What the correlation analysis shows Sugar ↔ IR: r = 0.87–0.93 IR ↔ Hypertension: r = 0.88 Grain ↔ IR: r = 0.21 (weak)

The sugar–insulin resistance correlation is very strong (r = 0.87–0.93 across all four regions in the post-1985 measured period). This is an ecological correlation, so it does not prove individual causation, but it is consistent with the mechanistic literature: fructose in refined sugar and HFCS drives hepatic de novo lipogenesis, hyperuricaemia, and endoplasmic reticulum stress — all of which impair insulin receptor signalling independently of caloric intake.

The insulin resistance ↔ hypertension link is equally strong (r = 0.88). This is well-established mechanistically: IR reduces nitric oxide bioavailability (endothelial dysfunction), activates the renin-angiotensin-aldosterone system, increases sympathetic nervous system tone, and promotes sodium retention — all direct pathways to raised blood pressure. People with IR have a ~2× higher risk of developing hypertension (Reaven, Diabetes 1988; Ferrannini et al., NEJM 1987).

The grain correlation is weak (r ≈ 0.21) and is explained almost entirely by covariance with overall caloric load rather than a specific grain-IR pathway. Refined grain starch does raise postprandial insulin, but the glycaemic load per gram is substantially lower than fructose-containing sugars, and the effect is greatly modulated by fibre content, processing degree, and food matrix.

Italy again stands out as the critical case. With sugar rising +300% from a very low base (9 → 36 kg/cap), Italy's IR prevalence rose from an estimated ~3% to ~28% — tracking almost perfectly with its sugar curve. Yet its grain consumption was the highest of all four regions throughout. This is your strongest single piece of ecological evidence for the programme rationale.

On the pre-1985 estimates: Back-projections use obesity prevalence (BMI >30), type 2 diabetes incidence data going back to the 1930s, and historical glucose tolerance surveys. These are proxy markers, not direct HOMA-IR measurements, and should be labelled clearly as estimated when used in presentations. The confidence intervals would be wide. However, the directional trend — low IR in the low-sugar era, rising with the sugar surge — is well supported by the historical epidemiological record.