Use of electronic clinical documentation: time spent
and team interactions
George Hripcsak,1,2David K Vawdrey,1,2Matthew R Fred,2Susan B Bostwick3
Objective To measure the time spent authoring and
viewing documentation and to study patterns of usage in
Design Audit logs for an electronic health record were
used to calculate rates, and social network analysis was
applied to ascertain usage patterns. Subjects comprised
all care providers at an urban academic medical center
who authored or viewed electronic documentation.
Measurement Rate and time of authoring and viewing
clinical documentation, and associations among users
Results Users spent 20e103 min per day authoring
notes and 7e56 min per day viewing notes, with
physicians spending less than 90 min per day total.
About 16% of attendings’ notes, 8% of residents’ notes,
and 38% of nurses’ notes went unread by other users,
and, overall, 16% of notes were never read by anyone.
Viewing of notes dropped quickly with the age of the
note, but notes were read at a low but measurable rate,
even after 2 years. Most healthcare teams (77%)
included a nurse, an attending, and a resident, and those
three users’ groups were the first to write notes during
Limitations The limitations were restriction to a single
academic medical center and use of log files without
Conclusions Care providers spend a significant amount
of time viewing and authoring notes. Many notes are
never read, and rates of usage vary significantly by
author and viewer. While the rate of viewing a note
drops quickly with its age, even after 2 years inpatient
notes are still viewed.
The nation is increasingly focused on the adoption
of electronic health records. Clinical documenta-
tiondwhich defines the patientdis the core of the
electronic health record, forming the foundation on
which sit most other functions, including informa-
tion access, interprovider communication, auto-
mated decision support, and registry functions.1
Physicians in hospital settings spend much of
their daily working time on documentation.2 3In
a broad study of the perceptions of medical resi-
dents about the time they spend on documenta-
tion,4two-thirds reported spending over 4 h, which
was more than they reported spending on direct
patient care. Most studies assessing time efficiency
have noted a substantial increase in time spent
documenting when using an electronic health
record compared to paper.5
Compared to paper, electronic provider docu-
mentation allows faster and more complete access
to the patient record, and may improve communi-
cation among members of the healthcare team.6 7
Some evidence exists that electronic documenta-
tion may be associated with improved patient
outcomes and decreased costs.8In contrast to these
benefits, evaluators have also identified unintended
documentation, including changes to workflow,
increased time spent writing notes, and an adverse
effect on documentation quality.5 9 10Furthermore,
residents perceive that they get feedback on their
documentation less than 30% of the time.4
Many electronic health records have detailed
audit logs that permit the monitoring of user
behavior, including authoring and viewing clinical
notes. These logs are often used to infer suspicious
activity with respect to patient privacy,11but they
may also be used to learn about healthcare. For
example, they have been used to infer clinicians’
We believe that audit logs of the authoring and
viewing of documentation can advance the infor-
matics field in three ways. First, by better under-
standing the use of documentation facilities, it may
be possible to improve the electronic health record.
For example, its user interface can be tailored to
make access to frequently used features or data more
efficient. Second, the audit logs can reveal informa-
tion about individual clinician’s behavior. For
example, while there is a body of literature on
creating documentation,13e15few have focused in
detail on the viewing of documentation. Third, the
audit logs can also reveal information about inter-
actions among team members, potentially diag-
nosing and offering solutions to communications
gaps. In this study, we use clinical documentation
audit logs to characterize the generation and use of
clinical documentation, as well as the formation of
clinical teams and the communication within them.
The study was carried out for the inpatient area of
an academic medical center. All clinical notes at the
center are authored and viewed in the Eclipsys XA
(Eclipsys Corporation, Atlanta, Georgia) electronic
health record product. We used detailed usage logs
for the Eclipsys product (Corman Technologies,
Santa Rosa, California), which show user identity,
note type, and time spent viewing the note. Insti-
tutional review board approval was obtained for
We tallied the rates and timing of authoring and
viewing clinical notes among user groups. We
employed social network analysis16to characterize
the relationships among users. Social network anal-
ysis has been used in a wide variety of contexts,
1Department of Biomedical
Informatics, Columbia University
Medical Center, New York, USA
2Department of Information
Services, New YorkePresbyterian
Hospital, New York, USA
3Department of Pediatrics, Weill
Cornell Medical College, New
Dr George Hripcsak, Department
of Biomedical Informatics,
Columbia University Medical
Center, 622 West 168th Street,
VC5, New York, NY 10032,
Received 13 September 2010
Accepted 18 December 2010
Published Online First
2 February 2011
112 J Am Med Inform Assoc 2011;18:112e117. doi:10.1136/jamia.2010.008441
Research and applications
including scientific coauthorship, citations, email, sexual contacts,
and metabolic networks. Three properties
relevant: the degree of the network, which reflects the mean
number of viewings by a viewer of an author’s notes; assortative
mixing among user groups, which reflects pairwise associations
among user groups; and hierarchical community structure, which
defines larger groupings of users that view each other’s notes. We
used APL2 (APL2000, Rockville, Maryland) and igraph (http://
igraph.sourceforge.net/) to calculate network parameters.
Documentation authoring and viewing
Table 1 shows the number of authors and the mean number of
notes they authored per week; and the number of users who
viewed notes and the mean number of notes they viewed per
week. The table is organized by user group, where groups are
defined in the system. The row marked ‘billing/HIM’ includes
billing compliance personnel, the health information manage-
ment department, quality assurance, and administrative coor-
dinators. ‘Other clinical’ includes pastoral care and certain
psychiatry roles. The ‘notes authored per week’ column shows
that the bulk of documentation in the electronic health record
consists of nursing, resident, and attending notes. Nevertheless,
respiratory therapists, dieticians, and social workers generate
more notes per user (‘notes authored per user per week’). The
groups that view the most notes (‘notes viewed per week’) are
residents and ‘billing/HIM,’ illustrating the importance of both
clinical and administrative functions.
The amount of time users spent authoring and viewing notes
is shown in table 2 for the larger, clinically relevant user groups.
The first numeric column shows the time per day authoring
notes for those days where the user authored at least one note,
and the second numeric column shows the time per day for
those days with more significant involvement to ensure that the
user was actually on service. The latter includes runs of days in
which the user authored notes on at least two patients per day,
for at least 2 days in a row, and with at least 1 day with at least
four patients. The third and fourth columns show time per day
viewing notes for those days in which at least one note was
viewed, and those days with more significant involvement, defined
similarly to column 2, but with viewing instead of authoring.
Other than dieticians, users averaged less than 90 min per day
authoring notes and 30 min per day viewing notes, even on days
with significant involvement. Nurses can author structured or free
text notes, which are included in table 2, and they can document
in ‘flowsheets,’ which are not included in the table. The latter
include plan of care, vital signs, intake and output, treatment
parameters (isolation, wound care, etc), and data for quality
initiatives such as falls risk assessment. Therefore, total nursing
documentation time is greater than that shown in the table.
We assessed the proportion of notes that were viewed within
3 months and who viewed them. Figure 1 shows the percentage
of notes written by an author within one of five user groups that
were viewed by users in the same or other groups. To emphasize
communication rather than recall, an author viewing their own
note was not counted in this graph. The designation ‘anyone’
refers to all the users in table 2 except the ‘billing/HIM’ group.
This was done to emphasize clinical rather than administrative
communication. If billing and records were included, then
attending notes were viewed 97% of the time, and resident notes
were viewed 99% of the time; the other user groups were less
affected. Fewer than 20% of nursing notes were read by
attendings or residents, and only 38% were read by other nurses.
Attending and resident notes were read more often, but not
consistently and not necessarily by attendings, nurses, or resi-
dents. Medical students’ notes were viewed by at least one
attending or resident 72% of the time on the first day after it
was written and 81% eventually. The rate of viewing medical
students’ notes depended on service, with pediatrics and
psychiatry lagging behind other services. For all authors, 16% of
notes were never viewed by anyone, not even the author or
billing and records.
mean number of notes they viewed per week
Rate of authoring and viewing notes: number of authors and mean number of notes they authored per week, and number of viewers and
per user per week Viewers*
Notes viewed per
user per week
Nurse (non-nurse practioner)
Child life specialist
Billing/health information management
*Authors and viewers overlap.
yExcludes ‘flowsheet’ documentation, which includes plan of care, vital signs, intake and output, treatment parameters (isolation, wound care, etc), and data for quality initiatives such as falls
J Am Med Inform Assoc 2011;18:112e117. doi:10.1136/jamia.2010.008441113
Research and applications
Figure 2 shows information-seeking behavior. Figures 1 and 2
complement each other: whereas figure 1 plots the probability of
a note being viewed, figure 2 plots the probability of a viewer
seeking out a note. The bars show how often someone in a user
group viewed at least one note by someone in a second user
group, given that the note was authored within the previous
day. Attendings and residents tended to view notes by other
attendings or residents (65e93%) more often than notes by
nursing or social work (22e42%).
Notes are generally viewed soon after authoring. For example,
as shown in figure 3, most viewings of attending and resident
notes by attendings and residents occur within the first day. In
this sample, about 26000 were read in the first day, and 19000
were read in all subsequent days combined. Nevertheless, notes
do get viewed months later at a measurable rate. The nearly
linear relationship on the logelog scale indicates a power law
relationship,17which is common to many natural phenomena.
At about 18 months, the fall appears to level off at a rate of
0.01% per day, but more data are needed to confirm the trend.
Table 3 shows the notes most frequently viewed by attend-
ings and residents, and how note viewing evolved with time. We
compared immediate viewing of notes during the hospital stay
(at 1 day) with subsequent clinical viewing (at 1 year). The top
of table 3 shows the 10 physician notes most frequently viewed
by physicians by midnight of the following day, and it shows an
admission and a progress note, which are both commonly
written notes. Internal medicine consult notes were viewed
most frequently. The bottom of table 3 shows the five physician
notes that were viewed at least 10 times in a 10-day period after
1 year (ie, days 365 to 374). After 1 day, the ‘Medicine Resident
Admission Note’ and the ‘Medicine Follow-up Free Text Note’
(the new progress note) were viewed about equally frequently.
After 1 year, the ‘Medicine Resident Admission Note’ and
‘Medicine Resident Daily Progress Note’ (the old progress note)
differ in viewing by more than a factor of 5. Therefore, the
relative use of the progress note fell faster than that of the
We studied the constitution of care teams that formed during
admissions. Admissions were defined as runs of documentation
with no more than 2 days consecutive without new notes where
groups from table 1
Time spent authoring and viewing notes: mean time (min) per day in spent authoring and viewing notes for the larger, clinically relevant user
Time spent (mean no of minutes per day)
Authoring notes Viewing notes
Days with at
least one note authored
multiple notes authored*
at least one note viewed
multiple notes viewed*
Nurse practitioner/physician assistant
Nurse (non-nurse practioner)
*Days with ‘significant involvement,’ defined as authoring or viewing at least two patients per day, at least 2 days in a row, and at least 1 day with at least four patients
yExcludes time spent on ‘flowsheet’ documentation, which includes plan of care, vital signs, intake and output, treatment parameters (isolation, wound care, etc), and data for quality initiatives
such as falls risk assessment
user group, the graph shows the percent of their notes that were viewed
within 3 months by any user in the same or another user group. Viewing
of a note by its author was not counted, and the ‘billing/HIM [health
information management]’ user group was excluded from ‘anyone.’
Use of notes for clinical communication. For authors in each
the bars show how often such users sought out information authored by
the same or another user group when it was available. Viewing of a note
by its author was not counted.
Information-seeking behavior. For each viewer user group,
114J Am Med Inform Assoc 2011;18:112e117. doi:10.1136/jamia.2010.008441
Research and applications
at least 10 notes were written. This was necessary because the
deidentified audit logs lacked the visit data; this iteratively
derived definition allowed for documentation outside of the
official admission window and ensured significant admissions.
We only included those members who had actually written
a note, not just viewed one. All but eight of 69205 so-defined
admissions had nurses on the team. The most frequent team
(9%) comprised a nurse, attending, resident, and social worker;
and the core of a nurse, attending, and resident occurred in 77%
To estimate the delay before team members joined a care
team, we measured the time from the first note for an admission
to any note by an author in a user group. On average, nurses,
residents, and attendings wrote a note within the first day.
Social workers’ notes were documented just over 1 day after the
first note, and nurse practitioners and physician assistants
documented 2 days later. Dieticians and therapists documented
3e5 days later. Medical students also took 3e4 days, possibly
because they were assigned to cases after admission.
We looked for pairwise associations among user groups on care
teams: which user groups were more likely to write a note if
another one did. Due to the large numbers of observations,
almost all pairings were statistically significantly different from
chance, but only one pairing showed a significant deviation.
Social workers and physical or occupational therapists were
more likely to be on a team together with an increased frequency
of 10%; all other pairs differed from chance by 4% or less.
Residents and nurse practitioners or physician assistants were
negatively associated by 4%, most likely because residents did
not participate in the daytime care of patients who were covered
by a nurse practitioner or physician assistant.
We also studied interactive communication among users. We
assessed the RR of a pair of users reading each other’s notes.
That is, user 1 writes a note, user 2 reads it, user 2 writes a note
on the same patient, and user 1 reads it. These cycles are
indicative of two-way communication. Table 4 shows the
observed probability of a pair of users from two user groups
engaging in a cycle, divided by the probability expected by
chance. Bolded entries exceed one at p<0.05, corrected for the
multiple hypotheses contained in the table (all possible pair-
ings). Attendings and residents show the strongest cross-group
cycles, and attendings, residents, occupational and physical
therapists, dietary workers, and social workers had significant
Hierarchical clustering can reveal large-scale organizational
structures. We found that users are generally well interconnected.
Looking only at sets of users who are extremely highly inter-
active (viewing at least 50 of each other’s notes in 2 weeks), one
sees a preponderance of small teams that are likely associated
with hospital floors and service teams, most with two to three
residents at the core. As soon as one drops the threshold of
interactivity (say to 20 views per 2 weeks), a single main set
with most of the users in the hospital emerges. That is, there is
some link among teams, perhaps consultants, or perhaps
workers who cross medical teams (eg, social work). One does
not see a preponderance of divisions along medical specialties
or buildings, nor does one see a separation between user groups
(eg, attending, resident, nurse, etc).
A broad variety of healthcare workers author and view notes in
the electronic health record. Clinical notes serve several
purposes. They augment the author’s memory, create a longi-
tudinal record for continuity of care, provide communication
between users, support quality assurance, substantiate billing,
serve as legal documentation, and support research and educa-
tion. About a sixth of notes overall go unread, thus playing no
direct role in communication, but an unread note may still be
useful because of its potential for future use and because of legal
requirements. About a third of nursing notes go unread; this
higher rate may be due to the use of oral communication
between successive nursing shifts and because much of the
critical information is contained in flowsheets instead of nursing
notes. Medical student notes were read by physicians at a rela-
tively high rate of 81%, implying that supervision is substantial
if not perfect.
The rate of attendings’ and residents’ notes being viewed by
other attendings and residents (37e74% in figure 1) corroborates
earlier findings at a different medical center campus where
physicians in the emergency department viewed clinical notes
from previous encounters 47% of the time, given that they knew
a note was available.18The perception that clinical notes are
difficult to find within an electronic health record
contribute to incomplete note review.
Users spent a moderate amount of time authoring and
viewing notes, with most less than 90 min per day in aggregate.
In the literature, documentation has been reported as 21% of
residents’ time, 12% of attendings’ time, and 7% of emergency
nurses’ time.13Other studies have reported that family practi-
tioners spend 1.2 h per day documenting,14and oncologists
spend 1.4 h per day documenting.15In 1997, internal medicine
residents who were on call were reported to have spent 2.6 h on
paper chart review and 2.2 h on paper documentation, with an
additional half hour on the computer.20
Our results contradict a study of medical resident’s percep-
tions of time spent on documentation.4We found 65 min per
day, whereas residents perceived spending over 4 h. This is likely
due to the survey including order entry in the definition of
‘documentation.’ There is also likely a difference between
perception and measured rates. The same study4found that
medical residents received feedback on documentation less than
50% of the time. Our findings, shown in figure 1, reveal that
attendings review residents’ notes less than 50% of the time,
corroborating this earlier study.
The use of notes over time, shown in figure 3, has implications
for system builders. The use of notes drops off rapidly after the
first day, but even old notes (up to almost 2 years in this study)
got viewed at a low but consistent rate. This implies that recent
of a physician (attending or resident) viewing a physician note on 1 day
given that the note was authored some number of days in the past. Both
scales are logarithmic. The probability of a note being viewed drops
quickly but predictably over time.
Use of notes over time. Each point represents the probability
J Am Med Inform Assoc 2011;18:112e117. doi:10.1136/jamia.2010.008441115
Research and applications
notes should be the most easily accessible, but that older notes
should remain accessible for at least 2 years. As one might
expect, summary notes like admission notes become relatively
more important than progress notes over time. The proven
redundancy among progress notes21explains some of the per-
note drop off: once one progress note is viewed, adjacent ones
become less important. Nevertheless, progress notes are still
viewed at a measurable rate after a year. Therefore, even progress
notes should be retained in the record for some time.
Figure 1 provides relevant information about intergroup
communication. Attendings and residents review nursing and
social work notes less than a third of the time. This may point
to an opportunity for the electronic health record to summarize
information and make it readily available, perhaps with the
ability of the author to highlight information that may be
critical and that has a high priority for communication. Nurses
appear to be reviewing attending and resident information at
a higher rate. Communication rates within groups tend to be
higher, perhaps reflecting the strong relevance of clinical infor-
mation within groups.
In this medical center, the core care team comprised one or
more attendings, residents, and nurses, with a social worker
joining the team soon after. Dieticians and therapists tended to
join the team later in the stay. Within these teams, the tightest
bilateral communication occurred between members of the same
user group and between attendings and residents.
Most frequently viewed physician (attending or resident) notes
Notes viewed after 1 day*
No viewed (1 day) No authored (1 day)
of being viewedNote type
Infectious disease consult note
Heart-failure free text note
Pulmonary fellow consult note
Hematology/oncology attending consult
Rheumatology attending consult note
Medical intensive care unit attending
Medical intensive care unit resident
Nephrology attending admission note
Neurology resident admission note
Pediatric critical care attending note
Medicine resident admission note
Medicine follow-up free text note
Notes viewed at 1 yeary
No viewed (10 days) No authored (10 days)
of being viewedNote type
Neurosurgery resident follow-up note
MD ob labor and delivery progress note
Medicine resident admission note
Medicine resident daily progress note
The top of the table shows the 10 physician notes most frequently viewed by physicians within 1 day; admission and follow-up notes
are added for comparison. The bottom of the table shows the physician notes most frequently viewed after 1 year by physicians. The
per-day probability of being viewed is shown for each note.
*Only the 10 most frequently viewed notes and medicine admission and progress notes are shown.
yAll those with at least 10 notes viewed in 10 days are shown.
reads the other’s note; see text)
Bilateral communication between user groups: RR of a pair of users (one from each group) reading each other’s notes in a cycle (each user
Nurse practioner or physician assistant
Nurse (non-nurse practioner)
*Bold indicates statistically significantly greater than 1 at p<0.05, corrected for multiple hypotheses.
116J Am Med Inform Assoc 2011;18:112e117. doi:10.1136/jamia.2010.008441
Research and applications
The study has several limitations. The study did not capture
in-person or telephone conversations. Therefore, some of the
uncovered lapses in communication may be met orally. For
example, critical results may be more likely to be communicated
orally and immediately. While this study limitation is important,
the results remain useful, as they represent trends, and the rates
may be compared to each other. Another limitation is that we are
inferring actions and intentions from artifacts like notes and logs
of keystrokes; for example, system idle time while a user carries
out other tasks may be counted as authoring or viewing time. A
full timeemotion study with concrete observation of user
activity coupled with a think-aloud protocol to elicit intentions
would complement this one. Such a study would reveal deep
information about a small number of clinical interactions, while
our present study permits a broad view of activity across many
user types. As noted above, the documentation analysis excluded
nursing flowsheets because of the way they are incorporated into
the application, and this will cause an underestimate of nursing
documentation time and can affect the analysis of communica-
tions. The study was conducted at a single academic medical
center, reflecting the workflow of a single institution; a multi-
center study would improve generalizability.
A detailed log of authoring and viewing clinical notes in an
electronic health record revealed documentation workload, team
structure, and levels of communication among team members.
Communication appeared strong in some cases (eg, supervision
of medical students) but weaker in others (eg, nurse to physician
information transfer), which may point to opportunities for
Funding This project was partially supported by grants from the Agency for
Healthcare Research and Quality (R03 HS018250) and from the National Library of
Medicine (R01 LM006910).
Competing interests None.
Ethics approval Ethics approval was provided by the IRBs of Columbia University
Medical Center and Weill Cornell Medical College.
Provenance and peer review Not commissioned; externally peer reviewed.
Schiff GD, Bates DW. Can electronic documentation help prevent diagnostic errors?
N Engl J Med 2010;362:1066e9.
O’Leary KJ, Liebovitz DM, Baker DW. How hospitalists spend their time: insights on
efficiency and safety. J Hosp Med 2006;1:88e93.
Ammenwerth E, Spotl HP. The time needed for clinical documentation versus direct
patient care. A work-sampling analysis of physicians’ activities. Methods Inf Med
Oxentenko AS, West CP, Popkave C, et al. Time spect on clinical documentation:
a survey of internal medicine residents and programs directors. Arch Intern Med
Poissant L, Pereira J, Tamblyn R, et al. The impact of electronic health records on
time efficiency of physicians and nurses: a systematic review. J Am Med Inform
Embi PJ, Yackel TR, Logan JR, et al. Impacts of computerized physician
documentation in a teaching hospital: perceptions of faculty and resident physicians.
J Am Med Inform Assoc 2004;11:300e9.
Stein DM, Wrenn JO, Johnson SB, et al. Signout: a collaborative document with
implications for the future of clinical information systems. AMIA Annu Symp Proc
Amarasingham R, Plantinga L, Diener-West M, et al. Clinical information
technologies and inpatient outcomes: a multiple hospital study. Arch Intern Med
Weir CR, Hurdle JF, Felgar MA, et al. Direct text entry in electronic progress notes:
An evaluation of input errors. Methods Inf Med 2003;42:61e7.
Hirschtick RE. A piece of my mind. Copy-and-paste. JAMA 2006;295:2335e6.
Asaro PV, Ries JE. Data mining in medical record access logs. Proc AMIA Symp
Chen ES, Cimino JJ. Automated discovery of patient-specific clinician information
needs using clinical information system log files. AMIA Annu Symp Proc 2003:
Hollingsworth JC, Chisholm CD, Giles BK, et al. How do physicians and nurses
spend their time in the emergency department? Ann Emerg Med 1998;31:87e91.
Gottschalk A, Flocke SA. Time spent in face-to-face patient care and work outside
the examination room. Ann Fam Med 2005;3:488e93.
Einhorn LH, Levinson J, Li S, et al. American Society of Clinical Oncology 2001
Presidential Initiative: impact of regulatory burdens on quality cancer care. J Clin
Newman MEJ. The structure and function of complex networks. SIAM Review
Clauset A, Shalizi CR, Newman MEJ. Power-law distributions in empirical data.
SIAM Review 2009;51:661e703.
Hripcsak G, Sengupta S, Wilcox A, et al. Emergency department access to
a longitudinal medical record. J Am Med Inform Assoc 2007;14:235e8.
Payne TH, tenBroek AE, Fletcher GS, et al. Transition from paper to electronic
inpatient physician notes. J Am Med Inform Assoc 2010;17:108e11.
Moore SS, Nettleman MD, Beyer S, et al. How residents spend their nights on call.
Acad Med 2000;75:1021e4.
Wrenn JO, Stein DM, Bakken S, et al. Quantifying clinical narrative redundancy in an
electronic health record. J Am Med Inform Assoc 2010;17:49e53.
J Am Med Inform Assoc 2011;18:112e117. doi:10.1136/jamia.2010.008441 117
Research and applications