{
  "id": "2006-july-august-poor-conditions-plague-medical-helicopters",
  "slug": "2006-july-august-poor-conditions-plague-medical-helicopters",
  "url": "https://dwillis.github.io/uplink/article/2006-july-august-poor-conditions-plague-medical-helicopters.html",
  "year": 2006,
  "month": "July-August",
  "monthNum": 7,
  "headline": "Poor conditions plague medical helicopters",
  "kicker": "AIR SAFETY",
  "author_name": "Alan Levin",
  "author_title": "Aviation safety reporter",
  "affiliation": "USA Today",
  "summary": "A USA Today investigation using a custom database discovered that lax industry safeguards and poor visibility lead to a high rate of fatal air ambulance helicopter crashes.",
  "keywords": [
    "aviation safety",
    "air ambulance",
    "NTSB"
  ],
  "topics": [
    "Public Safety",
    "Investigative Methods",
    "Data & Databases"
  ],
  "technologies": [
    "Access",
    "spreadsheet"
  ],
  "full_text": "There's no substitute for precision on a big project. USA Today set out more than a year ago to learn more about why air ambulance flights were crashing at an alarming rate. A quick review of the clips revealed several fatal crashes that had killed patients. The numbers of crashes were climbing. Our guts told us we had a good story. But you can't attribute major findings to gut instinct. That started several months of searching for hard data that eventually led us to create our own database of crashes from scratch. Along the way, we occasionally cursed the sometimes archaic ways of aviation regulation and recordkeeping, which can make data retrieval very awkward. But, in the end, we had a roadmap laying out the worst crashes and the major safety shortfalls. Our data was more complete than anything the government had compiled. Since the Vietnam War, air ambulances have gradually become commonplace. Every major metro area has a steady stream of helicopters landing on the roofs of major trauma centers, bringing the most severely injured and gravely ill patients to the best facilities. In recent years, air ambulances have helped rural areas cope with major changes in the health care system. As more and more small rural hospitals shut down, helicopters have become increasingly important. For millions of Americans, the only quick way to an emergency room is by air. During the 1980s, the number of air ambulance flights surged. But in the race to provide more service, safety standards suffered. In the years 1985 through 1988, there were 15 fatal air ambulance helicopter crashes, compared to only seven in the previous four-year period. The National Transportation Safety Board (NTSB) attributed the spike in crashes and deaths to carelessness that had crept into overstressed operations and to inadequate federal oversight. The crashes fell rapidly under the glare of attention from investigators and federal aviation regulators. As we began our reporting last November, we suspected that many of the same dynamics had crept into the industry over the past few years, but the NTSB was just beginning its own review of the crashes and had not issued any findings. So we started our own data collection. As a general rule, the NTSB keeps one of the cleanest databases in government. Because the agency has no regulatory power and relies solely on the bully pulpit, it is also unusually generous with its public records. But when it comes to niches such as the air ambulance world, the database can be difficult to work with. In order to be included in the NTSB's database, an accident must meet a test that requires factors such as fatalities, serious injuries or significant damage to the aircraft. All air ambulance accidents that met the NTSB's statutory requirements were in its database, but there is nothing in the data to flag these flights. They are simply listed under a broad category of charter flights. We did our best to collect lists of the accidents from a variety of sources. NTSB provided us with a spreadsheet of accidents after 1995 and we combed reports and other sources. Paul Overberg, a USA Today database editor, helped design the database and created a data entry form in Microsoft Access. We set our able intern, Marie Skelton, loose on the NTSB's database. The web version of the database includes a fairly robust search function that allows broad text searches. She began plugging in words like \"nurse,\" \"ambulance,\" and \"hospital\" to look for any cases we might have missed. To be as thorough as possible, we also entered into our database the accidents involving medical flights on planes, not just helicopters. Working one or two days a week, Marie finished in about two months. I don't recall the exact number, but by using this crude-but-effective manual search, she found a handful of accidents that others had missed. At our conclusion, we had 279 accidents from 1978 to the early months of 2005. It turned out that finding the crashes was the relatively easy part of the job. Next, I had to audit Marie's work (which was extremely clean) and to categorize all of the accidents. The NTSB provides detailed information about the cause of each accident. But that information comes in the form of a fairly lengthy string of text, and over the years, the agency has used different formats. We needed to boil this down to key components, namely the pilot's error, in order to know what patterns were behind the crashes. How important were errors by pilots versus mechanical failures? Were crashes occurring in bad weather? At night? The only way to identify whether these broad types of issues were leading to crashes was to read over every single NTSB cause finding. I then added a bunch of additional fields to our database for every conceivable type of accident, from pilot error and mechanical failure to factors such as the time of day. I went through each record twice to ensure consistency and accuracy. Later on, I met with a statistician at the Federal Aviation Administration to compare our database with one they had compiled to make certain we had not missed any accidents or included cases that didn't belong. I also went through the data and made sure we had accurate counts of the dead. In particular, we needed to know how many patients had died on flights. It wasn't pretty, but I hand-entered crash totals and various pieces of information into a Microsoft Excel spreadsheet. That allowed us to help visualize some of the trends. For example, it became obvious very quickly that poor visibility from bad weather and flying at night was frequently a contributing factor to crashes. We were far from the first news organization to notice that air ambulance flights were crashing at alarming rates. But the painstaking collection of data (and my editors' patience while I completed the work) allowed us to write about the problem with authority. Even though the NTSB had not yet issued its own findings about the growing number of air ambulance accidents, we were able to reach strong findings on our own. Here are two of our main conclusions: \"Industry safeguards are so lax that pilots have repeatedly caused accidents by knowingly flying into bad weather, failing to check weather conditions or otherwise violating federal or company regulations.\" \"About two-thirds of fatal air ambulance crashes occur in poor visibility, the newspaper's analysis shows. Even so, pilots are not required to have special training about what to do when they encounter fog, snow or darkness.\" The information on these crashes was available for anyone to see, but no one had bothered to compile it in one place. The raw number of crashes was the highest in history, but the industry had grown considerably in the past decade. Of course, the data was only one of the elements we relied on to report the story. But without our computer analysis, the story would not have been half as good.",
  "provenance": {
    "extracted_by": "pre-v2",
    "extracted_at": "",
    "verified": true
  },
  "similar": [
    {
      "slug": "1998-march-the-conversion",
      "score": 0.204
    },
    {
      "slug": "1999-julyaugust-ambulance-response-times-cardiac-crapshoot",
      "score": 0.204
    },
    {
      "slug": "2004-march-response-times-fall-short",
      "score": 0.156
    },
    {
      "slug": "1996-december-ive-got-a-secret",
      "score": 0.151
    },
    {
      "slug": "1997-february-counting-the-crashes",
      "score": 0.135
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/2006-july-august-poor-conditions-plague-medical-helicopters.html",
  "json_url": "https://dwillis.github.io/uplink/article/2006-july-august-poor-conditions-plague-medical-helicopters.json"
}