Big data expert Pam Baker says that offense is the best defense when it comes to a big data strategy for problem solving in healthcare.
The most common problem thwarting big data projects in healthcare is the same as found in other projects: a tendency to play defense rather than offense. A defensive big data strategy is typically focused on score carding whereas an offensive strategy is more attuned to problem solving. In healthcare this tendency is arguably more obstructive and considerably more harmful than in other sectors.
Why playing defense is weak and ineffectual
An example of defensive strategy in healthcare big data projects is the recent rush to gather intrusive data on patients and potential patients such as is seen in gathering data from financial institutions to mine an individual’s purchases. These “life logging” activities are intended to score a person on risky behaviors such as poor food choices and alcohol and tobacco abuse.
Ultimately, such scoring leads to shifting blame to the patient for illness and injury rather than in developing a preventative measure, treatment or cure. Hence, it is a defensive play as it deflects responsibility from the healthcare provider or institution to the patient. In the U.S. for example, the new healthcare law places accountability for patient outcomes directly on providers. Shifting blame to the patient can relieve much of that responsibility, at least in a legal argument before a court or in a compliance argument before a governing body.
But this maneuver achieves little else as blaming the patient cures nothing.
Even if the information is used to change patient behavior, there will always be something in any given patient’s behavior that enables scapegoating. Humans are not robots. They can’t simply be reprogrammed to behave perfectly.
Further, the conclusions drawn from such data as product purchases is often deeply flawed. For example, a mother may purchase sugary foods at the grocery store because there are children in the house and not because she personally consumes it. Someone else may buy adult diapers for an elderly neighbor or family member rather than for their own use. And on it goes. Just because a product purchase is tied to a specific bank account does not tell you anything reliable about what the bank card owner is actually doing.
EMRs and EHRs as a big data defensive play
Another defensive big data play is found in the design of many electronic medical records (EMRs) and electronic health records (EHRs) systems. Typically these too are focused on score carding but this time the scoring is usually on physicians and the health team members, usually in regards to costs, compliance, use of resources, and productivity (or profitability). These technologies often intrude on patient care rather than enhance it and can even add costs to healthcare even though their intent is to reduce costs.
"The designs of many electronic health records do not meet the needs of physicians and too often detract from valuable time with patients,” said Christine Sinsky, M.D.
American Medical Association (AMA) Advisory Committee on EHR Physician Usability in a statement to the press.
“As a practicing physician, my desire is that EHRs will help me focus on patient care. They can do this by providing concise, context sensitive and real time data that is uncluttered by extraneous information. This will help in eliminating the current information overload and unnecessary administrative data entry that is overwhelming today's physicians and interfering with patient care."
Indeed, in the U.S. many physicians, particularly internists, hire extra staff just to follow them around and enter the data in the EMR. This is an additional healthcare cost in itself. But the burden continues beyond that. Often a patient’s last office visit can’t easily be found by the physician on the next visit, for example. The flow of information between providers and staff is often also impeded.
"The ultimate measure of a well-designed electronic health record is how it helps physicians take better care of patients,” said Gary Botstein, M.D.
AMA Advisory Committee on EHR Physician Usability in that same statement to the press.
“It is critical that enhancing quality patient care is the first priority of an electronic health record and data collection is second. Particularly for physicians in solo and small practices, digital data collection has become overwhelming and interferes with and detracts from time with patients."
Because the focus of many of these systems is on the business end of healthcare rather than on patient-facing care delivery, this is a defensive rather than an offensive tactic. It is defensive of resources, costs, liability, and productivity.
I’m not saying defense of these things is fruitless, as indeed they are necessary. Resources are not limitless, productivity does matter, and controlling healthcare costs is a must.
I am saying that an offensive big data play does all of this—and more—better than a defensive tactic.
Offense is the best defense
An example of an offensive tactic in big data use in healthcare is to escalate and leverage the study of factors we’ve always known to affect human health and then use that information proactively to better patient outcome, resource allocation, and profitability.
By controlling those factors, or predicting their effects if we can’t control them outright, we can prepare better treatments and cures; predict the need for specific resources; realign supply orders and chains; reallocate space for additional labs, services or beds; reset the list of healthcare services offered; staff more efficiently; and take other proactive measures to improve the lot of patients and providers both.
Examples of taking the offense
For example, mobile phone metadata can and does reveal patterns in any local or regional area such as a tendency toward alcohol abuse, high or low prevalence of specific diseases, high or low vaccination rates, fast food versus fresh food consumption, and other collective behavioral patterns in that population. In addition, that metadata is helpful in tracking and predicting disease spread because it exposes the movement and contact between people in the community.
An ongoing, real-time read on air and water quality, poverty levels, pest populations, weather conditions, housing and building environs, and other environmental factors are also useful in tracking and predicting human health impacts that can be used by providers to plan accordingly. Data from social media can further reveal emerging conditions in the area that are likely to affect human health and corresponding demand for specific health services.
Further, importing data from past clinical studies, DNA findings in disease susceptibility and treatment effectiveness, epidemics, pandemics, public health data, comparative studies on numerous patients with same health conditions and treatments for effectiveness measurements, and other such medical and health data can vastly improve both patient outcome and healthcare cost reductions.
Fortunately most of this data is easily and cheaply importable from other sources such as the World Health Organization (WHO) data repository and the U.S. Centers for Disease Control (CDC) scientific data repository as well as many other readily available and reliable sources.
Redesigning EMRs and EHRs and other data collection to achieve a better offense
EMR and EHR data collection and use can be reconfigured to focus on treatment effectiveness combined with cost and resource efficiency—rather than primarily focusing on cost, productivity and resource containment—to vastly improve patient outcomes, provider performance, and provider or institution reputation (which coincidentally attracts more business and improves patient compliance).
One thing that can be done to facilitate EMR and EHR efficiency is to improve machine data reporting. Instead of the burden of reporting being placed directly on the physician or staff to enter the data, machines can be programmed to report the data directly to the patient’s record. This would include vitals and medicine bottle scanners to diagnostic machines and everything in-between. Surgery sensors can record procedures and instrument counts. Robotic surgery devices can also automatically report the entire procedure. Electronic medicine delivery systems can report dosages and administration times.
A change in EMR design will also improve the data entry burden.
"User-Centered Design (UCD) is critical to advancing electronic health record usability to meet the cognitive and workflow needs of physicians,” said Raj Ratwani, PhD, AMA Advisory Committee on EHR Physician Usability. “While some electronic health record vendors have implemented UCD, their results have been inconsistent and many others do not utilize UCD."
Even so, data liquidity must be improved so all patient data is instantly readable by all physicians and healthcare workers charged with that patient’s care. It is this complete and instant access that will ramp up care and radically decrease costs the most.
“Lack of interoperability between the nurse and physician workflows in its electronic health record system was the reason Texas Health Presbyterian Hospital Dallas initially sent home Ebola patient Eric Duncan, according to the healthcare facility,” writes Alison Diana in her post in InformationWeek. “And similar interoperability issues could threaten other healthcare organizations, industry executives caution.”
Blocked data liquidity is dangerous to the patient, the organization, the provider, and the community.
Patient intake can similarly be streamlined with patient online or mobile check-in and integrated software that can automatically fill forms and flag anything that requires attention. Other patient processes can also be streamlined so that data entry and costs are minimized, delivery of care is optimized, and detections of abuses of the healthcare systems are easier and faster.
Why any of this matters
Data-driven healthcare leads to a brighter future for all of humanity, but healthcare-driven data is a nightmare for us all. In other words, data should be the servant and the healthcare provider the master, not the other way around.
A too-narrow focus on costs typically drives up costs, as we saw in the example above where doctors hire extra staff just to input data in EHRs and where doctors are required by hospitals to enter so much data that the quality of patient care actually falls. Further, productivity suffers because doctors are spending more time on data entry than in front of patients.
Doctors have become the servant of data. We have to change that so that data serves the doctors.
Further, proactive data strategies actually do improve patient care AND reduce costs and resource drains. And that, ultimately, was what we were all trying to do in the first place.
That’s why shifting from a defensive, score carding strategy to an offensive, proactive problem solving strategy is not only optimal, but absolutely necessary.
Pam Baker is a regular nuviun contributor, the editor of FierceBigData and author of Data Divination: Big Data Strategies. For more expert insights from Pam, follow her on Twitter @bakercom1 and at FierceBigData.
The nuviun blog is intended to contribute to discussion and stimulate debate on important issues in global digital health. The views are solely those of the author.