Medication Adherence and Its Discordance with Glycemic Control in Type 2 Diabetes: A Real-World Study in Primary Health Care in the Brazilian Amazon

Background/Objectives: Medication adherence is a critical determinant of therapeutic outcomes in type 2 diabetes mellitus (T2DM); however, its relationship with glycemic control remains inconsistent, particularly in real-world and socially vulnerable settings. This study aimed to evaluate medication adherence using multiple validated instruments, assess disease-related knowledge, and examine their relationship with glycemic control, with a focus on potential discordance between self-reported adherence and objective metabolic outcomes. Methods: A cross-sectional analytical study was conducted with 237 adults with T2DM receiving care in a primary health care (PHC) unit in the Brazilian Amazon. Medication adherence was assessed using the Almeida Adherence Scale, ARMS-12, and the Haynes–Sackett test, while disease-related knowledge was evaluated using the Batalla–Martínez questionnaire. Glycemic control was determined based on glycated hemoglobin (HbA1c) values obtained from clinical records within the previous three months. Descriptive and comparative analyses were performed. Results: The study population was predominantly female (64.1%) and aged 40–59 years (55.7%), with a high prevalence of socioeconomic vulnerability. Non-adherence was identified in 55.7% of participants using the Almeida Adherence Scale, whereas higher adherence rates were observed with ARMS-12 (91.1%) and the Haynes–Sackett test (72.2%). Inadequate disease-related knowledge was found in 77.2% of participants. Among individuals with available HbA1c data (n = 116), the mean HbA1c was 8.63% (SD = 1.65), and 81.9% presented inadequate glycemic control (HbA1c ≥ 7%). Notably, among participants classified as adherent by the ARMS-12 scale (91.1%), inadequate glycemic control was nonetheless present in 81.9% of those with available HbA1c data, illustrating the magnitude of the observed discordance between self-reported adherence and objective metabolic outcomes. Cross-tabulation of each adherence instrument against glycemic control showed no statistically significant associations (chi-square with Yates correction; ARMS-12: p = 0.631, φ = 0.045; Almeida Adherence Scale: p = 0.301, φ = 0.096; Haynes–Sackett: p = 0.800, φ = 0.024). Multivariable logistic regression (Nagelkerke R2 = 0.321; AUC = 0.834) identified older age (aOR = 0.92; 95% CI: 0.87–0.96; p < 0.001) and higher income (aOR = 9.96; 95% CI: 2.05–48.32; p = 0.004) as independent predictors of glycemic outcome, while no adherence measure was independently associated with HbA1c ≥ 7%. A sensitivity analysis using HbA1c ≥ 8.0% revealed poor control in 59.5% of participants (n = 69/116). Conclusions: Despite varying levels of self-reported medication adherence, inadequate glycemic control was highly prevalent. The absence of statistically significant associations between self-reported adherence and HbA1c, combined with the high prevalence of poor glycemic control regardless of adherence status, is consistent with the hypothesis that adherence alone does not fully explain metabolic outcomes in T2DM. Given the cross-sectional design, no causal inferences can be drawn. These findings highlight the need for integrated care strategies in primary health care, including improved health literacy, structured pharmacotherapeutic follow-up, and the use of multiple adherence assessment tools to better inform clinical decision-making.