Objective To investigate the factors which influence effects of noninvasive positive pressure ventilation(NIPPV) in treatment of acute exacerbations of chronic obstructive pulmonary disease (AECOPD) and II type respiratory failure and to generate a multiple predictive model.Methods Initially,176 AECOPD patients with II type respiratory failure treated by NIPPV were enrolled and were divided into effective group and failure group according to the curative effect, age,gender,arterial blood gas analysis,oxygenation index,forced expiratory volume in one second (FEV1),forced expiratory volume in one second (FEV1)/forced vital capacity(FVC)% in stable phase were analyzed comparatively between two groups; arterial blood gas analysis,oxygenation index,Glasgow coma score(GCS),acute physiology and chronic health evaluationⅡ(APACHEⅡ),the clinical pulmonary infection score(CPIS,sodium,creatinine,blood urea nitrogen,random blood sugar,albumin,body mass index (BMI) before NIPPV were analyzed comparatively between two groups; arterial blood gas analysis,oxygenation index,GCS after 2 h of ventilation were analyzed comparatively between two groups; arterial blood gas analysis,oxygenation index,GCS were analyzed comparatively between before and after 2h of ventilition in effective group and failure group. Multi-variable logistic regression analysis was used to find the predictive factors which influence effects of NIPPV and establish the regression equation that could predict the prognosis of noninvasive ventilation. In the second part of the study, we prospectively applied the regression equation to another 20 consecutive patients admitted to our hospital withⅡtype respiratory failure due to AECOPD, the aim is to validate its efficacy.Result There are 136 patients in effective group and 40 patients in failure group.1) In COPD stable phase:FEV1% and PaCO2 in failure group were statistically different compare to the effective group all P<0.05); age,gender,pH,oxygenation index and FEV1/FVC% in the two groups were no statistically significant different (P>0.05). 2) Before NIPPV:GCS,albumin,BMI and APACHEⅡin failure group were statistically different compare to the effective group(all P<0.05); arterial blood gas analysis,oxygenation index,CPIS,sodium,creatinine,blood urea nitrogen,random blood sugar in the two groups were no statistically significant different (P>0.05).3) After 2h of ventilation:pH,PaCO2,GCS in failure group were statistically different compare to the effctive group(all P<0.05); oxygenation index was no statistically significant different (P>0.05)4) In effective group:pH and PaCO2 after 2h of ventilation were statistically different compare to before NIPPV group(all P<0.05); oxygenation index and GCS were no statistically significant different (P>0.05).5) In failure group:pH,PaCO2,oxygenation index and GCS after 2h were no statistically significant different compare to before NIPPV group (P>0.05).Multivariate regression was used to analyze the significant variables of the univariate analysis, the possibility of success was dependent variable, the rest variables were independent variable. A backward stepwise procedure was used to construct the final model. Multi-variable logistic analysis suggests statistical significance in PaCO2 value and GCS after 2h of ventilation and elicits the regression equation:Logit(P)=4.587-0.064×(PaCO2 after 2h of ventilation)+4.351×(GCS after 2h of ventilation). Then we applied the regression equation to another 20 consecutive patients admitted to our hospital withⅡtype respiratory failure due to AECOPD for whom the same inclusion criteria were used. NIPPV successfully treated 16 patients (80%), and the model correctly classified 16 patients (80%).Conclusions1) In uninvarite analysis:statistical significance in FEV1%,PaCO2 value in stable phase and GCS,albumin,BMI,APACHEⅡbefore NIPPV and pH,PaCO2,GCS after 2h of ventilation were relative factors which influence effects of NIPPV in treatment of AECOPD andⅡtype respiratory failure. In effective group:pH and PaCO2 after 2h of ventilation were statistically different compare to before NIPPV group. In failure group:pH,PaCO2,oxygenation index and GCS after 2h of ventilation were no statistically significant different compare to before NIPPV group.2) The Logistic regression analysis:PaCO2 value and GCS after 2h of ventilation were independent risk factors which influence effects of NIPPV in treatment of AECOPD andⅡtype respiratory failure. The Logistic regression equation: Logit(P)=4.587-0.064×(PaCO2 after 2h of ventilation)+4.351×(GCS after 2h of ventilation=1)3) The predictive model can successfully applied into clinical practice, and instructed clinicians to improve the successful ratio of NIPPV. |