In this issue of CHEST (see page 359), Rose and colleagues8 examine the relationship between the length of clinic INR follow-up and TTR at 100 centers in 104,451 people receiving warfarin from the Veteran Affairs Study to Improve Anticoagulation (VARIA) cohort. The INR target was 2.0 to 3.0, with the predominant indications for OAC therapy being atrial fibrillation (AF) (65%) and VTE (27%). Mean individual and site risk-adjusted TTR were calculated as a measure of anticoagulation control over time, excluding INR values within the first 6 months of OAC therapy and during hospitalization. At the patient level, as the number of days between INR tests increased, TTR increased, even after adjustment for covariates, with the mean interval after an in-range INR of 29.8 days. However, at a site level the results appear to suggest the opposite, with longer mean follow-ups between INRs being associated with worse TTR: For each additional day of follow-up, the risk-adjusted TTR was 0.51% lower (P = .004). However, this relationship was moderated as the number of consecutive INR values within range rose; with three or more INR values within range, this relationship was no longer statistically significant (−0.12%, P = .46). Subgroup analyses at the site level among patients with AF and those aged >75 years demonstrated similar results to the whole group. Rose and colleagues8 suggest that 28 days is the optimal period between INR tests, but for patients with three or more consecutive INRs within range, a longer period between INR checks may be both safe and cost-efficient. This contradiction between site and individual TTR reflects the practice of monitoring more regularly those patients who clinicians suspect may not have their next INR in target range and increasing the length of time between INR visits for those patients with good INR control. At the individual level, this demonstrates that less frequent INR visits are associated with better INR control, whereas at a site level, longer delays between INR visits reveal poorer INR control, a reflection of the overall population studied (ie, those with good and poor INR control).