External validation of the prognostic index in acute paraquat poisoning

Abstract

Objective:

Some studies have evaluated the prognostic indicators associated with acute paraquat (PQ) poisoning. In this study, we externally validated the Yamaguchi index, which showed a good prognostic relevance in predicting the outcome of PQ poisoning.

Methods:

A retrospective analysis of 297 patients was performed. The Yamaguchi index was calculated using the following equation: Eq1 = (K⁺ × HCO3⁻)/(Creatinine × 0.088)(mEq/L) against time from PQ ingestion (T). The patients were divided into three groups: group A: Eq1 > 1500 − 399 × log T, group B: 930 − 399 × log T < Eq1 ≤ 1500 − 399 × log T, and group C: Eq1 ≤ 930 − 399 × log T).

Results:

The overall mortality rate was 65.3% (194 of 297). The mortality rates of the three groups stratified by the Yamaguchi index were 7.1% (2 of 28), 22.4% (15 of 67), and 87.6% (177 of 202). The area under the receiver–operating characteristic curve for predicting mortality from the external validation of the Yamaguchi index was 0.842 (95% confidence interval: 0.795–0.882).

Conclusion:

The Yamaguchi index is a reliable prognostic factor and could be helpful in predicting mortality due to PQ poisoning.

Keywords

Emergency department validation outcome herbicide

Introduction

Paraquat (1,1′-dimethyl-4,4′-bipyridylium dichloride; PQ) is an organic nitrogen heterocyclic contact herbicide that is widely used in agriculture. PQ is not harmful to the environment due to its rapid deactivation after absorption in soil,¹ but it is highly toxic to humans when ingested. Suicide by PQ poisoning is a major concern, and the mortality rate due to PQ poisoning is high (60–80%).²

Several clinical prognostic markers and tests that have been used to evaluate the severity of PQ poisoning, including plasma PQ concentration, arterial lactate, lactate metabolic clearance, the Acute Physiology and Chronic Health Evaluation II score, the Sequential Organ Failure Assessment score, and the severity index of PQ poisoning.^3
–5

Similarly, Yamaguchi et al. suggested an index calculated from the levels of serum creatinine, potassium, arterial blood bicarbonate, and time from PQ ingestion to arrival at the hospital.⁶ Using this index, patients were divided into three categories with estimated probabilities of death of 10, 42, and 97%. In this study, we externally validated the Yamaguchi index for the prediction of hospital mortality.

Methods

Ethics statement

This study followed the statements of the Declaration of Helsinki and was approved by the institutional review board of Gyeongsang National University Hospital, with an informed consent exemption.

Study setting

We conducted a retrospective analysis of patients who were admitted to a single emergency department (ED) with PQ poisoning. This database was originally derived from our previous study, which showed that the absolute lymphocyte count was a prognostic factor for PQ poisoning.⁷ After that study, we collected more data before and after the beginning of the previous study.

This study was conducted between January 2006 and March 2014 in an 800-bed academic hospital with an annual ED census of 30,000 patients. We collected data from ED patients with a diagnosis of PQ poisoning who were at least 18 years of age. PQ poisoning was diagnosed from a clinical history of intentional ingestion of PQ and confirmation by a urine test. The exclusion criteria for this analysis included cases transferred to other hospitals, patients discharged against medical advice, prehospital cardiac arrest, and unintentional poisoning. Patients were also excluded if they had severe underlying diseases including malignancies and chronic kidney or liver diseases. Urine PQ analysis was conducted semiquantitatively using a sodium dithionite method. Detoxification of PQ poisoning patients was performed by gastric lavage with normal saline followed by Fuller’s earth or activated charcoal. Charcoal hemoperfusion was also conducted for detoxification after obtaining informed consent from the patient or his/her immediate family.

Data collection

All data were collected from medical records during the study period. Demographic variables such as age and gender were collected. The time interval between PQ ingestion and ED admission was evaluated. The initial vital signs and laboratory data, including arterial pH, partial pressure of arterial carbon dioxide (PaCO₂), partial pressure of arterial oxygen (PaO₂), bicarbonate $({H C O}_{3}^{-})$ , white blood cell (WBC) count, hematocrit concentration, platelet count, and levels of blood urea nitrogen (BUN), creatinine, sodium, potassium, aspartate aminotransferase (AST), and alanine aminotransferase (ALT), were collected. The primary end point of this study was mortality before hospital discharge.

External validation of the Yamaguchi index

Yamaguchi et al. suggested an equation using the levels of serum creatinine, potassium, arterial blood ${H C O}_{3}^{-}$ , and the time interval from PQ ingestion.⁶ The formula is defined as: Eq1 = (K⁺ × HCO3⁻)/(Creatinine × 0.088)(mEq/L). Patients were divided into three groups: group A, Eq1 > 1500 − 399 × log T; group B, 930 − 399 × log T < Eq1 ≤ 1500 − 399 × log T; and group C, Eq1 ≤ 930 − 399 × log T; and T is time from PQ ingestion to ED admission. Furthermore, the probability of death in each group was calculated and compared.

Statistical analysis

Continuous data were expressed as the mean ± standard deviation. The normality of the distribution was determined using the Kolmogorov–Smirnov test. Student’s t-test and Mann–Whitney U tests were performed to analyze normally and non-normally distributed data, respectively. Categorical data were also summarized in percentage according to the frequencies of occurrences and compared using a χ ² test. Receiver–operator characteristic (ROC) curves were constructed, and the areas under the ROC curves (AUC) were calculated together with a 95% confidence interval (CI) for mortality. The p values <0.05 were considered statistically significant. All analyses were performed using SPSS 21.0 (SPSS, Inc., Chicago, Illinois, USA).

Results

During the study period, a total of 393 cases were screened, of which 96 were excluded. The final analysis included 297 cases (Figure 1). The baseline demographics, clinical, and laboratory data for the survivors and nonsurvivors are described in Table 1. The mean age of the patients was 57.8 ± 16.8 years, 196 (66.0%) of whom were male, and the overall mortality rate of the patients was 65.3% (194 of 297). Nonsurvivors were older and had elevated heart and respiratory rates compared to survivors. Furthermore, the arterial blood pH, PaCO₂, ${H C O}_{3}^{-}$ , and the serum potassium were significantly lower in nonsurvivors. The WBC count, hematocrit, platelet count, serum BUN, creatinine, AST, and ALT were higher in nonsurvivors than that in survivors.

Figure 1.

Flowchart describing the study patients.

Table 1.

Baseline characteristics and laboratory findings of the patients.^a

	Total (n = 297)	Survivors (n = 103)	Nonsurvivors (n = 194)	p Value
Age (year)	57.8 ± 16.8	48.8 ± 15.6	62.6 ± 15.1	<0.001
Male gender, n	196 (66.0)	59 (57.3)	137 (70.6)	0.244
Time from ingestion to ED (h)	5.5 ± 13.6	6.1 ± 9.4	5.2 ± 15.4	0.126
Initial vital signs
MAP (mm Hg)	94.4 ± 19.7	93.6 ± 15.9	94.8 ± 21.5	0.582
Heart rate (beats/min)	89.7 ± 19.6	86.0 ± 15.6	91.7 ± 21.2	0.044
Respiratory rate (breaths/min)	22.3 ± 6.4	19.9 ± 4.4	23.5 ± 6.9	<0.001
Laboratory findings
pH	7.3 ± 0.1	7.4 ± 0.1	7.3 ± 0.1	<0.001
PaCO₂ (mm Hg)	29.0 ± 9.5	33.1 ± 7.9	26.9 ± 9.6	<0.001
PaO₂ (mm Hg)	95.0 ± 36.0	93.3 ± 23.8	95.9 ± 41.1	0.826
${H C O}_{3}^{-}$ (mmol/L)	15.5 ± 5.7	19.8 ± 4.5	13.2 ± 4.9	<0.001
WBC (×10³/mm³)	16.1 ± 8.3	11.8 ± 5.2	18.3 ± 8.7	<0.001
Hematocrit (%)	41.8 ± 5.4	40.6 ± 4.8	42.4 ± 5.6	0.005
Platelet (×10³/mm³)	298.8 ± 100.7	278.9 ± 91.9	309.4 ± 103.7	0.013
BUN (mg/dL)	18.0 ± 12.8	16.5 ± 15.5	18.8 ± 11.0	0.001
Creatinine (mg/dL)	1.6 ± 1.3	1.0 ± 1.2	1.9 ± 1.2	<0.001
Sodium (mmol/L)	139.3 ± 3.9	139.0 ± 3.3	139.5 ± 4.1	0.290
Potassium (mmol/L)	3.0 ± 0.7	3.5 ± 0.6	2.8 ± 0.6	<0.001
AST (U/L)	57.6 ± 63.6	39.1 ± 36.0	67.4 ± 72.4	<0.001
ALT (U/L)	31.0 ± 40.3	26.4 ± 36.3	33.5 ± 42.1	0.038

MAP: mean arterial pressure; BUN: blood urea nitrogen; AST: aspartate transaminase; ALT: alanine transaminase; WBC: white blood cell; PaCO₂: partial pressure of arterial carbon dioxide; PaO₂: partial pressure of arterial oxygen; ${H C O}_{3}^{-}$ : bicarbonate; SD: standard deviation; ED: emergency department.

^aData are presented as the means with SDs or number with percentage.

We validated the Yamaguchi index and compared the mortality rates of the three different groups. The mortality rates were 7.1% (2 of 28), 22.4% (15 of 67), and 87.6% (177 of 202), and there was a significant difference between the three groups (p < 0.001). The AUC for predicting hospital mortality from the external validation of the Yamaguchi index was 0.842 (95% CI: 0.795–0.882; Figure 2).

Figure 2.

ROC curve for external validation of Yamaguchi index. The diagonal line is reference line. ROC: receiver–operating characteristics.

Discussion

The herbicidal effect of PQ is much stronger than that of other products. It is a very fast acting, nonselective contact herbicide that kills a wide range of grasses and weeds. However, PQ is fatal to humans when ingested. Due to the high mortality of acute PQ poisoning, many intensive therapeutic modalities for PQ poisoning such as adsorbents, pharmacological treatment, lung radiotherapy, extracorporeal detoxification, and immunosuppressive therapy have been performed.^8

–11 However, the effectiveness of these therapeutic methods remains uncertain.

There are many markers of poor prognosis for PQ poisoning, and plasma PQ concentration level is the most reliable factor.^3
–5,12 However, testing for the level of plasma PQ concentration is not always available in every institution, including our hospital. Yamaguchi et al. proposed an equation calculated from different variables, including serum potassium, creatinine, arterial ${H C O}_{3}^{-}$ , and the time from PQ ingestion.⁶ Patients were divided into three groups based on the equation, and the mortality rates were 10, 42, and 97%. A number of studies have validated the Yamaguchi index for predicting the prognosis in PQ poisoning patients and have found the test to have good reliability and validity.^13
–15 However, these studies were not reported in English-language literature. In the present study, we performed an external validation study of 297 acute PQ poisoning patients, and the mortality rates were 7.1, 22.4, and 87.6%. This study shows that the Yamaguchi index is a significant predictor of outcomes in patients with acute PQ poisoning. It is fast and simple to use this index because the Yamaguchi index is derived from laboratory tests that can be performed easily in almost every hospital.

The Yamaguchi index includes serum potassium, creatinine, and arterial ${H C O}_{3}^{-}$ . Hypokalemia, one of the factors associated with poor prognosis from the equation of the Yamaguchi index, could be induced by several mechanisms including (a) renal potassium depletion resulting from decreased potassium reabsorption possibly due to direct oxidative injury in renal tubules, (b) a large amount of intracellular shifting of potassium by Na⁺/K⁺ pumps as a result of increased catecholamine and cortisol secretion, and (c) gastrointestinal potassium losses from emesis and diarrhea caused by PQ ingestion.¹⁶ Decreased arterial ${H C O}_{3}^{-}$ and elevated serum creatinine levels were also associated with poor prognoses because acute PQ poisoning is closely related to acid–base disturbances and impaired renal function.¹⁷

There are several limitations of the current study. First, the study is a retrospective analysis of data collected from a single institution. Therefore, the results of the study may lack generalizability. Second, all data, including vital signs and laboratory measures, were only collected at the initial ED presentation. Differences in patient’ data between initial and follow-up tests are not considered in this study. However, that data might be important for determining organ function and helpful for predicting the prognosis of acute PQ poisoning. Third, the Yamaguchi index does not include the amount of PQ ingested. As mentioned previously, plasma PQ concentration is highly associated with a poor outcome of PQ poisoning, and the concentration of plasma PQ is significantly influenced by the amount ingested. Fourth, the primary end point of this study was hospital mortality before discharge. It is possible that patients died after discharge. However, our hospital is the largest and only tertiary referral center in our community. Consequently, it is unlikely that critically ill PQ-poisoned patients visited another hospital if his/her condition worsened after discharge, and there was no case of a return to the hospital in the current study.

Conclusions

We conducted an external validation study of the Yamaguchi index for acute PQ poisoning patients. In our cohorts, the Yamaguchi index showed good reliability and validity. Therefore, the Yamaguchi index could be a reliable prognostic factor for predicting mortality due to PQ poisoning.

Footnotes

Conflict of interest

The authors declared no conflicts of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Lee

Hwang

Yang

. Predictors of survival after acute paraquat poisoning. Toxicol Ind Health 2002; 18: 201–206.

Weng

Lin

. Sequential organ failure assessment score can predict mortality in patients with paraquat intoxication. PloS One 2012; 7: e51743.

Dinis-Oliveira

Duarte

Sanchez-Navarro

. Paraquat poisonings: mechanisms of lung toxicity, clinical features, and treatment. Crit Rev Toxicol 2008; 38: 13–71.

Liu

. Prognostic value of initial arterial lactate level and lactate metabolic clearance rate in patients with acute paraquat poisoning. Am J Emerg Med 2013; 31: 1230–1235.

Min

Ahn

Chan

. Prediction of prognosis in acute paraquat poisoning using severity scoring system in emergency department. Clin Toxicol (Phila) 2011; 49: 840–845.

Yamaguchi

Sato

Watanabe

. Pre-embarkment prognostication for acute paraquat poisoning. Hum Exp Toxicol 1990; 9: 381–384.

Kang

Kim

Lee

. Absolute lymphocyte count as a predictor of mortality in emergency department patients with paraquat poisoning. PloS One 2013; 8: e78160.

Bateman

. Pharmacological treatments of paraquat poisoning. Hum Toxicol 1987; 6: 57–62.

Talbot

Barnes

. Radiotherapy for the treatment of pulmonary complications of paraquat poisoning. Hum Toxicol 1988; 7: 325–332.

10.

Hampson

Pond

. Failure of haemoperfusion and haemodialysis to prevent death in paraquat poisoning. A retrospective review of 42 patients. Med Toxicol Adverse Drug Exp 1988; 3: 64–71.

11.

Sydenham

Chaudhary

. Glucocorticoid with cyclophosphamide for paraquat-induced lung fibrosis. Cochrane Database Syst Rev 2012; 7: Cd008084.

12.

Senarathna

Eddleston

Wilks

. Prediction of outcome after paraquat poisoning by measurement of the plasma paraquat concentration. QJM 2009; 102: 251–259.

13.

Kim

Choi

. The significance of Yamaguchi index in patients with acute paraquat poisoning. J Korean Soc Emerg Med 1995; 6: 459–462.

14.

Bae

Kim

Sohn

. Prognostic factors in patients with paraquat poisoning and effect of hemoperfusion. J Korean Soc Emerg Med 2002; 13: 229–236.

15.

Park

Lim

. Application of sequential organ failure assessment (SOFA) score in the prognostic prediction of paraquat poisoning. J Korean Soc Emerg Med 2009; 20: 385–391.

16.

Chang

Lee

. Hypokalemia and hypothermia are associated with 30-day mortality in patients with acute paraquat poisoning. Am J Med Sci 2008; 335: 451–456

17.

Bismuth

Garnier

Dally

. Prognosis and treatment of paraquat poisoning: a review of 28 cases. J Toxicol Clin Toxicol 1982; 19: 461–474.