Evaluation of Various Scoring Systems in Prediction of Acute Aluminum Phosphide ( ALP ) Poisoning Outcome

Introduction: Aluminum phosphide (Alp) poisoning is a major health problem developing countries because of the high rates of morbidity and mortality even in well-equipped and experienced hospitals.. The aim of this study: Was to evaluate various scoring system (Acute Physiology and Chronic Health Evaluation (APACHE) score, Rapid Emergency Medicine Score (REMS), The Sequential Organ Failure Assessment (SOFA) score) at admission for outcome prediction in acute aluminum phosphide poisoned patients. Patients and methods: The present study is a cohort study that was carried out on acute Alp poisoned patients at Poison Control Unit, Tanta University Emergency Hospital. It was composed of two periods; one year retrospective (from April 2015 to March 2016) and six months prospective (from April 2016 to September 2016).All admitted patients over 16 years with acute aluminum phosphide poisoning were included in this study. Patients were grouped according to their outcome into survivors and non survivors. From the collected data, APACHE II, SOFA and REMS scores were calculated for every patient. Results: The present study was carried out on admitted fifty acute Alp poisoned patients. Out of them, survivors represented 44% and 56% were non-survivors. The majority of patients were in the age group 16-20 years, females, student, from rural areas (90%) and most of patients had ingested phosphides intentionally in a median dosage of one tablet. Hypotension, tachypnea, metabolic acidosis, hyperglycemia and higher serum creatinine levels and alanine transaminase (ALT) at admission, were risk factors of mortality from Alp. Non survivors had significantly higher APACHE II, SOFA and REMS score values than survivors. SOFA score had the best AUC (0.989) followed by APACHE II score then REMS score (0.987 and 0.970 respectively). However, no significant differences between AUC values of compared scores could be demonstrated. Conclusion: The clinical scores (APACHE II, SOFA and REMS) were similar and effective tools for determination of the severity of acute Alp poisoning. However, REMS proved to be more applicable than other scores owing to its simplicity, less time-consuming and effectiveness in emergency situations. Therefore, REMS score is suggested to be used in the emergency situations to predict outcome in Alp poisoned patients.


Introduction
luminum phosphide (Alp) is known as "Rice tablet" which is a solid pesticide that rapidly became one of the most commonly used grain fumigants because of its ideal properties; easy application, high efficacy, toxic to all stages of insects, does not affect seed viability, free from toxic residues and leaves little residue on food grains (Bogle et al., 2006 andWahab et al., 2009).
The majority of intoxicated cases are intentionally while accidental exposure by inhalation of phosphine gas is less commonly recorded (Shadnia et al., 2008).
The fatal dose of aluminum phosphide poisoning is 0.15-0.5 gram (wahab et al., 2009).Mortality rate in various studies ranged from 37-100 % (Goel and Aggarwal, 2007(.The exact mechanism of action of aluminum phosphide poisoning is still unknown.However, phosphine gas (PH3) is liberated when Alp comes in contact with moisture.Then it is rapidly absorbed causing systemic complications and multi-organ failure through suppression of cytochrome oxidase resulting in histotoxic hypoxia and production of free radicals (Proudfoot, 2009 andAbedini et al., 2014).
Acute Physiology and Chronic Health Evaluation (APACHE II) score was described by Knaus et al in 1985.It is the most frequently used scoring system by physicians because it is reliable, inexpensive and produces accurate results in prediction of the patients' outcomes.However, it includes several blood chemistry variables making it not suitable for quick scoring in the emergency situations (Ratanarat et al., 2005).
Rapid Emergency Medicine Score (REMS) is an attenuated version of APACHE II.It is simple (it does not require invasive or time consuming laboratory values) that allows rapid calculation in emergency conditions (Olsson et al., 2004 andGoodacre et al., 2006 ).
The Sequential Organ Failure Assessment (SOFA) Score was created in a consensus meeting of the European Society of Intensive Care Medicine in 1994 and further revised in 1996.It determines the extent of a person's organs function or rate of failure (Ferreira et al., 2001).
In the field of clinical toxicology, only few studies had evaluated the efficacy of scoring systems in predicting the outcome of acute aluminum phosphide poisoned patients (Mathai and Bhanu, 2010).So, the aim of this study was to evaluate various scoring system (APACHE II, REMS, SOFA) at admission for outcome prediction in acute aluminum phosphide poisoned patients.

Study design
The present study is a cohort study that was carried out on ALP poisoned patients who were admitted at Poison Control Unit, Tanta University Emergency Hospital.It was composed of two periods; one year retrospective (from the start of April 2015 to the end of March 2016) and six months prospective (from the start of April 2016 to the end of September 2016).
The study was carried out following approval of the Medical Research Ethical Committee of Tanta Faculty of Medicine.In retrospective cases, patients᾿ data were collected from patients᾿ files after taking the permission of the head of the poison control unit.In prospective cases, a written informed consent was taken from every patient or his/her guardians.The privacy and confidentiality of patients᾿ data were considered by making code number for every patient.

Patients Inclusion criteria
All admitted patients over 16 years with acute aluminum phosphide poisoning were included in this study.Diagnosis of acute aluminum phosphide poisoning was based on history of exposure to Alp, the suggestive clinical manifestations (symptoms and sign) and silver nitrate test.
From the collected data, the following scores were calculated for every patient: APACHE II score was calculated from 12 routine physiologic values; heart rate, mean arterial pressure, body temperature, respiratory rate, oxygenation of arterial blood (PaO2), arterial pH, white blood count, hematocrit value, serum sodium level, serum potassium level, serum creatinine level and Glasgow Coma Scale (GCS).The score for each parameter was assigned from 0 to 4, with 0 being normal and four being the worst value.The sum of these values were added to a mark adjusting for patient age and a mark adjusting for chronic health problems (severe organ insufficiency or immune-compromised patients) to arrive at the APACHE II score which ranged from 0 to 71 (Knaus et al., 1985 andSungurtekin et al., 2006).
Rapid Emergency Medicine Score (REMS) composed of 6 variables; pulse rate, mean arterial pressure, respiratory rate, GCS, peripheral oxygen saturation and age.Age was graded a score from 0 to 6 and the remaining five variables were assigned a score from 0 (normal) to 4 (the worst value) providing a daily score from 0 to 26 (Goodacre et al., 2006).

Sequential Organ Failure Assessment (SOFA)
score assessed the function of six organs; respiratory (PO2 divided by FIO2), renal (serum creatinine), hepatic (serum bilirubin), cardiovascular (mean Arterial Pressure and the dose of the administrated vasopressors graded from 2 to 4, hematologic (platelet count), neurologic (Glasgow Coma Score).Each organ was graded from 0 to 4 providing a daily total score from 0 (normal) to 24 (the worst value) point (Vincent et al., 1998 andJones et al., 2009).
After then, patients were grouped according to their outcome into survivors and non survivors Statistical analysis For quantitative data, the Shapiro-Wilk test for normality was performed.For data that were not normally distributed, median and interquartile range (expressed as 25th-75th percentiles) were calculated and Mann-Whitney test was used.For normally distributed data, values were expressed as mean ± standard deviation and Independent samples T test was performed for comparison between two groups.For qualitative data Pearson's Chi square test was used to examine association between two variables.Receiver operation characteristics (ROC) curve analysis was carried out to predict outcome.Areas under ROC curve, sensitivity, and specificity, were calculated.Significance was adopted at p < 0.05.

Results
The present study was carried out on fifty acute Alp poisoned patients.Out of them, survivors represented 44% and non-survivors represented 56%.
Table (1) showed that, the majority of patients (64%) were in the age group 16-20 years with a significant association with poor patients' outcome.The highest percentage of patients were females, students and from rural areas.
In the present study, the majority of patients (94%) had ingested phosphides intentionally in a suicidal attempt.The median amount was one tablet.The delay time post ingestion ranged from 0.75 to 14 hours and the median of length of hospital stay was 2 days.There was a significant association between the route of intoxication, delay time, hospital stay and mortality (Table 2).
Table (3) illustrated that, the majority of patients were hypotensive (52%).Tachycardia was detected in 44% of patients while, bradycardia was noticed in 6% of patients and 4%of patients had undetected pulse clinically.Tachypnea was detected in 72% of all patients and in 86% of non survivors.The majority of patients had normal temperature (64%).There was significant association between patients' outcome and each of blood pressure, respiratory rate and temperature.
The majority of patients (68%) had normal level of consciousness.12% of cases were presented with agitation, 56% with GIT disturbance (vomiting and abdominal pain) and only one case had wheeze.
Table (4) revealed that non survivors had significantly higher level of serum ALT and serum creatinine level than survivors.The arterial blood gases (arterial PH, partial oxygen pressure (PO2), partial carbon dioxide pressure (PCO2) and bicarbonate level (HCO3) decreased in non survivors compared to survivors.Serum sodium and random blood sugar were significantly raised in non survivors group when compared to survivors.
Regarding complication, fifteen patients needed mechanical ventilation, four patients were only intubated and Adult Respiratory Distress Syndrome (ARDS) developed in one patient.
Concerning the studied scores (APACHE, SOFA and REMS) in the current study, non survivors had significantly higher APACHE II, SOFA and REMS score values than survivors (Table5).
Table (6) illustrated that, APACHE II score had 92.9 % sensitivity and 90.9% specificity at a cut off value ≥ 9.5.SOFA score ≥ 5.5 can predict the mortality with sensitivity 96.4% and specificity 95.5 %.Furthermore, REMS score≥ 4.5 was predictor of mortality; the sensitivity and specificity was 89.3 % and 95.5 % respectively.
SOFA score had the best AUC (0.989) followed by APACHE II then REMS (0.987 and 0.970 respectively).However, Their AUC values didn't show significant difference (Figure 2).

Discussion
The present study revealed that 28 patients out of 50 patients died with overall mortality rate of 56%.This finding coincide with Taramsari et al., (2013) in Iran ;and Mathai& Bhanu, (2010) in India who reported that, mortality rates of Alp poisoning were 58.61%and 60% respectively.The high mortality rate could be attributed to extremely toxic property of Alp and absence of a specific antidote.
In the current study, the majority of patients (64%) were in the age group 16-20 years, followed by the age group 20-30 years (26%).This is in agreement with the findings of El-Ebiary et al., (2015) who reported that, 67.50% of Alp poisoned patients were < 20years followed by age group 20-40 years.Moreover, the present study indicates that most of the cases were students (48%); this could be explained by the majority of the studied patients in the present study were younger than 30 years.This is also in agreement with Dash et al., (2005) who reported that the majority of patients in their study were students at school.The high incidence of acute intoxication with Alp in young age groups (10-29 years) could be explained by that younger persons tend to be easily excited, depressed, failure in love, family scolding from parents or teachers, more exposed to failure in education and lack of employment opportunities problems (Kapoor et al., 2006;Khurana et al., 2011 andChaudhary et al., 2013).
Females of this study represented 66% of the studied patients and 94% of cases were intentionally ingested phosphides in suicidal attempts.This coincides with the findings of Brahim et al., (2013) who reported that, 63.8% of patients intoxicated with Alp were females and 95.7% of cases were suicidal.El Nagger and El Mahdy, (2011) referred this to the high stress in women's life caused by marital disharmony, economic hardship, social problems and scolding from other family members.
The majority of patients (90%) in this study were from rural areas.This is partially in line with the findings of Karamjit et al., (2003) and Kapoor et al., (2006) in India who demonstrated that, 65.12% and 64% of Alp poisoned patients were from rural areas respectively.The high incidence of acute intoxication with Alp in rural areas could be attributed to farming activity in rural areas with easy access to toxic pesticides and lack of awareness about its hazards (El Naggar and El Mahdy, 2011).
The median delay time in this study was 3 hours.This coincides with Wahdan and Elmadah, (2016) in Tanta.The short duration between exposure and arrival to the hospital could be explained by the presence of the Tanta Toxicology Unit in the center of Delta region with easy transportations and many of the cases who attempted suicide did not actually intend to end their lives but they just try to draw attention, gain sympathy from their families, so they rapidly seek medical support.
More than half of patients in the present study were hypotensive (52%), non survivors were more hypotensive than survivors.Teimoory et al., (2013) and El-Ebiary et al., (2015) reported that hypotension was detected in 92.0%and80% of their cases respectively.Furthermore, Soltaninejad et al., (2012) concluded that systolic blood pressure is a key risk factor for prediction of Alp poisoning mortality at admission to the hospital.The mechanism of shock is multifactorial (myocardial damage, fluid loss and adrenal gland damage) (Proudfoot., 2009).Clinical, biological and electrical observations suggest that, the main cause of hemodynamic failure was myocardial lesion (Siddaiah et al., 2009).
Tachypnea was detected in 72% of all patients.This in accordance with Jaiswal et al., ( 2009 ) andEl Ebiary et al.,(2015) who found that 100% and 70% of their patients had tachypnea respectively.Tachypnea could be a compensatory response to metabolic acidosis disorder.
The current study revealed significant elevation of serum ALT and serum creatinine in non survivors when compared to survivors.This is in accordance with Mathai & Bhanu, (2010) and Masoud & Barghash, (2013) These effects could be explained by the fact that, phosphine gas cause inhibition of cytochrome oxidase and free radical release that damage various organs especially those with the greatest oxygen requirement (Masoud and Barghash, 2013).
All parameters of ABG were decreased in non survivors than survivors.Mehrpour et al., (2008) and Mostafazadeh et al., (2011) recorded a significant association between metabolic acidosis and mortality.Accumulation of lactic acid that caused by blockage of oxidative phosphorylation and poor tissue perfusion may be the cause of metabolic acidosis (Agarwal et al., 2014).In the present study, partial oxygen pressure (PO2) was significantly decreased in non survivors when compared to survivors.Accordingly, Wahab et al., (2008) revealed that hypoxia is associated with high mortality rate.Valmas et al., (2008) and Proudfoot, (2009) stated that, phosphine gas inhibits 70% of mitochondrial oxidative respiration by suppression of cytochrome oxidase resulting in histotoxic hypoxia.
Regarding serum electrolytes, serum sodium was significantly increased in non survivors group compared to survivors which is agreement with the findings of Shadnia et al., (2010), This study revealed that non-survivors had a statistically significant increase in random blood sugar (the median was 158.50 mg/dl) when compared with survivors.In accordance with that, Mehrpour et al., (2008) stated that, Alp poisoned patients with glucose levels greater than 140 mg/dl at admission were at high risk of mortality.Suggested mechanisms for this hyperglycemia include impaired oxidative phosphorylation and glucose utilization with involvement of the adrenal axis and/or pancreas (Singh et al., 2006 andVerma et al., 2007).However, previous study by Mehrpour et al., (2008) reported that hypoglycemia in aluminum phosphide poisoning is a rare finding and it is dangerous.Hypoglycemia might be due to impairment of glycogenolysis and gluconeogenesis secondary to adrenal gland damage and decrease circulating cortisol level (Chugh et al., 2000).Moreover, Mathai&Bhanu, (2010) and Masoud&Barghash, (2013) did not find any significant association between random blood sugar and mortality.
Various scoring systems have been developed during the last few decades to evaluate the severity of illness and to predict patients' outcome (Kelly et al., 2002).
APACHE II score is a general mortality prediction model in critically ill patients (Strand and Flaatten, 2008).Non survivors of the present study had significantly higher APACHE II score values at admission than survivors (medians = 15 vs 3.5 respectively).The ability of APACHE II to predict patients who died was 92.9 % and patients who survived was 90.9% at ≥ 9.5 cut off value.In line with this, Mathai and Bhanu, (2010) found that, survivors of Alp poisoned patients had significantly lower score than non-survivors (8.64 ± 5.27 vs 14.56 ± 6.66, respectively).Moreover, Hajouji et al., (2006) revealed that APACHE II was positively correlated with poor outcome in Alp poisoned patients.
In the current study, REMS score had higher values in non survivors when compared to survivors (median was 8.0 vs. 2.00 respectively), with a sensitivity and specificity, 89.3 % and 95.5 % respectively at a cut off value ≥ 4.5.No other studies evaluated the use of REMS as a predictor of outcome in Alp poisoned patients.
Other previous studies by Olsson et al., (2004), Cattermole et al., (2009) and El-Sarnagawy and Hafez, (2017) reported that REMS was a good predictor of bad outcome in ICU admitted patients, long term mortality in patients attending non-surgical emergency department and the need of mechanical ventilation in drug-overdosed patients with disturbed conscious level respectively.
Regarding SOFA score, it was developed as a tool for assessment of organ failure.The present study revealed that SOFA score showed higher value in non survivors than survivors with median of 9.00 vs 1.00 respectively with 96.4 % sensitivity and 95.5 % specificity at a cut off value ≥5.5.To the best of author's knowledge, no previous studies evaluated the use of the SOFA as a predictor of outcome in Alp poisoned patients.Vosylius et al., (2004) and Halim et al., (2009) demonstrated that, SOFA score was significantly higher in non-survivors than survivors in surgical &intensive care patients and in patients with severe sepsis respectively.
The current study demonstrates that, SOFA score had the best A UC (0.989) followed by APACHE II score then REMS score (0.987 and 0.970 respectively.However, no significant differences between AUC values of compared scores.APACHE II and SOFA includes several blood chemistry variables.Subsequently, they are not suitable for quick evaluation in the emergency conditions.On contrary, REMS score is a simple clinical score (neither requires any laboratory parameters nor any staff training and extra skills).

Conclusion
Aluminum phosphide is a remarkably toxic compound with a high mortality rate reached in the current study to 56%.The presence of hypotension, tachypnea, metabolic acidosis, hyperglycemia and higher serum creatinine and ALT levels at admission, were risk factors of mortality from Alp poisoning.The clinical scores (APACHE II, Sofa and REMS) were similar and effective tools for determination of the severity of acute Alp poisoning.However, REMS proved to be more applicable than other scores owing to its simplicity, less time-consuming and effectiveness in emergency situations.Therefore, REMS is suggested to be used in the emergency situations to predict outcome in Alp poisoned patients.

Table ( 6): The best cut-off values, sensitivity, specificity, and AUC of the applied scoring systems APACHE, SOFA, and REMS Variables Cut-off value Sensitivity % Specificity % Accuracy % ROC-AUC P value
*significant at p <0.05, AUC: area under the curve, APACHE: Acute Physiology and Chronic Health Evaluation II, SOFA: Sequential Organ Failure Assessment, REMS: Rapid Emergency Medicine Score