Cartography has been used since antiquity, when the Greek astronomer Ptolemy created early maps of the Roman Empire. His landmark atlas “Geographia,” published around 150 AD, helped to establish the foundation of modern geography, and to codify the use of pictures to better understand the world around us. One needs look no further than the phones in our pockets to appreciate how ubiquitous maps have become, and how central their role is in our daily lives. Although indispensable as navigation tools, maps are also used to better understand our environment and its reciprocal interactions with society.
Modern computing, including tools like geographic information systems, now allow geographers and other researchers to identify spatial patterns in a host of natural and artificial phenomena, enabling better planning, resource allocation, and environmental stewardship. Medical research, including most notably in epidemiology, has also benefited from insights gleaned through mapping. But for the most part, this methodology remains underutilized, particularly in critical care, where the literature includes only a few maps, mostly related to the geographic distribution of ICU beds (1 , 2).
In this issue of Critical Care Medicine, Szakmany et al (3) present a study using population-level data to examine practice patterns and trends in critical care. Drawing on a unique dataset that merges patient registry data with U.K. government administrative data, the authors detail 1-year mortality patterns for patients discharged from Welsh ICUs.
Although thoroughly modern in its approach, this work draws on precedents for mapping in medicine, some of which also stem from the United Kingdom. The earliest maps of the British Isles in fact date back to Ptolemy’s “Geographia” itself. Centuries later in the 1830’s, the British journal The Lancet published an unsigned review detailing outbreaks of cholera across Europe, Asia, and Africa, complete with a map illustrating the location of outbreaks as circled dots in the affected cities (4). A short time later, in the 1850’s, John Snow famously used detailed mapping techniques to identify the Broad Street pump as the source of a cholera outbreak in London (Fig. 1). Studies such as these were instrumental in the birth of modern public health.
Today, maps play an increasingly visible role in health services and outcomes research. This is perhaps best exemplified by the Dartmouth Atlas of Healthcare, which since its inception in the mid-1990’s has generated numerous maps illustrating many important findings about healthcare in the United States. Some of these observations are directly relevant to critical care, including the substantial variation in healthcare spending between different regions of the United States (5), regional and temporal differences in the intensity of healthcare at the end of life (6), differences in the quality of end-of-life care between regions with high versus low ICU utilization (7), and differences in the number of ICU beds per capita across regions (8).
The Welsh study presented in this issue extends these findings further, with a look at outcomes up to 1 year after discharge from the ICU. Not surprisingly, mortality was greater among older patients with comorbidities, and healthcare utilization was high among survivors in the year following discharge. Almost half of all patients who died following discharge from the ICU passed away before ever leaving hospital. Some of these insights replicate findings from previous studies—including population-scale analyses from Canada, which like the United Kingdom has government-funded healthcare systems that facilitate population-wide data collection (9 , 10). But unlike in previous work, the Welsh results are presented stratified by subregions of the population in question—in this case at the county level—rather than for the population as a whole.
Beyond the insights gleaned into long-term ICU survivorship, the study by Szakmany et al (3) contributes meaningfully to a few key domains of critical care research. First, the authors used the Secure Anonymized Information Linkage (SAIL) databank, a unique repository of patient-level health data from Wales that provides analytic tools—including the ability to link datasets—in a computational safe haven that facilitates data science research. By harmonizing clinical, administrative, and demographic data in a deidentified format, SAIL brings much-needed context in which to interpret not only a patient’s ICU stay, but the larger care environment. The study by Szakmany et al (3) illustrates how analyzing datasets together, rather than in isolation, can generate unique insights, and enhance the generalizability of findings. Having more jurisdictions that can emulate the Welsh approach to data curation and stewardship would be a boon for critical care data science.
Second, the use of maps in the study Szakmany et al (3) helps to illuminate its findings in a clear and rapidly interpretable visual format. To better highlight this advantage, Critical Care Medicine has generated an online data exploration tool to accompany the article, enabling readers to interact more directly with the dataset. This platform, which can be reached from the article’s webpage (https://lippincott.shinyapps.io/WelshStudy2/), is the latest in a series of online tools designed to connect readers to the data resources that increasingly underpin critical care research (11–14). In this case, data regarding ICU survival are combined with publicly available datasets describing a number of demographic, economic, and environmental features. Exploring the maps, the magnitude and velocity of declining survival rates are clarified, as are the slight variations by region.
The graphical merger of clinical, demographic, and geographic data employed in the study Szakmany et al (3) suggests ways in which other research areas related to ICU care might benefit from similar mapping approaches. These include questions about regionalization of specialized ICU services and about imbalances in outcomes within and between countries. Mapping tools may also prove useful in tracking outbreaks of relevant emerging diseases such as pandemic influenza and Ebola, with a number of web-based mapping tools already poised to be informative in this regard (15).
Although the findings presented by Szakmany et al (3) cannot fully account for the complexity of interactions that contribute to long-term survival, their study provides an early look at how care processes and outcomes related to critical care might vary at the regional and population levels. By literally putting ICU survival on the map, the Welsh study provides us a view of outcomes beyond our own communities and challenges us to identify and ameliorate factors contributing to these.
1. Gooch RA, Kahn JM. ICU bed supply, utilization, and health care spending: An example of demand elasticity. JAMA 2014; 311:567–568
2. Fowler RA, Abdelmalik P, Wood G, et al; Canadian Critical Care
Trials Group; Canadian ICU Capacity Group: Critical care
capacity in Canada: Results of a national cross-sectional study. Crit Care 2015; 19:133
3. Szakmany T, Walters AM, Pugh R, et al. Risk Factors for 1-Year Mortality and Hospital Utilization Patterns in Critical Care
Survivors: A Retrospective, Observational, Population-Based Data Linkage Study. Crit Care Med 2019; 47:15–22
4. Koch T. 1831: The map that launched the idea of global health. Int J Epidemiol 2014; 43:1014–1020
5. Fisher ES, Bynum JP, Skinner JS. Slowing the growth of health care costs–lessons from regional variation. N Engl J Med 2009; 360:849–852
7. Teno JM, Mor V, Ward N, et al. Bereaved family member perceptions of quality of end-of-life care in U.S. regions with high and low usage of intensive care unit care. J Am Geriatr Soc 2005; 53:1905–1911
8. Carr BG, Addyson DK, Kahn JM. Variation in critical care
beds per capita in the United States: Implications for pandemic and disaster planning. JAMA 2010; 303:1371–1372
9. Hill AD, Fowler RA, Pinto R, et al. Long-term outcomes and healthcare utilization following critical illness–a population-based study. Crit Care 2016; 20:76
10. Garland A, Olafson K, Ramsey CD, et al. A population-based observational study of intensive care unit-related outcomes. With emphasis on post-hospital outcomes. Ann Am Thorac Soc 2015; 12:202–208
11. Delahanty RJ, Kaufman D, Jones SS. Development and evaluation of an automated machine learning algorithm for in-hospital mortality risk adjustment among critical care
patients. Crit Care Med 2018; 46:e481–e488
12. Sweeney TE, Azad TD, Donato M, et al. Unsupervised analysis of transcriptomics in bacterial sepsis across multiple datasets reveals three robust clusters. Crit Care Med 2018; 46:915–925
13. Shappell C, Snyder A, Edelson DP, et al; American Heart Association’s Get With The Guidelines-Resuscitation Investigators: Predictors of in-hospital mortality after rapid response team calls in a 274 hospital nationwide sample. Crit Care Med 2018; 46:1041–1048
14. Salter R, Bailey M, Bellomo R, et al; Australian and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation (ANZICS-CORE): Changes in temperature management of cardiac arrest patients following publication of the target temperature management trial. Crit Care Med 2018; 46:1722–1730
15. Brownstein JS, Freifeld CC, Reis BY, et al. Surveillance Sans Frontières: Internet-based emerging infectious disease intelligence and the HealthMap project. PLoS Med 2008; 5:e151