hospital readmissions and LOS articles (5 articles in last month)

there have been an array of articles recently on hospital readmissions and length-of-stays, one in the annals and several in the latest JAMA.

1. the question has come up that by our attempts to decrease hospital length-of-stay, are we creating more hospital re-admissions and perhaps increased mortality? (seehospital LOS and readmission rate annals 2013 in dropbox).  big study in the VA system of 129 hospitals from 1997-2010 with >4M admissions, looking at 2 chronic diseases (chf, copd) and 3 acute dz (acute MI, comm-acq pneumonia, GI hemorrhage). results:

    --LOS decreased from 5.44 to 3.98 days overall

    --LOS also decreased for the 5 target admissions above, especially for acute MI (dec 2.85 days) and pneumonia (2.22 days)

    --during this 14 year period, the 30-day readmission rate decreaesd from 16.5% to 13.8%, including in these 5 target diagnoses (greatest reduction in MI from 22.6% to 19.8% and COPD from 17.9% to 14.6%).

    --all-cause mortality also decreased 3% annually.

    --however, those hospitals with the lowest LOS did in fact have an increase in readmission rate at 6% increase for each day lower than the mean.

so, a few issues

--therapies have changed over this time period, which could explain both the reduction in LOS and readmission rates.  this is observational study only

--this is only the VA system, so may not be generalizable

--but, it is somewhat reassuring that mortality and readmission rates have not increased. 

 

2. medicare database of 1.3M admissions for heart failure, 550K for acute MI, 1.2M for pneumonia with readmission rates within 30 days in 24.8%, 19.9% amd 18.3% respectively (see hospital readmissions for chf MI pna jama 2013 in dropbox). of those readmissions, 35.2%, 10% and 22.4% respectively were for the same condition as the original admission, and 60% of each were within 15 days of discharge. (median time interval 12 days for chf. 10 days for MI and 12 days for pneumonia. no signif diff by age, sex or race.

 

comment. not really clear why 30days is the cutpoint to assess (seems to be totally arbitrary). also, not clear to me that readmission is necessarily a bad thing. many patients are in and out of the hospital at that age/with those co-morbidities. these conditions are by nature at high risk of decline, esp for heart failure. not clear that keeping them in the hospital longer is better. oftentimes, they can be discharged earlier with close followup, though this does not always work (and staying in the hospital may be bad: increased confusion, deconditioning, falls, exposure to bad bugs, etc. )  meds for heart failure often take a while to titrate to optimal dose and take awhile to work (eg ace-i, b-blockers). as i mention at the end, i think the key to decreasing hospitalizations is  better coordinated care, with an emphasis on the outpatient setting.

 

3. review of discharges in 5M adults from 3 huge areas (in california, florida and nebraska). 18% had an acute care encounter within 30 days, of these 40% went to the ER only, 60% admitted. 1/3 within 7 days of discharge. (see hospital ED and readmits post discharge jama 2013 in dropbox). raises issue that we need to work to decrease ER visits post-discharge.

 

4. for pedi pts, looked at >500K admission in 72 childrens hospitals. (see hospital readmissions pedi jama 2013 in dropbox).  30-day unplanned readmission rate of 6.5%, but varied dramatically from hosp to hosp (30% higher in those with high vs low readmission rates). and if look at the top 10 dx's, variance of 17-66% higher in high vs low hosps (though some of this is hosp-dependent because of different demography -- eg sickle cell readmission rates varied a lot -- and, no doubt, varied widely by socioeconomic differences of the communities). but this is likely an area for QI interventions, perhaps as with adults -- see below

 

5. the last JAMA article looked at the efficacy of various QI interventions to reduce rehospitalization postdischarge  (see hospital readmissions and QI initiatives jama 2013 in dropbox).  they looked at 14 intrervention communities as compared to 50 comparison communities from before (2006-8) and during (2009-2010) the intervention implementation.  intervention communities had 20-90K medicare beneficiaries each. there was a small but significant decrease in rehospitalizations overall in the intervention communities (in nonintervention ones, mean hosp rate of 82.09/1000 pts in 2006-8 decreasing to 79.48/1000, and in the intervention communities from 82.27 down to 77.54. mean community-wide rehospitaliztions in nonintervention inc from 18.97% to 18.91% in the different time periods, and dec 18.97% to 18.91% in the intervention ones, nonsignificant). these interventions were for up to 3 years, were different interventions (so unclear validity in adding them all together), were supposed to be hospital/community partnerships (the RED intervention at bmc was one of them -- which was mostly sending the discharge note and med list to us in primary care , which were helpful but did not really address the bigger issue of the need for a fully coherent intpatient/outpatient coordinated plan.

 

our experience with the sco model (coordinated, integrated community-based care in the medicare senior care option) is as follows: we have significantly decreased hospital readmission rate (and, i think, significantly improved the health and quality of life of the participants) in our senior care program at the health center, by following our own patients in the hospital, discharging them as quickly as appropriate to the appropriate setting (mostly home), ensuring that there is medication reconciliation between the in-hospital and out-patient meds (and, in several cases, limiting the aggressive changing of medications in-hospital, also likely decreasing adverse events), and, perhaps foremost, arranging either outpatient appointments right away and/or aggressive nursing care at home (often 3x/week initially, occasionally daily for some patients) and home-maker/social support services. one weak link is if the patient goes back to the ER, where it seems that several of the BMC attendings pretty much automatically re-admit the patient (it would be much better to call us and arrange the appropriate disposition, since more-often-than-not, in my experience, the readmission was unnecessary).  my understanding (and adam burrows should feel free to comment) is that the results of the very tightly coordinated care at our PACE sites is even better for keeping patients out of the hospital (ie, out of the emergency room) and decreasing readmissions.