| Objective:To evaluate effectiveness of pneumonia vaccination by providing free pneumococcal vaccination for the COPD patients. Outpatient visits, times of hospitalization, number of total treatment, days of hospital stay, average length of stay, EQ-5D index score and EQ-VAS index score were collected before and after the pneumococcal vaccination。Different statistical models were used to analyze the data to provide evidence-based basis for advancing pneumonia vaccine into immunization programs or incorporating health care. Moreover, analytical strategy of the different statistical models for such data provided the reference on methodology for multilevel data in public health.Methods:Before-and-after quasi-experimental design was employed to analyze and compare the effectiveness of the intervention and health status caused by the AECOPD before and after the vaccination. Univariate analysis, multiple linear regression, Poisson regression, multilevel Poisson regression, variance component model, interrupted time series analysis and multilevel ITS model were used to carry out statistical analyses. Data management was implemented using Excel 2007, statistical analyses were performed by SAS 9.4 for Windows. The statistical significant level was set at the level of 0.05.Results:Univariate analysis of the prevention of pneumonia vaccine on AECOPD onset, number of the hospitalized was reduced by 7.03-11.81 days per patient and number of the hospitalized per patient per time was decreased by 5.16-7.59 days compared to the pre-intervention, the difference was statistically significant. Different regions demonstrated different results for the EQ-5D health index score, the EQ-5D health index of combined regions increased by 0.02 points, EQ-VAS score increased by 1.20 points. The multivariate analysis showed that pneumonia vaccine was effectiveness by reducing 0.108 days of clinic visiting and treatment per person per time (P=0.030) after adjusting for age, sex, income, occupation, region, and vaccination. The number of clinic visiting and treatment was increased with the increase of the income. Pneumococcal vaccination could considerably reduce days of stay in hospital and average days of stay in hospital. Health index score and EQ-VAS score before and after the vaccination had no significant effect on the patients with COPD. Since the clustering effects existed among regions and the difference between regions was statistically significant, multilevel data structure was considered by using multilevel statistical models in the data analysis, number of clinic treatment and hospitalization was significantly reduced after the vaccination.With respect to the ITS analysis, the number of clinic visiting and treatment, number of hospitalization, and number of clinic treatment and hospitalization presented different results in different regions. The length of hospital stay, the number of total treatments and the average length of stay in Wenxian County and Jiaonan were significantly decreased. In order to further consider multilevel structure of the data the multilevel ITS analysis suggested monthly outpatient visits, the number of hospitalization per month, the total number of treatment per month and the average days of hospitalization were significantly reduced.Conclusions:Pneumococcal vaccination could effectively reduce times of the hospitalization, times of the total treatment, as well as the length of hospital stay. The recommendation is that the PPV23 vaccination for the COPD patients should include in the new rural co-operative medical system or immunization programs. For the before-and-after quasi-experiment design multilevel statistical models should be used to deal with hierarchical data after adjusting for the effect of regions for the program evaluation in public health;,moreover, multilevel interrupted time series analysis is a powerful statistical method to analyze time-series data to illustrate the effectiveness of the intervention. |
PDF Full Text Request