Table 1.
Outcomes and statistical analysis of the study
| Outcome measures | Statistical analysis | ||
|---|---|---|---|
| Outcomes | Indicators or constructs | Description | |
| RUM | Antibiotic prescription percentage | Calculated by dividing the number of antibiotic prescriptions to patients by the total number of prescriptions in a certain period of time. This value is then multiplied by 100. | Propensity Score Matching (PSM): This method reduces the influence of bias and confounding variables and reasonably assesses the intervention and control groups; |
| Injection prescription percentage | Calculated by dividing the number of injection prescriptions to patients by the total number of prescriptions in a certain period of time. This value is then multiplied by 100. | Difference-in-difference (DID): The variations in an index of the two groups before and after intervention are calculated to reflect the net effect of intervention. The differences lie in the cluster-level summaries of the two groups. | |
| Average drug cost per prescription | Calculated by dividing the total cost of all drugs prescribed by the number of prescriptions in a certain period of time. | Generalized estimating equations (GEE): For multivariate outcome analyses, we used GEE to assess the effect of PRPRI intervention with repeated measures at the individual participant level. For continuous dependent variables, we used the GEE model with a normal apply and an identity link function. For dichotomous dependent variables, we employ the GEE model with a binomial distribution and a logit link function. | |
| Prescription percentage of drugs listed in the essential drug list or formulary | Calculated by dividing the number of prescribed products listed on the essential drugs list or local formulary (or of those that are equivalent to drugs on the list) by the total number of products prescribed in a certain period of time. This value is then multiplied by 100. | ||
| Percentage of prescription with duplicate or more prescribed antibiotics | Calculated through dividing the number of patient prescriptions during the time a duplicate or more antibiotics were prescribed by the total number of prescriptions in a certain period of time, multiplied by 100. | ||
| Robustness of the transparent mechanism | Information accessibility | Accessibility score | Factor analysis: This method evaluates the reliability and validity of the questionnaire. |
| Information perception | Perception score | Structural equation modeling (SEM): This technique builds the inner link of each construct of the transparent mechanism and tests the robustness of the mechanism by applying the questionnaire data from two groups four times. | |
| Stress level | Stress score | ||
| Behavior intention | Behavior intention score | ||
| Development of the transparent mechanism | Constructs, such as accessibility, perception, and behavior intention | The code of information accessibility, perception, behavior, and other related influence factors. | Grounded theory: The qualitative data were analyzed through a thematic framework. Codes were then developed based on the viewpoints that were derived from these data. |