Is passenger rail in Japan at least an order of magnitude safer than in the U.S. (as measured by fatalities per passenger-km)?

Question

I've seen several statements online praising the safety of passenger rail in Japan and China compared to the US, implying that rail in the US is at least 10x less safe. For example, pedestrianobservations.com claimed in 2011 that the fatality rate in Japan was "one death per 51.4 billion passenger-km", while

US: 27.26 billion passenger-km/year (both Amtrak and commuter rail), 159 deaths over 20 years. Note the rate is more than twice that of China per capita, let alone per rail passenger. This is one death per 3.4 billion passenger-km.

But is it really true that passenger rail is an order of magnitude less safe in the US? To be specific, we can

• Focus on Japan (to avoid potential concerns about the quality of reporting on on China's rail safety).
• Focus on one specific metric: R = fatalities per billion passenger kilometers.
• Include the number of all fatalities (e.g. level crossing users) in passenger rail accidents, not just the number of passenger fatalities.
• On the other hand, the question is specifically about passenger trains, so freight train safety is explicitly out of scope (and freight train fatalities are not included in the numerator of R).

Is it true that this metric is at least 10 times higher in the U.S. than in Japan?

Clarification regarding the reference "Comparative Rail Safety"

Comparative Rail Safety blog post by Alon Levy shows explicit numbers about the fatalities from Wikipedia's List of rail accidents and List of countries by rail usage. The question is not about whether the author correctly collected the data from these Wikipedia articles, but about whether the overall conclusion about comparative safety of the US passenger rail is correct. There can be multiple reasons to doubt that this is the correct data to make such conclusion:

• List of rail accidents may only list notable accidents and, in particular, may not list the majority of pedestrian fatalities caused by a contact with passenger train. It also may omit other less notable accidents.
• List of rail accidents is an English-language Wikipedia page, and its coverage of accidents in non-English-speaking countries may be less comprehensive than that of the U.S. (i.e., it may list a larger proportion of accidents in the U.S. than in Japan).

---

Here are some references and information I found to fact check this statement myself.

1. According to Wikipedia, Japan had 446.7B passenger-km in 2019 while the US had 32.5B passenger-km. Wikipedia refers to the "Transport - Passenger transport" table in OECD Data. Looking at the OECD data directly, it seems that the number of annual passenger-km in Japan is about 13.4..15.3 times larger than that in the US. E.g. if I use the table "Passenger transport" (quarterly) for the last 4 quarters for which the data is available for both countries (2022'Q4..2023'Q3), I get a ratio of 378 / 24.8 = 15.2. It should be noted, however, that the table comes with a warning "Although there are clear definitions for all the terms used in this survey, caution is required when comparing data between countries." Also, they state that the unit is "passenger-km" while it appears to be "million passenger-km".
2. JR East's "Comparison of Safety Between Japan and EU Railways" states that in 2013 there were 274 total fatalities (when discussing railway operational accidents in Japan). It also states "The annual average number of fatalities in accidents related to railways in the latest five years is 311 persons, likely evoking a social demand on Japanese railways for further improvement of safety." perhaps referring to the period 2009--2013 given that it was published in 2014.
3. The Journalist's Resource "Transportation safety over time: Cars, planes, trains, walking, cycling" states that in 2000-2009 the number of average annual fatalities in the US were 38 for railroads 25 for rail transit.

These numbers do not appear to support the stated Japan rail safety claim. For example, combining the data from #1 and #2 we may conclude that Japan had one death per 1.49B passenger-km in 2013 (R = 1 / 1.49 = 0.67 where 1.49 = 408.7 / 274), while combining #1 and #3 we may conclude that the US had R = 6.9 or 4.1 depending on whether "rail transit fatalities" are included or excluded from the numerator. However, it is not clear to me whether this is the correct computation. In particular, (1) OECD passenger-km metric for the US jumps by a factor of 3 between 2013'Q4 and 2014'Q1, which seems suspicious; (2) it is not clear whether OECD US passenger-km includes rail transit or not, hence not clear whether it should be included in the numerator; (3) OECD data has explicit warning for being careful when comparing it between countries; (4) both fatality data points are more than 10 years old, and come from difference periods (one might hope that safety improves over time in both countries).

Answer 1

Using 2014 data the fatality rate for US long-haul rail service is 0.43 per BPMT (1 death per 2.32B miles traveled); excluding non-passenger deaths brings that to 0.15 per BPMT (1 passenger death per 6.66B miles traveled).

Japan, meanwhile, has the Shinkansen aka 'Bullet Train' which claims a 50-year, no fatalities service record. That, by itself, gets Japan's numbers well beyond the US since the Shinkansen routinely logs north of 70 billion passenger km traveled per year. The remainder of their train service would have to be a meat grinder for the US stats to have any hope of catching up.

Keep in mind, the Shinkansen is a source of national pride for Japan, and it's safety record is a matter of prestige worldwide. The US rail system, meanwhile, is more vestigial than anything, and struggles to receive support against the backdrop of the Federal Interstate System.

So the comparison is akin to some guy you interviewed at a bar about his weightlifting routine and then comparing it to an Olympic athlete.