End of the Spectacular Decrease in Fall-Related Mortality Rate: Men Are Catching Up

Abstract

Objectives. We determined time trends in numbers and rates of fall-related mortality in an aging population, for men and women.

Methods. We performed secular trend analysis of fall-related deaths in the older Dutch population (persons aged 65 years or older) from 1969 to 2008, using the national Official-Cause-of-Death-Statistics.

Results. Between 1969 and 2008, the age-adjusted fall-related mortality rate decreased from 202.1 to 66.7 per 100 000 older persons (decrease of 67%). However, the annual percentage change (change per year) in mortality rates was not constant, and could be divided into 3 phases: (1) a rapid decrease until the mid-1980s (men −4.1%; 95% confidence interval [CI] = −4.9, −3.2; women −6.5%; 95% CI, −7.1, −5.9), (2) flattening of the decrease until the mid-1990s (men −1.4%; 95% CI = −2.4, −0.4; women −2.0%; 95% CI = −3.4, −0.6), and (3) stable mortality rates for women (0.0%; 95% CI = −1.2, 1.3) and rising rates for men (1.9%; 95% CI = 0.6, 3.2) over the last decade.

Conclusions. The spectacular decrease in fall-related mortality ended in the mid-1990s and is currently increasing in older men at similar rates to those seen in women. Because of the aging society, absolute numbers in fall-related deaths are increasing rapidly.

Unintended falls present a major public health problem worldwide.^1–3 The World Health Organization (WHO) estimated that approximately 392 000 people worldwide died because of an unintended fall in 2004.⁴ According to the Centers for Disease Control and Prevention in the United States, unintended falls are the leading cause of fatal injuries among older adults.⁵

Falls in older adults mainly occur in or around the home, and their cause is often multifactorial.⁶ Risk factors associated with fall incidents include older age, female gender, use of medications that increase fall risk, and comorbidities.⁷ Approximately one third of all community-dwelling persons aged 65 years or older fall annually,^7,8 which leads to high health care demand, morbidity, and mortality.^2,9–13 The incidence rate of falls is age- and gender-dependent. The majority of falls and related injuries occur in older women (75 years or older).^9,14,15 Furthermore, the incidence rates of fall-related injuries among older adults increased over the past 30 years.³ New guidelines and preventive strategies were developed to reduce falls among older adults.¹⁶

Absolute numbers of fall-related deaths are increasing because of aging societies worldwide.¹⁷ In The Netherlands, the absolute number of persons aged 65 years or older is expected to double (up to 25% of the population) in 2040 (this percentage was 15% in 2008).¹⁸ The figures of aging societies are comparable to worldwide trends.¹⁹ Because of the expanding older population, based on an increasing life expectancy, an increase in absolute numbers of fall-related deaths might be expected in the near future. To investigate how fall-related mortality in persons aged 65 years or older developed over time in The Netherlands, absolute numbers of fall-related deaths and adjusted mortality rates corrected for demographic changes were quantified from 1969 to 2008. Significant health differences between both genders were demonstrated in the older population,²⁰ and it was known that temporal trends in disease mortality might show large gender differences as well.^21,22 Therefore, separate analyses for men and women were conducted, which were further specified by age group.

METHODS

Mortality data were obtained from the public accessible electronic database²³ of Statistics Netherlands (CBS), including the Official-Cause-of-Death-Statistics (OCDS), in which the data were stored in 5-year age groups for both genders during the study period. Data on unintended fall-related deaths of persons aged 65 years or older were collected between 1969 and 2008. An unintentional fall was defined using the WHO's International Classification of Diseases (8th revision, code E880-E887; 9th revision, code E880-E888; 10th revision W00-W19, X59.0)^24–26 as slipping, tripping, and stumbling.

CBS collects mortality data with a uniform classification system and has almost complete national coverage (99.7% in 2008). The OCDS is based upon the registration of the cause of death, by an official death certificate, which is required in all cases of death.²⁷ A death certificate may only be completed by the consulting physician who declares the patient dead, or by a coroner. The death certificate includes data on age, gender, location, cause of death, and comorbidities that contributed to the death. Completeness and acceptability of the death certificates are verified with the national birth registry. In cases of uncertainty, the physician who completed the death certificate is contacted to provide additional information. In case of a nonnatural death (e.g., traffic accident, fall, or suicide), the coroner can perform an official autopsy to determine the official cause of death.

Numbers of fall-related deaths were specified by age and gender. The age-specific mortality rates were calculated in 5-year age groups using the number of the fall-related deaths in that specific age group, divided by the population size within that specific age group for male and female patients, and was expressed per 100 000 persons in that age group. Age-adjusted incidence rates allowed us to compare the incidence rate for a standardized population during the study period and were performed by direct standardization to correct for demographic changes throughout the study. The age-adjusted incidence rate was therefore the outcome of interest. The midyear population was used for each year of the study. Data on demographics were obtained from CBS. Mortality data were absolute numbers covering the whole Dutch population.

To model the trend in fall-related mortality, a regression model with Poisson error and log link was built with log midyear population size of each year of the study as an offset factor. This model gave evidence of increasing linear trends, stable trends over time, or decreasing trends. To assess whether the trend was stable or changed during the study period, a Joint-Point regression analysis was used. This analysis identified points where a statistically significant change over time occurred in the linear slope of the trends in fall-related mortality rates (Join-Point Regression Program, Version 3.4.3; Statistical Research and Applications Branch, National Cancer Institute, Bethesda, MD). The Joint-Point function accommodated piecewise linear fits, connected with one another at the best Joint-Point,²⁸ and showed the necessity for assuming a spline instead of a simple linear model. In Joint-Point analysis, the best-fitting points corresponded to where the rate changed significantly (i.e., increased or decreased). The analysis started with the minimum number of Joint-Points, and tested whether one or more Joint-Points were statistically significant in the model and should be added. In the final model, each Joint-Point indicated a statistically significant change in trend, and an annual percentage change was computed for each of those segments by means of generalized linear models, assuming a Poisson distribution. The best Joint-Points were used for the analysis, and divided the study into 3 periods (1969–1983, 1984–1996, and 1997–2008). The parameter for calendar year, corrected for gender and age group, was transformed into percentage annual change (PAC). Interactions of the spline for gender were added and tested to investigate differences in trends for genders. Statistical analyses were performed using SPSS software version 16.1.1 (SPSS Inc, Chicago, IL). A P value < .05 was considered statistically significant.

RESULTS

The population aged 65 years or older increased from 1.3 million in 1969 to 2.4 million in 2008, representing 10.0% and 14.7% of the Dutch population, respectively. Approximately 90% (63 463) of all fall-related deaths in The Netherlands during that time occurred in the population aged 65 years or older (Table 1).

TABLE 1—

Absolute Number of Fall-Related Mortality in the Dutch Population: 1969–2008

Period	Aged < 65 Years, No. (%)	Aged 65–69 Years, No. (%)	Aged 70–74 Years, No. (%)	Aged 75–79 Years, No. (%)	Aged 80–84 Years, No. (%)	Aged ≥ 85 Years, No. (%)	Total No.
1969–1974	1322 (10.7)	525 (4.2)	960 (7.8)	1906 (15.4)	3087 (24.9)	4577 (37.0)	12 377
1975–1979	831 (9.0)	355 (3.8)	632 (6.8)	1267 (13.7)	2178 (23.5)	3989 (43.1)	9252
1980–1984	774 (9.6)	257 (3.2)	490 (6.1)	1027 (12.8)	1700 (21.1)	3804 (47.2)	8052
1985–1989	729 (9.6)	207 (2.7)	375 (5.0)	864 (11.4)	1505 (19.9)	3879 (51.3)	7559
1990–1994	704 (8.9)	189 (2.4)	369 (4.7)	807 (10.3)	1480 (18.8)	4320 (54.9)	7869
1995–1999	846 (10.6)	231 (2.9)	417 (5.2)	844 (10.6)	1451 (18.2)	4176 (52.4)	7965
2000–2004	1068 (11.4)	269 (2.9)	476 (5.1)	1052 (11.2)	1726 (18.4)	4794 (51.1)	9385
2005–2008	813 (10.0)	257 (3.2)	430 (5.3)	846 (10.5)	1583 (19.6)	4162 (51.4)	8091
Total	7087 (10.0)	2290 (3.2)	4149 (5.9)	8613 (12.2)	14 710 (20.9)	33 701 (47.8)	70 550

The absolute number of fall-related deaths in persons aged 65 years or older annually decreased from 1927 in 1969 to 1312 deaths in 1985, but since the mid-1980s the number of deaths showed a persistent increase to 1892 in 2008 for men and women together (Figure 1). The majority of fall-related deaths occurred in the population aged 85 years or older, which represented an increasing share in the proportion of fall-related mortality among older persons, from 37% in 1969 to more than 50% since 1985 (Table 1).

FIGURE 1—

Absolute number of fall-related deaths and age-adjusted fall-related mortality rates in persons aged 65 years or older in the Netherlands; 1969–2008.

Note. The percent annual change was: Period 1: rapid decrease, men (1969–1983) −4.1% (95% confidence interval [CI] = −4.9, −3.2), and women (1969–1985) −6.5% (95% CI = −7.1, −5.9); Period 2: flattening decrease, men (1984–1996) −1.4% (95% CI = −2.4, −0.4), and women (1986−1996) −2.0% (95% CI = −3.4, −0.6); Period 3: slight increase, men (1997–2008) 1.9% (95% CI = 0.6, 3.2), stabilized rate, women (1997–2008), 0.0% (95% CI = −1.2, 1.3).

Gender- and age-specific fall-related mortality rates in persons aged 65 years or older decreased in all age groups, both for men and women throughout the study period. However, age-specific rates of fall-related mortality increased with age, and continued to be highest among adults older than 95 years for both men and women (Table 2).

TABLE 2—

Age-Specific Incidence Rates of Fall-Related Mortality in the Older Dutch Population per 100 000 Persons: 1969–2008

	Aged ≥ 65 Years Overall	Aged 65–69 Years		Aged 70–74 Years		Aged 75–79 Years		Aged 80–84 Years		Aged 85–89 Years		Aged 90–94 Years		Aged ≥ 95 Years
Period		Men	Women	Men	Women	Men	Women	Men	Women	Men	Women	Men	Women	Men	Women
1969–1974	129.5	18.3	17.5	38.9	45.2	94.6	140.7	242.7	414.8	504.7	867.4	1062.7	1617.0	1727.9	2713.3
1975–1979	110.0	15.5	11.9	30.3	29.2	73.0	93.8	193.0	279.6	409.0	681.2	862.5	1268.5	1565.9	2223.0
1980–1984	87.0	11.3	8.0	23.7	20.0	59.5	62.0	149.9	174.9	338.1	449.9	721.9	915.4	1311.1	1702.7
1985–1989	75.0	9.2	5.4	18.3	14.1	53.7	43.3	129.5	126.1	311.4	327.1	733.7	753.1	1457.0	1319.2
1990–1994	72.7	7.7	4.7	16.7	12.7	47.7	38.4	123.6	109.2	298.3	305.1	670.4	717.5	1323.9	1330.3
1995–1999	67.9	9.4	5.4	17.6	13.3	46.5	36.2	108.8	103.6	274.2	251.8	596.1	580.3	1294.5	1202.4
2000–2004	75.2	10.4	6.2	20.4	13.8	51.1	42.8	124.5	106.5	294.7	264.7	659.1	603.4	1405.0	1162.6
2005–2008	76.7	11.2	7.0	24.0	13.5	54.0	37.5	132.4	107.7	292.8	248.7	784.6	548.7	1357.8	1048.4

During the whole study period (1969–2008), the PAC for the age-adjusted fall-related mortality rate decreased −1.54% (95% confidence interval [CI] = −1.66, −1.43; P < .001) per year in older men and by −3.61% (95% CI = −3.69, −3.53; P < .001) per year in older women. The Joint-Point regression analysis revealed that the PAC in mortality rate was not stable over time and could be divided into 3 phases (Figure 1). Until the mid-1980s, the PAC in mortality rates because of a fall decreased strongly in both men (−4.1%; 95% CI = −4.9, −3.2; P < .001) and women (−6.5%; 95% CI = −7.1, −5.9; P < .001). Subsequently, after that period, until the mid-1990s, the decrease leveled off in both men (−1.4%; 95% CI = −2.4, −0.4; P < .001) and women (−2.0%; 95% CI = −3.4, −0.6; P < .001). Finally, over the last decade, the age-adjusted mortality rate remained stable among older women (0.0%; 95% CI = −1.2, 1.3; P = .97) and started to rise again in older men (1.9%; 95% CI = 0.6, 3.2; P = .005).

In 1969, the age-adjusted mortality rate for women (260.4 per 100 000 persons) was twice as high as for men (119.4 per 100 000 persons). However, the rates for men are catching up to those of women. In 2008, the age-adjusted mortality rates were 69.8 per 100 000 older women and 62.8 per 100 000 older men (Figure 1).

DISCUSSION

This observational study was based upon the OCDS of CBS. The aim was to determine secular trends in fall-related mortality in persons aged 65 years or older. The spectacular reduction in the age-specific and age-adjusted fall-related mortality rate in the 1970s and 1980s ended in the mid-1990s. Moreover, since then, mortality rates because of falls have been rising among older men in similar rates to those seen in women.

A strength of this study was the accurate OCDS monitoring of nearly all deaths in The Netherlands (99.7% in 2008), which allowed us to assess long-term trends over 40 years, covering the whole Dutch population. Because the database was verified with the national birth registry,²⁷ the mortality rates and numbers presented were real figures rather than estimates of the entire older Dutch population.

Petridou et al.²⁹ found decreasing trends in unintended fall-related mortality rates in persons aged 65 years or older for most European countries, but this study did not go beyond 2002 and might have been conducted too early to pick up unfavorable trends within the first decade of the 21st century. Similar trends in fall-related mortality as observed in our study were seen in Finland, with a recent increasing trend for men and a stable rate for women.³⁰ The observed change in the trend in fall-related mortality rates in the final period of our study might be explained by an increasing life expectancy, which is accompanied by a prolonged period of comorbidities and disabilities.³¹ Our hypothesis was that the fall-related mortality mainly occurred among older adults who are frail, resulting in an increased fall risk and less functional reserve to cope with the consequences of falls. It was not likely that migration significantly affected the observed mortality trends to a large extent because the proportion of migrants was only 4% of the older Dutch population, and no significant differences in fall-related mortality by ethnic background exist in The Netherlands.³²

Although the age-specific mortality rates remained highest among women, the mortality rates of men are catching up to the mortality rates of women. A possible explanation for this finding could be that the life expectancy in men increased more rapidly than in women over the last decades, resulting in a smaller gap in the life expectancy between men and women.³¹ Consequently, men became more vulnerable for age-related (co)morbidities over a prolonged period of life. This assumption was supported by our previous report on a more rapid increase in fall-related hospitalizations in older men than in older women in The Netherlands.^3,33 Because age itself is an independent risk factor for falls,³⁴ the number of falls will grow with an increasing life expectancy.

The shift in mortality rates might also be based upon gender-specific injury patterns that were demonstrated for men and women.⁹ Fractures are the main fall-related injuries in women. Fracture treatment, especially for hip fractures, improved rapidly over the last few decades, and might have contributed to the decreasing fall-related mortality rates. However, men presented more often with severe skull–brain injuries,¹ which have fewer therapeutic options. These gender-specific injury patterns, together with the differential increase in life expectancy, might have contributed to the (near) closure of the gap between the fall-related mortality rates between men and women.

A limitation of the present study was that the numbers and incidence rates of fall-related mortality among older Dutch adults might not be directly generalizable to other populations. However, absolute numbers and rates of fall-related mortality also increased in the United States³⁵ and Finland² over recent years, which showed that our findings were not unique and might possibly indicate a universal phenomenon for western countries. Further studies are required to show similar trends for other countries.

In addition, during the study period, new versions (9th and 10th revision) of the International Classification of Diseases (ICD) system were published and implemented in the OCDS, which may have led to changes in the coding of subgroups of falls. The code for “falls” in the ICD-8 and ICD-9 included a category, “Fractures unspecified.”^24,25 In the ICD-10,²⁶ this code was replaced by code ×59.0 (i.e., fracture, cause unspecified). Analysis of coroner reports by CBS showed that nearly all fractures contributing to the mortality cases in older adults were caused by a fall. Because of these findings, after introduction of ICD-10,²⁶ CBS decided to include all cases with the code ×59.0 within the category of falls. In our study, we followed this decision and included all codes ×59.0 in the analysis to avoid artificial trend interruptions in our time series analyses because of changes in registration practice or coding. However, the WHO and European Injury Prevention Network analytical databases did not include ×59.0 in the category of falls, which must be considered when comparing our findings with studies using these databases.

For comparative purposes, a sensitivity analysis of trends in fall-related deaths was conducted from the introduction year of ICD-10 (1994)²⁶ onwards under different assumptions concerning the cause of fractures, cause not specified (×59.0). The sensitivity analysis showed that the addition of the code ×59.0 did not inflate our time trend figures, but instead produced a conservative estimate of unfavorable developments in the most recent period of our study (1997–2008).

Mortality rates were a combination of injury incidence and case fatality rates.³⁶ In a recent study, we showed that the incidence of severe fall-related injuries requiring hospitalization increased 61% to more than 30 000 annually in The Netherlands between 1981 and 2008.³ The decreasing rate of fall-related mortality made it therefore likely that the case fatality rate of fall incidents declined. The decrease might be explained by several factors. During the study period, new diagnostic tools like computed tomography and magnetic resonance imaging were introduced. Standardized protocols for treatment of trauma patients at the Emergency Department were also introduced, such as the Advanced Trauma and Life Support.³⁷ Furthermore, new and better treatment possibilities were developed during the study period, (e.g., osteosynthesis material and intensive care units). The possible effects of all these factors could not be specified, but it was unlikely that the introduction of falls prevention guidelines had a strong effect on fall-related mortality because the guidelines were only introduced in The Netherlands after 2000.³⁸ Fall risk and, hence, fall-related mortality in general, might be increasing because of increased mobility of frail elderly resulting from the use of walking aids and other equipment (i.e., electric mobility scooters), which was seen in The Netherlands and the United States.^39,40 As a result of sustained walking abilities, these older persons remained at risk for fall incidents. Another potential cause that was shown to contribute to an increased fall risk in older adults was a more active lifestyle (e.g., sports participation).⁴¹

In conclusion, the spectacular decrease in fall-related mortality since 1969 ended around the mid-1990s. Currently, the mortality rates in women are stable, but the rates are increasing in older men. Falls are expected to become an even more serious cause of death in the near future because of aging societies worldwide. In recent years, the health care system has not been able to compensate fully for the aging society and for rising numbers of fall-related injuries. These findings indicate the need for further research identifying factors that can be used to reduce fall-related mortality among older adults and to control for the recent unfavorable trend.