Revista Română de Statistică (Dec 2016)
Likelihood Estimation of the Systemic Poison-Induced Morbidity in an Adult North Eastern Romanian Population
Abstract
Purpose: Acute exposure to a systemic poison represents an important segment of medical emergencies. We aimed to estimate the likelihood of systemic poison-induced morbidity in a population admitted in a tertiary referral center from North East Romania, based on the determinant factors. Methodology: This was a prospective observational cohort study on adult poisoned patients. Demographic, clinical and laboratory characteristics were recorded in all patients. We analyzed three groups of patients, based on the associated morbidity during hospitalization. We identified significant differences between groups and predictors with significant effects on morbidity using multiple multinomial logistic regressions. ROC analysis proved that a combination of tests could improve diagnostic accuracy of poison-related morbidity. Main findings: Of the 180 patients included, aged 44.7 ± 17.2 years, 51.1% males, 49.4% had no poison-related morbidity, 28.9% developed a mild morbidity, and 21.7% had a severe morbidity, followed by death in 16 patients (8.9%). Multiple complications and deaths were recorded in patients aged 53.4 ± 17.6 years (p .001), with a lower Glasgow Coma Scale (GCS) score upon admission and a significantly higher heart rate (101 ± 32 beats/min, p .011). Routine laboratory tests were significantly higher in patients with a recorded morbidity. Multiple logistic regression analysis demonstrated that a GCS < 8, a high white blood cells count (WBC), alanine aminotransferase (ALAT), myoglobin, glycemia and brain natriuretic peptide (BNP) are strongly predictive for in-hospital severe morbidity. Originality: This is the first Romanian prospective study on adult poisoned patients, which identifies the factors responsible for in-hospital morbidity using logistic regression analyses, with resulting receiver operating characteristic (ROC) curves. Conclusion: In acute intoxication with systemic poisons, we identified several clinical and laboratory variables, such as GCS, WBC, ALAT, myoglobin, glycemia and BNP, as early predictors for in-hospital morbidity, allowing the physician to provide an accurate diagnosis and management.
