Determinants of Tetanus, Pneumococcal and Influenza Vaccination in the Elderly

A Representative Cross-Sectional Study on Knowledge, Attitude and Practice (KAP)

Carolina J. Klett-Tammen; Gérard Krause; Linda Seefeld; Jördis J. Ott

Disclosures

BMC Public Health. 2016;16(121) 

In This Article

Methods

Study Design and Data Management

We analysed data from a German nationally representative cross-sectional survey on infection prevention, initiated by the Federal Centre for Health Education (BZgA) in 2012. Reference population for representativeness was the German national Census[20] including an error tolerance of +/−3 %. Inclusion criteria for the study were: ability to understand and speak German, age between 16 and 85 years, living in a private household in Germany. The sample was drawn via the ADM-telephone-master-sample for both, household phones and mobile phones. In each of the reached households, the household member whose birthday was last, was chosen as participant.[21] Computer-Assisted Telephone Interviews were conducted using a questionnaire that contains 101 questions on vaccination-related KAP leading to 112 variables. The full list of variables of the data set with original coding and recoding is available from Additional file 1 https://static-content.springer.com/esm/art%3A10.1186%2Fs12889-016-2784-8/MediaObjects/12889_2016_2784_MOESM1_ESM.pdf. For all variables in the data set, missing values were less than 5 % and were at random and excluded using list wise deletion. Further methodological details of the overall survey are described elsewhere.[21]

Data Analysis

Outcomes we considered were self-reported influenza-, pneumococcal-, tetanus- and any vaccination in the last 5 years. Independent variables were socio-demographic, health-, knowledge-, attitude- and practice-related variables. We conducted descriptive analysis and bivariate analyses including chi2-tests to assess associations between socio-demographic and vaccination-related KAP-variables, and self-reported vaccine uptake (crude risk ratios for dichotomous variables). Potential effect modification and/or confounding by age, sex, migration-status, place of residence, and education was addressed by Mantel-Haenszel tests. To generate odds ratios (OR) for effects of KAP-related variables on the self-reported uptake of influenza-, pneumococcal-, tetanus- and any vaccination, we applied adjusted logistic regression, and used hierarchical backward elimination of non-significant (α > 0.05) variables and interactions (if change of estimate for residual variables = <10 %) for best model fit,[22,23] separately for each vaccine and their combination. We assessed validity of final models by likelihood-ratio test and Nagelkerke's r-square.

In a second step, we composed scores of KAP. Given the large sample size (>1000 participants) and the high participant-variable ratio (1:18 at minimum),[24] explanatory principal axis factor analyses with oblique promax-rotation and Kaiser-normalization was used. We selected and aggregated most relevant individual predictors within each score according to scree-plot and the Kaiser-criterion, as available from Additional file 2 https://static-content.springer.com/esm/art%3A10.1186%2Fs12889-016-2784-8/MediaObjects/12889_2016_2784_MOESM2_ESM.pdf. Factors with loadings <0.4 were excluded and confirmatory factor analyses were conducted. We applied a Cronbach's alpha threshold of 0.70 to assure internal validity.

After dichotomization of each score by median (higher versus lower vaccination-related KAP), we conducted multiple logistic score regression analyses for associations between scores and influenza-, pneumococcal-, tetanus- and any vaccination, respectively. We compared results from score regression models to those from regression models conducted using individual variables.

Statistical analyses were done using Stata IC 12; for factor analyses we used SPSS 20.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....