How Population Health Management and Big Data Can Rock Your World

Hospitals and health systems like yours have been aggressively pursuing a range of information systems over the last several decades. Cited goals are often efficiency, lower costs, better decisions, and better patient outcomes. But how do these systems purportedly lead to population-level improvements in care? In this column, we address the connections that are anticipated as well as challenges to be expected along the way.Let’s start with some definitions. Population health management, according to a leading outcomes management provider, is the “aggregation of patient data across multiple health information technology resources, the analysis of that data into a single, actionable patient record, and the actions through which care providers can improve both clinical and financial outcomes.”¹ While we like this definition because systems are used in a way that patient care is provided to individual patients, we think that the intelligence gained from each patient encounter can concurrently be applied throughout the continuum of care for any population being served. Big data is a buzzword in health care, even though other industries have been using the analysis of huge quantities of digitized data for many years. In health care, the rapid adoption of the electronic health record (EHR) provides an opportunity to finally having a real chance for improving health outcomes and controlling costs.The definition of big data varies, but we will define it as the “ability to access and analyze information that holds the key to more efficient, higherquality health care while significantly shortening the time between research and translation into practice.”² Big data is made possible because health care is now moving toward being a real digital enterprise to leverage the collective power of information. In our examination of health system technology devices that have been deployed for the last 10 years, we discovered that some had the ability to be networked but many were not. The EHR can now be the data hub for providers while supporting care provision by consolidating and analyzing these digital warehouses of real-time data to discover trends and make predictions.In a previous column, we described these processes as enterprise performance management. At a strategic level, a health system would generate critical success factors and key performance indicators that would lead to outcomes improvement. At an operational level, data would be gathered as a byproduct of rendering patient care to determine how well these indicators of success were being met. The system would generate e-mails to managers to give them feedback on any success factors assigned to them. Exception reports could include deficiencies, meeting of goals, and exceeding expectations. When best practices were identified within the enterprise, the methods being utilized to exceed expectations could then be used to address the problems experienced in units where expectations were not being met.In our experience, niche industries are being generated by the inability of EHR vendors to address both the developmental needs to improve their core product for its primary purpose of patient care and to add all of the population health and data analysis capabilities required. Add to this the fact that the individuals who are needed at the health system level to work with data analysis are the same people that Google and Microsoft are recruiting as quickly as possible. Thus, entrepreneurs look at the needs of health care and bring the skills and expertise necessary to the task. The expectation is that the EHR vendors who are going to cooperate by providing the needed data will eventually wrap the capabilities of these consultants into the everyday functions found in their systems.The complexity inherent in population health management is quite high. The data sources and their divergent information standards bring about the first challenge. Again, starting with a specific EHR, integration or interfaces must be established with any ambulatory electronic medical record being utilized by employed or affiliated providers. Each of these medical records could utilize one of 10 standards to include HL7, CCR, CCD, and so on that will need to be translated and normalized to be of any use for analysis. Next, we have separate computerized prescriber order entry systems, labs, imaging, health information exchanges, payers, and claims data. Each of these data sources must be integrated and normalized before they provide any real utility.Now we need to talk about clinical decision support systems. As a provider, you are probably already aware of the problem we call flag fatigue where alerts and warnings interrupt your provision of care for your patients. The challenge for an enterprise decision support system will be to ensure that the right provider is involved in the appropriate intervention at the appropriate point in the care process in the appropriate facility for the appropriate patient at the appropriate time. Get your mind around this complexity. Now think about multidisciplinary care team coordination and communication. How are we going to know who did what, when, and how?Alerts that are needed in population health management can also start when care gaps are identified. They can start when a patient steps on a digital scale that transmits a 10-lb weight gain due to heart failure–related edema. The alert may take place because patient outreach is indicated and an assignment for this task must be made. Action may be needed due to a patient’s entry in a notes section of a patient portal. Alerts may occur because quality reporting is either missing or the values entered have triggered the need for a response.Right now, we’re spending most of our time putting these data in and straddling the current reimbursement system that is so heavily based on fee-forservice care provision while preparing for anticipated, future ways of providing care. To understand how life will be different as these changes take place, look at those health systems that have already gone through significant population health management transitions and who use big data routinely to improve their operations.We have been attending presentations by health systems that have started with the care provision of their own employees as a way to get some small population experience in the area and then moved on to larger populations they were able to attract. Just Google “population health management” and explore testimonials on how care provision has changed among these frontrunners. Some will definitely rock your world or at least give you a few “ah-ha” moments. We would enjoy hearing your comments and questions on this topic. You can reach Bill at felkebg@auburn.edu or Brent at foxbren@auburn.edu.     

IT Maximization for Population Health Management

Those of you who have been in the health-system setting for several decades have seen many changes. Some of the changes originate within the care setting, whereas many others are brought on by external circumstances. Even those who have been in health systems for only 5 years can recall a recent change in the organization that greatly impacted pharmacy. More change is coming. In this installment, we explore critical technology-related changes of which you should be aware.Health systems are increasingly finding themselves involved in something they describe as “straddling.” They are straddling current reimbursement drivers and practices while concurrently getting ready for population health strategies to replace what has been the norm for decades. We have been saying for years that we cannot imagine a scenario where we will use less technology in any foreseeable future. When it comes to operating a health system focused on managing populations, we cannot imagine anything more important than the effective use of the following top 5 technologies.     

Microbiological,biological, and chemical weapons of warfare and terrorism

Microbiological, biological, and chemical toxins have been employed in warfare and in terrorist attacks. In this era, it is imperative that health care providers are familiar with illnesses caused by these agents. Botulinum toxin produces a descending flaccid paralysis. Staphylococcal enterotoxin B produces a syndrome of fever, nausea, and diarrhea and may produce a pulmonary syndrome if aerosolized. Clostridium perfringens epsilon-toxin could possibly be aerosolized to produce acute pulmonary edema. Ricin intoxication can manifest as gastrointestinal hemorrhage after ingestion, severe muscle necrosis after intramuscular injection, and acute pulmonary disease after inhalation. Nerve agents inhibit acetylcholinesterase and thus produce symptoms of increased cholinergic activity. Ammonia, chlorine, vinyl chloride, phosgene, sulfur dioxide, and nitrogen dioxide, tear gas, and zinc chloride primarily injure the upper respiratory tract and the lungs. Sulfur mustard (and nitrogen mustard) are vesicant and alkylating agents. Cyanide poisoning ranges from sudden-onset headache and drowsiness to severe hypoxemia, cardiovascular collapse, and death. Health care providers should be familiar with the medical consequences of toxin exposure, and understand the pathophysiology and management of resulting illness.     

Are lipid-lowering drugs also antiarrhythmic drugs? An analysis of the Antiarrhythmics versus Implantable Defibrillators (AVID) trial

OBJECTIVES: This study sought to evaluate the antiarrhythmic effects of lipid-lowering drug therapy as assessed by ventricular tachyarrhythmia (ventricular tachycardia [VT]/ventricular fibrillation [VF]) recurrences recorded by an implantable cardioverter defibrillator (ICD) in patients with atherosclerotic heart disease (ASHD). BACKGROUND: Randomized trials of lipid-lowering drugs suggest reduction of sudden death (SD) in patients with ASHD. Because SD is usually secondary to VT/VF, this observation suggests that lipid-lowering therapy has antiarrhythmic effects. METHODS: The probability of VT/VF recurrence in patients with ASHD treated with an ICD in the Antiarrhythmics Versus Implantable Defibrillators (AVID) trial who did not receive lipid-lowering drug therapy (n = 279) was compared with that in patients who received early and consistent lipid-lowering therapy (n = 83). In addition, all-cause mortality and cardiac mortality of all patients in the AVID trial with ASHD who did not receive lipid-lowering therapy (n = 564) were compared with that of those who received early and consistent lipid-lowering therapy (n = 149). RESULTS: Using multivariate analyses, lipid-lowering therapy was associated with a reduction in the relative hazard for VT/VF recurrence of 0.40 (95% confidence interval [CI] 0.15 to 0.58) (adjusted p = 0.003) in the ICD subgroup. Lipid-lowering therapy was also associated with a reduction in the relative hazard for all-cause mortality of 0.36 (95% CI 0.15 to 0.68) (adjusted p = 0.03) and a reduction in the relative hazard for cardiac mortality of 0.39 (95% CI 0.16 to 0.78) (adjusted p = 0.04) in the larger study population. CONCLUSIONS: In patients with ASHD who have received an ICD, lipid-lowering therapy is associated with reduction in the probability of VT/VF recurrence, suggesting that part of the benefit of lipid-lowering therapy may be due to an antiarrhythmic effect.     

Risk factors for active trachoma and Chlamydia trachomatis infection in rural Ethiopia after mass treatment with azithromycin

Objectives To investigate risk factors for ocular Chlamydia trachomatis infection and active trachoma, comparing communities receiving or not receiving an intervention programme of community‐wide azithromycin treatment and health education. Methods In a 3‐year post‐intervention follow‐up survey, 1722 children aged 3–9 years, from randomly selected households in 37 communities, were examined for signs of active trachoma and had samples taken to test for ocular C. trachomatis by polymerase chain reaction. Multivariate random effects logistic regression analyses considered interventions at community level, adjusting for other independent risk factors as appropriate. Results Younger age, ocular discharge and flies on eyes were risk factors for active trachoma in communities with and without antibiotic treatment. After azithromycin treatment, odds of active trachoma were lower in children aged 6–9 years than in children aged 3–5 years (OR 0.48, 95% CI: 0.36–0.66) and higher for children with ocular discharge (OR 4.5, 95% CI: 2.6–7.7) or flies on their eyes (OR 2.5, 95% CI: 1.6–3.7). Odds of C. trachomatis infection were lower in children aged 6–9 years than in younger children (OR 0.47, 95% CI: 0.23–0.96); and in children who received 2 or 3 doses rather than 1 (OR 0.26, 95% CI: 0.08–0.88). Conclusions In communities that received or did not receive the mass antibiotic treatment, the same risk factors for C. trachomatis and active trachoma were identified. Education and environmental improvements need to supplement antibiotic campaigns in order to positively impact on these remaining child level risk factors.     

Adenosine versus regadenoson comparative evaluation in myocardial perfusion imaging: results of the ADVANCE phase 3 multicenter international trial.

BACKGROUND: Earlier phase 1 and 2 studies have shown that regadenoson has desirable features as a stress agent for myocardial perfusion imaging. METHODS AND RESULTS: This multicenter, double-blinded phase 3 trial involved 784 patients at 54 sites. Each patient underwent 2 sets of gated single photon emission computed tomography myocardial perfusion imaging studies: an initial qualifying study with adenosine and a subsequent randomized study with either regadenoson (2/3 of patients) or adenosine. Regadenoson was administered as a rapid bolus (<10 seconds) of 400 mug. The primary endpoint was to demonstrate noninferiority by showing that the difference in the strength of agreement in detecting reversible defects, based on blinded reading, between sequential adenosine-regadenoson images and adenosine-adenosine images, lay above a prespecified noninferiority margin. Other prospectively defined safety and tolerability comparisons and supporting analyses were also performed. The average agreement rate based on the median of 3 independent blinded readers was 0.63 +/- 0.03 for regadenoson-adenosine and 0.64 +/- 0.04 for adenosine-adenosine-a 1% absolute difference with the lower limit of the 95% confidence interval lying above the prespecified noninferiority margin. Side-by-side interpretation of regadenoson and adenosine images provided comparable results for detecting reversible defects. The peak increase in heart rate was greater with regadenoson than adenosine, but the blood pressure nadir was similar. A summed symptom score of flushing, chest pain, and dyspnea was less with regadenoson than adenosine (P = .013). CONCLUSIONS: This phase 3 trial shows that regadenoson provides diagnostic information comparable to a standard adenosine infusion. There were no serious drug-related side effects, and regadenoson was better tolerated than adenosine.     

Characterization of the Zebrafish atxn1/axh Gene Family

In mammals, ataxin-1 (ATXN1) is a member of a family of proteins in which each member contains an AXH domain. Expansion of the polyglutamine tract in ATXN1 causes the neurodegenerative disease, spinocerebellar ataxia type 1 (SCA1) with prominent cerebellar pathology. Toward a further characterization of the genetic diversification of the ATXN1/AXH gene family, we identified and characterized members of this gene family in zebrafish, a lower vertebrate with a cerebellum. The zebrafish genome encodes two ATXN1 homologs, atxn1a and atxn1b, and one ATXN1L homolog, atxn1l. Key biochemical features of the human ATXN1 protein not seen in the invertebrate homologs (a nuclear localization sequence and a site of phosphorylation at serine 776) are conserved in the zebrafish homologs, and all three zebrafish Atxn1/Axh proteins behave similarly to their human counterparts in tissue-culture cells. Importantly, each of the three homologs is expressed in the zebrafish cerebellum, which in humans, is a prominent site of SCA1 pathogenesis. In addition, atxn1a and atxn1b are expressed in the developing zebrafish cerebellum. These data show that in zebrafish, a lower vertebrate, the complexity of the atxn1/axh gene family is more similar to higher vertebrates than invertebrates with a simple central nervous system and suggests a relationship between the diversification of the ATXN1/AXH gene family and the development of a complex central nervous system, including a cerebellum.