I wrote in this blog last year that the focus of work of clinical informatics itself will likely change from electronic health record implementation to analytics, i.e., optimizing the systems we now have so all of the data accumulating in them can be used to improve health and healthcare delivery. I have also taken, like many other informatics researchers, a great interest in data-related issues [1, 2].
By the same token, I believe we need to be careful about the hype around analytics, something I alluded to in a recent posting reminding data scientists that they also needed to be research methodology scientists. Since that time, a colleague provided some additional examples of this, noting a recent talk she attended by a CEO of a data analytics company who proudly declared that he did not need to know anything about the underlying domain of medicine; he was only a data scientist. He proceeded to describe some correlations that his product had recently uncovered, such as a diagnosis of chronic kidney disease being associated with an ateriovenous (AV) fistula and septicemia being associated with hospitalization. (To which the medically knowledgeable people replied, "Duh!") In other words, finding correlations among things we already know clinically is hardly an advance for analytics. This is also borne out in another recent paper on "Why big won't cure us," with the author pointing out a myriad of technical, financial, and ethical issues that must be addressed before we will be able to make use of big data routinely in clinical practice .
Another issue is that while the use of data analytics is important and will grow, we cannot say with certainty what proportion of informatics professionals will actually be working in this area. There is no question that most informatics professionals will be working with data and information in some way, as these are the "lifeblood" of healthcare . We also know that for every person who explicitly "does" analytics, there are some number of people, probably a larger proportion, who are either supporting analytics or doing other health IT work, whether with EHRs or other technologies. This is borne out by a recent report from the McKinsey consulting firm, which forecasts a demand for 140,000 to 190,000 positions for all data scientists in all fields (not limited to healthcare), and also another need for an addition 1.5 million managers and analysts who "can ask the right questions and consume the results of the analysis of big data effectively" .
What our informatics educational program needs to teach was borne out in a recent conversation with Brian Sikora, who is Director of Data & Information Management Enhancement in the Kaiser-Permanente Northwest Region. His service currently employs 110 analysts and plans to add more. Also good news for our students was our discussion about possible internship opportunities.
Mr. Sikora described four types of analysts in his department to our career development specialist, Ms. Virginia Lankes:
- Data analysts - the most technical group who do the architecture
- Report analysts – design reports, dashboards, key performance indicators; wider background, business intelligence domain.
- Business systems analysts – understand system, workflow and the people they are working with to process and know how data is produced in a specific area. “Here’s the problem - help us solve it.” Deep expertise in operational areas.
- Informatics analysts – (the largest group) which includes statisticians; they help operations leaders interpret and analyze data. Data mining, text mining.
- Programming skills - analytics professionals must have programming skills, especially in data-related areas for locating and extracting data, using tools such as SQL and SAS.
- Understanding the healthcare environment
- Communication skills - ability to work with clinical, administrative, and financial staff to understand their programs and present solutions in written and oral form
- Critical thinking - including the ability to understand a business problem, identify the appropriate data elements, extract and aggregate the data, and use it to solve the practical problem
The jury is clearly out on who and how many will be doing what in analytics in the future work of informatics and larger healthcare. Just as I called for research delineating the informatics workforce years ago , we need a similar analysis with regards to analytics. We need to answer questions such as, how many people do we need to do the work of analytics, how many do we need to support it, what fraction of the informatics field will be working explicitly in analytics, and what training is needed for those who work directly in it and those who need to know it as part of a well-rounded informatics education.
1. Hersh, WR, Weiner, MG, et al. (2013). Caveats for the use of operational electronic health record data in comparative effectiveness research. Medical Care. 51(Suppl 3): S30-S37.
2. Hersh, WR, Cimino, JJ, et al. (2013). Recommendations for the use of operational electronic health record data in comparative effectiveness research. eGEMs (Generating Evidence & Methods to improve patient outcomes). 1: 14. http://repository.academyhealth.org/egems/vol1/iss1/14/.
3. Neff, G (2013). Why big data won't cure us. Big Data. 1: 117-123.
4. Manyika, J, Chui, M, et al. (2011). Big data: The next frontier for innovation, competition, and productivity, McKinsey Global Institute. http://www.mckinsey.com/insights/business_technology/big_data_the_next_frontier_for_innovation.
5. Hersh, WR (2006). Who are the informaticians? What we know and should know. Journal of the American Medical Informatics Association. 13: 166-170.