Study population

To evaluate the association of GnRH and a PCa diagnosis per se with atherogenic lipids and LLT, we created two cohorts: “PCa-Exposure cohort” (men with PCa on or not on GnRH vs. PCa-free men) and “GnRH-Exposure cohort” (men with PCa receiving GnRH vs. men with PCa but not receiving GnRH) (Supplementary Fig. 1).

In the PCa-Exposure cohort, we selected men with T2DM who had at least four NDR registration data and were diagnosed with PCa on/not on GnRH after their third registered date in NDR (exposed men). The start of follow-up of this cohort was the date of PCa diagnosis. For each exposed man in this cohort, five men with T2DM but without PCa (non-exposed men) were randomly selected from NDR. The non-exposed men were matched to corresponding exposed men in this cohort on number of NDR registrations before inclusion and average time between NDR visits. Start of follow-up for each non-exposed man was inherited from the exposed man (Supplementary Fig. 1).

The GnRH-Exposure cohort included men with T2DM who had at least four NDR registrations data and were diagnosed with PCa and treated with GnRH after the third NDR registration (exposed men). The start of follow-up was the date of the first filled prescription for GnRH. This cohort also included five men with T2DM and PCa but not on GnRH for each man treated with GnRH (non-exposed men). The non-exposed men were randomly selected from the NDR and matched to corresponding exposed men in this cohort based on the number of previous NDR registrations and average time between NDR visits. Start of follow-up for these non-exposed men was inherited from the corresponding exposed men in this cohort (Supplementary Fig. 1).

PCa-Exposure cohort enabled us to explore the association between PCa diagnosis per se and worsening dyslipidaemia. Additionally, we grouped exposed men in this cohort into (1) men with PCa on GnRH and (2) men with PCa but not on GnRH. It allowed us to examine the relationship between the use of GnRH and worsening dyslipidaemia as a first step in this cohort, through making comparison between men with PCa on GnRH and non-exposed men. Nevertheless, this estimated association might be affected by PCa diagnosis in this cohort. Meanwhile, the health seeking behaviours may be differed between men with PCa on GnRH and non-exposed men in the PCa-Exposure cohort, which might affect findings on the association between the use of GnRH and dyslipidaemia. Therefore, we further created the GnRH-Exposure cohort with the primary aim to investigate the association of lifelong use of GnRH and dyslipidaemia. The GnRH-Exposure cohorts precluded the potential impact of PCa diagnosis and reduced the impact of difference in healthcare seeking behaviours by matching non-exposed men to exposed men on the number of previous NDR registrations and average time between NDR visits.

In each cohort, we included men with at least one NDR registration in the follow-up, aiming to maximise use of data from the NDR. Meanwhile, to reduce the selection bias caused by the missing data at baseline, we include three NDR registrations prior to the start of follow-up as baseline characteristics and used last observation carried forward to impute the missing data at baseline (see below). Given above, in each cohort, we included men with at least four NDR registrations. The data on baseline characteristics was collected from the three last NDR records prior to the start of follow-up.