  |
|
|
||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||
| Journal of the New Zealand Medical Association,
24-September-2004, Vol 117 No 1202
|
||||||||||||||||||||||||||||||||||||||
|
Circumstances and consequences of falls in residential care:
the New Zealand story
Meg Butler, Ngaire Kerse, Maree Todd
Older people living in residential care facilities are
frail, have a high prevalence of co-morbidities and age-related impairments, and
are prone to falls1,2—this combination
makes even a simple fall particularly
dangerous.3
A high proportion of falls result in injuries, with as many
as 25% of falls resulting in lacerations, fractures, and the need for hospital
care.3 Non-injurious falls also have a negative
impact on residents, with as many as 75% of fallers experiencing loss of
confidence and/or a fear of further falls (with a consequent loss of physical
activity and function).4
The problem of falls in residential care facilities warrants
attention because falls are common and distressing for both residents and their
carers. Recent research indicates that factors related to the
environment,5 and to nursing or services
provision, could place older people at risk of falls. Among less mobile
residents, accidental falls have been found to occur around equipment use or
during the ‘transferring process’.6
Descriptions of the time and place of fall occurrences, and the consequences of
falls among this frail population are rare in the literature.
This study reports falls sustained in a cohort of
residential care facilities in the Auckland region. Risk factors related to
resident characteristics are reported
elsewhere.7 This paper describes the fall
events, circumstances, and timing under which they took place as well as the
consequences of the falls in terms of injury and health services utilisation.
Association between characteristics and timing of falls and
injury were investigated. An understanding of the nature of falls is an
important step in preventing falls among residents in long term residential
care.
MethodsSelection
of rest homes—In the greater Auckland (New Zealand) region, a
random selection was made of rest homes or low-level dependency facilities and
private hospitals or high level dependency facilities (as defined by New Zealand
Ministry of Health criteria) using random number tables. Rest homes residents
who require assistance with most instrumental activities of daily living and at
least two basic activities of daily living, can usually ambulate to some degree
and feed themselves. In private hospitals or high-level dependency facilities,
residents require assistance with most activities of daily living and usually
need daily nursing care. Written informed consent was obtained from managers and
residents in the selected homes. The Auckland Ethics Committee approved the
study.
All the participating homes agreed to undertake falls
monitoring within their homes for the study duration (18 months) starting in
December 1999. Half of the study homes (n=7) participated in a fall prevention
intervention that began on 28 April 2000. As the intervention may have had an
impact on fall incidence rates in intervention homes,8
falls occurring during the intervention period in intervention homes are
excluded. This paper reports all falls that occurred during the 5-month period
prior to the intervention as well as falls occurring in control group homes for
the 12 months of the intervention period (18 months in total).
Study
variables—After enrolment, information was collected about each
resident’s medical diagnoses, medications, and previous falls, from the
resident’s medical record. Information listed on the summary sheet as
being a diagnosis, problem, or disability was considered a
‘diagnosis’ for the purposes of this study. Functional status was
recorded by the registered nurse or lead nursing assistant. This was measured by
a validated scale developed in Sheffield,9
found to be reliable, and then used in the New Zealand Long Term Care
Survey.10,11
The composite scale comprises three level scales. One
of the validated level scales, self care (mobility, dressing, feeding, bathing,
and toileting) was used in this study. Composite scales for mobility (transfers,
mobility within the home, and ability on stairs) and behaviour (needing night
care, social behaviour, memory, wandering, and awareness) were constructed. Each
scale is the summed score of its items, rating the resident as independent, able
to do self care with a little help, or needing considerable help. Information
was also sought concerning staffing levels using a structured interview with the
home’s management staff, and this was expressed as the ratio of total
staff (nursing and nursing assistants)/total residents.
Falls—A
one-page standardised falls surveillance form was developed for staff use, and
completed for each fall that occurred. All fall forms were faxed or posted to
the research centre and read by an independent researcher using a standard
definition of a fall: ‘unintentionally
coming to rest on a lower
level’.12 Falls not meeting this
definition were excluded. Reporting of falls was audited by a visit to the home
and by a hand search of the medical record and incident reports to identify
unrecorded falls. Falls records were audited for duplicate reports of the same
fall event.
Information sought
included:
Activity at
the time of the fall was recorded and included:
Space for a written description of the fall
was also provided. Fall-related injuries were documented, as was the type of
medical attention required.
Injuries were
categorised into minor, moderate, and major injuries:
Injury categories such as mild, moderate, or
severe were created for each fall. To make these categories mutually exclusive,
the most severe injury sustained was used to decide the most appropriate
category. For example, major injuries were not classified as moderate or mild,
even if a skin tear was sustained during a fall resulting in fracture.
Analysis—As
there was a potential intervention effect, all falls occurring during the
intervention period in intervention homes were
excluded8. Descriptive statistics were used to
describe circumstances and frequencies of falls and injuries by level of care
using SPSS.13
To describe the timing of falls, a rate of falls was
calculated. This was the number of falls occurring per hour (averaged over
2-hour periods) per person, using 648 persons as the number enrolled over the
study period, for the overall rate. To calculate fall rates for rest homes, the
number of falls was divided by the number enrolled from rest homes. This process
was repeated for high-level dependency residents in private hospitals. To adjust
for the variable amount of time residents were observed (and the falls being
accrued over 18 months), rates were then divided by the average number of days
residents were enrolled (overall and for each of the dependency levels),
multiplied by 3650 and expressed as falls/hour/10 resident years occurring
during each 2-hour period. Graphs are presented in the Results showing
distribution of fall rates by level of injury severity over 24 hours.
For this analysis, falls occurring while standing,
walking, stretching were combined into one variable ‘on feet’. Falls
occurring at a lower lever such as ‘a
fall on/off bed, chair, or toilet’ and
‘transferring on/off bed, chair, or
toilet’ were combined to create a variable
‘fall from a lower level’.
All falls that were described as a slip, trip, trip over object, or having a
hazard involved were coded as involving ‘a hazard’.
To establish relationships between fall circumstances
and level of injury, logistic regression models were constructed with an
injurious fall as the dependent variable. Independent variables were: location,
activity, direction of fall, presence of restraints, presence of hazards, and
time of day (categorised into 2-hourly time units).
As there are many patient characteristics that predict
falls, the following variables were controlled for in the models: level of
dependency (type of home), gender, presence of a diagnosis of dementia, number
of medications, number of diagnoses, and previous falls. As falls were not
independent (for example, several falls occurred in many residents), the model
was adjusted for clustering by resident using STATA 7.0 software. An exposure
term was included in the models—the variable indicating follow-up time in
days for each individual in the trial.
ResultsStudy population—the residential facilitiesOf a
possible 206 residential care homes in the Auckland region, 27 rest homes (low
level dependency homes) and 13 private hospitals (high level dependency homes)
were excluded as being too far from the study centre. Of the remaining 121 rest
homes and 45 private hospitals, 14 were randomly selected. One declined to
participate and another was randomly selected yielding a 93% response rate.
Overall these facilities contained 25 units consisting of 9 rest home units, 9
private hospital units, 6 secure units for those with psychogeriatric disorders,
and 1 unit for disabled young people.
Study population—the residentsAt baseline, 18 residents (out of a
possible 648 residents) in the homes refused to participate, thus yielding a
response rate of 97%. Before surveillance began, 10 residents died and 3 were
transferred from these homes. Data on the functional status and health of these
13 residents were not available.
Fall surveillance began with 617 consented residents in
December 1999 and continued until the end of April 2001. Residents admitted into
the homes until November 2000 (n=63) were enrolled in the study. During the
study period (18 months), 128 residents died (20%), and 83 (12%) residents were
transferred to another residential facility or discharged home. Fall
surveillance data was collected on a total of 680 residents. Table 1 shows
demographic, health and functional status information on the residents.
Table 1. Patient characteristics in a sample of
residential care residents
SD=standard deviation; m=mean;
n=demographic and health data available on 652 residents only; *self care,
mobility, and behavioural score were calculated from subscales of the dependency
questionnaire (higher score means higher level of function); †diagnoses
and medications established from the summary sheet in the medical record and
medication chart.
The fallsDuring the 18 months of
surveillance, 2,021 falls forms were faxed to the study office. Of these, 32
were excluded as the fall was described as being due to an epileptic
seizure—this yielded a total of 1,989 fall forms. Falls occurring in
intervention homes during the intervention period were excluded, leaving 954
falls included in this analysis.
Using residents as a denominator, 271 of 680 residents (40%
of all residents) sustained one or more falls and 218 (32% of all residents, 80%
of fallers) had an injurious fall; 183 (67% of fallers, 27% of all residents)
sustained a minor injury; 66 (24% of fallers, 9.7% of all residents) sustained a
moderate injury; and 39 (14% of fallers, 5.7% of all residents) sustained a
major injury.
Using falls as the denominator (rather then residents), 605
of the 954 falls (63%) resulted in some form of injury. Self-reported falls
totalled 271 (28%). Most fall-related injuries were minor (46% of falls), with
12% of falls being of moderate severity and 44 falls resulting in major injury
(5% of falls). About one-third of falls were routinely assessed by the GP and 5%
resulted in an urgent visit. One-quarter of falls resulted in skin tears and 20%
bruises. Of the major injuries, hip fractures were the most common fracture
(n=12) with 36 falls resulted in a radiological examination (4%). The details of
injuries sustained from falls and treatments provided is described in Table 2
.
Table 2. Falls and injuries in a cohort of 680
residential care residents over an 18-month period
Falls were equally distributed between the days of the week
with an average of 136 falls on any particular day of the week. Fall rates
tended to be higher during the day when residents were active—with the
highest rates observed between 4pm and 6pm (2.2 falls/hour/10 resident years).
Falls were mainly contributed by residents in rest homes (Figure 1).
Using logistic regression (adjusted for time in the study,
presence of dementia, self-care score, mobility score, behavioural score, age,
gender, type of home, and clustering [by resident]), no 2-hour period, compared
with the time period 12midnight–2am, was more likely to result in any
injury.
Staffing levels in all homes were lowest at night, at
intermediate levels in the afternoons, and were highest in the mornings. There
was a tendency for falls rates to be higher in the afternoon when staffing
ratios were intermediate.
Figure 1. Rate of falls and fall-related injury over 24
hours in a sample of residential care homes in Auckland, New Zealand by (a) type
of residence and (b) severity of injury
(Using
logistic regression, adjusted for time in the study, presence of dementia,
self-care score, mobility score, behavioural score, age, gender, type of home,
and clustering (by resident); no single 2-hour time period was more likely to
result in injury)
(a) Fall rate by type of
home
Staffing
ratio
 hi
level=private hospitals;
lo
level=rest homes;
staff
ratio=staff per shift/total
residents. (b)
Fall rate by severity of injury
Figure 2. Injury severity over time in (a) rest homes
and (b) private hospitals
(Staffing
ratio plotted for comparison)
(a) Rest homes
 (b) Private
hospitals
 Injury severity varied throughout the day and differed
between the type of home. For example, in rest homes, major injuries from falls
were more common in the mornings whereas minor injuries occurred at a higher
rate in the late afternoon. Staffing ratios may have been lower when falls with
minor injury rates were highest in the afternoons. In the private hospitals and
secure units (Figure 2b), falls resulting in moderate and major injuries
occurred at similar rates throughout the 24-hour period.
Minor injuries occurred at a lower rate at night compared
with during the day. Using logistic regression, adjusted for health and
demographic factors, previous falls, time of day and clustering (by resident)
falls occurring in private hospitals OR 2.82 (CI 1.19–6.66) and
secure units OR 1.59 (CI 1.02–2.48) were more likely to result in any
injury compared with rest homes.
Table 3. Location and activities at the time of falls
and fall related injuries in a cohort of residents living in long term
residential care facilities
*Not
seen means the fall was not witnessed, and
includes self report falls.
†Percentages do not add
to 100 due to missing values.
Logistic regression results
(1-4), adjusted for time in the study, presence of dementia, self-care score,
mobility score, behavioural score, age, gender, type of home, and clustering (by
resident).
1OR
0.54 (CI 0.33–0.88) less likely to sustain injury in the dining
room
2OR
4.67 (CI 1.02–21.36) more likely to sustain injury in the bathroom
3OR
4.92 (CI 1.50–16.11) more likely to sustain injury
outside
4OR
1.70 (CI 1.17–2.46) more likely to sustain injury falling
sideways.
The location of falls, and activities noted at the time
falls are described in Table 3. Nearly all
falls occurred indoors (96%)—the majority of which occurred in the
residents room (63%). Adjusting for resident health and functional factors and
the type of home, falls sustained in the bathroom were more likely to result in
injury OR 4.67 (CI 1.02–21.36). In contrast, falls occurring in the dining
room (n=69) were less likely to result in injuries compared with other locations
OR .54 (CI 0.33–0.88).
Only 35 falls occurred outdoors, and these were more likely
to result in injury OR 4.92 (CI 1.50–6.11). A sideways fall was more
likely to result in injury, OR 1.70 (CI 1.17–2.46) whereas other
activities and fall circumstances were not associated with injury. Using the
same model with major injury as the
dependent variable the following circumstances were associated with major
injury: fall while on feet OR 2.36 (CI 1.22–4.52) and fall to the side OR
1.69 (CI 1.16–2.47). Falls occurring in between 4pm and 6pm were less
likely to be associated with major injury OR .16 (CI 0.03–0.96).
Half of all falls, n=473 (50%), occurred while the residents
were on their feet (either standing or walking). A sizeable minority (142, 15%)
of falls were recorded as having a hazard involved. These included cords, steps,
thresholds, bedroom cluster, mats (n=135), and dim light (n=16). These hazards
resulted in falls associated with tripping over an object (n=44) and slipping on
wet floors (n=37).
During the study period, 35 falls occurred with hip
protectors being worn at the time. No hip fractures occurred as a result of
these 35 falls whereas 12 hip fractures resulted from 917 falls occurring
without hip protectors in place. Restraints were recorded as in use in 17
falls.
DiscussionFalls in residential care are a
major public health concern. The purpose of this study was to describe the
circumstances and injuries related to falls sustained by residents living in
long-term residential homes in Auckland, New Zealand. Almost half (40%) of the
residents sustained a fall during 18 months of surveillance; consistent with
other studies.2
The majority of these falls resulted in injuries (63%) with
5% of all falls resulting in injuries of major severity, a higher rate for any
injury but lower rate for serious falls than other
studies.6,14 Previously, the timing of falls
has not been accurately described. When residents were active during the day,
they were more likely to sustain injury, and the more active residents fell more
but were less likely to injury themselves.
This study identified all falls occurring in the homes,
including self-reported falls. We chose to accept self-reported falls and
recorded any form of injury. This is justified, as even a minor injury can
result in not only decreased mobility and increased
morbidity4 but also increased need for care.
Our definition of injury categories has resulted in a higher proportion of falls
being categorised as ‘injurious’ (63%) than in other studies which
report between 40% and 60% of falls resulting in
injury.12,15
This paper also uses the fall as the denominator for
analysis, rather then the resident, meaning that direct comparison to other
studies reporting the proportion of residents injured is less appropriate.
A limitation of this study is the potential for inaccuracy
of information regarding self-reported falls that were unwitnessed. These
occurred mostly in rest homes, where residents have (in the main) sufficient
cognition to report events.
There was no way to validate information about these
unwitnessed falls. This was unavoidable as the alternative of running videotapes
at all times in all locations within the homes was beyond the scope of this
study.
Information about direction and location of the fall was
evident to staff when the residents were found and data were recorded. Staff
ratios were reported to researchers by the principal nurse manager and averaged
over the three usual working-shifts, therefore losing some precision in
description.
In addition, injury severity categories were arbitrary, but
created after consultation with experts. As the numbers of residents in secure
units was small, they were combined with those in private hospitals for the
regression analyses. Almost all secure units were placed within private
hospitals, thus under similar environmental and staffing conditions, and the
fall rates were the most similar between these groups.
It is possible that relationships between circumstances of
falls and injury would be better examined in larger samples of secure unit
residents separately; however, this was beyond the scope of this study. This
study highlights the need for the creation and standardisation of injury
severity categories, which would allow for more accurate comparisons between
studies.
This paper provides important descriptive information about
injuries sustained and treatments provided to a large group of residents which
will be useful for health care planners, medical staff and the residential care
industry. Most injuries were minor (with bruises and skin tears being most
common) as confirmed in other studies.16 The
doctor involved in the routine care of the resident was often involved in the
management of fall-related injuries.
This study confirms the finding of peak falls times reported
by less rigorous research.16 The distribution
of injurious falls in a 24-hour period is intriguing. The emphasis on higher
fall rates in the afternoon in rest homes is important, as this is the time of
shift change. It is reasonable to assume that consideration of different ways to
deliver care at these times may impact on fall rates, as (notably) the afternoon
staffing ratios are lower than the mornings.
In addition, there is a trend for major injuries to occur in
the mornings across all homes, a time when maximal staff is available. Mornings
are a time of high activity in all homes with toileting, showering, dressing,
and breakfasting. Staff often take a morning-tea break, which may make them less
available to residents and potentially compounds the risk of major injury at
that time. Other studies have shown that non-ambulatory residents are more
likely to sustain injury associated with use of equipment and
transferring6 and this may also be a part of
the explanation of this finding. Further study is needed to establish the impact
of staffing level changes on fall and injury in residential care.
Rest home residents in this study had more falls but were
less likely to sustain injury than residents in private hospitals. Furthermore,
rest home residents are less frail than those in private hospitals and this will
protect them from injury. It must be accepted that active residents are at risk
of falls and activity is important for quality of life. The trade off between
level of activity and risk of falls deserves more attention in future
studies.
The prevalence of falls among residents (40%) in this study
is similar to other studies which report 40%–52% of residents sustained a
fall.17 However, our prevalence of injurious
falls is higher (23% of residents), perhaps owing to our acceptance of
self-reported falls. Thus, a lower rate may have been recorded if a different
study design been employed.
While falls in the dining room were less likely to result in
injury, falls in the bathroom were more likely to result in injury. This
provides additional information useful for planners, managers, and caregivers.
It is reassuring that, in a protected regulated environment,
only 15% of falls involved a slipping or tripping hazard. While routine
surveillance for such hazards may impact fall rates, it can be postulated that
the majority of falls were from intrinsic factors that will require strategies
to reduce individual’s fall related risk factors.
Protection from fall injuries is very important. The finding
that hip fractures were associated with a sideways fall, and that no hip
fractures occurred from falls while residents were wearing hip protectors,
further supports the efficacy of hip protectors among high-risk individuals in
rest homes and private hospitals.18,19 This is
important as an upward trend in the incidence of hip fractures occurring in New
Zealand has recently been reported.20
The strength of this study is that, in comparison to other
studies,15,16 ascertainment of all falls was
achieved and a large sample of residents were prospectively followed. Indeed,
the information gathered and analyses techniques used (in investigation of the
timing of falls) in this study are unique.
ConclusionsTo date, few interventions have
been shown to reduce falls among older people living in long-term residential
care settings. An understanding of the timing and location of falls within homes
provides useful information for healthcare providers concerning staffing levels
at differing times of the day—as peaks times of falls and injury may be
related to periods of high activity within residential care homes.
It is important for clinical staff to consider intrinsic
risk factors for falls, as most falls did not involve environmental hazards.
Protection from injury is emphasised by this study—as a sideways fall was
related to serious injury and no hip fractures occurred in residents wearing hip
protectors.
A variety of strategies involving interventions for
individual residents and strategies (about staff, time, and locations in
facilities) will be needed to reduce injury. These strategies will, in turn,
need to be tested for efficacy.
Acknowledgements: We
thank the staff in the participating Auckland residential care homes (for their
hard work and enthusiasm); the Health Research Council (NZ), Auckland Medical
Research Foundation, and Royal New Zealand College of General Practitioners
Auckland Faculty Trust (for contributing project grants); and The Commonwealth
Fund (for supporting this study through a Harkness Fellowship for Ngaire Kerse).
Clare Robertson provided comments on earlier drafts of the
manuscript and Elizabeth Robinson provided statistical advice.
References:
|
|
||||||||||||||||||||||||||||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|