Evaluation of the Effectiveness of Implementing a Screening Strategy for the Diagnosis of Prediabetes and Type 2 Diabetes Mellitus in Primary Health Care in a Brazilian Municipality

Methodology: Pragmatic, quasi-experimental, single-arm effectiveness and implementation study with a historical comparator. In each period, the analytic population will be defined by eligibility on the respective index date. In the historical period, data will estimate the reference proportion of the primary outcome and historical values of applicable secondary outcomes. In the post-intervention period, individuals eligible on the index date will be followed for 12 months via care records. Individuals who become eligible after the post-intervention index date may receive care actions under the strategy and be included in operational, implementation, and exploratory analyses, but not in outcome denominators fixed at the index date.

The study will take place in Afogados da Ingazeira, Pernambuco, covering all 15 UBS with an affiliated Family Health Team. Research-specific technological components will incur no additional cost to the municipality.

ARISE comprises two components:

1. Digital analytics layer, built on the existing epHealth infrastructure. It uses structured data from e-SUS PEC to apply predefined T2DM screening eligibility rules. Identified individuals appear in operational lists and may generate alerts for UBS managers and community health workers, supporting visits, scheduling, test requests, follow-up, and patient return. It may also flag: delays over 60 days between test request and completion, availability of lab results, and need for active case-finding. No test, diagnosis, prescription, referral, or treatment is automated; clinical decisions remain with APS professionals and managers. 2. Professional training. UBS managers, community health workers, physicians, nurses, and nursing technicians will be trained on the PCDT-DM2 and ARISE workflows, before the digital layer's activation. Professionals are research participants only for evaluating educational impact, acceptability, and usability, provided they attend the relevant training and sign the Informed Consent Form (ICF). Training participation is independent of research participation; declining data use carries no functional or institutional penalty.

A census-based approach will define population size. The primary analysis includes all individuals eligible on the post-intervention index date; final numbers depend on municipal database findings. Power calculations covered 8,000-30,000 eligible individuals and historical T2DM diagnosis proportions of 0.80%-0.90%. A population of ~14,000-15,000 eligible individuals yields ~80% power to detect a ~25% relative increase, for historical proportions of 0.85%-0.90%. For professional-related components, a census approach covers all eligible categories per the updated roster of active professionals directly involved.

Statistical analysis: The post-intervention proportion of confirmed T2DM diagnosis will be compared to the historical reference using a two-tailed one-proportion test (5% significance level). Point estimates, 95% CIs, absolute differences, and proportion ratios (where applicable) will be reported. Adjusted analyses may use logistic regression. Change in professionals' knowledge will be analyzed via paired intra-individual pre/post-training score differences. RE-AIM indicators will be mainly descriptive. Acceptability, training reaction, and workflow perception blocks will be scored separately (1-5). Usability will be assessed via the System Usability Scale (0-100) among professionals who directly used the digital component.

The study was approved by the Research Ethics Committee (REC), CAAE n. 00260026.5.0000.5330, approval opinion n. 8.705.723.

Primary outcome: Confirmed T2DM diagnosis during the observation period, treated as a binary individual indicator.