Study design and data source
We retrospectively analyzed the administrative claims database of patients diagnosed with acute infectious diarrhea between January 1, 2013, and December 31, 2018, using a database from the Japan Medical Data Center (JMDC) Co., Ltd. (Tokyo, Japan). The JMDC Claims Database is a nationwide electronic database of completely anonymized individual records extracted from insurance societies throughout Japan. Its population comprises employees and their family members; as of June 2018, its cumulative population size is about 5.8 million [18]. Because of the nature of the database population, persons of working age predominate and few data are available on persons older than 65 years. This database contains patients’ sex, year and month of birth, date of diagnosis, diagnosis codes, prescriptions, medical procedures, laboratory tests (results are not included) and other medical services, and inpatient or outpatient status. Diagnosis codes are recorded as International Classification of Diseases, 10th Revision (ICD-10) codes. Prescribed medications are recorded using the codes of the standard Anatomic Therapeutic Chemical Classification System (ATC codes). In addition, the route of drug administration is coded separately as oral, injectable, topical, dental, and others. In this study, we extracted only the code for oral administration.
Study population and definition of acute infectious diarrhea
Eligible patients were selected according to the ICD-10 code-identified diagnosis. Acute infectious diarrhea was defined in this study as any diarrhea resulting from any kind of microorganism and food. We chose the ICD-10 codes corresponding to the diagnosis of intestinal infection based on Clinical Classification Software Refined because these codes are regarded to be more clinically meaningful [19]. The ICD-10 codes selected for acute infectious diarrhea were categorized into A00, A01, A02.0, A02.9, A03, A04, A05, A18.3, A21.3, A22.2, and T62.9 as clinically suspected bacterial acute infectious diarrhea and A06.0, A06.2, A06.9, A07, A08, A09, and B37.8 as nonbacterial acute infectious diarrhea, meaning that the use of antimicrobials would not be appropriate (see Additional file 1: Table S1). If a single visit was associated with several of the above ICD-10 codes for acute infectious diarrhea diagnosis and had both bacterial and nonbacterial codes, the patient was regarded as having bacterial infection and was categorized in the appropriate use group.
Inclusion and exclusion criteria
Patients’ visits due to acute infectious diarrhea were identified by the above ICD-10 codes (Fig. 1). To avoid repeated visits for a single episode of acute infectious diarrhea, we determined if the same codes had been used in the previous 30 days; if so, we included only the first one. We used a time limit of 30 days to exclude repeat visits for follow-up consultations [18]. When examining the prescription of oral antimicrobials for acute infectious diarrhea, we limited the visits with diagnosis to only those for acute infectious diarrhea alone as defined in this study in order to exclude the prescription of antimicrobials for other diseases.
We excluded patients older than 65 years old because the JMDC database has few data on this population, as well as patients who were hospitalized because our aim was to identify visits in primary care. The study population was categorized into groups aged 0–17 years old (children) and 18–65 years old (adults). The age for the former group was defined based on the Convention on the Rights of the Child by the United Nations [20] and the pediatric terminology proposed by the Eunice Kennedy Shriver National Institute of Child Health and Human Development [6, 21].
To evaluate the use of antimicrobials in otherwise healthy patients for who an antimicrobial prescription was not justified at the first visit, we stratified patients by medical background—otherwise healthy, chronic bowel disease, and immunocompromised conditions—and excluded the latter two backgrounds in addition to those patients with multiple disease diagnoses. The definitions of these three backgrounds are shown in Additional file 1: Table S2.
Epidemiological evaluation of outpatients aged 0–65 years with acute infectious diarrhea
We extracted the number of eligible visits for acute infectious diarrhea by year and month, by age group, and by sex within the 2013 to 2018 period. The rate of the number of eligible visits/number of JMDC registrants as of October of each year from 2013 to 2018 was defined as the visit rate and used in the analysis.
To assess the seasonality of acute infectious diarrhea, we plotted the visit rate as a monthly trend. In this study, we defined March–May as spring, June–August as summer, September–November as autumn, and December–February as winter.
Evaluation of antimicrobial prescriptions to otherwise healthy outpatients aged 0–65 years with acute infectious diarrhea only
In only otherwise healthy individuals, bacterial and nonbacterial acute infectious diarrhea were evaluated separately with respect to antimicrobial prescription rates. The antimicrobial prescription rate was calculated and evaluated as the number of visits with antimicrobial prescriptions divided by the number of visits targeted for the evaluation of antimicrobial prescriptions. The categories of selected antimicrobials and monthly trends in antimicrobial prescription rates were analyzed specifically for nonbacterial acute infectious diarrhea.
Information on antimicrobials was extracted by ATC codes J01, A07AA, and P01AB, based on the definition of antimicrobials by the WHO surveillance program [22]. Data were collected at ATC level 5 and were arbitrarily classified into seven categories as penicillins (J01CA, J01CE, and J01CR), cephalosporins (J01DB, J01DC, and J01DD), macrolides (J01FA), fluoroquinolones (J01MA), fosfomycin (J01XX), metronidazole (J01XD and P01AB; in the relevant data, J01XX was only fosfomycin and J01XD and P01AB were only metronidazole), and others (A07AA, J01AA, J01BA, J01DH, J01DI, J01ED, J01EE, J01FF, and J01MB), based on ATC levels 4 and 5 (see Additional file 1: Table S3).
Statistical analysis
We evaluated the median and interquartile range for continuous variables. We also calculated numbers and percentages for the categorical variables.
The visit rate and antimicrobial prescription rate as defined in this study were calculated. The visit rate was calculated by dividing the number of eligible visits by the total number of JMDC registrants as of October of each year from 2013 to 2018. The number of eligible visits was defined as the “epidemiological evaluation” (Fig. 1). The antimicrobial prescription rate was calculated using the results of the “antimicrobial prescription evaluation” (Fig. 1) by dividing the number of visits with antimicrobial prescriptions by the number of visits targeted for the evaluation of antimicrobial prescriptions.