Professional Documents
Culture Documents
C Goldfrad, K Rowan
Summary Introduction
It is generally believed that pressure for beds on intensive
Background It is generally believed that pressure for beds on
care units (ICUs) has increased. Although no rigorous
intensive-care units (ICUs) has increased in the UK. This
research evidence exists, the increasing number of reports
study used discharge at night as a proxy measure to
and correspondence related to this subject does suggest
investigate pressure.
that the pressure on bed availability in ICUs may be
Methods Night was defined in two ways: “out of office hours’ greater now than before.
from 2200 to 0659 h and “the early hours of the morning” Increasing numbers of admissions with a concomitant
from 0000 to 0459 h. The rate of discharge at night was decrease in length of stay in the ICU have been
compared for 21 295 adult admissions to 62 ICUs covering reported.1–3 Occupancy rates have been described as very
the period 1995–98 with 10 806 admissions to 26 ICUs high.4 The shortage of available ICU beds in London,
covering the period 1988–90. With data solely from UK, and elsewhere2 has been detailed and a wide variation
1995–98, the consequences of discharge at night and in provision of facilities has been recorded.4–6 Cancelled
premature discharge were investigated.
operations due to the lack of available ICU beds4 and high
rates of refused admissions have been reported, both
Findings Overall, 2269 (21·0%) admissions did not survive regionally7 and nationally.5,6 The transfer of patients over
the ICU in 1988–90 compared with 4487 (21·1%) in long distances in search of an ICU bed has also been
1995–98. Of ICU survivors, 2·7% were discharged at night reported8–10 and the potential dangers of transferring
(2200–0659 h) in 1988–90 compared with 6·0% in critically ill patients has been highlighted.11 Premature
1995–98. In 1995–98, night discharges (2200–0659 h) had discharge of patients has been described.3,12,13
a higher crude (odds ratio 1·46, 95% CI 1·18–1·80) and A national bed register covering England was
case-mix adjusted (1·33, 1·06–1·65) ultimate hospital introduced in 1996 by the Department of Health to ease
mortality. Higher odds ratios were observed when the the problem of finding a suitable bed for a critically ill
definition of night was 0000–0459 h. Premature discharge patient.14 In addition, the Intensive Care Society, the
was commoner at night, 42·6% vs 5·0% and its importance professional organisation of intensive-care doctors, has
was apparent when incorporated into the logistic-regression called for both an increase in ICU facilities in areas of low
model (premature discharge 1·35, 1·10–1·65; night
provision and a formal transport system.15
High dependency units (HDUs) have been proposed
discharge 1·17, 0·92–1·49).
for low-risk, short-stay ICU admissions to ease the
Interpretation Night discharges from ICU are increasing in pressure.1,3,12,16,17 However, a shortage of such intermediate
the UK. This practice is of concern because patients facilities also seems to exist.
discharged at night fare significantly worse than those With the results of the Intensive Care Society’s UK
discharged during the day. Night discharges are more likely APACHE II study from July, 1988 to September, 1990,
to be “premature” in the view of the clinicians involved. The which indicated a high death rate on the ward after
implication of these results is that many hospitals have discharge from ICU (variation 6–16% across ICUs),18 and
insufficient intensive-care beds. In deciding whether or not to the belief that discharging patients from ICUs at night
invest more resources in intensive care we must, however, does not constitute good quality care and, where possible,
consider the cost-utility of this particular service compared is to be avoided, our study used discharge at night as a
with other ways that additional resources could be used. proxy measure to investigate pressure on ICUs. The aim
was to investigate the change, over time, in the rate of
Lancet 2000; 355: 1138–42 discharge at night from ICUs and to find whether there
were any adverse consequences following discharge at
night.
By the use of two high-quality clinical databases, the
UK APACHE II study database,18,19 and the Intensive
Care National Audit & Research Centre’s Case Mix
Programme Database (CMPD),20 the rate of discharge at
night for 1988–90 was compared with the rate for
December, 1995, to April, 1998. We used data solely
from the CMPD to compare the consequences of
discharge at night with discharge during the day.
Methods
Databases
Intensive Care National Audit & Research Centre (ICNARC), Data were extracted for the 10 806 admissions to 26 ICUs in the
Tavistock House, Tavistock Square, London WC1H 9HR, UK UK APACHE II study database, covering 1988–90, and for the
(C Goldfrad MSc, K Rowan PhD) 22 059 admissions to 62 ICUs in the CMPD, covering 1995–98.
Correspondence to: K Rowan All these data had been collected prospectively. Age exclusion
(e-mail: kathy@icnarc.demon.co.uk) criteria (age <16 years) used in the UK APACHE II study were
APACHE II CMPD
All ICUs Same ICUs All ICUs Same ICUs
ICUs 26 9 62 9
Adult (肁16 years) admissions 10 806 4131 21 295* 4064†
ICU survivors (% adult admissions) 8528 (78·9%) 3219 (77·9%) 16 789 (78·8%) 3255 (80·1%)
Discharges 2200–0659 h (% ICU survivors) 234 (2·7%) 94 (2·9%) 1009 (6·0%) 182 (5·6%)
Range of discharges at night across ICUs 0·6–9·6% 1·9–4·0% 0·7–17·2% 2·2–15·0%
Discharges 0000–0459 h (% ICU survivors) 91 (1·1%) 43 (1·3%) 433 (2·6%) 81 (2·5%)
Range of discharges at night across ICUs 0–4·0% 0–2·3% 0–9·4% 1·1–7·0%
Admissions before applying age exclusion criteria (admissions <16 years) used in APACHE II study=*22 059 and †4206.
Table 1: Proportion of discharges at night from ICUs participating in UK APACHE II study compared with CMPD
applied to the CMPD resulting in 21 295 adult admissions. patient’s readiness for discharge in the light of the needs of other
Deaths in ICU were excluded from the analyses. patients for the available ICU beds. No attempt was made to
impose standard explicit criteria for this variable.
Data Length of stay in the original ICU was calculated, in part days,
by the use of data for the variables “date/time of admission to
Data were extracted for “time of discharge from ICU”. We
ICU” and “date/time of discharge from ICU”. For discharges
consulted with ICU colleagues and, before the analysis was done,
directly transferred to another ICU, in either the same or another
defined night as “out of office hours” (2200–0659 h) and second
hospital, data for the variables “date of discharge from ICU” and
as “the early hours of the morning” (0000–0459 h). Each
“date of ultimate discharge from ICU” were used to calculate, in
admission surviving the ICU stay was categorised as either a
days, length of stay in subsequent ICUs after discharge from the
night or day discharge according to the two definitions.
original ICU. Similarly, for all discharges, we used data for the
Solely by the use of the CMPD, data were extracted on the
variable “date of ultimate discharge from hospital” to calculate
case mix (age, medical history, acute severity). A severe [past
the length of stay in hospital after discharge from the original
medical] history was defined by the presence of one or more of 16
ICU. In order to verify whether patients were being discharged at
severe chronic conditions. Acute severity was measured by the
night due to pressure on ICU beds, the time, in hours, to the next
APACHE II score,19 which encompasses weighting for age,
admission (turnover time for the bed) was calculated from the
medical history, acute severity (defined by derangement from the
“date/time of discharge from ICU” for one admission to the
normal range, in the first 24 h in the ICU, for 12 physiological
“date/time of admission” for the next admission to the same
variables), and for surgical status. Given that two admissions with
ICU.
the same APACHE II score but with different reasons for
admission can have very different risks of death,21 the APACHE
II probability of hospital death, defined as death before discharge Analyses
from hospital after intensive care, was used to describe case mix The overall proportion of discharges at night, and the proportion,
overall. Probabilities of hospital death were estimated by the UK by hour, were compared for the two periods for all ICUs and,
APACHE II model.22 additionally, for the ICUs common to both databases. For
Subsequent readmissions to ICU were identified by the 1995–98 solely (CMPD), discharges at night were compared
variables “postcode”, “date of birth”, and “sex”, and confirmed with discharges during the day for case mix, readmission rates
by the participating ICUs. For discharges directly transferred to and crude and case-mix-adjusted, ultimate hospital mortality.
another ICU, in either the same or another hospital, survival data Time to death, reason for discharge and time to next admission,
(alive/dead) at final discharge from ICU or hospital were destination after discharge, length of stay in the original ICU, in
extracted. For deaths at final discharge from hospital, the time to subsequent ICUs and in hospital were also compared.
hospital death, in days, was calculated. Case-mix adjustment was done by the APACHE II method.
Data were also extracted for the variables “reason for discharge Admissions staying less than 8 h in the ICU were excluded from
from ICU” and “destination following discharge from ICU”. The calculation of APACHE II scores. In addition, admissions with
former was based on a clinician’s subjective assessment of a burns, after cardiac surgery, directly transferred from another
20
1988–1990 (n=8 528)
18 1995–1998 (n=16 789)
16
Proportion of discharges (%)
14
12
10
0
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5 6
Hour of discharge (07:00–06:59 h)
Proportion of discharges by hour for ICUs in UK APACHE II study compared with CMPD
*Ultimate discharge status from hospital missing for 3·7% of night discharges and for *Ultimate hospital mortality for night versus day discharges.
2·5% of day discharges. Table 5: Outcome for night versus day discharges—impact of
Table 3: Outcome for night versus day discharges premature discharge