Intraosseous access is often necessary to provide rapid resuscitation during pediatric emergencies. This study aimed to evaluate the current utilization, clinical characteristics, and complications of IO access in children among a large sample of hospitals using an EHR database, TriNetX. We found that more IO lines were placed in pediatric subjects ≥ 1 year of age, and a higher mortality rate was seen in subjects < 1 year. Additionally, our findings show a higher frequency of IO placement in subjects < 1 year of age with cardiac arrest and a higher frequency of IO placement in patients ≥ 1 year of age with convulsions.
By evaluating two different age groups within the pediatric population in a multicenter fashion, while restricted to the limitations of this dataset, we have expanded our knowledge of IO usage and outcomes. Furthermore, due to physiological reasons and the different approaches that are taken in emergency settings, it may be more reasonable to examine younger children separately from older children whenever possible. For example, it is more difficult to gain emergent vascular access in smaller and younger children due to smaller veins and involuntary movements. Thus, more IOs are placed in older subjects, and a higher rate of complications and deaths are observed in younger subjects, possibly due to the inability to achieve prompt access. By stratifying our population by age, we demonstrated differences in outcomes between these two groups, which is novel to this study.
While IO access is reported as a safe and rapid approach to pediatric resuscitation, there is minimal population-based data at present showing its utilization outside of cardiac arrest. Based on the lower frequencies of noncardiac diagnoses found in the study subjects at the time of IO placement, IO access may be underutilized in noncardiac settings such as convulsions, shock, and respiratory failure. Given the current AHA PALS guidelines and its interchangeable efficacy profile with other forms of vascular access, this should not be the case [9, 11, 19].
In the setting of neurological disease, a higher frequency of IO placement was seen in older pediatric subjects than in those less than 1 year of age in this study. There are many possible reasons for these findings. For example, administering intramuscular (IM) medications for seizure control is generally effective, but older children may need larger doses of medications, which may be easier to achieve with an IO. The lower frequency seen in subjects less than 1 year of age may highlight the difficulty in obtaining IO access in younger children, especially during excess involuntary muscle contraction. It is also possible there was a reluctance or discomfort in obtaining IO access in a young child. Alternatively, IO access may not have been considered necessary as intravenous access may have been successfully obtained, especially if the patient is otherwise healthy. Nevertheless, quick vascular access is essential to administer medications necessary to treat convulsions and prevent further complications, regardless of age. Thus, IO placement should be considered as soon as possible if other vascular access attempts appear to be futile when children experience convulsions, especially given the relatively low risk of complications in our study.
Despite including diagnoses other than cardiac arrest and respiratory failure in this study, the mortality rate was similar to other studies evaluating IO usage [23, 24]. Given the American Heart Association’s PALS recommendation, the mortality rate recorded in our study may seem high based on the similar safety profiles and efficacy of IO access compared to other intravascular access forms. For example, in the pediatric intensive care unit (PICU), the overall mortality rate has historically been reported to be 2% [11, 27]. However, children who present to the emergency department and inpatient setting often have preexisting medical conditions, comorbidities, and often are extremely ill (even before they reach the PICU). Vascular access in these patients can be challenging to achieve. Thus, the high mortality rate seen in both groups in this study population is likely a marker of the inherently poor prognosis of critically ill and medically complex pediatric patients who require IO access.
We identified possible complications 1 day to 3 months after IO use in this population, including lower extremity thrombosis, fracture, cellulitis/abscess, pulmonary embolism, and osteomyelitis. Our study demonstrates similar complication rates as previously reported [17, 18, 23, 28]. In this data set, the low incidence of complications reported may be explained by the mortality rate. More complications likely occurred with patients who did not survive and thus were not reported within the EHR. It should also be considered that not all complications were recorded, and perhaps the reported complication rate is lower than the actual rate. Finally, it is possible that the reported complications were unrelated to IO placement. Nonetheless, because complication rates continue to be reported as low, IO use should be considered early on for urgent vascular access.
The study was not without limitations. The exact temporal relationship between IO placement and outcomes is unknown due to constraints within the TriNetX database, which only allows users to determine when CPT and ICD codes are billed. We can only assume that the patient receives IO placement and is diagnosed on the same day they are billed. Additionally, we would have expected a higher mortality rate among this population due to the likely prognosis of critically ill patients receiving IO placement. Therefore, it is possible that not all IO placement codes were recorded correctly within the EHR, or some procedures took place but were not documented at all. It should also be considered that IO procedural codes are not necessarily recorded for missed attempts. In our study, the location of IO placement for 40% of the encounters was not reported. It is possible that the patient received an IO line in a clinical setting that was not the emergency department or inpatient area. Even though it was unknown where these encounters occurred, we opted to include these subjects as IOs are typically utilized as an emergency tool in the ED or inpatient setting. Additionally, we must consider that patients who survived in a hospital represented within the TriNetX database may not have survived in a non-TriNetX hospital if they were transferred out of the TriNetX hospital. Due to dataset limitations, the severity of the illness was unknown. However, the patient population was likely critically ill due to IO usage in emergencies. Furthermore, the data was de-identified such that no information was provided about the hospitals within the database (including whether they were pediatric-based hospitals). Despite these limitations, the data set provided a large sample size using population-based sampling of 37 different HCOs across the USA.