The tip was vague: There had been a series of unusual personnel moves in the heart surgery unit at a beloved local hospital, Johns Hopkins All Children’s. Maybe someone should check it out.
It’s easy to see how the story might have ended there. But Tampa Bay Times reporters Kathleen McGrory and Neil Bedi were determined to get to the bottom of it.
They embarked on what would become a yearlong journey to understand what had happened in the surgical unit, which was widely considered among the most prestigious in Florida. They began the arduous process of developing sources with firsthand knowledge and scouring every data source imaginable to look for a hint of what had happened.
Behind the scenes, the hospital’s powerful leaders raised ferocious objections.
McGrory had learned early on that surgeons at the hospital had left a surgical needle in a baby’s chest. The child’s parents said they were not told about the needle until after their daughter was discharged.
When McGrory asked about the incident, the hospital’s CEO conceded the program had experienced “challenges” and its mortality rate had gone up. But he would not elaborate. Instead, he came to the Times’ Editorial Board to chide the paper for its coverage. He deflected questions, citing the hospital’s long history and vital mission to say it deserved the public’s trust. “We’ve already self-policed our way out,” he said.
It would take months of reporting, scores of interviews and a first-of-its-kind analysis to reveal the truth he was trying to hide:
A stunning number of surgeries had suddenly gone wrong.
Administrators had brushed aside clear warnings about the problem.
And children had died as a result.
What makes this project innovative?
It began with old-fashioned reporting to build a massive pool of records. McGrory and Bedi criss-crossed the region to collect thousands of pages of medical records from families who had taken children to the Heart Institute. (Some parents couldn’t afford them, so the reporters secured permission to purchase them on the family’s behalf, at a cost of thousands of dollars.) They mined the records for similarities and cross-referenced them against peer-reviewed research to identify unusual patterns. At times, they found mistakes that the parents themselves didn’t know about. Next, they set out to develop a method to place those failures in context. The reporters quickly determined there was an obscure federal methodology to calculate mortality rates for children’s heart-surgery patients with state billing data. But budget cutbacks meant it hadn’t been updated in years, even though the state data’s format had changed, rendering it unable to produce up-to-date results. They developed their own version of the federal methodology with handwritten Python code that accounted for the updates by reviewing research papers, studying hospital billing codes and examining hundreds of records by hand. It was high-stakes work. The hospital wouldn’t discuss its numbers, and Dr. Jacobs was a leader in the field of heart-surgery data who had already suggested in our earliest conversations that what we were attempting was impossible. Once the analysis seemed solid, the reporters spent weeks trying to come up with every potential objection they could. One by one, they knocked each down with additional reporting and analysis. Top experts ultimately reviewed and blessed their work. When it was time to tell the story, they presented their analysis with tremendous transparency. It was seamlessly integrated into the digital story as interactive charts. They also wrote a lengthy explainer, readable by laymen but containing interactive annotations for experts. And they published all of their results and the computer code.
What was the impact of your project? How did you measure it?
Our investigation ran barely two months ago, but the impact has already been profound: * Top management resigned. The CEO, the Vice President in charge of risk management and the chief heart surgeon all departed less than two weeks after we published. The surgery department chairman stepped down from his leadership role, and three other top administrators would later leave the hospital. * An internal probe began. Johns Hopkins, the hospital’s parent company, hired a former federal prosecutor at a top global law firm to lead an investigation into what went wrong at the heart surgery program. It said it would perform no new heart surgeries until it was sure the procedures were safe. It hired a group of outside experts to consult on restarting the program and vowed to hire an external monitor to ensure the reforms stick. * Johns Hopkins pledged improvements. When asked about problems at other Johns Hopkins hospitals, the health system said it would prioritize safety. “The Tampa Bay Times has identified occasions where it is apparent that as an organization we failed to act quickly enough, we failed to listen closely enough and, in some instances, we failed to deliver the care our patients and their families deserve,” the health system said. “This is unacceptable.” * Regulators investigated. Our initial reporting last spring on the needle left in the baby, prompted a state inspection. The hospital was cited for breaking Florida law related to proper disclosure of medical mistakes. * State health officials are considering reform. Policy makers are now considering a series of changes, including some that would increase transparency around mortality rates, to ensure a situation like this cannot happen again in secret. Influential state lawmakers have promised those recommendations will be backed by action in Tallahassee. The project also ended up being the most read Tampa Bay Times series in years in terms of engagement time and number of users. Analytics showed that the readership was extremely local and this story on a revered hospital resonated with readers across the Tampa Bay area.
Source and methodology
To be as transparent and open as possible, we outlined our complete methodology here: http://www.tampabay.com/projects/2018/investigations/heartbroken/data-methodology/. We also published the Jupyter Notebook with the Python code behind our analysis, and the results on GitHub: https://github.com/tbtimes/florida-heart-surgery-analysis/.
Jupyter notebooks, Python, Microsoft Access, Microsoft Excel, Google Sheets, JS, HTML, Sass
Kathleen McGrory, Deputy Investigations Editor Neil Bedi, Investigative reporter