To investigate the impact of survivor treatment selection bias, also known as immortal time bias, on the association between virtual transition of care (VToC) follow-up and 30-day readmission rates.
Approach:
Study Design: A retrospective cohort study examining readmission rates among patients receiving VToC follow-up compared to those who did not.
Bias Analysis: Assessment of survivor treatment selection bias, also known as immortal time bias, affecting exposure classification.
Methodological Recommendations: Suggestions for intent-to-treat analysis and time-varying Cox regression models to address bias.
Key Findings:
Patients receiving VToC follow-up had a significantly lower 30-day readmission rate.
Survivor treatment selection bias led to an overestimation of readmissions in the non-VToC group due to misclassification.
Conventional risk adjustment could not correct for bias caused by outcome-dependent exposure assignment.
Interpretation:
The observed protective association of VToC follow-up on readmission rates may be spurious due to misclassification of patients experiencing early readmissions.
Early readmissions disproportionately affected the classification of patients in the non-VToC group, leading to misclassification.
Conclusion:
Temporal concordance must be explicitly addressed in studies of virtual or postdischarge interventions to avoid misclassification and ensure accurate assessment of intervention effects.
