Abstract / Summary
In type 2 diabetes mellitus (T2DM) management, high nutritional knowledge often fails to translate into appropriate dietary practices due to complex barriers. This study quantified the knowledge-practice gap and identified multi-level barriers influencing dietary non-adherence among adults with diabetes in Accra, Ghana, using a Socio-Ecological Model (SEM) framework. A hospital-based cross-sectional study was conducted among 392 patients with T2DM at the 37 Military Hospital in Accra between May and October 2024. Methodological reporting was aligned with STROBE guidelines. Participants were recruited via systematic random sampling (k = 5). Nutritional knowledge was assessed using the 24-item Diabetes Knowledge Questionnaire (DKQ-24), and dietary practice was measured using the Summary of Diabetes Self-Care Activities (SDSCA) scale. The knowledge-practice gap was defined as the percentage difference between patients achieving "Good Knowledge" (≥ 70%) and those achieving "Good Practice" (≥ 4 days/week, a pragmatically calibrated threshold for resource-constrained food environments). Multi-level perceived barriers were grouped using the SEM. Multivariable logistic regression was used to identify independent predictors of dietary practice, reporting both crude and adjusted odds ratios alongside model diagnostics. Of the 392 participants (mean age 55.4 years, SD = 10.9; 72.4% female), 74.4% exhibited "Good Knowledge" of diabetes nutrition, whereas 68.4% achieved "Good Practice", resulting in a population-level knowledge-practice gap of 6.0 percentage points. The individual-level "Failure to Translate" (FTT) rate was 21.9%, calculated as the proportion of knowledgeable patients who nonetheless exhibited poor practice (64/292). Main barriers spanned the SEM: structural food unavailability in local markets (84.6%), clinical diet unpalatability (65.9%), financial constraints (60.1%), long outpatient waiting times (54.8%), and inconsistent dietitian access (50.5%). Good nutritional knowledge was associated with a higher odds of dietary adherence (Adjusted Odds Ratio = 2.80, 95% Confidence Interval [CI]: 1.61–4.16), but this association was significantly attenuated by low daily household income (< 100 GHS/day; approximately US$6.70 at the time of the study; AOR = 0.54, 95% CI: 0.31–0.94) and the absence of family support (AOR = 0.48, 95% CI: 0.26–0.89, p = 0.019) was associated with approximately half the odds of dietary adherence compared to those with family support. Multivariable model diagnostics confirmed excellent fit (Hosmer–Lemeshow p = 0.554), minimal multicollinearity (mean VIF = 1.24), and strong discriminative capacity (Area Under the ROC Curve = 0.76). Theoretical knowledge is necessary but insufficient for dietary change when structural and socioeconomic barriers remain unaddressed. Structural food insecurity, low household purchasing power, and culinary mismatches with traditional Ghanaian cuisine prevent patients from translating knowledge into action. Public health policies must shift from generic patient education toward targeted nutritional safety nets and culinarily competent, culturally tailored dietary counselling.