Discussion

The differences between Group 1 and Group 2 are likely attributable to a combination of multiple factors. Regarding the decrease in ophthalmology consultation volume, physical, and emotional stressors ^{Reference Jacob, Arnold and Hunleth23} may have encouraged Houstonians and volunteers to delay eye care, which may have been perceived by patients as lower priority ^{Reference Taber, Leyva and Persoskie24} when compared to flooded homes or stranded family members. However, Pines et al. ^{Reference Pines25} reported significant volume increases at Houston freestanding emergency departments during the 9 days after Harvey, with peak levels reaching 125% of the baseline. A similar volume spike was reported by Long Island emergency departments following Hurricane Sandy in 2012. ^{Reference Kim, Schwartz and Hirsch26} Thus, accessibility may have also contributed to the decrease in volume at BTGH. Flooded streets may have reduced hospital accessibility in Houston, forcing Houstonians to instead visit closer, freestanding emergency departments. Accessibility issues may have also driven the rise in privately insured patient percentage and decline in county-insured and uninsured patient percentage. Residents who may have preferred a private hospital could have been forced to go to BTGH by transportation limitations. In contrast, county-insured and uninsured patients who may have relied on public transit could have been stranded without means of transportation to BTGH. However, it should be noted that our analysis did not find statistical differences in the distance from patient homes to BTGH, and our heat map did not appear to show correlation between flooding severity and patient volume. Further research is needed to clarify the weight of geography on natural disaster eye morbidity volume. Nevertheless, for future natural disaster preparation, geographic variables should be considered when forecasting ophthalmology case volume.

The proportional rise in patient consultations associated with immunosuppression medication is also likely multifactorial. Immunosuppression can increase susceptibility to infection, including the variety of fungal, viral, and bacterial diagnoses documented in Group 2. However, this surge may have also arisen from refill needs. After Hurricane Katrina, Jhung et al. reported that 68% of all evacuee prescriptions were for chronic medications. ^{Reference Jhung, Shehab and Rohr-Allegrini27} Similar refill demand has been documented after Hurricanes Iniki, ^{Reference Gnauck, Nufer and LaValley28} Wilma, ^{Reference Taylor, Stokes and Bajuscak29} Andrew, ^{Reference Gnauck, Nufer and LaValley28} and Ike, ^{Reference Noe, Schnall and Wolkin30} although systematic review has concluded that medication replenishment needs vary by disaster. ^{Reference Ochi, Hodgson and Landeg31} Given the similar medication uptrend observed in our ophthalmology consultation patients, ophthalmologists should be mindful of medication inventory when preparing for natural disasters, particularly for chronic eye diseases. In addition, public education may also improve disaster eye health. Heslin et al. reported that only 60.1%–81.9% of California residents carried a 2-week emergency supply of their chronic medications ^{Reference Heslin, Gin and Afable32} ; thus, preemptive awareness and planning may reduce emergency medication needs.

Perhaps most notably, infectious eye diagnoses rose dramatically after Hurricane Harvey, increasing over 150% from 8 to 22, with an odds ratio of 4.43. In comparison, the odds ratio for traumatic eye diagnoses was 2.13. Previous literature has also described ocular infections after natural disasters. ^{Reference Sridhar, Lam, Pasol and Sternau9,Reference Davies, Smith, Hink and Durairaj10,12,Reference Bich, Quang and Ha14,Reference Osaadon, Tsumi and Pokroy16} However, some authors have also reported increases in traumatic eye injuries following typhoons/hurricanes ^{Reference Jones11,Reference Rotheray, Aitken and Goggins33} and other natural disasters such as earthquakes and avalanches. ^{Reference Osaadon, Tsumi and Pokroy16} These discrepancies between infectious and traumatic morbidities may be correlated with the characteristics of specific natural disasters. While Hurricane Harvey was a true Category 4 hurricane, ³ the majority of its damage stemmed from flooding, rather than wind speed. It is possible that flooding may correlate more closely with infectious eye diseases, while high wind speeds may cause more eye trauma. Further research is needed, but consideration for the specific type of natural disaster and even the specific characteristics of a natural disaster category, such as hurricanes, may be helpful for eye health preparation.

To our knowledge, an analysis of natural disaster eye health with this depth and inclusivity has never been performed before in the English literature. Our study is unique in its combination of control group utilization, statistical analysis, and ophthalmology-specific data. Most importantly, our study provides guidance for natural disaster preparation in ophthalmology and lays the groundwork for reducing the research gap in natural disaster eye health.

As with any study, our analysis has limitations. Our methodology relied on EHR documentation for data inclusion; thus, only consultations recorded in the EHR system were analyzed. Difficulties with comprehensive case documentation have been noted in literature, ^{Reference Wolf, Taylor and Placek34} even without hurricane-related disruptions. Therefore, realistically, our analysis is likely an underrepresentation of the full ophthalmology consultation data set. However, while our post-hurricane data may be underreported, we collected our control group data in an identical manner to ensure fair statistical comparison. Since the majority of our analysis was proportion-based, we remain confident that our statistical conclusions between groups are valid, although specific numbers may be variable. For future studies, prospective consultation recording, rather than retrospective review, may provide stronger data. While our analysis is one of the largest and most detailed studies of its kind, only 1 physical location (BTGH) was assessed, and the 2-month study duration may have limited the capture of indirect health consequences, particularly those related to postponed consultations, rescheduled appointments, and overall delayed care. BTGH is the main county hospital in Houston and one of the busiest trauma centers in the United States, but a larger location sample size and longer study period may strengthen the analysis of future studies. Finally, although a short time period between Harvey’s landfall and this analysis provides a swift conclusion and guidance, our analysis was focused on short-term, rather than long-term ocular morbidity. Future research may provide more insight into long-term natural disaster eye health by incorporating longitudinal models used by other medical specialties ^{Reference Schwartz, Tuminello and Kerath35,Reference Horney, Casillas and Baker36} or analyzing longitudinal natural disaster databases such as the Hurricane Harvey Registry. ^{Reference Miranda37} In addition to direct eye health effects, these future studies could also investigate indirect ophthalmological complications driven by social determinants of health and delayed care.