The evidence suggesting coffee is good for you is of the most unreliable sort.
While reports from cohort studies suggest coffee (caffeinated coffee, unless otherwise specified) may have health benefits, especially in terms of diabetes risk, cohort studies are a weak and unreliable form of evidence.
Stronger evidence, from laboratory experiments on human subjects and from randomized-clinical trials (RCTs), suggest that coffee consumption is either irrelevant for, or potenitally harmful to, human health.
Laboratory experiments on human subjects find that caffeine ingestion reduces insulin sensitivity (meaning it increases insulin resistance). When subjects consume caffeine, the insulin released in their body has a harder time doing its job: clearing glucose from their blood, and depositing that glucose (in the form of glycogen) into muscle and liver tissue. With caffeine ingestion, blood sugar levels stay higher, and insulin levels are higher (and more glucose gets packaged into triglyceride and stored in fat tissue).
These lab experiments tend to be brief so their applicability to regular coffee use is unclear.
But they should make you nervous. They directly contradict the cohort studies – increased insulin resistance takes you a step closer to diabetes, not a step further away.*
How can this be? Cohort studies chronically confuse correlation for causation. Epidemiologists who assert that coffee reduces diabetes risk either ignore, or haven’t thought of, the possibility that both increased likelihood of coffee consumption and reduced type II diabetes risk might be caused by a third factor. Like being affluent, which might also correlate with eating better food, or being more education, or having better access to healthcare, or displaying a greater willingness to follow medical advice…
That’s why cohort studies generally do so badly when tested in randomized clinical trials, and this trend continues with coffee. One recent RCT divided participants into test groups – some drinking coffee, some not – and found that insulin levels were higher (that’s bad) among coffee drinkers over the 4-week trial. It is worth noting that participants drank a serious amount of coffee – about the equivalent of 3-4 normal Starbucks drip coffees per day.
Another, longer RCT found that insulin levels essentially stayed normal – and blood glucose disposal improved marginally – for coffee drinkers consuming very modest amounts of coffee for 16 weeks.
Might the improved glucose disposal support the idea that drinking normal amounts of coffee protects against diabetes?
Well, no. The participants in this study did NOT consume normal amounts of coffee.
Participants in this longer study consumed the equivalent (in terms of caffeine) of one normal sized Starbucks drip coffee per day. In the real world, coffee drinkers consume an average of 2.5 cups of coffee per day.
These randomized clinical trials remind us that we should be skeptical of cohort studies, especially those suggesting coffee prevents diabetes.
And you should probably cut back on the coffee.