# Effects of population age structure on estimates of parenteral antimicrobial use

### Study data

This retrospective, population-based study was conducted using the National Database of Health Insurance Claims and Specific Health Examinations in Japan (NDB). NDB contains nationwide medical data from the Japanese health insurance system since April 200924,25. The NDB contains data on all patients enrolled in any health insurance policy, but does not include data from public assistance recipients (about 1.7% of Japan’s population)21. The age structure data of the national population from 2013 to 2019 were obtained from the Statistics Bureau, Ministry of Internal Affairs and Communications of Japan.13. Population projections from 2020 to 2030 were obtained from the National Institute of Population and Social Security Research14, which provides three separate population projections computed according to low, medium, and high birth and death rates. In this study, we used a population projection based on median values ​​(differences in these three projections up to 2030 were marginal)14.

From the NDB, we extracted parenteral AMU data from 2013 to 2018. We chose to analyze data from 2013 as the Japanese National Action Plan for Antimicrobial Resistance was developed based on AMU data from 20134. We did not include data from 2019 or later due to the occurrence of events that could affect parenteral AMU in Japan, such as cefazolin deficiency and the COVID-19 pandemic.26, 27. Parenteral antimicrobials were identified using the WHO Anatomical Therapeutic Chemical Code J017. To analyze age-specific temporal trends in parenteral AMU, we divided the population into the following age groups: children (<15 years), persons of working age (15–64 years), and the elderly (65 years).

### analysis

First, we describe temporal trends of parenteral AMU in Japan between 2013 and 2018 using DIDs (as an indicator of population-adjusted AMU) and DDDs (as an indicator of total unadjusted antimicrobial consumption). These measures were calculated for the total population as well as for each age group. We also calculated annual rates of change in DID, DDD, and the national population as proportions compared to 2013 values. Linear regression analyzes were performed to calculate 95% confidence intervals and s Trend values ​​in AMU as time in years from 2013. Statistical significance was set at s <0.05.

To illustrate the effects of population shift on DDDs, we simulated DDD trends from 2019 through 2030 based on population projections. DDD predictors of parenteral AMU were modeled for the rate of specific DID (in 2018) in each age group. In these models, DID was applied as a covariate that reflects the effectiveness of antimicrobial stewardship interventions. DDDs were calculated using the following equations:

$$DDDs \left (t \right) = {\sum }_{i}DD{Ds}_{i} (1) And$$DD{Ds}_{i}

(2)

where \({DDDs}_{i}