After the last flurry of posts I realized my deadline for a “Year in Review” on heart failure for the Cardiopulmonary PT Journal (CPTJ) was coming up. I had to bunker down and turn the 36 papers selected to be included from 2014 into a review - providing a broad overview of what each paper presents, and commentary with the assistance of my co-author Konrad Dias, PT, DPT, CCS. We expect to have this finalized and submitted by Friday. If accepted for publication it will appear in the June issue of the CPTJ.
The article reviews outcome prediction models for patients with chronic HF, particularly models that include hospital admission as an outcome, and more specifically three recently published models that attempt to predict hospital “readmission.” A rather hot topic right now. I found the most interesting article one by Bakal et al (2014) which is freely available if you click here. They identify in a rather large sample of patients with chronic HF (40,677) with a 5 year follow up that hospital admissions occur at increasing frequency. The “the unadjusted time between hospitalizations was reduced ~40% between each successive hospitalization. After adjustment each additional hospitalization was associated with a 28 day (95% CI: 22-35) reduction in time spent out of hospital.”(Bakal et al, 2014). I would be shocked if this does not resonate with the clinical impression of physical therapists working in facilities that patients with chronic HF frequent (hospitals, rehabilitation center, skilled nursing facilities, home care). It provides a new dynamic to the term “frequent flyers”, to “increasingly frequent, frequent flyers.” The major take home message from this and other prediction models (2 others directly and about 150 others through 2 systematic reviews) is that prediction models do not work well yet, and that the best predictor is whether someone has been in the hospital before! So being admitted to the hospital a lot is a predictor of being readmitted to the hospital! While circular - possibly useful. In the paper we argue - along with the authors of several of the prediction models - that social, economic and environmental factors (we propose unmet ADL need) are likely underlying this phenomenon (interacting with the disease indicator variables).
Of course - during the article it is obvious that many of the studies reviewed cannot make a claim to have uncovered a truly causal association :)