Accepted Manuscript Title: Editor's Page: Got Big Data? Author: Paul J. Hauptman PII: DOI: Reference:
S1071-9164(16)00030-0 http://dx.doi.org/doi: 10.1016/j.cardfail.2016.01.010 YJCAF 3709
To appear in:
Journal of Cardiac Failure
Please cite this article as: Paul J. Hauptman, Editor's Page: Got Big Data?, Journal of Cardiac Failure (2016), http://dx.doi.org/doi: 10.1016/j.cardfail.2016.01.010. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Editor’s Page: Got Big Data?
Paul J. Hauptman MD Saint Louis University School of Medicine
Address for Correspondence:
Paul J. Hauptman MD Saint Louis University Hospital 3635 Vista Avenue Saint Louis MO 63110 USA P: 314 268 5293 F: 314 577 8861 E:
[email protected]
Financial support: None Conflict of interest: None Word count: 571
Page 1 of 4
“If you torture the data long enough, it will confess." - Ronald Coase, Economics, Nobel Prize Laureate https://en.wikiquote.org/wiki/Ronald_Coase “Everybody gets so much information all day long that they lose their common sense”. - Gertrude Stein https://www.goodreads.com/author/quotes/9325.Gertrude_Stein
A number of years ago, the advertising campaign “Got Milk?” caught fire and suddenly milk was “in.” Although Big Data has no celebrity supporters (that we know of), everyone seems to be asking the same question: Got Big Data? On paper it certainly appears to be a major step forward, helping to put the “Outcomes” in Outcomes Research. Previously, “linkage” was the key, providing a way to connect patient data across multiple administrative databases (e.g. Medicare and UNOS). Big Data is linkage on steroids. Indeed, the ability to “follow” patients longitudinally across multiple sites and even systems of care (including clinic visits, hospitalizations, pharmacy records, long-term care facilities and laboratories) may sound too good to be true, and it likely is.
Theoretically, heart failure is an excellent target for Big Data: it’s a chronic disease with significant resource utilization across different sites over extended periods of time. The prevalence of comorbidities is high, which provides ample opportunity to evaluate both processes and outcomes of care. Further, while a lot can be learned from community-based studies (e.g., Olmstead County), Big Data can more readily expand beyond a particular geographical region.
It is reasonable to point out, however, that a unifying definition of Big Data does not exist currently. From 30 000 feet, it is probably best described as a mechanism that provides “…things one can do at a large scale that cannot be done at a smaller one, to extract new insights or create new forms of value.” (1).
Page 2 of 4
For our purposes, there are a number of significant potential benefits that can accrue from Big Data, both in a generic sense and for a particular health system evaluating quality and performance. The oftused phrase “real world medicine” can be evaluated. Even with the added power that Big Data delivers, however, we have not reached a tipping point in outcomes research. To begin with, not all data are good data. Just ask yourself the next time you enter ICD-10 codes in your electronic medical record. Free text is difficult to incorporate. The data are not actionable in real time. And there are other measurements that cannot be easily captured, if at all (Table). Big Data will still be plagued by residual confounding. It is also reasonable to ask: to what end are we engaging in this research? Clearly, we might gain more insight into types of interventions that might work to improve access and outcomes. Implementation science may benefit. Comparative effectiveness research can be performed using novel methods. Moving forward, at the JCF we will be interested in publications that use Big Data in constructive ways. Almost certainly, we can learn a lot about the heart failure care we deliver and the outcomes of the patients we treat. We will be surprised by what we find.
However, at the same time, we should remember that Big Data will not provide answers to many of our pressing questions. Indeed, there is a risk that the questions will be forced to comply with Big Data, but the nature of the question being asked should drive the method, not the other way around. Big Data may inform us about resource use but only from afar. On the individual patient basis, it will still be up to us, to our own individual knowledge about heart failure, our ability to communicate and our individualized approach to the patient. In that way, Gertrude Stein’s admonishment is not far off.
References 1. Mayer-Schonberger V and Cukier K quoted in: http://www.forbes.com/sites/gilpress/2014/09/03/12big-data-definitions-whats-yours/#2715e4857a0b2acdab7d21a9
Page 3 of 4
What Big Data Does Not Do (Easily or At All) · Incorporate Quality of Life · Incorporate the Patient experience · Control for Implicit Bias · Evaluate Patient self-efficacy or knowledge · Evaluate physician decision-making · Measure severity of co-morbidities like depression · Eliminate Residual Confounding
The author thanks Dr. John T. Chibnall for his critical review.
Page 4 of 4