In Victorian London, people were used to death. But the cholera epidemic of 1849 had people afraid to breathe, believing that cholera was an air borne illness. One letter from the time describes a hard-hit area, “The inhabitants themselves show in their faces the poisonous influence of the mephitic air they breathe. ”
John Snow, MD, learned otherwise. When he placed small rectangular bars representing cholera deaths on a map of London, it showed clearly that the deaths were clustered around an infected water well. The day officials removed the handle from the Broad St. pump in London was the last day of the cholera epidemic, which had taken over 50,000 lives. Snow’s map is now included among important texts of information visualization and is considered the birth of epidemiology.
Today, the press has been focusing on another epidemiological concern, this time ‘hospital acquired infections.’ The stakes are high both for patient care and for the hospital with the advent of bundled payments; yet until recently it seems that the data analysis behind this was like something out of Victorian England. Reporting on Health has an article asking “Why is there such an accurate, granular map of heads of cattle but not of MRSA infections?” Consumers Union is working on data collection to garner the attention that will help hospitals implement practices that will make MRSA as rare as cholera in the United States. According to Consumers, "Hospitals have started reporting surgical site infection rates to the CDC and that information will be posted on Hospital Compare every quarter beginning in 2013 ...Public disclosure drives hospitals to improve care and helps patients choose hospitals with better safety records.”
Even with this progress, results can be hard to measure. Not only is there a tremendous disincentive for hospitals to report hospital acquired infections, there also is tremendous disagreement about which infections are acquired in the hospital. It doesn’t have the binary, clear quality of something easier to measure (for example, “dead by cholera” in 1854 London). Even mortality rates at hospitals, which seem to be very clear cut, are subject to debate.
Techniques like crowdsourcing, log files of online behavior, and the accessibility and organization of data from EMRs are considered a huge opportunity for emergent discoveries across the Acute Care Continuum.
Uniformity of data collection across organizations is difficult even if the task is clear. For example, in my department we measure turnaround times to discharge, and we know that some hospitals collect arrival times, and some only collect triage times. Triage times are often 15-30 minutes later than arrival times. We are knowingly ranking hospitals against each other and against a goal where superior data collection leads to a lower score, and we have to console ourselves knowing that the hospitals who perform better data collection have only the humble opportunity to authentically improve for their patients.
But with ingenuity, opacity of measurement does not make visualization impossible. Less than 20 years after John Snow’s cholera map, a famous visualization by Florence Nightingale is considered the impetus for improved cleanliness in hospitals. She used a pie chart to illustrate seasonal sources of patient mortality. She presented her charts to Members of Parliament and to civil servants, some of whom were illiterate and needed her pictures to help them understand the situation. Her work had dramatic life-saving results for soldiers: "After 10 years of sanitary reform, in 1873, Nightingale reported that mortality among the soldiers in India had declined from 69 to 18 per 1,000."[33]
Adding today’s technology to the techniques used by Snow and Nightingale can produce powerful results. As described in “The Hot Spotters”, Jeffrey Brenner, MD, produced a map detailing buildings with the highest accident and violence rates in Camden, NJ. In order to do this, he transferred data from the medical billing records of three Camden hospitals onto a computer, and created a searchable database. He then looked at records of the ED visits of victims of serious assault, and used the multiple sets of data to create maps where these patients had lived. He concluded, “Between January of 2002 and June of 2008 some nine hundred people in the two buildings accounted for more than four thousand hospital visits and about two hundred million dollars in health-care bills.”
His map made it possible to place health resources exactly where needed, resulting in both accessible timely care and huge reductions in healthcare bills. By taking advantage of data that was already collected and organized, Dr. Brenner used his creativity and modern technology to generate important conclusions without having to incur the expense and effort of hitting the pavement himself or organizing teams of data collectors.
In the future, we can expect other resourceful researchers to use the riches of EMRs, billing databases, and other data sources to generate pictures that will help solve the well of problems we face today.