Every few months an emergency manager gets asked about a transportation accident, either by a reporter, a county commissioner or a firefighter who wants to know what really happened. The National Transportation Safety Board publishes an enormous amount of material on closed cases, and almost none of the people quoting it have read past the summary. This is a working guide to the difference between a preliminary report, the factual record in the public docket, a probable cause finding and a safety recommendation, and to what an emergency manager should actually pull out of a case the Board has finished.

What the Board is, and what it does not have the power to do

The National Transportation Safety Board is a small independent federal agency with five presidentially appointed, Senate-confirmed members. It was created in 1967 inside the Department of Transportation and was separated from that department by the Independent Safety Board Act of 1974, with the separation taking effect the following year, because an agency investigating the FAA’s oversight of an airline cannot sit under the same secretary as the FAA. It investigates every civil aviation accident in the United States along with selected highway, railroad, marine, pipeline and hazardous materials accidents, and it participates in foreign investigations where a U.S. operator, manufacturer or registry is involved.

The Board writes no regulations, issues no fines and revokes no certificates. Its entire output is information: reports, findings, probable cause determinations, safety recommendations, safety alerts and research studies. When something changes after an accident, the change comes from the FAA, the Pipeline and Hazardous Materials Safety Administration, a state legislature, a manufacturer or an operator, and the Board’s role is to have made the case in public and then to track whether anyone acted on it.

Major investigations run on the party system. The investigator-in-charge designates parties that can supply technical expertise, which typically means the operator, the airframe and engine manufacturers, the relevant labor unions and, in aviation, the FAA as a party by statute. Parties work the investigation under NTSB direction, and they sign an agreement not to release investigative information on their own. Parties do not vote on findings and do not determine probable cause, though they may file a written submission arguing for their view of the evidence, and that submission goes into the public docket where you can read it and weigh it for what it is.

Preliminary, factual, final: three different documents

The preliminary report is a short factual notice, typically posted within a few weeks of an aviation accident, that records the date, location, aircraft, operator, injuries and a bare narrative of what is known so far. It carries no analysis and no probable cause, and the NTSB says on the face of it that the information is preliminary and subject to change. People misuse these constantly, treating a sentence about weather or a witness description as though the Board had reached a conclusion, when a preliminary report is closer to the first size-up transmitted from a working structure fire, useful for orienting yourself and worthless as a basis for judgment about why anything happened.

The factual material comes next and is much larger. In a major investigation the work is divided among groups covering operations, human performance, air traffic control, meteorology, structures, powerplants, systems, maintenance records, the cockpit voice and flight data recorders, and survival factors. Each group chairman writes a factual report that lays out what the group found, and those reports are deliberately stripped of analysis, because the Board separates the establishment of facts from the argument about what the facts mean. Aircraft performance studies, sound spectrum studies, interview transcripts and photographs sit alongside them in the record.

The final report is where analysis appears. It restates the factual history, analyzes it, lists numbered findings, states a probable cause with any contributing factors, and carries the safety recommendations issued as a result. For major accidents the report is adopted at a public Board meeting where the members vote and can file concurring or dissenting statements that are printed in the report. Most accidents, including the great majority of general aviation cases, never get a Board meeting at all and are closed with a much shorter final report produced by staff under delegated authority, which still contains a probable cause statement.

The public docket, and why it beats the press coverage

The public docket is the assembled evidentiary record of an investigation, and for a major case it can run to thousands of pages across dozens of exhibits. It usually opens months into the work, well before the final report is adopted, and the Board’s practice is to release the factual material without any accompanying analysis so that the record is public before the argument about it begins. You reach it through the Board’s online case system, CAROL, which is searchable by date, location, mode, operator and accident number from ntsb.gov.

Press coverage of an accident is built almost entirely from the Board’s media briefings during the first week and from the probable cause line in the final report. Those two sources are correct as far as they go, and they leave out everything an emergency manager would want. The docket is where you find the minute-by-minute notification timeline, the interview with the airport operations supervisor, the radio traffic transcript, the hospital distribution list, the photographs of the exits that did not open and the maintenance record showing when a part was last inspected.

Two limits are worth knowing before you go looking. Cockpit voice recorder audio is not released to the public because federal law forbids it, so what you get is a written transcript prepared by the CVR group, and the transcript covers only the portion the group judged pertinent. Medical and personal information about individuals is redacted, and some material is withheld or partially released for other legal reasons. If a document you expect is missing, the docket index will usually tell you it exists, and the Board’s records office can tell you the status of anything you cannot see.

Probable cause, findings, and what the statement is for

The probable cause statement is a formal determination, made under the Board’s statutory mandate, of the cause or probable cause of an accident. It is usually one or two long sentences naming the cause, followed by a list of contributing factors. The findings that precede it are numbered statements of what the Board concluded on each issue, including negative findings that rule things out, and the findings often matter more to a working reader than the cause line does, because that is where the Board records things like the adequacy of the emergency response or the quality of a training program.

A probable cause statement exists to support prevention. It is written to justify recommendations rather than to assign legal liability, and federal law bars the use of NTSB accident reports as evidence in civil suits for damages arising from the accident, a restriction found in Title 49 of the U.S. Code that any attorney working the subject can cite for you precisely. Courts, insurers, prosecutors and regulators run their own processes with their own standards of proof, and they sometimes reach conclusions that do not match the Board’s. When someone tells you the NTSB blamed a pilot, the honest translation is that the Board determined the pilot’s action to be the probable cause of the accident for safety purposes, which is a narrower claim than it sounds.

Read the whole statement rather than the first clause. The Board frequently names an operator’s procedures, a manufacturer’s documentation or a regulator’s failure to require something among the contributing factors, and the press coverage almost always drops those and keeps the crew. Also read the analysis section that supports the statement, because that is where you learn how confident the Board actually was, and where uncertainties that could not be resolved are written down plainly.

Safety recommendations and their status codes

Safety recommendations are the operative product. Each one is addressed to a specific recipient, carries its own number, and asks for a specific action. They are not confined to final reports, since the Board can issue recommendations at any point during an investigation and issues urgent recommendations when it believes a hazard requires action before the case closes. Recommendations also come out of safety studies and special investigation reports that are not tied to a single accident at all.

Every recommendation gets a status, and the status is public. The classifications include Open with the response awaited, Open with the response judged acceptable or unacceptable, and several forms of Closed, including Closed as Acceptable Action, Closed as Unacceptable Action, Closed as Reconsidered, Closed as No Longer Applicable and Closed as Superseded. Those distinctions carry most of the useful information, because a recommendation closed as acceptable action means the recipient did what was asked, while a recommendation closed as unacceptable action means the Board gave up on that recipient and the underlying hazard is very likely still there. Check the current status in CAROL before you quote a recommendation in a plan or a briefing, because a recommendation from 2011 may have been superseded twice since.

Alongside recommendations, the Board publishes safety alerts, which are short plain-language documents aimed directly at pilots, mechanics, mariners, truck drivers and small operators rather than at agencies. They are free, they are usually two pages, and they are the easiest NTSB product to put in front of a crew at a shift briefing. The Board also maintains a Most Wanted List of priority safety improvements, published in multi-year cycles, and the current edition on ntsb.gov tells you where the Board thinks the largest unaddressed gaps are across all modes.

Survival factors: the chapter written for your side of the job

In an aviation investigation, the survival factors group covers occupant protection, seat and restraint performance, cabin configuration, exits and evacuation, crashworthiness, injury patterns and the emergency response. In rail, highway, marine and pipeline cases the equivalent material appears under emergency response or occupant survivability headings. This is the part of the record written about your work, and it is almost never covered by the press, which means an emergency manager who reads it knows things the local officials quoted in the news coverage do not.

What you find there is unusually specific: notification times from first alarm to airport rescue and firefighting response, the sequence in which mutual aid units arrived and where they staged, whether unified command was established and who was in it, how patients were counted and where they were transported, which radio channels worked, how long the evacuation took, whether the airport emergency plan matched the actual geography of the crash site, and how the family assistance center was stood up. Group factual reports include the interviews, so you can read the airport operations supervisor describing in his own words what he did not know in the first ten minutes.

Family assistance is a specific federal responsibility worth understanding before you need it. The Aviation Disaster Family Assistance Act of 1996 assigned the NTSB the role of coordinating federal family assistance after a major aviation accident, and Congress extended a comparable structure to passenger rail in 2008. The Board’s Transportation Disaster Assistance division works alongside the operator, the medical examiner and local government, and the local pieces of the job, meaning facility space, security, credentialing, behavioral health support, translation and traffic control, land on the county. The docket in a major case will show you how those pieces were assembled, and you can compare that to what you would actually be able to produce.

Certificated airports operate under the FAA’s Part 139 rules, which require an airport emergency plan and periodic full-scale exercises, with the current interval and content requirements set out in the regulation and interpreted by your airport certification safety inspector. Verify the current text before you write it into anything, and get the last exercise after-action report from the airport rather than assuming your agency’s copy of the plan is the current one.

Four closed cases and what each one gives a local agency

All four cases below are closed, meaning the Board has adopted a final report in each one, and everything attributed here comes from those adopted reports rather than from contemporary coverage. I am reporting what the Board determined, not offering an opinion about what caused any of them, and readers who want the reasoning behind a determination should read the analysis section of the report and the supporting exhibits in the docket.

ValuJet Flight 592 crashed in the Everglades on 11 May 1996, killing all 110 people aboard, and the Board’s final report, adopted in 1997, determined the probable cause to be a fire in the class D cargo compartment initiated by the actuation of chemical oxygen generators improperly carried as cargo. What makes it worth reading is the set of contributing factors the Board named, which included the maintenance contractor’s failure to properly prepare, package and identify the generators, the airline’s failure to oversee that contractor, and the FAA’s failure to require smoke detection and fire suppression in class D compartments. The FAA subsequently required detection and suppression in those compartments, which effectively ended the class D compartment in passenger service.

US Airways Flight 1549 ditched in the Hudson River on 15 January 2009 after both engines lost thrust shortly following takeoff from LaGuardia, which the Board’s final report, adopted in May 2010, attributed to the ingestion of Canada geese, and all 155 people aboard survived. The survival factors and emergency response material is a study in what a fast, largely improvised waterborne rescue by commercial ferry operators and municipal agencies looks like when it works, and the recommendation list addresses engine bird ingestion certification, checklist design for a dual engine failure at low altitude, and passenger flotation equipment, among other subjects. Read those recommendations with their current status attached rather than as they appeared in 2010.

Asiana Flight 214 struck the seawall short of Runway 28L at San Francisco on 6 July 2013 with 291 passengers and 16 crew aboard, and three passengers died. In its final report, adopted in June 2014, the Board determined the probable cause to involve the flight crew’s mismanagement of the approach and their inadequate monitoring of airspeed, and it named among the contributing factors the complexity of the airplane’s autoflight system, the adequacy of Boeing’s documentation and Asiana’s training. For an emergency manager, the survival factors work is the reason to open this docket at all, because two evacuation slides inflated inside the cabin and trapped flight attendants, and one of the three fatally injured passengers, who had been ejected from the airplane and was lying in firefighting foam on the ground, was run over by aircraft rescue and firefighting vehicles. The Board addressed ARFF procedures for locating occupants outside the aircraft in its findings and recommendations.

The Merrimack Valley over-pressurization of 13 September 2018 is the case to hand to anyone who thinks NTSB material is only about airplanes. During a Columbia Gas cast iron main replacement project in Lawrence, Andover and North Andover, the low-pressure distribution system over-pressurized and fires and explosions damaged more than a hundred structures. The Board’s final report, adopted in 2019, records one fatality and injuries to more than twenty people, gives the exact counts, and centers its probable cause on the utility’s weak engineering management and its failure to account in the work package for regulator sensing lines that remained connected to a main being abandoned, with state oversight named among the contributing factors. The emergency response sections cover the mass evacuation, the difficulty of shutting the system down when the control center in Ohio could see the alarms without being able to close the valves, and the unified command that local officials assembled across three communities.

While an investigation is open, you have nothing to say about cause

The rule I follow, and the one I would put in a public information annex, is that no one in a local agency comments on the cause of a transportation accident while the investigation is open, and that rule rests on something firmer than caution. The Board is the single release authority for investigative information during an active case, parties sign an agreement to that effect, and a local official speculating on camera about a mechanical failure or a crew error is publishing an unverified claim under the apparent authority of a government position. Your agency can and should describe its own operations, meaning what it was dispatched to, what it did, how many patients it transported and what the current road closures are, which is a substantial and useful thing to say.

The same discipline applies to writing. When a case is open, anything published about cause is speculation, including a retired investigator’s opinion on a cable panel, which is a report of a suspicion rather than a finding, and including a fragment of ATC audio circulating online. Preliminary reports are preliminary and may change. If you must write about an open case, write about what is known to have happened, label the unconfirmed material as unconfirmed in plain words, and point the reader to the docket and to the eventual final report.

There is one more thing worth saying to elected officials and to your own staff. The Board’s product is slow on purpose, because the alternative to taking twelve to twenty-four months on a major case is publishing a cause that later turns out to be wrong, and a wrong probable cause produces wrong recommendations that consume real money and attention at hundreds of agencies. The waiting buys a record that holds up, and the docket exists so that anyone who wants to check the Board’s work can do it without asking anyone’s permission.

What to do at your agency

• Have the emergency manager open the Asiana Flight 214 docket in CAROL this month, print the survival factors group factual report, and read the emergency response timeline against your own airport or mass casualty plan, then write a half-page note of the differences for the next planning meeting.
• Ask the airport director or the airport certification safety inspector, in writing, for the date of the last full-scale Part 139 exercise, the after-action report from it, and the current version of the airport emergency plan, and confirm that the copy in your EOC matches.
• Have the public information officer add one paragraph to the existing crisis communications annex stating that during an NTSB investigation the Board is the single release authority for investigative information, that your agency comments only on its own operations, and naming who is authorized to make that statement.
• Have the communications or dispatch supervisor verify the current NTSB 24-hour reporting number from ntsb.gov and place it on the dispatch notification card alongside your state warning point.
• Put a twenty-minute item on the next scheduled LEPC or EOC working group agenda in which one person walks the group through the emergency response section of the Merrimack Valley pipeline report and identifies which utility in your jurisdiction would face the same shutdown problem.
• Have whoever maintains your plans check every NTSB recommendation cited in them against its current status in CAROL, and annotate each citation with the classification and the date, deleting any that have been superseded.
• Identify by name, before you need one, who in your county would run a family assistance center, what building it would use, and who provides behavioral health and translation support, and record those three answers in the mass casualty annex.

Takeaways

• The NTSB is an independent five-member board that investigates and recommends, and it has no authority to write rules, levy fines or revoke certificates, so every change that follows an accident comes from a regulator, a legislature, a manufacturer or an operator.
• A preliminary report is a short factual notice with no analysis and no cause, it is explicitly subject to change, and quoting it as though it settled anything is a misuse of the document.
• Group chairman factual reports are deliberately written without analysis, because the Board separates establishing the facts from arguing about what they mean, and the argument appears only in the final report.
• The public docket usually opens months before the final report is adopted and contains the interviews, transcripts, studies and photographs that press coverage never reaches, and it is searchable through the Board’s CAROL system at ntsb.gov.
• Probable cause is a safety determination rather than a liability finding, federal law bars the use of NTSB accident reports as evidence in civil damage suits, and other investigations with different standards of proof can and do reach different conclusions.
• Safety recommendations carry a public status classification, and the difference between Closed as Acceptable Action and Closed as Unacceptable Action tells you whether the hazard was addressed or abandoned, so check the status before citing one.
• Survival factors and emergency response sections are written about local response work, covering notification times, unified command, evacuation, patient distribution and family assistance, and they are the highest-value part of an NTSB report for an emergency manager.
• Speculation about cause while an investigation is open has no place in professional material, and the correct response to a question during an open case is to describe your own agency’s operations and point the questioner to the NTSB.