How I did it: Reporting on delays in newborn testing Date: 12/06/13
The investigative team of Ellen Gabler, John Fauber and Mark Johnson at the Milwaukee Journal Sentinel used data to uncover a national tragedy: infants dying or becoming permanently disabled because labs don't process their newborn blood tests in time to save them. Here, Gabler describes how she got the numbers that led to the multi-part series.
By Ellen Gabler
Our series on newborn screening began when my colleague, Mark Johnson, a health and science reporter at the paper, heard about a baby who nearly died from a rare genetic disorder. Doctors had to go to great lengths to save the child because he got very sick before results from his newborn screening test were returned. Had his condition been identified earlier, the baby could have been put on a special diet and sugar water, avoiding massive medical complications.
A source said that the baby’s blood sample taken shortly after birth had been “batched,” or held at the hospital for a few days instead of being sent to the state lab within 24 hours of when the blood was collected.
Hospitals sometimes batch samples because it saves money. In other instances, hospitals have not established newborn screening protocols or staffers don't follow them when they have been set.
Newborn screening is required throughout the country. Nearly every baby has their heel pricked within a day or two of birth. Spots of blood are collected on a filter card and that filter card is supposed to be sent within 24 hours to a state lab where it is tested for rare, yet deadly, disorders.
Labs must track those cards and babies – that means data. (Our national investigation was based on an analysis of nearly 3 million newborn screening tests from babies in 31 states. More on how that data was gathered in a bit.)
Some poking around the Internet showed that “batching” is a widely warned-against practice. Nearly every state’s newborn screening website had some variation of this message: “Don’t batch samples. Doing so seriously increases the risk of irreversible harm or death in infants.” It seemed like delayed samples might be an issue nationally.
Important note: Many advocates and medical experts contacted initially said delays with babies’ blood samples happened very rarely or “used to” happen. But no one really knew the scope of the problem because it had never been quantified.
Making the request
The first data request went to the Wisconsin State Lab of Hygiene but they were slow to respond. That wasn’t a good sign and my attempts to have a meaningful discussion with someone at the lab failed. The lab director said he would determine what information I needed to tell my story and then he’d let me know.
Tip: Try to talk with someone at an agency before making a records request. Often they will help you craft a request that won’t be denied and can help you understand what information exists and what is publicly available. Public officials frequently want to be helpful and understand they have to follow the law.
Follow-up tip: Except for when they don’t ... if your request is denied, don’t take “no” for an answer the first time (or even the second or third time). Many state officials said immediately that all newborn screening data was confidential and would not be released. I asked again and tweaked the request (more on that later). Insist on having a phone or face-to-face conversation to explain your case to people who are actually in charge – the decision makers who you will hold accountable if you have to write that the agency denied the records request.
Widening the scope
When Wisconsin balked at that first records request, I thought it would be smart to get the information from other states. Each state has to collect the same type of data for every baby’s test – and all that information is kept in databases. The data fields I was requesting did not include names or outcomes of the test, so confidentiality shouldn’t have been an issue.
Another reporter, John Fauber, found a list of lab directors from all state labs in the country. The list had email addresses, phone numbers, titles, etc. I built a Google spreadsheet listing the lab directors for each state and the District of Columbia, then created 51 records requests using a “mail merge” so I could pop in the name of the lab director and the corresponding state to say I was requesting the records under “Arizona” public records laws, for example.
Tip: Stay organized. This project was made possible by spreadsheets. After I did the mail merge and sent out all the requests, I tracked each state’s response and every interaction I had with them in a spreadsheet. There is no way my brain could keep track of conversations and emails with officials in 50 states and the District of Columbia. I kept notes to remind myself what occurred during every conversation, such as: “Very hostile lab director. Wanted to know why this was any of my business. Hung up. Twice.”
Follow-up tip: Don’t be afraid to “go big.” Having data from more than half the country – including the states that deliver the most babies – made the findings from this investigation hard to ignore. I also think several states released data because I was able to tell them other places already had complied. (Despite what your mom told you, peer pressure isn’t always a bad thing.)
Be persistent and polite. Negotiate and don’t give up.
I’m convinced that officials in half the states just figured I’d never follow up so they ignored my request at first.
None of the public records requests were fulfilled immediately and without negotiations. Only one or two were "easy" to get, meaning they didn't take several phone calls and new records requests, dozens of emails, and endless prodding to shake the records loose. Quite often I thought I'd come to an agreement with an agency on what they'd be sending, but then something different would arrive in my inbox. I went back and asked again. Several times I opened a spreadsheet to find that a state had omitted hospital names even though that redaction was never discussed in our multiple phone conversations. I went back and asked again.
I insisted on having phone conversations to discuss the requests. Then the negotiating started. Initially I had requested de-identified data for every single baby born in that state. Some states gave it to me. Others balked and when that happened I said, “OK, how about hospital-specific data.” I knew I needed hospital names so we could hold the institutions accountable.
Tip: Remember that you are requesting records and not just general information. You don’t get to give an agency a homework assignment. Ask about their processes and how they keep their records. My first request to Minnesota was denied in about five seconds. When I asked the guy in charge if they tracked how quickly hospitals sent in blood samples, he happily produced the records (in PDFs) the next day. It also really helps if you have someone in your newsroom who is a computer whiz. From scraping PDFs to building awesome interactives, this project would have been impossible in its current form without the work of savvy news apps developer, Allan James Vestal.
Follow-up tip: Make sure you are talking to the right people. As I mentioned, the initial records requests were all sent to public health labs. In some states, the lab is under the purview of the state health department. When I re-routed my records request through that agency, the information was released. Our own state of Wisconsin was the 26th state to release hospital-specific data – the result of a detailed email and phone conversation with higher-ups at the institution that oversees the labs.
Don’t take “HIPAA” for an answer
Many states tried to claim that releasing data would violate HIPAA (Health Insurance Portability and Accountability Act), a common excuse used with health-related records. While HIPAA is relevant in some situations, it is grossly overused and reporters must fight back.
I asked for data that I knew was collected for each baby but would not identify them. Most importantly that included the date the blood sample was collected and the date it arrived at the lab. I knew if I had those two fields, I could calculate how long it took a sample to arrive at the lab. The name of the hospital was also crucial as we wanted to hold the institution accountable. This was a sticking point for many states that did not want to call out individual hospitals.
Tip: Get the agency to explain why releasing the information would violate HIPAA. See if it is reasonable.
Follow-up tip: If it’s not reasonable, state your case and consider using their own data against them. One state said releasing the name of a hospital and the total number of babies born at that hospital would allow me to identity individuals, a violation of HIPAA, they said. I went to that health department’s site and looked at health statistics the agency regularly publishes. In an email I politely explained that it seemed strange they would withhold the information I was seeking while they frequently publish the number of individuals within small Indian tribes who are diagnosed with specific sexually transmitted infections. I got the data a few days later.
Find the perfect people
To tell a story you need people who have been affected by whatever you are writing about. Search everywhere for these people and do not settle for anything less than a perfect example.
Tip: Some places to look: Lawsuits (use all kind of search words – “inborn metabolic error” is the technical way to talk about these genetic disorders); advocacy organizations; Facebook (there is a group for everything now – find every one); blogs; meeting minutes; Congressional hearings. Ask sources for other sources.
Follow-up tip: Put these people to work for you. Most doctors and hospitals refused to talk and wouldn’t hand over records so I had the parents request them. They certainly are entitled to those records. My colleagues faced similar challenges. It’s not smart to just rely on a parent’s recollection so you need hospital records, lab reports, etc. to back up their claims. In one case, a parent called a lab and asked for a detailed explanation of what happened with their child’s blood sample. I could not have gotten the same information. Do not be afraid to call back your sources several times to clarify something or just to get more details. “Was it giraffes or monkeys on the baby’s shirt?” It’s exhausting but will make your story better and ensure accuracy, which is what your sources want, too.
Read the entire Deadly Delays series at www.jsonline.com/deadlydelays
Ellen Gabler is a reporter and assistant editor on the Milwaukee Journal Sentinel’s investigative team where she incorporates data analysis into her reporting. She says of her job: "Just about everything fires me up about what I do. I love it and still can't believe it's an actual career. I love getting to the bottom of difficult stories, especially when others say it can't be done." IRE hosted a Google Hangout to talk about the project with Gabler and Vestal.