Association of Intravenous Acetaminophen Administration With the Duration of Intravenous Opioid Use Among Hospitalized Pediatric Patients

Key Points Question Does intravenous (IV) acetaminophen administered prior to IV opioids reduce the total duration of IV opioids administered during a patient’s inpatient hospitalization? Findings In this comparative effectiveness research study including 893 293 hospitalized children, administration of IV acetaminophen prior to IV opioids was associated with a significant 15.5% reduction in total IV opioid duration when compared with administration of IV opioids alone. Meaning The results of this study suggest that IV acetaminophen administered prior to IV opioids should be considered in multimodal pain regimens because it may potentially reduce the total duration of IV opioids administered to hospitalized children.


eAppendix 1. Overview, Data Cleaning and Data Definitions for the Health Facts Database
Health Facts® Data warehouse: The analytic dataset used for this analysis ranged from January 2009 to June 2016. The initial date corresponded to the point when the Health Facts® Data warehouse expanded to include, in general, the current set of data elements. Data cleansing included: a) eliminating duplicate and null values; and b) eliminating data inconsistent with valid entries (e.g. admission times of zero, data inconsistent with life, or negative values). Data cleansing relevant to each data type is detailed in the section for each data type.
Pediatric Patients: < 22 years of age 1 Inpatient: Inpatients were defined as those having a Health Facts® inpatient classification code or with evidence of care from an Intensive Care Unit (ICU) location including discharge from an ICU care setting, medication requested or dispensed to an ICU care setting, or a laboratory order from an ICU care setting.
Pediatric Intensive Care Unit (ICU) Patients: Patients were identified as the encounters that satisfy all the following characteristics: • Pediatric patient (above).
• Age in hours not equal to 0.
• ICU patients were identified by the care setting associated with laboratory orders, cares setting to which the medication was requested or dispensed and/or the discharge care setting. The care setting required an ICU designation of at least one of the following: intensive care unit, intensive care unit -cardiac, intensive care unit -medical, intensive care unit -neurology, intensive care unit -surgical. Neonatal Intensive Care Unit (NICU) patients were excluded.
• The hospital outcome was known based on the hospital discharge disposition.
Intensive Care Unit Entry and Exit Times: The ICU entry and exit times were determined as follows. The ICU entry time was recorded as the earliest hour based on the laboratory order, medication request or medication dispensed associated with an ICU location (above). Following ICU entry, if there were no laboratory orders, medication requests or medication dispersals, the location was assumed to be the ICU.
We identified the ICU exit time using the following two conditions: First, the medication request, medication dispensed, or laboratory order came from a non-ICU location following ICU care. If the care setting was null or not mapped, the previous care setting was assumed. If multiple care settings including an ICU were observed during the same hour, the care setting was assumed to be an ICU care setting. Second, if the patient was discharged from the hospital from the ICU, the hospital discharge time was used as the ICU discharge time.
Since we recognized that there is imprecision in the assignment of entry and exit times, we assigned a minimum of 10 hours to each ICU stay for survivors (deaths could have a time period < 10 hours). For patients readmitted to the ICU during a hospitalization, a minimum of 40 hours was assigned to each non-ICU time period between ICU admissions. This resulted in 97.0% of PICU patients having only 1 ICU admission, and 2.3% have 2 ICU admissions, and 0.7% have 3 or more. For the rare patient that had more than 5 ICU admissions (0.13%), the ICU admissions after the 5 th were not included in the ICU admission.

Health Facts® Laboratory and Clinical Events (Vital Signs, Respiratory) Data:
Laboratory, vital sign, and respiratory data were assessed for measurement units and each test and its specific measurement units were assessed with descriptive statistics and data distributions. Units such as missing or null were combined with the appropriate data when the assessment of the data values, descriptive statistics, and data distributions indicated the identical measurement unit. Laboratory data with several units of measurement were converted to the standard units and combined. This is described in detail in Appendices C and D.

eAppendix 2. Creation of the Medication Database
The Health Facts® database contains the drug name (generic and brand), national drug code (NDC), medication volume, administration times, and administration frequency that were requested and dispensed from the pharmacy. Infusion times were available, but sparsely recorded and were therefore not utilized. Medications were assumed to be administered to patients during the times the pharmacy dispensed them to be administered. The duration of medication administration was rounded to the nearest hour. If a single dose of medication was administered in an hour -the patient was considered to have received one hour of medication.
To assess medications by class (ie. IV Opioid medications), the medication table was linked to an industry standard for medication classification, Multum™ , via the NDC number.
30,31 Multum™ provides information about medications therapeutic action in three categories from general mechanism of action to a specific therapeutic category. Table A provides a summary distribution of the first IV opioid administered in the intervention and control groups before and after propensity score matching.