Who Leaves DRCRnet Eye Studies, and What Predicts It?

Researchers reviewed public records from the Diabetic Retinopathy Clinical Research Retina Network, or DRCRnet. They wanted to see who stopped taking part and which signs at the start might help explain it. The records covered 7,353 participants. Their average age was 60. Female participants made up 48.0%. The group was 64.1% non-Hispanic White, 17.8% non-Hispanic Black, and 14.2% Hispanic.

Over a median follow-up of 24 months, 1,110 participants discontinued. That is 15.1% of the total. The 95% confidence interval ran from 14.3% to 15.9%. The researchers calculated this rate, then checked demographic traits, health measures, and study design features. They used both univariable and multivariable logistic regression. Generalized estimating equations helped account for connections within the same study.

After adjustment, some groups had significantly higher odds of discontinuing. Hispanic participants had an adjusted odds ratio, or aOR, of 1.73, with a 95% CI of 1.38 to 2.18. Non-Hispanic Black participants had an aOR of 1.36, with a 95% CI of 1.14 to 1.62. Higher odds also appeared among participants younger than 50, those aged 50 to 59, and those aged 70 or older. Their aORs were 1.46, 1.26, and 1.20, with 95% CIs of 1.21 to 1.77, 1.05 to 1.51, and 1.03 to 1.39.

Starting health measures mattered too. For each one-unit rise in baseline HbA1c, the aOR was 1.09, with a 95% CI of 1.06 to 1.12. Participants whose baseline blood pressure was at least 140/90 mmHg had an aOR of 1.24, with a 95% CI of 1.09 to 1.41. Study length was another signal. Those in studies lasting 24 to under 36 months had an aOR of 1.64, while studies lasting 36 to 48 months had an aOR of 2.68. Their 95% CIs were 1.02 to 2.64 and 1.87 to 3.84.

Future study planning may need to examine these starting factors instead of assuming everyone has the same chance of staying enrolled. Racial and ethnic identity, age, HbA1c, blood pressure, and follow-up length all deserve attention when researchers design later DRCRnet studies. That makes discontinuation not just an administrative problem, but a clue about whether a study is equally workable for different participants.