Lifestyle counselling was performed either in group sessions or individually. Local resources had an effect on the number and frequency of intervention visits. However, individual visits were planned according to individuals’ needs, and the following topics and items were included: weight control, meal frequency and quality, for example, use of dietary fibre, fat and salt, physical activity, smoking, alcohol use and diabetes as a disease in general. Group sessions were mainly weight maintenance or exercise groups and lectures concerning diabetes and lifestyle changes.^{15 17}

After identification individuals at high risk for T2D were referred to health check-ups. They were asked to fill in a questionnaire on background health and lifestyle information before the health check-up. Local nurses in primary healthcare measured weight (to the nearest 0.1 kg), height (to the nearest cm), systolic and diastolic blood pressure (BP) (mm Hg) and gave brief motivation-building and counselling for lifestyle changes,¹⁷ based on the principles of the Finnish Diabetes Prevention Study (DPS).²⁰ Body mass index (BMI) was calculated by weight divided by height (kg/m²).¹⁵ Weight loss during follow-up was dichotomised by using median to ≤0.80 or >0.80 kg for both genders. Oral glucose tolerance test (OGTT) was carried out in the morning after overnight fast with a glucose load of 75 g. Fasting and 2 h samples were collected from venous or capillary blood samples. At the baseline 80% and at the follow-up 85% were plasma venous samples. Corresponding figures for capillary samples were 20% and 15%, respectively.¹⁵ Glucose tolerance status was classified by using World Health Organization (WHO) 1999 criteria.²¹ Enzymatic methods were performed locally to determine plasma lipids and lipoproteins (mmol/l).¹⁵

The use of lipid-lowering medication, antihypertensive and coronary artery disease medication was verified at the health check-up at baseline from prescriptions if possible. Individuals whose lipid-lowering medication had been initiated 120 days before the first visit or before the midway of the 1-year follow-up (no later than 270 days after the first visit) were regarded as users at the baseline. Statins are used almost exclusively for the treatment of dyslipidaemias in Finland.²² In 2010, 93 035 individuals had reimbursement of drug costs under the Special Refund Category for dyslipidaemia associated with chronic coronary artery disease for HMG-CoA reductase inhibitors, while the corresponding numbers for fibrates and bile acid sequestrates were only 573 and 104, respectively.²³ Thus, we had a good reason to assume that lipid-lowering medication included statins only. If an individual used antihypertensive or coronary artery disease medication or both at the baseline he/she was classified as a user. CVD (heart failure, angina pectoris, coronary artery disease, myocardial infarction, heart bypass surgery and angioplasty) were asked by the questionnaire. If an individual had one of these diseases or abnormalities, he/she was considered as having CVD.

In total, 10 149 (3379 men and 6770 women) high-risk individuals aged 18–87 years were contacted between 17 January 2004 and 28 August 2007; 8353 of them had OGTT. Of them 5523 individuals participated in the follow-up visits, and 1-year follow-up information (defined by visits between 17 January 2005 and 12 June 2008) was available for 3880 individuals (1339 men and 2541 women). Of these individuals, 444 had screen-detected T2D, while 638 individuals did not participate in the OGTT at the baseline and were excluded from this study. Thus 2798 individuals participated in this study.¹⁵ There were altogether 484 (17.3%) statin users at baseline among them. Twenty-six of them (5.4%) had started statin use no later than 270 days after the first visit.

Figure 1 presents the changes in fasting and 2 h glucose in statin users and non-users according to weight loss. In individuals whose weight loss was ≤0.8 kg, fasting glucose increased more in statin users compared to non-users (p=0.055 after adjustment for age, sex, weight, CVD, medication and fasting glucose at the baseline). In those with weight loss of >0.8 kg, the reduction in fasting glucose was not statistically significantly different between statin users and non-users (p=0.330 after adjustment, figure 1).

As far as we know, this is the first study examining the association of lifestyle intervention on the risk of T2D according to the use of statins. This question is of utmost clinical importance since we now know that T2D is preventable by lifestyle changes.^10–14 A slight increase in fasting glucose during the follow-up in statin users in our study is in line with the findings from a prospective follow-up study comprising both diabetic and non-diabetic individuals without any lifestyle intervention.²⁴ In addition, another study also showed slight increase in fasting glucose in diabetic patients.²⁵ In our study fasting glucose increased by 0.08 mmol/l in statin users, whereas there was a slight decrease in 2 h glucose values in both groups after 1-year of follow-up. An increase in fasting glucose in statin users suggests deterioration in insulin secretion capacity in statin users.²⁶ Besides its putative effect on insulin secretion statins may also interfere with insulin sensitivity.²⁷ Nevertheless, in this study, 2 h glucose values, reflecting insulin sensitivity, were similarly decreased in statin users and non-users. It is not known whether statins have any acute effects on fasting glucose level, and we do not know whether people in our study had taken statin tablets in the morning or evening. From the clinical point of view, we want to emphasise even more intensive lifestyle interventions,^{14 28} for high-risk patients using statins in order to prevent T2D. We also highlight that the role of lifestyle in the prevention of T2D in statin users should be investigated in larger intervention studies with longer follow-up.

The strength of our study is that glucose tolerance status was based on the OGTT,²¹ and that IGT, that is, post-challenge hyperglycaemia, was strongly related to the development of T2D. Note that this information is missing from all previous studies that have investigated the relation between statins and T2D risk. In addition, the advantage of our study was that we were able to examine the changes in various cardio-metabolic risk factors according to the use of statins. The main shortcomings were that a high number of individuals who were screened at baseline did not participate in the intervention programme,¹⁵ and thus the present results can be generalised only to a subgroup of high-risk individuals who also may present highly motivated individuals. Furthermore, the number of statin users was quite small and the use of statins was self-reported and not based on health records, which is why we did not have detailed information on the agents and doses used. We are also unable to say how long these individuals had used statins. We also conducted analyses on the use of statins and their association to changes in glucose metabolism and T2D risk excluding those 26 individuals who started to use statins no later than 270 days after the first visit. These results were very similar (data not shown) to findings presented here. Due to the nature of our data we cannot elicit a causal relationship. In addition, our study is likely to be underpowered to demonstrate any statistical difference in T2D risk according to the use of statins due to the short follow-up period. There may also be other factors (eg, variations in how the interventions were implemented in the centres) that may have an effect on the associations of statins and fasting glucose which we were not able to take into account in our study.