16/7/2013

How Wired Italy Made the ‘Check Your Hospital’ Interactive Map

 

By Liliana Bounegru, editor at DataDrivenJournalism.net.

Check Your Hospital, an interactive map that enables readers to check mortality rates for 19 different causes in 1.200 hospitals across Italy, and one of the shortlisted entries in the Data Journalism Awards, is one of the first data projects to address the issue of quality of care in Italian hospitals by using large previously inaccessible datasets. We spoke to Guido Romeo, the journalist who led this work at Wired Italy, to learn how the project was made.

How did the idea for the project come about?

"Healthcare is an issue that is very relevant to everyone and as a journalist I was naturally interested to know more and to possibly debunk some myths about healthcare services in Italy. Italy prides itself as having the world’s second best healthcare system according to World Health Organisation estimations. While this is certainly true for access to care, I have always wondered how mortality and cost of healthcare actually vary across the country. Additionally, given that Italy is facing a massive spending review, I wanted to see if we could show how mortality rates of patients are affected by the dramatic cuts in public expenditure imposed by the latest spending reviews."

Screen_shot_2013-07-12_at_14.25.53.png

Mortality rate for several causes for Ospedale Di Locri in Calabria, Italy via Check Your Hospital

Talk us through how the map can be used.

"The map can be used in a number of ways and can be accessed from a desktop or from mobile devices as we've designed an elastic interface. The most obvious approach is the self-centred one: how good is the hospital next door? What about the centre where one of my family members or friends is being cured? At this level the map allows me to visualize the data for the all the available causes of death for the centre I've searched for. It shows me how it compares to the mean national survival rate for each disease and, once I select a particular one, a pop-up window also shows me the top three and bottom three hospitals for that specific cause.

On a more general level the map confirms that there is a North-South descending gradient in survival rates. It also shows that there is a consistent variation among individual centres and in some hospitals in Palermo you're much better off than in others in Turin. Comparing this data allowed us to show that regions such as Lazio and Lombardy, in spite of the recent scandals linked to healthcare, have excellent survival rates but are overpriced. Others, such as Campania and Calabria, are still overpriced, but rank among the lowest in terms of survival rates. Finally, there are the regions that rank well in terms of quality and financial efficiency, such as Umbria and Emilia Romagna.

Our hope is that readers will be using the map and the data both for making their own choices in terms of healthcare and to start questioning their decision makers about the quality and cost of care."

Walk us through how the project was made: How much time did it take? What data did you use? How did you analyse and present it? What were the challenges?

"The actual work took about three weeks but it was spread out over two months. This didn’t include retrieving the data, which was actually the really hard bit. I had started asking the Health Ministry for data in 2010 as I knew it had been gathered. After a long line of requests and lobbying together with another colleague, I was finally given access to a password-protected site in 2012 (this was presumably related to the change of government and the “transparency” programme of former prime minister Mario Monti). The sad part was the data I had access to (1500 hospitals with 19 causes of death) was viewable but not downloadable as a dataset. I teamed up with a developer friend (Marco Montanari, now a regular contributor to Wired Italy) and one of our journalists, Marco Boscolo, to scrape all of it. To our surprise, while scraping one year, we ended up with data for five years from 2007 to 2012. We used data from Agenas, the Italian Agency for Regional Health Services controlled by the Italian Health Ministry to show the standardized mortality of patients in 1500 hospitals (99% of the country’s healthcare infrastructure is covered) during 2011. This has also been an exercise in opening up data, as Agenas’ data had been, up to that point, only accessible to registered medical doctors and, more recently, to a selected number of journalists (who had been basically hand-picked by the Ministry). The beauty of the data is that it is statistically adjusted to compensate for age, sex and other parameters, which makes all healthcare structures comparable (with a few common sense precautions on the size of hospitals). We then built a database in Excel and worked on it with filters and pivot tables. The last leg of the work was geo-mapping the hospitals and designing the interface, which we did partly in-house at Wired. Mapping was done with QGIS. The map was designed using a Mapbox kit. For the software part, we teamed up with Mitecube, a small Rome based developer team, with whom we had worked previously on our #safeschools project."

What was the impact of this project?

"The map accounted for about 25% of the overall traffic to the Wired Italy website in the first three days after it was published and attracted a lot of attention in Italy. Many colleagues, especially from the local press, use the map to compare local and national data to see where their local hospital sits in terms of mortality risk.

Surprisingly, most critiques came from health reporters who generally considered our data to be wrong. Their main objection was that patients from different hospitals and different areas couldn’t be compared because we are dealing with different populations. While this is generally true, the objection doesn't hold in this case as we used the adjusted mortality risk calculated by the Ministry of Health in order to normalize data and make all hospitals comparable. In fact, the Ministry itself uses this data to work out its policies and the same data is sent out to the World Health Organisation. Medical doctors have generally been very supportive of the effort."

Was there anything that didn’t really work in the project, that you wish you’d done better, or that you’d do differently?

"The interface could have been made more efficient had we had a better budget (overall we spent less than 2000 Euros on this project). I’d also like to add actual stories about specific hospitals and patients to have more impact."

Comments