June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Locality is the strongest predictor of performance in image-based differentiation of bacterial and fungal corneal ulcers from India
Author Affiliations & Notes
  • Christopher R Rosenberg
    Oregon Health & Science University Casey Eye Institute, Portland, Oregon, United States
  • Travis K Redd
    Oregon Health & Science University Casey Eye Institute, Portland, Oregon, United States
  • Footnotes
    Commercial Relationships   Christopher Rosenberg None; Travis Redd None
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 3248 – A0283. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Christopher R Rosenberg, Travis K Redd; Locality is the strongest predictor of performance in image-based differentiation of bacterial and fungal corneal ulcers from India. Invest. Ophthalmol. Vis. Sci. 2022;63(7):3248 – A0283.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : Infectious keratitis (IK) is a major cause of blindness. Clinical recognition of the underlying cause of infection is necessary to guide empiric treatment but is often difficult even for cornea specialists. Our prior work showed India-based experts were significantly better than an international cohort at differentiating Indian cases of bacterial and fungal keratitis (BK/FK) on an image-based test. FK is much more common in tropical regions like India compared to more temperate environments like much of the US and Europe. In this secondary analysis, we explored whether this difference in performance could be explained by regional variability in FK burden.

Methods : We used a test set of 100 culture-proven photos of IK, 50 each of FK and BK. 66 cornea specialists from 16 countries (Table 1) graded each image, providing a predicted probability of FK or BK. Area under the curve (AUC) was calculated as the primary performance metric for each respondent. Because FK prevalence data are not available in most regions, the following known risk factors were used as surrogates of FK burden: tropical climate (based on latitude, average annual temperature, and dew point) and agricultural work (based on percent of gross domestic product). These predictors were incorporated into a multivariate linear regression model along with local (Indian) vs. external (non-Indian) practice location to predict respondent AUC.

Results : Bivariate analyses showed significant associations between AUC and Indian practice location (P<0.01), latitude (P<0.01), agricultural economy (P<0.01), dew point (P=0.01), and average temperature (P<0.01). However, multivariate regression revealed that only Indian practice location was statistically significant after controlling for all other predictors (P=0.01).

Conclusions : Local experts remained significantly better at differentiating Indian cases of FK and BK compared to their international counterparts, regardless of regional variability in FK burden. Future work should determine whether similar regional effects occur elsewhere, or if India-based cornea specialists are simply better at this task due to the extraordinarily high volume of IK in their patient population. The results may inform educational efforts to enable earlier initiation of appropriate antimicrobial therapy and improved visual outcomes.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

 

Table 1. Respondent location based on latitude.

Table 1. Respondent location based on latitude.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×