In sensitivity analysis, including workers whose weekly working hours changed during follow-up, the association was slightly attenuated but still borderline significant.
2
—
—
The sentences
In sensitivity analysis ( Table S1 ), when Cox proportional hazards regression analyses were performed for 69,516 participants, including workers whose group of working hours had changed during the follow-up period, the association between weekly working more than 52 h at baseline and incident glucose intolerance, compared with working 35–40 h, was still observed with borderline significance in Model 3. 4.