Findings

Overall statement or summary

They have retrospectively analysed an 11-year data base from the England and Wales General Practitioners Research and Surveillance Centre. The authors compare the rates of ischaemic strokes, gastrointestinal or cerebral hemorrhages, and all-cause deaths among 2424 newly diagnosed AF patients started on anticoagulation versus those of 2424 newly diagnosed AF patients on no anticoagulation. The pairs were matched by propensity score. Over a median follow-up of 1.4 years, the rates of stroke and haemorrhage were significantly increased, whereas mortality was significantly lower, among anticoagulated patients. The authors conclude, that anticoagulating older people with AF and coexisting CKD is perhaps associated with an increased risk of stroke and hemorrhage but a lowered risk of all-cause mortality. They call for adequately powered randomized trials in this patient-population (older pts with AF and CKD) to provide more clarity on best clinical management

Strengths

The study is the largest of its kind to report on individual ischaemic, bleeding and fatal outcomes. Within the limitations of a retrospective observational analysis, it appears well conducted and is clearly presented

Weaknesses

Within the overall-cohort, only 35% of pts with new AF were anticoagulated. It would be of interest and importance to learn more about reasons for this notablelow number. - Additionally, separate analyzes on gastrointestinal and cerebral hemorrhages should be provided to enable better interpretation of results. - What was the reason to define exposure to anticoagulation within 60 days? This time period appears quite wide to me and it could be speculated, that events occurred before starting anticoagulation, hence influencing results. - Including also pts who received heparin compounds is not helpful in my point of view and these patients should be deleted; in particular, also because no information on the degree of therapy/medication is provided (i.e.: full dose for anticoagulation or low dose regimes?). - While TIA is often falsely diagnosed, I would recommend to include only definite ischemic stroke as an endpoint. Moreover, more information on the definition and procedure to diagnose the chosen endpoints has to be provided and information on the severity of these event would be desirable. - What was the definition for cerebral hemorrhage? Is it about intracranial or intracererbral hematomas? Were traumatic bleeds included? Were subdural/subarachnoid hemorrhages included or were only intraparenchymal hematomas included?

Major points

1. This is a very interesting study and the topic is of high clinical relevance for daily clinical practice
2. The pairs were matched by propensity score.
3. Large study which addresses many important points regarding ischemic stroke

Minor points

1. It should be made more clearly that adherence data were not available and the consequence regarding data interpretation should be made more clearly to readers.
2. It would be of interest to present results also for the different substances used.
3. Statistical review should be performed