Background
Diabetes constitutes a major public health challenge and numerous studies suggest that the prevalence is increasing in most countries [
1]. However, reliable national estimates for the occurrence of diabetes are difficult to obtain, particularly time trends in incidence. Population-based health studies have limitations, since study samples tend to have low and therefore potentially biased participation [
2,
3]. A few countries such as Denmark, Sweden and Scotland have established national diabetes registers to monitor trends in the occurrence of diabetes [
4,
5], but few of these have complete coverage.
While most approaches to estimating the total number with diagnosed diabetes in a population are likely to have important sources of error, data from nationwide prescription drug databases can overcome the problems with biased participation and limited sample sizes to estimate gender- and age-specific trends. All diagnosed patients with Type 1 diabetes and a large majority of patients with Type 2 diabetes in most Western countries are treated with blood glucose-lowering drugs. In European children, insulin use essentially indicates a diagnosis of Type 1 diabetes [
6].
Blood glucose-lowering drugs are used for other indications than diabetes, e.g. metformin for the treatment of polycystic ovary syndrome (PCOS) and this has to be taken into account. Furthermore, diabetes not treated with blood glucose-lowering drugs and undiagnosed diabetes will not be covered but important trends can nevertheless be monitored with high quality population-based prescription databases [
7].
The primary aim of this study was to describe time trends in prevalent and incident use of blood glucose-lowering drugs by age group and gender in Norway during 2005–2011.
Discussion
We have monitored the use of blood glucose-lowering drugs in Norway, and found an increase in the overall prevalence from 2.5% in 2005 to 3.2% in 2011. Interestingly, the number of new users was, for the first time in the period, significantly lower in 2011 than the year before and this was primarily due to a reduction in incident use of OAD in the 70 years and older age groups.
The prevalence of diabetes depends on the incidence and mean duration of the disease, both of which are affected by several factors. For example, increased survival among those with the diagnosis will increase the prevalence [
11] and higher life expectancy (ageing) in the general population could influence the incidence. Others have shown that declining mortality among patients may contribute, but that increasing incidence was the main driver of the increasing prevalence over time, at least in Denmark 1992–2003 [
7] and in Scotland 1993–2004 [
5]. We did not have mortality data in our study but others have shown a secular decrease in mortality from ischaemic heart disease among patients with diabetes in Norway [
12], suggesting that this could have contributed to the observed increasing prevalence with concurrent decline in incidence.
The strength of this study is the complete coverage of the Norwegian population [
8]. NorPD should therefore provide good estimates for the number of patients with diabetes treated with blood glucose-lowering drugs (100% reimbursed), with a few caveats that are discussed below. The study does not cover diabetes that is not treated with blood glucose-lowering medication, and classification of diabetes into Type 1 or Type 2 diabetes based on available data is difficult in the older age groups. Individuals with at least one prescription of blood glucose-lowering drugs dispensed were included but the vast majority had actually had several prescriptions. In 2010, of incident users in both groups (insulins only and OAD) recorded, 88% redeemed more than one prescription during the following 12 months. Institutionalisation, emigration, switch to lifestyle treatment only, or off-label use (PCOS, obesity) could partly confound the results (see below). The high incident use of insulins in women in reproductive age might be explained by gestational diabetes.
Another limitation is lack of individual level data of institutionalised patients with diabetes. About 13% of the population above 80 years of age (approximately 30,000 people) lived in nursing homes in Norway in 2011 (figures from
Statistics Norway,
http://www.ssb.no). Thus, the prevalence and incidence in the oldest patient groups will be underestimated. For instance, recalculating for users of blood glucose-lowering drugs in 2011, the prevalence increased from 9.0% to 10.4% in the over 80 age group. However, data from the NorPD show that in 2011 only 2.4% of total DDDs (Defined Daily Doses) [
10] prescribed of blood glucose-lowering drugs were dispensed to institutions. For insulins, 3.4% of total DDDs prescribed were dispensed to institutions. This is probably too little to have any major influence on our observed time trends.
Migration of patients has not been taken into account in the study. In principle, a prevalent user of blood glucose-lowering drugs, moving from another country to Norway, may therefore have been mistakenly counted as an incident user. Again, this is likely to be a minor source of error, particularly for time trends.
The use of dispensed drugs in a prescription database as a proxy for diabetes will obviously result in an underestimation of the total number of patients with diabetes in the population, as a considerable proportion of Type 2 diabetes patients are treated with lifestyle measures alone. Studies from Norway in the period 1995–2005 report that 20-35% of prevalent patients are treated by diet only [
13‐
15]. Data from 33 general practices in two representative areas of Norway showed a decline in the proportion of patients with diabetes not treated with blood glucose-lowering drugs, from 30.7% in 1995 to 28.4% in 2005 [
14]. Data from the Swedish Diabetes Register show a decline in the proportion of patients with Type 2 diabetes treated with diet only from 25.3% in 2008 to 23.4% in 2011 [
16]. We do not have nationally representative data on the proportion of diabetes patients not treated with blood glucose-lowering medication for our study period (2005–2011) but if there is a real decline in this proportion, it is particularly interesting that the incident use of blood glucose-lowering drugs did not increase during the study period. The proportion of patients not treated may also be influenced by the number of new blood glucose-lowering drugs available on the market promoted by the pharmaceutical industry.
On the other hand, we cannot rule out the possibility that the proportion of diagnosed Type 2 diabetes patients treated with OAD has declined during the study period; for instance, as a result of patients being diagnosed earlier and increasing focus on beneficial dietary changes. Safety concerns in the elderly regarding polypharmacy and hypoglycaemic episodes could also contribute to such a trend.
The large majority of patients receiving OAD are likely to have Type 2 diabetes. On the other hand, there will be some misclassification from prescriptions for patients with PCOS, pre-diabetes and the metabolic syndrome [
17‐
19]. Some women of reproductive age who are treated with metformin are probably suffering from PCOS rather than diabetes. In 2011, the total number of women in this age group treated with metformin was 3,482 (0.54% in this age group population), and a total of 3,712 (0.58%) received oral blood glucose-lowering drugs. In European studies, the occurrence of PCOS has been reported to be 6.5-8% [
20].
Since insulins and analogues are used by both patients with Type 1 diabetes and Type 2 diabetes, the definition applied in the outcome measure “prevalent user of insulins only” is aimed at estimating a group of patients with Type 1 diabetes. Some of the defined insulins only users may have received treatment with OAD prior to the 24 month period, as suggested in our sensitivity analysis that extends the period to six years. This could be explained by switching from OAD to insulins only e.g. due to the relative contraindication of OAD in severe heart and kidney failure. Thus, the estimated prevalence of Type 1 diabetes in the population in this study may be overestimated to some extent in the elderly.
There was a stable incidence of use of insulins only. Recent data on childhood-onset Type 1 diabetes from the Norwegian Childhood Diabetes Registry show little or no increase in the past few years, which is consistent with the current results [
6]. In the only previous Norwegian study of Type 1 diabetes incidence in individuals over 15 years of age from Norway, Joner & Søvik estimated an incidence of 17/100,000 person years in the age group 15–29 years during 1978–1982 [
21]. The current study showed 32/100,000 person years across all age groups. In the age groups over 30–40 years, diabetes classification becomes more challenging. It is nevertheless clear that a substantial proportion of incident Type 1 diabetes cases arise in adulthood [
22].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
HS conceived the study and drafted and edited the manuscript. KIB, HSc, TJB, AKJ, KM, CB and LCS developed the design, interpreted the results and revised the manuscript. RS participated in the design of the study and performed the statistical analysis. All authors reviewed the manuscript for intellectual content and approved the final manuscript.