Atrial fibrillation, CKD, and bad outcomes

Patients with atrial fibrillation and chronic kidney disease (CKD) may not benefit from anticoagulants (see afib CKD outcomes BMJ2018 in dropbox, or doi.org/10.1136/bmj.k342 ). this follows on the tails of a recent blog finding no benefit and perhaps harm in those getting ICDs for heart failure but have CKD (see below)

Details:

--large database from the Royal College of General Practitioners (2.73 million patients from 110 general practices across England and Wales), evaluated over the 11 year period of 2006-2017

--6977 patients aged >65yo, with chronic kidney disease (eGFR <50)and newly diagnosed atrial fibrillation were identified

    ​--2434 on anticoagulants within 60 days of developing atrial fib were compared to 4543 not anticoagulated (ie, only  35% of patients in the overall cohort were put on anticoagulation), followed for a median of 506 days.

    --anticoagulation was by vitamin K antagonists (1739 patients, 72%), rivaroxaban (307 pts, 13%), apixaban (261 pts 11%), dabigatran (69 pts, 3%), heparin (44 pts 2%).

    --mean number of days til first script for anticoagulant: 18

--only  35% of patients in the overall cohort were put on anticoagulation

--prior to matching, those on anticoagulants tended to be younger (81.7 vs 83.2 yo), female (54.7%) and not current smokers (8.6% vs 12.1%), and less likely to have had a prior CNS bleed (0.7% vs 1.6%) or GI hemorrhage (2.8% vs 5.1%).  more CAD in anticoag group (14% vs 10%) along with meds for that

--the mean CHA₂DS₂-VASc score was 4.2 for each group, and eGFR was similar at 38. 

--they attempted to match the 2 groups by:

    --choosing patients on anticoagulation who had it started within 60 days of the onset of afib

    --matching them with patients who had new onset afib, not put on anticoagulants, but making sure to incorporate the date they received the diagnosis of afib plus the time from when their matched counterpart had the initial diagnosis and then started anticoagulation

    --and they did propensity score matching, using an array of demographic variables (age, sex, year of afib diagnosis, and a validated socioeconomic instrument incorporating household income, education, healthcare provision, and living environment per patient postcode), an array of clinical variables (smoking, eGFR) and baseline medical comorbidities (MI, CAD, heart failure, diabetes, htn, history of stroke/TIA, prior CNS or GI bleed, PAD), and array of meds (mostly to treat these comorbidities, but not including aspirin, but of note aspirin was used more in those who were anticoagulated: 16% vs 12%)

        --in the propensity score matched cohort of 2424 pairs, there were no differences in baseline characteristics between the groups

Results:

--309 ischemic strokes (6.4%)

    ​--crude rate for ischemic stroke (on anticoag vs not): 4.6 vs 1.5 per 100 person-yrs

    --per regression model analysis: hazard ratio 2.60 (2.00-3.38), ie. 2.6 times the rate in those anticoagulated (the curves separated after about 6 months, then slowly splayed apart over 10 years)

--79 GI or CNS hemorrhages (1.6%)

    ​--crude rate for hemorrhage (on anticoag vs not): 1.2 vs 0.4 per 100 person-yrs

    ​--per regression model analysis: hazard ratio 2.42 (1.44-4.05), ie. 2.42 times the rate in those anticoagulated (the curves separated after about 2 years, then pretty dramatically splayed apart over 10 years)

--1410 all-cause fatalities (29.1%)

    ​--per regression model analysis: hazard ratio 0.82 (0.74-0.91), ie. 18% lower rate in those anticoagulated (the curves separated a little from 6 months until 4 years, then merged until 10 years)

Commentary:

--atrial fib is extremely common (33.5 million people >55yo around the world), and accounts for 1% of UK's Natl Health Service budget and $22 billion in the US annually

--CKD is also common, affecting 10-15% of adults globally. its incidence also increases with age, affecting 40% of US adults >60yo,. And the intersection of the CKD and AF is particularly common: 20% of those with CKD not on dialysis and 27% with ESRD, perhaps related to common risk factors

--part of the risk/benefit analysis is that patients with CKD who were anticoagulated are at higher risk for stroke and hemorrhage, as the eGFR decreases (in part from increasing platelet dysfunction with worsening renal function/uremia)

--2 retrospective Canadian studies of patients with AF and CKD have found conflicting results on the benefits of anticoagulation

--?reason why decreased all-cause fatality rate in the anticoag group? perhaps there were unassessed differences between the groups. after all, only 35% of the overall cohort were put on anticoagulation.  were they healthier in ways not covered in the propensity matching? or was there a diagnosis shift: lower rate of fatal strokes with anticoagulation but more nonfatal ones? and/or protection by anticoagulation from MIs or other vascular events?  No data presented on nonfatal nonstroke cardiovascular events

--and, such a large data-mining study is bound to have potential problems: were the ischemic strokes accurately diagnosed? were those on vitamin K antagonists in the appropriate therapeutic range?

--would be interesting to know if anticoagulation by non-vitamin K antagonists yielded similar results to the above.

--another issue raised by the above study is that 16% of the anticoagulated patients were also on aspirin. See http://gmodestmedblogs.blogspot.com/2014/11/aspirin-plus-warfarin-for-afib-and-cad.html  for an argument that aspirin may not be necessary and could significantly increase the risk of bleeding.

so, this is a bit like the recent blog finding that patients with heart failure with reduced ejection fraction but also had CKD did not have benefit from implantable cardioverter defibrillators (see http://gmodestmedblogs.blogspot.com/2018/02/icds-plus-ckd-no-benefit-but-more.html ). Again, this raises the issue that when we have patients with several comorbidities (much more often than not, as the patients get older), some of these comorbidities (chronic kidney disease, in the above) may counteract the potential benefit of an intervention.  And, the studies showing benefit may well exclude those with CKD or other important comorbidities. As i write in the ICD blog: 

    --"Unfortunately, the results of these studies are typically generalized: most clinicians are unaware of the details of the exact exclusion criteria of the study (or forget them, given the onslaught of medical studies, guidelines, etc we get daily), but the headline the “drug xxx helps in patients with yyy” is what is remembered, reinforced, and used in clinical practice (much to the delight of the drug makers).This article brings to light that there may well be very real limitations to the generalizability of their results: clinical studies have exclusions to make it easier for the researchers to interpret the findings in smaller studies (not so many variables), which basically undercuts the real utility of the results in actual patient care, which may lead to the results being inappropriately generalized."

--so, how should this negative finding on anticoagulation be put into practice???  I do have a few very elderly patients (mid-90s) on anticoagulation for afib but also with CKD. I had discussed the potential risks and benefits of anticoagulation​ with them but highlighted the potentially devastating effect of a stroke. I am strongly considering stopping the anticoagulation based on the above, especially the combination of perhaps increased stroke risk but also that hemorrhagic complications become increasingly common over time.