Hospital Readmission — effect of condition at discharge

What aremr-vital-signse the causes of hospital readmission?  A previous post details the most significant factor:  fragmentation of health care.  This post focuses on disease and the condition of the patient at discharge.

Until the past 5 years the prevailing view of re-hospitalization was it is unavoidable due to the nature of the chronic disease.  The view changed when studies showed strong outpatient social and medical management vastly reduce re-admissions.

And yet, disease does catch up with those afflicted causing eventual death.  Sometimes patients are discharged from a hospital and much to the surprise of clinicians the patient is back in the hospital before outpatient care can engage to deal with a potentially unstable situation.  Even before the outpatient check 4 weeks after discharge the person decompensates and is hospitalized again.

Frustration born out of rapid re-hospitalization lead Michael Rothman to develop a statistical method to alert clinicians before hospital discharge to the presence of high risk.   Whether the outpatient care network is sufficiently robust to act on the information is possibly the bigger issue.  But, consideration of the “Rothman Index” is worth a few comments.

Mr. Rothman published the statistical findings “Development and validation of a continuous measure of patient condition using the Electronic Medical Record”  (Journal of Biomedical Informatics 46 (2013) 837–848).

Rothman found he could predict bad outcome (and hospital re-admission) based on routine measurements done in every hospital, just combined in a statistical way.  He divided the measurements into 3 equally weighted groups of (1) lab tests, (2) nursing observations and (3) vital signs.  Overall there were 26 items such as potassium level, nursing charting by exception (like normal or abnormal respiration), and blood pressure.  The index started at 100 and if all items were normal the index stayed at 100.  But, if items were abnormal an amount was subtracted.  The lower the score the worse the situation.  In fact, he showed his index correlated with the 1-year mortality of the patient.  The lower the index at discharge the more likely was re-hospitalization and even death. Mr Rothman started a company to calculate the index for interested hospitals.

The idea is great:  identify high risk patients and focus more outpatient resources quickly.

Doctors always try to normalize abnormal findings.  If the heart rate is too high find out why and correct it.  If the potassium is too low find out why and correct it.  If the blood pressure is too low find out why and correct it.  When the end of the hospitalization comes hopefully everything is in the normal range.  The Rothman Index basically says that if a concerted effort by doctors fails to normalize findings it means the patient will do poorly (an abnormality caused by a stroke just can’t be normalized).  The very definition of chronic illness is that it can not be resolved by modern medicine.

Rothman’s research shows some interesting findings.  The 1-year predicted mortality is increased by 10% if any the following findings are present at discharge:

Clinical item absent or below present or above
 Respiratory rate  14  20
 Heart rate  40  90
 Temperature  96.9  100.2
 Oximetry  96%  –
 Systolic BP  100  190
 Diastolic BP  50  105
 Sodium  138  145
 Potassium  2.9  4.7
 Creatinine  0.4  1.5
 Chloride  98  102
 Hemoglobin  10  17
 Blood Urea Nitrogen  –  25
 White Blood Count  4,000  14.000
 Heart Rhythm    anything other than
sinus rhythm
 Braden Scale (link)  19  
 Nursing Assessment
excluding pain
   any body system


The Rothman Index is important and either that index or other similar index should be calculated at discharge.  If the index indicates an increase in mortality then questions need to be asked and answered:

  1. Have the abnormal findings been investigated and treatment started — if an available treatment has not been started it should be.
  2. Has enough time passed for the abnormal finding to normalize — if not the patient should either stay in the hospital or be seen as an outpatient in just a few days.
  3. The estimate of a poor prognosis should be discussed with the patient and family to make sure they understand why follow up is important.
  4. If the estimate of poor prognosis is very high (over 50% 1-year mortality) and not expected to improve then planning for death should be started.

, , , , , ,

  1. #1 by David Bittleman on May 12, 2014 - 12:56 PM

    Excellent review of the Rothman Index!

%d bloggers like this: