Key Questions and Concepts
Healthcare hotspotting is a data-driven process for the timely identification of extreme patterns in a defined region of the healthcare system. It is used to guide targeted intervention and follow-up to better address patient needs, improve care quality, and reduce cost. Most often the data in question are hospital reimbursement claims that are continuously produced, standardized by an operational imperative, and bring together high-stakes diagnostic, temporal, financial, spatial, and demographic data in a single set of records. Through hotspotting, claims data can help reveal both a community’s healthcare problems and their solutions.
Why is the history of our techniques important at all? Because the particular technologies of the present should extend our reach, not constrain our imaginations. Systemic change requires systemic understanding and a sense of what might have been. The unrealized possibilities of the past can often be the most realistic path to transforming present institutions. The above description of Healthcare Hotspotting briefly defines it as it has currently come to be known. To better understand all that this technique can offer in the future, let’s look at how the current form has developed.
The United States now spends roughly 18% of our GDP on healthcare, a number that is expected to continue growing in the coming years.1 On the global scale, this percentage is staggering, with American health expenditures per person at 51% more than any other industrialized nation. That comes out to over $8,000 per person each year.2
Meanwhile, a small number of patients in the United States account for a large percentage of healthcare costs, with just 5% of Americans accounting for nearly 50% of costs and 1% accounting for more than 20% of costs.3In Camden, a city where 38% of the population lives below the poverty level (compared to the national average of 14.3%), 13% of the population accounts for 80% of healthcare costs citywide.4 This small subset, or segment, is what we identify as “super-utilizers” in healthcare hotspotting.
The Camden Coalition
Primary care physician Dr. Jeffrey Brenner needed to understand what was driving apparently perverse patterns of healthcare use in Camden, New Jersey. After years of building relationships with local providers and facilities, Brenner was able to obtain hospital claims data from all three Camden hospitals. When he started working with the data, concentrations of cost, geography, utilization, diagnosis, and other variables began to emerge. Brenner explains,
“Hotspotting is segmentation. It’s taking big data sets, [and] segmenting them into strategy so that you can target different pockets of need… Hotspotting is making sure that people who are in need get their needs met… in a rigorous, data-driven way.” Dr. Jeffrey Brenner5
He founded the Camden Coalition of Healthcare Providers in part to create a local community of practice that could learn to act on these insights. Now, with over twelve years of all-payer claims data for Camden residents, the Camden Health Database (CHDb) has become a source of knowledge that can at any time upend, validate, or generalize the intuitions of care providers in the field. The Database is the deep well of insight to which the coalition returns to study trends, investigate patient segmentation variables and produce the visualizations of systemic crisis which have demonstrated a singular capacity to motivate serious stakeholder commitment to community based interventions.
Cholera, Bureaucrats, Chromosomes & Crime
While “healthcare hotspotting” as we use the term here is a relatively new technique, distinct from other forms of hotspotting and other uses of geography in the broader health and medical research fields, it draws on and reacts to each of these. Healthcare hotspotting also returns to themes from even earlier eras of graphical and analytic innovation for inspiration.6 In the visualization and analysis of healthcare system hotspots, four major historical antecedents converge: epidemiological disease maps beginning in 1798,7,8 the rise of national statistical bureaus from the 1820’s, the rise of national statistical bureaus from the 1820’s,9,10 the invention of matrix clustering techniques in the 1950’s,11 and the crime hotspotting trend beginning in the 1980’s.12,13
Hotspotting In Practice
The Camden Coalition’s initial work with claims data demonstrated the need for multi-hospital collaboration and real-time data sharing across the city’s hospitals. Because the creation of claims data only occurs after a person visits the hospital and a bill is generated, any analysis performed is retrospective in nature. To actively find and engage super-utilizers, real-time data is also essential.
In 2011, the Camden Coalition implemented a Health Information Exchange (HIE) that pulls Admission/Discharge/Transfer (ADT) data from the city’s three hospitals. The Coalition uses its HIE as a powerful tool for real-time population health surveillance and to trigger interventions based on individual utilization patterns.
The HIE produces daily reports identifying individuals currently admitted at a Camden hospital who have also had 2 or more admissions in the past 6 months. Coalition staff review the identified individuals’ medical and social histories to determine their eligibility for one of the Coalition’s community-based Care Management Interventions. Once an individual has been identified as eligible for an intervention, a Camden Coalition enrollment staff member reaches out to the individual at the hospital bedside, a catalytic moment where patients are more amenable to change. There will be more detailed discussion of CCHP’s interventions in Module 9.
Atul Gawande’s 2011 New Yorker piece: “The Hot Spotters: Can we lower medical costs by giving the neediest patients better care?”
Elizabeth Bradly and Lauren Taylor: The American Health Care Paradox: Why Spending More is Getting Us Less.
The Kaiser Family Foundation: Health care cots: A primer.
- Healthcare Costs, A Primer: Key Information on Healthcare Costs and Their Impact. 2012. The Kaiser Family Foundation. http://politicsandmedicine.ezweb123.com/download/i/mark_dl/u/4010056499/4575054013/KFF%%20costs%%20primer.pdf.
Analysis performed using the CHDb ↩
- Altonen, Brian. 2012. “Valentine Seaman, 1797 (1804) – the Black Plague or Yellow Fever of New York City”. Wordpress. Brian Altonen, MPH, MS. http://brianaltonenmph.com/gis/historical-disease-maps/valentine-seaman-1804-the-black-plague-or-yellow-fever-in-new-york-city/.
- Friendly, Michael, and Gilles Palsky. 2007. “Visualizing Nature and Society.” In Maps: Finding Our Place in the World, edited by James R. Akerman and Robert W. Karrow, 207–53. University of Chicago Press. http://www.datavis.ca/papers/Maps_Chapter5-ocr.pdf.
- Friendly, Michael. 2006. “A Brief History of Data Visualization.” In Handbook of Data Visualization. Vol. 3. Springer Handbooks of Computational Statistics. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.69.4737&rep=rep1&type=pdf.
- Friendly, Michael. 2008. “The Golden Age of Statistical Graphics.” Statistical Science 23 (4): 502–35. doi:10.1214/08-STS268.
- Wilkinson, Leland, and Michael Friendly. 2009. “The History of the Cluster Heat Map.” The American Statistician 63 (2): 179–84. doi:10.1198/tas.2009.0033.
- Weisburd, David, and Cynthia Lum. 2005. “The Diffusion of Computerized Crime Mapping in Policing: Linking Research and Practice.” Police Practice and Research 6 (5): 419–34. doi:10.1080/15614260500433004.