Total number of fall-related hospital admission days in the older Dutch population (1981-2008).
Hartholt KA, van der Velde N, Looman CWN, van Lieshout EMM, Panneman MJM, van Beeck EF, Patka P, van der Cammen TJM. Trends in Fall-Related Hospital Admissions in Older Persons in the Netherlands. Arch Intern Med. 2010;170(10):905–911. doi:10.1001/archinternmed.2010.106
Fall-related injuries, hospitalizations, and mortality among older persons represent a major public health problem. Owing to aging societies worldwide, a major impact on fall-related health care demand can be expected. We determined time trends in numbers and incidence of fall-related hospital admissions and in admission duration in older adults.
Secular trend analysis of fall-related hospital admissions in the older Dutch population from 1981 through 2008, using the National Hospital Discharge Registry. All fall-related hospital admissions in persons 65 years or older were extracted from this database. Outcome measures were the numbers, and the age-specific and age-adjusted incidence rates (per 10 000 persons) of fall-related hospital admissions in each year of the study.
From 1981 through 2008, fall-related hospital admissions increased by 137%. The annual age-adjusted incidence growth was 1.3% for men vs 0.7% for women (P < .001). The overall incidence rate increased from 87.7 to 141.2 per 10 000 persons (an increase of 61%). Age-specific incidence increased in all age groups, in both men and women, especially in the oldest old (>75 years). Although the incidence of fall-related hospital admissions increased, the total number of fall-related hospital days was reduced by 20% owing to a reduction in admission duration.
In the Netherlands, numbers of fall-related hospital admissions among older persons increased drastically from 1981 through 2008. The increasing fall-related health care demand has been compensated for by a reduced admission duration. These figures demonstrate the need for implementation of falls prevention programs to control for increases of fall-related health care consumption.
Approximately one-third of all people 65 years or older fall each year, and half of those fall repeatedly.1,2 Even a low-energy trauma, such as a fall incident, causes physical injuries. Around 10% of all fall incidents result in serious injuries and require a hospital admission. Fall-related hospital admissions in older patients are generally due to hip fractures (50%), fractures of the arm (13%), and head injuries (10%).3,4 Falls do not only lead to physical injuries but may also have a large impact on the quality of life of elderly individuals2,3,5- 10 and on health care costs.11- 13
The cause of a fall incident in the older age groups is often multifactorial. Risk factors associated with fall incidents include a higher age, female sex, the use of fall-risk–increasing medication, and certain comorbidities. Many older patients visit their general practitioner or the emergency department after a fall.7 Treatment is usually restricted to the direct fall-related injuries. Because the cause of the fall is often not investigated, the patient may remain at risk for future falls, which calls for the implementation of preventive services.5
In the Netherlands, the proportion of people 65 years or older is expected to increase to comprise up to 25% of the general population in 2040 (the proportion was 15% in 2008).14 These figures are comparable with worldwide trends,15 and it can be expected that this will have an enormous impact on the fall-related health care consumption. Therefore, we aimed to quantify time trends of fall-related hospitalization and in-hospital length of stay (LOS) in older patients over the past decades.
For this study, all accidental fall-related hospital admissions during the period of 1981 through 2008 were collected. Throughout the study period, an accidental fall was defined using the International Classification of Diseases, Ninth Revision (ICD-9), codes E880 to E888 (Table 1), for external causes of injury. Older persons were defined as persons 65 years or older. The age-specific fall-related clinical incidence rates were calculated in 5-year age groups. Data were retrieved from the Statistics Netherlands,16 which combines information of the National Medical Registration (Landelijke Medische Registratie [hereinafter, LMR]) with the National Hospital Discharge Registry. The LMR collects hospital data of all hospitals in the Netherlands. Data regarding hospital admissions, admission diagnosis, LOS in days, sex, age, causes of external injuries, main diagnosis (such as fractures), and mortality are stored in this database. Data on hospital admissions, mortality, and population numbers were verified with the Dutch Birth Registry. The LMR uses a uniform classification and coding system for all hospitals and has an almost complete national coverage (missing values are <5%).17 These figures were extrapolated to full national coverage for each year. An extrapolation factor was estimated by comparing the adherence population of the participating hospitals with the total Dutch population in each year. Specific data on admission duration, fall mechanisms, and main injuries were not fully available in the first years of the study owing to data aggregation in the past. The medical ethical review board of the Erasmus University Medical Center, Rotterdam, approved the study.
Numbers of fall-related hospital admissions, main injuries, and LOS were specified for age and sex in 1981, 1986, 1991, 1996, 2001, 2006, and 2008. The age-specific incidence was calculated using the number of the fall-related hospital admissions in that specific age group, divided by the number of total population within that specific age group for male and female patients, and was expressed per 10 000 persons in that age-group. Overall growth in the number of hospital admissions was calculated for 2008 in percentages relative to the year 1981. The population numbers as of January 1 were used in each year of the study.
To model the trend in hospital admissions, a linear regression model with Poisson error and log link was built with log (population size as of January 1 of each year of the study) as the offset factor. A linear spline model, with age, year, sex, and population size, was built to assess whether the annual growth changed over the study period for both sexes. The parameter for calendar year, corrected for sex and age group, was transformed into percentage of annual change (PAC). Our spline function accommodated 2 piecewise linear fits, connected with one another at the knot.18 The knot was placed in the middle of the study period (January 1, 1995). The analysis including splines yielded estimates of annual changes in admission rates within each 14-year period (1981-1994 and 1995-2008). Comparison of these 2 periods enabled us to detect and quantify changes in the secular trend in admission rates, such as a stagnation or an increase of admission rates. A likelihood ratio test was performed to assess the significance of the spline over a single trend for the study period. Interactions of the spline vs sex and age were added and tested to investigate differences in increase for sex and age. All statistical analyses were performed using SPSS software (version 16.1.1; SPSS Inc, Chicago, Illinois). P <.05 was considered statistically significant.
In the Netherlands, the number of fall-related hospital admissions in older persons grew from 14 398 in 1981 to 34 091 in 2008 (a 137% increase). This reflects an annual growth of 778 fall-related hospital admissions during the study period. In particular, the number of persons 80 years or older showed a strong increase (from 6535 in 1981 to 20 253 in 2008, an increase of 210%). The mean age of the patients increased from 79.3 years in 1986 to 81.0 years in 2008, and three-quarters of the patients were female (Table 2). Fall incidents resulted mainly in hip, wrist, and upper arm fractures as well as skull and brain injuries and superficial injuries (Table 2). Numbers of the main injuries all increased, but the proportion of hip fractures slightly decreased (from 48.0% in 1986 to 41.8% in 2008). The incidence rate of hospital admissions due to a fall with a hip fracture in older Dutch adults increased from 51.2 per 10 000 older adults in 1986 to 63.6 in 1996, while in the period 1997 through 2008 the incidence rate of hospital admissions for hip fractures caused by a fall decreased to 59.0 per 10 000 older adults. The proportion of wrist and upper arm fractures together with superficial and head injuries increased from 13.6% (2563) in 1986 to 23.8% (8095) in 2008. Mechanisms of fall incidents hardly changed over time. Most fall incidents happened on the stairs, or near a bed or chair; were due to slipping, tripping, or stumbling; and remained stable at 96% throughout the study period.
The annual growth of fall-related hospital admissions for men and women differed (P < .001) and was 1.3% for men vs 0.7% for women, respectively, throughout the study period, corrected for age and population at risk. The annual growth, however, was not constant during the whole period. The linear spline model revealed an annual growth of 1.62% (95% confidence interval [CI], 1.47%-1.78%) in men and 1.10% (95% CI, 1.01%-1.18%) in women in the period 1981 through 1994. The growth decreased to 1.07% (95% CI, 0.94%-1.20%) in men and 0.37% (95% CI, 0.30%-0.44%) in women in the period 1995 to 2008. The age-specific annual change is shown in Table 3 and was highest in men 80 years or older.
For all age-specific groups, the incidence of admissions in the female population exceeded the incidence in men. A 5- to 10-year shift was noticed in the age-specific incidence between men and women. For example, in 2008 the admission rate for men aged 85 to 90 years was 280.4 per 10 000 compared with 277.0 per 10 000 women aged 80 to 85 years. The incidence rate was consistently higher for women than for men (Table 4).
Sex- and age-specific incidence rates of fall-related hospital admissions increased in all age-groups, both for men and women (Table 4). The overall incidence rate increased from 87.7 in 1981 per 10 000 older adults to 141.2 per 10.000 in 2008 (an increase of 61%). Regarding the age-specific incidence in women, the largest absolute and relative increase was seen in patients 95 years or older (57.0%; 95% CI, 33.8%-84.3%). For men, the largest absolute and relative increase was also seen in patients 95 years or older (157.7%; 95% CI, 87.2%-254.6%).
The LOS for fall-related hospital admissions was age related and peaked in patients 85 to 90 years old. However, the LOS of a fall-related hospital admission in older persons decreased over the past 25 years for all age-specific groups. Overall, the mean (SD) LOS was reduced from 26.3 (33.3) days in 1991 to 11.1 (12.4) days in 2008. The mean LOS in days is shown in Table 5.
The change in LOS affected the total number of hospital bed–days. The total number of fall-related hospital bed–days decreased from 487 769 days in 1981 to 388 650 days in 2008. The overall number of hospital bed–days remained stable for male patients 65 years or older (approximately 103 000 days) between 1981 and 2008. For women between 65 and 80 years old, the number of hospital bed–days decreased gradually from 192 360 days in 1981 to 87 475 in 2008 (a 54% reduction). In women 80 years or older, however, the number of hospital bed–days increased from 191 077 in 1981 to 297 825 in 1993 and decreased to 202 343 in 2008 (Figure).
To gain insight into the absolute numbers, incidences, and trends of fall-related hospital admissions in older patients, registration data of all persons 65 years or older in the Netherlands were studied from 1981 through 2008. Both the absolute numbers and the incidences of fall-related hospital admissions in older people had strongly increased over time. This increase was more pronounced in male patients than in female patients, although the total incidence rate remained consistently higher for female patients in all age groups. The main injuries at admission were hip, wrist, and upper arm fractures, and, together with superficial and skull or brain injuries, these accounted for two-thirds of all injuries. Because the LOS was reduced over time, the increased numbers of fall-related hospital admissions did not lead to an increased overall number of admission days until now.
A major strength of this study is the availability of population-based in-hospital data for an extensive period of 28 years (recording started in 1981). Since that year, absolute numbers of fall-related hospital admissions and hospital bed–days in all hospitals in the Netherlands have been recorded in a highly accurate electronic database with an almost complete national coverage. A possible limitation of this study is that the data describe a national situation for 1 country, which may not directly translate to other countries because demographics and health care system characteristics may differ. Nevertheless, because falls data in older populations in western societies are comparable with the data of the Netherlands,19 we have no reason to assume that the prevalence of fall-related hospital admissions is different.20 However, additional studies are required to confirm if these trends in fall-related injuries and hospital admissions are comparable with those in other populations. Furthermore, patients who were not admitted to the hospital were not registered in the National Hospital Discharge Registry database and, consequently, were not included in this study. Therefore, this study mainly reflects trends regarding the incidence rates of serious fall-related injury and excludes isolated minor fall-related injuries. The actual societal impact of all fall-related injuries, both major and minor, is most likely to exceed the burden as described in this study.
A limitation of the use of this linked administrative database is that it does not contain data regarding underlying diagnosis, comorbidities, treatments, injury severity, lifestyle, or medication use of the patients. This hampers the interpretation of causal mechanisms behind the observed trends.
Readmissions in 1 calendar year were not excluded and could potentially lead to some “double registrations.” However, readmissions most likely did not influence our results because readmissions constitute only 2.6% (at the maximum) in the Netherlands, as was found in a study by Polinder et al.21
Trends in hospital admission rates and LOS as observed in this study are important for 2 reasons: first, the population at risk is increasing worldwide,15 and second, the age-specific incidence of fall-related hospital admissions is increasing. Multiple studies, focusing on fall-related injuries, such as hip fractures, proximal humeral fractures, and severe head injuries in elderly individuals, have shown a comparable trend over time.20,22- 26
Throughout the study period, no major policy changes that might have affected the increase in admission rates were introduced in the Netherlands. The Dutch health care system was, and continues to be, characterized by full health insurance coverage and full accessibility for the whole population during the study period. As in other countries, clinical practice has changed during the study period (eg, introduction of geriatric medicine, improved anesthetical care and surgical techniques for older adults), but this probably only marginally affected the admission policies because of the low general admissions threshold in the Netherlands. However, rapidly increasing rates of admissions for wrist and upper arm fractures may be partly explained by improved surgical procedures and techniques in the oldest old and an associated drop in admission threshold.
Another potential cause for the observed increase in fall-related hospital admission rates might be “the aging society.” Because life expectancy is increasing in the Netherlands27 and older persons are living longer with multiple medical problems, the risk of falls and fall-related injuries can be expected to increase. Nevertheless, population health analyses in the Netherlands have shown that these patients are reporting fewer problems of mobility,28,29 which may be explained by improved medical care and by the use of walking aids and other equipment (ie, electric mobility scooters). The consequence of sustained walking abilities among older persons with multiple morbidities is that these patients remain at risk for fall incidents. This trend of improved mobility in older adults is seen in other countries as well.30,31 Another cause for the observed increased incidence of fall-related injuries might be an increasingly active lifestyle (ie, cycling, jogging, walking). All these factors do have an influence on fall risk and outcomes and may (partly) explain the observed rise in incidence rate of fall-related hospital admissions. Besides a more active lifestyle, other fall mechanisms, a lower surgical intervention threshold, and improved anesthetical care for older adults may have also contributed slightly to the changed distribution of main injuries.
The deceleration of growth in admission rates in the most recent period might partly be explained by a reduction in hip fracture rates, as observed in our study and other studies worldwide.22,32- 34 The possibility of a birth cohort effect resulting in a healthier aging population with improved functional abilities and a reduced level of injurious falls has been suggested.22 Our findings raise the question of which other factors could be associated with a more favorable trend in fall-related admission rates in recent years. We have no clear answers to this specific question.
The total number of hospital bed–days due to falls has decreased from 1981 through 2008. This can be explained partly by the larger decline in the trend of the number of hospital admissions in women compared with men, which has been shown previously in fall-related injury studies outside the Netherlands.25,32- 34 An additional explanation for the phenomenon may be that during the past decade, multiple medications with proven positive effects on bone quality have become available, and clinical practice guidelines, focusing primarily on older women, have been introduced.35,36 Additional research is needed to determine the causes behind the observed differences between trends in men and women. In addition, lifestyle interventions and falls prevention programs have been implemented following the recent introduction of falls prevention guidelines.37 This might be one of the reasons for the decline in age-specific incidence rates in fall-related hospital admissions in older women in the most recent period. However, the total burden is still increasing owing to an increased population of older women.
Despite the fact that the society is aging and comorbidities are increasing in number, the total number of hospital bed–days decreased because the mean LOS decreased in all age groups. This reduction may, at least partly, be explained by improvement in surgical techniques (eg, minimally invasive procedures), the development of clinical guidelines (eg, hip fracture guidelines), and more efficient treatment strategies.36- 39 With respect to the LOS, in the Netherlands, specific and readily available rehabilitation facilities for older patients with hip fractures, supported by multidisciplinary teamwork, have been instituted in nursing homes in recent years. However, the oldest old seem to need more time in hospital, most probably owing to more comorbidities and reduced functional reserves.
In summary, our data highlight the importance of monitoring hospitalization trends over time across age groups and sex in order to visualize possible changes in health care needs and usage. Insight into fall-related hospital admission numbers can contribute to optimization of planning, resource allocation, and staff distribution in the future. Improved implementation of falls prevention programs in older populations and specialized in-hospital tracks for older fallers seem required to control for possible further increases in fall-related morbidity and health care consumption.
Correspondence: Tischa J. M. van der Cammen, MD, PhD, Section of Geriatric Medicine, Department of Internal Medicine, Erasmus University Medical Center, Room D442, PO Box 2040, 3000 CA Rotterdam, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: November 21, 2009.
Author Contributions: Dr Hartholt and Mr Panneman had full access to the data. Study concept and design: Hartholt, van der Velde, and van der Cammen. Acquisition of data: Hartholt and Panneman. Analysis and interpretation of data: Hartholt, van der Velde, Looman, van Lieshout, Panneman, van Beeck, Patka, and van der Cammen. Drafting of the manuscript: Hartholt, van der Velde, and Looman. Critical revision of the manuscript for important intellectual content: van der Velde, van Lieshout, Panneman, van Beeck, Patka, and van der Cammen. Statistical analysis: Hartholt, van der Velde, Looman, and Panneman. Administrative, technical, and material support: van Lieshout. Study supervision: van der Velde, van Beeck, Patka, and van der Cammen.
Financial Disclosure: None reported.
Funding/Support: Dr Hartholt is a research fellow at the Erasmus University Medical Center, appointed on a research grant from the Netherlands Organization for Health Research and Development (ZonMw), project No. 170.885.607.