2018 年禁用巴拉刈對農藥自殺的初期影響評估： 一所中臺灣主要區域醫院之病歷回溯研究

Abstract

背景和目標：農藥自殺占全球自殺人數的14-20%，在我國則位居自殺人數排名第四的方式。「以固體或液體自殺」死亡的人數為雲林縣所有自殺方式的第一位，其中多數為農藥自殺個案。巴拉刈中毒的致死率較其他農藥高出許多，致死人數約佔台灣農藥自殺身亡者的一半。台灣政府採行兩階段的巴拉刈禁用，於2018年2月1日起禁止進口和製造、2020年2月1日起禁止銷售和使用。本研究的目標包括探討（一）曾在雲林縣一所主要區域醫院接受治療的農藥自殺個案的特徵、使用農藥種類；（二）使用不同種類農藥自殺的個案觸發事件和精神病理的差異；（三）第一階段禁用巴拉刈後使用於農藥自殺的農藥種類、預後（插管和死亡）的變化、（四）農藥種類和預後的相關性。 方法：本研究回溯收集2009至2018年因農藥自殺而在中台灣一所區域醫院接受治療的個案，從病歷中萃取人口學、中毒農藥、觸發事件、精神科診斷、預後等資訊。中毒農藥分為六組，包括固殺草、嘉磷塞、巴拉刈、有機磷、其他農藥、不明農藥。本研究使用Chi-Square test和Fisher’s exact test比較禁用巴拉刈前後的預後和用於自殺的農藥種類，使用羅吉斯迴歸分析探討用於自殺的農藥種類和預後的相關。 結果：本研究共收集304位個案（禁用巴拉刈前286位、禁用巴拉刈後18位）。最常使用於自殺的農藥為除草劑和殺蟲劑。巴拉刈、嘉磷塞、固殺草為最常使用的除草劑。少數個案在農藥自殺前有人際衝突或併用酒精，精神疾病是最常見的觸發因子。在166位接受精神科會診的個案中，使用不同種類農藥自殺者其人口學特徵、自殺前的人際衝突、精神科診斷、精神科就醫史、自殺史無顯著差異。禁用巴拉刈後，使用巴拉刈自殺的比例從30.1%減少為11.1%（p=0.085），而固殺草和嘉磷塞則分別從4.2%增加為22.2%（p=0.001）、從13.6%增加為33.3%（p=0.022）。整體致死率從28.0%減少為16.7%（p=0.42）。排除使用不明農藥自殺的個案後，和其他農藥相比，巴拉刈與較高的死亡率顯著相關（校正後勝算比827.34，95%信賴區間65.30-10481.71）。相對地，固殺草、有機磷則和較高的插管率有關 （固殺草校正後勝算比4.91，95%信賴區間1.60-15.08；有機磷校正後勝算比2.27，95%信賴區間1.13-4.54），但和較高的死亡率無顯著相關 （固殺草校正後勝算比0.21，95%信賴區間0.02-2.05；有機磷校正後勝算比0.05，95%信賴區間0.01-0.45）。使用固殺草與嘉磷塞自殺的插管發生時間較巴拉刈和有機磷晚，且插管持續時間較長，但統計檢定無顯著差異。 結論：使用不同農藥自殺的個案有相似的人口學背景、精神病理、自殺前的人際衝突，然而使用不同種類農藥的預後（插管和死亡）卻有顯著差異。巴拉刈禁用似乎初步在台灣有減少致死率的效果。禁用高致命性的農藥，同時提供其他農藥中毒併發症（如呼吸衰竭）的相應治療，將能避免許多農藥自殺的死亡。

Background and objectives: Pesticide ingestion accounts for 14-20% of global suicide and ranks the fourth most common method of suicide in Taiwan. Poisoning by solid/liquid substances is the most common method of suicide in Yunlin County, Taiwan, and pesticides are the most commonly used poisons for suicide. Paraquat poisoning shows very high fatality rate compared with other pesticides and accounts for approximately half of all pesticide suicides in Taiwan. Taiwanese government banned paraquat in two stages - the ban on import and production became effective from 1st February 2018 and the ban on sale and use from 1st February 2020. We aimed to investigate i) the characteristics of patients who attempted suicide by pesticide poisoning and were treated at a major regional hospital in Yunlin County as well as the types of pesticides ingested; ii) the triggering events and psychopathology of these patients; iii) the changes in pesticides ingested and clinical outcomes of pesticide self-poisoning (intubation and death) after the first stage of paraquat ban; and iv) the associations between the types of pesticides ingested and clinical outcomes. Methods: We collected data of patients who ingested pesticides and were treated at a major regional hospital in Middle Taiwan between 2009-2018. Data for demographic characteristics, pesticides ingested, triggering events of suicide attempt, psychiatric diagnoses, and clinical outcomes were extracted from hospital records. Pesticides were grouped into six categories, namely glufosinate-ammonium, glyphosate, paraquat, organophosphates, others, and unknown pesticides. Chi-square test and Fisher’s exact test were used to compare pesticides ingested and outcomes before and after the paraquat ban. Logistic regression analyses were conducted to investigate the associations between the types of pesticides ingested and clinical outcomes. Results: A total of 304 patients (pre-ban n=286; post-ban n=18) were identified. The most commonly used pesticides were herbicides and insecticides. Paraquat, glyphosate, and glufosinate-ammonium were the mostly frequently herbicides ingested. A minority of patients had interpersonal conflicts before pesticide self-poisoning or had concomitant alcohol use. A psychiatric disorder was the most common precipitant of self-poisoning. Amongst patients who received psychiatric assessment (n=166), there was no difference in demographic characteristics, the presence of interpersonal conflict before pesticide self-poisoning, psychiatric diagnoses, or the past history of psychiatric treatment and suicide attempt across the groups of patients who ingested different types of pesticides. The proportion of paraquat ingestion decreased from 30.1% to 11.1% after the paraquat ban (p=0.085), whilst that of glufosinate-ammonium and organophosphates ingestion increased from 4.2% to 22.2% (p=0.001) and from 13.6% to 33.3% (p=0.022), respectively. Overall fatality decreased from 28.0% to 16.7% (p=0.42). Compared to other four categories of pesticides (the unknown category excluded), paraquat was associated with a markedly increased mortality risk (adjusted odds ratio [aOR]= 827.34, 95% confidence interval [CI] 65.30-10481.71). By contrast, glufosinate-ammonium and organophosphates were associated with an increased risk of intubation (aOR=4.91 [95% CI 1.60-15.08] and 2.27 [95% CI 1.13-4.54], respectively) but not mortality (aOR=0.21 [95% CI 0.02-2.05] and 0.05 [95% CI 0.01-0.45]). Glufosinate-ammonium and glyphosate poisoning showed a longer interval between self-poisoning and intubation (compared with paraquat and organophosphate) and a longer duration of intubation, but the differences did not achieve statistical significance. Conclusion: Patients ingested different types of pesticides had similar demographic and psychiatric profiles and the percentage of interpersonal conflict before pesticide ingestion. However, clinical outcomes (intubation and death) following the ingestion of different types of pesticides varied. Paraquat ban appeared to show an initial effect of reducing pesticide poisoning fatality in Taiwan. Bans on highly lethal pesticides while providing adequate treatment for adverse outcomes such as respiratory failure associated with alternative pesticides ingested could prevent many pesticide suicides.