Preliminary Logistic Regression Analysis Of The Prognosis Of The Heat Stroke Patients

Objective To explore high risk factors which affect the prognosis of severe heatstroke patients and to establish a new prediction model.Methods30cases of severe heatstroke patients from emergency department of thefirst affiliated hospital of Nanchang university during June2008to August2011, wereretrospective analysized. Those patients were divided into death group and live groupaccording to their outcome. Clinical data between the two cases were single analyzedand muti-factor logistic analyzed. Factors studied in the research were listed asbellow: age, sex, underlying disease, body temperature (T), pulse (P), Respiratory rate(R), whether shock or not, white blood cell count (WBC), platelet count (PLT), redblood cell deposited (HCT), alanine aminotransferase (ALT), aspartateaminotransferase (AST), blood sugar (GLU), creatinine (SCR), urea nitrogen (BUN),creatine kinase (CK), creatine kinase isozymes (CK-MB), blood sodium (Na),potassium (K), prothrombin time (PT), activate part blood coagulation live enzyme(APTT), D Dimer (D-Dimer),3P test, blood potential of hydrogen (PH), partialpressure of carbon dioxide (PCO2), partial pressure of blood oxygen (PO2),hydrogen carbonate root (HCO3-), Glasgow score (GCS), Acute physiology andchronic health score (APACHEⅡ). We established forecasting model to predictsevere heatstroke patients’ death, described the receiver-operating characteristic curve（ROC curves）and worked out the area under ROC curve (AUC). We then comparedthe new established modle with GCS score and APACHE Ⅱscore in the prediction ofthe death of severe heatstroke patients. At last, we put relevant data from the cases inthis study into the new modle to further tested its diagnosis ablitity. Statisticalanalysis was accomplished through the SPSS17.0and MEDCALC9.2software.Results Single factor analysis showed that: in the studied29factors, underlyingdisease, R, PLT, AST, CK-Mb, PT, APTT, PH, HCO3-,GCS score, APACHE Ⅱscorein20patients live group was statistically significant different with that of10deathgroup(P <0.05).the other factors between the two groups was not statisticallydifferent(P>0.05). Multivariate Logistic regression analysis showed that: AST and PH is the independent risk factors which influence the severe heatstroke patientsprognosis. Logistic regression equation (i.e. prediction model) is: Y=78.821+0.007×AST-11.17×PH. Prediction model was statistical significance（P<0.001）,according to the ROC curves, the area under the ROC curves (AUC) was0.935±0.057,with95%confidence interval0.781to0.990, the truncation point was0.2684. The sensitivity and specificity of judge model in predict severe heatstrokepatients death was90%and85%respectiely. And compared with APACHE ⅡscoreGCS score, the AUC of the new diagnosis model is larger than both (P <0.05), thesensitivity of the diagnosis model is higher than GCS score and similar toAPACHE Ⅱ score, the specific degrees is higher than APACHE Ⅱscores, and aslightly lower than GCS score.Conclusion Whether with underlying disease, R, PLT, AST, CK-Mb, PT, APTT,blood PH, blood HCO3-,GCS score and APACHE Ⅱ score score bloodcan helpjudge the prognosis of patients with severe heatstroke. Among them, AST and PH areindependent risk factors related with the patients’ death. The new established forecastmodel has higher sensitivity and specificity, worth more than GCS score andAPACHE Ⅱscore. Therefore it may have the important meaning in evaluating therisk of death for severe heat stroke patients.