Sas 9.4 ts level 1m5

We calculated the distribution of patients for each categorical variable including age in 10-year intervals, for both the total population and by municipality type. We also calculated mean and standard deviation for age as a continuous variable. The proportion of patients who experienced at least one outpatient pneumonia or pneumonia hospitalization was determined for each municipality and for each municipality type. We excluded four small island municipalities from analysis due to few observations.
We estimated risk of pneumonia between municipality types for both outpatient pneumonia and pneumonia hospitalization with a multinomial logistic regression model with the following groups as outcome variable: 1) No pneumonia during the year, 2) at least one outpatient pneumonia (but no pneumonia hospitalization) during the year, and 3) at least one pneumonia hospitalization during the year. The ‘No pneumonia’ group was used as reference, and the model was adjusted for age, sex, educational level, co-habitation status, and CCI. We used a significance level of 5%. SAS 9.4 TS Level 1M5 (SAS, Inc., Cary, NC, USA) was used for all analyses.

The data management and the statistical analyses were performed with R version 3.6.3 R (R Core Team, version 3.6.3) and SAS 9.4 TS Level 1M5 (SAS Institute, Inc., Cary, NC, USA). The data was presented as percentages with 95% confidence intervals. Univariable and multivariable logistic regression models were created to evaluate the associations between binary dependent variables and covariates. The multiple logistic regression model included the independent variables age, sex, education level, financial situation, quality of life, place of residence, stroke, hypertension, diabetes, obesity, tobacco consumption, alcohol consumption, hypercholesterolemia, and poor nutritional status. The independent variables were chosen based on previously reported potential risk factors from the existing literature and data available in the Polsenior2 study. Highly correlated variables, such as abdominal obesity and high BMI, were excluded from the analysis. Sampling weights were included in the statistical calculations to account for the complex survey design using the R survey package. The poststratification procedure was used to match the age–sex sample distribution to the population of Poland. The level of significance was set at p < 0.05. The Variance Inflation Factor (VIF) was calculated, indicating a slight multicollinearity among the predictors. In each case, the VIF values ranged from 1 to 1.5.

Baseline characteristics were described as mean (standard deviation [SD]) for continuous variables and absolute and relative frequencies for categorical variables. Baseline characteristics were described during the baseline year or at the date of first mMRC measurement (index date) (sFigure 2). Baseline treatment was defined as respiratory medication collected during the four months prior to index.
Follow up for moderate exacerbations, severe exacerbations and death started on the day after the index date between January 1, 2008 and December 31, 2014. Follow up ended on the date of death, emigration, 36 months after index date, or end of study Dec 31, 2017, whichever occurred first. If a patient had several mMRC measurements during the study period, the first measurement was used as index. Patients who developed an OCS-related diagnosis were censored from analyses on the date of diagnosis.
Logistic regression models were applied to estimate odds ratios (OR) with 95% CIs of moderate exacerbation, severe exacerbation and death in B1 with B0 as reference for each year of follow up and adjusted for age, sex, cohabitation status, comorbidity, BMI and smoking. A cumulated measure of outcomes over the entire three year follow up was calculated with each patient being categorized according to the most severe event they experienced during follow up.
The hazard ratio of exacerbations accounting for recurrent exacerbations was estimated using a Cox proportional hazards model with 95% confidence intervals and GOLD B0 as the reference group. Recurrent events were included in the model as a covariate. Furthermore, the model was adjusted for age, sex, cohabitation status, comorbidity, BMI and smoking and death was handled as a competing event using the Fine & Gray method.20 (link)
All analyses were performed using SAS 9.4 TS Level 1m5, and all p-values <0.05 were considered statistically significant.

This was a descriptive study. For categorical variables, the proportion of patients within each group was calculated. Age was the only continuous variable in our study, and we calculated both mean and standard deviation as well as proportions in each 10-year age group (40–49 years, 50–59 years, 60–69 years, 70–79 years, and 80+ years). Short college, medium college, and masters/PhD were summed into one group (college) to account for too few observations in the groups. CCI was grouped into four categories (0, 1, 2, and ≥3). Because ICD-10 code registration in the DNPR is delayed, and we only have access to data until 2018, the last valid CCI is from 2015 [10 (link)].
We calculated the proportion of patients who had at least one outpatient pneumonia or pneumonia hospitalization within each year, and we investigated the relationship between pneumonia groups and ICS treatment dose groups. All analyses were performed using SAS 9.4 TS Level 1M5 (SAS, Inc., Cary, NC, USA).