Advances in life-saving technology have led to the unintended side effect of prolongation of intensive care unit (ICU) treatment while awaiting irreversible cardiopulmonary death. In contrast, brain death has well-defined criteria and diagnostic technologies which allow for precise determinations that may reduce unnecessary care. In the final coding addendum to the International Classification of Disease Ninth Revision, Clinical Modification (ICD-9-CM), the National Center for Health Statistics created a five digit diagnostic code for brain death, effective October 1, 2011. We evaluated the association of this new code with hospital length of stay and inpatient charges in ICU patients who suffered inpatient demise using a large, publically available dataset.

We utilized the Nationwide Inpatient Sample, the largest all-payer dataset of inpatient hospitalizations in the United States, comprising a 20% stratified sample of nonfederal community hospitals, to search for occurrences of brain death coding in the last quarter of 2011 among mechanically ventilated adult patients who died during hospitalization.  Data extracted included outcomes of length of stay and total hospital charges, demographic information (age, race, gender, payer, zip-code related income quartile), primary diagnoses and major diagnostic categories on discharge, comorbid medical conditions, and hospital information (annual discharge volume, teaching status, urban/rural location, and region of country).  Among decedent ICU patients, we matched exposed (coded for brain death) records with controls with the same ICD-9 coded primary discharge diagnoses. Analyses utilized the NIS complex sample design (probability weights, stratification, and clustering). We conducted descriptive analysis of reported variables (totals, proportions or means) for differences in exposure and control. We performed univariate and multiple regression modeling for length of stay (ordinary least squares) and total charges incurred from hospitalization (generalized linear model, gamma distribution with log linkage).

A total of 1830 deceased adult inpatient ICU discharges in the weighted sample (382 unweighted) had ICD9 coding for brain death, and were matched with 24,300 (5,030 unweighted) ICU discharges with the same primary ICD-9 discharge diagnoses. Brain death codes were most often used with neurological diseases (55%) and respiratory failure (11%). Discharges using brain death coding were younger (68.2 vs. 52.8), nonwhite (46% vs. 31%) privately insured (45% vs. 25%), had fewer comorbidities (42% vs. 68% with 3 or more comorbidites), and came more often from teaching hospitals (61% vs. 52%) and hospitals with higher annual discharge volumes (30,000/yr vs. 26,000/yr).  Unadjusted mean length of stay was 5.0 days less in the brain death group (3.1 days vs. 8.1 days), with 29% less total charges ($78,000 vs. $109,000). Adjusted for demographics, comorbidities, and hospital characteristics, the difference in length of stay was 4.2 days (p<0.001) and total charges were 31% less for the brain death group (p<0.001).

Brain death coding was associated with significantly smaller hospital charges and shorter length of stay in decedent ICU patients. The introduction of specific diagnostic coding should be critically examined in associations with healthcare costs and utilization, especially as ICD-10 is implemented.