Abstract
Background:
The mortality rate associated with human glufosinate poisoning is high. We evaluated the usefulness of serum ammonia and sequential organ failure assessment (SOFA) and acute physiology and chronic health evaluation II (APACHE II) scores for early prediction of in-hospital mortality in glufosinate ammonium poisoning.
Methods:
A prospectively collected pesticide poisoning registry at a single academic medical center was retrospectively analyzed. Data were collected from consecutive patients diagnosed with glufosinate ammonium poisoning between May 2007 and February 2018. The initial serum ammonia level was defined as the highest serum ammonia level measured within 12 h after emergency department (ED) arrival. The SOFA and APACHE II scores were calculated using data obtained within the first 12 h after ED arrival. The patients were divided into survivor and nonsurvivor groups by in-hospital death status.
Results:
In total, 110 patients were included. Ten patients (9.1%) died in the hospital despite treatment. Median initial serum ammonia level was significantly higher in the nonsurvivor group than in the survivor group (219 (range: 158–792) versus 100.5 (range: 25–317) µg/dL, p < 0.001). Median SOFA and APACHE II scores in the survivor and nonsurvivor groups were 2 (range: 0–10) versus 5 (range: 1–8) (p = 0.044) and 7 (range: 0–28) versus 16 (range: 8–22) (p = 0.001), respectively. In the multiple logistic regression analysis, the initial serum ammonia level was the only independent predictor (cutoff value: 151 µg/dL).
Conclusion:
An initial serum ammonia level >151 µg/dL was an independent early predictor of in-hospital mortality in glufosinate ammonium poisoning.
Introduction
The use of glufosinate herbicides has gradually increased worldwide. 1 In Korea, commercial products contain 18% glufosinate in the form of ammonium salts, along with anionic surfactants. Glufosinate poisoning has been known to cause severe complications following ingestion, such as decreased mental status, seizures, amnesia, and respiratory failure, 2 –4 and the associated mortality rate in humans is high at 6.1–17.7%. 1 However, clinicians cannot predict which patients are likely to become critically ill in the early stages of treatment. Therefore, the early prediction of in-hospital mortality can guide physicians in the application of optimizing care and efficiently allocating resources from the emergency department (ED). In addition, easily performed laboratory tests or scoring systems that can predict in-hospital mortality can aid clinicians.
A few studies reported that measuring serum ammonia levels at the ED or serial ammonia levels are useful for predicting neurologic complications, such as decreased mental status, seizure, or amnesia in glufosinate poisoning. 5 –7 However, there is no study of whether serum ammonia concentration predicts in-hospital mortality in the early phase of glufosinate poisoning. The sequential organ failure assessment (SOFA) and acute physiology and chronic health evaluation II (APACHE II) scoring systems are tools widely used for predicting the clinical outcomes of patients admitted to the intensive care unit (ICU). 8,9 However, there is also no study of whether the SOFA and APACHE II scoring systems predict in-hospital mortality in the early phase of glufosinate poisoning, although a few studies have assessed the early prediction of mortality in organophosphate or paraquat poisoning using scoring systems including the SOFA and APACHE II scoring systems. 10 –13
We hypothesized that in the early phase of poisoning, serum ammonia levels, as well as SOFA and APACHE II scores, would be high in patients with in-hospital mortality, and these may be used to predict in-hospital mortality in the early stage of glufosinate poisoning. Therefore, we evaluated the usefulness of serum ammonia and the SOFA and APACHE II scoring systems for early prediction of mortality in glufosinate poisoning.
Methods
Study design and population
A prospectively collected pesticide poisoning registry at a single academic medical center was retrospectively analyzed. Data were collected from consecutive patients diagnosed with glufosinate ammonium poisoning between May 2007 and February 2018. The ED was located in a suburban, tertiary-care hospital (Wonju, Republic of Korea), which has >46,000 annual patient visits. Moreover, this hospital employs board-certified emergency physicians who are available 24 h a day. The institutional review board of the Wonju Severance Christian Hospital approved this study (approval number: CR318028) and the study protocol conformed to the ethical guidelines of the Declaration of Helsinki (1975) and its later amendments. Since this was a retrospective study, requirement for informed consent was waived, and patient records and information were anonymized before the analyses.
Emergency physicians verified the ingested agents using the following methods: (1) investigation of the named pesticide if the patients or guardians remembered the pesticide’s name, (2) confirmation of the pesticide classification through a search of the official pesticide website if a bottle of the ingested pesticide was brought to the ED, and (3) confirmation of glufosinate poisoning via analysis of serum, urine, or stomach contents at the Forensic Toxicology Division of the National Forensic Service, Wonju, Republic of Korea since March 2017. All patients with hyperammonemia upon ED arrival received a lactulose enema to lower their ammonia levels based on a serum ammonia level of 86 µg/dL, as high levels have been suggested to cause neurologic complications. 6,7
Some of the participants were excluded based on the following criteria: (1) age below 19 years, (2) poisoning involving other agents except alcohol, (3) transfer to another hospital after ED admission, and (4) insufficient data.
Study variables and data collection
The following variables were obtained for each patient: age, sex, ingested dose (mL), intentionality, period from ingestion to ED arrival (hours), use of gastric lavage and activated charcoal, initial vital signs and Glasgow Coma Scale (GCS) score, shock at the ED, comorbidity (diabetes mellitus, liver disease, chronic kidney disease, cancer, and cerebrovascular disease), early symptoms and complications within 12 h after arriving at the ED (abdominal pain, headache, dizziness, dyspnea, mental change defined as GCS score <13, respiratory failure, seizure, aspiration pneumonia, acute kidney injury, use of ventilator, and shock), and complications (mental change, respiratory failure, shock, aspiration pneumonia, and acute kidney injury) and procedures (use of mechanical ventilator and continuous renal replacement therapy) during hospitalization. We investigated the number of admission days in the ICU as well as in-hospital mortality. We defined the amount of ingested glufosinate as follows: “a little” or “a spoonful” as 5 mL, “a mouthful” as 25 mL, “a small cup” as 100 mL, and “a bottle” as 300 mL. 6 Respiratory failure was defined as hypoxemia (PaO2 <70 mmHg) or hypercapnia (PaCO2 >60 mmHg). Aspiration pneumonia was diagnosed based on the appearance of a new infiltrate on chest imaging combined with either the significant growth of meaningful bacteria, which included all bacteria except for bacterial contaminants in the sputum culture, or accompanied by a body temperature of at least 38°C or lower than 36°C. 14 Acute kidney injury was defined as an increase in serum creatinine levels of ≥0.3 mg/dL within 48 h or an increase of ≥50% within 7 days or a urine output <0.5 mL/kg/h for >6 h. 15 Mechanical ventilation was applied in (1) patients with an unprotected airway and (2) patients with respiratory failure. Shock was defined as a systolic blood pressure <90 mmHg requiring the administration of a vasoactive drug.
Laboratory parameters included serum ammonia levels and pH, HCO3 −, and lactate values in the arterial blood gas analyses. Serum ammonia levels were measured using a Dimension Vista® System (Siemens, Munich, Germany). Venous or arterial blood was collected in a tube containing lithium heparin for measuring ammonia concentrations. The Dimension Vista Ammonia method is an enzymatic method that uses glutamate dehydrogenase and a stabilized nicotinamide adenine dinucleotide phosphate oxidase analog. The serum ammonia level was serially measured in all included patients at least twice more within 12 h after ED arrival, and the initial serum ammonia level was defined as the highest serum ammonia level measured within 12 h after ED arrival. The SOFA and APACHE II scores were calculated using data obtained within the first 12 h after ED arrival. 8,9 The SOFA scoring system is based on six different scores, that is, one for each system (the respiratory, cardiovascular, hepatic, coagulation, renal, and central nervous systems). 8 The APACHE II scoring system uses a point score based on the initial values of 12 routine physiological measurements, patient age, and medical history for assessing disease severity. 9,16 The patients were divided into a survivor group and a nonsurvivor group.
Data analysis
The general characteristics of the participants, laboratory findings, and clinical scoring systems were presented as median (range) for continuous variables and frequencies (percentages) for categorical variables in glufosinate poisoning. Differences between the survivors and nonsurvivors were assessed using an independent Mann–Whitney U test for continuous variables and a χ2 test for categorical variables. Multiple logistic regression was conducted to assess predictors of in-hospital mortality. In addition, receiver operating characteristic (ROC) curves were generated, and the area under the curve (AUC) was calculated to compare the predictive powers of variables. The optimal cutoff value was calculated using the Youden index. All statistical analyses were performed using SPSS version 23 (IBM Inc., Chicago, Illinois, USA) and MedCalc version 18.2 (MedCalc Inc., Mariakerke, Belgium), and the significance level was set at 0.05.
Results
During the study period, 122 consecutive patients with glufosinate poisoning were identified, and 12 were excluded based on the following exclusion criteria: co-ingestion of sedative drugs (5 patients), transfer to another hospital after ED admission (2 patients), and insufficient data (5 patients). Finally, 110 patients were included. Of these, 85 patients were investigated by verifying the name of the pesticide through a search of the official pesticide website; this was done if the patients or guardians remembered the pesticide’s name or if they brought a bottle of the ingested pesticide to the ED. Twenty-five patients were verified at the Forensic Toxicology Division of the National Forensic Service.
Sixty-eight patients (61.8%) were men, and median age for all included patients was 56 (range: 24–89) years. Poisoning was intentional in 90% of patients, and median ingested dose was 27 g. Median time from ingestion to ED admission was 3 h. Four patients had liver disease such as chronic hepatitis (no advanced stage and normal liver function test), five patients had cancer, and there was no liver malignancy. The early serious complications that developed within 12 h were mental changes (56.4%), respiratory failure (9.1%), seizures (36.4%), acute kidney injury (21.8%), and shock (2.7%). During hospitalization, a mechanical ventilator was used in 65 patients (59.1%) and the most common complications were respiratory failure (41.8%) and aspiration pneumonia (33.6%). Ten patients (9.1%) died in the hospital despite treatment (Table 1).
General characteristics, laboratory findings, and clinical scoring systems in patients with glufosinate poisoning.a
ED: emergency department; MAP: mean arterial pressure; PR: pulse rate; RR: respiratory rate; GCS: Glasgow Coma Scale; SOFA: sequential organ failure assessment; APACHE II: acute physiology and chronic health evaluation II; CRRT: continuous renal replacement therapy; ICU: intensive care unit.
aAll variables are expressed as frequency (percentage) or median (range).
bWithin 12 h after arriving at the ED.
A comparison of the patients’ characteristics and laboratory results is shown in Table 1. The patients in each group significantly differed in terms of age; presence of dyspnea, mental changes, seizure, aspiration pneumonia, and acute kidney injury developing within 12 h after arriving at the ED; the use of mechanical ventilation; and all the investigated complications during hospitalization. The median initial serum ammonia level was significantly higher in the nonsurvivor group than in the survivor group (219 (range: 158–792) versus 100.5 (range: 25–317) µg/dL, p < 0.001). The SOFA and APACHE II scores were significantly higher in the nonsurvivor group than in the survivor group. The SOFA scores in the survivor and nonsurvivor groups were 2 (0–10) and 5 (1–8), respectively (p = 0.044). The APACHE II scores in the survivor and nonsurvivor groups were 7 (0–28) and 16 (8–22), respectively (p = 0.001).
Multiple logistic regression analysis was performed after adjustment for age, sex, and the factors that significantly differed between the survivor and nonsurvivor groups among the factors that could be measured within 12 h after ED arrival. The initial serum ammonia level (odds ratio 1.026, 95% confidence interval (CI): 1.003–1.050, p = 0.027) was the only early independent predictor of in-hospital mortality in glufosinate poisoning (Table 2). To evaluate the predictive values of initial serum ammonia levels and of the scoring systems, we evaluated the AUC of the ROC curves. The AUC of initial serum ammonia level (0.911, 95% CI: 0.842–0.957) was higher than that of the SOFA and APACHE II scores (0.691, 95% CI: 0.595–0.775 and 0.813, 95% CI: 0.727–0.881, respectively) (Figure 1). The cutoff value for initial serum ammonia level was 151 µg/dL (sensitivity: 100%, 95% CI: 69.2–100; specificity: 78%, 95% CI: 68.6–85.7; positive predictive value: 31.3%; negative predictive value: 100%).
Early independent predictors of in-hospital mortality in patients with glufosinate poisoning.
SOFA: sequential organ failure assessment; APACHE II: acute physiology and chronic health evaluation II.
aWithin 12 h after arriving at the emergency department.
Receiver operating curves for the prediction of mortality according to the APACHE II and SOFA scoring systems and initial serum ammonia level. The areas under the curve and 95% confidence intervals for these indicators were 0.911 (0.842–0.957) for initial serum ammonia levels, 0.691 (0.595–0.775) for the SOFA score, and 0.813 (0.727–0.881) for the APACHE Ⅱ score. APACHE II: acute physiology and chronic health evaluation II; SOFA: sequential organ failure assessment.
Discussion
In glufosinate ammonium poisoning, glufosinate, solvent, and surfactant may affect the prognoses of patients by inducing complications, including neurological complications. We believe that ammonia in the brain parenchyma is one of the major causes of neurological complications. In addition, we think that blood ammonia levels may reflect ammonia levels in the brain parenchyma. In this study, initial serum ammonia levels were associated with in-hospital mortality in glufosinate poisoning cases. In our previous studies, we reported that serum ammonia levels may predict neurologic complications. 6,7 Neurologic complications compromise the cough reflex and glottic closure and increase the risk of aspiration pneumonia, as well as the risk of ventilator use. In one of those studies, the use of mechanical ventilation, shock, pneumonia, and cardiac arrest were observed only in patients with neurologic complications including mental changes and seizures. 6 The complications that developed, such as pneumonia, shock, respiratory failure, acute kidney injury, or cardiac arrest, can lead to organ dysfunction, which is associated with high morbidity and mortality. 17,18 In the current study, the nonsurvivor group had a significantly higher number of neurologic symptoms and signs (mental changes and seizures) than the survivor group from the early stage of poisoning. As a result, the larger number of neurologic complications led to many other complications and ultimately caused higher mortality among glufosinate poisoning cases. In addition, aspiration pneumonia and acute kidney injury occurred within 12 h after ED arrival. Therefore, it may be reasonable to assume that serum ammonia levels may also have potential as an early predictor of in-hospital mortality. Although advanced liver disease, such as liver cirrhosis and liver malignancy, can affect serum ammonia levels, none of the included patients had advanced liver disease or liver malignancy. Four patients had liver disease, and all four patients were hepatitis B virus carriers. In addition, four patients who were hepatitis B virus carriers showed normal liver function. In addition, there was no significant difference between nonsurvivor and survivor groups in terms of the presence of liver disease. Other blood tests besides serum ammonia have been reported to predict neurologic complications. Although we did not investigate serum S100B, Lee et al. report that S100B was a significant predictor of neurologic complications in patients with glyphosate and glufosinate poisoning. 19 Study that perform serum ammonia and S100B simultaneously may be needed.
In this study, although the SOFA and APACHE II scores were significantly different between the two groups, they were not independent early predictors of in-hospital mortality in glufosinate poisoning. In addition, their predictive powers were lower than that of the initial serum ammonia level (Figure 1). Lee and Kim also found that although the SOFA and APACHE II scores differed significantly between the survivor and nonsurvivor groups, they were not independent early predictors of 90-day mortality in patients with glufosinate poisoning. 20 Our negative results may have arisen from the investigation of the SOFA and APACHE II scores within the first 12 h after ED arrival. APACHE II scores are generally used within 12 h after ICU admission 9,16 and SOFA scores are usually used to monitor a person’s status during ICU admission to determine the extent of their organ dysfunction. 8 In addition, the two scoring systems, especially the APACHE II scoring system, involve complicated calculations, unlike simple serum ammonia measurements.
In the present study, the nonsurvivor and survivor groups differed in terms of age, dyspnea, mental changes, seizures, aspiration pneumonia, and acute kidney injury. Older patients may more frequently have serious symptoms and complications due to a lack of physiological functional capacity to overcome toxicity. Therefore, clinicians should keep in mind that older patients are more likely to develop serious symptoms and in-hospital mortality.
The prevalence of aspiration pneumonia (33.6%) in this study appeared to be much higher than that (11.3%) reported by Mao et al. in 2012. 1 It may be reasonable to speculate that the higher prevalence of aspiration pneumonia observed in this study may be partially attributable to inadequate airway protection during the treatment with charcoal because the use of activated charcoal was quite common in this study (87.3% among all patients). It is a problem that some local hospitals in Korea often use charcoal in poisoned patients regardless of whether airway maintenance has been ensured.
This study had several limitations. First, the study was limited by its small sample size owing to the single-center design. Therefore, we could not adjust for many confounding factors in the multiple logistic regression analysis. However, we evaluated all patients with glufosinate poisoning to reduce possible bias during the study period (10 years and 9 months). Second, we could not investigate the exact content of diets and drugs being taken that may have affected serum ammonia levels because this was a retrospective study. Third, the variations in the time intervals from ingestion to ED admission (0.3–24 h) may limit the interpretation of our results. However, there was no significant difference between the survivor group and nonsurvivor group in the time interval from ingestion to ED admission. Fourth, owing to the long study period, the possibility of treatment modalities having been changed in the hospital cannot be ruled out. However, there were no significant differences in the use of specific modalities such as continuous renal replacement therapy, vasopressor administration, and extracorporeal membrane oxygenation. Fifth, before March 2017, we did not measure blood levels and screen for toxicants other than glufosinate through the Forensic Toxicology Division of the National Forensic Service and we relied only on the history given by patients and the bottle labels. Further studies are necessary to clarify these issues.
Conclusions
An initial serum ammonia level >151 µg/dL was an independent early predictor of in-hospital mortality in glufosinate poisoning. Therefore, initial serum ammonia levels may be adjunctive markers for predicting in-hospital mortality in glufosinate poisoning.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a Grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health and Welfare, Republic of Korea (Grant No. HI18C2196) and by a National Research Foundation of Korea (NRF) Grant funded by the Korea government (Ministry of Science, ICT and Future Planning) (Grant No. NRF-2017R1C1B5017695).