Safety won’t improve with incomplete, inaccurate data

Sometimes I wonder: “Are FRA and NTSB ‘friends’ on Facebook? Can they, do they, like each other’s tweets?” I only wonder that when I’m feeling particularly cynical and jaded, like the New Yorker I’ve become in the past 40 years.

Back story: One day, a not-totally unsympathetic Chicago police officer says to young man: “You know, with a mouth like yours, you’d be much happier in New York City, and so would the Chicago Police Department.” Wham-o! Epiphany. Blinding light of satori.* And I’m on my way to a career, to success.

Here, I’m a member of the co-op board, a pillar of the community. Anywhere else, I’d be confined to a halfway house and wearing an electronic ankle bracelet. “Don’t stray too far, Dave. You know how painful those shocks can be.”

So what am I worried about, other than my own mortality? Come on, you know—if you’ve been a railroad operating officer, then you know that we worry about everything. We worry when the phone rings. We worry when the phone doesn’t ring, like maybe there’s been massive communications disruption, and they can’t call. We worry when FRA announces a new regulation. We worry when there is no regulation to establish a minimum acceptable standard. Sweat the details? Of course we do. Operating a safe railroad is nothing but detail, detail, detail.

Which is why we worry when those with our best interests at heart, like the NTSB, get the details wrong.

NTSB produces, annually give or take, a Most Wanted List (MWL) of the safety programs, technologies and practices it deems most critical to improve safe operations across the various modes of transportation.

The chairman of the NTSB, Robert Sumwalt, has provided his thoughts on the circumstances triggering, and the applications of, these most wanted elements for safe train operations.

Who could argue with NTSB producing an MWL of safety improvements? Not even I could argue with that. After all, that’s why NTSB exists—not simply to investigate accidents, but to distill from the causes uncovered by that investigation the practical, material improvements that reduce the likelihood of repeating those accidents.

One of the items on the MWL is for railroads to adopt programs and procedures that will “eliminate distractions” around those responsible for safe train operations. I’m all for that. We’re all for that.

However, Chairman Sumwalt relates this item to the collision in Chester, Pa., of an Amtrak train with a backhoe causing the death of two maintenance-of-way employees. Says Sumwalt: “… like many accidents we’ve investigated, distraction played a role. When the accident occurred, the dispatcher was speaking to his spouse on a landline.”

That sounds bad, but the fact is the conversation on the telephone had absolutely nothing to do with the cause of the accident in Chester, because the train dispatcher’s operating decisions and actions were all in compliance with operating rules and procedures, and had nothing to do with the cause of the accident. The accident was caused by the actions of employees located on the tracks at Chester, by the removal of working limits protection when the track was still obstructed by equipment.

Why target the train dispatcher’s use of the telephone as a contributing factor? Beats the hell out of me.

A train dispatcher oversees operations of multiple trains over multiple miles of track, and at any moment may not be focused on a specific area. Suppose Train 123 at MP 321 makes an inquiry or has a problem. The dispatcher is not going to be focusing on an area where m/w has reported itself clear and allowed restoration of the track for normal service. Is Train 123 now a “distraction”? Of course not.

Can there be personal phone calls that might constitute a distraction? Of course, and there can be company phone calls that constitute a distraction. Railroads are supposed to teach their train dispatchers to exercise professional responsibility, and to terminate conversations that might pose a distraction. And that—professional responsibility, the acceptance of and fidelity to professional responsibility—is the most important factor in the execution of safe train operations.

NTSB in this instance has made the issue of distraction a distraction itself from the cause of the accident and the mitigations required to prevent its recurrence.

Focused on the Chester accident, Chairman Sumwalt identifies another of the MWL items: end drug and alcohol impairment in transportation. No argument with that, right? Right, but what about this from chairman Sumwalt: “The Chester accident also illustrated the fact that drug use by rail workers has been on the rise in recent years, playing a part in seven accidents in the past three years, and nine accidents in the past decade, compared to only one accident in the prior decade.”

Really? Sources please!

If Chairman Sumwalt is referring to the railroad accidents NTSB investigated, then we have to be very clear that NTSB investigates very few of the total number of FRA-reportable accidents. It’s quite possible that in the accidents investigated by NTSB, there were greater numbers of individuals impaired by drug and/or alcohol use while at the same time the rate of drug and alcohol use by all rail workers has declined.

Using accidents as a sample representative of the entire industry is, to speak plainly, nonsense, as the accident is the exception, the non-ordinary event on the railroad. If we want to use accidents to try and detect a trend, then at the very least we would have to look at the trend for all accidents over the time periods, and the trend for human factor accidents in those same time periods.

So, for the 2008-2017 period, the overall frequency of reportable accidents declined by 28%, while the absolute number of human factor (which presumably includes drug and alcohol impairment) accidents declined 33%. The rate of human factor accidents as a percentage of total train accidents remained constant at 37%, indicating what a tough nut we have to crack with human factor accidents—a nut we won’t crack until we have medical fitness-for-duty standards and properly deal with fatigue and obstructive sleep apnea.

For the previous decade, 1999-2008, the overall frequency of accidents declined 17%, while the absolute numbers attributed to human factors declined only 11.7%. Again, the percentage of human factor-caused accidents sticks at 37% for the time period.

So, comparing the two periods, the rate of accident reduction is greater in the 2008-2017 period, and the reduction in the number of human factor-caused accidents is also greater. I think this might indicate that drug and alcohol testing has been successful in its impact on employee behavior.

To my knowledge, FRA does not specifically “break out” for the public the number of accidents where employees tested positive for drug and alcohol use. To my knowledge, FRA also does not provide, to the public, the number of mandatory and random drug and alcohol tests performed in the industry, and the “failure” rates for employees so tested. I’d make a Freedom of Information Act request, but I’ve got two requests that have been in the queue for the past two months, and I’m still waiting to hear.

Maybe NTSB has access to that data, but if that’s so, then why use the small sample of investigated accidents? Without employee failure rates for random and mandatory testing, what are we to make of Chairman Sumwalt’s assertion that because the locomotive engineer in the Chester accident did test positive for marijuana, his failure shows that “despite DOT random drug testing requirements for locomotive engineers, such a program did not deter his use of an illicit drug”?

Well, we can scratch our heads and go, “Huh?” The program is not designed to deter any particular individual’s use of an illicit drug. An individual can always exercise professional irresponsibility and violate the rules. The program is designed to deter the class of employees from using the prohibited substance.

Can you imagine anyone arguing that “despite the requirements of Rule 292 for locomotive engineers to stop at a signal displaying ‘Stop,’ such requirement did not deter this specific locomotive engineer from proceeding through the signal and into the interlocking”?

Unfortunately, I can imagine someone arguing just that way, over just that issue.

The real problem here isn’t the MWL. The Most Wanted List is a useful index to the problems we encounter and the solutions we must consider. The problem isn’t the desire of Chairman Sumwalt and the NTSB to make safe railroads even safer. We all welcome that desire. We all share that passion.

The problem is that we cannot improve rail safety if we use incomplete or inaccurate data.

That might be good enough for Facebook and Twitter. It’s not good enough for us.

* Satori is a Japanese Buddhist term for awakening, comprehension, understanding. It is derived from the Japanese verb satoru. In the Zen Buddhist tradition, satori refers to the experience of kenshō, “seeing into one’s true nature.” Satori and kenshō are commonly translated as “enlightenment.”

The post Safety won’t improve with incomplete, inaccurate data appeared first on Railway Age.

Continue reading...