Measuring utility values of eye conditions among children in India using the EQ-5D-Y instrument

Abstract

Background

Vision impairment and blindness are significant global public health challenges, particularly in low- and middle-income countries, where access to eye care services remains limited. India has significantly reduced the prevalence of Blindness and Vision Impairment (VI) over the last two decades. This was achieved with the help of greater investments towards blindness control programs. The use of utility values helps in conducting economic evaluations of various eye health programs and empirically justify investing in these programs. This study aimed to estimate utility values for various childhood eye conditions in central India using the EuroQol-Five-Dimension-Youth (EQ-5D-Y) instrument.

Methods

This is a before and after study with data collected at two time points for few participants and at only one time point for others. This study was undertaken at Shri Sadguru Netra Chikitsalaya (SNC) and included children representing central and north India. Participants were randomly sampled in the hospital. After comprehensive eye examination, participants completed the EuroQol-Five-Dimension-Youth (EQ-5D-Y) questionnaire along with EuroQol Visual Analogue Scale (EQ VAS) measurement to elicit their health state for their condition which was repeated after six months post-intervention to measure the change in utility value. We have used Indonesian value set to analyze the preference scores of each dimension of EQ-5D-Y.

Results

Utility values of 16 eye conditions were estimated at baseline and seven conditions were followed up for post-intervention utility value estimation. There is a statistically significant improvement in the utility values post-intervention amongst six conditions. Blindness and Pediatric cataract had the greatest change (0.23 and 0.2 respectively) in utility value whereas mild Vision Impairment (VI) showed the least change (0.02) in the utility value post-intervention. Blindness had the lowest baseline (0.62) and post-intervention (0.85) utility value.

Conclusion

The utility values estimated in this study showed that generic measures such as EQ-5D-Y may be used to elicit health states for various eye conditions amongst children. These estimates are helpful in undertaking cost-utility analyses of eye health programs and interventions aimed at these eye conditions.

Introduction

Globally, significant strides have been made in improving public health, including eye health, over the past few decades [1,2,3,4]. However, the burden of blindness and vision impairment remains a major public health issue, especially in low- and middle-income countries such as India, where access to eye care services is often limited [5, 6]. This burden is not only a health issue but also an economic one, as visual impairment can lead to loss of productivity and increased healthcare costs, which is especially true in case of children with vision impairment and blindness [5, 6]. In India, the incidence of childhood blindness is 0.4-1.0 in every 1,000 children and about 35% of childhood blidness is preventable or treatable [3, 7]. The Sustainable Development Goals (SDGs), particularly SDG 3, emphasize the importance of ensuring healthy lives and promoting well-being for all, which includes the prevention and treatment of vision impairment [8, 9].

In recent years, there has been a growing emphasis on integrating eye health into broader public health systems [8]. This horizontal integration aims to ensure that eye care is not treated in isolation but is part of comprehensive health service delivery [8]. Early intervention in eye health, particularly for children, is critical as it can prevent long-term vision impairment and its associated economic consequences [5,6,7]. Despite these advancements, challenges such as affordability and accessibility to eye care persist, hindering progress toward achieving the SDGs. Over the past two decades, substantial investments by governments, non-governmental organizations (NGOs), and other stakeholders have contributed to reducing the prevalence of blindness [10]. However, these stakeholders often face tough decisions regarding resource allocation, requiring evidence-based approaches to determine the effectiveness and priority of various health interventions [11, 12].

To evaluate the effectiveness of eye care programs, measures such as improvements in visual acuity are commonly used [13, 14]. However, these measures may not fully capture the individual’s quality of life, particularly when co-morbidities are present [13, 14]. This highlights the need for more comprehensive and standardized measures like the Quality Adjusted Life Year (QALY), which combines both health-related quality of life and length of life into a single metric [11, 12, 15]. QALYs allow for comparisons across different health conditions and interventions, aiding in more informed decision-making [11, 12, 15]. In this context, our study aims to estimate the utility values of various eye conditions in children, both before and after interventions, in central India. By using the EuroQol-Five-Dimension-Youth (EQ-5D-Y) instrument [16], we seek to provide valuable data that can inform the cost-utility analysis of eye health programs and interventions, ultimately supporting better resource allocation and improved outcomes for children with vision impairment.

Methods

Study location

This study was undertaken in a remote and rural part of Madhya Pradesh and Uttar Pradesh, Central and North India to represent the socio-economic status of the country as a whole. Children visiting the paediatric ophthalmology clinic of “Shri Sadguru Netra Chikitsalaya (SNC)”, Chitrakoot, Madhya Pradesh, India. Since it is the only tertiary care hospital in this area, people come from various parts of India including the states of Madhya Pradesh, Uttar Pradesh, Bihar, and Jharkand.

Study design

This study compares the change in utility values before and after intervention.

Study population

Children aged between 4 and 16 years with or without an ocular condition or co-morbidities, and their caregivers were included in the study. Since our primary focus was on eye health conditions, we did not measure co-morbidities at baseline but focused on the eye conditions that each child has presented with. Children were randomly selected using randomization of their registered Medical Record Number (MRN).

Sample size

We have used the prevalence estimates of childhood ocular morbidity for sample size calculation: The prevalence of ocular morbidity in children, and the prevalence of blindness and vision impairment under the age of 49 years were 6.54% and 23.8% respectively [10, 17]. We have used 23.8% as the prevalence measure to calculate the sample size as this is the only nationally representative estimate we could find in our literature review. We have used the following formula to calculate the sample size for the study:

$$\begin{aligned} \text {n}=\frac{\text {4pq}}{\text {l}^2} \end{aligned}$$

(1)

Where:

• ‘p’ is the prevalence of blindness and vision impairment in India

• ‘q’ is the proportion of people with blindness and vision impairment calculated as $1-p$

• ‘L’ is the allowable tolerable error set at 5% to have a confidence level of 95%.

The estimated sample size was n = 290. Considering a 10% non-response rate, the final sample size was adjusted to 320.

Baseline data collection

Participants underwent comprehensive eye examination by an ophthalmologist and or an optometrist. A demographic form was filled which documented the demographic details such as age, sex, socio-economic status, village name, refractive status, direct and indirect costs of treatment(s). We have used Euro-Qol-Five-Dimension Youth (EQ5D-Y) three level questionnaire to measure the child’s health state as it is relevant to be used amongst this particular age group [16]. The EQ-5D-Y tool is available in English and Hindi and is validated [16]. The EQ-5D-Y tool measures “an individual’s health state across five dimensions namely Mobility, Self-care, Usual activities, Pain/ Discomfort, Anxiety/ Depression” [18]. The tool also has a “Visual Analogue Scale (EQ VAS)” which measures the current overall health state of the individual on an interval scale of 0 and 100; “where ‘0’ means a health state equal to being dead and ‘100’ means a health state equal to being in full health.” Proxy interviews of the caregiver who is usually the mother or the father of the child were undertaken for children aged 4 to 8 years and where the child could not comprehend the questions. Hindi tool was used where the participant was not able to comprehend English questions. Participants had various diagnosed conditions at the time of data collection. Untreated children and their respective health states were considered as baseline data.

Post-intervention data collection

The participants were followed up after six months post-intervention. The interventions included various treatments such as spectacles to treat refractive errors, pseudophakia to treat cataracts, and patching and orthoptic exercises to treat amblyopia and strabismus respectively. The EQ-5D-Y tool along with the EQ VAS scores was re-administered on the participants after undertaking a comprehensive eye examination and eliciting the refractive status of the participants. The tool was self or proxy administered. Only those participants who have received an intervention were included in the post-intervention assessment of the health state preference.

Calculating utility based on EQ-5D-Y

The EQ-5D-Y questionnaire measures health-related quality of life across five dimensions: “mobility, self-care, usual activities, pain/discomfort, and anxiety/depression”. Each dimension has three levels of responses. The responses are converted into a summary index (utility value) using a predetermined scoring algorithm which is closely related to the population of study; we have used Indonesian value set to convert the scores to the index value [11, 19].

EQ-5D-Y proxy version

For children aged 4-8 years, we utilized the proxy version of the EQ-5D-Y questionnaire, completed by their caregivers. This approach was necessary as younger children might struggle to comprehend and accurately respond to the questions. This ensured reliable and valid data collection for this age group.

The data collection instrument, patient recruitment strategy, and the ocular conditions and the child’s perceived health states were tested as part of a pilot study with 57 participants. We made few amendments after the pilot to improve the survey process; We included proxy responses for children aged 4-8 years and those unable to comprehend the questionnaire. Additionally, we categorized children into broader visual impairment categories rather than specific types of refractive errors to simplify data analysis and interpretation. The subsequent full study was undertaken on 308 participants.

Data analysis

Individual conditions were categorized according to the diagnosed condition as per the ‘World Health Organization (WHO) International Classification of Diseases 11th edition (ICD-11)’ [20]. Participants were categorized into blindness and vision impairment using best corrected visual acuity in better seeing eye definitions. “best corrected visual acuity worse than 3/60 in the better seeing eye is defined as blindness”, “best corrected Visual Acuity in better seeing eye worse than 6/60 (0.1 LogMAR) and equal to or better than 3/60 (0.05 LogMAR) is defined as severe vision impairment”, “best corrected Visual Acuity in better seeing eye worse than 6/18 (0.3 LogMAR) and equal to or better than 6/60 (0.1 LogMAR) is defined as moderate vision impairment”, “best corrected Visual Acuity in better seeing eye equal to or better than 6/18 (0.3 LogMAR) is defined as mild vision impairment.” Normality was tested using Kolmogorov-Smirnov test (KS-test).

We have used descriptive statistics to summarize utility and demographic data. Wicoxon Signed Rank Test was used to compare mean or median utility scores and EQ VAS scores at baseline and post-intervention. Visual acuity measured using other notations were converted to LogMAR visual acuity values for data analyses using “eye” and “eyedata” packages on R software [21,22,23]. Conditions were classified according to the refractive status, BCVA, and diagnosed pathology. Any pathological condition may have refractive error associated with it, and may have been classified under either category. All patients were also classified using visual impairment classification. Individual scores for each dimension were analysed using Indonesian value set for EQ-5D-3L-Y tool [19]. EQ VAS scores were also analysed and compared with utility values derived from dimensions using the Indonesian value set. To evaluate the variability of Best Corrected Visual Acuity (BCVA) between eyes, we used the Generalized Estimating Equation (GEE) model [24]. The GEE model accounts for the correlation between measurements taken from the same subject (both eyes).

Bootstrapping with 1,000 to 10,000 replicates was used to estimate the confidence intervals around the median utility value, EQ VAS score and median BCVA according to the methods in published literature [15]. We addressed missing data by omitting observations that lacked responses to any or all dimensions of the EQ-5D-Y tool. This approach ensures the integrity and accuracy of our final analysis, though it may reduce the sample size.

We adhered to COVID-19 protocols during the entire process of the study data collection.

Results

We had randomly selected 470 participants out of which a total of 308 children including their caregivers consented to participate in the study. Data from two responses were excluded from the final analysis due to data inconsistency. Most of the children were male. The demographic details of the participants are mentioned the Table 1

Table 1 Demographic details

We have identified a total of 46 eye conditions at baseline. 14 conditions had sufficient sample size to be followed up for post-intervention estimation of EQ VAS scores and utility values. The baseline EQ VAS scores are mentioned in Table 2 and the comparison of baseline and post-intervention EQ VAS scores is mentioned in Table 3.

Table 2 Baseline EQ Visual Analogue Scale (EQ VAS) Score

Table 3 Baseline and Post-Intervention EQ VAS Scores

The median baseline EQ VAS score before the intervention is 0.8 (IQR 0.65-0.9). The median baseline best corrected LogMAR visual acuity in better eye is 0.2 (IQR 0.0-0.5). There is a negative correlation between best corrected visual acuity (BCVA) and EQ VAS scores (-0.37).

Using the Indonesian value set, the estimated range of baseline utility values for all eye health conditions was-0.086 to 1.00; and the estimated range of post-intervention utility value for all eye health conditions was 0.08 to 1.

Using Generalized Estimating Equations, we have found that there is no statistically significant difference in Utility values for a person with a specific condition between both eyes or only one eye ($\beta$ = 0.034, p = 0.65), after controlling for age and gender. The baseline and post-intervention utility values for different conditions and vision impairment are mentioned in Table 4.

Table 4 Baseline and post-intervention utility values for various eye conditions

The baseline and post-intervention utility values for different conditions and vision impairment are mentioned in Table 4

Discussion

The utility values estimated in this study pave way to economic evaluations of various interventions aimed at addressing vision impairment, pediatric cataract, and other conditions. Children were classified into various conditions according to the pathological diagnoses of their eyes and also according to the BCVA into one of the four visual impairment categories; this may mean that the same children may have been classified into more than two conditions. There were 46 unique diagnoses identified in this study. We have included 13 conditions for further analyses as other conditions had very few observations. Children with seven conditions were followed-up post-intervention. The various interventions included cataract surgery (Small Incision Cataract surgery (SICS) or Phaco-emulsification with foldable IOL)), Spectacles (eyeglasses) for refractive errors, Vision Therapy for Amblyopia, and Orthoptic exercises for Strabismus. We have used “best corrected visual acuity in better seeing eye” as the criterion to categorize patients into one of the visual impairment categories. Using GEE model, we have found that there is no significant difference in the utility value when compared with the visual acuity between both eyes and BCVA of single eye after controlling for age and gender. Our estimated utility value for blindness is 0.62 (IQR 0.50-0.71) which is similar to other studies [25, 26], it changed to 0.85 (IQR 0.69-0.98) post-intervention as mentioned in Table 4; this may be due to higher number of children with bilateral congenital cataract being operated only in one eye at the time of data collection for this study; however, we have not tested this hypothesis. As mentioned in Table 4, we have observed that children with amblyopia showed significant improvement in their utility values post-intervention from 0.85 to 0.98 indicating that amblyopia is one of the conditions which affects their quality of life negatively if untreated. There are various Vision Related Quality of Life (VRQoL) tools [27, 28] available but we have not used them as the utility value would have been specific only to eye health and not general health, making it impossible to compare and evaluate with other general health conditions for scarce resource allocation. This study shows that the utility value has increased amongst all conditions as the visual acuity improves, with an average utility gain of 10% in all health conditions studied. These utility values for various conditions are particularly useful in bridging the gap in knowledge and facilitate economic evaluations of child eye health interventions and outcomes using cost-utility analyses and cost-effectiveness analyses and make informed decisions related to scarce resource allocations [11, 29, 30].Further research is needed to understand the long-term impact using population-based samples of conditions and their interventions on the utility values of children with congenital conditions such as congenital glaucoma, cataract, RoP, and others as these conditions may have long-term negative impact on the HRQoL if untreated early in life. We have observed that the utility value changes with geographic location of the individual, socio-economic conditions, and education of the child’s family; however, we could not statistically prove these relationships. Hence future research should focus on the variability of HRQoL and utility values in relation to changes in socio-economic conditions, and education, and has to be region specific as India has a federal structure and health is a state subject and hence utility values for the same condition may not necessarily apply in a different population setting.

Close to 50% of the study population was lost to follow-up due to remote location of the hospital, transportation barriers including cost, poor compliance to treatment, and people just choosing to come to the hospital on a later date due to ongoing COVID-19 pandemic situation and its associated mobility restrictions prevalent in their respective locations. This study is conducted in a hospital setting and although we have considered all the potential confounders in the data including region wise variation in the utility values, and the socioeconomic conditions, questions may arise about the generalizability of the data to a larger community context and hence future focus should be made on studying if there will be any variation in the results between hospital or provider based data and community based data and results. This study estimated utility values using generic preference-based models and it paves the way for larger population-based studies to be carried out as our utility value estimates were calculated using Indonesian value set of EQ-5D-Y and comparing EQ VAS scores with that value set as there is no India specific value set for this tool. The utility values estimated using a different country’s value set may have a significant difference in their estimation due to the various reasons such as valuation methodology, varied health system support between the countries, differences in the way societies perceive their health state and quality of life, and cultural differences [31]; this is one of the limitations of this study and we recommend updating the utility values once the India specific value sets are available for EQ-5D-Y instrument.

Conclusion

This study contributes valuable scientific evidence related to utility values specific to various ocular conditions in a hospital setting, allowing researchers, programme staff, NGOs, government, and various stakeholders; these values are used to understand the gain in utilities as well as conduct health economic evaluations on interventions and programs to address these eye conditions. It also provides evidence that early interventions to address ocular conditions such as amblyopia and severe VI leads to improved quality of life of the child; and that these interventions have to be prioritised over others. Utility estimates combined with cost related data pave way to evidence based health economics and outcomes research in eye care, and it should be routinely practised to advocate for resource prioritization and allocation.