Climate Determinants of Keratoconus: Insights From a Systematic Review of Prevalence

Abstract

Purpose: The reported prevalence of keratoconus varies widely worldwide, but the causes of this variation are not well understood. We therefore aimed to explore the potential impact of local climate variables on keratoconus prevalence.

Methods: The worldwide prevalence of clinical keratoconus in the general population was systematically reviewed. In each eligible prevalence area, four climate variables deemed possibly relevant to keratoconus were assessed: daily maximum temperature, relative humidity, ultraviolet radiation, and wind speed. Climate variables were calculated using worldwide gridded climate datasets from the European Center of Medium-Range Weather Forecasts. Population density weighting was applied to enhance exposure accuracy. The average of each climate variable was calculated over the 10 years preceding data collection of each study. The potential impact of those climate variables was investigated using multiple linear regression adjusted for the gross domestic product per capita (based on purchasing power parity) with the natural logarithm of prevalence as the outcome variable.

Results: Sixteen eligible studies were identified. After filtering to retain one prevalence estimate per region, 11 studies including datapoints from 61 areas were analyzed. The median (interquartile range) prevalence of keratoconus was 0.10% (0.07%-0.19%). Multiple regression revealed a significant negative association between humidity and keratoconus prevalence (β = -0.03; 95% confidence interval, -0.06 to -0.01; P = 0.004). In contrast, the other analyzed climate variables were not significantly associated with keratoconus prevalence.

Conclusions: Using global gridded climate maps, we observed a significant and biologically plausible link between low humidity and keratoconus. This suggests that humidification could benefit patients and at-risk groups.

Figure 1.

Distribution of humidity (top) and prevalence (bottom) across the included study regions. The maps were created using Microsoft Excel 365. The entire state or country was shaded uniformly, even when data were derived from a smaller subregion. Hawaii was manually enlarged and repositioned to improve visibility. In the United Kingdom, two prevalence estimates were available; this graph displays data from the study that directly reported prevalence values.

Figure 2.

Suggested pathways for the relationship between low humidity and keratoconus. The letter on each arrow describes relevant findings from the cited studies. (A) Allergic conjunctivitis visits were significantly less when relative humidity increased (odds ratio = 0.998; P < 0.001) based on a multiple logistic regression model. (B, F–H, J, K, N, P, Q) Descriptive/review articles. (C) Participants who lived in places with <70% relative humidity showed a higher prevalence of DED than others (17.7% vs. 13.6%; P < 0.01) based on a log-linear model adjusted for age and sex. (D) Average relative humidity mean in the last 40 years was significantly correlated with minimum corneal thickness (Pearson correlation r = 0.564; P = 0.010). (E) First applanation (A1) length and the stiffness parameter at first applanation (SP-A1) were significantly lower in the vernal keratoconjunctivitis group (P = 0.002 and P = 0.04, respectively) based on the independent samples t-test. (I) In DED patients, conjunctival staining scores were significantly associated with second applanation velocity (P = 0.04), and corneal staining scores were significantly associated with the second applanation length (P = 0.01). This was assessed using a mixed-effects linear regression model adjusted for spherical equivalent refraction, intraocular pressure and central corneal thickness. (L) Dry eye patients had significantly lower central corneal thickness (CCT) compared to age- and gender-matched healthy controls (mean CCT = 537 vs. 561 µm; P < 0.01) based on the paired t-test. (M) After 1 minute of eye rubbing, smaller SP-A1 (P < 0.001), higher deformation and deflection amplitudes (P < 0.001 and P = 0.012, respectively), higher peak distances (P < 0.001), earlier A1 times (P < 0.001), faster velocities (P < 0.001), and lower maximum inverse radii (P = 0.004) were observed. These results are from the paired t-tests. (O) Inverse-variance weighted Mendelian randomization (IVW-MR) suggested that the corneal resistance factor has a negative causal relationship with keratoconus (β = −1.37 ± 0.086). (R) Similarly, IVW-MR suggested that CCT has a negative casual relationship with keratoconus (β = −0.027 ± 0.0021).