Background/aims To identify and evaluate risk factors associated with the need to convert intended deep anterior lamellar keratoplasty (DALK) to penetrating keratoplasty (PK).

Methods Retrospective institutional cohort study including all consecutive eyes undergoing intended DALK between May 2015 and October 2018 at ‘Villa Igea’ Hospital (Forli, Italy). The indications for surgery were categorised as (1) keratoconus without scarring; (2) keratoconus with scarring; (3) non-keratoconus without scarring; and (4) non-keratoconus with scarring. Multivariate binary logistic regression analysis was performed, introducing, as independent variables, those that reached a significance level of less than 0.05 in univariate analysis. The main outcome measure was whether or not conversion to a PK occurred.

Results 705 eyes were included, with conversion to PK occurring in 16.2% (n=114) of cases. The factors that remained significant in multivariate analysis were corneal scarring (OR=3.52, p<0.001), manual dissection (OR=42.66, p<0.001), type 2 bubble (OR=90.65, p<0.001) and surgeon inexperience (OR=10.86, p<0.001). A receiver operating characteristic (ROC) curve based on the factors significant in the multivariate binary logistic regression analysis achieved a sensitivity of 89.5% (95% CI 82.3% to 94.4%) and a specificity of 80.2% (95% CI 76.8% to 83.3%) with an area under the ROC curve of 0.91 (95% CI 0.88 to 0.93) (p<0.001)

Conclusion Occurrence of a type 2 bubble, the need for manual dissection, the presence of scarring and surgeon inexperience are independent risk factors for the need to convert intended DALK to PK. Correct identification and management of the type of bubble achieved during pneumatic dissection is instrumental in minimising the rate of conversion to PK.