We found a decline in the excess overall mortality rate for non-insulin-treated people with diabetes, especially with regard to CVD, compared to the people without diabetes between the periods from 1998–2002 to 2003–2007. However, the excess overall mortality rate showed no change for those with insulin-treated diabetes. Furthermore, a statistically significant increase in excess mortality from cancer was observed.
According to this study the all-cause mortality among people with diabetes decreased in almost all age groups from 1998 to 2007 as it has decreased among the general Finnish population . Similar development has also been reported in other countries [11, 12]. One explanation might be that the population with diabetes increases more rapidly due to more effective screening of type 2 diabetes, which was emphasized in the national diabetes prevention program in 2000–2010 in Finland. That might have brought people earlier to diagnosis and to better care. Although mortality among people with diabetes decreased, a significant excess mortality was still found. Some previous studies have also showed that the rates in all-cause mortality, in CHD mortality and in deaths from cancer are higher among people with diabetes than among people without diabetes [14–20].
To our knowledge, the present study is the first to examine the changes and associations in excess mortality from specific causes using the whole population together with all insulin- or non-insulin-treated individuals with diabetes. Tierney et al.  assessed changes in the excess mortality rate over time in the US for those with diabetes compared to those without diabetes using specific causes of deaths. They found that the excess mortality among men and women with diabetes decreased from 1992–1998 to 1999–2003 especially with regard to CVDs . In spite of different study periods, these findings were consistent with our results concerning persons with non-insulin-treated diabetes who normally form the main part of the population with diabetes.
Insulin-treated people with diabetes
The excess mortality rate for those with insulin-treated diabetes compared with those without diabetes remained stable except for increasing excess mortality from neoplasms.
Contrary to non-insulin-treated people with diabetes, the excess mortality associated with CVD did not decrease among insulin-treated people. In 2003–2007, for example the excess mortality from CHD was almost eightfold for women and almost fivefold for men with insulin-treated diabetes compared to people without diabetes. Our study is in line with earlier results describing the major impact of diabetes in increasing the risk of CHD mortality [13–16, 18]. We found that excess CHD mortality was particularly high among insulin-treated people with diabetes. One reason for higher excess mortality among insulin-treated compared to non-insulin-treated individuals might be the longer duration of the disease. A previous study concerning insulin-dependent people with diabetes in Finland reported that the relative mortality, compared with the general population, was more strongly influenced by duration of diabetes than by age . Also smoking prevalence might be one reason for higher CHD mortality among people with diabetes. A Finnish study conducted in the early 1990’s, discovered that young adults with diabetes smoked more than non-diabetic individuals .
Among insulin-treated people with diabetes, mortality from neoplasms increased almost in all age groups between the study periods. At the same time, mortality from neoplasms decreased among people without diabetes and showed also a decreasing trend among people with non-insulin treated diabetes. The increase in deaths from cancer was more obvious among insulin-treated women, among whom mortality increased especially with regard to cancer of digestive organs, breast and urinary tract, and lymphoid tissue and blood. Many studies have indicated that diabetes is associated with an increased risk of death from malignancies [19, 20, 27, 28]. According to Barone and colleagues  there are several potential explanations for this. First higher insulin levels may contribute to increased tumor growth [29, 30]. Another reason can be that cancer patients with diabetes might be treated less aggressively than those without diabetes . Third, patients with diabetes may have poorer response to cancer treatment, including increased infection risk and intraoperative mortality . However, it is difficult to explain the most obvious increase in cancer mortality among women with insulin-treated diabetes found in this study. Changes in treatment practices might have influenced in these worse outcomes. For example, in Sweden during 2006 and 2007, women using insulin glargine alone had an increased incidence risk of breast cancer as compared with women using types of insulin other than insulin glargine . The same finding was also made in a Scottish study . Further investigation is needed concerning the increased cancer mortality among insulin-treated diabetic individuals.
Non-insulin-treated people with diabetes
The excess mortality rate for persons with non-insulin-treated diabetes compared to those without diabetes appears to be clearly diminishing. The most obvious decrease found in excess mortality was that of CVD mortality. This may mainly have been due to improved preventive and acute treatment practices reported in an earlier study by the study group . The proportion of people with diabetes using cholesterol lowering and antihypertensive drugs increased between 1997 and 2007 in Finland . Another study compared the use of secondary preventive medication among newly diagnosed CHD patients with and without diabetes between 1997 and 2002 . It was found that the use of a new class of antihypertensive drugs, angiotensin II receptor antagonists and ACE inhibitors was more common among people with diabetes compared to people without diabetes. The use of lipid lowering medication and β-blockers was almost similar in both groups. Also there has been found improvements in the proportion of people with type 2 diabetes achieving glycemic target levels in Finland . Similar development has also been reported in a US study concerning the proportion of people with diagnosed diabetes achieving glycemic and LDL targets .
Our data were based on large national administrative datasets, namely hospital discharge and cause of death registers, and reimbursement for medication and prescription data derived from the Finnish Social Insurance Institution registries. The reliability of Finnish health registers has been evaluated good . The reliability of the Finnish Cause of death registers is also considered good, due to fairly high overall autopsy rate . Between 1996 and 2007 a medical or medico-legal autopsy was performed in about 30% of deaths .
A major strength of our study was the use of comprehensive national registers to follow the mortality for all treated individuals with diabetes in Finland over two five-year time period. In addition, we were able to compare different treatment groups of people with diabetes. Some methodological limitations still exist. Individuals treated with diet only were classified as non-diabetic individuals, if diabetes was not reported in any of the registers used. However, those treated with diet only are mainly among people aged over 65 . The different treatment groups were not identified from the medical records but from medication and hospital discharge registers. However, those recorded insulin-treated have for the most type 1 diabetes and those with non-insulin treated type 2 diabetes. Despite these limitations this material is unique. Our data encompass the whole population of a country; all diabetic individuals receiving medication have been identified, and all cause of deaths in 1998–2007 have been registered and analyzed according to sex and diabetes treatment groups.