Study design
We performed a cross-sectional study using an online questionnaire.
Setting and participants
Participants were physicians who were members of the Japan Primary Care Association (JPCA). The JPCA is an academic organization which comprises many Japanese primary care physicians and has around 10,470 members (as of February 2019). We recruited 2,000 physicians using random sampling, who were at least three years postgraduate. The first and second years after graduation from medical school fall under the category of initial postgraduate clinical training [12], and were thus excluded.
In Japan, primary care services are provided by both clinics and hospitals [13]. Also, the boundary between primary care and specialty areas is ambiguous and the census reports that approximately 32% of physicians work in clinics and 64% work in hospitals among all physicians [14]. Therefore, the respondents included physicians working in hospitals.
In addition, there are two important characteristics of the healthcare system in Japan, which are universal health insurance and free access [13]. All residents of Japan are required by law to be enrolled in a health insurance program, and co‐payment rates range from 10 to 30%. Patients are free to choose any healthcare facility, regardless of the severity of their disease and their insurance status.
Study period
Data were collected from February 10 to March 10 in 2021.
Data collection
Participants answered the questionnaire on the Web using Microsoft Forms. We provided URLs via email and sent a total of three reminder emails over the period of data collection. The first 300 respondents received an Amazon gift certificate worth JPY 2,000 (approximately USD 17.4).
Primary outcome: having experience in treating their family members or relatives
The following question was used to explore participants’ experience in treating family members. “Looking back over the past few years, have you ever provided medical care (including medical examinations, prescriptions, procedures, etc.) to your family members or relatives at least once?”.
Based on past studies and reports, we defined “family members or relatives” to include parents, children, siblings, grandparents, grandchildren, spouses, and spouses’ first-and second-degree relatives [2, 7, 15].
Explanatory variable: field of practice (clinic or hospital)
Participants chose their fields of practice from the following: clinics without admission, clinics with admission (1–19 beds), small hospitals (less than 200 beds), medium hospitals (200–499 beds), large hospitals (500 beds or more), and others. In Japan, a hospital is defined as a medical institution with 20 beds or more, whereas a clinic is defined as having less than 20 beds or no beds. In addition, there is a difference in the medical fees among hospitals with more than 200 beds and those with less than 200 beds. Therefore, this classification was used in a previous study conducted in Japan [12].
Covariates
Based on the previous studies, we included covariates for age, gender, geographic location, and whether or not physicians had relatives who were doctors [7, 10, 15]. These covariates were evaluated as categorical variables through a self-administered questionnaire.
Types of treatment, reasons and factors behind the treatment, and physicians’ feelings during the treatment
We also collected information on the type of medical treatment families and relatives received, the reasons and factors behind the treatment, and their feelings (satisfaction or hesitation) during treatment. Some of the questions were: “What were your reasons for providing medical care to your family or relatives?;” “Have you felt satisfied/hesitant while treating family or relatives?” For the latter question, we asked participants to choose from “Often,” “Sometimes,” “Rarely,” and “Never.” These questions and alternatives were based on past studies [7, 15].
Statistical analysis
Nominal data were expressed as percentages, whereas medians with interquartile ranges were calculated for continuous variables. Responses were excluded if there were missing or apparent inconsistencies in them.
We conducted a chi-square test with field of practice (binary variable of hospital or clinic) as the independent variable, and the percentage of doctors who had treated family members or relatives as the outcome. In addition, we conducted a logistic regression analysis with the following factors: age, gender, physician’s geographic location, and presence of a doctor in the family. Age was classified into three groups: < 45 [7], 45–64, ≥ 65 years, and included in the model as a categorical variable.
As four variables, that is, age, gender, physician’s geographic location, and presence of a doctor in the family were included in the final model, and the percentage of “having experience in treating family members or relatives” was estimated to be 80% from previous studies, we calculated that we would require at least 250 responses for this study. All statistical analyses were conducted using StataCorp software (Stata Statistical Software, Release 15, College Station, TX, StataCorp LLC, 2017).