The main findings of this study was that anticoagulants were superior in reducing mortality among individuals with diabetes and AF below 80 years of age, while antiplatelet therapy seemed equally effective as anticoagulants among individuals 80 years and above. An unexpected finding was that mortality was reduced in women who were prescribed digitalis, but not in men. In addition, mortality was increased in men prescribed loop diuretics. Furthermore, we found that mortality was reduced in patients aged below 80 years prescribed statins, and also among those below 80 years prescribed beta blockers.
One might have expected that all prescribed pharmacotherapies would decrease mortality among the population in this study. However, the benefits of a drug could be offset by side effects, as in the case of disopyramide, quinidine and sotalol, which are known to be associated with increased mortality . The studied subgroup of subjects, i.e. AF patients with diabetes, could show a risk pattern different to what could be expected in AF patients in general.
An interesting finding was the equally positive effects of antiplatelets and anticoagulants on mortality among individuals ≥80 years of age, making findings from a Cochrane review accountable also for patients with diabetes and AF . Although anticoagulants are superior to antiplatelets in stroke prevention of AF patients , the current study indicates that among elderly individuals antiplatelets could be useful for those who do not tolerate anticoagulants because of equal effects on all-cause mortality.
Prescription of digitalis was associated with decreased mortality only among women. It seems, according to a review, that beneficial effects of digitalis are obtained at low digoxin concentrations, while high serum levels were associated with increased mortality . Positive effects of digitalis in AF and CHF seem to be related to attenuation of sympathetic activation and neurohumoral alterations. Diabetes is associated with an impaired myocardial energy production affecting myocyte contraction and diastolic function , leading to diastolic heart dysfunction, and little is known about the optimal treatment of this condition . The female myocardium could possibly benefit more from the effects of digitalis than the male. Another explanation could be that the gender differences might reflect different prescription patterns in men and women with AF. The previously reported higher mortality among men < 80 years  could be due to more pronounced CHF , and consequently higher digitalis doses.
The finding of an association between prescription of loop diuretics with an increased mortality among men could be explained by concomitant presence of CHF, as CHF in AF patients do increase mortality . Prescribed loop diuretics may be a marker of more advanced CHF, but this is contradicted by the fact that other possible markers of this, i.e. digitalis , and aldosterone antagonists , were not associated with significantly increased mortality. Another possibility is that loop diuretics could have increased arrhythmia risks due to potassium depletion [32, 33].
The low mortality associated with prescribed beta blockers was expected, considering the effect in reducing mortality in myocardial infarction with AF , and also in CHF [35, 36].
The reduced mortality associated with prescription of statins is in line with evidence of positive effect of statins on diabetic patients . The mortality reduction associated with statins was more pronounced compared to a meta-analysis of randomized trials , as is commonly the case in observational studies using propensity scores . Statins have been shown to have anti-arrhythmic properties , which could be relevant for patients with AF. However, it cannot be excluded that the positive finding for survival could be due to confounding by indication, i.e., severely ill patients not being prescribed statins to the same extent.
Crude mortality was higher in women than in men, in agreement with others’ findings . In an earlier Swedish study, we showed that men with AF had a relative mortality risk of 1.3 and women with AF a relative mortality risk of 1.9 when they were compared to men and women in the general population . The women in this study were on average five years older than the men (mean age 76.3 vs. 71.7), and could consequently be expected to have 11% higher mortality rate than men (data from Statistics Sweden). However, we found a 46% higher mortality rate in women compared to men, which could be explained by the excess mortality risk in women due to both diabetes , as well as to AF , or the co-morbidity.
A limitation of this study was the small sample size, which prevented us from detecting small differences, even though the analyses included almost 9,000 person-years at risk. Since this was an observational study of the association between prescription of drugs and mortality, the findings may have been subject to confounding by indication [41, 42]. Furthermore, drugs prescribed by other caregivers were not included in the patient records, which may have weakened the associations between prescription of certain medications and mortality in this study. We did not have access to doses of the prescribed drugs. Severity of CHF and CHD were not classified in the patient records. Moreover, AF could not be classified as paroxysmal, persistent or permanent and heart rhythm could not be classified as sinus rhythm or fibrillation rhythm. We had not access to data on renal disease, and besides not of non-cardiovascular diseases or pharmacotherapies. Thus, we cannot exclude the possibility that residual confounding may explain at least partly some of the results.
A major strength of this study was that we were able to link clinical data from individual EPRs to data from national demographic and socioeconomic registers with less than 1% of information missing. In addition, an earlier study of the diagnoses recorded in the EPRs showed that less than 2% of all diagnoses per individual were missing . The use of a propensity score with covariates relevant to mortality as the outcome was an additional strength of this study as it enabled adjustment with a low risk of over-fitting regression models . Besides, we used Laplace regression to confirm results of Cox regression in the entire sample.
In conclusion, we found significantly positive effects of both antiplatelets and anticoagulants on survival among individuals aged ≥80 years, which could imply that antiplatelets despite little effect in stroke prevention could be beneficial on survival in elderly diabetes patients with AF when anticoagulants cannot be prescribed. We also found gender differences, indicating a positive effect by digitalis on survival among women, and a negative effect by loop diuretics among men, both needing further attention. The decreased mortality risk associated with prescription of statins underlines the positive effect of statins in diabetes patients with AF, and treatment by statins could be as important as antithrombotics in these patients. More studies focusing on diabetes patients with AF are needed to confirm or reject our findings.