Keywords
electronic health records - documentation burden - oncology - physician - interfaces
- usability
Background and Significance
Background and Significance
The widespread use of electronic health records (EHRs) has increased substantially
and dramatically changed modern medical care. The use of EHRs could lead to many advantages
such as improved access to data, improved data quality, and faster documentation.[1]
[2] However, most health care providers are not yet experiencing these benefits of EHR
use.[3] Whereas the most important task of health care professionals is to deliver patient
care, the transition from paper-based to computerized documentation has led to increased
documentation time.[4] This might be caused by the increased need to fulfill regulatory, reimbursement,
and quality-measurement requirements.[5] Consequently, increased EHR time can result in less time for direct patient care,
decreased physician job satisfaction, and increased burnout rates among physicians.[6]
[7]
[8] Moreover, the time spent on desktop medicine is increasing at the expense of face-to-face
visits, with time currently evenly split between both categories.[9] On the other hand, benefits of EHR use, such as improved access to and quality of
information, have also been reported.[10]
Some papers have quantitatively described how much time and effort physicians spend
on the EHR during consultations in the outpatient clinic. A time–motion study (TMS)
investigating documentation time in 14 different specialties reported a mean percentage
of documentation time per consultation of 33%, while another TMS describing time allocation
in four specialties reported similar results with 37% spent on the EHR.[7]
[11] A study that used EHR activity logs found that the EHR is used for an average of
16 minutes and 14 seconds per encounter, with chart review (33%), input (24%), and
ordering (17%) accounting for most of the time.[12] One study revealed that daily EHR time can vary significantly between surgical (45.6 minutes),
medical (85.7 minutes), and primary care specialties (115.0 minutes).[13] These studies consistently show a high percentage of time spent on the EHR. However,
detailed data on EHR activity measures such as mouse clicks, keystrokes, and mouse
movement are limited. These data might give insights into where the usability of EHRs
can be improved. Additionally, physicians make significantly less eye contact with
patients when using an EHR than a paper chart.[14] Patients are also less likely to actively participate in consultations when a physician
is physically engaged with the computer (e.g., keyboard activity) than when a physician
is merely gazing at the EHR.[15] This implies that less effort required for documentation during consultations could
be beneficial to doctor–patient interaction. Besides, a survey study investigating
the relationship between EHR design and use factors with high stress and burnout identified
interference with the patient–clinician relationship and excessive data entry as significantly
associated factors with high stress and burnout.[16] The findings of these studies suggest that not only the amount of time spent on
the EHR is relevant for the experienced documentation burden, but also the actual
effort put in by the health care professional is an important factor, which is also
stated in a recent scoping review by Moy et al.[17] The authors discussed the clinical documentation burden among health care providers
and identified time and effort as the two main concepts that underlie the documentation
burden in EHRs. The study concluded that the documentation burden remains understudied
and undermeasured in both inpatient and outpatient settings, indicating that further
research is warranted. As stated, time spent on the EHR can vary depending on specialty
or setting. Little is known about the documentation burden in the more specific, oncological
setting.
Objectives
This study investigated the current state of the documentation burden within the EHR
during consultations in a tertiary oncology center. Furthermore, we assessed perceptions
of head and neck cancer (HNC) care providers on various aspects regarding EHR documentation
and EHR satisfaction.
Methods
A cross-sectional TMS was conducted at the Department of Head and Neck Oncology at
the Antoni van Leeuwenhoek Cancer Centre in Amsterdam, the Netherlands. In the outpatient
clinic, patients were routinely seen and examined by a HNC care provider. These consultations
were recorded and analyzed with video-analytic software Morae version 3.1 (TechSmith,
Michigan, United States). Furthermore, providers were invited to complete an online
questionnaire regarding various concepts underlying the documentation process and
system satisfaction. Data were collected between April and July 2020. The procedures
of this study were approved by the Antoni van Leeuwenhoek Cancer Center local ethics
committee (IRBd19–312).
We included patients scheduled for an initial oncological consultation (IOC; N = 47) or a follow-up consultation (FUC; N = 50). Participating providers were head and neck surgeons, fellows, residents, and
physician assistants. Providers with less than 3 months of experience with the EHR
(Chipsoft HiX, custom build, version 6.1), which was implemented in 2012, were excluded.
After obtaining informed consent, Morae Recorder was used to capture the routine workflow
during outpatient consultations. A consultation was defined as the time that a patient
was present in the consultation room. Furthermore, the wrap-up time, defined as the
time providers need to complete tasks after a patient has left the room, was recorded.
The software simultaneously captured the screen of a provider, generated usability
metrics, e.g., mouse clicks and keystrokes, and used a webcam to record audio as well
as video of the mouse and keyboard. Recordings started at the beginning of a consultation
and stopped when the provider finished the consultation, including the wrap up. At
the end of a consultation, recordings were password-protected and stored in a secured
folder, ensuring a double layer of protection. Subsequently, recordings were imported
into the video-analytic software program Morae Manager. Following this, detailed video
analysis was performed while using time–motion methodology. During playback of the
recordings, time spent on various tasks during consultations was measured by a single,
independent researcher using the app Time Motion Study version 2.3 (Graphite Inc.),
which is similar to the TMS capture tool TimeCAT, but available on mobile devices.[18] The categories and subtasks used ([Table 1]) were based on a similar study conducted by Joukes et al.[7] When a provider was multitasking, both subtasks were measured simultaneously.
Table 1
Categories and subtasks used in the measurement app
|
Category
|
Subtask
|
Explanation
|
|
1. EHR
|
Chart review
|
When the physician is looking for or reading information from the patient record.
|
|
1. EHR
|
Input
|
When the physician is entering information into the patient record.
|
|
1. EHR
|
Ordering
|
The physician orders tests, e.g., imaging, laboratory, or medication.
|
|
1. EHR
|
Other
|
Used when the observer cannot discern whether the task falls in one of the four other
(more specialized) EHR tasks.
|
|
2. Communication
|
Physician–patient communication
|
All communication between physician and a patient.
|
|
2. Communication
|
Discussion with colleague
|
All communication between the physician and a colleague.
|
|
3. Other
|
Other computer tasks
|
All tasks on the computer that are not in the EHR program (e.g., reading mail).
|
|
3. Other
|
Other activities
|
All tasks that do not fit in one of the other categories.
|
Abbreviation: EHR, electronic health record.
Furthermore, the number of mouse clicks, scrolls, keystrokes and EHR mouse path length
in meters, consultation duration, and supervision time were extracted from the recordings.
Subsequently, data from the recording software, the time–motion capture tool, and
data extracted from the EHR regarding order entry were combined in a database.
A validated questionnaire was used to assess perceptions of HNC care providers on
concepts regarding EHR documentation and EHR satisfaction.[19] All questions were answered on a 5-point Likert scale, ranging from strongly disagree
(1) to strongly agree (5). The questionnaire can be found in [Supplementary Appendix A] (available in the online version). All HNC care providers working at the department
were invited by mail to complete this questionnaire in the online environment of the
electronic data capture tool CasterEDC. Twenty-two (84%) providers completed the questionnaire,
of which 14 (64%) were supervising staff, 5 (23%) were residents, and 3 (14%) were
physician assistants.
Continuous variables are presented as median and quartiles, mean and standard deviation
(SD), and categorical variables as numbers and percentages. Descriptive statistics
were performed using IBM SPSS Statistics software version 25.0 (IBM Corp).
Results
After excluding three incomplete recordings, a total of 97 valid outpatient consultations
were used for analysis, of which 47 were IOCs, and 50 were FUCs. Provider and patient
demographics are shown in [Table 2].
Table 2
Physician and patient demographics and details of the observed consultations
|
Physician characteristics
|
Initial oncological consultation
|
Follow-up consultation
|
All
|
|
Total HNC care providers
|
8 (66.6%)
|
4 (33.3%)
|
12 (100%)
|
|
Physician assistant
|
2 (16.6%)
|
0 (0.0%)
|
2 (16.6%)
|
|
Resident
|
4 (33.0%)
|
0 (0.0%)
|
4 (33.3%)
|
|
Fellow
|
2 (16.6%)
|
0 (33.0%)
|
2 (16.6%)
|
|
Head and neck surgeon
|
0 (0.0%)
|
4 (33.0%)
|
4 (33.3%)
|
|
Patient characteristics
|
|
Age (mean)
|
67.6
|
64.6
|
66.1
|
|
Sex (n)
|
|
|
|
|
Male
|
30
|
26
|
56
|
|
Female
|
17
|
24
|
41
|
|
Observations
|
|
Number of consultations
|
47
|
50
|
97
|
|
Total recording time
|
44h:19m
|
13h:01m
|
57h:20m
|
|
Total duration of consultations
|
41h:18m
|
09h:26m
|
50h:44m
|
Abbreviation: HNC, head and neck cancer.
The median duration of an IOC with a patient present was 52:38 (43:43–62:05) and 54:27
(47:04–63:45) including wrap-up time. The median duration of a FUC with a patient
present was 09:54 (06:12–15:14) and 11:55 (07:40–17:21) including wrap-up time. During
an IOC, a resident or physician assistant usually consults with a supervisor outside
of the room. In most cases, this provider has to wait for the supervisor. The median
duration for this supervision time during an IOC was 07:29 (05:15–13:50). The clean
consultation duration, in which the supervision time outside of the room is subtracted
from the total consultation duration, was also calculated. This was 42:51 (36:55–48:51)
with the patient present and 43:59 (38:20–52:15) including wrap-up time. [Table 3] shows how much time was spent on each of the main categories. The median percentage
of time spent on a specific task relative to the total consultation time is also shown.
Because some tasks are regularly conducted simultaneously, such as communicating with
the patient and EHR tasks, the total percentage exceeds 100%. Furthermore, not all
subtasks were used in every consultation.
Table 3
Time spent on tasks during consultations
|
Initial oncological consultation
|
N
|
Median
|
(Q1–Q3)
|
Mean
|
(SD)
|
Median % of consultation spent on task
|
|
Consultation duration (including wrap-up, excluding supervision time)
|
47
|
43:59
|
(38:20–52:15)
|
45:56
|
(12:25)
|
100%
|
|
EHR tasks—total
|
47
|
19:16
|
(14:42–24:02)
|
19:20
|
(07:15)
|
44.0%
|
|
EHR tasks—chart review
|
47
|
01:36
|
(00:37–02:32)
|
01:57
|
(01:46)
|
3.1%
|
|
EHR tasks—input information
|
47
|
11:10
|
(07:40–14:28)
|
11:06
|
(04:23)
|
24.7%
|
|
EHR tasks—placing orders
|
44
|
05:59
|
(04:08–09:10)
|
06:37
|
(03:51)
|
12.2%
|
|
EHR tasks—other
|
16
|
00:05
|
(00:04–00:24)
|
00:14
|
(00:15)
|
0.2%
|
|
Other computer tasks
|
20
|
00:46
|
(00:18–01:54)
|
01:08
|
(00:46)
|
1.8%
|
|
Physician–patient communication
|
47
|
31:47
|
(28:02–40:09)
|
34:48
|
(11:14)
|
80.0%
|
|
Peer communication
|
43
|
01:37
|
(00:48–02:25)
|
02:32
|
(03:37)
|
3.1%
|
|
Other tasks
|
15
|
00:12
|
(00:05–00:19)
|
00:13
|
(00:10)
|
0.4%
|
|
Follow-up consultation
|
N
|
Median
|
(Q1–Q3)
|
Mean
|
(SD)
|
Median % of consultation spent on task
|
|
Consultation duration (including wrap-up)
|
50
|
11:55
|
(07:40–17:21)
|
13:18
|
(06:34)
|
100%
|
|
EHR tasks—total
|
50
|
03:45
|
(02:28–05:32)
|
03:56
|
(01:57)
|
30.7%
|
|
EHR tasks—chart review
|
49
|
01:12
|
(00:33–01:48)
|
01:23
|
(01:00)
|
9.8%
|
|
EHR tasks—input information
|
47
|
01:49
|
(01:13–02:19)
|
01:57
|
(00:57)
|
14.9%
|
|
EHR tasks—placing orders
|
47
|
00:24
|
(00:12–01:18)
|
00:42
|
(00:39)
|
3.7%
|
|
EHR tasks—other
|
16
|
00:11
|
(00:08–00:16)
|
00:12
|
(00:06)
|
2.0%
|
|
Other computer tasks
|
12
|
00:36
|
(00:14–01:31)
|
01:04
|
(01:06)
|
4.9%
|
|
Physician–patient communication
|
50
|
07:29
|
(04:23–13:01)
|
08:56
|
(05:37)
|
67.9%
|
|
Peer communication
|
29
|
00:58
|
(00:35–02:00)
|
01:34
|
(01:43)
|
8.4%
|
|
Other tasks
|
9
|
00:17
|
(00:11–00:28)
|
00:22
|
(00:17)
|
1.8%
|
Abbreviations: EHR, electronic health record; SD, standard deviation.
The time spent on EHR tasks had a median duration of 19:16 (14:42–24:02) for IOC and
03:45 (02:28–05:32) for FUC. Furthermore, during IOC, 44.0% of the total consultation
time was spent on EHR tasks, and during FUC, 30.7%. The input of information into
the EHR was the most time-consuming EHR task, with 24.7% (IOC) and 14.9% (FUC) of
total consultation time. When comparing time spent on EHR tasks by residents, physician
assistants, and fellows, no significant differences were found. [Table 4] summarizes the usability metrics measured within the EHR during consultations.
Table 4
Usability metrics required per consultation
|
Metric
|
Initial oncological consultation including wrap-up
|
Follow-up consultation including wrap-up
|
|
Mean
|
SD
|
Mean
|
SD
|
|
Total mouse events, mean (SD)
|
593
|
(300.0)
|
140
|
(89.3)
|
|
Mouse clicks, mean (SD)
|
215
|
(91.6)
|
55
|
(28.4)
|
|
Scrolling, mean (SD)
|
378
|
(233.9)
|
86
|
(67.0)
|
|
Keystrokes, mean (SD)
|
1,664
|
(896.3)
|
450
|
(290)
|
|
Mouse travel distance in meters, mean (SD)
|
56
|
(25.9)
|
14
|
(8.2)
|
|
Other
|
|
|
|
|
|
Orders per consultation, mean (SD)
|
6.9
|
(3.4)
|
1.6
|
(1.1)
|
|
Time per order, mean (SD)
|
00m:53s
|
(00m:20s)
|
00m:20s
|
(00m:17s)
|
Abbreviation: SD, standard deviation.
This table shows that providers required 1,664 (SD = 896) keystrokes and 593 (SD = 300)
mouse events per IOC, and providers required 450 (SD = 290) keystrokes and 140 (SD = 89)
mouse events per FUC. [Table 4] also displays the number of orders placed per consultation, the mean time per order,
and the time to complete all orders after consultation.
Perceptions of HNC care providers on different aspects regarding EHR documentation
and EHR satisfaction were measured using the validated questionnaire. Relevant results
are displayed in [Fig. 1]. Most respondents (78%) felt that they properly mastered working with the EHR, while
4% disagreed with this statement and 18%, all attendings, were neutral. Over half
of respondents (55%) said that the EHR supports their personal work processes, 44%
indicated that they can always find the information they need in the EHR, and 50%
agreed that the EHR facilitates agreement with colleagues on the treatment plan of
the patient. However, only a minority indicated that they thought the EHR was user-friendly
(32%) and had a clear interface (27%). Furthermore, less than a quarter of respondents
(23%) agreed that there is enough time to properly document patient data in the EHR,
and that they can easily and timely send all required information when referring a
patient (23%). Despite this, over two-thirds of respondents said that the EHR helps
them provide good quality patient care (73%), a vast majority indicated that they
can trust that the EHR always works (86%), and only 9% disagreed with the statement
that their organization has a high-quality EHR. The full questionnaire results can
be found in [Supplementary Appendix B] (available in the online version).
Fig. 1 Perceptions of HNC care providers on EHR documentation and EHR satisfaction. EHR,
electronic health record; HNC, head and neck cancer.
Discussion
This study aimed to quantify the time and effort currently spent on the EHR by providers
in an outpatient clinic of a Head and Neck Oncology care center. Our analysis shows
that a significant proportion of time is spent on EHR tasks during consultations.
We found that 44.0% of the time during an IOC and 30.7% of the time during a FUC is
spent on the EHR. In contrast, during 80.0 and 67.9% of the IOC and FUC, respectively,
there was active communication between the patient and the provider. On average, providers
require 593 mouse events, 1,664 keystrokes, and 56 m of mouse travel distance during
an IOC and 140 mouse events, 597 keystrokes, and 14 m of mouse travel distance during
a FUC. Additionally, despite that over one-third to just under half of the available
time during consultations is spent on the EHR, a majority of providers still feel
there is not enough time for proper documentation.
Comparison with Previous Literature and Interpretation
Our results on time spent on the EHR in Head and Neck Oncology during consultations
are consistent with findings of earlier studies. A study conducted at an ophthalmology
department found similar results regarding documentation time during consultations,
reporting 27% of time during consultations spent on EHR use.[20] A study conducted at four different departments reported 37% of consultation time
spent on the EHR.[11] Another study investigating physician time allocation in various specialties during
a whole day found percentages for documentation tasks ranging from 11 to 39%, stating
that the distribution of time spent by providers using EHRs varies between specialties.[12] Furthermore, de Hoop and Neumuth reported that 37.1% of time during consultations
was spent on the EHR.[21] In this study, physicians reported that the spread of patient information, poor
integration of information into workflow, and limited information exchange were problematic.
Only a few studies investigated usability measures such as keystrokes and mouse clicks.
One study describing how physician EHR activity influences patient participation reported
similar results, with a mean of 216 (SD = 174) mouse events and 729 (SD = 768) keystrokes
required in consultations lasting 20.3 (SD = 10.5) minutes on average.[14]
Our results suggest that while already spending a large proportion of their time on
the EHR, providers are also actively engaged with the EHR. Based on our results, a
provider requires almost 40 keystrokes and 13 mouse clicks or scrolls for every minute
of consultation time. In contrast, we found that during a large proportion of the
consultations, there is active communication between providers and patients, which
is beneficial to the provider–patient relationship. However, based on our results,
we cannot determine whether the provider was actually talking or listening. It could
also mean that the patient is talking and the provider is multitasking and conducting
an EHR task while listening. While this is common practice, a high level of multitasking
adds to the experienced documentation burden.[17]
[22]
Health care providers mainly had concerns regarding the available time for recording
data, timely sending referral information, and finding relevant information within
the EHR. All of these factors can contribute to spending additional time on the EHR
and therefore cannot be spent on direct patient care. Additional concerns were expressed
regarding the extent to which the EHR supports structured data capture. Lack of structured
data capture can impede data reuse.[23] Surprisingly, only one respondent disagreed with the statement that they properly
mastered working with the EHR. This indicates that the vast majority considered themselves
skilled with the EHR. This could be either the result of proper training, but overestimating
their own efficiency with the EHR could also contribute to this result. Furthermore,
our survey results suggest that whereas most providers are optimistic regarding the
usefulness of the EHR, most also think that the usability (e.g., ease of use) of the
EHR should be further improved. This suggests the EHR as a solution, rather than consider
it the primary reason for the documentation burden.
Comparing our results to other studies must be done with caution because of various
factors, such as differences in consultation types and complexity, different EHR vendors,
EHR maturity, and study methods. Nevertheless, this study further corroborates that
the high documentation burden is widespread.
Strengths and Limitations
The main strength of this study is that this study evaluated the time spent on the
EHR combined with EHR usability measures. It also quantifies the time and effort required
to document and review information in the EHR while also describing provider perceptions
regarding EHR satisfaction and the documentation process. This allows for comparison
between quantitative data and the opinion of health care providers on this topic.
Another strength is the chosen methodology for our study. While time-consuming, TMSs
are still generally considered the gold standard methodology for accurately measuring
a process.
A limitation of this study is that, as expected, we found variation in consultation
duration and usability metrics between consultations in both IOC and FUC. This can
probably be attributed to differences in various factors, such as patient complexity
and provider variation. Another limitation is that, due to the chosen methodology,
we did not investigate time spent on the EHR outside of consultation hours, which
is also a construct underlying the documentation burden. However, only a minority
of providers indicated that they felt that the amount of time they spent on the EHR
outside of consultation hours is high (14.3%), whereas most providers rated this as
acceptable (61.9%). Nevertheless, this does not rule out that health care professionals
still spent a considerable amount of time on the EHR outside of consultation hours.
Lastly, as stated in the Introduction, a high level of interaction with the computer
can negatively influence the doctor–patient relationship. In this study, the measure
patient satisfaction was not measured. However, it can be expected that patient satisfaction
can increase when EHR time decreases, as more time can be spent on the patient, which
was also established by Marmor et al.[24]
Implication for Practice
While our results indicate that the burden of documentation during consultations is
already high, accurate and complete documentation is becoming increasingly important
as information recorded by providers is increasingly reused for other purposes, such
as research, quality registries, and other improvements that rely on structured data,
such as clinical decision support. However, policy makers should be critical as to
which information should be recorded by health care providers while providing care.
If information is not relevant for providing care and solely documented for secondary
purposes, it is better to minimalize the burden for providers and collect it in different
ways. For example, employing coding staff or using patient-entered before-visit questionnaires
that are automatically integrated into provider documentation could be a solution
that increases data collection and also reduces documentation burden by relieving
physicians.[25] The challenge is to reduce the documentation burden while simultaneously increasing
the accuracy and completeness of recorded data in the EHR. For this reason, a national
program, “Facilitating Clinical Documentation at the Point of Care,” has started in
the Netherlands. This program urges hospitals and EHR vendors to optimize EHRs to
support unambiguous, single registration of data during the care process. It also
stimulates that data are stored as discrete, coded data to enable reuse for various
purposes. This should lead to a decrease of the documentation burden for health care
providers and simultaneously increase the accuracy and completeness of data in EHRs.
Furthermore, streamlining workflow and aligning the documentation process with clinical
workflow might also be effective in reducing the documentation burden.[22]
[26] Lindsay and Lytle found that this can result in an 18.5% reduction in documentation
time.[26] Minimizing interruptions of workflow, for example, by being critical of which decision
support alerts should and which should not be used, can also contribute to reducing
the burden.[27] Other solutions that have been suggested are, for example, telehealth expansion,
changing compliance rules and performance metrics, and EHR optimization sprints.[28]
The optimal strategy to reduce the burden could differ based on the primary underlying
reason. This might vary based on region or setting. A recent study evaluated the difference
in EHR use between United States and non-United States clinicians and found that U.S.
clinicians daily spent over 50% more time using the EHR.[29] This might be attributed to additional documentation requirements for billing or
administrative functions. Policy makers could also consider such nontechnical aspects
when developing a strategy. Future studies should focus on implementing and evaluating
innovations and developments within EHRs that aim to decrease documentation burden
while increasing the quality of EHR data. Providing evidence is important in identifying
the best practices that should be implemented.[30] To make this type of research more scalable, it might be better suitable to use
EHR log studies instead of TMSs.[31] However, the process of turning raw audit logs into insights is still complex and
can result in largely under- or overestimating of time spent on the EHR.[32] This might be helpful to conduct more studies in which audit log data are compared
with time–motion data to further validate the reliability of audit log studies and
define validated standards.
Conclusion
This study found that HNC care providers spent up to 44.0% of consultation time on
EHR tasks. During these consultations, providers require up to 40 keystrokes and 13
mouse clicks for every minute of consultation time. These results quantify the widespread
concern of high documentation burden for health care providers, which is known to
lead to potential burnout and decrease of patient–clinician interaction. Despite the
significant amount of time spent on documentation, most providers still feel this
is insufficient for proper documentation. The challenge is to decrease documentation
burden while increasing the quality of EHR data.
Clinical Relevance Statement
Clinical Relevance Statement
While the results of this study further corroborate a high documentation burden, accurate
and complete documentation is becoming increasingly important as information recorded
by providers is increasingly reused for secondary purposes, such as measuring the
quality of care. The challenge is to reduce the documentation burden while simultaneously
increasing the accuracy and completeness of recorded data in the EHR.
Multiple Choice Questions
Multiple Choice Questions
-
How much time is spent on EHR tasks during outpatient consultations in head and neck
cancer?
-
Up to 24%
-
Up to 34%
-
Up to 44%
-
Up to 54%
Correct Answer: The correct answer is option c. In initial oncological consultations, up to 44% of
the time a provider is interacting with the electronic health record. Multitasking
is common, other tasks might also be conducted simultaneously.
-
Which task is the most time-consuming during consultations, according this study?
-
Chart review
-
Information input
-
Placement of orders
-
Other
Correct Answer: The correct answer is option b. According to our results, providers spent the most
time on entering information into the EHR. However, other studies have shown that
provider spent the most time on chart review. This might vary based on specialty or
appointment type.
