Performance assessment of new Poisoning Mortality Score and PGI score for predicting mortality in patients with acute aluminum phosphide poisoning

Abstract

Background

Until now, no definite standardized method has been used to promptly assess the severity and outcome of acute aluminum phosphide (ALP) poisoning. The current study aimed to evaluate the performance of the new Poisoning Mortality Score (PMS) and PGI score for predicting mortality in acute ALP-poisoned patients, highlighting the accuracy of new PMS components.

Patients and Methods

A 2-year cross-sectional study was conducted on ALP-poisoned patients admitted to Tanta University Poison Control Centre from April 2021 to March 2023. Socio-demographics, poisoning data, and initial vital signs were recorded. Additionally, new PMS and PGI scores were calculated on admission. Patients were categorized according to the mortality outcome into survivors and nonsurvivors.

Results

Out of 160 included ALP poisoned patients, mortality was recorded in 112 (70%) patients. The nonsurvivors had significantly higher median PGI and new PMS values than survivors. New PMS, vital signs component of new PMS, and PGI conveyed good discriminatory power for predicting mortality (AUC = 0.883, 0.873, and 0.817, respectively). Although the new PMS outperformed PGI in all predictive metrics, no significant difference in AUCs was observed between the new PMS and its vital signs component.

Conclusion

The new PMS vital signs component is closely aligned with the new PMS. Thus, it can be used as a valid, comprehensive, and practical tool to substitute the whole score calculation for rapid ALP-poisoned patient assessment to enhance emergency clinical decision-making.

Keywords

Acute aluminum phosphide poisoning mortality outcomes new poisoning mortality score PGI score

Introduction

Aluminum phosphide (ALP), known as rice tablets or wheat pills, is a popular over-the-counter insecticide and rodenticide in some developing countries.^1,2 Although it effectively protects crops from all stages of insects without leaving toxic residues on grains, it accounts for many suicidal deaths.^3,4

ALP toxicity exists when ALP tablets are exposed to moisture or gastric acidity, releasing the rapidly absorbed phosphine gas.⁵ Mechanistically, phosphine targets mitochondria, disrupting cytochrome C oxidase activity and cellular respiration. Additionally, it enhances the production of reactive oxygen species and impairs the endogenous antioxidant scavenging activities, resulting in lipid peroxidation or even cellular death.^6,7

Clinically, ALP-poisoned patients often present initially with sudden onset of abdominal pain, vomiting, refractory hypotension, metabolic acidosis, and dysrhythmias. However, cases may rapidly deteriorate to multiorgan failure.^8,9 Owing to the absence of a convenient antidote, the ALP mortality rates range from 37% to 100% within the first 24 hours due to cardiogenic shock and inevitable dysrhythmias.¹⁰ Accordingly, recent statistical scales in various poison centers pointed to a tremendously increased incidence of ALP-poisoned cases with considerably alarming morbidities and mortalities.^11–13

Currently, the mortality prediction in ALP poisoning depends on various baseline clinical and laboratory factors such as blood pressure, metabolic acidosis, abnormal electrocardiogram, presence of low Glasgow Coma Scale (GCS), and alteration in blood glucose levels and white blood cell counts.^14,15 Additionally, other studies suggested scoring systems to provide more accurate assessment by incorporating more than one parameter, including Poisoning Severity Score (PSS), Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA), and Simplified Acute Physiology Score II (SAPS II).^16–18 However, most of them are not routinely used in emergencies due to time-consuming calculations resulting from including many laboratory parameters that impede the rapid decision-making of high-risk ALP-poisoned patients.¹⁹

As a result of previous scores shortcomings, Pannu et al. (2020)²⁰ have suggested a simplified clinical-laboratory prognostic “PGI score” including low pH (P), low Glasgow Coma Scale score (G), and impaired systolic blood pressure (I) for assessing poisoning severity and predicting mortality. Previous studies have demonstrated its effectiveness in evaluating ALP poisoning severity as well as predicting mortality.^21,22

Han et al. (2021)²³ have also constructed a new Poisoning Mortality Score (new PMS) by considering demographic factors, poisoning-related characteristics, and the patient’s physiological conditions to be more applicable to all age groups of acutely poisoned patients. Although it was validated for predicting mortality in acute organophosphate poisoning, it has not been evaluated in acute ALP poisoning.

Hence, the current study aimed to assess the performance of PGI and new PMS scores for predicting mortality in acute ALP poisoning. Owing to the wide range of new PMS (0–137 points), we further evaluate the predictive ability of each new PMS component to simplify the assessment process and assist the early treatment allocation for high-risk ALP-poisoned patients.

Patients and methods

Study design and setting

This cross-sectional study was conducted on ALP-poisoned patients admitted to Tanta University Poison Control Centre (TUPCC), Tanta Emergency Hospitals, Egypt, from April 2021 to March 2023.

Ethical consideration

This study was performed after the agreement of the institutional research ethics committee of the Faculty of Medicine - Tanta University (Approval number: 34577/3/21) and according to the Declaration of Helsinki guidelines. Written informed consent was attained from each patient or the legal guardian of the incompetent patients. Coding of the patients was considered to maintain their confidentiality and ensure anonymous data analysis.

Inclusion criteria

The current study enrolled all adult symptomatic acute ALP patients above 18 years of both sexes. The diagnosis was based on the history of exposure, the availability of the pesticide containers, and suggestive toxidrome features of metabolic acidosis, hypotension, and garlic breath odor, as well as detecting phosphine gas by applying a silver nitrate test on the gastric aspirate. A silver nitrate test was performed by mixing 5 mL of gastric aspirate with 15 mL of water in a flask and covering its mouth with 0.1 N silver nitrate-soaked filter paper. The positive result is the blackening of filter paper when heating this mixture at 50^o for 15 to 20 minutes, indicating the existence of phosphine gas.²⁴

Exclusion criteria

Patients with co-ingestion of other agents and comorbidities such as cardiovascular, hepatic, or renal disorders were excluded from this study. Additionally, our study excluded dead patients on admission, patients with unknown poisoning, incomplete data, persistent asymptomatic patients, transferred patients, and those who received any medical intervention pre-hospital admission.

Patient data collection

A standardized sheet was conducted to gather data from the enrolled patients’ medical records. Data included socio-demographics (age, gender, and residence) and poisoning data (alleged route, mode, amount, and delay time). Initial vital data including heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), respiratory rate (RR), and temperature, as well as level of consciousness using GCS, and central venous pressure (CVP) were recorded.

Additionally, the results of laboratory investigations were documented on admission and before receiving any treatment. Substantially, arterial blood gas analysis (ABG), including pH, partial arterial oxygen pressure (PaO₂), partial arterial carbon dioxide pressure (PaCO₂), and serum bicarbonate (HCO₃), were retrieved. Additionally, the first serum electrolytes (sodium and potassium levels), liver enzymes (aspartate aminotransferase and alanine aminotransferase), kidney function tests (urea and creatinine), blood indices including complete blood count, and random blood sugar were recorded. Furthermore, electrocardiography was done for each patient.

Scoring calculations

On patient admission, in addition to PSS, two studied scores were calculated for every admitted patient as follows:

Poison severity score (PSS)

The Poisoning Severity Score is the most popular conventional scoring for grading the degrees of poisoning severity depending on the most severe patient’s symptoms or signs as follows: grade 0; asymptomatic, grade 1; mild transient spontaneous resolving symptoms, grade 2; pronounced or prolonged symptoms, grade 3; severe or life-threatening symptoms, grade 4; death²⁵.

PGI score

The PGI score uses three characteristics of three systems: metabolic status, blood pressure, and consciousness level, including blood pH < 7.25, GCS score < 13, and SBP < 87 mmHg. Each variable gives one point, and the total score ranges from 0 (best) to 3 (worst).²⁰ This score can stratify the risk of ALP patients into three categories: zero points (low risk), one point (intermediated risk), and two to three points (high risk).²¹

New poisoning mortality score

The new PMS consists of 10 predictors by incorporating demographic factors (age and sex), poisoning-related factors (intent of poisoning, route of exposure, and poison category), initial vital signs (SBP, HR, RR, temperature), and consciousness level assessment using AVPU scale (A-alert, V-verbal, P-painful, and U-unresponsive). The total points of new PMS range from 0 to 137, where 0 is normal, and 137 is the worst. Additionally, the new PMS categorizes the patient’s mortality risk into four groups: very low (0 – 27), low (28 – 40), intermediate (41 – 55), and high risk (≥ 56).²³

Treatment strategy and patient grouping

All patients received the standard treatment according to the TUPCC protocol, including ensuring a patent airway with supporting breathing and circulation. Additionally, hypotensive patients were treated by continuous intravenous fluid infusion and vasopressors (norepinephrine). Although some patients initially have stable hemodynamic status, they have rapidly deteriorated due to ALP-induced cardiotoxicity.²⁶ Accordingly, upon detecting hypotension through vital sign monitoring, we inserted a central venous catheter for further hemodynamic monitoring and administered intravenous fluids, potentially in combination with vasopressor infusion based on CVP recording, to sustain the SBP over 80 mm Hg as well as to prevent fluid overloading and pulmonary edema development.^27–29 Furthermore, metabolic acidosis was corrected according to its degree and treatment of other expected manifestations.

Paraffin oil was used to perform gastric lavage according to recent treatment guidelines. The gastric lavage technique was performed by placing the nasogastric tube from the nose to the stomach after positioning the patients in the left lateral decubitus to reduce the risk of gastric content aspiration. Upon confirming the proper position of the nasogastric tube, the gastric contents were first aspirated for medicolegal preservation and silver nitrate test performance. Subsequently, 50 mL of paraffin oil was administered through the nasogastric tube with frequent aspiration after 3 to 5 min till the aspirate was clear. Finally, another 50 mL of paraffin oil was administered and left in the stomach.³⁰ Although the usefulness of gastric lavage with paraffin oil in cases of ALP poisoning has been documented in current literature,³¹ there are certain situations in which it should not be used. Substantially, paraffin oil’s petroleum-based composition renders it unsuitable for disturbed-conscious patients and those with respiratory disorders to prevent the development of aspiration pneumonitis. Alternatively, coconut oil has been recommended for those patients as it is more safe and tolerable.³²

According to the primary outcome (mortality), patients were categorized into survivors and nonsurvivors.

Sample size

The sample size was calculated based on the mortality outcome (Pannu et al., 2022)²¹ according to the methods described by Buderer (1996)³³ and Negida et al. (2019).³⁴ We calculated the number (N) required for sensitivity and that required for specificity using the following equations: N required for sensitivity = TP + FN/P, and N required for specificity = TN + FP/1-P, where TP, FN, TN, FP, P stand for true positive, false negative, true negative, false positive, and prevalence, respectively. Accordingly, the minimum required number was 140 patients, which increased to 160 to increase the predictive scores’ performance accuracy.

Statistical analysis

Analyses were conducted using the R Statistical language (version 4.4.0; R Core Team, 2024), using the packages pROC (version 1.18.5; Robin X et al., 2011³⁵), gtsummary (version 1.7.2; Sjoberg et al., 2021³⁶), ggplot2 (version 3.5.1; Wickham H, 2016³⁷). The distribution of numerical data was tested using the Shapiro-Wilk test for normality and the Q-Q plots. Normally distributed data were summarized using the mean and standard deviation (SD), and comparisons were performed using the unpaired t-test. Abnormally distributed data were summarized using the median and interquartile range (IQR, expressed as 25th – 75th percentiles), and between-group comparisons were done using the Mann-Whitney test (Wilcox rank sum test). Data were expressed as numbers and percentages for categorical variables and compared using the Cochran-Armitage test for trend and the Chi-squared Test for comparative analysis. The predictive performance of the studied scores was analyzed through receiver operating characteristic (ROC) curves. The area under the ROC curve and its 95% confidence interval (CI) were calculated. The optimal cutoff point was identified using the Youden J index, and the associated sensitivity, specificity, predictive values, and accuracy were computed. Pairwise testing of the AUCs of each two score was done using the DeLong method. A p-value <0.05 was the significance level of the statistical tests.

Results

The total number of patients who fulfilled the inclusion criteria and enrolled in the study was 160. Mortality was recorded in 112 patients, which constituted 70% of the participants. Table 1 shows that the age of the studied patients ranged from 18 to 55 years, and the mean age of nonsurvivors was significantly higher than survivors (25.73 versus 20.52 years; p = 0.001). Males constituted more than half of the nonsurvivors (51.8%), with a significant difference between the two groups (p < 0.001). However, there is no significant difference between the two groups regarding sociodemographic status (educational level, occupation, marital status) and psychiatric comorbidities (p > 0.05). The median of the ingested ALP amount was significantly higher in nonsurvivors than survivors (1 tablet versus 0.75 tablets; p = 0.001), with no significant difference regarding the delay time (p = 0.098). All the included patients alleged intentional exposure and were referred from rural areas.

Table 1.

Comparison of basic characteristics and complications among survivors and nonsurvivors of acute aluminum phosphide-poisoned patients (n = 160).

Characteristics		Survived (n = 48)	Non survived (n = 112)	Total (n = 160)	p-value
Age (years)	Mean ± SD (Range)	20.52 ± 5.68 (18.00 – 40.00)	25.73±10.23 (18.00 – 55.00)	24.17±9.40 (18.00 – 55.00)	0.001 * t
Gender	Female	38 (79.2%)	54 (48.2%)	92 (57.5%)	<0.001 * X²
Gender	Male	10 (20.8%)	58 (51.8%)	68 (42.5%)	<0.001 * X²
Education	Illiterate	0 (0%)	1 (0.89%)	1 (0.63%)	>0.999 X² _L
	Primary school	3 (6.25%)	7 (6.25%)	10 (6.25%)
	Secondary school	41 (85.4%)	93 (83.0%)	134 (83.8%)
	University	4 (8.33%)	11 (9.82%)	15 (9.38%)
Occupation	Housewife	6 (12.5%)	21 (18.8%)	27 (16.9%)	0.188 FE
	Worker	2 (4.17%)	13 (11.6%)	15 (9.38%)
	Farmer	7 (14.6%)	24 (21.4%)	31 (19.4%)
	Employee	2 (4.17%)	4 (3.57%)	6 (3.75%)
	Student	31 (64.6%)	50 (44.6%)	81 (50.6%)
Marital status	Single	31 (64.6%)	71 (63.4%)	102 (63.8%)	0.403 FE
	Married	16 (33.3%)	41 (36.6%)	57 (35.6%)
	Divorced	1 (2.08%)	0 (0%)	1 (0.63%)
Psychiatric comorbidities	Yes	5 (10.4%)	12 (10.7%)	17 (10.6%)	0.955 X²
Psychiatric comorbidities	No	43 (89.6%)	100 (89.3%)	143 (89.4%)	0.955 X²
Amount	Median [IQR] (Range)	0.75 [0.50 – 1.00] (0.25 – 1.00)	1.00 [0.50 – 1.00] (0.25 – 3.00)	1.00 [0.50 – 1.00] (0.25 – 3.00)	0.001 * Z
Delay time (hours)	Median [IQR] (Range)	2.00 [1.50 – 3.00] (0.50 – 6.00)	2.25 [2.00 – 3.50] (0.50 – 34.00)	2.00 [1.50 – 3.00] (0.50 – 34.00)	0.098 Z
Heart rate (Beats/min)	Mean ± SD (Range)	100.96 ± 17.96 (74.00 – 148.00)	86.05 ± 21.82 (36.00 – 136.00)	90.58 ± 21.78 (36.00 – 148.00)	<0.001 * t
Systolic blood pressure (mmHg)	Mean ± SD (Range)	101.30 ± 16.68 (80.00 – 150.00)	81.61 ± 17.94 (40.00 – 120.00)	90.00 ± 19.91 (40.00 – 150.00)	<0.001 * t
Diastolic blood pressure (mmHg)	Mean ± SD (Range)	58.26 ± 15.10 (40.00 – 100.00)	47.74 ± 14.65 (30.00 – 90.00)	52.22 ± 15.67 (30.00 – 100.00)	<0.001 * t
MAP (mmHg)	Mean ± SD (Range)	72.90 ± 13.38 (53.33 - 106.67)	59.02 ± 14.44 (33.33 - 100.00)	64.93 ± 15.55 (33.33 - 106.67)	<0.001 * t
Respiratory rate (breaths/minute)	Mean ± SD (Range)	21.17 ± 5.30 (14.00 – 40.00)	26.68 ± 6.67 (16.00 – 48.00)	25.02 ± 6.76 (14.00 – 48.00)	<0.001 * t
Temperature (°C)	Mean±SD (Range)	37.00 ± 0.00 (37.00 – 37.00)	37.00 ± 0.00 (37.00 – 37.00)	37.00 ± 0.00 (37.00 – 37.00)	1.000 t
Glasgow Coma Scale grading	Normal	48 (100.0%)	86 (76.8%)	134 (83.8%)	0.002 * X²_L
	Mild	0 (0.0%)	16 (14.3%)	16 (10.0%)
	Moderate	0 (0.0%)	4 (3.6%)	4 (2.5%)
	Severe	0 (0.0%)	6 (5.4%)	6 (3.8%)
Central venous pressure (mmHg)	Mean ± SD (Range)	15.67 ± 5.68 (7.00 – 30.00)	24.22 ± 5.41 (12.00 – 33.00)	21.54±6.77 (7.00 – 33.00)	<0.001 * t
Cardiac arrest	Yes	2 (4.17%)	50 (44.6%)	52 (32.5%)	<0.001 * X²
Cardiac arrest	No	46 (95.8%)	62 (55.4%)	108 (67.5%)	<0.001 * X²
ICU admission	Yes	26 (54.2%)	106 (94.6%)	132 (82.5%)	<0.001 * X²
ICU admission	No	22 (45.8%)	6 (5.36%)	28 (17.5%)
Mechanical ventillation	Yes	2 (4.17%)	94 (83.9%)	96 (60.0%)	<0.001 * X²
Mechanical ventillation	No	46 (95.8%)	18 (16.1%)	64 (40.0%)	<0.001 * X²
Poisoning Severity Score	Minor	8 (16.7%)	0 (0.0%)	8 (5.0%)	<0.001 * X²_L
	Moderate	26 (54.2%)	8 (7.1%)	34 (21.3%)
	Severe	14 (29.2%)	104 (92.9%)	118 (73.8%)

FE: Fisher’s exact test; n: number, Z: Mann-Whitney test, X²: Pearson’s Chi-square test for independence, t: Independent samples t-test, X²_L: Cochrane-Armitage test for trend, IQR: interquartile range, SD: standard deviation, MAP: mean arterial blood pressure, ICU: intensive care unit, *significant at p < 0.05.

Regarding vital signs, nonsurvivors had significantly lower mean values of HR, SBP, DBP, and mean arterial pressure (MAP) than survivors (p < 0.001 for each). Likewise, GCS was significantly lower in nonsurvivors than survivors (p = 0.002), where all survivors had normal GCS, and all patients with severe GCS did not survive. Conversely, the mean value of RR was significantly higher in nonsurvivors than in the survivors (26.68 versus 21.17 cycles/minute; p < 0.001). Similarly, in nonsurvivors, CVP was significantly higher than in survivors (24.22 mmHg versus 15.67 mmHg; p < 0.001). Furthermore, there was a significant difference regarding PSS grading (p < 0.001), where severe grading was observed in 92.9% of the nonsurvivors versus 29.2% of the survivors. Accordingly, the nonsurvivors had a significantly higher rate of admission to the ICU and mechanical ventilation than the survivors (p < 0.001). Cardiac arrest was significantly more prevalent in nonsurvivors than in survivors (p < 0.001), as shown in Table 1.

Table 2 reveals that the nonsurvivors had significantly lower mean O₂ saturation, pH, and HCO₃ values than survivors (p < 0.001). Compared to survivors, the nonsurvivors had substantially higher serum Na and significantly lower serum K levels (p = 0.026 and 0.019, respectively). Although there was no significant difference between the two groups regarding liver enzymes, nonsurvivors had significantly higher blood urea, serum creatinine, and RBS than survivors (p = 0.025, < 0.001, and < 0.001, respectively). Furthermore, abnormal ECG and shortened hospitalization periods were significantly recorded in nonsurvivors (p < 0.001 for each). The nonsurvivors had considerably higher percentages of sinus bradycardia, ST-segment elevation, ST-segment depression, and extrasystole than survivors (p = 0.006, 0.041, 0.006, and 0.044, respectively).

Table 2.

Comparison of initial laboratory investigations, electrocardiographic findings, and hospitalization period among survivors and nonsurvivors of acute aluminum phosphide-poisoned patients (n = 160).

Characteristics		Survived (n = 48)	Non survived (n = 112)	Total (n = 160)	p-value
Oxygen saturation (%)	Mean ± SD (Range)	95.71 ± 3.66 (86.00 – 100.00)	84.41 ± 12.73 (35.00 – 100.00)	87.80 ± 12.00 (35.00 – 100.00)	<0.001* t
PaCO₂ (mmHg)	Median [IQR] (Range)	25.80 [22.15 – 31.70] (12.70 – 38.00)	26.00 [19.00 – 33.20] (7.00 – 53.00)	26.00 [20.00 – 32.35] (7.00 – 53.00)	0.671 Z
pH	Mean ± SD (Range)	7.40 ± 0.09 (7.10 – 7.54)	7.29 ± 0.12 (6.90 – 7.57)	7.32 ± 0.12 (6.90 – 7.57)	<0.001* t
Serum HCO₃ (mmol/L)	Mean ± SD (Range)	16.58 ± 4.30 (7.20 – 23.00)	13.10 ± 4.80 (3.90 – 26.00)	14.14 ± 4.91 (3.90 – 26.00)	<0.001* t
Serum sodium (mmol/L)	Mean ± SD (Range)	140.25 ± 4.70 (133.00 – 152.00)	142.66 ± 6.65 (120.00 – 160.00)	141.88 ± 6.18 (120.00 – 160.00)	0.026* t
Serum potassium (mmol/L)	Mean ± SD (Range)	3.78 ± 0.62 (2.07 – 5.20)	3.58 ± 0.41 (2.60 – 4.35)	3.64 ± 0.49 (2.07 – 5.20)	0.019* t
AST (U/L)	Median [IQR] (Range)	24.00 [18.00 – 29.00] (10.00 – 45.00)	23.50 [15.00 – 37.50] (7.00 – 151.00)	24.00 [15.00 – 33.00] (7.00 – 151.00)	0.793 Z
ALT (U/L)	Median [IQR] (Range)	17.00 [12.00 – 21.00] (7.00 – 44.00)	17.50 [13.50 – 32.50] (8.00 – 161.00)	17.00 [13.00 – 27.00] (7.00 – 161.00)	0.054 Z
Blood urea (mg/dL)	Median [IQR] (Range)	30.00 [23.00 – 35.00] (17.00 – 51.00)	32.00 [28.00 – 39.00] (16.00 – 64.00)	31.00 [26.00 – 38.00] (16.00 – 64.00)	0.025*Z
Serum creatinine (mg/dL)	Mean ± SD (Range)	0.88 ± 0.28 (0.40 – 1.60)	1.27 ± 0.35 (0.60 – 2.20)	1.13±0.38 (0.40 – 2.20)	<0.001*t
Random blood sugar (mg/dL)	Mean ± SD (Range)	118.79 ± 53.51 (72.00 – 348.00)	162.39 ± 77.44 (42.00 – 441.00)	149.31±73.72 (42.00 – 441.00)	<0.001*t
Electrocardiography	Normal	30 (62.5%)	10 (8.9%)	40 (25.0%)	<0.001*X²
Electrocardiography	Abnormal	18 (37.5%)	102 (91.1%)	120 (75.0%)	<0.001*X²
Sinus tachycardia	Absent	31 (64.6%)	81 (72.3%)	112 (70.0%)	0.328 X²
Sinus tachycardia	Present	17 (35.4%)	31 (27.7%)	48 (30.0%)
Sinus bradycardia	Absent	48 (100.0%)	97 (86.6%)	145 (90.6%)	0.006*FE
Sinus bradycardia	Present	0 (0%)	15 (13.4%)	15 (9.38%)
ST segment elevation	Absent	47 (97.9%)	98 (87.5%)	145 (90.6%)	0.041*FE
ST segment elevation	Present	1 (2.08%)	14 (12.5%)	15 (9.38%)
ST segment depression	Absent	45 (93.8%)	84 (75.0%)	129 (80.6%)	0.006*X²
ST segment depression	Present	3 (6.25%)	28 (25.0%)	31 (19.4%)
Inverted T wave	Absent	47 (97.9%)	99 (88.4%)	146 (91.3%)	0.066 FE
Inverted T wave	Present	1 (2.08%)	13 (11.6%)	14 (8.75%)
Extrasystole	Absent	41 (85.4%)	107 (95.5%)	148 (92.5%)	0.044 * FE
Extrasystole	Present	7 (14.6%)	5 (4.46%)	12 (7.50%)
Atrial fibrillation	Absent	47 (97.9%)	110 (98.2%)	157 (98.1%)	>0.999FE
Atrial fibrillation	Present	1 (2.08%)	2 (1.79%)	3 (1.88%)
Hospitalization period (hours)	Median [IQR] (Range)	72.00 [58.00 – 104.00] (16.00 – 172.00)	6.00 [4.00 – 11.50] (2.50 – 25.00)	10.50 [5.00 – 47.50] (2.50 – 172.00)	<0.001* Z

FE: Fisher’s exact test; n: number, Z: Mann-Whitney test, X²: Pearson’s Chi-square test for independence, t: Independent samples t-test, IQR: interquartile range, SD: standard deviation, PaCO₂: partial arterial carbon dioxide pressure, HCO₃: bicarbonate, AST: aspartate aminotransferase, ALT: alanine aminotransferase, *significant at p < 0.05.

Regarding the scoring system, nonsurvivors had significantly higher PGI score median value than survivors (1 versus 0; p < 0.001). Similarly, the median value of new PMS in nonsurvivors was substantially higher than that of survivors (59 versus 47.50; p < 0.001). Regarding severity grading, the majority of nonsurvivors had significantly more high-grade new PMS (67.9%) and intermediate to high PGI grades (80%) than survivors (p < 0.001), as demonstrated in Table 3.

Table 3.

Comparison of PGI score and new poisoning mortality score between survivors and nonsurvivors of acute aluminum phosphide-poisoned patients (n = 160).

Variables		Survived (n = 48)	Non survived (n = 112)	Total (n = 160)	p-value
PGI	Median [IQR] (Range)	0.00 [0.00 – 0.00] (0.00 – 2.00)	1.00 [1.00 – 2.00] (0.00 - 3.00)	1.00 [0.00 – 1.50] (0.00 - 3.00)	<0.001 * Z
PGI	Mean rank	45.04	95.70		<0.001 * Z
New PMS	Median [IQR] (Range)	47.50 [43.00 – 49.50] (43.00 – 71.00)	59.00 [54.00 – 67.50] (43.00 – 79.00)	57.00 [49.00 – 67.00] (43.00 – 79.00)	<0.001 * Z
New PMS	Mean rank	37.58	98.89		<0.001 * Z
New PMS grading	Very low	0 (0.0%)	0 (0.0%)	0 (0.0%)	<0.001 * X²_L
	Low	0 (0.0%)	0 (0.0%)	0 (0.0%)
	Intermediate	42 (87.5%)	36 (32.1%)	78 (48.8%)
	High	6 (12.5%)	76 (67.9%)	82 (51.2%)
PGI grading	Low	38 (79%)	22 (20%)	60 (38%)	<0.001 * X²
	Intermediate	8 (17%)	52 (46%)	60 (38%)
	High	2 (4%)	38 (34%)	40 (25%)

n: number, Z: Mann-Whitney test, IQR: interquartile range, X²_L: Cochran-Armitage test for trend; X²: Chi-squared Test for Trend in Proportion, new PMS: new Poisoning Mortality Score; *significant at p < 0.05.

Studying the new PMS components showed that the nonsurvivors had significantly higher median values of demographic and vital signs points than survivors (4, 15 versus 0 and 4, respectively; p < 0.001 for each). However, no significant difference was observed between the two groups regarding poisoning points, as illustrated in Table 4.

Table 4.

Comparison of components values of new poisoning mortality score between survivors and nonsurvivors of acute aluminum phosphide-poisoned patients (n = 160).

New PMS risk grades		Survived (n = 48)	Non survived (n = 112)	p-value
Demographic points	Median [IQR] (Min – Max)	0.00 [0.00 – 2.00] (0.00 – 7.00)	4.00 [0.00 – 4.00] (0.00 – 11.00)	<0.001 * Z
Poisoning points	Median [IQR] (Min – Max)	43.00 [43.00 – 43.00] (43.00 – 43.00)	43.00 [43.00 – 43.00] (43.00 – 43.00)	1.000 Z
Vital signs points	Median [IQR] (Min – Max)	4.00 [0.00 – 6.00] (0.00 – 24.00)	15.00 [10.00 – 20.00] (0.00 – 32.00)	<0.001 * Z

n: number, New PMS: new Poisoning Mortality Score, IQR: interquartile range, Min: minimum, Max: maximum, Z: Mann-Whitney test, *significant at p < 0.05.

ROC analysis’s scores assessment revealed that new PMS and PGI displayed good discrimination for predicting ALP mortality (AUC = 0.883 and 0.817, respectively). Although the vital signs component of the new PMS had good predictive power (AUC = 0.873), the demographic component had fair discriminatory power for mortality prediction (AUC = 0.674). The new PMS at cutoff ≥ 51 exhibited the highest AUC (0.883) and best sensitivity (87.50%) for predicting the likelihood of mortality, with an overall accuracy of 81.7% and the highest NPV (73.08%). However, at a cutoff of more than 11, the vital signs component of new PMS had the highest specificity (87.50%), yielded more precisely in expecting patients at risk of mortality (PPV of 93.33%), with overall accuracy (78.8%), as shown in Table 5 and Figures 1–4.

Table 5.

Receiver operating characteristics (ROC) curve analysis comparing the accuracy of PGI, new Poisoning Mortality Score, and its demographic and vital signs components in predicting mortality in acute aluminum phosphide-poisoned patients (n = 160).

Variables	Cutoff	Sensitivity	Specificity	AUC^a	95% CI	Accuracy	p-value
PGI	≥ 1	80.36	79.17	0.817	0.751 to 0.882	79.5	<0.001*
New PMS	≥ 51	87.50	79.17	0.883	0.820 to 0.946	81.7	<0.001*
Demographic points of new PMS	≥ 2	58.93	75.00	0.674	0.597 to 0.752	63.8	0.000*
Vital signs points of new PMS	> 11	75.00	87.50	0.873	0.811 to 0.935	78.8	0.000*

AUC: area under the curve, CI: confidence interval; New PMS: New Poisoning Mortality Score, ALP: aluminum phosphide, *significant at p < 0.05.

^aPairwise comparisons of AUCs of the studied predictors: new PMS versus PGI (p = 0.018*); demographic points versus New PMS (p < 0.001*); demographic points versus PGI (p = 0.004*); demographic points versus vital signs points (p < 0.001*); vital signs points versus PGI (p = 0.031*); vital signs points versus new PMS (p = 0.237).

Figure 1.

Receiver operating characteristics (ROC) curve of PGI score for predicting mortality of acute aluminum phosphide poisoned patients.

Figure 2.

Receiver operating characteristics (ROC) curve of new Poisoning Mortality Score for predicting mortality of acute aluminum phosphide poisoned patients.

Figure 3.

Receiver operating characteristics (ROC) curve of demographic component of new Poisoning Mortality Score for predicting mortality of acute aluminum phosphide poisoned patients.

Figure 4.

Receiver operating characteristics (ROC) curve of vital signs component of new Poisoning Mortality Score for predicting mortality of acute aluminum phosphide poisoned patients.

Although the AUC of new PMS was significantly higher than PGI and the new PMS components (vital and demographic), no significant difference in AUCs was observed between the new PMS and its vital component (p = 0.237) by pairwise comparison analysis of AUCs of the studied predictors as demonstrated in Table 5.

Discussion

To date, ALP has become a prevalent suicidal agent with an alarming increase in related morbidities and mortalities, especially in developing countries.³⁸ The great challenge is the wide availability of highly fatal poison without restricted purchasing rules.³⁹ Additionally, the rapidly deteriorated clinical course with the lack of specific antidotes augmented this issue’s complexity despite performing many trials of adjuvant therapies with standardized supportive strategies.^32,40,41

As a result, our study recorded a high mortality rate (70%) out of enrolled ALP-poisoned patients. Consistent with our findings, Khalaf et al. (2023),¹³ Ghonem et al. (2024),³⁸ and Dorooshi et al. (2022)⁴² reported nearly ALP fatality incidences (72%, 72.3%, and 77.4 %, respectively). However, lower mortality rates (39%) were reported by Anbalagan et al. (2023).⁴³ They justify this decrease by consuming expired tablets, variation of the exposure formulation (tablet, pellet, and powder), multiple vomiting episodes postexposure, and rapid treatment initiation.

Accordingly, determining the high predictive power score containing readily available parameters is crucial to prioritizing risky patients for intensive management while allocating limited hospital resources. Hence, our study goal was to compare new PMS and PGI scores, focusing on their clinical applicability in predicting mortality to provide an immediate, accurate tool for decision-making in high-risk ALP-poisoned patients.

Our study recorded a significant increase in medians of new PMS and PGI scores among nonsurvivors than survivors (59 versus 47.50 and 1 versus 0, respectively; p < 0.001 for each). In accordance, Saeed and Elmorsy, 2024⁴⁴ and Lee et al. (2024)⁴⁵ demonstrated a significant increase of means and medians new PMS in nonsurvivors compared to survivors (59.32 ± 13.86 versus 47.29 ± 12.93; p < 0.001 and 66 [57-77] versus 31 [23-40]; p < 0.001, respectively) in acutely poisoned patients.

Substantially, Moorthy et al. (2023)⁴⁶ demonstrated a significant increase of new PMS in nonsurvivors compared to survivors among patients with acute organophosphate poisoning (70.12 ± 12 versus 51 ± 7, respectively; p < 0.001). Likewise, Pannu et al. (2022)²¹ revealed that nonsurvivors had higher PGI medians than survivors (3 [2-3] versus 0 [0-0]; p < 0.001) in acute ALP-poisoned patients.

Accordingly, the higher severity grades of PSS, new PMS, and PGI scores on admission in the nonsurvivors could explain the increased median values of the studied scores in our mortality group. Noticeably, 92.9%, 67.9%, and 80% of nonsurvivors had severe PSS, high new PMS, and intermediate to high PGI score grading. Likewise, Pannu et al. (2022)²¹ and Sakr et al. (2023)²² recorded 82.60% and 95.38% of nonsurvivors had high PGI scores in patients with acute ALP poisoning, respectively. Although no previous studies assessed new PMS in acute ALP poisoning, Saeed and Elmorsy, 2024⁴⁴ recorded that 57.1% of nonsurvivors had high new PMS grading in acutely poisoned patients. Furthermore, Lee et al. (2024)⁴⁵ demonstrated a good calibration curve for new PMS for predicting in-hospital mortality, especially in high-risk groups of acutely poisoned patients. Accordingly, the risk for mortality is increased in acutely poisoned patients with intermediate to high scores of the new PMS.⁴⁶

Additionally, the new PMS demographic and vital components were significantly higher among nonsurvivors than survivors in the current study (4, 15 versus 0 and 4, respectively; p < 0.001 for each). This significant association was attributed to substantial differences in age, gender, vital signs, and consciousness level between the two groups. However, there was no significant difference between the two groups regarding poisoning variables, as all patients were exposed intentionally to the same substance category (ALP) by oral route.

Our study recorded that the mean age of nonsurvivors was significantly higher than survivors (25.73 versus 20.52 years; p = 0.001), with male predominance (51.8% versus 48.2%; p < 0.001). Accordingly, Khodabandeh et al. (2014)⁴⁷ observed that the majority of the nonsurvivors of ALP poisoning (54.1%) were males aged 15–43 years. Consequently, a systematic review and meta-analysis by Bagherian et al. (2021)⁴⁸ in Iran concluded that the majority of nonsurvivors were males (62.3%) with significantly higher mean age than survivors (p < 0.01).

Although the gender proportion may vary among studies, numerous studies recorded a young age group with suicidal intentions, highlighting the importance of assessing the psychiatric status of those patients.^49,50 Substantially, El-Farouny et al. (2020)⁵¹ recorded that the predominant instances of intentional self-poisoning (38.4%) were caused mainly by pesticides and occurred in the youngest age group (10–20 years), with personality traits being a significant risk factor for suicidal attempts. Furthermore, the 5-year cross-sectional study by Majidi et al., (2021)¹⁴ demonstrated that the highest incidence of suicidal ALP poisoning (67.9%) belonged to the age group (14–30) with a mean age of 28.6 years. However, Wang et al. (2020)⁵² revealed that middle-aged 40-year-olds had the highest frequency of suicidal pesticide ingestion (3.43%). Furthermore, Song et al. (2020)⁵³ demonstrated increases in the percentage of intentional pesticide poisoning with age; hence, the mortality rate in the elderly patients was higher than in the younger patients.

The contradictory age-related findings among studies may be attributed to various stresses affecting distinct age groups. Noticeably, the youngest age group may subjected to psychic troubles in the family, as well as emotional and educational issues.⁵⁴ Alternatively, the higher age group may experience marital concerns, occupation obstacles, and financial difficulties.⁵¹ Another justification may be related to geographical diversity, as well as variation in sample size and eligibility criteria.

In consistent with previous studies,^21,38,55 all hemodynamic parameters (HR, SBP, DBP, and MAP) were significantly decreased in nonsurvivors compared to survivors (p < 0.001). Hypotension, bradycardia, and ECG changes are the most prominent presentations of ALP-induced cardiotoxicity, which is the leading cause of mortality during the first 24 h.^32,56 The hypovolemic shock could occur due to dehydration, substantial intravascular fluid loss resulting from increased capillary permeability, and adrenal damage inducing peripheral vasodilatation, as well as direct myocardial tissue damage and life-threatening arrhythmias.^38,57,58

However, a significant increase in RR was observed in nonsurvivors (26.68 versus 21.17 cycles/minute; p < 0.001). It may be explained by a compensatory increase in the respiratory rate in response to metabolic acidosis in the nonsurvivor group to wash the excess CO₂ by exhalation. Correspondingly, the blood pH and serum HCO₃ were significantly decreased in the fatal group compared to survivors. Consistent with these findings, Pannu et al. (2022)²¹ observed the concomitant increase in RR and decrease of pH and HCO₃ in the nonsurvivor group compared to survivors in acute ALP-poisoned patients (25.6 ± 3.6 versus 20.6 ± 5.2, 7.04 ± 0.23 versus 7.34 ± 0.10, and 9.7 ± 5.8 versus 16.7 ± 5.0, respectively, with p < 0.001 for each).

In agreement with prior studies,^58,59 all survivors had normal GCS. However, all patients with severe GCS belonged to the non-survivors group (p < 0.001). Consciousness level impairment in nonsurvivors may be explained by phosphine gas-inducing cerebral hypoxia and hypoperfusion resulting from refractory shock.^38,58

The current study demonstrated that new PMS had reasonable discriminatory power for mortality in acute ALP-poisoned patients (AUC = 0.883). Approximately similar, Moorthy et al. (2022)⁴⁶ recorded a valid discrimination of new PMS for mortality prediction in acute OP-poisoned patients (AUC = 0.917). However, Han et al. (2021)²³ recorded higher AUC of new PMS in derivation and validation groups for predicting mortality in acutely poisoned patients (AUC = 0.941 and 0.946, respectively). On the contrary, Saeed and Elmorsy, 2024⁴⁴ reported a lower predictive value of new PMS (AUC = 0.735) in patients with acute intoxication.

The higher predictive power in Han et al. (2021)²³ study is attributed to significant differences in all new PMS components (demographic, poisoning, and vital signs) among survivors and nonsurvivors (p < 0.001). However, Saeed and Elmorsy, 2024⁴⁴ did not observe significance in all new PMS components between survivors and nonsurvivors except for poisoning categories and heart rate parameters (p = 0.046 and 0.025, respectively).

Regarding the PGI score, our study demonstrated good discrimination in predicting mortality of ALP-poisoned patients (AUC = 0.817) at cutoff value ≥1 with a sensitivity of 80.36%, specificity of 79.17%, and overall accuracy of 79.5%. However, Sakr et al. (2023)²² demonstrated that the PGI score had excellent power of mortality discrimination (AUC = 0.951) at cutoff point ≥ 2, with a sensitivity of 95.4%, specificity of 87.5%, and overall accuracy of 94.5% in ALP-poisoned patients. The increased number of patients with PGI grade 3 (high risk) in Sakr et al. (2023)²² study could explain the higher predictive power of PGI in their research than ours. Likewise, Pannu et al. (2020)²⁰ recorded that PGI > 2 had the best sensitivity, negative predictive value, and diagnostic accuracy (93.9%, 93.5%, and 84.8%, respectively) for anticipating mortality in acute ALP-poisoned patients.

In our study, the new PMS significantly outperformed the PGI score in all ROC metrics (p = 0.018). The superiority of the new PMS over the PGI score could be attributed to incorporating demographic and poisoning variables with the vital assessment with a grading scale to stratify patients into four risk groups (very low, low, intermediate, and high risk). Additionally, it provides subgrouping analysis for different poison categories to reinforce its predictive performance in coingestions.

In contrast, the PGI relies on three particular variables with definite cutoff levels (blood pH < 7.25, GCS score < 13, and SBP < 87 mmHg). Accordingly, many previous studies have recorded mortality risk at higher blood pH and GCS than the abovementioned cutoff levels. Substantially, Wahdan and Khalifa, 2020,⁵⁸ Abdelkader et al. (2023),⁶⁰ and Sagah & Elhawary, 2022¹⁵ revealed that pH ≤ 7.28, 7.30, and ≤ 7.33 had good predictive ability for mortality discrimination (AUC = 0.881, 0.752 and 0.938, respectively) in patients with ALP poisoning. Furthermore, El-Sarnagawy (2017)⁵⁵ and Majidi et al. (2021)¹⁴ observed that 25.5% and 35% of nonsurvivors had normal GCS. Alternatively, Erfantalab et al. (2017)⁶¹ did not find a significant difference in GCS between survived and non-survived ALP-poisoned patients, as consciousness may be maintained till late poisoning stages depending on hypotension and hypoxia severity.⁶² Therefore, some ALP-poisoned patients with mortality risk may be unidentified by PGI score.

Unanticipatedly, the current study recorded comparable predictive power between the new PMS and its vital signs component (AUC = 0.883 versus 0.873; p = 0.237) by pairwise comparison analysis. This finding may be explained by the equality of poisoning-related factors between groups and the poor discriminatory power of demographic variables in forecasting ALP mortality (AUC = 0.674).

Strengths and limitations

This study’s strength is demonstrating the effectiveness of the new PMS and its vital signs-related variables over PGI score in predicting ALP mortality. Accordingly, the hemodynamic and AVPU variables can be calculated instead of the total score, especially in intentional acute ALP-poisoned cases by oral route. However, the drawback of this study is the inability to evaluate other poisoning circumstances, such as non-intentional or inhalational exposure. This limitation impeded assessing the association between poisoning-related factors and in-hospital mortality. Additionally, owing to the unavailability of gas chromatography in our institution, the silver nitrate test is a promising preliminary diagnostic tool for ALP poisoning, demonstrating 100% sensitivity to gastric content. However, false positive results may occur in the presence of sulfide derivatives due to the production of silver sulfide. To exclude the presence of sulphides, filter-soaked paper with one drop of 10% lead (II) acetate is used. If it is further blackened, sulfides may be present, leaving the results equivocal. As a result, the confirmatory ammonium molybdate test was carried out by dropping ammonium molybdate suspension over the dark filter paper, causing bluish discoloration in the presence of phosphine gas.^63,64 Furthermore, this study is a single center with a relatively small sample size, limiting the generalization of our findings. Therefore, further multicenter studies should be conducted to assess the applicability and reliability of new PMS and its vital component in various ALP poisoning circumstances.

Conclusions

Although the new PMS and PGI scores performed well in predicting ALP mortality, the new PMS outperformed the PGI score in various metrics. The new PMS is a simple, valid score based on objective predictors reflecting the demographic and poisoning characteristics, as well as hemodynamic and mental status. However, the vital signs component of the new PMS provided a comparable result to the whole score. Therefore, determining the initial vital signs related to new PMS factors in intentional ALP poisoning offers rapid triaging for high-risk patients requiring intensive care management and excludes others not in need. Additionally, it is easily applicable as it depends on routine clinical parameters that all emergency clinicians can apply.

Footnotes

Acknowledgments

All authors express gratitude to the staff members of Tanat University Poison Control Center (TUPCC) for their support and help in this study.

Author contributions

Ghada N. El-Sarnagawy and Mona M. Ghonem formulated the study’s conception and design. Material preparation and data collection were accomplished by Ghada N. El-Sarnagawy, Amira A. Abdelnoor, and Mona M. Ghonem. The first draft was written by Ghada N. El-Sarnagawy, Amira A. Abdelnoor, and Mona M. Ghonem. All authors approved the final manuscript.

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) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Ghada N. El-Sarnagawy

Amira A. Abdelnoor

Mona M. Ghonem

References

Mwaheb

Hassan

. Fatal aluminium phosphide poisoning in fayoum governorate, Egypt (2012-2019). Egypt J Forensic Sci Appl Toxicol 2021; 21: 47–58.

Yadav

Bhattacharyya

Banerjee

. Acute aluminum phosphide poisoning: the menace of phosphine exposure. Clin Chim Acta 2021; 520: 34–42.

Deraz

Elrafey

Mesallam

. Acute aluminium phosphide poisoning in east delta, Egypt: a growing public Health problem over the last five years. Egypt Soc Clin Toxicol J 2022; 10: 49–61.

Salimi

Kheiripour

Fathi Jouzdani

, et al. Nanocurcumin improves lipid status, oxidative stress, and function of the liver in aluminium phosphide-induced toxicity: cellular and molecular mechanisms. BioMed Res Int 2022; 2022: 7659765. DOI: 10.1155/2022/7659765.

Petkovska

Babulovska

Simonovska

, et al. Fatal acute aliminium phosphide poisoning - case report and literature review with reference to current treatment protocols and outcome. Georgian Med News. 2021; 320: 111–115.

Elbadawy

Metwaly

Mokhtar

, et al. The protective role of erdosteine and exogenous reduced nicotinamide adenine dinucleotide on acute aluminium phosphide induced hepatotoxicity in adult albino rats. Biannu Forensic Sci Toxicol J 2023; 1: 24–36.

Niknahad

Heidari

Jangjou

, et al. The therapeutic effect of a novel parenteral formulation of dihydroxyacetone in aluminum phosphide-intoxicated patients. Heliyon 2023; 9: e22165.

Shaker

Rageh

OES

Alnajar

, et al. Efficacy of intravenous N acetylcysteine as an adjuvant therapy in the treatment of acute aluminum phosphide Poisoning: a systematic review and meta-analysis. BMC Pharmacol Toxicol 2023; 24: 59–110.

Katwal

Malbul

Mandal

, et al. Successfully managed aluminum phosphide poisoning: a case report. Ann Med Surg 2021; 70: 102868.

10.

Bogale

Ejigu

Muche

. Clinical profile and treatment outcome of aluminum phosphide poisoning in felege hiwot referral hospital, northwest Ethiopia: a retrospective study. Open Access Emerg Med 2021; 13: 239–248.

11.

Rageh

OES

Sabra

Alammar

, et al. Profile and outcomes of acute poisoning in the toxicology treatment and control center at Tanta University Hospital, Egypt. BMC Pharmacol Toxicol 2023; 24: 6. DOI: 10.1186/s40360-023-00650-5.

12.

Elfeky

AKE

Elgamal

Etman

, et al. Assessment of the pattern, management and outcomes of acute unstable poisoning patients in emergency department of menoufia university hospitals. Menoufia Med J. 2024; 37: 1–6. DOI: 10.59204/2314-6788.1061.

13.

Khalaf

Amin

Hishmat

. Outcomes of acute aluminium phosphide poisoning and role of magnesium sulfate as an antidote in minia poisoning control center, minia governorate, Egypt. Egypt J Forensic Sci Appl Toxicol 2023; 23: 135–148.

14.

Majidi

Jamalpour

Solmaz

. Research paper: prognostic factors of aluminum phosphide poisoning in urmia: a-five-years cross-sectional study. Int J Med Toxicol Forensic Med 2021; 11: 32663. DOI: 10.32598/ijmtfm.v11i2.32663.

15.

Sagah

Elhawary

. Prognostic significance of acid base disturbances among patients with acute aluminum phosphide poisoning. Egypt J Forensic Sci Appl Toxicol 2022; 22: 113–125.

16.

El-Sarnagawy

Abdelnoor

Abuelfadl

, et al. Comparison between various scoring systems in predicting the need for intensive care unit admission of acute pesticide-poisoned patients. Environ Sci Pollut Res Int 2022; 29: 33999–34009.

17.

Abd Elghany

Heshmat

Oreby

, et al. Evaluation of various scoring systems in prediction of acute aluminum phosphide (ALP) poisoning outcome. Ain Shams J Forensic Med Clin Toxicol 2018; 30: 117–127.

18.

Farzaneh

Ghobadi

Akbarifard

, et al. Prognostic factors in acute aluminium phosphide poisoning: a risk-prediction nomogram approach. Basic Clin Pharmacol Toxicol 2018; 123: 347–355.

19.

Pannu

Bhalla

. A simple tool predicts mortality in aluminum phosphide self-poisoning. Indian J Crit Care Med 2020; 24: 755–756.

20.

Pannu

Bhalla

Sharma

, et al. “PGI score”: a simplified three-point prognostic score for acute aluminum phosphide poisoning. Indian J Crit Care Med 2020; 24: 790–793.

21.

Pannu

Jhuria

Bhalla

, et al. PGI score: prospective validation and correlation with SOFA, SAPS-II, and Apache-II scores for predicting outcomes in acute aluminum phosphide poisoning. Toxicol Res 2022; 11: 361–366.

22.

Sakr

Atef

Shalaby

. PGI score as a predictor of cardiotoxicity and mortality in patients with acute aluminum phosphide poisoning. Zagazig J Forensic Med 2022; 21: 32–48.

23.

Han

Kim

Lee

, et al. Development and validation of new poisoning mortality score system for patients with acute poisoning at the emergency department. Crit Care 2021; 25: 29–111.

24.

Wahab

Zaheer

Wahab

, et al. Acute aluminium phosphide poisoning: an update. Hong Kong J Emerg Med 2008; 15: 152–155.

25.

Persson

Sjöberg

Haines

, et al. Poisoning severity score. Grading of acute poisoning. J Toxicol Clin Toxicol 1998; 36: 205–213.

26.

Sharma

Acharya

, et al. Extracorporeal membrane oxygenation in aluminum phosphide poisoning in Nepal: a case report. J Med Case Rep 2018; 12: 311. DOI: 10.1186/s13256-018-1864-z.

27.

Datey

Chakrabarty

Agarwal

. A retrospective descriptive study of aluminium phosphide poisoning in a tertiary care Centre of Central India. J Evol Med Dent Sci 2018; 7: 2584–2587.

28.

Seifert

Jansen

Widmer

, et al. Central-venous catheters. Catheter-Related Infections. 2nd ed. Saint Petersburg: StatPearls Publishing, 2004, pp. 293–326.

29.

Louriz

Dendane

Abidi

, et al. Prognostic factors of acute aluminum phosphide poisoning. Indian J Med Sci 2009; 63: 227–234.

30.

Elbastawesy

Elmansy

. Comparison between gastric lavage with paraffin oil versus coconut oil in acute aluminum phosphide poisoning: a randomized controlled clinical trial. Ain Shams J Forensic Med Clin Toxicol 2023; 40: 1–14.

31.

Hafez

ASAF

Elgazzar

Sobh

, et al. Gastrointestinal decontamination using oil-based solutions in patients with acute aluminum phosphide poisoning: a systematic review and meta-analysis. Crit Rev Toxicol 2024; 54: 235–251.

32.

Elsharkawy

Ghonem

El-Sarnagawy

, et al. Cardioprotective role of the coenzyme Q10 and coconut oil in acute aluminum phosphide poisoning: a randomized controlled clinical trial. Toxicol Res 2023; 12: 507–519.

33.

Fenn Buderer

Statistical methodology

. Incorporating the prevalence of disease into the sample size calculation for sensitivity and specificity. Acad Emerg Med 1996; 3: 895–900.

34.

Negida

Fahim

Negida

. Sample size calculation guide - Part 4: how to calculate the sample size for a diagnostic test accuracy study based on sensitivity, specificity, and the area under the ROC curve. Adv J Emerg Med 2019; 3: e33.

35.

Robin

Turck

Hainard

, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinf 2011; 12: 77–78.

36.

Sjoberg

Whiting

Curry

, et al. Reproducible summary tables with the gtsummary package. R J 2021; 13: 570–580.

37.

Wickham

. ggplot2. Cham: Springer International Publishing, 2016. DOI: 10.1007/978-3-319-24277-4.

38.

Ghonem

Abdelnoor

Hodeib

. Shock and modified shock indices in predicting poisoning severity and outcomes in acute aluminum phosphide poisoned patients. Toxicol Res 2024; 13: tfad124. DOI: 10.1093/toxres/tfad124.

39.

Abdelghafar

Farrag

Zanaty

, et al. Pattern and predictors of death from aluminum and zinc phosphide poisoning using multi-kernel optimized relevance vector machine. Sci Rep 2023; 13: 8268–8313.

40.

Abdelhamid

Sakr

Mostafa

, et al. Comparing the effectiveness of L-carnitine and paraffin oil in acute aluminum phosphide poisoning using predictive biomarkers and scores: a randomized controlled clinical trial. Hum Exp Toxicol 2023; 42: 9603271221149650. DOI: 10.1177/09603271221149650.

41.

Gheat

Fayed

Elgazzar

, et al. The possible therapeutic role of intravenous lipid emulsion in acute aluminium phosphide poisoning: a randomized controlled clinical trial. Toxicol Res 2024; 13: tfae090. DOI: 10.1093/toxres/tfae090.

42.

Dorooshi

Mirzae

Fard

, et al. Investigating the outcomes of aluminum phosphide poisoning in khorshid referral hospital, Isfahan, Iran: a retrospective study. J Res Pharm Pract 2021; 10: 166–173.

43.

Anbalagan

Pannu

Bhalla

, et al. Prognostic significance of poison-related factors and consumption patterns in acute aluminum phosphide poisoning. Turk J Emerg Med 2023; 23: 088–95.

44.

Saeed

Elmorsy

. Evaluation of the new poisoning mortality score in comparison with PSS and SOFA scoring systems to predict mortality in poisoned patients admitted to the intensive care unit. Toxicol Res 2024; 13: tfad113. DOI: 10.1093/toxres/tfad113.

45.

Lee

Kim

Han

, et al. Comparison of the new-poisoning mortality score and the modified early warning score for predicting in-hospital mortality in patients with acute poisoning. Clin Toxicol 2024; 62: 1–9.

46.

Moorthy

DGSRK

Priya

Rajesh

, et al. Performance of new Poisoning Mortality Score in comparison with SOFA and Apache II scores in acute organophosphate poisoning. J Emerg Med Trauma Acute Care 2023; 2023: 9.

47.

Khodabandeh

Kahani

Soleimani

. The study of fatal complications of “rice tablet “poisoning. Iran J Forensic Med 2014; 20: 27–36.

48.

Bagherian

Kalani

Rahmanian

, et al. Aluminum phosphide poisoning mortality rate in Iran; a systematic review and meta-analysis. Arch Acad Emerg Med 2021; 9: e66.

49.

Lumpe

Schurr

Rabe

, et al. Socio-demographic and psychiatric profile of patients hospitalized due to self-poisoning with suicidal intention. Ann Gen Psychiatry 2022; 21: 16–114.

50.

El-Sarnagawy

Hafez

Amer

. Characteristics of suicidal poisoned patients admitted to tertiary care center during COVID-19 pandemic. Egypt J Neurol psychiatry Neurosurg 2022; 58: 138 DOI: 10.1186/S41983-022-00577-4.

51.

El-Farouny

Helmy

. Suicidal self poisoning and its relation with personality traits among admitted cases to menoufia poisoning control center. Egypt J Forensic Sci Appl Toxicol 2021; 21: 13–30.

52.

Wang

Han

Wen

, et al. Self-poisoning with pesticides in Jiangsu Province, China: a cross-sectional study on 24,602 subjects. BMC Psychiatr 2020; 20: 545–611.

53.

Song

Park

Lee

, et al. The characteristics of elderly individuals who attempted suicide by poisoning: a nationwide cross-sectional study in Korea. J Korean Med Sci 2020; 35: e286. DOI: 10.3346/JKMS.2020.35.E286.

54.

Kasemy

Sharif

Amin

, et al. Trend and epidemiology of suicide attempts by self-poisoning among Egyptians. PLoS One 2022; 17: e0270026. DOI: 10.1371/journal.pone.0270026.

55.

El-Sarnagawy

. Predictive factors of mortality in acute aluminum phosphide poisoning: 5 Years retrospective study in Tanta poison control unit. Ain Shams J Forensic Med Clin Toxicol 2017; 29: 70–79.

56.

El-Sarnagawy

Shaban

Lashin

. Validation of corrected and dispersed QT as predictors of adverse outcomes in acute cardiotoxicities. Cardiovasc Toxicol 2022; 22: 1–13. DOI: 10.1007/s12012-020-09629-8.

57.

Ghalavand

Javanshir

Nahandi

, et al. Prevalence and life-threatening electrocardiographic changes in aluminum phosphide poisoned patients. J Iran Med Counc 2023; 6: 514–524.

58.

Wahdan

Khalifa

. Clinical data, laboratory investigations and electrocardiographic changes as predictors of mortality in acute aluminum phosphide poisoning. Mansoura J Forensic Med Clin Toxicol 2020; 28, 111–123.

59.

Sheta

El-Banna

Elmeguid

, et al. A study of the predictive factors of mortality in acute poisoning with aluminum phosphide with special reference to echocardiography and SOFA score. Environ Sci Pollut Res 2019; 26: 33135–33145.

60.

Abdelkader

Elsayed

Shokry

. Pattern and outcome of acute aluminium phosphide intoxicated cases at sohag university hospitals: special consideration to paraffin oil as a supportive treatment. Zagazig J Forensic Med 2023; 21: 133–156.

61.

Erfantalab

Soltaninejad

Shadnia

, et al. Trend of blood lactate level in acute aluminum phosphide poisoning. World J Emerg Med 2017; 8: 116–120.

62.

Ghonem

El Sharkawy

Lashin

. Predictive variables of acute aluminum phosphide poisoning outcome: a new proposed model. Egypt J Forensic Sci Appl Toxicol 2020; 20: 45–60.

63.

Mohammadinejad

Rizi

Oskuee

, et al. Development of detection methods for the diagnosis and analysis of highly toxic metal phosphides: a comprehensive and critical review. Biotechnol Appl Biochem 2022; 69: 1121–1147.

64.

Banagozar Mohammadi

Zaare Nahandi

Solhi

. Utility of silver nitrate test for post-mortem diagnosis of aluminum phosphide poisoning. Iran J Toxicol 2013; 7: 804–807.