Methods

Mortality data came from death certificate information, based on the underlying cause of death, recorded by the national vital statistics system and accessed via the Centers for Disease Control and Prevention (CDC) web-based injury statistics query and reporting system (WISQARS) online database.1 WISQARS mortality reports provide tables with the numbers of injury-related deaths and the mortality rates per 100 000 population according to cause (circumstances or mechanism) and intent of injury by sex, age group, race and state for the years 1981 to 2006. These mortality data are from annual data files of the National Center for Health Statistics, CDC, and are derived from multiple cause of death data.7 Age-adjusted rates are calculated by the direct method standardised to the total US population. The year 2000 was chosen as the standard year.

WISQARS non-fatal injury reports2 were used to examine the change in morbidity rates among population subgroups that showed significant increases in mortality rates during 2001–7. WISQARS non-fatal is based on data from hospital ED. The numbers of non-fatal injuries presented in WISQARS non-fatal are national estimates based on weighted data from the US Consumer Product Safety Commission's national electronic injury surveillance system.2 Because data collection for the national electronic injury surveillance system all injury programme did not begin until July 2000, and national estimates for 2000 were computed using data obtained from July to December of that year, we used national estimates for 2001 and later years.2

The classification scheme of race or Hispanic origin for WISQARS non-fatal injuries reports assumes that most white Hispanic individuals were recorded on the ED record as Hispanics, and most black Hispanic individuals were recorded as black.2 Further reasons for not categorising Hispanic individuals by race were that most mortality rates for Hispanic subgroups were unstable due to small numbers of deaths and preliminary analyses of Hispanic individuals by race (not shown here) revealed that significant increases in injury rates in the elderly occurred only in Hispanic whites, who comprised 93% of Hispanic individuals during 1999–2005.8 WISQARS non-fatal injuries reports include results for five racial/ethnic categories: non-Hispanic white (‘white’); black (including Hispanic and non-Hispanic black); Hispanic (excluding black Hispanic); other, including non-Hispanic Asians and Native Americans and unknown (correspondence from L Annest, CDC, May 13 2009). To ensure consistency of race/ethnic classification, we used the first three race/ethnicity groups above (white, black and Hispanic) for both mortality and morbidity. (Using a group that combines Asians and Native Americans is not informative because they have strikingly different injury rates.)

The overall trend in cause-specific mortality rates from 2000 to 2006 was first examined to determine the significance of changes. The causes of common injuries showing significant increases were further subjected to subgroup analysis by cause, sex and race/ethnicity in order to compare changes among subgroups. In addition, changes in mortality rates from unintentional falls were examined for three age groups (65–74, 75–84 and 85 years and over).

Linear regression was used to examine the statistical significance of trends in mortality rates in the study period. The percentage change in rates was used to measure the linear trend, which was calculated as regression coefficient × 100 × 6 divided by the rate of 2000.9 Unlike a percentage change calculated as ‘(the rate in the ending year − the rate in the beginning year)/the rate in the ending year × 100/%’, this formula takes into account fluctuations in the intervening years by fitting the linear regression between death rate and year in this period. An alpha of p<0.05 was selected as the level of statistical significance. Stata version 10.0 was used for data analysis.

Results

The overall change in injury mortality for individuals aged 65 years and older during the 7 years, 2000–6, was 3%, a non-significant increase (table 1). Significant increases occurred in death rates from all unintentional injuries (6%), motorcycle crashes (145%), machinery (46%), falls (42%), poisoning (34%) and drowning (19%). Small but significant decreases occurred in death rates from motor vehicle crashes (−13%), suffocation (−12%) and suicide (−8%).

Morbidity from non-fatal injury, as reflected in emergency room visits, showed a non-significant 7% increase for both all injuries and unintentional injuries (table 2). The rate of visits for falls fluctuated in this period, without a significant increase. Significant increases were seen in non-fatal poisoning (143%) and injuries related to motorcycles (86%), machinery (48), bicycles (24%), struck by/against (13%) and overexertion (11%). The only significant decrease was in motor vehicle occupant injury (−16%).

Subgroup analyses

Falls

The death rates from falls increased by at least 38% in all racial/ethnic groups, with the greatest increase seen in white individuals (45%) (figure 1). Subdivision analysis by age group (table 3) revealed that in white individuals of each age group, death rates from falls increased by at least 37%; significant increases were seen only for Hispanic individuals aged 65–84 years and for black individuals aged 75 years and over. In contrast to fall mortality, rates of visits to ED for fall-related injury did not increase (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Percentage change in mortality rate (2000–6) and emergency department visits (2001–7) from falls by race/ethnicity among individuals aged over 65 years, Unites States. *p<0.05.

Traffic-related crashes

Motor vehicle traffic-related mortality declined by 13%, primarily due to a decrease in death rates of occupants, the largest group (table 1). Eighty-five per cent of the motorcyclist deaths in 2006 occurred in white men, whose death rate increased from 0.5 per 100 000 in 2000 to 1.2 in 2006 (a 139% increase). Rates of ED visits for motorcycle crash injury increased by 77% for white men (not shown).

Poisoning

Poisoning deaths increased by 21% in white men. Increases were seen in Hispanic (51%) and white (34%) women. Even larger increases in rates of ED visits for non-fatal poisoning occurred in white men, Hispanic women and white women, with increases of 107%, 168% and 124%, respectively (not shown).

Machinery

Machinery deaths occurred almost exclusively (95%) in white men, among whom the rate increased by 55% (from 0.9 per 100 000 to 1.4). ED visits of Hispanic men showed a 23-fold increase for machinery, from 3.3 per 100 000 in 2001 to 77.0 in 2007 (not shown).

Drowning

Only white men experienced a significant increase (17%) in death rates from unintentional drowning. For non-fatal drowning, the rates fluctuated greatly during 2001–7 and no significant increase was found (not shown).

Discussion

Our findings reveal significant and important increases between 2000 and 2006 in death rates from unintentional falls, machinery, motorcycle crashes, poisoning and drowning among Americans aged 65 years and over. Increases also occurred in non-fatal injuries from machinery, motorcycle crashes and unintentional poisoning. These increases are important and merit further investigation, but have been largely unrecognised or ignored, perhaps because of the small change (a 6% increase) in the overall unintentional injury mortality rate.

Several studies describe injury risk factors that are common among the elderly. Brophy et al10 reported that adults with disabilities are at increased risk of injury. Rohr11 found that the use of alcohol is a major contributing risk factor for unintentional injury and death; the risk is even greater in the older population, among whom smaller amounts of alcohol may result in significant impairment and injury. Given the association between alcohol and injury, recent increases in alcohol problems among the elderly suggest a partial explanation for the increase in fatal falls: admissions for alcohol treatment among adults aged 55 years or older increased by 19% for men and 24% for women, from 1995 to 2002.12 A positive association between body mass index and the probability of sustaining an injury was found by Finkelstein et al.13 However, the prevalence of overweight and obesity did not increase in the US population aged 65 years and over between 1999–2000 and 2003–4,14 suggesting that this is not a factor in the increase in fall mortality.

Falls are the most common cause of fatal and non-fatal injuries among older adults.15 16 Approximately 5.8 million persons aged 65 years and older fell at least once during the preceding 3 months, and 1.8 million of those who fell sustained an injury that resulted in a medical visit or restricted activity for at least 1 day.17 The direct medical costs of falls totalled US$0.2 billion for fatal injuries and US$19 billion for non-fatal injuries in 2000.18 Due to these facts, any significant changes in falls mortality are important. We found that the increase in fatal falls occurred in all the racial groups except Hispanic women.

Before our study, Paulozzi et al19 reported an increase in the falls death rate in the age group 65 years and over from 1992 to 2002. Internationally, an increasing trend in both mortality and morbidity rates among elderly Finns aged 50 years and older between 1970 and 1995 was reported by Kannus et al,20 a rise that could not be explained merely by demographic changes.21 Several studies have attempted to explain the recent increases in falls mortality.

For all causes of falls in the age group 65 years and over, however, emergency room visits did not show a significant increase in either overall or subgroup rates. Data on hospital discharges between 2000 and 200622 revealed that neither fractures of the neck of the femur nor other fractures increased among individuals aged 65 years and over during 2000–6. The absence of increases in ED visits for falls or hospital admissions for fractures strongly suggests that the increase in reported fatal falls in the elderly might result from: (1) an artefact, prompted by a trend in recent years to report the underlying cause of death as a fall, rather than as pneumonia or other sequelae of a fall; (2) the increased occurrence of severe falls, for example, if older adults fall more often and with more serious consequences due to exposure to new risks for falls—for example, risks associated with greater physical activity and (3) the fact that many elderly individuals are now living longer with health conditions that may predispose them to falling or fractures. These questions merit future study. The importance of differential diagnosis (reporting a death as a fall vs as the result of ensuing complications) is illustrated by the huge difference in fall mortality in the population aged 65 and over in Europe, where fall death rates per 100 000 person-years range from 15 or less in Spain and Greece to over 150 in Hungary and the Czech Republic.23

In addition to the increase in falls mortality, increases in fatal and non-fatal injury from machinery, motorcycle crashes and poisoning among older adults are revealed by this research. They have received little attention from researchers and policy-makers. Using traffic exposure data for the entire population, Beck et al24 found motorcyclists, pedestrians and bicyclists faced increased risks of both fatal and non-fatal injuries. Similar findings from non-fatal injuries suggest that the increase in injury mortality rates from machinery, motorcycle crashes and poisoning resulted mainly from the increase in exposure to these threats. It must be pointed out that the increase in poisoning should be interpreted with caution because a study in California showed that a portion of the 18% increase in the poisoning death rate between 2001 and 2002 was a result of stricter procedures concerning data processing that were implemented in 2002.25 However, Warner et al26 provided details about the specific substances involved in the dramatic increase in poisonings from 1999 to 2006. Although separate data are not available for the elderly, for all ages combined some of the greatest increases were seen in deaths from opioids coded as T40-2 (ICD 10)—morphine, oxycodone and hydrocodone. Deaths from these prescription narcotics, which are widely prescribed for the elderly, doubled in the reported 5-year period. To control the increasing trend effectively, future research is needed to identify risk factors that increase the exposure of older adults to these hazards and to develop specific interventions.

Our study has three main limitations. First, the lack of detailed data available from death certificates and thus from WISQARS prevents us from examining the association between elderly mortality rates and relevant influences, such as socioeconomic factors, which might explain changes in unintentional injury mortality or in exposure data, protective behaviours and risk factors that are critical for the development of effective policy interventions. The second limitation is the slight difference in time period between fatal data (2000–6) and non-fatal data (2001–7). A third limitation is that the WISQARS non-fatal injury data are estimated based on a sample, and for some specific causes are not stable and robust. In addition, we were not able to conduct analyses to determine whether the discrepancy between falls mortality and morbidity trends are caused by changing practices in coding injuries and deaths.

Dessypris et al27 recently estimated that for the US population as a whole, a reduction of approximately 25% in overall unintentional injury mortality, including a saving of approximately 5500 lives annually of elderly persons, could be achieved if all regions of the USA had death rates as low as rates in the regions with the lowest rates. Regardless of whether equalisation of rates is possible, the regional variations should trigger research to shed light on the reasons for the differences and the possibilities of successful preventive efforts.

In conclusion, US residents aged 65 years or older have recently experienced significant increases in fatal unintentional falls and in both fatal and non-fatal injury related to machinery, motorcycle crashes and poisoning. These changes may have been ignored because of the relatively small increase in the overall mortality rate for injuries. Future studies are needed to identify risk factors contributing to these increases and effective preventive measures.