Under-recording of diabetes on medical record may be associated with adverse in-hospital outcomes

Abstract

Materials and methods This is a retrospective analysis including 62136 adults (> 18 yrs.) admissions in groups NL (69%), DM (17%), DMNR (5%), and SH (9%), hospitalized between 2010 and 2013. Patients were classified according to five coded discharge diagnoses registered in electronic medical record. The inclusion criteria for DMNR were registration of DM in previous admissions and omission of diagnosis in consecutive readmission. Inclusion criteria were length of stay between 2 and 120 days, availability of capillary glucose (at least one measurement for NL patients and two measurements for DM, DMNR and SH patients). Exclusion criteria were pregnancy. In hospital outcomes included in the retrospective analysis were nosocomial infection, sepsis, intensive care unit admission and death. To compare the proportion of adverse outcomes we perform chi-square test. Results Average age was 54,8 yrs., the proportion of females was 52%. The proportion of hospitalizations with surgical interventions was 44%. The average length of stay among NL, DM, DMNR and SH, were 4,7; 7,3; 13,9 and 17,1 days respectively. The incidence of nosocomial infection were 0,4%; 0,9%; 2,3% and 3,9% and the incidence of sepsis were 1,5%; 4.0%; 13,7% and 16% in NL, DM, DMNR and SH, respectively. The mortality rate were 0,8%, 1,4%, 10,2% and 12,7%, respectively. P values were < 0.05 for every outcome described.