Brian Altonen, MPH, MS

Part I.  Introduction

Over the past twelve years, I have worked on a number of electronic medical records systems, exploring their uses primarily as a spatial analyst.

Now of course, my employment for each of these positions was not as a spatial analyst.  Rather, it was as a typical research associate, quality improvements specialist or compliance officer, data miner or analyst, BI or population health specialist, working for a state MCD/MCR/CHP/COM/Insurance program devoted to assessing the health of anywhere from 50,000 to 100 million members or patients.

Each time, I was engaged in the exploration, mining and production of baseline reporting at the population health level, there were several aspects of these systems that had to be corrected for in order to produce a highly effective spatial analysis tool for medical GIS work.

What stands out about these requirements for engaging in the spatial analysis of population health data is the ease at which such a program could be initiated.  The fact that population health analysis is not carried out in any significant fashion by nearly all healthcare programs, insurance companies and PBMs, tells me that there is some sort of barrier to learning this new technology.  However, working alongside others in this EMR industries, it becomes obvious to all of us that these barriers are linked primarily to middle and upper level management.

If you ask anyone familiar with GIS what such a system can do, you will get an endless list of examples, and most likely be told to go back to the library and look it up, since obviously you are not in touch with this new technological field.

If you ask the same of a healthcare technician, you have some chance of connecting with someone who is either familiar with or experienced with the values GIS offers the healthcare profession.  Any sizable corporation with sufficient medical records, data entry and analytic staff, will probably have two or three individuals familiar with GIS utilization, mostly at the descriptive level.

People engaged in this task, or experimenting with its use, in the standard Cognos or Tableau tools will have some familiarity with how maps can be used as part of a powerful reporting tool.  A more specific use of this technology could once again be lacking in these individuals.

If you ask someone associated with upper level Watson and EPIC tools are unlikely to be familiar with the value of surveillance mapping, at least at the professional, experiential level.

In recent months, there have been outbreaks monthly that could be evaluated using spatial analytic tools.  Some of these outbreaks were very local, like the NYC Legionnaire’s outbreak.  Others have been of national significance, like the Ebola transmission worldwide and the “discovery” that measles vaccinations programs have become a failure in recent decades.  Still other outbreaks have emerged as regional concerns, their distributions predictable and explainable with GIS, by the use of climate, weather, temperature and human ecology mapping methods.

Very few of the tools we currently utilize, originating from the above defined businesses, are capable of depicting the detail of these outbreaks, defining their routes of passage and potential places of origin, or leading to the production of a valid predictable modeling algorithm for detailing their routes of travel, and times of passage versus outbreak.  The current tools being used to handle managed care EMR serve mostly and introspective to retrospective surveillance tools, with just enough resolution to predict models at a fairly reasonable area-size defined level–namely the county, township or zipcode tract level.

Yet, an HIT-GIS is capable of producing valid population and public health data at the small area, very local, neighborhood-based, prediction modeling level.  Exact routes, possible places (including homes) for outbreaks, and even future pathways of travel and diffusion across the country, can be evaluated and defined using the right tools and the right algorithms.

In the typical GIS research program environment, if you ask individuals for examples of these uses, you can easily come up with several dozen examples.  In the health care environment, if you asked the same of these analysts, the experts will more than likely be the public health and epidemiology experts, and teams devoted to disease surveillance, prevention and healthcare maintenance programs, it is atypical to find a program engaged in much more than a couple dozen different types of analyses (PPACA requires 40-6o about).

Still, it is possible to use spatial analysis for managed care population health management and surveillance, for which hundreds to thousands of population health statistics are generated monthly, weekly or even daily. (Although, “thousands” might be pushing it!)

In a number of places, I published my lists of reports that I know by experience are possible.  In particular, I am interested in reports that focus on the most important issues, like social sciences and health, health education/health promotion, cost/savings, inequity and healthcare practices, and quality of care rights for patients.  A program that cannot meet the basic metrics proposed by PPACA (the “obamacare” plan) are in essence a failure when it comes to reaching the agency’s fullest potential.

This means the one third of insurance companies like BCBS, UHC, etc. that have or are about to fail to make the ends meet regarding the ongoing higher costs for care, are in fact the lowest third, the “last leg” in the innovation curve, the worst examples for the managed care insurance profession.

These problems in healthcare, healthcare insurance and certain managed care programs are reviewed and supported by other examples of my work at:

• httpd://brianaltonenmph.com/gis/populations-and-managed-care/applying-gis-to-managed-care/        An overview of this process that I developed years ago.
• https://brianaltonenphd.wordpress.com/2015/10/25/orchestrating-the-hit-gis/     A general overview of why to engage in this new form of managed care HIT-GIS population health monitoring.
• https://brianaltonenphd.wordpress.com/2015/10/26/more-on-the-plan-for-hit-gis/   The method for implementing quality of care and prediction modeling algorithms as a standard part of the MC HIT-GIS system.  In my current position, I have the tool for analyzing any population, racially, ethnically and age-genderwise, for any ICD subclass applied to overall population health analyses, and special ICD subclasses predefined by past researchers for the evaluation of individual health risks (Charlson Score, Elixhauser score etc.).
• https://brianaltonenphd.wordpress.com/2015/10/19/a-policy-for-the-development-of-a-hit-gis-system-in-managed-care-dissertation-related-work/     This delineates further the steps an agency, business and/or healthcare facility need to undergo to develop a spatial medical GIS program as part of its institution-wide programs.  Ironically, and quite unfortunately, the way managed care works, is that policies prevail when making important changes.  Adding GIS to a managed care system in order to explore thousands of health metrics per day seems to be asking quite a lot of these programs.  The irony is, it takes only a few hours to run a well-scripted program that can pull data, evaluate it, compare it, undergo statistical analysis, produce tables, graphs and maps, and then output these in logical format into a powerpoint or word report that can be mailed to management–ON A DAILY BASIS.   Such a process allows specific days of the week and month to be designated for producing and distributing specific reports, like those devoted to program comparisons (i.e. Monday through Friday as: MCD, MCR, CHP, the Rest, ALL), cultural health (White, African American, Asian, Other, All), special age group studies (Genomic diseases, Developmental, Childhood, MidAge Health, Elderly Health), Culturally bound and rare diseases (White vs. others, Black vs. others, Hispanic vs others, Asian vs. Non, Outstanding or Unusual rare ethnic groups), SES patterns (Area specific reporting, program specific reporting for CHP and other programs successes, Low Income needs special services, QOC, visits and noncompliance reporting for each of these groups, Upper income group health care strategies compared with lowest income groups), and with a financial focus (comparisons across the board to cost, relative to all in-hospital, emergent/urgent care, preventive care/visit procedures, effects of cost on follow up to recommendations, impacts of cost to specific regions, relative to annual amounts allocated to each facility/region.)
• https://brianaltonenmph.com/2015/04/15/hex-grids-are-essential-to-developing-a-more-effective-medical-gis-workstation/   This is the exploratory, introductory page to a new technique that I developed the algorithms and math for in 2004.  After posting mention of that algorithm in 2009, the number of people visiting my page skyrocketed almost overnight.  I suspected I hit upon a growing popular topic, and after a few months of seeing the following grow, decided to release my algorithm to the public (I was saving it for a job opportunity), since the GIS community in the workforce setting was still catching up with the basics.  Since then, the Hexagonal Grid page and the Download page for this tool (the original Excel I used, which I keep promosing to upgrade soon to an SQL and SAS version as well), has been viewed by about one fifth of my visitors–the major country interested in this technology is, of all places, Canadian Urban development GIS specialists.  Most other managed care or even traditional health care agencies have yet to understand the value of being 27% more correct in the tool that you use for spatial analysis (which I prove at https://brianaltonenmph.com/gis/population-health-surveillance/grid-mapping-disease-in-the-united-states/ ).
• https://brianaltonenmph.com/2015/10/29/mcs-goal-for-2016-solving-the-financial-crisis-of-healthcare/     My criticism of the current system regarding its inability to develop a comprehensive spatial analysis tool and program.  This criticism is not only directed at the programs and overseers (PPACA, NCQA), but also the companies, COs/Pres/VPS, and other agencies with the ability to bring PPACA into a new generation of HIT-GIS success.
• https://www.linkedin.com/pulse/algorithms-brian-altonen   In 2015 I transitioned from national health spatial analysis to densely populated, megalopolis regions analytics processes.  This transition enables me to retest my algorithms at the detailed spatial level, even street and neighborhood level.  A number of highly controversial topics were evaluated in the very beginning, in particular local data confirming the asymmetric behaviors of some genomic  disease states, like the male-female survival rate differences noted for one commonly carried disease.  This work also demonstrated THE ROLES THAT RELIGION DOES PLAY on whether or not certain religious families are more likely to bring their patient in for unusual disease screenings and rule outs.  It also allowed me to confirm my suspicions about most of the culturally-linked diseases I have identified, and the unique religious-ethnic post-violence behavioral patterns of patients when it comes to reporting child and adult/wide/spousal abuse; in other words, some religions are more likely to report than others.
• http://www.scoop.it/t/an-episurveillance-researchers-guide/p/4022245870/2014/05/30/why-innovation-in-health-care-is-so-hard-harvard-business-school-ge     A brief recital of the six forces that influence managed care and race-ethnicity focused research approaches, to date.
• https://brianaltonenmph.com/gis/populations-and-managed-care/big-data-and-managed-care/    This page starts with an overview to the value of big data in facilitating what I termed “ACER” analysis of patient populations.  At the bottom are links to my pages on how to design to design a cultural health analytics and public health screening program that is more complete in its approach than the typical NCQA program.  Contemporary HEDIS and NCQA interpretations of the value and meaning of health and EMR analysis for non-caucasion groups is, needless to say, still too “ethnocentric” or focused on white population health metrics.
• https://wordpress.com/stats/insights/managedcareinnovations.wordpress.com    As the title implies – – my innovations.  Or to be more exact, those “secret” IP related sql and SAS codes that I use to generate my 3D mapping results without GIS.  The majority of algorithms you’ll see at the methods I developed for reclassifying data.  The most unique algorithm is how I recode the religious religious groups, to put those with similar philosophies together in terms of how they relate to disease, its meaning and purpose in life, the reason for healthcare practitioners and their responsibilities.  There are also algorithms here for ICD regroupings (of which I use many) for risk related analyses, the three standard risk scoring algorithms for patient risk assessment (Charlton, Elixhauser, and the Federal Chronic Disease Score), infectious and zoonotic disease surveillance reclass algorithms, race-ethnicity reclass, several culturally-linked and -bound ICDs, etc. etc. Click on the ‘Codes and Population Health’ for a drop down with more on these.
• https://nationalpopulationhealthgrid.com/author/altonenb/   My site devoted to just the non-GIS spatial mapping technology.  How to produce spatial imagery without depending upon a GIS or GIS related cost.  This process requires beginners to intermediate level experience, with intermediate SAS programming experience preferred for complex comparisons between race, ethnicity, gender, religion, age range, gender, SES status, and area.
• https://brianaltonenmph.com/about/surveillance-3d-modeling/   A site I developed to document this work on the internet.  This was produced following a heated “interview” with on the the nation’s primary EMR companies, responsible for developing HIT systems for numerous managed care facilities and companies.  In essence, this company felt that HIT-GIS was not needed for this profession and the healthcare process to develop, which I contested.  This company simply stated my technology would go nowhere.  Its rivals, also in contact with me, threatened to develop their own IT process for producing these end products devoted to spatial analysis of population health.  Those arguments by the way occurred in 2012.
•  https://nationalpopulationhealthgrid.com/   To date, no major company has been able to produce an effective HIT-GIS tool for analyzing data down to the square inch of an urban setting.  This site was developed and opened up in order to document my discovery of this algorithm and process for 3D modeling of disease and public health patterns, by then a nearly ten year old skillset.
• https://www.surveymonkey.com/r/HZ7MH7Q?sm=Dw%2bY%2fHiQ3dYv0ZjM%2fFbOSA%3d%3d    I developed a survey (long, 25+ questions) to document the details about the use (lack thereof) for GIS in the managed care or university hospital/clinical workplace setting.  This is the link to that survey.

Part 2.  Creating your own Managed Care Population Health Surveillance HIT-GIS Program

To produce an effective MC HIT-GIS program, the following steps and actions are recommended.

Step 1.  Evaluate.  Determine, Define and Describe Needs and Potentials.

• Prepare the EMR for spatial analysis and Medical GIS performance.

To perform spatial analysis, two things are needed:

1. location or spatial information, preferably in the form of  a latitude-longitude dataset for each member of the patient population (we’ll skip lat-longs for the providers and facilities for the moment).
2. An analytic tool or process in place that enables researchers to produce spatial information about these data, in the form of descriptive, exploratory, summary visualization products, namely the production of maps depicting the  content and meaning of the environmental, demographic, health, disease, service and financial data stored within the system.

These are the only two additions to an EMR that are required for Medical GIS to be incorporated into a managed care [MC] health information system.

According to the Precision Medical Initiative (PMI) about to be underway,  managed care[MC] system will ideally have several spatial analytic processes in place and being used or experimented with.

The current ways Medical GIS is being promoted is via a single platform, single analytic system process.  Large companies like IBM, Cerner, have platforms in which patient care can be entered, kept track of, monitored, and even reported using primarily their tool.  Reasonable approaches for these tool enable some related company-invested or non-invested software  packages to be included, such as by enabling parts of the ESRI GIS process to be meshed or engaged in as part of the larger HIT platform.

Multiple tools however mean the data itself has to be utilizable as well by other unattached software products or packages.

SAS spatial analysis (versus SAS-GIS), for example, is an easy route to take in producing semi-automated, highly powerful analytics that cannot be produced using the common platforms and packages now being promoted.  SAS alone can be used to produced visualizations that neither EPIC, nor most traditional GISs, nor IBM Watson-Cognos, nor Alteryx-Tableau are capable of producing.  SAS has the ability to produce these visualizations within the everyday work setting at hundreds to thousands of objects (products or maps) per day.  The advantage to self-scripted queries using this method is that the math is traceable, provable, and can be modified or upgraded with ease.  There are no limits to the type of math use, nor the details at which the data are developed.  Whereas very large platforms provide the data at an areal level that is hard to change (i.e. attached to town and zip centroids, usually not to patient place of work or stay), it is up ti the developer/user of SAS programming and tools to determine the limits of the applications for any new algorithms that are developed.

Step 2. EMR/HIT Preparation.

• Of what value is the current HIT EMR system?

There are a number of related factors for HIT and the EMR that impact whether or not MC can develop a functional GIS.  Nearly all of these are frequently reviewed and re-reviewed in respect to HIT, PPACA requirements, and the EMR.

Quality of data within the EMR is the most basic example of this.   We have the three most basic problems with EMR data:

• absence or presence of data,
• data content and structure, and
• data form.

Arguments concerning the absence or presence of data are obvious.  Data can be missing because they were not included in the plans to develop a system, or they were included, but their content, structure and format weren’t well planned and so they are inconsistently entered, or they are not entered for other reasons, such as

• a lack of engagement at the data entry (office or clinician) level,
• a lack of understanding of why data are important, resulting in reduced query for the data (worker:patient relationship), and subsequent entry of that by employee, or in some cases by the patient, and
• the unwillingness of the clinician to ask a question or unwillingness of the patient to answer the questions needed for the EMR to be of maximum use.

In a recent review five very densely populated urban government boundary settings within a single megalopolis contained within a single state border, one of the five areas had a 45% lack of data completion when compared with the others, that missed only 20% of the data or much less.  Any of the above three reasons, or a combination thereof, may be responsible for this severe lack of important population health related background data on the patient population.

Step 3.  The Planning Stage.

• Plan. Plan!  PLAN!!!

Developing a plan for the implementation of a Medical GIS for a managed care system requires that measures be developed and engaged in at multiple levels.

Contemporary MC systems use their EMR to monitor population health in accordance with PPACA requirements.  During the past ten or fifteen years, the ‘meaningful use’ [MU] part of the PPACA was preceded by quality improvement activities (QIAs), quality improvement programs (QIPs), performance improvement projects (PIPs).

In a managed care system where the plan is to monitor an entire population health, the HIT-EMR and Medical GIS processes must include projects focused on the individual patient, patients’ quality of life, disease groups, surveillance/monitoring, service industry, providers/programs, health care specialty, interventions, public health, environmental health, “precision”-based (genomic), regional, community, neighborhood, special interest group,  financial, and government/potential investors levels.  PPACA, HEDIS, NCQA programs alone do not satisfy this more complete population health focused approach to MC programs.  The “Big Data” version of EMR utilization requires that all these levels be assessed.

Step 4.  Implement HIT-GIS.

Implementation of a successful managed care HIT-GIS requires that a number of methods of analyses be established, maintained, further developed, and improved on a regular basis.  Ideally, these plans are somehow monitored and/or managed by a single director who is responsible for maintaining the breadth at which Medical GIS in being engaged in, and the various values it can be used to demonstrate and the technology, IT, HIT, mathematical, statistical, BI, and institutional corporate level.  BI, EMR, Medical GIS may together be used to produce an effective predictive modeling program for use in redirecting and improving upon long term quality of care standards and the resulting cost-effectiveness of care relationship.

Step 5  Testing and Production.

1. Diversify the implementation, testing and production process.
2. Have two levels of engagement established.
3. Utilize a variety of methods to establish spatial analysis as part of the MC program.

There are many HIT-EMR processes that are required of programs, such as the HEDIS, PPACA MU and internal QOCs/QIs used to define the value for a MC program.  These standard, usually annual assessment processes occur and recur at and institutional level, and should be integrated with a Medical GIS program developed to meet institutional needs.

Step 6 (if necessary).

Contact me!

Part 3. Critique and Conclusion

Can GIS be included in the next phase of Managed Care development, and the program known as the Precision Medical Initiative (PMI)?

The value of GIS in the PMI is that for the first time, we can assign particular human genomic traits related to illness to a map.  The mapping of illness has been around for several centuries.  When it was first developed in the 18th century, there were few comparisons made between cultures, and the focus was on latitude, climate and disease.

But during the 1800s, physicians went beyond the latitude theory for disease and began incorporating topographic and meteorological features into their pre-bacterial, pre-microbe way of understanding disease.  Once the bacteria was discovered and proven, our focus on disease patterns and causes turned to the microcosm of the human body, and focused on our anatomy, physiology and cellular chemistry.  All of these in turn were related to the bug, germ, microbe, bacterium, virus, what have you, that came to infect us, cause us to become ill.

With the human genome project now to the point where genetic counseling has led to the development of genomic pathway treatment modalities, the older notion of race and temperament, those causes which led to Asians and eastern Europeans bearing one type of illness, and the Prussians another, is now back in the medical literature to such an extent that “mapping” disease bears that double entendre–mapping the gene within and mapping the disease linked to that and where they are in the world, where they evolved as a part of some local society.

The Precision Medical Initiative is going to make GIS an even more important part of the HIT system, combined to form a HIT-GIS.  HIT=GIS programs may in fact have some of their predecessors already in place in the medical profession.  The younger more innovative, potential management and CO employees are more valuable in the long run than the current leaders of Managed Care.  Unless a CO, VP or President of a MC system is learned enough to profess expertise in how to design and HIT-GIS to successfully produce both microcosmic and macrocosmic maps depicting human health, such leaders may soon be seriously outdated in their intellect and understanding of the architectural design of human health, population health, genomic health.

There are many institutions that have smaller special topic, exploratory or experimental, and even self-defined HIT-GIS groups established that are testing the value of these newer, alternative methods of analyzing health data.

Once these smaller GIS programs are established in a MC system, they are usually functional at the office, group, even small department or sub-department level.  Moving this analysis process from lower level to upper level defines whether or not it is a success at the inventor, special team, or departmental level.    The ability of an institution to use a Medical GIS as a symbol of accomplishment should be the long term goal of any team or individual engaged in this process.  The limiter to this success is demonstrating some value for the end products developed using Medical GIS techniques.

The majority of MC programs so have internal small group programs using GIS in some way to evaluate a little of their data.  These programs use the Medical GIS for specific purposes.  One common example of this is the use of Medical GIS in a manner that is already employed by environmental health and disease intervention programs functioning within a public health program.

Common diseases or health conditions evaluated spatially include HIV, diabetes, asthma, smoking, and drug addiction.  The impacts of poverty, race/ethnicity, teen pregnancy, and SES on small regional health statistics are also common applications of GIS to a basic MC like setting.

The barrier to producing a larger Medical GIS in MC is the lack of interest in this new technology at the upper level and the unwillingness of programs to diversify their quality improvement/population health surveillance processes.  These special interest topics like HIV, street drugs, poverty, teen pregnancy, serve as an arguable reason to bring Medical GIS from the department level on down, into the corporate level of development, implementation, use and integration.  Combining the notion of mapping disease genomically and demographically, culturally could greatly impact the need for GIS in a managed care system.

The very common large software packages and Big Data HIT platforms in part allow some of these uses to be implemented, although not at an advantageous small area level. Disease mapping requires more than just point-arc GIS or a raster system that is GIS.  The GIS for HIT-GIS must be polymorphic in its capabilities, activities,  and performance.  These values are what turn partially useful spatial analysis tools, programs or systems serves into a reason for supporting (though not promoting) further development of GIS in Managed Care.  The costs attached to these products may or may not help the institution or corporation develop a medical GIS that is both highly productive and impressive, but they will help us improve our understanding of mankind and mankind’s health, past, present and future, as a product of natural and human ecology, and a result of our interactions with the macro- and microcosmos that make us who we are.

Space, disease and time . . . with only time at a standstill!  

Could 2016  become another year without change?