Such was the scene of a young, black university student shot by police in one of the most dangerous cities in the US. Tragic events like these often punctuate the daily interactions between the health and criminal justice systems. And they rarely have happy endings. But, it was also the moment when a young GP named Jeffrey Brenner began an investigation that would link the concepts of law enforcement and population health. It was the start of “hotspotting” for health care.
Hotspotting is a concept first pioneered by law enforcement in New York City during the 90s. The underlying principle is that police should focus on the pockets of the city where crime is the highest. To do that, we need to create block-by-block crime maps that show the "hot spots."
Brenner, working in primary care and community health, realised a similar methodology could be applied to health care. Armed with billing information from three surrounding hospitals, the young GP cut the data multiple ways, looking for outliers, trends, and anomalies.
In one data-cut looking at ambulance dispatch data, Brenner discovered a single building sent more elderly patients to the ED with falls than any other in the city. Brenner knew he was onto something.
Brenner then expanded his focus and rolled out a series of colour-coded city maps to show where hospital costs per patient were the highest. In that analysis, Brenner identified one large nursing home and a low-income housing tower that cumulatively accounted for 4,000 hospital visits and $200,000,000 USD in health care bills.
In essence he found a hotspot within a hotspot.
Brenner has since become a celebrity within health care circles. Espousing the theory of data-driven targeting, his approach to health can be summarised in the simple concept that marshalling finite resources requires data to target interventions.
How to hotspot
Far and away, the most common question we get on the Global Forum is: "What algorithm do we use to segment our population?" And as much as I would like to point to a sure-fire calculation, the truth is that data governance and population specifics mean that everyone will have to complete a similar exercise of starting with the data pieces you already have available.
In fact, our research shows that beyond the perfect dataset or algorithm, the first and most important step an organisation can take is to establish a team of curious and data-capable individuals to assemble the datasets available to you. In some cases, hospitals have used university students to compile data and run computations to support their hotspotting.
The good news is that Brenner's organisation, the Camden Coalition of Healthcare Providers, published a great step-by-step resource to replicate his approach. For Global Forum members, our recent study on rising-risk patients includes a metric picklist that highlights useful datasets and algorithms to consider.
I invite you to email me at email@example.com if you're interested in sharing your ideas or insights.