Fasting plasma glucose is an independent predictor for severity of H1N1 pneumonia

Data were collected from suspected H1N1-infected patients who had complications of pneumonia in the Emergency Ward of the Second Hospital of Harbin Medical University in 2009. The diagnostic criteria for suspected H1N1, which were similar to those of the Centers for Disease Control and Prevention (CDC) criteria [7], were defined by the Chinese Ministry of Health: fever, known exposure to H1N1, respiratory symptoms, and infiltrates observed on chest radiography [8]. Pharynx swab samples of all suspected cases were sent to the provincial CDC, where H1N1 virus infection was detected and confirmed by real time PCR.

In this retrospective analysis, the clinical and biochemical characteristics of 34 patients positive for H1N1 and 67 patients with non-H1N1 pneumonia were compared. Patients were excluded if they were younger than 16 years of age, were pregnant, had a history of diabetes, or died. Data from laboratory examinations were collected on admission and before steroid treatment, including FPG, glycosylated hemoglobin (HBA1c), function tests for liver and kidney, myocardial enzymes, blood cells, and arterial blood gas analysis. Arterial oxygen saturation (SaO₂) < 93% (measured on supplemental oxygen) was also used to classify cases as 'serious'; the duration of a hospital stay was the main indicator to judge severity of the disease. Plasma insulin (FINS) was randomly tested among 21 H1N1-positive and 31 H1N1-negative patients at the same time of FPG determinations.

To account for possible H1N1 effects on pancreatic β-cell function, a homeostasis model assessment (Homa) and IAI were used to measure insulin resistance and sensitivity [9, 10]. These included: an insulin resistance index: Homa-IR = Ln(FPG × FINS/22.5); β-cell function index: Homa-β = 20 × FINS/(FPG-3.5); and insulin sensitivity index : IAI = -Ln(1/(FPG × FINS). This study was approved by the Ethics Committee of the Harbin Municipal Bureau of Health.

Upon admission, all patient blood samples had been screened with hematology, biochemistry, and virology tests. All patients were empirically administrated antivirals, broad-spectrum antibiotics, and mechanical ventilation therapy before certification of viral or bacterial infections. Baseline and serial chest radiographs and CT scans, if necessary, were performed for all suspected or confirmed H1N1 patients. For suspected or confirmed H1N1 patients with severe pneumonia, as indicated by persistent fever and deteriorating lung opacities on chest X-ray, intravenous therapy with methylprednisolone (120-500 mg every day for 5 days followed by a decreasing dose based on the patient's condition) was used as necessary treatment.

Results are given as means ± SDs or percentages and compared between different groups by independent t-tests or ANOVA, as appropriate. Associations between FPG and H1N1 virus infection, expressed as Odds Ratios (OR) and 95% confidence intervals (CI), were examined by logistic regression analysis after controlling for sex, age, and BMI. Correlation analysis was used to evaluate possible relationships between FPG and biochemical test results. Associations between number of admission days and significant data among the two groups were further assessed by single factor linear regression. Associations between FPG and SaO₂ were assessed by correlation and regression analyses. To evaluate possible H1N1 effects on pancreatic β-cell function, homa-IR, homa-β, and IAI were compared by t-tests between H1N1 and non-H1N1 groups after adjusting for age, gender, and BMI. All analyses used SPSS version 13.0 (SPSS Inc., Chicago, IL, USA). P < 0.05 (two-tailed) was considered to be statistically significant.

All data for the patients diagnosed as suspected H1N1 and confirmed H1N1 were collected in the Emergency Ward of the Second Hospital of Harbin Medical University in 2009. All patients were Han Chinese, except for one Man Chinese, and all were from Harbin City or surrounding areas. Table 1 summarizes the clinical and biochemical characteristics of the H1N1 and non-H1N1 patients. There were no statistically significant differences for age, gender, or BMI between the two groups. There were no detectable differences in WBC counts and differential counts, and hepatic and renal functions between the two groups.

However, significant differences of biochemical exam results were found in plasma glucose levels (7.93 ± 3.44 vs. 6.13 ± 1.54, H1N1 vs. non-H1N1; p < 0.01), LDH (1139.63 ± 1086.24 vs. 501.59 ± 613.70, H1N1 vs. non-H1N1; p < 0.01), BUN (6.57 ± 4.46 vs. 4.54 ± 2.42, H1N1 vs. non-H1N1; p < 0.05), and SaO₂ (86.29 ± 13.23 vs. 93.23 ± 7.30, H1N1 vs. non-H1N1; P < 0.05). The number of admission days for H1N1 patients were obviously longer than for non-H1N1 patients (17.65 ± 8.28 vs. 11.91 ± 8.77, H1N1 vs. non-H1N1; p < 0.01). Logistic regression analysis indicated significant positive associations between FPG (OR = 1.377, 95%CI:1.062-1.786; p = 0.016), LDH (OR = 1.001, 95%CI: 1.000-1.002; p = 0.010), BUN (OR = 1.184, 95%CI: 1.024-1.369; p = 0.022) and H1N1 virus infection.

Defining SaO₂ of 93% as a threshold, we further investigated the relationship between FPG and SaO₂. Those with high FPG levels (≥7 mmol/L) had a significantly lower SaO₂ level than those with low FPG levels (<5.6 mmol/L) among H1N1 patients (Figure 1). The association between FPG and SaO₂ is shown in Figure 2. In the H1N1 group, patients with hypoxia (SaO₂< 93%) had remarkably higher FPG levels than those of patients with SaO₂ ≥ 93% (Table 2). Elevated FPG also paralleled a trend for prolonged hospitalization, although there was no correlation between FPG and hospitalization duration (data not shown). Figure 3 shows the relationships between FPG levels and other biochemical indicators in the H1N1 group. Significant Pearson correlation coefficients were found between FPG and AST (r = 0.215; p = 0.039), LDH (r = 0.400; p = 0.000), BUN (r = 0.28; p = 0.005), and SaO₂ (r = -0.416; p = 0.000).

Random determinations of FPG and FINS were made for suspected H1N1 patients on admission; information was collected for 21 confirmed-H1N1 patients and 34 non-H1N1 patients. The possible effects of H1N1 on pancreatic islets function was evaluated by calculating an insulin resistance index (homa-IR), an insulin function index(homa-β), and an insulin sensitivity index(IAI). The mean FPG level was significantly higher in the H1N1 group (8.29 ± 3.57 vs. 6.22 ± 1.62, H1N1 vs. non-H1N1; p < 0.05; Table 3). However, no statistically significant differences were found for homa-IR, homa-β, or IAI between the H1N1 and non-H1N1 groups after adjusting for age, gender, and BMI (Table 3).

In the H1N1 group, the FPG levels of patients with hypoxemia (SaO₂ <93%) were higher than those of patients without hypoxia (9.82 ± 4.14 vs. 6.64 ± 1.78; p < 0.05; Table 2). After adjusting for age, gender, and BMI, a significant negative correlation was found between FPG levels and SaO₂ in the H1N1 group with SaO₂ < 93%(r = -0.497; p = 0.041). The mean course of H1N1 patients with SaO₂ ≥ 93% was 12.26 ± 7.63 d, whereas for patients with SaO₂<9 3% it was 19.22 ± 11.15 d; there was a significant difference between the 2 groups (p < 0.05). In addition to FPG, some biochemical indicators that might influence or predict the disease course were analyzed by linear regression analysis. These results showed that number of hospitalization days were correlated with LDH (β = 0.000297 ± 0.0000; p = 0.049), BUN (β = 0.0891 ± 0.036; p = 0.013), PaO₂ (β = -0.0119 ± 0.004; p = 0.003), and SaO₂ (β = -0.0287 ± 0.013; p = 0.036).