| Sumario: | Aim: To study trajectories of HbA1c in type 2 diabetes (T2D) patients referred to diabetes outpatient clinics (DOCs), and to explore characteristics of referrals and patient pathways in patients treated in DOCs. Methods: We retrospectively followed T2D patients from the Norwegian population-based ROSA 4 study to identify persons with T2D who were referred to a DOC. We used latent class trajectory modelling to identify subgroups of patients with similar patterns of HbA1c one year before to one year after the first consultation at a DOC. We performed multinomial regression analyses to identify baseline characteristics associated with group membership. Results: Four hundred and two of 6716 T2D patients started treatment at a DOC, constituting a yearly starting rate of 1.5%. We identified three classes of HbA1c trajectories: (1) stable moderate hyperglycaemia (75%); (2) severe hyperglycaemia with a decline in HbA1c around referral (14%) and (3) severe hyperglycaemia with a decline in HbA1c after starting treatment at the DOC (11%). HbA1c trajectories were associated with diabetes duration RRR 0.92, CI (0.87, 0.97) in class 2 vs. 1 and 0.93 (0.88, 0.98) in class 3 vs. 1. Some differences were found between clinics in rejection rate, processes of care, and duration of follow-up. Conclusions: Norwegian GPs handle most T2D patients themselves. Those with T2D and severe hyperglycaemia had a considerable benefit from being referred to a DOC, though with two separate trajectories: One where HbA1c improved around the time of referral, and another that improved after starting in a DOC. Key Points: The data originate from a high-quality database of type 2 diabetes patients. The strength of the data-driven latent class trajectory modelling (LCTM) is that it can uncover hidden patterns of HbA1c changes in heterogenic populations. On the other hand, LCTM have concerns about generalisability, as it may be difficult to replicate. Among other limitations are a rather small number of patients referred during the study years, increasing the risk of type 2 errors in further analyses.
|
|---|