METHODS

Computerized national mortality data were obtained from the Department of Health of Taiwan for the years 1975 to 1998. Nine age groups (<4, 5–14, 15–24, 25–34, 35–44, 45–54, 55–64, 65–74, and >75 years) were used to calculate age adjusted injury mortality rates by intent and mechanism of injury. The standard world population of 1976 was used for adjustment.²

The ICD codes for the matrix by intent and mechanism groupings were based on CDC recommendations.⁵ However, because the Taiwan Office of Statistics, Department of Health uses only three digit codes for external causes of death, many categories in the matrix (for example, motor vehicle traffic injuries, fire/burn, and firearm) could not be further specified by intent. Thus the detailed analyses by both intent and mechanism were only available for five mechanisms of injury: cut/pierce, drowning, fall, poisoning, and suffocation.

RESULTS

On average, more than 10 000 people die from injuries each year in Taiwan, a country with a population of 21 million in 1998. The number of injury deaths in 1975–78 was 44 046 and increased to 59 552 in 1995–98. The age adjusted death rate increased from 106.4 per 100 000 in 1975–78 to 124.2 in 1987–90 and leveled off in 1990s (table 1). Unintentional injuries accounted for 80% of all injury deaths in each period. Motor vehicle traffic, drowning, falls, poisoning, and suffocation were the top five mechanisms in each period and comprised 80% of all injury deaths.

The various mechanisms of unintentional injury were ranked as motor vehicle traffic followed by drowning, fall, poisoning, and suffocation in most periods. Drowning was the second leading mechanism of death in most periods in Taiwan except in 1995–98 when falls ranked second. If all intents are combined, motor vehicle still ranked first in each period, although the relative importance of poisoning and suffocation increased. Poisoning became the second leading mechanism of death from 1979–82 to 1991–94. The number of deaths from suffocation also surpassed those dying from falls from 1975–78 to 1987–90. In 1991–94, 4589 people died from suffocation; this increased by 22% to 5587 in 1995–98 (table 1).

The trends of age adjusted death rates for four selected mechanisms of injury are illustrated in fig 1. The patterns of changes in mortality trends for falls and drowning were similar between unintentional only and all intents combined. However, for poisoning and suffocation, the pattern of changes showed large differences between unintentional only and all intents combined. The poisoning death rate for all intents combined was three times as high as for unintentional poisoning, and that for suffocation there was a fivefold difference between total and unintentional death rates. Mortality trends of poisoning by all intents combined revealed a steeper increase from 1975–78 to 1983–86 and a steeper decrease from 1987–90 to 1995–98, in contrast to the generally flat pattern for unintentional poisoning only. An increase in suffocation with all intents combined was noted in 1995–98.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Trends of age adjusted death rates for four selected mechanisms of injury in Taiwan: unintentional only (left) and all intents combined (right).

Intent was further classified into four subcategories. A large surge in death rate for “intent unknown” was seen in the 1990s (fig 2). The increase in suicide death rates in the mid-1990s differed by different mechanisms: poisoning was the main mechanism before the 1990s, accounting for three fifths of suicide deaths, with suffocation (hanging) becoming the most common means of suicide in the 1990s. Cut/pierce was the main mechanism of homicide in Taiwan throughout the entire study period, although the proportion decreased yearly, from 78% in 1975–79 to 58% in 1995–98 (table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Patterns of changes in age adjusted death rates for selected injuries by mechanism and intent, Taiwan, 1975–78 to 1995–98.

DISCUSSION

The results show that the relative importance of poisoning and suffocation increased if all intents are combined. Given the same mechanism of injury, different intents showed different patterns of change during the study period. And, given the same intent, the patterns of change for different mechanisms of injury also differed across time. The simultaneous presentation of mortality data by both intent and mechanism of injury provides quite different profiles from those that present mechanism only within unintentional injuries.

One important rationale underpinning the presentation of mortality data by both intent and mechanism is that an effective passive measure will prevent unintentional as well as intentional injuries.⁵ Baker, a major proponent of this idea, points out that the large declines in both unintentional and suicidal poisonings in England and the United States were probably due to the replacement of poisonous coal gas by natural gas. She further illustrates the experience of Sri Lanka and China where agricultural poisons are used as a means of suicide by many despondent women, while at the same time killing many curious children.¹⁴ One recent example was legislation restricting package sizes of paracetamol and salicylate in the United Kingdom. This produced substantial reductions of mortality and morbidity associated with self poisoning with these drugs.¹⁵

In 1972, the government of Taiwan passed the Pesticides Regulation Act. Most of the regulations are targeted at manufacturers to decrease exposures to pesticide during the manufacturing process. In 1989, the Council of Agriculture promulgated a series of regulations on the use of pesticides. In the early 1990s, the Council of Agriculture required that manufacturers of paraquat, the most common agent of fatal poisoning in Taiwan,¹⁶ add an unpleasant odor and color, as well as an emetic which make paraquat difficult to drink regardless of whether it is done unintentionally or intentionally. The decreasing proportions of people using poisoning as a means of suicide might in part be due to these measures.

Some people might argue that if we make it harder to commit suicide by one means, people will choose another. As shown in fig 2, we did not see any compensated increasing of suicide death rate by other means. We thus confirm Hassall and Trethowan's argument that the total suicide rate went down when the preferred means became impossible.¹⁷

With regard to suffocation (mainly hanging), cut/pierce, drowning, and fall, it is impossible to design a passive preventive measure to prevent both unintentional and intentional injuries in all kinds of settings. We have to “nibble away” at them according to specific circumstance and high risk group. For example, after identifying the high incidence of suicide attempts in jails among Hopi American Indians the authors suggested removing belts, shoelaces, and other means of attempting suicide from recently arrested men.¹⁸ The same would apply to people in other institutions, such as mental facilities. Bars on windows in mental facilities can prevent suicide, just as bars on windows in apartments prevent unintentional falls. The idea also applies to motor vehicles—when someone tries to kill themselves by driving into a tree, they are much less likely to succeed now that cars have airbags. The design of personalized handguns can also prevent both unintentional injuries from children and intentional assaults from others.¹⁹

When injury mortality data are presented using both intent and mechanism of injury, we need to be cautious about the possibility of misclassification of intent for different mechanisms.^20–24 Rockett and Smith proposed that suicide by elderly Japanese women is relatively underestimated because of the misclassification of suicidal drownings as unintentional.²⁵ Fingerhut and McLoughlin also recommend that the criteria used to rule if a death is intentional or unintentional or undetermined vary by country.⁴ These points should be taken into consideration when interpreting cross national comparisons of mortality rates by intent.

When interpreting differences in temporal trends in injury mortality by both intent and mechanism of injury, we also need to consider the possibility of changes in the people who certify deaths by providing information on death certificates, as well as changes in the habits of people who code the assignment of external causes of death. There has been a surge in death rates with unknown intention (fig 2) which may reflect these attitudinal changes.

This study shows that presentation of mechanism of injury within intentional injuries yields different profiles of injury problems. Thus the simultaneous tabulation of injury mortality data by both intent and mechanism is a necessary step for identifying and prioritizing injury problems. We also confirmed the argument that good preventive measures could prevent both unintentional and intentional injuries.