Purpose: It is well known that the average length of stay in psychiatric care beds in Japan is the longest among OECD countries. In order to examine the historical and institutional backgrounds of this long length of stay in Japanese psychiatric care beds, we analyzed the relationship between the length of stay and the type of payment for admission in 1956 using historical survey records. Our primary goal was to investigate whether the length of stay paid for by public assistance was longer than that paid for by social insurance after controlling for possible confounding factors.
 Data: We used the patient records of "Report of the Statistical Survey of Mentally Disordered Inpatients in 1956", which was conducted by the Ministry of Welfare. We randomly sampled 1,000 inpatients from the original records and then selected inpatients who had been admitted for the first time, resulting in a sample of 808 patients with schizophrenia, bipolar disorder, or epilepsy. We also analyzed the survey records of discharged patients for robustness.
 Method: Multiple regression analysis was carried out to investigate how the length of stay differed by the type of payment for psychiatric hospitalization. We regressed length of hospital stay on the dummy variables of payment types, controlling for possible confounding factors such as gender, age, disease status, nursing situation, type of guardian (s), marriage, duration before hospitalization, and hospital fixed effects.
 Results: When controlling for all of the observed covariates, the length of stay paid for by public assistance was, on average, approximately 9 months longer than the length of stay paid for by social insurance for patients with schizophrenia, approximately 10 months longer for patients with bipolar disorder, and approximately 17 months longer for patients with epilepsy. In addition, although standard errors were often large, the positive correlation between length of stay and payment by public assistance was slightly stronger when the guardians of patients were parents, spouses, or siblings.
 Discussion: In 1956, only six years after the current Public Assistance Act was enforced, the length of stay in psychiatric care beds paid for by public assistance was slightly longer than that paid for by social insurance, even after controlling for the patient background and hospital fixed effects. This longer length of stay paid for by public assistance may be explained by familial factors of welfare recipients, such as poorer economic conditions and lower caring capacity, and institutional factors, such as free admission through public assistance and the so-called "(family) consent admission" system, which was often used in cases of psychiatric admission with public assistance. The long length of stay in psychiatric care beds paid for by public assistance has persisted since the period we examined and has become an essential characteristic of post-war psychiatric hospitalization in Japan.
 <Authors' abstract>