This was a single institutional retrospective cohort study performed by chart review. This study was conducted at Chonnam National University Hospital, a tertiary care teaching facility comprising an emergency department (ED) that provides care for approximately 45,000 patients per year.

The inclusion criteria consisted of 18 yr of age or older who presented to our ED after pyrethroid ingestion between 2003 and 2014 and underwent the glucose testing at presentation. The diagnosis of pyrethroid ingestion was made based on the history of pyrethroid ingestion that was provided by the patients or the witness, and the respective hospital records with the physician's notation of "pyrethroid" after he had checked the bottle brought by the patients or the witness. Through this review, 132 patients were identified.

The exclusion criteria consisted of mixed ingestion, including ethanol, type I pyrethroid ingestion and a history of renal failure. We excluded patients who had a history of co-ingestion of pyrethroid with ethanol because ethanol can influence glucose levels (11). Additionally, patients who were given fluids containing glucose before the measurement of glucose levels at presentation were excluded. Thus, data from a total of 104 patients were included in the analysis.

The medical records were reviewed by an assistant, who was unaware of the purpose of the study, and the following data were collected: age, gender, comorbidity, chemical compound and the amount of type II pyrethroid ingested, history of administration of gastric lavage or activated charcoal, the time interval from ingestion to arrival at our ED, initial vital signs, electrocardiographic (ECG) findings, blood glucose level at presentation, results of laboratory tests, complications, length of hospitalization, and the cause of death. The amount of type II pyrethroid ingested was estimated as a spoonful (5 mL), a mouthful (25 mL), a cup (100 mL), and a bottle (300 mL) (12). The initial blood glucose level at presentation was obtained by performing a routine blood test. Hyperglycemia was defined as a glucose concentration >200 mg/dL (13). ECGs were interpreted by an emergency physician who was blinded to the patients' clinical data. The corrected QT interval (QTc) was calculated using the Bazett's formula (QT divided by the square root of RR). The prolongation of a QTc interval was defined as a QTc interval greater than 440 ms. Medical complications included the following: acute kidney injury, respiratory failure, hypotension (the need for pressor support to maintain blood pressure after admission), acidosis (pH<7.20), seizures, cardiac arrest or death. Acute kidney injury was defined as an absolute increase in serum creatinine (Cr) of ≥0.3 mg/dL (26.52 µM/L) or a percentage increase in the serum Cr≥50% from baseline. Respiratory failure was defined as the need for mechanical ventilation. In addition, any other conditions encountered during the review of patient medical records that were deemed to be serious and clinically significant by the investigator were considered as medical complications.

To demonstrate the association between hyperglycemia at presentation and patient outcome, the patients were divided into two groups according to the presence of DM, because the cut off value of blood glucose level differs according to the presence of DM in other acute illness (14). The presence of DM was determined based on a previous clinical or biochemical diagnosis of DM, or treatment with oral hypoglycemic agents or insulin. Also, when the fasting plasma glucose level was higher than 120 mg/dL or the postprandial plasma glucose level was 200 mg/dL at discharge, the patients were considered as having DM. Then the patients were subsequently divided into two groups based on the development of complications. Patients who developed a medical complication during hospitalization were assigned to the complicated group.

The predictive value of hyperglycemia at presentation was determined by univariate analysis. Also, for identifying other predictive factors associated with a high risk for complications, objective variables that could be easily assessed at presentation or that were previously reported as predictor variables such as age, intentional ingestion, Glasgow coma scale (GCS), radiologic findings, and WBC were also evaluated (15). However, variables such as the result of arterial blood gas analysis, and systolic blood pressure were excluded because they were used to define complications such as hypotension and acidosis, and also because the patients who had these conditions at presentation were assigned to the complicated group. Univariate analysis of the association between the covariates and the outcome was performed. Covariates with P values less than 0.05 were considered sufficient for the inclusion of a variable in the multivariate logistic regression analysis. Odds ratios (ORs) and confidence interval (CI) were calculated for all significant variables. All statistical analyses were performed using the Statistical Package for the Social Sciences version 21.0.

The study design was approved by the institutional review board at Chonnam National University Hospital (IRB No. CNUH-2014-248). Informed consent was exempted by the board.

Baseline characteristics and clinical course of the 104 patients who were enrolled in this study are summarized in Table 1. The complication rate and mortality rate was 26.9% and 2.9% in type II pyrethroid poisoned patients. Eleven (10.6%) patients had DM. Diabetic patients had a higher frequency of complications compared to non-diabetics patients (63.6% in diabetics vs. 22.6% in non-diabetics, P value=0.004), but there was no significant difference in mortality between the two groups. (9.1% in diabetics vs. 2.2% in non-diabetics, P value=0.193).

In the 93 non-diabetic patients, complicated patients showed a higher frequency of hyperglycemia, abnormalities on the initial radiograph, depressed mentality (GCS≤13), lower PaCO₂, and HCO₃^- levels, and higher WBC and AST levels at the time of admission than non-complicated group (Fig. 1, Table 1).

In the 11 diabetic patients, complicated group had a higher frequency of depressed mentality and a lower blood urea nitrogen (BUN) level at admission. However, the glucose level at presentation and frequency of hyperglycemia did not differ according to the development of complications (Fig. 1, Table 1).

Twenty eight (26.9%) patients experienced at least one complication during hospitalization, and the most common complications were respiratory failure (21.2%) (Table 2). The incidences of acidosis, hypotension, and renal failure were significantly higher in diabetic patients than in non-diabetic patients.

All of the complications developed within 24 hr after presentation. Three patients died because of refractory hypotension and azotemia accompanied by hyperkalemia and metabolic acidosis. Cardiac arrest developed in a 26-yr-old male who had ingested 600 mL of cypermethrin. He underwent a therapeutic hypothermia, and regained an alert mentality with a normal lifestyle at discharge.

In the 93 non-diabetic patients, hyperglycemia, GCS ≤13, WBC, AST, and X ray abnormality were included in multivariate analysis. The ingested amount variable was not included in the multivariate analysis, even though a significant difference was found for the development of complications in a univariate analysis, because it is possible the reported information is inaccurate. Hyperglycemia at presentation and GCS ≤13 were independent factors for predicting complications in non-diabetic type II pyrethroid poisoned patients. The ORs for hyperglycemia were higher than the ORs for GCS score (Table 3).

However, in 11 diabetic patients, none of the investigated variables showed a significant difference in the development of complications in multivariate analysis probably because of small number of diabetic patients (data not shown).