Poisoning deaths in China, 2006–2016

Abstract Objective To provide a comprehensive overview of poisoning mortality patterns in China. Methods Using mortality data from the Chinese national disease surveillance points system, we examined trends in poisoning mortality by intent and substance from 2006 to 2016. Differences over time between urban and rural residents among different age groups and across external causes of poisoning were quantified using negative binomial models for males and females separately. Results In 2016, there were 4936 poisoning deaths in a sample of 84 060 559 people (5.9 per 100 000 people; 95% confidence interval: 5.6–6.2). Age-adjusted poisoning mortality dropped from 9.2 to 5.4 per 100 000 people between 2006 and 2016. Males, rural residents and older adults consistently had higher poisoning mortality than females, urban residents and children or young adults. Most pesticide-related deaths (34 996 out of 39 813) were suicides among persons older than 15 years, although such suicides decreased between 2006 and 2016 (from 6.1 per 100 000 people to 3.6 for males and from 5.8 to 3.0 for females). In 2016, alcohol caused 29.3% (600/2050) of unintentional poisoning deaths in men aged 25–64 years. During the study period, unintentional fatal drug poisoning by narcotics and psychodysleptics in individuals aged 25–44 years increased from 0.4 per 100 000 people to 0.7 for males and from 0.05 to 0.13 for females. Conclusion Despite substantial decreases in mortality, poisoning is still a public health threat in China. This warrants further research to explore causative factors and to develop and implement interventions targeting at-risk populations.


Introduction
According to the Global Burden of Disease 2015 update, approximately 86 353 people died from unintentional poisonings worldwide in 2015, with 78 054 (90%) deaths occurring in low-and middle-income countries. 1 However, if the global estimates of the numbers of intentional poisonings were freely accessible, the poisoning mortality numbers would be higher. 1 Despite the implications for public health and the impact described in high-income countries, 2-10 poisonings in low-and middle-income countries are poorly understood. We are aware of just two published studies on the epidemiology of poisoning at the national level in low-and middle-income countries: from Fiji and the Islamic Republic of Iran. 11,12 As the most populated country in the world, China had 16 179 unintentional poisoning deaths in 2016, 31% of the world's total of 52 077. 1 Available knowledge about poisoning incidents in China is scattered. A study using Global Burden of Disease 2015 update data examined trends in unintentional poisoning deaths. 13 The authors reported a substantial reduction in unintentional poisoning mortality from 1990 to 2015. A second study used the Chinese disease surveillance points system data and reported decreases in suicide by poisoning from 2006 to 2013. 14 The authors also reported that suicide by pesticide poisoning was the leading method of suicide. Other published studies have examined poisoning patterns within one hospital catchment area or one province 15,16 or focused on the epidemiology of a single cause of poisoning, such as pesticide poisoning. 17 We found no recent, comprehensive published studies of the epidemiology of fatal poisonings in China. To address these gaps, we used national disease surveillance data to examine changes in poisoning mortality from 2006 to 2016 by loca-tion (urban or rural), age group, intent and type of substance. Analyses were conducted separately for males and females.

Data source
We designed a population-based longitudinal study based on the data from the national disease surveillance points system, initiated in 1978 by the Chinese government. The surveillance system has undergone three major improvements since 1978. First, it was expanded from 145 to 161 points in [2004][2005][2006], yielding coverage of about 73 million residents. 18 Second, a web-based approach was introduced to report deaths in 2008, a development that greatly improved the timeliness of data reporting. 18 Third, the Chinese government combined the system with the national vital registration system in 2013, creating a data collection system from 605 surveillance points. 18 The population of China in the most recent census in 2010 was around 1332 million (682 million males and 650 million females).
The data for the disease surveillance points system are collected using a standard protocol by trained persons. 18 Trained staff members oversee data collection of all deaths occurring in the hospital. For deaths occurring outside the hospital, village health workers or community hospital professionals use verbal autopsy strategies to collect the relevant data. Local centres for disease prevention and control report all data to their next-level office (from county to prefectural to provincial to national) and routine quality checks are conducted by coders at each centre. Quality checks assess completeness, coding and internal logic across items reported on death certificates. 18 Any unqualified reports that are detected Objective To provide a comprehensive overview of poisoning mortality patterns in China. Methods Using mortality data from the Chinese national disease surveillance points system, we examined trends in poisoning mortality by intent and substance from 2006 to 2016. Differences over time between urban and rural residents among different age groups and across external causes of poisoning were quantified using negative binomial models for males and females separately. Results In 2016, there were 4936 poisoning deaths in a sample of 84 060 559 people (5.9 per 100 000 people; 95% confidence interval: 5.6-6.2). Age-adjusted poisoning mortality dropped from 9.2 to 5.4 per 100 000 people between 2006 and 2016. Males, rural residents and older adults consistently had higher poisoning mortality than females, urban residents and children or young adults. Most pesticide-related deaths (34 996 out of 39 813) were suicides among persons older than 15 years, although such suicides decreased between 2006 and 2016 (from 6.1 per 100 000 people to 3.6 for males and from 5.8 to 3.0 for females). In 2016, alcohol caused 29.3% (600/2050) of unintentional poisoning deaths in men aged 25-64 years. During the study period, unintentional fatal drug poisoning by narcotics and psychodysleptics in individuals aged 25-44 years increased from 0.4 per 100 000 people to 0.7 for males and from 0.05 to 0.13 for females. Conclusion Despite substantial decreases in mortality, poisoning is still a public health threat in China. This warrants further research to explore causative factors and to develop and implement interventions targeting at-risk populations.

Data analysis
We considered three demographic factors in the analyses: location (urban or rural), age group and year. Based on preliminary analysis and human development theory, we classified age into six groups: 0-4, 5-14, 15-24, 25-44, 45-64 and ≥ 65 years.
We calculated age-adjusted mortality rates and 95% confidence intervals (CI), using the 2010 population of China as the reference. To overcome over-dispersion of count data 22 we ran univariate negative binomial regression to quantify poisoning mortality changes between 2006 and 2016, using percentage change in mortality rate and its 95% CI. We first calculated the mortality rate ratio by dividing the mortality rate for 2016 with the corresponding rate of 2006, and then calculated the percentage change as: (mortality rate ratio−1) × 100.
We found that the location-specific analysis showed similar changes to that in overall poisoning mortality and that multistrata subgroup analysis yielded unreliable subgroup mortality rates for many groups due to the small numerators (deaths < 20). 23 We therefore analysed age-specific fatal poisonings by intent and type of substance.
We conducted all analyses separately for males and females. Statistical analyses were completed using Stata version 12.1 (StataCorp LLC, College Station, United States of America; USA).

Overall trends
Between 2006 and 2016, 67 713 poisoning deaths were reported to the disease surveillance points system (Table 1 and  Table 2 (Table 3; Table 4). Males consistently had higher poisoning mortality than females across the study period (male to female ratio range: 1.4-1.8). Poisoning mortality per 100 000 people decreased by 36% (from 10.7 to 6.9) for males and 50% (from 7.6 to 3.8) for females between 2006 and 2016.
Across the study time period, poisoning mortality in rural areas was 2.1-2.8 times greater than in urban areas. Male poisoning mortality per 100 000 people decreased over 2006-2016 by 30% (from 13.7 to 9.6) for rural and 38% (from 6.3 to 3.9) for urban residents. Female poisoning mortality per 100 000 decreased by 47% (from 9.6 to 5.1) for rural and 50% (from 4.6 to 2.3) for urban residents.
Poisoning mortality generally rose as age increased. Substantial reductions in mortality occurred over 2006-2016 in all sex-and age-specific groups except for children aged 0-4 years.
Among males, pesticides and alcohol were the most common substances involved, accounting for 48% (1516 of 3175) and 22% (711 of 3175) of poisoning deaths in 2016, respectively. Deaths per 100 000 people due to poisoning by pesticides and alcohol among males decreased by 44% (from 5.9 to 3.3) and 20% (from 2.0 to 1.6) respectively over the study period. Among females, pesticides were the most commonly used substances in poisoning deaths, accounting for 1209 of 1761 (69%) deaths in 2016. Deaths from poisoning by pesticide among females decreased by 54% (from 5.6 to 2.6 per 100 000) between 2006 and 2016.
Suicidal poisoning mortality per 100 000 people decreased by 50% (from 6.3 to 3.1) in males and 57% (from 5.8 to 2.5) in females. This contrasts with a non-significant decrease in unintentional fatal poisonings in males (5%, from 3.8 to 3.6) and in females (7%, from 1.4 to 1.3).

Location-and sex-specific trends
Analysis by location showed that mortality in all subgroups by intent and by type of substance were higher in rural than urban areas. Similar patterns of change over time were observed for urban and rural areas (data available from the corresponding author).
Sex-specific analysis demonstrated that males had higher subgroup poisoning mortality (by intent and by type of substance) than females, particularly for alcohol-related poisoning (data available from the corresponding author). The patterns of change from 2006 to 2016 by intent and by type of substance were generally similar among males and females.

Age-specific trends
Subgroup analyses by age group and intent show great variations across age groups (Fig. 1). Unintentional poisoning was the leading intent in children      younger than 5 years, accounting for 40 of 42 (95%) poisonings for both sexes in 2016. Suicidal poisoning was the leading cause of fatal poisonings in individuals older than 24 years, and was especially prominent in the oldest individuals, constituting 59% and 68% of poisonings in males and females aged ≥ 65 years in 2016. Suicide poisoning mortality decreased substantially in most age groups for both males and females (data available from the corresponding author). Subgroup analyses by age group and type of substance demonstrated inconsistent changes over time from 2006 to 2016 (Fig. 2). For both sexes, pesticides, other gases and vapours, and drugs were the most common substances causing poisoning in children younger than 5 years. Among all other age groups, pesticide poisoning was most common. Alcohol poisoning deaths also occurred frequently among males aged 25-44 years and 45-64 years, accounting for 203 of 796 (26%) and 397 of 1254 (32%) deaths in 2016, respectively.
Analysis of changes over time found that decreases in pesticide poisoning mortality ranged from 38% to 66% for males and from 43% to 72% for females in all adolescent and adult age groups. Drug poisoning mortality per 100 000 people decreased by 11% (from 0.9 to 0.8) and 50% (from 2.8 to 1.4) in males and 43% (from 0.7 to 0.4) and 63% (from 2.4 to 0.9) in females in age group 45-64 years and ≥ 65 years respectively. Alcohol poisoning decreased in two male age groups: 25-44 years (from 2.2 to 1.6) and ≥ 65 years (from 4.2 to 2.3; data available from the corresponding author).
Subgroup analysis examining the combination of intent and type of drugs revealed various age-specific patterns in drug poisoning (Fig. 3). For both sexes, drug poisonings were rare in young children (Fig. 2). In 2016, for example, mortality per 100 000 in children was 0.17 for ages 0-4 years and 0.04 for ages 5-14 years. Deaths were primarily due to unintentional and suicidal poisonings by other drugs in age group 5-14 years. In males, unintentional poisoning by narcotics and psychodysleptics was the leading cause of drug poisoning for age groups 15-24 years and 25-44 years. Unintentional poisoning by narcotics and psychodysleptics or by other drugs and suicide poisoning by other drugs were most frequent in males aged 45-64 years and ≥ 65 years. Unintentional and suicide poisonings by other drugs were most common in females aged 45-64 years and ≥ 65 years. Changes in drug poisoning mortality varied greatly across sex-and age-specific groups (data are available from the corresponding author).
Subgroup analysis by both intent and type of substance revealed similar patterns of pesticide and alcohol poisoning deaths across the four adolescent and adult age groups (Fig. 4). For both sexes, unintentional poisoning by pesticide and other gases and vapours were most common in age group 0-4 years. Poisoning mortality varied differently between subgroups over the study period. Despite small numbers, unintentional poisoning by alcohol increased 217% in females aged 45-64 years from 0.06 to 0.19 per 100 000 over 2006-2016.

Discussion
We generated five major findings. First, age-adjusted poisoning mortality dropped from 9.15 to 5.40 per 100 000 people between 2006 and 2016 in China. The reduction was present in both sexes and both urban and rural areas. Second, a reduction in suicidal poisoning by pesticides was the primary driver of recent decreases in overall poisoning mortality. Third, changes in overall poisoning mortality varied greatly across age groups in the analyses of particular intents and substances and patterns of change over time. Fourth, unintentional poisonings from alcohol represented a major subcategory of fatal poisonings for males ages 25-44 years and 45-64 years, and did not show downward trends in frequency over time. Fifth, despite the small numbers of deaths, large increases in unintentional drug poisoning by narcotics and psychodysleptics were found for individuals ages 25-44 years. We discuss each point below.
The study replicates the findings in two previous publications from China, 13,14 and extends previous work. We found that the largest decreases in poisoning mortality in China occurred in unintentional poisoning and suicide by pesticide. The 2016 crude poisoning mortality in China (7.4 per 100 000 people in males and 4.3 in females) is lower than previous reports for northern European countries (ranging from 7.9 to 22.4 per 100 000 people) 6,8 and the USA (16.2 in per 100 000 people). 2 Variations across countries may reflect cultural differences, including societal reactions to suicide and suicide risk, management of lethal and poisonous substances, living and work environments and the reporting quality of poisoning mortality data (such as under-reporting due to the stigma attached to suicide). 24,25 Consistent with previous studies both in China 14 and elsewhere, 2 our study reports higher poisoning mortality risk in males, rural residents and older adults. Higher risk among males could result from a combination of a greater tendency towards risk-taking and greater opportunity for exposure to poisoning, including being more likely to take on high-risk jobs and having higher alcohol consumption rates. 26,27 Similarly, rural residents have greater access and exposure to poisonous substances such as pesticides. Greater fatality rates in rural areas of China may be also due to relatively underdeveloped prehospital aid and hospital treatment services for poisoning cases compared with those of urban cities. 28 Older adults may face greater unintentional poisoning risk because of poorer physical health (such as weakened eyesight or sense of smell) and greater suicide risk from depressive disorders in older age. 1 As reported in other Asian countries, 24,29,30 suicide by pesticide was the most common cause of fatal poisonings in China. This may result primarily from easy access to pesticides that are used in agricultural production. Unlike the situation in high-income countries, where there is more mechanization and larger-scale farming, rural populations in low-and middle-income countries farm on small plots and have immediate access to pesticides. 31 Encouragingly, suicidal poisoning by pesticides demonstrated a distinct decreasing trend from 2006 to 2016, which was an important component of the declining overall trend in suicides for China over that time period. 32 The decrease likely reflects the effect of rapid urbanization that caused many citizens to migrate from rural to urban areas, reducing exposure and access to lethal pesticides. 33,34 It may also reflect the impact of national changes in regulations on the production, circulation and sale of lethal pesticides. 35,36 Social and economic development has improved living standards across both urban and rural China, 32 reducing small-household farming practices. National development has also improved the capacity of mental health services, 37    Notes: All age-and drug-specific poisoning mortality rates were presented to show the changing pattern although most rates were unstable because of small numerators < 20 deaths. Mortality for overall and subgroup poisoning rates were age-adjusted using the population of China in 2010.

Poisoning mortality in China
Lijun Wang et al. reduce suicides associated with chronic or acute mental illness. Older Chinese adults are less likely to migrate into urban areas and the number of older adults living in rural areas is actually increasing relative to previous years because of increased lifespans. Prevention efforts to reduce pesticide poisoning in rural China have been piloted, such as household lock-boxes and community education, but have so far been implemented in only a few areas. 38 Poisoning deaths by pesticide and other gases and vapours among children aged 0-4 and 5-14 years remained relatively stable, a finding that deserves attention from researchers and policy-makers. Young children are full of curiosity, traits essential to learning and growth, but may lack the knowledge to judge the risks from lethal substances. Subgroup analysis by ICD-10 code showed that unintentional pesticide ingestions, children failing to recognize ingestion risks, was a major cause of unintentional poisoning by pesticide. Deaths from gases and vapours were primarily from carbon monoxide exposure. A study in Wuhan city reported multiple causes of carbon monoxide poisoning, including firerelated incidents, gas leaks, liquefied gas, gas-related poisoning while showering (from natural gas heaters) and coal or charcoal burning. 39 Many of these risks can be prevented through established prevention strategies. 40 Alcohol poisoning declined steeply between 2006 and 2016 among Chinese men older than 65 years. This success is probably related to prevention efforts to reduce harmful use of alcohol, in-cluding introduction of strict penalties for drink-driving since 2007 40,41 and advocating moderate drinking as a component of healthy lifestyles. 42 Between 2006 and 2016, we observed overall increases in unintentional drug poisoning by narcotics and hallucinogens among Chinese men aged 25-64 years and Chinese women aged 25-44 years, a trend that replicates concerns elsewhere in the world. 43,44 These results may indicate the growing use of narcotics and psychodysleptics through both legal and illegal channels to treat pain and other conditions, as well as the rising production of fentanyl in China. All-age prevalence rates increased from 112.5 to 124.2 per 1000 persons for low back neck pain and from 0.76 to 0.80 per 1000 persons for drug use disorders between 2005 and 2015 in China. 1 China is a primary producer of non-pharmaceutical fentanyl, 45 creating potential for easy access among Chinese citizens.
Our analyses were limited by the lack of data on non-fatal poisoning and relevant risk factors such as access to lethal poisonous substances for vulnerable children and older adults, sale of prescription drugs, alcohol drinking behaviours and enforcement of relevant laws on pesticides, alcohol and prescription drugs. Without these data, causal inferences cannot be made. Additionally, our results are affected by the reporting quality of the disease surveillance data, including completeness, validity of classification and stability. A previous study reported that a small proportion of suicide deaths (5%), including those by poisoning, were misclassified as other specified injuries or injuries with unknown external causes. 46 In addition, the introduction of a web-based reporting system in 2008 caused a slight fluctuation in reported mortality rates in the disease surveillance data set around 2008. 47 Our findings have at least two policy implications. First, the findings offer evidence that some recent poisoning prevention policies and interventions may have achieved success. For example, new policies related to pesticide production, sale and storage may have contributed to a reduction in suicidal poisoning from pesticides. Similarly, there have been efforts to reduce harmful use of alcohol, 42 including efforts to alter traditional attitudes of drinking alcohol as a way to enhance social connections, 26 efforts to standardize and manage advertising of alcoholic beverages in the mass media, 48 and the introduction of more severe drinking and driving laws. 49 These may have contributed to a reduction in fatal alcohol poisoning incidents. Policy efforts should be continued.