{
  "id": "2005-may-june-911-data-shows-systems-shortcomings",
  "slug": "2005-may-june-911-data-shows-systems-shortcomings",
  "url": "https://dwillis.github.io/uplink/article/2005-may-june-911-data-shows-systems-shortcomings.html",
  "year": 2005,
  "month": "May-June",
  "monthNum": 5,
  "headline": "911 data shows system's shortcomings",
  "kicker": "EMERGENCY RESPONSE",
  "author_name": "Scott A. Zamost",
  "author_title": "Reporter",
  "affiliation": "NBC",
  "summary": "After investigating a drowning where a nanny's three 911 calls went unanswered, journalists discovered a systemic failure in Broward County's 911 emergency response system by obtaining call records that revealed lengthy delays before operators answered calls.",
  "keywords": [
    "drowning",
    "911",
    "investigation"
  ],
  "topics": [
    "Investigative Methods",
    "Public Safety",
    "Data & Databases"
  ],
  "technologies": [
    "AccuTrak"
  ],
  "full_text": "It started with an intriguing tip. And it ended with major changes in the works for the 911 emergency response system in Broward County, Fla.\n\nA source suggested we examine a drowning in which a nanny called 911 three times. Instead of getting an operator, no one answered the phone, so she rushed the child to a nearby fire station. The toddler later died.\n\nThis was something entirely new. Reporter Patricia Andreu and I knew there had been periodic stories about problems in 911 systems around the country. Most of those stories had focused on delays in getting to locations depending on where someone lived. But none had shown a systemic failure to answer calls in a timely manner using police records.\n\nWe first obtained the police file of the drowning, which confirmed the nanny had dialed 911 three times. The police reports did not specify how long the phone rang each time. To get that, we asked the Broward Sheriff's Office, which handles most of the 911 calls for Broward County for the call records the day of the drowning.\n\nPrior to our request, a source told us that every time someone calls 911, a record of the call is logged into the sheriff's computer system. This information, stored in a system called AccuTrak developed by Gemini Software Technologies, includes the time the call came in, when an operator answered the phone, when the operator disconnected the call and if the caller hung up. That call data would prove to be a gold mine for our story.\n\nAt the same time, we learned that any time there is a question or complaint about a 911 call, an official investigation is opened. We obtained the Broward Sheriff's Office investigative files about the drowning and found remarkable information: The nanny's three phone calls to 911 lasted 37 seconds, 23 seconds and 11 seconds before she hung up and took the child in her own car to the fire station. This was the kind of detail that would form the basis for a broader investigation. If this happened in a drowning where a child died, how long do other emergency phone calls ring before they are answered?\n\nAt first, the sheriff's office told us it wasn't sure such records existed. We then made a public records request for the AccuTrak records for the drowning.\n\nWe received the paper records for the drowning incident, but there was a problem. It was impossible to decipher them without a law enforcement expert from the agency we were investigating. Some of the lines were coded with information about which operator picked up. For instance, \"C08\" meant the operator in the eighth position. The public information officer gave us a basic explanation, but we needed more.\n\nWe decided to ask for a telephone interview with the head of the police communications division to explain the records and how the 911 system worked. Since they knew we were looking into answer times anyway, we figured knowing this information was critical to deciding whether we had a bigger story.\n\nOver the phone, I spoke with Maj. Paul Lauria of the Broward Sheriff's Office. He explained what the fields meant - when a call came in, when it was answered and when the operator disconnected it. The records also showed how long \"abandoned calls\" rang before the caller hung up.\n\nLauria confirmed the call answer times in the drowning incident. He told me the operators answered calls \"as quickly as possible\" and there was no standard they had to meet because it would be \"self-defeating.\"\n\nHowever, we later learned there was a standard the sheriff's office agreed to meet: 90 percent of 911 calls must be answered within 10 seconds. It's a state standard we discovered in a phone interview with Florida's 911 coordinator. It would turn out to be a key element in our story.\n\nWe looked at the AccuTrak records for the day the toddler drowned. The call answer times for that day were well below the 90 percent standard. It appeared we were onto something so we asked for six months of records. The records are kept electronically and the Florida public records law states that agencies should be able to redact non-public information. Nevertheless, the sheriff's office insisted it could not electronically redact the confidential phone numbers of the callers. We consulted someone in our own information technology department who told us this was possible. The sheriff's spokesman insisted it couldn't be done and said the office could provide the records on paper with the numbers crossed out, a process that would take weeks or even months.\n\nIn the interest of moving the story forward, we asked for six months of AccuTrak records on paper. The sheriff's office, as it had done throughout our investigation, responded in a way that kept delaying the story. It told us it would take several months to cross out the phone numbers and estimated the cost could be higher than $8,000. That was unacceptable to us. We continued to press the office about getting the records in electronic form but kept getting denials.\n\nWe ultimately decided to look at two randomly selected weeks, which would amount to nearly 29,000 emergency calls. The state 911 coordinator said that would be an excellent sampling and more than anyone had ever analyzed. The paper records, which still took nearly a month to get, cost about $600.\n\nKeying the call information into an electronic format was a monumental task. A group of dedicated interns helped input the time each call came in and the time it ended into a Microsoft Excel spreadsheet. We then wrote formulas that calculated the answer time for every call, and the daily totals. We broke down the calls into 10-second increments. We included a column for the abandoned calls and the times it took before the callers hung up.\n\nThe office's performance was much lower than the state standard. Only 65 percent of the calls were answered within 10 seconds. Hundreds of other calls took 20, 30, 40 seconds - even more than a minute. When we finally interviewed Lauria on camera, he not only acknowledged his department could not meet the standard, he said it doesn't do what we did - track call answer times.\n\nWhile we were gathering the numbers, Andreu concentrated on persuading the parents of the boy who drowned to go on camera. Still in shock, they finally agreed with the hope that the attention would ensure this wouldn't happen again. We also tracked down the nanny who made the call as well as an elderly woman who hung up after her call to 911 rang for more than one minute. We ended up with two compelling examples on camera as well as all the strong data we had assembled.\n\nReaction was swift. Shortly after the story aired in November, two cities agreed to study whether they would be better off handling 911 calls on their own.\n\nIn late January, Broward Sheriff Ken Jenne credited our investigation with major changes in how 911 calls are handled. The sheriff's office installed software that allows it to track call answer times on a daily basis. Keeping track of calls and better training have already improved answer times. In addition, the sheriff told us he would ask for additional staff to handle non-emergency calls. He hopes to set up a system in which those calls are immediately routed to special operators to free up emergency lines. Finally, the sheriff insisted his agency must work toward reaching the state standard of 90 percent.\n\nContact Scott Zamost by e-mail at Scott.Zamost@nbcuni.com",
  "provenance": {
    "extracted_by": "anthropic/claude-sonnet-4-6",
    "extracted_at": "2026-07-05",
    "verified": true
  },
  "similar": [
    {
      "slug": "2004-march-sprawl-outpaces-private-ambulances",
      "score": 0.266
    },
    {
      "slug": "1999-november-learning-the-hard-part",
      "score": 0.217
    },
    {
      "slug": "1996-december-whats-the-911",
      "score": 0.212
    },
    {
      "slug": "1999-julyaugust-ambulance-response-times-cardiac-crapshoot",
      "score": 0.172
    },
    {
      "slug": "2004-march-response-times-fall-short",
      "score": 0.124
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/2005-may-june-911-data-shows-systems-shortcomings.html",
  "json_url": "https://dwillis.github.io/uplink/article/2005-may-june-911-data-shows-systems-shortcomings.json"
}