There are two noteworthy limitations of this study: representativity and the possibility of underreporting of unsafe abortions.
Statistical methods are introduced to study the representativity of specimen.
Our study cannot claim representativity for Japanese health care; the more rural areas in particular were not represented.
Strengths of this study include nationwide representativity over more than 30 years, extensive follow-up time, and minimal loss to follow-up.
Delphi surveys do not strive for representativity in sampling, but for the highest possible degree of expertise [25, 26].
This was confirmed in our study (Karnofsky performance score: trend only) arguing for the representativity of the study population.
This survey has some limitations, in particular the representativity of the sample.
One of the strong points of this study is the large sample size (3,710 workers) and the representativity of the study population.
For the specific purpose of our study we additionally analysed the representativity of the networks to cover all deaths in the country.
Although the surveys in all countries attempted representative sampling, determinants as level of education, working hours and socio-economic class, that are known to affect participation in surveys, might also have an impact on the representativity of our findings.
