Risk of retinal detachment and exposure to fluoroquinolones, common antibiotics, and febrile illness using a self-controlled case series study design: Retrospective analyses of three large healthcare databases in the US

Abstract

Objective: The risk of retinal detachment (RD) following exposure to fluoroquinolone (FQ) has been assessed in multiple studies, however, results have been mixed. This study was designed to estimate the risk of RD following exposure to FQ, other common antibiotics, and febrile illness not treated with antibiotics (FINTA) using a self-controlled case series (SCCS) study design to reduce risk of confounding from unreported patient characteristics.

Design: Retrospective database analysis-SCCS.

Setting: Primary and Secondary Care.

Study population: 40,981 patients across 3 US claims databases (IBM® MarketScan® commercial and Medicare databases, Optum Clinformatics).

Outcome: RD.

Methods: Exposures included FQ as a class of drugs, amoxicillin, azithromycin, trimethoprim with and without sulfamethoxazole, and FINTA. For the primary analysis, all drug formulations were included. For the post hoc sensitivity analyses, only oral tablets were included. Risk windows were defined as exposure period (or FINTA duration) plus 30 days. Patients of all ages with RD and exposures in 3 US claims databases between 2012 to 2017 were included. Diagnostics included p value calibration and pre-exposure outcome analyses. Incidence rate ratios (IRR) and 95% confidence interval (CI) comparing risk window time with other time were calculated.

Results: Our primary analysis showed an increased risk for RD in the 30 days prior to exposure to FQ or trimethoprim without sulfamethoxazole. This risk decreased but remained elevated for 30 days following first exposure. Our post-hoc analysis, which excluded ophthalmic drops, showed no increased risk for RD at any time, with FQ and other antibiotics.

Conclusion: Our results did not suggest an association between FQ and RD. Oral FQ was not associated with an increased risk for RD during the pre- or post-exposure period.

Trial registration: ClinicalTrials.gov identifier: NCT03479736-March 21, 2018.

Fig 1. Graphical representation of all time intervals described in this study.

For example: patient A had a baseline observation period that began approximately 6 months before 01/01/2012, a FQ exposure after the end of the baseline observation period, a retinal detachment after the end of the risk window for that exposure and left the study before 03/31/2017. Patient C had a baseline observation period that began after 04/01/2012, a FQ exposure after the end of the baseline observation period, a retinal detachment within the risk period for that exposure and left the study before 03/31/2017. Patient D had a baseline observation period that began after 04/01/2012, a FQ exposure after the end of the baseline observation period, a retinal detachment after the end of the risk period for that exposure, second exposure after the retinal detachment and ceased to have observations at the end of the study period, i.e., 03/31/2017.

Fig 2. Timeline of RD Events, in each database, from 60 days before to 60 days after first FQ exposure.

FQ exposure included all types of formulations. Red bars indicate that events occurred concurrently with the FQ risk window, whereas blue bars indicate event occurrences outside of the risk window. A spike of RD events was observed prior to the first day of FQ exposure. 2A: Exposure timeline in OPTUMEXTDOD; 2B: Exposure timeline in IBMCOM; 2C: Exposure timeline in IBMMDCR.

Fig 3. Timeline of RD Events, in each database, from 60 days before to 60 days after first FQ exposure.

FQ exposure included oral tablets only. Red bars indicate that events occurred concurrently with the FQ risk window, whereas blue bars indicate event occurrences outside of the risk window. 3A: Exposure timeline in OPTUMEXTDOD; 3B: Exposure timeline in IBMCOM; 3C: Exposure timeline in IBMMDCR. In contrast to Fig 2, there is no exposure peak in the days just before the retinal detachment. RD events were randomly distributed before and after the first date of FQ exposure.

Fig 4. Incidence risk ratios (IRR) by exposure and database.

Asterisks identify statistically meaningful findings based on calibrated p values. Risks associated with exposures that included tablets only are shown in blue lines whereas risks associated with exposures that included all formulations are shown in orange lines. For fluoroquinolone exposures and for trimethoprim without sulfamethoxazole exposures, there is a substantial and statistically significant association of RD with exposure across all databases when all formulations, including topical ophthalmic formulations are included. When only oral exposures are included, the magnitude of this this association for these two medications becomes much smaller across all databases and the association loses statistical significance except in the panel for trimethoprim without sulfamethoxazole in the IBMMCDR database.