Vital Times 2017 - 59

systematic bias of the measurements as a group. Precision
on the other hand, informs us of the frequency and
degree of error in a measurement system. If a series of
measurements were to be repeated under unchanged
conditions, precision tells us whether the measurements
would show the same results, i.e. the reproducibility and
repeatability of a measurement system.
For the data in our EHR to be useful, it must be both
accurate and precise. For example, in clinical decision
support (CDS), when a provider is presented with a
best practice advisory notification based on clinical
data-if the data that informs that CDS and its resulting
recommendation are wrong, the provider is likely to lose
all faith in that CDS system.5

Most of us reading this article remember our days as
a medical student or intern when we hand wrote our
notes, and the clinical note was a succinct document used
primarily for communication with current and future
healthcare team members. Now with the push to capture
as much billing data as possible, many EHR notes have
become an endless stream of data copied from one place
to the next (witness the auto-populating of labs, problem
lists, and medical history lists into notes). Meaningful
information is now scattered throughout a sprawling
patient chart. Thus, modern EHRs end up with lots of
helpful data mixed among repetitive or irrelevant data,
often making finding all of the relevant information for a
single disease quite difficult and/or time consuming.

Consider a hypothetical example of a real-time CDS that
notifies a provider of the occurrence of blood pressure
below a set threshold. What if the non-invasive blood
pressure cuff reading is systematically 5mmHg lower
than the invasive blood pressure reading. This represents
a lack of accuracy that must be accounted for in the
CDS algorithm. Say the invasive arterial blood pressure
waveform repeatedly dampens and requires flushing,
resulting in wide ranges in the pulse pressure. This is an
example of lack of precision that needs to be accounted
for as well. Creating a system that fails to account for
either of these scenarios not only makes the CDS system
ineffective, it actually inhibits safe patient care.

And what about the quality of the data? If you were to
find "everything", would it be accurate and precise? The
answer here is somewhat murkier, but the answer is "sort
of." We do know from our research that individual people
will often miss something when trying to find data from
across the medical record.8 For example, how many
anesthesiologists look through every hemoglobin A1C
to determine if their patient has diabetes. This means,
that things like problem lists, or list of medical history
often contain outdated diagnosis and fail to contain some
current ones. Efforts by hospitals to "encourage" doctors
to maintain these lists have often failed and generally lead
to significant impacts on physician satisfaction. Thus, on
the level of an individual patient the data lacks precision.

With this background in mind, let us look to the current
state of the data in our EHRs and what is necessary for
us to reach the "promised land". Most EHRs had their
start as billing systems and thus were designed to capture
data primarily for billing purposes. In this context, it is
of critical importance that measurements for billing err
on the side of being more specific than sensitive. Ideally
such a system only "bills" for a disease or entity if it most
certainly is present or occurred, and is also meticulously
engineered to avoid "upcoding" or over-billing.

Despite these shortcomings, in aggregate, the data from
one hospital probably can be accurately compared
to the data from another hospital. This comes back
to the meaning of accuracy - the systemic bias of
the measurements as a group. While the data on
each individual patient may be imprecise (i.e. not
reproducible), however in aggregate those differences
even themselves out. This is why current pay-forperformance schemes usually focus on the outcomes of
a hospital as a whole, rather than on the outcomes of an
(continued)

Annual	Publication	2017	|	 	59



Table of Contents for the Digital Edition of Vital Times 2017

Vital Times 2017 - Cover1
Vital Times 2017 - Cover2
Vital Times 2017 - 1
Vital Times 2017 - 2
Vital Times 2017 - 3
Vital Times 2017 - 4
Vital Times 2017 - 5
Vital Times 2017 - 6
Vital Times 2017 - 7
Vital Times 2017 - 8
Vital Times 2017 - 9
Vital Times 2017 - 10
Vital Times 2017 - 11
Vital Times 2017 - 12
Vital Times 2017 - 13
Vital Times 2017 - 14
Vital Times 2017 - 15
Vital Times 2017 - 16
Vital Times 2017 - 17
Vital Times 2017 - 18
Vital Times 2017 - 19
Vital Times 2017 - 20
Vital Times 2017 - 21
Vital Times 2017 - 22
Vital Times 2017 - 23
Vital Times 2017 - 24
Vital Times 2017 - 25
Vital Times 2017 - 26
Vital Times 2017 - 27
Vital Times 2017 - 28
Vital Times 2017 - 29
Vital Times 2017 - 30
Vital Times 2017 - 31
Vital Times 2017 - 32
Vital Times 2017 - 33
Vital Times 2017 - 34
Vital Times 2017 - 35
Vital Times 2017 - 36
Vital Times 2017 - 37
Vital Times 2017 - 38
Vital Times 2017 - 39
Vital Times 2017 - 40
Vital Times 2017 - 41
Vital Times 2017 - 42
Vital Times 2017 - 43
Vital Times 2017 - 44
Vital Times 2017 - 45
Vital Times 2017 - 46
Vital Times 2017 - 47
Vital Times 2017 - 48
Vital Times 2017 - 49
Vital Times 2017 - 50
Vital Times 2017 - 51
Vital Times 2017 - 52
Vital Times 2017 - 53
Vital Times 2017 - 54
Vital Times 2017 - 55
Vital Times 2017 - 56
Vital Times 2017 - 57
Vital Times 2017 - 58
Vital Times 2017 - 59
Vital Times 2017 - 60
Vital Times 2017 - 61
Vital Times 2017 - 62
Vital Times 2017 - 63
Vital Times 2017 - 64
Vital Times 2017 - 65
Vital Times 2017 - 66
Vital Times 2017 - 67
Vital Times 2017 - 68
Vital Times 2017 - 69
Vital Times 2017 - 70
Vital Times 2017 - 71
Vital Times 2017 - 72
Vital Times 2017 - 73
Vital Times 2017 - 74
Vital Times 2017 - 75
Vital Times 2017 - 76
Vital Times 2017 - 77
Vital Times 2017 - 78
Vital Times 2017 - 79
Vital Times 2017 - 80
Vital Times 2017 - 81
Vital Times 2017 - 82
Vital Times 2017 - 83
Vital Times 2017 - 84
Vital Times 2017 - 85
Vital Times 2017 - 86
Vital Times 2017 - 87
Vital Times 2017 - 88
Vital Times 2017 - Cover3
Vital Times 2017 - Cover4
https://www.nxtbook.com/allen/csvt/vital-times-2021
https://www.nxtbook.com/allen/csvt/2020
https://www.nxtbook.com/allen/csvt/2019
https://www.nxtbook.com/allen/csvt/2018
https://www.nxtbook.com/allen/csvt/2017
https://www.nxtbook.com/allen/csvt/2016
https://www.nxtbook.com/allen/csvt/2015
https://www.nxtbook.com/allen/csvt/2014
https://www.nxtbookmedia.com