Introduction
In preterm infants, early recognition of neonatal brain injury and assessment of risks of later impairment is a challenging goal of current neuroimaging studies [
1‐
3]. Magnetic resonance imaging (MRI) provides clinicians and researchers objective, high-quality, in vivo information about brain anatomy, pathology and, due to recent advances, functional and physiological characteristics [
4‐
10]. Early cerebral MRI scans at 30 weeks’ postmenstrual age and at term-equivalent age are increasingly being incorporated into standard care for very low birth weight (VLBW) infants (birth weight < 1,500 g). This provides early biomarkers for studying preterm brain injury related to neurodevelopmental outcome. These early determinants may contribute to the design of pharmacological and behavioural interventions to improve outcome [
6,
7,
11,
12].
MRI is considered a safe imaging technique. No evidence exists of serious harm to human tissue, besides loud acoustic noise, tissue heating and peripheral nerve stimulation [
13‐
16]. Performing early MRI scans in VLBW infants is challenging, as they frequently require respiratory support and are vulnerable to haemodynamic instability. Consequently, early MRI scans of VLBW infants should be performed in a safe and controlled environment with the use of a dedicated protocol. Studies on the methods that promote patient safety and health care quality are ongoing. Previous studies regarding safety of MRI in VLBW infants suggest that MRI procedures are safe [
17‐
19]. However, population size and maturity range varied widely in these studies, and in some works, only adverse events during the scan were assessed [
17,
18].
Our aim was to study the safety of routine MRI scans in preterm infants at a postmenstrual age of 30 weeks. To accomplish this, we collected data of these infants regarding safety incidents and (avoidable) adverse events over a long time period: 24 h before and 24 h following the MRI scan.
Discussion
Our study stresses the importance of providing a controlled environment for early MRI procedures for preterm infants. Despite the presence of a multi-disciplinary guideline specifically designed for preterm infants, minor adverse events, such as hypothermia and the need for increased respiratory support after the scan, were encountered regularly: these events occurred in 26 infants, 50% of our study population. In total, 39 minor adverse events occurred. Therefore, caution needs to be taken regarding the safety of VLBW infants during MRI procedures. Critical incident review and continuous re-evaluation of the guidelines are essential in this process.
MRI is becoming increasingly important for accurately evaluating brain injuries and the consequent effects on neurodevelopment in preterm infants [
9,
11,
21,
22]. Compared with cranial US, MRI has proven to be more sensitive for the detection of diffuse white matter injury (DWMI) [
3,
23,
24], and allows objective quantification of brain injury at a micro-structural level [
4,
25]. MRI is considered a safe imaging technique, independent of ionizing radiation, and it enables high-resolution neuroimaging in a non-invasive manner [
26]. The use of MRI scanning is limited in preterm infants because of their cardio-respiratory instability and predisposition to hypothermia [
17,
26‐
28]. Performing an MRI scan in this vulnerable population requires a comprehensive guideline that includes all the essential elements: good preparation, optimal monitoring of vital parameters, open communication between the involved parties, individualised care and continuous efforts to improve the quality of care. Neonatal intensive care must obviously be maintained throughout the procedure, which requires the use of MR-compatible equipment that ensures optimal monitoring of vital parameters without causing injuries, such as burning, or image degradation as a result of radiofrequency interference with the static magnetic field.
Because of the increased risk of respiratory and circulatory compromise, no sedation was used in this study. To reduce motion artefacts, we use other strategies to comfort the infant, such as those according to the principles of the Newborn Individualised Developmental Care and Assessment Program [
29,
30].
Safety incidents in (neonatal) health care are generally related to poor preparation, equipment failure and human error [
31,
32]. Studies on interventions to improve healthcare quality, such as staff training, implementation of a time-out-procedure (TOP) and the use of checklists and tailored guidelines, have shown that such preventable incidents can be reduced [
33‐
35]. In addition, adverse events should always be reported in order for the guideline to be adjusted [
31]. Comparable to operative procedures, a systematic pre-procedural briefing, such as a TOP, can be implemented for MR procedures as well. A pre-procedural TOP ensures that all involved caregivers agree that the correct procedure is being carried out properly for the correct patient.
We have shown that adverse events related to MRI scans in this vulnerable population are common. This is in contrast to other studies [
17‐
19], in which no significant adverse events were found. However, these studies primarily investigated serious adverse events that occurred during the scan itself, and the MRI scans had short acquisition times [
17], or the study population consisted of patients with a wide range of gestational ages [
17,
18]. In contrast, the results of the current study only include data on VLBW infants with a mean postmenstrual age of 30 weeks ± 4 days. In addition, we collected data for the 48 h surrounding the MRI procedure and total acquisition time was approximately 39 min.
Although it is reasonable and logical to assume that a longer total acquisition time is likely to increase the number of adverse events, we were unable to demonstrate this relationship in our study, possibly related to the small sample size.
The limitations of this study include selection bias, as our data consist only of infants considered haemodynamically stable enough for an MRI scan. In our setting, the medical team decided whether the infants were haemodynamically stable enough to undergo an MRI scan.
Perhaps if more strict criteria for haemodynamic stability were applied, the incidence of adverse events might be less frequent. In contrast, the incidence of adverse events might increase if more critically ill preterm infants were scanned, emphasising the importance of a comprehensive guideline with strict contraindications and staff training to ensure the safe execution of MR procedures.
Another limitation could be the retrospective design and the lack of temperature measurement during the MRI scan. Despite the use of an MR-compatible incubator, which provides controlled temperature and humidity, we encountered an increased incidence of hypothermia after the MRI scan. This could be explained by the mode of respiratory support: the infants were supported with cold air or oxygen during the procedure, which is in contrast to the setting at our wards, where infants are supported with pre-heated (40°C) air or oxygen. In order to decrease the high incidence of hypothermia after the MRI scan, we propose using an optical temperature probe to measure temperature continuously during the scan. Although minor adverse events were encountered more frequently after the MRI scan, it is not with certainty established that this in fact can be attributed to having undergone an MRI scan. Being transported from the NICU alone could in fact be stressful enough. However, due to the lack of evidence against causality and in the context of patient safety, we argue that each adverse event should be considered as a result of the procedure. Moreover, in order to avoid this possible bias, vital parameters, mode of respiratory support and the number of episodes of bradycardia, apnea or oxygen desaturation that occurred within 24 h before the MRI scan were compared with the same details occurring within 24 h following the MRI scan of each infant individually.
Logistical regression to weigh gestational age, birth weight and mode of respiratory support with the increased need for respiratory support was not performed given the small sample size (n = 12) in that group.
Finally, no serious adverse events occurred during the procedures, but the clinical significance of minor adverse events for future neurodevelopmental outcome remains unclear. Until empirical evidence shows that these events do not adversely affect neurodevelopment, we argue that adverse events should always be considered potentially harmful, and maximal efforts to prevent them must be undertaken.