Study Characteristics
A total of 2452articles were retrieved, of which 33 articles met the study criteria for inclusion (Figure 1). After construing the full text, we finally included 23 studies for the review (Table 1). We did not include data for pooled compliance estimate from three studies8, 26,34as the proportion of children compliant with spectacle use was not available. However, descriptive data from these studies was included.
All the 23 studies included were cross-sectional in nature. The studies are geographically diverse and covered 14 countries. Most (34.7%) of the studies were from South-East Asian region8,10-12,15,19,23,followed by Eastern Mediterranean region9,13,18,25 (17.4%). More number (39.1%) of studies was from lower-middle income country (LMIC) groups. The per capita GDP of the countries ranged from 835 USD for Nepal to 59532 USD for USA. Percentage expenditure on health ranged from 2.69 for Pakistan to 16.84 for USA. In most studies, compliance with spectacles was defined if the child was wearing spectacles at the time of surprise visit by the investigators or had the spectacles in bag (65.2%). For a few studies compliance was defined by taking interviews of the children and asking questions regarding their patterns of spectacle use (34.8%). The prescription patterns varied across studies. Some studies used the cutoff of >0.5 Diopters for defining myopia and assessing compliance while a few defined myopia with a cut off of >1 Diopters. Similarly, for hyperopia the cut offs ranged from +1.0 D to 2.5 Diopters. A few studies defined the prescriptions and cutoffs in terms of visual acuity and included children with visual acuity <6/9 to <6/12 across studies.
Population characteristics
The total number of children studied was 7859. The children enrolled were from all age groups ranging from preschool to end of school. The period of follow up ranged from ‘no follow up’ to follow up of 18 months. The variation in ages was vast and contributed to significant heterogeneity.
Quality Assessment
The studies were assessed for the methodological quality based on the tool developed by Wong WC et al31, 32. Close to half (43.5%) of the studies were of good quality while 9 studies (39.1%) were of satisfactory quality and 4 studies (17.4%) were poor quality studies with score <30%. The tool used for quality assessment is as shown in eTable 1 (supplementary file). The results of quality assessment of studies are attached as a supplementary file (eTable 2).
Compliance with Spectacle use:
The overall compliance with spectacle use was 40.14% (95% CI- 32.78-47.50) for 20 studies. The compliance varied from 9.84%(95% CI=2.36-17.31) to 78.57 %(95% CI=68.96-88.18). Four studies had extreme values and their confidence intervals did not overlap those of other studies. Two of these studies reported very low compliance16, 24 while 2 reported very high compliance23, 25with spectacle use. A forest plot of the studies and the compliance rate is as shown in Figure 2. The compliance across almost all income groups was poor. The correlation coefficient for relationship between per capita GDP and percentage compliance was -0.051, indicating a decrease in compliance with an increase in per capita income. However, this was not statistically significant (p=0.830). Similarly, the correlation coefficient for relation between percentage expenditure on health and percentage compliance was -0.238, indicating that an increase in percentage expenditure on health was associated with a decrease in compliance. This relationship also, was not statistically significant (p=0.312). The prescription cut offs used for measuring compliance had an effect on the compliance pattern across studies. Based on the setting of the study the compliance was pooled for screening vs clinical care. It was observed that the compliance was 45.84% in the setting of clinical care and 39.33 in the setting of screening (eFig.1). This difference was not statistically significant. The method of measurement of compliance also varied across groups. A few studies used observation/ surprise visit as a measurement method while others used interviews/questionnaires for measurement of compliance. We grouped and analyzed the studies by compliance measure. The pooled compliance where observation was the method of measurement was 39.24% while the pooled compliance where interview was the method of measurement of compliance; the pooled compliance was 43.23% (eFig.2). This difference was not statistically significant, indicating that the measurement method did not significantly alter the compliance. A few studies that used a higher prescription cut off, reported better compliance as compared to studies that had lower cut offs for measurement of compliance, indicating lesser compliance at lower refractive errors. However most studies did not report any specific cut offs for prescriptions and some studies took cut offs using visual acuity, hence the pooled effect could not be observed.
Sensitivity Analysis and Investigation for heterogeneity
Sensitivity analysis was done to know whether the poor quality studies10, 12, 13, 17 had any effect on the overall compliance. The overall compliance with spectacle use did not vary significantly and was 40.09% when these studies were removed from the analysis. The forest plot excluding these studies is as shown in Figure3. The Galbraith plot (eFig.3) has shown that the studies had significant heterogeneity. The heterogeneity could be attributed to varying sample sizes and prescription cut offs (degree of refractive errors).
Reasons for Non-compliance with Spectacle Use
The most commonly cited reasons for non-compliance with spectacle use in the studies were broken9-12,14,15,17-20,23,24,27/lost spectacles9-12,14,15,17-20,24,27, forgetfulness9-11,14,15,17-19,24,27 and parental disapproval8,9,11,12,15,16,18,23,24,34. These were followed by headache10-12, 15, 17,23,24,27, teasing by peers10,12,19and dislike for spectacles9,14,16-23. Other reasons mentioned in a few studies were use only when required10,12 unclear vision11,12,22,27,34, unattractive frames/poor appearance8,34, fear of injuries8,34, lack of affordability9,12,16, uncomfortable spectacles18,34 and negative attitude of the society8,15,21(eFig.4).The reasons for non-compliance were pooled for different studies, and is as shown in the forest plots (Figure. 4(a-d). The reasons were broadly classified into personal factors, social factors, visual problems and breakage/loss/forgetfulness. The non-compliance due to personal factors was25.78% (fig 4a), for social factors it was 13.18% (fig 4b), visual problems/headache it was 5.47% (fig 4c) and for breakage/loss/forgetfulness it was 23.34% (fig4d) of the total. The results clearly show that personal factors and breakage/loss/forgetfulness were the most commonly cited reasons for noncompliance.