الفهرس 
Title pageOnly 14 pages are availabe for public view
 |  | from | 197 |  |  |
| | | from | 197 | | |
Abstract
On NICU admission, antibiotics are commonly prescribed empirically to neonates with suspected infection or prophylactically with high risk of infection. Overuse of antibiotics and subsequent development of infections caused by MDRO threatens the achievements of modern medicine. MDR neonatal infections are therapeutically challenging, associated with increased mortality, extra cost, and longer NICU stay. Many factors influence clinicians’ antimicrobial decision making, they range from diagnostic uncertainty, lack of local updated evidence-based protocols and fear of poor prognosis.
We are unaware of resistance pattern of common pathogens causing infections in El-Shatby NICU, incidence rate to MDR infections and related mortality rates. We do not know if using more than one line of antibiotic therapy in NICUs increases the risk of developing infections caused by MDRO.
6.1.2 Objective
This study was conducted to evaluate pattern of antibiotic therapy (one line versus two or more lines) as regards its role as a risk factor for developing MDR in neonates admitted to El-Shatby NICU at Alexandria University Children Hospital. To identify antibiotic strategies, common pathogens, and their resistance pattern, and to estimate incidence rate of bacterial MDR and survival rate among those developing MDRBS.
6.1.3 Methods
Study Design: A retrospective cohort study
Study Setting: NICU of Alexandria University Children Hospital “El-Shatby”
Study Subjects: Neonates admitted to El-Shatby NICU between December 2017 and December 2019, survived for at least one day and received antibiotic therapy
Sample size: Hypothesizing that prevalence of MDR among NICU admitted neonates is 2% and accepting 1% margin of error, a minimum sample size of hypothetical cohort of 753 neonates was considered sufficient to achieve 80% power with 95% confidence level.
Data collection: Data of neonates who received antibiotics were extracted from a random sample of 754 medical records using two data collection forms. Collected data included personal, admission, maternal neonatal, culture, sepsis, and antibiotic data and outcomes (mortality, NICU length of stay). The collected data was analyzed using Excel® for Microsoft® 365, and SPSS® for Windows, version 25. WHONET® software version 20.16.15. was used to calculate percentages of susceptible, intermediate, and resistant isolates, to generate a cumulative antibiogram and frequencies of different types of resistance; MDR, XDR and PDR. “Bacterial isolates non-susceptible to at least one agent in three or more antimicrobial categories were considered MDR. While those susceptible to only one or two
categories were considered XRD and finally those non-susceptible to all tested agents in all tested antimicrobial categories were considered PDR.”
Statistical analysis: Unadjusted association between MDRBS and categorical variables was tested using chi-square test or independent t-test. Logistic regression analysis was applied for the assessment of bacterial MDR prediction factors. Significant variables from univariate analysis were tested to detect factors that can predict the probability a neonate will develop bacterial MDR. Kaplan-Meier analysis was performed to estimate the probability of developing EOMDRBS and LOMDRBS infection. Comparison of outcomes between neonates with versus without exposure to multiple lines of antibiotic therapy were made using Cox proportional hazards model. Probability decision-model analysis was created for the occurrence of MDR among those receiving different lines of antibiotic therapy. Analysis of antibiotic prescribing practice was done using AHRQ framework “4 moments of antibiotic decision making”.
6.1.4 Results
Neonates in our study stayed for at least one day in the NICU. Received empiric antibiotics within the first three days of life. Male neonates constituted 53.3% of the study sample. Study neonates were classified into two cohorts; exposed cohort included those who received ≥ two-lines of antibiotic therapy were 51.2% of study neonates and non-exposed cohort which included those who received one-line of antibiotic therapy were 48.8%. Both cohorts were followed up for three days for culture confirmed MDR EOS. The probability of developing EOMDRBS was 1% among neonates who received one line of antibiotic therapy all were confirmed in the first 24 hours after birth versus 5% among neonates who received two or more lines confirmed throughout the three days of follow up. This difference was statistically significant. Both cohorts were also followed up for 60 days for culture confirmed multidrug resistant late onset sepsis. The probability of developing LOMDRBS by the end of 2nd, 3rd and 9th week of admission was 15%, 20 % and 25% respectively among neonates who received ≥ two-lines of antibiotic therapy. While, among neonates who received one line of antibiotic therapy up till the end of 60 days follow-up period the probability of developing LOMDRBS remained 0%. This difference was extremely significant.
For calculation of survival rate of neonates developing MDR, we classified neonates into two cohorts; exposed cohort included neonates with confirmed MDRBS, and non-exposed cohort included neonates with confirmed bacterial sepsis. Both cohorts were followed up for 60 days to outcome either death of the neonate or improvement and discharge. We found that among neonates with confirmed bacterial sepsis cumulative proportion surviving at the end of first week of follow-up was 86%, 59% in bacterial sepsis without MDR versus with MDR, respectively. We noted at the end of 60 days follow-up a statistically significant difference between overall survival of both groups. Proportion of neonates surviving bacterial sepsis was 71% compared to 30% of neonates with MDRBS. The median survival time in MDRBS group was 12.95 days. Statistically significant risk factors for predicting MDRBS were pattern of antibiotic therapy OR=13.64; 95% CI. [4.80 – 38.79]. CVC/PICC OR=4.6; 95% CI. [2.72 - 7.8], Other tubes/catheters (chest, urinary, gastrostomy and dialysis) OR=4; 95% CI. [1.59 – 10.26], mechanical ventilation OR=3.73; 95% CI. [1.06 - 13.07], Gastric tube OR=2.5; 95% CI. [1.23 – 5.06] and TPN OR=1.87; 95% CI. [1.04 – 3.37]. Mortality due to MDR in EOBS was 45% compared to 75% in LOBS. We identified number of antibiotic lines, days of stay in NICU and birth weight (Kg) as mortality predictors.
Unit change in number of antibiotic lines significantly increased the risk of mortality by 36%. Unit increase in birth weight of neonates significantly decreased the risk of mortality by 47% and unit increase in days of stay in NICU significantly decreased the risk of mortality by 6%.
Prolonged PROM “>18 hours” was the only maternal factor significantly related to EOMDRBS. Neonates admitted for respiratory distress, preterm, LBW or assessed as SGA or IUGR were at a higher risk of LOMDRBS. Neonates of mothers who had vaginitis with or without treatment or had UTI were at a higher risk of LOMDRBS.
Two probability decision trees were created: the first for lines of empiric antibiotic therapy and probability of developing bacterial MDR. It showed that probability of prescribing empiric antibiotic therapy from first line Antibiotics (Aminoglycoside and Beta Lactam with or without Beta-lactamase inhibitor) and developing bacterial MDR compared to Second line antibiotics (Carbapenems) was 0.099 and 0.024, respectively. On the other hand, probability of prescribing third line antibiotics (Glycopeptide; Vancomycin) or fourth lines antibiotics (Fluoroquinolone) as an empiric therapy was less than 0.01 in both groups and probability of developing bacterial MDR was zero in both groups. The second probability decision tree was created for number of prescribed lines of antibiotic therapy and probability of developing bacterial MDR. Probability of prescribing one line of antibiotic therapy and developing MDR compared to five lines of antibiotic therapy was 0.005 (5 per 1000 neonate) and 0.029 (29 per 1000 neonate) respectively.
LOS was the most common form of neonatal sepsis. The incidence rate of MDR neonatal sepsis, EOMDRBS and LOMDRBS was 126 per 1000, 27 per 1000 and 101 per 1000 birth lives, respectively. Gram-negative bacteria were dominant in both EOS and LOS. In both EOS and LOS Klebsiella pneumoniae was the main isolated Gram-negative pathogen, while Staphylococcus aureus was the main isolated Gram-positive pathogen. Candida was the second most common pathogen in both EOS and LOS. Proportion of fungal sepsis caused by Candida was more common in LOS (22.8%) than EOS (11.4%).
Resistance data from NICU cultures were used to develop accumulative antibiogram showing overall profile of antimicrobial susceptibility and resistance pattern. High antibiotic resistance to the commonly used and available drugs was noted. High MDR was found among Gram-negative bacteria; 55.4%, 84.7%, 57.8% and 74.6% of Enterobacteriaceae isolates were ESBL producing, Carbapenem-resistant, Aminoglycosides resistant and Fluoroquinolone resistant, respectively. Alarming XDR and PDR was found among both Enterococcus, Acinetobacter and Klebsiella pneumoniae isolates. We also noted emerging non-susceptibility to both Colistin and Tigecycline (Last resort antibiotic therapy for Gram-negative bacterial infections) and to linezolid (Last resort antibiotic therapy for Gram-positive bacterial infections). Recorded urgent threats included Carbapenem-resistant Acinetobacter and Carbapenem-resistant Enterobacteriaceae. While serious threats included: Drug-resistant Candida, ESBL-producing Enterobacteriaceae, VRE and MRSA.
Analysis of antibiotic prescribing practice using AHRQ framework “4 moments of antibiotic decision making” showed that in our study 60% of documented indications correlated with possible infections. While 40% of prescribed empiric antibiotic therapy was prophylactic for anticipated risk of acquiring infection during NICU stay for LBW or ELBW, initiating mechanical ventilation and before double volume exchange transfusion. An 82.7% of empiric antibiotic therapy was first line (Aminoglycosides and Beta-lactam with or without Beta-lactamase Inhibitors). Second line empiric antibiotic therapy (carbapenems) represented 16.2%. Good antibiotic prescribing practices included ordering cultures prior to antibiotics being prescribed, reassessment and discontinuing antibiotic therapy after 48-72 hours. Laboratory contamination (5%) was high compared to recommended 1-2%.
Antibiotic prescribing practices that need improvement included spectrum of antibiotic therapy which was narrowed in less than 1% of neonates. Less than 1% of neonates were switched to oral antibiotic therapy. More than 60% of alternative antibiotic lines were add-on and included a combination of 3 or more antimicrobial therapy.