How much do operational processes affect hospital inpatient discharge rates?

Article (PDF Available)inJournal of Public Health 31(4):546-53 · June 2009with17 Reads
DOI: 10.1093/pubmed/fdp044 · Source: PubMed
Abstract
The objective of this study is to determine the effect of day of the week, holiday, team admission and rotation schedules, individual attending physicians and their length of coverage on daily team discharge rates. We conducted a retrospective analysis of the General Internal Medicine (GIM) inpatient service at our institution for years 2005 and 2006, which included 5088 patients under GIM care. Weekend discharge rate was more than 50% lower compared with reference rates whereas Friday rates were 24% higher. Holiday Monday discharge rates were 65% lower than regular Mondays, with an increase in pre-holiday discharge rates. Teams that were on-call or that were on call the next day had 15% higher discharge rates compared with reference whereas teams that were post-call had 20% lower rates. Individual attending physicians and length of attending coverage contributed small variations in discharge rates. Resident scheduling was not a significant predictor of discharge rates. Day of the week and holidays followed by team organization and scheduling are significant predictors of daily variation in discharge rates. Introducing greater holiday and weekend capacity as well as reorganizing internal processes such as admitting and attending schedules may potentially optimize discharge rates.

Figures

How much do operational processes affect hospital inpatient
discharge rates?
Hannah Wong
1,2,3
, Robert C. Wu
4,5
, George Tomlinson
4,6
, Michael Caesar
3,4
,
Howard Abrams
4,5,7
, Michael W. Carter
1,2
, Dante Morra
4,5
1
Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada
2
Centre for Research in Health Care Engineering, University of Toronto, Toronto, ON, Canada
3
Shared Information Management Services, University Health Network, Toronto, ON, Canada
4
Centre for Innovation in Complex Care, Toronto General Hospital, University Health Network, Toronto, ON, Canada
5
Department of Medicine, University of Toronto, Toronto, ON, Canada
6
Department of Public Health Sciences, University of Toronto, Toronto, ON, Canada
7
Department of Health Policy, Management, and Evaluation, University of Toronto, Toronto, ON, Canada
Address correspondence to Dante Morra, Email: dante.morra@uhn.on.ca
ABSTRACT
Background The objective of this study is to determine the effect of day of the week, holiday, team admission and rotation schedules, individual
attending physicians and their length of coverage on daily team discharge rates.
Methods We conducted a retrospective analysis of the General Internal Medicine (GIM) inpatient service at our institution for years 2005 and
2006, which included 5088 patients under GIM care.
Results Weekend discharge rate was more than 50% lower compared with reference rates whereas Friday rates were 24% higher. Holiday
Monday discharge rates were 65% lower than regular Mondays, with an increase in pre-holiday discharge rates. Teams that were on-call or that
were on call the next day had 15% higher discharge rates compared with reference whereas teams that were post-call had 20% lower rates.
Individual attending physicians and length of attending coverage contributed small variations in discharge rates. Resident scheduling was not a
significant predictor of discharge rates.
Conclusions Day of the week and holidays followed by team organization and scheduling are significant predictors of daily variation in discharge
rates. Introducing greater holiday and weekend capacity as well as reorganizing internal processes such as admitting and attending schedules may
potentially optimize discharge rates.
Keywords day of the week, discharge rate, holiday, operational efficiency, scheduling
Introduction
Improving patient flow in acute care hospitals is an impor-
tant issue in hospital management and research. Improved
patient flow can decrease wait times for care, ease
Emergency Department (ED) congestion, and increase the
effective capacities of the ICU and inpatient units.
1–4
One
way to improve patient flow is to remove variation in pro-
cesses along the care pathway that can block or delay flow.
5
One process in particular—patient discharge—has received
critical attention because variation and delays in this process
create ‘bottlenecks’ that ultimately delay most care pathways,
especially new admissions from the ED.
6,7
Although the decision to discharge an individual patient
from hospital should predominately be a clinical decision,
there may be non-clinical factors that influence decision-
making. These may include patient and family preferences,
8
physician practice preferences,
9
internal hospital inefficien-
cies,
10
post-acute care bed capacity
11
and healthcare finan-
cing arrangements.
12
Hannah Wong, PhD Candidate
Robert C. Wu, Assistant Professor
George Tomlinson, Associate Professor
Michael Caesar, Director of Information Management
Howard Abrams, Associate Professor
Michael W. Carter, Professor
Dante Morra, Assistant Professor
546 # The Author 2009, Published by Oxford University Press on behalf of Faculty of Public Health. All rights reserved
Journal of Public Health | Vol. 31, No. 4, pp. 546553 | doi:10.1093/pubmed/fdp044 | Advance Access Publication 22 May 2009
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
The impact of non-clinical hospital discharge delays on
costs, quality and appropriateness of care has gar nered inter-
national attention. In the United States where hospitals are
reimbursed prospectively, one study found that 7% of
hospital days were judged unnecessary and were due to diffi-
culty finding a bed in a skilled nursing facility.
13
In the UK,
where transfer from acute inpatient hospital care to ongoing
health and social care is subject to means-testing and user
charges, the Community Care Act 2003 was introduced that
allowed hospital trusts to charge social service depar tments
(providing ongoing health and social care in the community)
for hospital beds unnecessarily ‘blocked’ by people awaiting
social services.
14
In Canada, universal health insurance
covers medically necessary in- and out-patient hospital ser-
vices as well as extended health services (certain aspects of
long-term residential care and the health aspects of home
care and ambulatory care services).
15
Hospitals receive block
funding and chronically face budget deficits.
16
There is
increasing accountability on hospitals to meet benchmarks
on wait times in the ED, for diagnostic imaging, and for
selected elective surgeries, all whereas maintaining a
balanced budget.
17
Thus, patients occupying acute care beds
who are awaiting transfer to a community facility represent a
significant challenge in timely acute care hospital
discharge.
18
Although efforts have been made to better coordinate
care between acute and post-acute sectors,
19
understanding
factors within an acute care hospital’s control should be a
top priority, as reducing them may improve efficiency.
Several studies have focused on internal hospital oper-
ational factors including the day of the week of admission
in terms of inter nal resource availability,
20,21
team organiz-
ation and workload,
22 24
and clin ician behaviour
25
to
understand how they contribute to variations in discharge.
In these studies, the primary outcome measure was length
of stay (LOS). An alternative metric to measure the impact
of hospital operational factors on the discharge process is
daily discharge rate.
26
Daily discharge rate incorporates day
of the week explicitly and may highlight other operational
factors that exhibit temporal patterns including staffing and
scheduling.
The objective of this study was to determine the effect
that the following operational factors had on discharge rates:
day of the week, holiday status, team admission schedule,
resident scheduling, individual attending physicians and the
length of attending physician coverage. Our primary
outcome measure was mean daily team discharge rate,
expressed as the number of team discharges divided by
team census on a particular day.
Methods
We conducted a retrospective analysis of the General
Internal Medicine (GIM) inpatient service at Toronto
General Hospital for two consecutive years, 2005 and 2006.
We wanted to determine the effect of day of the week,
holiday status, team admission schedule, resident scheduling
and attending physician schedule on team discharge rates.
University Health Network (UHN) Research Ethics Board
approved this study.
Setting
This study was conducted on the GIM inpatient service at
the UHN’s Toronto General Hospital site, a 400-bed ter tiary
care centre and teaching hospital located in downtown
Toronto, Canada. GIM provides acute, non-surgical health
care to a patient population primarily composed of elderly
patients with complex, chronic illnesses. GIM receives 98%
of its inpatient admissions from the ED, other sources
being transfers from other services (CCU, ICU, surgery, etc.)
and other institutions. Of all patients requiring admission
from the ED, GIM receives the largest share (30 50% of
all ED admissions).
During the study period, there were four admitting teams.
Each team consisted of one attending physician, one resi-
dent and two interns. Each day, the team assigned to be
on-call accepted new admissions from 8 AM to 8 AM the
following day, at which time they transitioned to post-call
status. During weekends, the on-call team was responsible
for all admitting duties as well as clinical duties for all GIM
admissions. Teams (resident and interns) rotated every 2
months, whereas attending physicians normally rotated on a
monthly basis. Attending physician rotations could range
from half a month to two consecutive months. Attending
physicians were not assigned exclusively to one team.
Data collection
Data was collected from January 15th to December 15th for
the 2 years, 2005 and 2006. We excluded the period
December 16thJanuary 14th from our analyses due to
Christmas/New Year holiday disruptions to physician team
structures. To simplify analyses, we only selected holidays
that occurred on a Monday, thus excluding Easter (March
2429, 2005 and April 13 18, 2006) and Canada Day
(June 30July 4, 2005 and June 30July 4, 2006) from
analysis. The unit of analysis was a day, defined as the 24 h
period from 08:00 to 08:00. The time period of 08:00
08:00 was chosen because it better reflects the period when
decisions are made and work is completed.
OPERATIONAL PROCESSES THAT AFFECT DISCHARGE RATES 547
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
Patient-level data was obtained from UHN’s primary
patient care system Electronic Patient Record (EPR). EPR
contains information pertaining to sociodemographics, diag-
nosis, LOS, patient disposition, attending physician, admis-
sion and discharge dates and times. Attending physician
schedules, team admission schedules, and resident schedules
were obtained from team rosters maintained by the GIM
residency program. We included all GIM inpatient admis-
sions that were under the care of GIM services during the
study period (i.e. patients whose admission or discharge
dates were within the study period and patients admitted
before the start of the study period and discharged after the
end of the study period).
Outcomes
Our primary study outcome was mean daily team discharge
rate, expressed as the number of discharges divided by
census for a specific team on a particular day. Daily team
discharge rate was chosen as opposed to the overall dis-
charge rate because each team effectively acted as an inde-
pendent unit and it allowed us to look at variations caused
by scheduling. Daily team census was measured at 8 AM.
To retain a focus on operational factors that can act as bot-
tlenecks in discharge, we excluded discharges with disposi-
tion of either death or left against medical advice from daily
discharge rate calculations. These visits were however main-
tained for daily census calculations.
Predictors
Day of the week and holiday status
We examined if day of the week or a holiday period was a
predictor of daily team discharge rates. Holiday Mondays
included Victoria Day (23 May 2005 and 22 May 2006),
Civic Holiday (1 August 1 2005 and 7 August 2006),
Labour Day (5 September 2005 and 4 September 2006) and
Thanksgiving Day (10 October 2005 and 9 October 2006).
A priori, we believed that in anticipation of a Holiday
Monday, proactive measures to increase discharge rates
would be taken pre-holiday period (either by physician pur-
poseful behaviour or by request of patient and/or patient’s
family), or conversely reactive measures would be taken
post-holiday period. Fridays, Saturdays, and Sundays
immediately preceding a Holiday Monday were defined as
pre-holiday days. Tuesdays immediately following a Holiday
Monday were defined as post-holiday days. ‘Regular’ week-
days and weekends included all days in the study period
except pre-/post-/Holiday Mondays listed above.
Clinical scheduling
We examined whether the scheduling of team admissions,
resident scheduling or attending physicians was a predictor
of daily team discharge rates. According to the team admis-
sion schedule, each day, each team was assigned either
pre-call, on-call, post-call, postpost-call or no-call status
(team neither pre-call, on-call, post-call nor postpost-call)
(Fig. 1). Each call-status was equally likely to occur during
the 7 days of the week, with the exception of no-call
status, which only occurred on weekends. We anticipated
that discharge rates for post-call teams would be signifi-
cantly decreased since post-call teams were relieved of hos-
pital duties at 12 PM. We hypothesized that there would be
an increase in discharge rates for pre-call teams in antici-
pation of the increased workload of new admissions the
following day.
We also examined whether resident scheduling was a pre-
dictor of daily team discharge rates. Residents are scheduled
for durations of 2 months. We anticipated that there would
be differences in team discharge rates for three distinct
periods during the 2 months: first week, last week and all
remaining weeks. We hypothesized that in the first week of
the 2 month period, teams were newly acquainted with their
patients and as a result, would be less likely to discharge
them compared with remaining weeks. In contrast, in the
last week of the 2 months, teams were well acquainted with
their patients and would be more likely to discharge them
compared with other weeks.
Fig. 1 A typical 4-week period for one of the admitting teams. Note the other teams would have similar schedules. A team is on-call for a 24-h period
(Monday 8 AM to Tuesday 8 AM). After working for the 24 h, the team is now considered post-call, and would leave early after signing over their patients.
The next day is considered the postpost-call day. Note that while a team is on-call 1 in 4 days, a strict every fourth day on call is not followed. This results
in certain time when teams could be considered both postpost-call and on-call. For the purposes of this analysis, we considered that team on-call.
548 JOURNAL OF PUBLIC HEALTH
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
Finally, we examined whether individual attending phys-
icians had different daily team discharge rates and whether
the length of coverage assigned to an attending affected dis-
charge rates. We anticipated no major difference in aggregate
daily discharge rate among the different attending physicians
during the course of the study period; however, we hypoth-
esized that there would be differences in discharge rates for
physicians attending for shorter periods (i.e. 15 days)
versus longer coverage periods (i.e. 30 days). Physicians
who are attending for short periods may be less inclined to
be actively involved in discharge planning. In addition, phys-
icians nearing their end of their coverage period may be less
motivated to discharge patients. To account for the length
of attending service and proximity to end of service cover-
age for each day of the study period, we calculated the
numbers of days since the start of team coverage and the
numbers of days till the end of team coverage for each of
the four attending physicians on service.
Statistical analyses
Preliminary analysis of the data included the calculation of
descriptive statistics to summarize GIM patient admission
and service characteristics. We reported medians and inter-
quartile ranges for continuous variables and proportions for
categorical variables. All regression analyses used a linear
mixed random effects model with the number of discharges
by a team on a day treated as a Poisson random variable. In
all analyses, we accounted for the effects of clustering of the
four team outcomes within a day by including a random
effect for day. All analyses also included a six-degree-
of-freedom natural spline fitted to the calendar time variable
to account for seasonal fluctuation and cor relation between
outcomes on adjacent days.
In four separate univariable reg ressions, we assessed the
relationship between discharg e rate and each of the predictor
variables (type of day, team admission schedule, resident
scheduling and the length of attending physician coverage).
We also ran a model with a random effect for the attending
physician; this allowed each attending physician to have a
discharge rate that is either higher or lower than the average.
Finally, we fitted a model with all four predictors and the
random effect for attending physician. We report the rate
ratios and 95% confidence intervals, compared with a refer-
ence group for categorical predictors, and per 30 days for
attending physician coverage. We also report the standard
deviation between log-rate ratios for individual attending
physicians. All statistical analyses were performed using R
(version 2.8.0: R Foundation for Statistical Computing,
Vienna, Austria); and two-tailed P , 0.05 was considered
statistically significant in all analyses.
Results
Patient and service characteristics
During the 648-day study period, there were 5088 patients
under the care of GIM services (Table 1). About 98% of
patients were admitted from the ED, 2% were directly
admitted. The median age of patients was 68 years and 48%
were women. Patient dispositions following GIM care
during the study period included: discharge home (72%),
transfer to other facility including other acute care facilities,
rehabilitation, long-term care, respite care, complex continu-
ing care (14%), in-hospital death (7%), transfer to another
inpatient service within UHN (6%) and lastly left against
medical advice (2%). The overall median LOS was 5 days.
Measured at 8 AM, daily median GIM census was 66
patients. On a daily basis, both the median number of
admissions and discharges to and from GIM services was 7
patients. At the team level, median daily census was 16
patients, median number of daily discharges was one patient,
and median daily team discharge rate was 9% of team
census.
Day of the week and holiday status
The day of the week accounted for a significant amount of
variation in discharge rates in both unadjusted and adjusted
models (Fig. 2 and Table 2). Wednesday was chosen as the
Table 1 GIM patient and service characteristics
GIM patient characteristic Value N ¼ 5088 patients
Admitted via ED, no. (%) 4994 (98.2)
Median age, years (IQR) 68 (5480)
Female sex, no. (%) 2452 (48.2)
Inpatient Mortality, no. (%) 334 (6.6)
Left against medical advice, no. (%) 86 (1.7)
Median length of stay, days (IQR) 5 (310)
GIM service characteristic Value N ¼ 648 days
Median census in @ 8 AM, patients (IQR) 66 (6072)
Median team census @ 8AM, patients (IQR) 16 (1319)
Median admissions, patients (IQR) 7 (59)
Median discharges, patients (IQR) 7 (49)
Median team discharges, patients (IQR) 1 (13)
Median daily team discharge rate,
% team census (IQR)
9 (416)
OPERATIONAL PROCESSES THAT AFFECT DISCHARGE RATES 549
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
reference. For regular weekends, both Saturday and Sunday
had significantly lower (by 50% P , 0.001, and 71% P ,
0.001, respectively) adjusted rates of discharge relative to
Wednesday. For regular weekdays, whereas Monday, Tuesday
and Thursday had similar adjusted discharge rates relative to
Wednesday (8, 5 and 6% lower, respectively), Friday had sig-
nificantly higher discharges (by 24%, P , 0.001) relative to
Wednesday.
Adjusted rates of discharge on Holiday-Monday were sig-
nificantly lower than regular Mondays with Holiday-Monday
discharge rates being 65% lower than rates on a regular
Monday (P , 0.001). Discharge rates on a pre-holiday
Friday were 27% higher than rates on a regular Friday (P ¼
0.04). Although pre-holiday Saturday and Sunday had rates
that were higher than their corresponding regular weekend
days, these increases were not statistically significant
(Saturday: 40% higher, P ¼ 0.06; and Sunday: 34% higher,
P ¼ 0.22). In contrast, discharge rates on a post-holiday
Tuesday were similar to regular Tuesdays (P ¼ 0.17).
Clinical scheduling
Following type of day, team admission schedule ranked as
the next most important predictor of variation in discharge
rates in both unadjusted and adjusted models. Postpost-
call teams were chosen as the reference. Relative to post
post call teams, pre-call and on-call teams had significantly
higher (by 17 and 15%, respectively) adjusted rates of dis-
charge (P , 0.001, P ¼ 0.002, respectively), whereas post-
call and no-call teams had significantly lower (by 20 and
27%, respectively) adjusted rates of discharge (P , 0.001,
P ¼ 0.005, respectively).
Resident scheduling was not a predictor of discharge rates
in either unadjusted or adjusted models. Weeks other than
the first or last weeks of resident rotation were chosen as
the reference. Adjusted discharge rates during the first and
last weeks of resident rotation schedule were similar to
middle weeks (P ¼ 0.65 and P ¼ 0.67, respectively).
Individual attending physicians and their length of sched-
uled coverage was a predictor of discharge rates in the
adjusted model. There were 28 different attending physicians
that covered the four teams during the study period. There
was a small but significant random effect for the attending
physicians in the adjusted model, wherein attending phys-
icians had discharge rates that ranged from 8% below
average to 4% above average (P ¼ 0.006). Length of attend-
ing physician coverage ranged from a minimum of 14 con-
secutive days to a maximum of 64 consecutive days. There
was a small but significant increase in adjusted rates of daily
discharge as the term length (days from start of attending
coverage) increased (by 10% per 30 days attending, P ¼
0.04). Proximity to end of term coverage (days till end of
attending coverage) was not a significant predictor of daily
discharge rate (P ¼ 0.84).
Discussion
Main finding of this study
We found that discharge rates within a GIM inpatient
service were affected by the day of the week, holiday status,
team admission schedule, individual attending physicians
and their length of attending coverage. Adjusted discharge
rate ratios and 95% confidence intervals for the majority of
predictors were virtually identical to the unadjusted dis-
charge rate ratios, confirming that the effects of the predic-
tors are independent of each other. Weekend discharge rate
was significantly lower than weekday discharge rate. More
discharges occurred on Fridays than other days, possibly to
compensate for reduced weekend discharges. Holiday
Monday discharge rates were significantly lower than regular
Mondays with increases in pre-holiday discharg e rates. More
discharges occurred when a team was either pre-call or
on-call status, and significantly less discharges occurred for
teams on post-call status. Resident scheduling was not a sig-
nificant predictor of team discharge rates. Individual attend-
ing physicians had small but significant variation in team
Fig. 2 Adjusted discharge ratios and 95% confidence intervals evaluating
day of the week and holiday periods. Wednesday is reference variable.
Each measure reports the change in proportion of team census discharged
compared with reference.
550 JOURNAL OF PUBLIC HEALTH
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
discharge rates, and as the length of attending coverage
increased, so did the rate of discharge.
What is already known on this topic
We investigated the effect of weekends and holidays on dis-
charge rates because it has received considerable attention as a
source of hospital inefficiency.
27,28
Although discharges and
discharge rates should be predominately clinical decision,
there is significant variation caused by hospital processes.
Understanding these variations may be key to improving effi-
ciency. A study by Carey et al.
13
to quantify delays in care for
general medicine inpatients found that nearly 25% of
unnecessary patient-days involved an inability to access
medical services on a weekend day. The data on weekends
reported here are consistent with and extend findings from
previous studies that have evaluated the effect of weekends on
patient flow. It has been previously shown that discharge rates
are significantly higher on Fridays and lower over the week-
ends and that these results are independent of clinical indi-
cators of risk.
29
One explanation for this phenomenon is that
availability of hospital resources including physicians and hos-
pital staff are decreased on weekends. To avoid the weekend,
patients were preferably discharged on Friday. A similar expla-
nation may be used to explain increased pre-holiday discharges
relative to their regular counterpart days. Reduced weekend
capacity also applies to many community resources and post-
acute care institutions making it unfeasible to discharge
patients that require home care or transfer to other institutions
on weekends. Patient and family preferences may also be rel-
evant in determining the day of week of discharge.
Table 2 Univariate and multivariate analysis of operational factors on team discharge rates
Variable Univariate analysis (unadjusted) Multivariate analysis (adjusted)
Discharge Rate Ratio 95% CI P value Discharge Rate Ratio 95% CI P value
Day of the week
Holiday Monday 0.32 0.200.49 , 0.001 0.32 0.210.51 ,0.001
Monday 0.90 0.811.00 0.05 0.92 0.831.02 0.11
Postholiday Tuesday 1.12 0.891.43 0.34 1.13 0.891.44 0.32
Tuesday 0.95 0.861.05 0.31 0.95 0.861.05 0.35
Wednesday* 1 1
Thursday 0.93 0.851.03 0.19 0.94 0.851.03 0.19
Pre-holiday Friday 1.54 1.25 1.89 ,0.001 1.57 1.281.93 ,0.001
Friday 1.24 1.13 1.37 ,0.001 1.24 1.131.37 ,0.001
Pre-holiday Saturday 0.62 0.44 0.86 0.004 0.70 0.50 0.98 0.04
Saturday 0.45 0.39 0.51 ,0.001 0.50 0.430.57 ,0.001
Pre-holiday Sunday 0.36 0.24 0.55 ,0.001 0.39 0.260.6 ,0.001
Sunday 0.27 0.23 0.32 ,0.001 0.29 0.250.34 ,0.001
Attending physician coverage
Days from start of coverage
1.01 0.93 1.10 0.76 1.10 1.00 1.20 0.04
Days to end of coverage
1.02 0.94 1.12 0.58 1.01 0.92 1.11 0.84
Team admitting s chedule
Pre-call 1.15 1.06 1.25 0.001 1.17 1.07 1.27 ,0.001
On-call 1.06 0.97 1.15 0.21 1.15 1.05 1.25 0.002
Post-call 0.75 0.69 0.82 ,0.001 0.80 0.740.87 ,0.001
No-call 0.40 0.32 0.49 ,0.001 0.73 0.580.91 0.005
Postpost-call* 1 1
Resident rotation schedule
First week 0.94 0.81 1.09 0.427 0.98 0.88 1.09 0.651
Last week 1.08 0.94 1.24 0.298 1.02 0.92 1.13 0.668
Not first or last week* 1 1
Random effect of attending SD ¼ 0.054 0.04 SD ¼ 0.061 0.006
*Reference variable.
Per 30 days.
OPERATIONAL PROCESSES THAT AFFECT DISCHARGE RATES 551
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
What this study adds
Our study introduces daily team discharge rate, expressed as
the number of discharges divided by census for a specific
team on a particular day. Although daily discharge rate expli-
citly incorporates day of the week trends, there are implicit
factors operating at many different levels that influence its
value. Likelihood of discharge is affected by the composition
of the ultimate discharge destinations of the current patient
census (home versus transfer to another facility). Similarly
the proportion of team census considered ‘long-stay’ will
also affect daily team discharge rate.
Our study further highlights the importance of consider-
ing full utilization of hospital capital designed to operate 7
days a week. Bell and Redelmeier
30
found that even conser-
vative growth in weekend service can achieve an increase in
procedure volumes of 15%. Our study also showed that
team organization and scheduling is a significant predictor
of daily team discharge rates. Teams discharged a signifi-
cantly greater proportion of their patients during pre-call
and on-call days and significantly fewer on post-call days.
Clinically, pre-call teams have patients that have had at least
3 days of diagnostic and therapeutic care and therefore
would be expected to be more ready for discharge than
post-call teams that have a larger proportion of their patient
census consisting of new patients with active clinical issues.
A plausible non-clinical explanation for increased pre-/
on-call discharges is that in anticipation of increased work-
load/census due to new admissions, teams purposefully dis-
charged more of their patients. Although in our admission
system, we get a ‘bolus’ of patients on average every 4 days,
other systems have been proposed that have daily admis-
sions, a ‘drip’ admission process. If our traditional ‘bolus’
admitting system in place was replaced with ‘drip’ admis-
sions, workload would be more manageable and discharges
would occur in a more uniform and predictable manner.
31
We also compared team discharge rates in the first week,
last week and remaining weeks of a resident rotation schedule
and found no substantive differences. This finding may be in
part because of attending physician scheduling. Since attend-
ing physicians lead team decision-making but do not follow
the resident rotation schedule, similar rates of discharge
during the course of the resident rotation schedule may not
be sur prising. A study by Smith et al.
32
did find that the last 3
days of the month was an independent predictor of LOS.
Lastly, we observed that individual attending physicians had
small but significant differences in discharge rates and found
that attending physicians with shorter coverage periods
tended to have lower discharge rates compared with attending
physicians with longer coverage periods. Physicians who
attend for longer periods of time may be more invested in
discharge processes and discharge planning. Further study is
required to investigate optimum scheduling and durations of
attending coverage to improve efficiency.
Limitations of this study
This study has several limitations. It took place in one large
teaching institution with resident and attending physician
staffing, and as a consequence, may be less generalizable to
other community hospitals. Nonetheless, clinical scheduling
policies and availability of resources on weekends and holi-
days affect all hospitals. Also, we excluded the period
December 16thJanuary 14th from our analyses due to
Christmas/New Year holiday disruptions to physician team
structures and holidays not landing on a Monday. Therefore,
although our results may underestimate the impact of holi-
days on discharge rates, they still reveal important trends in
discharge rate during holidays.
Conclusion
In conclusion, our findings suggest that day of the week,
holiday status and team admission schedule significantly
influenced daily discharge rates. Individual attending phys-
icians and their length of coverage have a small influence in
discharge rate. Introducing greater holiday and weekend
capacity as well as reorganizing inter nal processes such as
admitting and attending schedules may potentially optimize
discharge rates. Discharge rate, expressed as the number of
discharges divided by census on a particular day, may be a
useful metric to measure the impact of operational factors
on the discharge process. One particular benefit of dis-
charge rate is that it can be used to actively monitor and
improve operations in more real-time as opposed to lagging
indicators of performance. Further study in other inpatient
settings is required to confirm the appropriateness of dis-
charge rate as a performance metric of discharge efficiency.
References
1 Forster AJ, Stiell I, Wells G et al. The effect of hospital occupancy
on emergency department length of stay and patient disposition.
Acad Emerg Med 2003;10(2):12733.
2 Hoot NR, Aronsky D. Systematic review of emergency department
crowding: causes, effects, and solutions. Ann Emerg Med
2008;52(2):12636.
3 Proudlove NC, Gordon K, Boaden R. Can good bed management
solve the overcrowding in accident and emergency departments?
Emerg Med J 2003;20(2):14955.
552 JOURNAL OF PUBLIC HEALTH
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
4 Van Houdenhoven M, van Oostrum JM, Wullink G et al.Fewer
intensive care unit refusals and a higher capacity utilization by using
a cyclic surgical case schedule. J Crit Care 2008;23(2):222 6.
5 Institute for Healthcare Improvement. Optimizing Patient Flow: Moving
Patients Smoothly through Acute Care Settings. Digital document available
at www.ihi.org, 2003.
6 Asplin BR, Magid DJ. If you want to fix crowding, start by fixing
your hospital. Ann Emerg Med 2007;49(3):273 4.
7 Canadian Institute for Health Information. Understanding emer-
gency department wait times: access to inpatient beds and patient
flow. Canadian Institute for Health Information (Ottawa: CIHI, 2007).
Digital document available at www.cihi.ca, 2007.
8 Bryan K, Gage H, Gilbert K. Delayed transfers of older people
from hospital: causes and policy implications. Health Policy
2006;76(2):194201.
9 She pperd S, Parkes J, McClaren J et al. Discharge planning
from hospital to home. Cochrane Database Syst Rev
2004;(1):CD000313, DOI: 10.1002/14651858.CD000313.pub2.
10 Baumann M, Evans S, Perkins M et al. Organisation and features of
hospital, intermediate care and social services in English sites with
low rates of delayed discharge. Health Soc Care Community
2007;15(4):295 305.
11 Glasby J, Littlechild R, Pryce K. All dressed up but nowhere to go?
Delayed hospital discharges and older people. J Health Serv Res Policy
2006;11(1):52 8.
12 Godden S, McCoy D, Pollock A. Policy on the rebound: trends and
causes of delayed discharges in the NHS. J R Soc Med
2009;102(1):22 8.
13 Carey MR, Sheth H, Braithwaite RS. A prospective study of reasons
for prolonged hospitalizations on a general medicine teaching
service. J Gen Intern Med 2005;20(2):108 15.
14 McCoy D, Godden S, Pollock AM et al. Carrot and sticks? The
Community Care Act (2003) and the effect of financial incentives
on delays in discharge from hospitals in England. J Public Health
(Oxf) 2007;29(3):281 7.
15 Health Canada. Canada Health Act Annual Report 20072008.
Digital document available at http://www.hc-sc.gc.ca, 2008.
16 Blake TJ, Carter MW. Physician and hospital funding options in a
public system with decreasing resources. Socio-Econ Plann Sci
2003;37(1):45 68.
17 Canadian Institute for Health Information. Waiting for health care
in Canada: what we know and what we don’t know. Canadian
Institute for Health Information (Ottawa: CIHI, 2006). Digital document
available at www.cihi.ca, 2006.
18 Ontario Hospital Association. Alternate Level of Care Challenges
and Opportunities. Ontario Hospital Association (OHA). Digital docu-
ment available at www.oha.com, 2006.
19 Henwood M. Effective partnership working: a case study of hospi-
tal discharge. Health Soc Care Community 2006;14
(5):4007.
20
Earnest A, Chen MI, Seow E. Exploring if day and time of admis-
sion is associated with average length of stay among inpatients from
a tertiary hospital in Singapore: an analytic study based on routine
admission data. BMC Health Serv Res 2006;6:6.
21 Fonarow GC, Abraham WT, Albert NM et al. Day of admission
and clinical outcomes for patients hospitalized for heart failure:
findings from the Organized Program to Initate Lifesaving
Treatment in Hospitalized Patients With Heart Failure
(OPTIMIZE-HF). Circ Heart Fail 2008;1:507.
22 Friedman DM, Berger DL. Improving team structure and com-
munication a key to hospital efficiency. Arch Surg 2004;139:1194 8.
23 Schuberth JL, Elasy TA, Butler J et al. Effect of short call admission
on length of stay and quality of care for acute decompensated heart
failure. Circulation 2008;117(20):2637 44.
24 Ong M, Bostrom A, Vidyarthi A et al. House staff team workload and
organization effects on patient outcomes in an academic general
internal medicine inpatient service. Arch Intern Med 2007;167(1):47 52.
25 Laing R, Coles C, Chambers S et al. Community-acquired pneumo-
nia: influence of management practices on length of hospital stay.
Intern Med J 2004;34(3):917.
26 Chow KM, Szeto CC. Impact of enforcing the Labour Ordinance,
with 1-in-7-day off for hospital doctors, on weekend hospital dis-
charge rate. J Public Health (Oxf ) 2005;27(2):18991.
27 Redelmeier DA, Bell CM Weekend worriers. N Engl J Med
2007;356(11):11645.
28 Weissman JS, Bendavid E. Should U.S. hospitals go 24/7? Am J
Med 2004;117(3):2023.
29 Varnava AM, Sedgwick JE, Deaner A et al. Restricted weekend
service inappropriately delays discharge after acute myocardial
infarction. Heart 2002;87(3):216 9.
30 Bell CM, Redelmeier DA. Enhanced weekend service: an affordable
means to increased hospital procedure volume. CMAJ
2005;172(4):503 4.
31 Arora VM, Georgitis E, Siddique J et al. Association of workload of
on-call medical interns with on-call sleep duration, shift duration, and
participation in educational activities. JAMA 2008;300(10):114653.
32 Smith JP, Mehta RH, Das SK et al. Effects of end-of-month admis-
sion on length of stay and quality of care among inpatients with
myocardial infarction. Am J Med 2002;113(4):28893.
OPERATIONAL PROCESSES THAT AFFECT DISCHARGE RATES 553
by guest on February 2, 2016http://jpubhealth.oxfordjournals.org/Downloaded from
    • "Administrators and healthcare providers are particularly focused on minimizing patient length of stay, ensuring a safe and high-quality patient discharge experience and reducing unnecessary readmissions (Connolly et al., 2009; Glasby, Littlechild, & Pryce, 2006; Ontario Ministry of Health Care and Long Term Care, 2010). The concern with length of stay is linked to organizational imperatives concerning patient flow, and the aims to optimize bed availability, minimize emergency department wait times, and reduce costs (Office of the Auditor General of Ontario, 2010; Wong et al., 2009). The emphasis on patient safety and quality of care reflects the potential for adverse events to occur during this period of transition and the goal to minimize readmissions (Davis, Devoe, Kansagara, Nicolaidis, & Englander, 2012; Romagnoli, Handler, Ligons, & Hochheiser, 2013). "
    [Show abstract] [Hide abstract] ABSTRACT: Patient discharge is a key concern in hospitals, particularly in acute care, given the multifaceted and challenging nature of patients' healthcare needs. Policies on discharge have identified the importance of interprofessional collaboration, yet research has described its limitations in this clinical context. This study aimed to extend our understanding of interprofessional interactions related to discharge in a general internal medicine setting by using sociological theories to illuminate the existence of, and interplay between, structural factors and microlevel practices. An ethnographic approach was employed to obtain an in-depth insight into healthcare providers' perspectives, behaviours, and interactions regarding discharge. Data collection involved observations, interviews, and document analysis. Approximately 65 hours of observations were undertaken, 23 interviews were conducted with healthcare providers, and government and hospital discharge documents were collected. Data were analysed using a directed content approach. The findings indicate the existence of a medically dominated division of healthcare labour in patient discharge with opportunities for some interprofessional negotiations; the role of organizational routines in facilitating and challenging interprofessional negotiations in patient discharge; and tensions in organizational priorities that impact an interprofessional approach to discharge. The findings provide insight into the various levels at which interventions can be targeted to improve interprofessional collaboration in discharge while recognizing the organizational tensions that challenge an interprofessional approach.
    Full-text · Article · Feb 2016
    • "We noticed a weekly pattern (Fig. 2a) and monthly pattern (Fig. 2b) in discharges from the ward. Other studies have also confirmed that discharges 340 peak on Friday and drop during weekends [14, 5, 15]. This " weekend effect " could be attributed to shortages in staffing, or reduced availability of services like sophisticated tests and procedures [52, 15]. "
    [Show description] [Hide description] DESCRIPTION: Modeling patient flow is crucial in understanding resource demand and prioritization. To date, there has been limited work in predicting ward-level discharges. Our study investigates forecasting next-day discharges from an open ward with no real-time clinical data. We propose nearest neighbor forecasting using the median of similar discharges in the past. We also identify a predictor set from patient demographics, ward data and discharge time series to derive a random forest model for forecasting daily discharge. Using data from a general ward of a large regional Australian hospital, we compared our models with the classical auto-regressive integrated moving average (ARIMA) for 12,141 patient visits over 1826 days. Forecasting quality was measured using Mean Forecast Error, Mean Absolute Error, symmetric Mean Absolute Percentage Error and Root Mean Square Error. When compared to the baseline model, next day discharge forecasts using nearest neighbor demonstrated 11.9% improvement, while random forests achieved 17.4% improvement in Mean Absolute Error, for all days in the year 2014.
    Full-text · Working Paper · Feb 2016 · Journal of Asthma
    • "One study showed that the weekend hospital discharge rate was more than 50% lower, compared with reference rates, and teams that were post-call had 20% lower discharge rates [59]. Interventions that have been shown to improve weekend care quality and reduce LOS include ensuring intensity of inpatient care and reorganizing internal processes, such as attending schedules [59,60], and these might also prove useful in reducing LOS for pediatric asthma hospitalizations. "
    [Show abstract] [Hide abstract] ABSTRACT: Objective: Asthma is a leading cause of pediatric hospitalizations, but little is known about factors associated with length of stay (LOS) for asthma hospitalizations. The aim of this study was to identify factors associated with LOS for pediatric asthma hospitalizations. Methods: The Pediatric Health Information System (PHIS) was used to cohort patients 2-17 years old with a primary asthma diagnosis discharged from 42 PHIS hospitals in 2011. Sociodemographic, temporal and health-status factors were examined. Bivariate and generalized-estimating-equation logistic regression analyses were performed to identify factors associated with LOS, after adjusting for severity of illness (SOI). Results: In total, 25,900 children were hospitalized, with a mean LOS of 1.9 days. In bivariate analysis, mean LOS was longer (p < 0.01) for patients with complex chronic conditions (CCC) (3.1 days versus 1.8 for non-CCC) and adolescents (2.3 versus 1.8 for 2-5 years old). In multivariable analysis, obstructive sleep apnea (OSA; OR 2.3; 95% CI: 1.8-2.9), older age (OR 1.3; 95% CI: 1.2-1.4), obesity (OR 1.3; 95% CI: 1.1-1.4), CCC (OR 1.3; 95% CI: 1.1-1.4), winter admissions (OR 1.2; 95% CI: 1.1-1.4), female gender (OR 1.1; 95% CI: 1.1-1.3), and weekend admissions (OR 1.1; 95% CI: 1.03-1.2) had higher odds of asthma LOS >2 days. Conclusions: OSA, older age, obesity, CCC, winter and weekend admissions, and female gender are associated with longer LOS for pediatric asthma hospitalizations, after adjustment for SOI. The study findings suggest that interventions focused on these at-risk groups may prove most useful in reducing LOS for pediatric asthma hospitalizations.
    Full-text · Article · Nov 2014
Show more