Patient Co-Morbidity and Functional Status Influence the Occurrence of Hospital Acquired Conditions More Strongly than Hospital Factors.

BackgroundNever events (NE) and hospital-acquired conditions (HAC) are used by Medicare/Medicaid Services to define hospital performance measures that dictate payments/penalties. Pre-op patient comorbidity may significantly influence HAC development.MethodsWe studied 8,118,615 patients from the NIS database (2002-2012) who underwent upper/lower gastrointestinal and/or hepatopancreatobiliary procedures. Multivariate analysis, using logistic regression, was used to identify HAC and NE risk factors.ResultsA total of 63,762 (0.8%) HAC events and 1645 (0.02%) NE were reported. A total of 99.9% of NE were retained foreign body. Most frequent HAC were: pressure ulcer stage III/IV (36.7%), poor glycemic control (26.9%), vascular catheter-associated infection (20.3%), and catheter-associated urinary tract infection (13.7%). Factors correlating with HAC included: open surgical approach (AOR: 1.25, P < 0.01), high-risk patients with significant comorbidity [severe loss function pre-op (AOR: 6.65, P < 0.01), diabetes with complications (AOR: 2.40, P < 0.01), paraplegia (AOR: 3.14, P < 0.01), metastatic cancer (AOR: 1.30, P < 0.01), age > 70 (AOR: 1.09, P < 0.01)], hospital factors [small vs. large (AOR: 1.07, P < 0.01), non-teaching vs teaching (AOR: 1.10, P < 0.01), private profit vs. non-profit/governmental (AOR: 1.20, P < 0.01)], severe preoperative mortality risk (AOR: 3.48, P < 0.01), and non-elective admission (AOR: 1.38, P < 0.01). HAC were associated with increased: hospitalization length (21 vs 7 days, P < 0.01), hospital charges ($164,803 vs $54,858, P < 0.01), and mortality (8 vs 3%, AOR: 1.14, P < 0.01).ConclusionHAC incidence was highest among patients with severe comorbid conditions. While small, non-teaching, and for-profit hospitals had increased HAC, the strongest HAC risks were non-modifiable patient factors (preoperative loss function, diabetes, paraplegia, advanced age, etc.). This data questions the val