Table 1 Definitions and estimation of cost categories in Trinidad & Tobago in 2014.

 a) Medical Direct costs

Costs included the resources used to treat an eye disease, including eye care services provided by ophthalmologists, optometrists, health centres, general practitioners and the emergency department, eye surgery, medication for eye care, laser therapy and other ophthalmic interventions.

EYE CARE SERVICE UTILISATION: We asked survey participants how many times in the past year they visited each group of eye care provider, and if more than once, how many times they visited. From this prevalence of utilization, and the mean number of episodes of utilization amongst those reporting any, both of which we adjusted, we estimated episodes of eye care service utilization per annum for each group of providers in 2014 (Supplementary Table 2).

EYE CARE SERVICE COSTS: We multiplied these by the mean unit cost for each eye care service (Supplementary Table 3). We estimated eye care service unit costs in a contemporaneous study on the eye care system in Trinidad and Tobago in 2014 [17]. In this study, we contacted all registered eye care providers in Trinidad and Tobago by post or email or telephone, inviting them to complete questionnaires. These included the public and private sector tariffs for their outpatient, inpatient and emergency eye services and eye treatments. Provider groups included ophthalmologists, optometrists, public hospital eye department administrators, health centres, and public sector GPs. Further cost information sources included personal communication with officers within the Ministry of Health. Unit costs were not available for public hospital day case and overnight admission, and public sector laser, and the cost of these was therefore assumed to be the same as private sector costs.

TOPICAL OPHTHALMIC MEDICATION USE AND COST: In the national survey, we ascertained the number and name of prescription topical ophthalmic medications used in each eye in the past 3 months, and asked about compliance (Supplementary Table 4). We assumed these applied to the past 12 months for the person. We obtained private sector unit costs for 13 commonly used topical ophthalmic medications in 2014, from three private companies operating in Trinidad and Tobago in 2014, who were invited to submit competitive tender to supply drugs to the NESTT study. We averaged these to produce a mean unit price per drug. We identified public sector unit costs from the Government Chronic Disease Assistance Program (CDAP) price list for the same year. We used public sector unit costs if participants reported free topical ophthalmic medications, private sector unit costs if participants reported paying for all topical ophthalmic medication, and an average unit cost if participants reported using a mixture of free and paid prescriptions.

LASER: NESTT participants were asked to report whether they had received any laser therapy (and what type) in the private sector in the past 12 months, or ever (and what type) (Supplementary Table 5). In the health system survey we asked public and providers to report laser procedure volumes for 12 m in 2013-2014, and fees.

OTHER EYE TREATMENTS: NESTT participants were asked to detail all surgical or other medical treatments for the eyes (Supplementary Table 6). In the health system survey we asked each of the 5 regional ophthalmology departments, and private ophthalmology clinic responders, to report on surgical services offered and costs.

 b) Non-medical direct costs

Costs included refractive correction, low vision aids, and transportation to attend eye care services.

REFRACTIVE CORRECTION: The health system survey determined that spectacles and contact lenses were exclusively available in the private sector in 2014, and in addition, basic reading spectacles could be purchased over-the-counter in some supermarkets. The cost of basic distance and near spectacles, and of bifocal, trifocal and varifocal spectacles and contact lenses was determined in the eye care system study, from 48 registered optometrists who responded to the questionnaire (Supplementary Table 3) [17]. We asked survey participants whether they had purchased spectacles or contact lenses in the past 12 months as an out of pocket expenditure (Supplementary Table 7). We used the adjusted prevalence to estimate the number of pairs of spectacles purchased in 2014. We assumed that 70% of spectacles purchased were basic distance or near spectacles, and 30% were bi, tri or varifocal spectacles to estimate total cost (Supplementary Table 8).

LOW VISION SUPPORT: We invited those with low vision (best-corrected visual acuity in the better seeing eye worse than 6/18) to complete an additional questionnaire, in which they were asked whether they had received a low vision assessment in the past year, and whether they owned any low vision aids (Supplementary Table 9). The list of low vision aids included devices to assist the individual in their personal, home and work environments, and was developed in consultation with the Low Vision Clinic at the University of the West Indies in St Augustine, Trinidad, and the Blind Welfare Association in Port of Spain, Trinidad, in 2013. The adjusted prevalence of low vision was used to estimate the number of cases in Trinidad and Tobago in 2014, and the crude proportion using each type of low vision aid was used to estimate low vision aid purchases in 2014. The unit cost of low vision assessment was ascertained directly from optometrists in the eye care system study [17]. The unit price of individual low vision aids in Trinidad and Tobago in 2014 was not determined. A literature review identified a study on the cost of low vision aids in four European countries in 2004 [46], which has been used in other cost of vision loss studies in the USA and UK [47, 48]. Additional unit costs were obtained from the Royal National Institute for the Blind UK online shop [49]. These costs were adjusted to 2014 values.

TRAVEL FOR EYE CARE: Survey participants who reported attending an eye care service in the past 12 months were asked what their usual mode of transportation was (e.g. private car, taxi, water taxi, bus), and the crude proportion was applied to the total estimated episodes of eye care to estimate the number of return journeys of each type (Supplementary Table 10). Mean unit costs associated with return journeys using different modes of transportation were obtained from 450 outpatients attending eye clinics in the five regional hospital ophthalmology departments [17].

 a) Productivity loss associated with VI

The value of lost labour output caused by reduced economic productivity resulting from VI in the affected individual.

We excluded the cost of allowances (benefits, financial support for income, residence) and consideration of the time that visually impaired people spend in prevention activities or self-help groups.

We invited survey participants to specify the category into which their household monthly income fell but individual income was not ascertained, as the latter was felt to be too sensitive a question for inclusion in the NESTT. It was therefore not possible to directly determine the mean reduction in income associated with different categories of VI.

Instead, we used a simple human capital approach to estimate productivity loss. We defined the employment rate (ER) as the percentage of the population of working age in this study (40 years to 64 years, inclusive) who reported being employed over the past 12 months. We estimated the adjusted ER for each vision category (Supplementary Table 11). To estimate lost productivity, we assumed that in the absence of VI, individuals would have been employed at the same rate as the average person aged 40 to 64 years in Trinidad and Tobago in 2014. We calculated the employment ‘gap’ as 1 minus the ER in that vision category divided by the overall ER in the population aged 40 to 64 years. We calculated productivity loss for individuals in each vision category as the product of the employment gap, median annual income in 2014, and overall ER [36, 50]. For the base cost case, we used a median annual income for all occupations in 2014 of TT$54,000 [51]. In sensitivity analysis we explored two alternatives from the same Central Statistics Office data, namely the average annual income for all occupations in 2014 of TT$66,960, and the average annual income for elementary occupations of TT$40,704. We multiplied the productivity loss for individuals in each vision category by the estimated number of VI cases in the 2014 population aged 40 to 64 years, using adjusted prevalence estimates for this age group, and summed to obtain total productivity loss.

To calculate the productivity loss associated with part-time work, we used the same analysis approach, assuming a 50% reduction in working hours (Supplementary Table 12). We took this approach to be conservative, because it did not account for the possibility that those with VI might experience slower promotion or restricted choice over employment type and associated lower earning potential than those with normal vision [52]

 Productivity loss associated with sick leave

The value of lost labour output caused by reduced economic productivity resulting from VI in the affected individual.

We asked participants who reported employment how many sick days they had taken in the past 12 months in total, and specifically whether they had taken any days off to attend healthcare services for their eyes or vision (Supplementary Table 13). We also asked them to specify the total value of any lost earnings over that period. Workers who are absent on account of illness for a protracted period are likely to be replaced at some point. This period, the ‘friction period’ (e.g. 90 days), can be used to make an adjustment to the productivity loss to avoid overestimating the actual loss [53]. We did not apply a friction period adjustment in this study because the number of days of sickness taken on account of incapacity from eye disease or VI was not directly ascertained, and was anticipated to be few days on average.

 Informal care

The value of lost labour output caused by reduced economic productivity resulting from the care of an individual with VI

We asked participants if any friends or family members provided them with informal care on account of their eyes or vision state in the past month, and if so, we asked them to specify how many hours (Supplementary Table 14). To estimate the value of productivity loss associated with informal care, we used the opportunity cost method [54]. Specifically, we used the proportion reporting need for informal care to estimate the number of people, multiplied by the mean hours of informal care per person needing any, and by the hourly wage rate for an individual in an elementary occupation, of TT$21.20. The latter was calculated from the mean annual income of an elementary occupation in 2014 of TT$40,704, assuming 40 working hours per week and paid annual vacation of 4 weeks per year [51].

 Transfer payments

Payments between economic agents, for example, social welfare payments made for distributional purposes rather than as payment for goods or services.

We asked participants about their employment status over the past 12 months, to estimate the prevalence of individuals having formal ‘disabled’ status (Supplementary Table 16). We assumed that these individuals were in receipt of disability allowance if they were aged between 40 and 64 years. The means-tested disability allowance was $1500 per month in 2014 and was available up to the age of 64 years, after which it was replaced by the senior citizens pension of TT$3000 per month. We reported the budgetary impact associated with this social welfare disability payment in the total cost estimate. There was no specific carers allowance in Trinidad and Tobago in 2014. We also reported the budget cost of Government-funded programs providing services for the blind, identified from documents in the public domain, in the total cost estimate, but did not include fiscal flows resulting from reduced income tax revenue. We excluded this category from final cost estimates.

 Dead weight losses

The excess allocative inefficiency on society associated with administering transfer payments and raising additional taxation revenue [15]. Examples of dead-weight loss include welfare payments resulting in reduced labour force participation, and taxation levels disincentivising people from working.

Dead-weight losses are challenging to estimate reliably, and we did not estimate them in this study, in line with most other cost studies.

 Intangible effects

The suffering associated with a condition or disease may greatly extend beyond financial costs. Intangible effects are defined as the loss of wellbeing experienced by affected individuals.

It is challenging to quantify loss of wellbeing, and even more challenging (and controversial) to assign a monetary value to the loss. Previous cost of illness, and cost of vision loss studies specifically, have seldom assigned a monetary value to loss of welfare (Koberlein et al., 2013), but current Consensus Guidelines recommend this for cost of VI studies (Frick et al., 2010). In this analysis we adopt the common approach of estimating Disability Adjusted Life Years (DALY, Murray, 1994) associated with vision impairment, and present total cost estimates both with and without intangible effects.

Disability weights:

A number between 0 (perfect health) and 1.0 (a health state as bad as death)

Disability weights were introduced in the 1990s to give a new population health measure, the disability adjusted life year (DALY). Disability weights are obtained from ordinal measurement of preferences (paired health state comparisons). Advanced modelling is required to transform these data into weights. At least eight studies have used a variety of methodological approaches to estimate disability weights associated with vision impairment, with blindness disability weights varying from 0.17 to 0.60 [27].

Disability adjusted life years (DALYs):

DALYs aim to capture a societal assessment of the burden of a disease resulting from premature mortality and the non-fatal consequences of disease, in terms of lost welfare, subjective wellbeing and quality of life.

DALYs facilitate explicit comparison of health outcomes for health sector planning and evaluation, and greater consistency in resource allocation decisions. DALYs differ from quality adjusted life years (QALYs), which measure individual preferences for time spent in different health states (Supplementary Table 15).