Trends in poisoning hospitalization and mortality in Taiwan, 1999-2008: a retrospective analysis

Data source

The mortality data in Taiwan from 1999-2008 were obtained from the cause-of-death statistics provided by the Department of Health, Executive Yuan [7]. Hospitalization data were extracted from the NHI Research Database [10]. Since its launch on March 1, 1995, the NHI has reached a coverage rate of 99% of hospitals in Taiwan. The NHI Research Database consists of data on both outpatient (including emergency treatment) and inpatient care. Medical institutions are required by law to make claims for outpatient costs and inpatient fees to the NHI Bureau. Hence, the NHI Research Database is an essential resource of representative and evidence-based data for studies in medical and health research, which may play a valuable role in future health policy [10].

Our research required secondary data analysis, and the identification of patients had already been encrypted in the official death statistics and the NHI Research Database in Taiwan; therefore, our research complied with the Helsinki Declaration and protected the privacy of patients.

Data analysis

We calculated the mortality and hospitalization rates using the mid-year population in the "population by age report" provided by the Ministry of the Interior as the denominator. Data were age-adjusted by the direct method to the 2000 World Standard population obtained from the WHO. Afterwards, we calculated the 10-year trends in poisoning mortality and hospitalization rates and stratified them by gender, age, and causative agent.

The Department of Health age classification, which uses 6 groupings, was adopted as follows: infants and toddlers, aged 0-4 years; children, aged 5-14 years; young adults, aged 15-24 years; mature adults, aged 25-44 years; middle-aged adults, aged 45-64 years; and the elderly, aged 65 years or older. The official death statistics and NHI database employ the ICD-9-CM coding system in Taiwan, so we selected study cases by ICD-9-CM external causes of injury codes (E-Codes), which include "unintentional poisoning, such as by medicinal drugs (E850-E858), alcohol (E860), cleaning solutions (E861), petroleum products (E862), pesticides (E863), corrosives (E864), foodstuffs (E865), other solid and liquid substances (E866), gas distributed by pipeline (E867), carbon monoxide (E868), other gases and vapors (E869), venomous animals and plants (E905), and medicinal drugs causing adverse effects (E930-E948)"; "intentional poisoning, such as suicide using solid or liquid substances (E950), suicide using gases in domestic use (E951), suicide using other gases and vapors (E952), and assault using poisoning (E962)"; and "substances, undetermined whether accidently or purposely inflicted (E980), by unspecified gases in domestic use (E981), and by other unspecified gases (E982)."

For statistical analysis, the independent sample Student t-test, one-way analysis of variance, and Scheffe's post hoc comparisons were conducted using SPSS 18.0 software to examine the differences in poisoning mortality and hospitalization rates according to gender, age, and poisoning agent. Poisson regression was used to identify 10-year trends.

Epidemiological characteristics

Long-term trends

Between 1999 and 2008, there was an increasing trend in overall poisoning mortality rate with a significant 47.2% increase (p = .002). Men demonstrated a larger increase in mortality rate than women during this time (57.6% versus 26.2%, p = .001 and .012, respectively) (Figure 1a). There was also a rising trend in overall poisoning hospitalization rate, with a 79.1% increase, p < .001. Men had a larger increase in hospitalization rate than women (98.1% versus 61.4%, p < .001 and p = .001, respectively) (Figure 1b).

The population aged 65 years or older demonstrated a declining trend in mortality with a drop of 19.9% over the 10-year period (p < .001). However, the 25-44 year age group showed a rising trend in mortality rate, with a 120.7% increase (p < .001), and in 2005, the rate in this group surpassed that of subjects aged 65 years or older to become the primary high-risk group for poisoning (Figure 2a). Considering hospitalization, the age group of 65 years or older demonstrated the largest rise in hospitalization rate, with an increase of 165.7% (p < .001), followed by the 45-64 and 25-44 year age groups, with increases of 96.1% and 30.8%, respectively (p < .001 and .055, respectively) (Figure 2b).

Mortality from unintentional poisoning declined by 35.8% between 1999 and 2008 (p < .001). While mortality from intentional poisoning showed an increase of 120.0% (p = .001), mortality from unspecified poisoning decreased by 42.7%, but it failed to reach statistical significance (p = .396) (data not shown). There was a 123.1% rise in the unintentional poisoning hospitalization rate (p < .001), which was in line with the overall hospitalization rate. While the intentional poisoning hospitalization rate decreased by 23.3% (p = .151), the unspecified poisoning hospitalization rate increased by 54.1% (p < .001) (data not shown).

After analyzing the causes of unintentional poisoning, we found that the primary agent of unintentional poisoning was medicinal drugs, followed by carbon monoxide, and pesticides. While most causes of unintentional poisoning demonstrated a decreasing trend in mortality rate, unintentional poisoning by medicinal drugs showed a 51.9% increase (p < .001) (Figure 3a). Unintentional poisoning by medicinal drugs constituted the leading cause of hospitalization due to unintentional poisoning, followed by venomous animals and plants, and pesticides. In consideration of trends in hospitalization rates, different methods of unintentional poisoning showed varying 10-year trends (Figure 3b).

Poisoning mortality and hospitalization by gender

Our analysis of overall poisoning mortality revealed that men had higher poisoning mortality than women, regardless of intention; this finding is consistent with the 2000 WHO report that indicated a higher global mortality from unintentional poisoning in men [1]. As for hospitalization, we found that men had a higher rate of hospitalization resulting from unintentional poisoning, and women had a higher rate of hospitalization resulting from intentional poisoning; our findings differed somewhat from the previous literature. In the US, both the 2001-2003 health statistics from Colorado [11], and a study on emergency care data from 2001-2004 [12] indicated a higher poisoning hospitalization rate in women than in men, regardless of intention.

Poisoning mortality and hospitalization by age

We found that older adults aged 65 years or older showed the highest poisoning mortality among all age groups, regardless of intention. Corresponding to our finding, the 2000 WHO report also demonstrated that subjects aged 80 years or older worldwide had the highest mortality rate from unintentional poisoning, followed by subjects in the 45-59 and 60-69 year age groups [1]. On the other hand, our study was similar to previous research in a hospital center in Taiwan, which showed that the mortality risk in subjects aged 60 and older was 4.3 times more than in those aged under 60 [13]. However, some studies found different trends from ours. In China, a hospital-based study found that the highest rates of poisoning mortality were in adults aged 30-39 years [14]. In south India, a study of poisoning suicide showed that youths aged 20-29 had the highest mortality rate [15]. Regarding poisoning mortality trends, our study found that the population aged 65 years or older demonstrated a significantly declining trend in mortality; the results were similar to the European study of Petridou et al. [16].

Older adults aged 65 years or older demonstrated the highest poisoning hospitalization rate among all age groups, regardless of intention; this finding was different from previous literature. The 2001-2003 health statistics in Colorado in the US suggested that adolescents aged 15-19 years were at highest risk of hospitalization for intentional poisoning, with 98.1 cases per 100,000 subjects [11]. On the other hand, an epidemiological study using emergency care data from 2001 to 2004 in Sri Lanka revealed that adolescents aged 15-19 years demonstrated the highest incidence of non-lethal intentional poisoning (248 per 100,000 subjects), a rate that decreased with age [17]. However, our study was similar to previous research that showed the elderly aged 65 or older had the highest medicinal drug poisoning hospitalization rate [18]. The phenomenon may be related to the fact that this population is less well educated and has a lower literacy rate, resulting in a lack of knowledge and skills related to handling items safely. In addition, the existence of multiple illnesses, multiple drug medication and poor physical conditions may also increase the risk of hospitalization or mortality as a result of poisoning.

Poisoning mortality and hospitalization by intention

The majority of unintentional poisoning deaths was caused by medicinal drugs, followed by carbon monoxide, and pesticides. The most frequent cause of unintentional poisoning hospitalization was medicinal drugs, followed by venomous animals and plants, and pesticides. A number of previous studies showed similar results that medicinal drugs [11, 19–21] and pesticides [13–15] were the primary causes of poisoning-related mortality and hospitalization. This phenomenon may result from the widespread prescribing and easy availability of medicinal drugs and pesticides, compared with other poisoning methods. Therefore, the following discussion will focus on the long-term trends in poisoning from medicinal drugs and pesticides.

Trends in mortality and hospitalization from poisoning by medicinal drugs

This study revealed a rising trend in mortality and hospitalization rates from unintentional poisoning caused by medicinal drugs. This result was similar to a previous study in Taiwan showing an increase in unintentional drug poisoning mortality rate of 51.9% [18]. Since the launch of the NHI in Taiwan in 1995, the general public has had better access to prescription drugs, such as sleeping pills, sedatives, and analgesics, which may increase the possibility of drug misuse and increase mortality resulting from unintentional drug poisoning [20]. The promulgation of the Drug Harvard Relief Act in Taiwan in 2000 allows timely relief, without resorting to lawsuits or pleads, to victims of unforeseeable adverse reactions from proper use of legal drugs that results in death, disability, or serious illness [21]. Therefore, the launch of the Act may have encouraged reporting of drug poisoning deaths and thus could have increased the death count from unintentional drug poisoning.

Trends in mortality and hospitalization rates from poisoning by pesticides

This study revealed decreasing trends between 1999 and 2008 in mortality and hospitalization rates from unintentional poisoning by pesticides. In Korea from 1996 to 2005, mortality from pesticides rose for both genders [22]. In Central America, there was also an increase in both incidence and mortality rates of pesticide poisoning from 1992 to 2000, which might be related to increased pesticide imports during the same period of time [23]. In Thailand from 1987 to 1996, both injuries and deaths from pesticide poisoning decreased; however, the numbers might be an underestimation because most locals failed to seek medicinal help in the event of poisoning [24]. In Japan from 1991 to 1996, poisoning by pesticides was on the decline, which could result from the decrease in paraquat poisoning [25].

As a result of social transformations in Taiwan in recent years, the area of farmland has been decreasing. The traditional agriculture-centered Taiwanese society depends more and more on business, traditional industry, and high-technology industry, and organic farming has also gradually replaced traditional farming, resulting in a large decrease in pesticide use. Between 1951 and 1992, the proportion of agricultural workers in the Taiwanese population decreased from 56% to 12%, while that of industrial workers rose from 16% to 40% [26]. According to the 2009 statistics from the Council of Agriculture, Executive Yuan, there has been a decreasing trend in the amount of people employed in the agricultural industry, from approximately 710,000 people in 2002 to about 540,000 people in June, 2009 [27]; this trajectory has coincided with the increasing importance attached to environmental protection and food safety issues by the public. Since 1972, a total of 119 agricultural chemicals have been banned due to high toxicity, teratogenicity after long-term exposure, tumorigenicity, and/or environmental pollution [28]. The abovementioned reasons may well account for the decline in pesticide poisoning mortality rate.

Limitations

The variables analyzed in this study were limited by the cause-of-death statistics in Taiwan and the hospitalization data retrieved from the NHI Research Database. Certain information was not available, such as clinical biochemistry data, education level of subjects, smoking habits, and use of alcohol.

Our study was limited with regards to cause-of-death statistics; when registering the cause of death, the Department of Health only records the first three digits of the E-Codes. The absence of the fourth digit hindered us from extracting additional detailed information. For example, we were unsuccessful in differentiating between organic phosphorus pesticides and organic chlorine pesticides in agricultural chemical poisoning cases and were unable to compare the mortality data against the hospitalization data.

After an injury, posttraumatic care can be categorized into no medical care or self-care, outpatient care, emergency care, inpatient care, and death. However, because the outpatient and emergency treatment data provided by the NHI Research Database failed to include E-Codes, we were unable to classify those patients receiving such care by the poisoning method based on the ICD-9-CM system. As a result, only deaths and hospitalized patients for poisoning injuries were analyzed in this study, while those receiving other types of care were excluded.