Study setting

We performed a retrospective cohort study of patients admitted to the pediatric medical intensive care unit (PICU) or pediatric cardiac intensive care unit (CICU) at a freestanding tertiary children’s hospital. Children’s Mercy Hospital-Kansas City (CMKC) is licensed for 338 inpatient beds including 43 PICU and 24 CICU beds resulting in more than 2,300 annual intensive care admissions.

Sample collection and processing

The order for a TA was placed, and a sample was collected by either the nurse or respiratory therapist and sent to the microbiology lab for processing. A Gram stain was generated, and the sample was then plated in triplicate on blood agar, chocolate agar, and MacConkey agar. Susceptibility testing was then performed on all identified pathogens. During the study period, there were no decision support tools utilized to inform TA collection. Nor were there quality improvement initiatives to modify the frequency of TA collection. There was no standardized procedure for obtaining the sample and no formalized protocol for first-pass suction or utilization of normal saline. There were no rejection criteria utilized by the microbiology lab during this time that deemed a sample inappropriate.

Data collection

Patients were eligible for inclusion if they were <18 years of age at the time of ICU admission and had ≥2 TAs obtained during the study period within a single ICU stay. The microbiology laboratory created a report of all patients with a TA during the study period and excluded patients with a single TA from further analysis; Bronchoalveolar lavage and sputum samples were not included. Additionally, TAs obtained for organ donation purposes were excluded. A comprehensive chart review was performed on all patients. Demographic information, antibiotic duration/indication, ICU admission/discharge dates, microbiologic details of all TA cultures including susceptibility profiles, and clinician-documented reasoning for obtaining a TA were collected.

The type of artificial airway was captured via manual review of physician and respiratory therapy documentation. Antibiotic days, calculated as days of therapy (DOT) via manual review of start and stop dates, were included if the antibiotic was administered for at least 24 hours with concerns of an acute bacterial process, including oral, inhaled, and intravenously administered antibiotics. ^{Reference Stanic Benic, Milanic and Monnier11} Antibiotics used for prophylaxis were excluded. An organism was identified as multidrug-resistant if it was resistant to >3 antibiotic classes or had a ≥4-fold increase in minimum inhibitory concentration (MIC) value for an antibiotic used to treat a prior TA result. ^{Reference Tamma, Turnbull, Milstone, Lehmann, Sydnor and Cosgrove12}

Additionally, a modified clinical pulmonary infection score (mCPIS) was calculated for each TA based on clinically available data up to 24 hours prior to TA collection. This included temperature data, serum white blood cell count, quality of tracheal secretions, chest radiograph results, TA white blood cell count, TA growth characteristics, and respiratory factors including escalation of respiratory support. ^{Reference da Silva, de Aguiar, de Carvalho and Machado Fonseca13} The mCPIS score was calculated retrospectively by our study team to assess antibiotic days and the corresponding score. These scores have correlated with the likelihood of a VAE with resultant treatment recommendations developed at our institution based on the available literature. For a mCPIS score <6 with no or few bacteria, there was no indication for continued antibiotics. For a score ≤6 with moderate bacterial growth or >6, treatment should have been completed at 5 or 8 days, respectively. ^{Reference Tamma, Turnbull, Milstone, Lehmann, Sydnor and Cosgrove12–Reference American Thoracic14}

Finally, the Pediatric Complex Chronic Care Conditions Classification System, a validated system to assess the medical complexity of children, was calculated utilizing all available ICD-10 coding on patients. ^{Reference Feudtner, Feinstein, Zhong, Hall and Dai15} All data were recorded in a secure Research Electronic Data Capture (REDCap) database with subsequent analysis. ^{Reference Harris, Taylor, Thielke, Payne, Gonzalez and Conde16} This study was classified as exempt research by the CMKC Institutional Review Board.

Analysis

Quantitative data were analyzed with descriptive statistics, including frequencies and percentages, using Microsoft Excel. Endotracheal tube (ETT) days were counted if an ETT was present and the child received mechanical ventilation. Each day a tracheostomy tube (TT) was in place following surgical placement was counted as a tracheostomy day. This includes patients who were mechanically ventilated via the TT but also those who no longer required mechanical ventilation with a TT in place. To understand the influence of an artificial airway being present (ETT or TT), regardless of the need for mechanical ventilation, these two groups were combined and labeled “artificial airway days.”

The total antibiotic DOT that was documented as treatment for TA was calculated for each culture. The summary distribution of DOT was then compared across the 3 mCPIS scores using the Wilcoxon rank sum test. Lab results from each repeat culture were compared with lab results from the prior culture to categorize whether the repeat culture yielded new results; identical lab findings were categorized as “same findings,” whereas any new findings, including a previously identified pathogen now being no longer detected, were categorized as “new findings.” The time between repeat cultures was then compared between the “same findings” and “new findings” groups using the Wilcoxon rank sum test. Lastly, Fisher’s exact test was used to compare select categorical demographic/clinical factors between the MDRO and non-MDRO groups.