Impact of Refractive Outcomes on Bias in Follow-up and Completion of Patient-Reported Outcome Measures after Laser Vision Correction.

PURPOSE

To examine factors contributing to completion of a patient-reported outcome (PRO) measure in patients undergoing laser vision correction.

DESIGN

Retrospective, population-based study.

PARTICIPANTS

All patients who underwent primary laser vision correction with a target of plano from July 1, 2014, to June 30, 2016, at a large refractive surgery center.

METHODS

Patients were asked to complete a PRO measure at the time of their preoperative and months 1 and 3 postoperative visits. Characteristics between patients who attended and did not attend the follow-up visits and completed and did not complete the PRO measure were compared. A logistic regression was performed to identify factors associated with likelihood of follow-up and completion of PRO measure. An inverse probability censoring weighted model was created to account for selective loss to follow-up and used to adjust the PRO satisfaction measure.

MAIN OUTCOME MEASURE

Completion of the PRO measure at 1 and 3 months.

RESULTS

A total of 37 043 patients were identified. Of these, 20 501 completed a 1-month postoperative PRO measure and 10 474 completed a 3-month postoperative PRO measure. Patients completing a PRO measure were more likely to be older, be female, have had photorefractive keratectomy (PRK), have completed a preoperative PRO measure, and have had a preoperative hyperopic correction (P < 0.001 for all comparisons). For every line of postoperative uncorrected acuity worse than 20/16, the odds ratio of completing a PRO measure was 1.33 (95% confidence interval [CI], 1.30-1.36, P < 0.001) at 1 month and 1.29 (95% CI, 1.26-1.33, P < 0.001) at 3 months. At 1 month, there was no difference between the raw and model-adjusted rates of satisfaction with vision, but at 3 months the adjusted rate was significantly higher than the raw rate.

CONCLUSIONS

Patients with worse objective visual outcomes were more likely to complete PRO measures in this population-based study. In a setting with loss to follow-up, PRO measures require methods to address missing data for correct interpretation.