the UPSIS-17 could be shortened without significant effect and finally, the questions 6, 8, 10, 11, and 13 were eliminated from UPSIS-17 to create UPSIS-12. The UPSIS-12 is as useful as the UPSIS-17 and is faster and easier for patients to complete.³

UPSIS-12 questionnaire is one of the few valid photophobia evaluation questionnaires and has been reviewed in some studies.^3,7 In this study, we intended to prepare the Persian translation of the UPSIS-12 questionnaire and to evaluate its validity and reliability based on Rasch model analysis.

This is the first study to translate and assess the psychometric aspects of this questionnaire in the Persian population. The valid Persian version can be useful in both research and clinical settings.

Materials and methods

This prospective cross-sectional research was reviewed by an independent ethical review board and conforms with the principles and applicable guidelines for the protection of human subjects in biomedical research (ethics approval code: 4010522). The translation and adjustment of the English language 12-item Utah Photophobia Symptom Impact Scale (UPSIS-12) was chosen to create a Persian culturally specific measurement of photophobia.

Respect the originality and cultural adaptation were noticed during translation. The translation was performed according to the approved international rules in translation and cultural adaptation of the second annual ISPOR congress.⁸ First, forward translation (from English to Persian) was done by two optometrists with Persian mother tongue and knowledge of English. Then its cultural adaptation was checked by an expert committee of methodologists, health professionals, and language professionals. So the first Persian version of the questionnaire was prepared. This version was translated into English by two English native persons with advanced level knowledge of the Persian language. These two translations were compared with the original one and necessary corrections were applied. To calculate the content validity ratio (CVR) and content validity index (CVI) (in terms of necessity, clarity, simplicity, and relevance), the Persian version of the questionnaire was reviewed by a panel of 10 experts (including optometrists, ophthalmologists, and statistician). In this stage, we asked them to determine whether the skill measured with each item is essential, useful but not essential, or not necessary. Then CVR was calculated by the formula below, in which, n_e is the number of panellists indicating ‘essential’ and N is the total number of panellists.

$$\mathit{CVR}=\frac{\mathrm{ne}-\mathrm{N}/2}{\mathrm{N}/2}$$

For calculating CVI, the experts were asked to rate the clarity and relevancy to the construct of each item on a 4-point ordinal scale (1 = not relevant, 2 = somewhat relevant, 3 = quite relevant, 4 = highly relevant). To obtain CVI, the number of those judging the item as relevant or clear (rating 3 or 4) was divided by the number of content experts. If the CVR and CVI of each item were less than 0.62 and 0.7 respectively, the item was removed from the questionnaire, and if the CVI was between 0.7 and 0.79, the item was modified. After essential modifications, the final Persian version of the questionnaire was presented.

Psychometric aspects of the Persian translation of the UPSIS-1 were checked by the Rasch model. In the next step, 61 Persian native participants who complained about photophobia in outdoor environments were enrolled using a simple sampling method. Patients with any eye disease, history of eye surgery, underlying systemic diseases, using medications that cause photophobia and blurred vision (mydriatics and cycloplegics, antihistamines, quinine, tetracycline, doxycycline, etc.), and disability to read, were excluded. The participants were asked to fill out the Persian translation of the photophobia questionnaire.

Statistical analysis

Psychometric assessment of a new instrument should present some information about the items and also participants to whom it was administered such as item measures, standard errors and fit statistics, person measures, and indicators of overall scale function such as unidimensionality and precision.⁹ Rasch modelling is a statistical modelling approach used to assess survey items that are meant to measure a latent construct.¹⁰ Our Rasch analysis was performed with WINSTEPS software (version 3.8.1). In the Rasch model, one parameter is defined for the person (ability), and one for each item (difficulty). As the first step, construct validity was checked. Next, fit statistics were investigated. The outlier-sensitive fit statistic (outfit < 2.0), and the inlier-pattern-sensitive fit statistic (infit < 2.0) were used.¹¹ Item separation and item reliability were used to assess the item hierarchy. The internal consistency (Cronbach’s alpha) was also calculated to assess the intercorrelation of the questionnaire items, ranging from 0 to 1, with a value of > 0.7 considered adequate.

Results

The CVR and CVI calculated for each item as presented in Table 1. According to CVR and CVI calculations, items 4, 10, 12(a), and 12(b) were removed from the questionnaire and items 3 and 6 were modified. Figure 1 shows these modifications.

Table 1.

CVR and CVI for each item.

Item	CVR	CVI
1	1	0.8
2	0.7	0.8
3	0.8	0.7
4	0.8	0.5
5	0.7	0.9
6	0.7	0.7
7	0.8	0.9
8	0.7	0.9
9	1	1
10	−0.4	0.6
11	0.8	0.9
11A	0.8	0.8
12A	0.2	0.8
12B	0.6	1
12C	0.7	1
12D	0.8	0.8

CVR: Content Validity Ratio, CVI: Content Validity Index.

Figure 1.

Modification of the original English questionnaire during Persian translation and cultural adaptation. (With permission of Melissa M Cortez et al).³

The Persian photophobia questionnaire was completed by 61 participants with a subjective report of photophobia with a mean age of 25.80 ± 10.18 years (ranging from 10 to 56 years). The demographic characteristics of the patients have been presented in Table 2.

Table 2.

Socio-demographic characteristics of the subjects.

Characteristic		Number (n, %)
Gender	Male	37 (60.7%)
	Female	24 (39.3%)
Education	High school	23 (37.7%)
	Graduated	29 (47.5%)
	Post graduated	9 (14.8%)
Job	Unemployed	3 (4.9%)
	Employed	19 (31.1%)
	Student	39 (63.9%)
Location	Rural	9 (14.8%)
	Urban	52 (85.2%)

Based on Rasch model measurements, point-measure correlation values varied from 0.41 to 0.77 and all factor loadings were found to be more than 0.4, providing adequate construct validity. Item fit statistics showed that all items obtained infit and outfit MnSq (mean square) values of < 2.0. Rasch fit statistics for each item have been presented in Table 3.

Table 3.

Rasch fit statistics for each item.

Survey item	Factor loading	Infit MnSq	Infit ZSTD	Outfit MnSq	Outfit ZSTD	Point-measure correlations
1. Generally, how do you assess the intensity of your light sensitivity?	0.736	0.60	−2.52	0.56	−2.87	0.75
2. How unpleasant are you with the light when you do not have a headache?	0.662	1.09	0.58	1.04	0.26	0.55
3. How uncomfortable are you with the severe light when you have a headache?	0.732	1.98	4.94	1.89	1.71	0.44
4. How often does intense light trigger a headache?	0.577	0.77	−1.41	0.86	−0.78	0.63
5. How tough is it for you to look at a computer screen for a while?	0.660	0.80	−1.20	0.87	−0.71	0.67
6. To what extent does the light sensitivity affect your capability to watch TV?	0.713	0.92	−0.41	0.93	−0.34	0.69
7. To what extent does light sensitivity affect your capability to work indoors or outdoors?	0.807	1.17	0.98	1.10	0.59	0.77
8. How much does light sensitivity affect your driving capability?	0.691	0.92	−0.38	0.860	−0.74	0.64
9. Do you wear sunglasses to moderate headaches?	0.766	0.29	−4.29	0.38	−3.35	0.70
10. Does the sensitivity to light effect your driving capability?	0.464	0.61	−2.04	0.84	−0.68	0.41

According to the fit statistics, from 61 participants, five had infit values > 2.0 (4.48, 2.47, 2.45, 2.26, and 2.19 infit values), while the infit values of the other participants were in the acceptable range. These participants were ignored and excluded from all subsequent analyses.

Based on the Rasch model, person reliability, person separation, item separation, and item reliability were more than 0.8, 0.2, 0.7, and 0.9, respectively, and all were in the acceptable range. Table 4 shows the aforementioned measurements. The category probability curves, where the probabilities (y-axis) are plotted against differences between person and item measures (x-axis) (Figure 2), show the relation between the probability of a given category as a function of person’s location

Table 4.

Item and patient separation and reliability.

Parameter	Infit		Outfit		Reliability	Separation
	MNSQ	ZSTD	MNSQ	ZSTD
Item	0.95	−0.6	0.97	−0.4	0.98	7.11
Person	0.97	−0.2	0.97	−0.2	0.82	2.16

Figure 2.

Category probability curves for the first question of the Persian version of the photophobia questionnaire.

To assess the item difficulty estimates of the items, an item person map was generated. Figure 3 demonstrates the person item map. Very few items were at the higher ability and lower ability levels. The items are listed on the right side of the vertical axis in a hierarchical order, from the most difficult (at the top of the map) to those which were least difficult (at the bottom of the map). Patient ability estimates have been presented on the left, from the highest ability level at the top to the lowest ability level at the bottom. According to this map, ceiling and floor effects are 19.6% and 12.5% respectively.

Figure 3.

Person item-map for 10 survey items. The items are listed on the right in a hierarchical order, from the most difficult (at the top of the map) to those that were least difficult (at the bottom of the map). Patient ability estimates have been presented on the left, from the highest ability level at the top to the lowest ability level at the bottom. M = mean, S = one standard deviation, T = two standard deviations, X = 1 patient.

Figure 4 shows patients’ ability relative to item difficulty.

Figure 4.

Person-item (wright) map of the instrument. In the upper map, the participants are shown on the left of the vertical axis, with less able participants located at the bottom. In the lower map, the number of difficult items on the left of the vertical axis, and the horizontal axis shows item difficulty.

Discussion

This study aimed to provide a Persian translation of the UPSIS-12 questionnaire and to assess the psychometric aspects to check its validity and reliability based on the Rasch modelling method.

UPSIS-12 was developed to improve the evaluation of the impact of photophobia on daily activities consisting of computer and television use, in or out-of-house activities, driving, and its effect on headache onset and severity. During the translation and cultural adjustment process of the UPSIS-12, CVR, and CVI calculations necessitated us to remove items 4, 10, and 12(b) and modify items 3 and 6. If the CVR and CVI of each item were less than 0.62 and 0.7 respectively, the item was removed from the questionnaire and if the CVI was between 0.7 and 0.79, the item was modified.^12,13 Finally, the Persian version contains 10 items. At first, construct validity was assessed. Construct validity describes whether items actually measure the underlying construct that they are intended to measure.¹⁴ Point-measure correlations are Pearson correlations based on the Rasch model which correlates individual item response values and the corresponding person ability estimates. This correlation shows whether the responses to each item are equal to the ability estimates of the persons. Point measure correlations range from −1 to 1. Problematic items can be identified by Point measure correlations. These items do not appear to map onto the test’s latent construct.¹¹ Point measure correlation values measured for our 10 survey items were all positive and varied from 0.41 to 0.77, providing a preliminary indication of adequate construct validity. This is an indicator of unidimensionality and shows that item scores were correlated with total scores.

The fit statistics of person ability estimates were also examined. The outfit is a measure of unexpected outlying observations, and outfit measures of a person indicate whether a series of responses are inconsistent with the Rasch model. Outfit and infit MnSq values more than 2.0 are concerning, and indicate when persons should be pruned from the data set.¹¹ Five out of 61 participants had outfit values > 2.0. This confirmed that according to the model and based on item difficulty estimates and person ability estimates these five participants had unusual and inconsistent responses. These participants were pruned from the data set and excluded from all subsequent analyses.

Item separation and item reliability were also evaluated. These factors examine the item hierarchy, measuring the ability to stratify persons and generate reproducibility of relative item location. Item separation ≥ 3 and item reliability ≥ 0.9 were considered acceptable. Person separation and person reliability verify that the instrument was able to classify a person ability (e.g. distinguish between high and low photophobia). Separation ≥ 2 and reliability ≥ 0.8 were considered acceptable.¹⁵ Our study showed acceptable ranges of reliability and separation measurement for both items and patients which means.

Item difficulty values were measured for all items. For each item, the response option ‘zero’ is likely to be endorsed by people with higher abilities (Figure 2). This response option can be considered to be the most difficult category because only the most able participants (i.e., the participants with less photophobia) can subscribe to this category. In a well-functioning rating scale, the thresholds are neither too close nor too far apart. Generally, accepted values for a threshold step are between 1.4 and 5.0 logits.¹¹

The item person map displays the model results along a scale of item difficulty, with both items and persons in the sample. When an item bank is aimed for use with a wide range of individuals, the items should be extended across a range of ability levels. When there are no nearby items to a given ability estimate, the item bank has less precision.¹⁶ This map showed better instrument targeting for the Persian UPSIS-10 (19.6%) compared to the UPSIS-17 (23.2%), UPSIS-12 (22.1%), and KUMC-8 (49.5%) which were reported in Cortez et.al study.³ These results confirmed that the Persian UPSIS can measure a wider range of factors. Furthermore, shortening the Persian version to 10 items did not have a negative effect on item targeting, ceiling, or floor effects. We believe that the Persian UPSIS-10 is as useful as the UPSIS-12 with the benefit of being faster and easier for patients to complete. The floor effect was measured near the 10% range which is considered acceptable.

In conclusion, the Persian UPSIS-10 questionnaire has excellent psychometric properties and it will be valuable for both clinical practice and research purposes to assess patients with symptoms of photosensitivity.

Funding Statement

This work was supported by Mashhad University of Medical Sciences [Grant code: 4010522]. No financial disclosures.

Disclosure statement

The study protocol adhered to the principles of the Declaration of Helsinki and was approved by the ethics committee of Mashhad University of Medical Sciences. All patients were informed about the study and provided signed informed consent to participate. The authors report there are no competing interests to declare.