The drink tied to 59% more diabetes and the foods tied to less

In a UK study of 148,858 adults, people who ate the most vegetables had a 20% lower risk of type 2 diabetes than people who ate the least, and people who ate the most fruit or whole grains a 16% lower risk. People who drank the most sugary drinks had a 59% higher risk than people who drank the least. The study watched what people already ate, so it can't show these foods caused the differences. The details are below, with how the whole diet score was linked with type 2 diabetes, heart disease or stroke, asthma and COPD, cancer and dying, and how far to trust the numbers. Sources are linked throughout.

The foods and drinks, and type 2 diabetes

The researchers split people into thirds by how much of each food or drink they reported, and compared the top third with the bottom third (Wang, 2026).

Each food was looked at in its own analysis, and people who eat more of one food often eat more or less of others, so the table doesn't show which single change matters most.

The diabetes numbers here and below count type 2 diabetes when it was the first of the four diseases the study tracked (type 2 diabetes, cancer, heart disease or stroke, asthma or COPD) that a person got (Wang, 2026).

The short version

• People who ate the most vegetables had a 20% lower risk of type 2 diabetes than people who ate the least, and people who ate the most fruit or whole grains a 16% lower risk (Wang, 2026).
• People who drank the most sugary drinks had a 59% higher risk of type 2 diabetes than people who drank the least (Wang, 2026).
• The quarter whose diets scored best on DASH, a diet made to cut blood pressure, had a 42% lower risk of type 2 diabetes than the quarter that scored worst (Wang, 2026).
• The study watched what people already ate, so it can't show the diet caused these differences.

The study of 148,858 adults

Wang and colleagues took people in the UK Biobank who had no type 2 diabetes, cancer, heart disease or stroke, asthma or COPD when their diet was recorded. They had joined the UK Biobank at ages 40 to 69. Each recalled everything they ate and drank the day before, online, on 1 to 5 occasions, and they were followed for a median of 12.5 years (Wang, 2026).

Each diet was scored on DASH, which gives points for more vegetables, fruit, nuts and legumes, whole grains and low-fat dairy, and for less sugary drink, red and processed meat and sodium (salt). The researchers compared the quarter with the best scores against the quarter with the worst (Wang, 2026).

Compared with the worst-scoring quarter, the best-scoring quarter had (Wang, 2026):

• a 42% lower risk of type 2 diabetes
• a 33% lower risk of long-term lung disease (asthma or COPD)
• a 20% lower risk of heart disease or stroke
• a 4% higher risk of cancer, which could be chance

Each of these counts the disease when it was the first of the four a person got. Heart disease or stroke means coronary heart disease, heart failure or stroke. Cancer means cancers of the digestive organs, lungs and airways, genital organs, breast and thyroid (Wang, 2026).

The risk of type 2 diabetes fell with each step up in diet score: 21% lower in the second quarter, 34% lower in the third and 42% lower in the best quarter, each compared with the worst (Wang, 2026).

The best-scoring quarter also had a 13% lower risk of getting a first one of the four diseases, and a 25% lower risk of dying with none of the four yet. Deaths from injuries and other outside causes were left out (Wang, 2026).

How far to trust these numbers

• The study watched what people already ate. It can't show that changing a diet would lower anyone's risk by these amounts.
• People recalled what they ate on 1 to 5 days, and 39% on only one day, so a score may miss how someone usually ate. In the main analysis, leaving out people with only one day gave similar results (Wang, 2026).
• The authors adjusted for smoking, exercise, income, education and more, but say other differences between people could still explain some of the link. The links with getting a first of the four diseases and with dying before any of them were weaker in people who had never smoked (Wang, 2026).
• Everyone lived in the UK, so the numbers may not carry over to other countries (Wang, 2026).

Sources

• Wang et al. Associations of key diet-quality indexes with trajectory of multimorbidity: a prospective study in the UK Biobank. BMC Public Health, 2026.