There are several reasons why this type of study should not frighten you. Don’t get me wrong, I’m not doubting the study itself or its methodology. But this type of observational epidemiological study should never be translated to any given individual. Here are the key reasons why: First, the study examines associations between a biomarker (FPG) measured at one time point (baseline) and the outcome (all-cause death). Association NEVER equates with causation. The observed associations may provide for hypothesis generation, which must then be tested in suitably designed intervention studies.

Second, the outcome parameter (dependent variable) is all-cause death which includes several ways of “kicking the bucket” that have nothing to do with glucose metabolism. What interests us in the (pre)diabetes context are the diseases directly linked to chronically elevated glucose levels resulting from impaired glucose metabolism, such as retinopathy, neurological and cardiovascular diseases, which may or may not be fatal.

Third, the one-time snapshot of FPG measurement does not give us any indication of the progression from baseline throughout the observation period. Think about it this way: are you with your long-term stable FPG truly represented by the comparison age group that inevitably contains people whose FPG values have changed over time? I don’t think so.

Fourth, the statistical adjustment for various other biomarkers (age, smoking status, BMI etc.) is necessarily incomplete. In this case, it includes self-reported PA levels, but nothing about dietary habits. Even if the latter were included, they are wildly inaccurate, as we know from nutrition science in general.

Then there is the reporting of hazard ratios, that is a special kind of relative risk ratio. Researchers like to report those, because these ratios tell us whether the observed associations are statistically significant or not. What interests you as an individual is your absolute risk and absolute risk increase or decrease that might come from increasing or lowering your FPG levels.

You can extract those from the supplementary data of this study. I have done a quick calculation based on the data presented in supplementary table S5. In your age group, the absolute risk of death per person-year was 1% for those in the reference range (94-99 mg/dL), the absolute risk was 1.1% for those in the range from 105-109 mg/dL. Over ten years, that translates to an absolute risk reduction of 1%, which means 100 people would have to reduce their FPG levels by 10 mg/dL to prevent one single death (always provided that all those deaths are causally linked to FPG, which is doubtful. This minor difference is unadjusted, which, therefore, is an overestimate.

In your case of essentially green-territory FPG, you would be far better off monitoring the vascular functional biomarkers (such as PWV and blood pressure) that more directly and causally correlate with death. You’ll find more on that subject in earlier posts of mine.