Video recordings of neonatal resuscitations reveal a gap between how clinicians think they perform and what actually happens in the delivery room. When researchers started filming the first moments of newborn life, they found that more than half of resuscitations deviated from established guidelines, with errors ranging from poor mask seal during ventilation to overly aggressive stimulation. These recordings have become one of the most powerful tools in neonatal medicine for improving care, training staff, and understanding why some babies struggle in their earliest minutes. What the footage shows is often uncomfortable but consistently useful.
What the Camera Captures in the First Minutes
A neonatal resuscitation video typically begins the moment a baby arrives at the warming table. The camera, usually mounted overhead or at an angle that captures the infant and the hands of the clinical team, records everything: the initial assessment, drying and stimulation, airway positioning, suctioning if needed, and any ventilation or advanced interventions. Physiological monitors often overlay heart rate and oxygen saturation data onto the recording or run in parallel, giving reviewers a synchronized view of what the team did and how the baby responded.
One of the earliest and most cited studies to systematically review these recordings examined 100 resuscitations and found that 54% included at least one deviation from Neonatal Resuscitation Program guidelines. Ten percent of newborns received overly aggressive stimulation. Among the 78 infants given supplemental oxygen, the decision was judged incorrect in 15% of cases. Of 18 babies who needed mask ventilation, about a quarter had poor chest expansion, and among 12 who were intubated, only 7 were successfully intubated on the first attempt, with just 4 completed within the recommended 20-second window.1Pediatrics. Video Recording as a Means of Evaluating Neonatal Resuscitation Performance These numbers were jarring when first published, and they set the stage for a broader movement toward video-based quality improvement in neonatal care.
The Gap Between Guidelines and Practice
One consistent finding across video studies is that clinicians frequently deviate from resuscitation protocols without realizing it. In a study that recorded preterm infant resuscitations while simultaneously monitoring physiological parameters, only 21% of cases followed guidelines fully. About three-quarters had at least one respiratory intervention that did not conform to protocol, and roughly 29% had deviations in non-respiratory steps, most commonly related to preventing heat loss.2Resuscitation. Auditing resuscitation of preterm infants at birth by recording video and physiological parameters
More recent work has broken these deviations down further. A prospective observational study using video to assess adherence to the Neonatal Resuscitation Program found that during positive pressure ventilation, nearly 17% of cases involved an inconsistent face mask seal, while about 12% showed improper finger positioning with the recommended C-E grip, and another 12% used an incorrect ventilation rate.3Resuscitation Plus. Assessment of adherence to the neonatal resuscitation program using video recording: a prospective observational study These are not obscure technical failures. Mask seal and ventilation rate are among the most fundamental skills taught in resuscitation training, and when they break down during a real event, they can delay the establishment of effective breathing.
The footage also shows what happens during suctioning, a routine step that video has helped reframe. A study that reviewed 200 suctioning episodes in both premature and term infants found that breathing improved in only about 2% of premature infants after suctioning, compared with 72% of term infants. Breathing remained unchanged in roughly 85% of premature infants. More striking, over 80% of the preterm and term infants who went on to need intensive care experienced bradycardia after airway suctioning, though many were already bradycardic before the procedure began.4PubMed Central. Video-Recorded Airway Suctioning of Clear and Meconium-Stained Amniotic Fluid and Associated Short-Term Outcomes in Moderately and Severely Depressed Preterm and Term Infants This kind of granular observation, where the camera shows what the clinician’s hands are doing and the monitor shows what the baby’s heart rate is doing simultaneously, has pushed the field to reconsider how aggressively and how early suctioning should happen.
Why Clinicians Miss What the Camera Catches
The delivery room during a complicated birth is a high-stress, fast-moving environment. Neonatal staff have described a phenomenon that video review makes vivid: tunnel vision. As one clinical nurse specialist put it during a qualitative study of video review programs, it is easy to focus entirely on providing ventilation to a baby and lose track of whether the mask is the right size, whether the pressures are set correctly, whether oxygen needs to be increased, or whether repositioning is needed.5PubMed Central. Impact of the Neonatal Resuscitation Video Review program for neonatal staff: a qualitative analysis The recording captures the full picture, but the person performing the resuscitation is processing only a narrow slice of it in real time.
This is not a failure of knowledge. Most neonatal staff can recite the correct steps when tested in a calm setting. The gap is between knowing the algorithm and executing it under pressure, with a tiny patient whose condition may change second by second. Video makes this gap measurable rather than anecdotal, which is part of why it has been so effective as a quality improvement tool.
How Video Debriefing Changes Performance
When teams watch their own resuscitation footage in structured debriefing sessions, performance improves measurably. One study that implemented a combined program of video debriefings and ventilation refresher training tracked outcomes over 12 months. Comparing resuscitation events before and after the program, the median time to effective spontaneous breathing dropped from about 196 seconds to 144 seconds. An overall resuscitation performance score rose from 77% to 89%.6PubMed. Video performance-debriefings and ventilation-refreshers improve quality of neonatal resuscitation The time for heart rate to recover above 100 beats per minute did not change significantly, and neither did the number of newborns transferred to intensive care, but the improvements in technique and timing were clear.
A related study implementing a video debriefing program found similar gains. Comparing 74 resuscitation events before the program to 45 events after, the overall performance score again improved from 77% to 89%. Adequate positive pressure ventilation rose from 43% to 64%, and pauses during initial ventilation, moments when ventilation stopped and should not have, dropped from 20% to 0%.7PubMed. Implementation and effectiveness of a video-based debriefing programme for neonatal resuscitation Eliminating pauses during the critical early ventilation window is exactly the kind of improvement that video makes possible. Before the footage existed, no one could say with certainty how often those pauses were happening.
Staff who participated in video review programs have described the experience as distinct from simulation training. Watching a simulated scenario on a mannequin is useful for practicing hand skills, but reviewing footage of a real resuscitation forces a different kind of reckoning. Participants reported that video review let them practice troubleshooting in their heads, though several noted the limitation that it does not replace hands-on skill practice.8Pediatric Research. Impact of the Neonatal Resuscitation Video Review program for neonatal staff: a qualitative analysis The two approaches complement each other: simulation builds muscle memory, while video review builds situational awareness.
What Video Reveals in Low-Resource Settings
The impact of video recording is arguably even more significant in hospitals with limited resources, where neonatal mortality rates are higher and fewer opportunities exist for formal debriefing. A study in Uganda that recorded 141 resuscitations and analyzed 99 of them found that the median time to initiate positive pressure ventilation was 66 seconds, with essentially no ventilation happening during the first minute. Once ventilation did begin, interruptions during the first minute lasted a median of 28 seconds. The most common reasons for those interruptions were stimulation, suctioning, and causes the reviewers could not identify from the footage.9PubMed Central. Neonatal resuscitation practices in Uganda: a video observational study
These findings matter enormously because the first minute of resuscitation, sometimes called the “golden minute,” is the window in which effective ventilation can prevent a cascade of harm. Knowing that ventilation frequently does not start until well past that window, and that it is interrupted for reasons the clinicians themselves may not be aware of, gives quality improvement teams something concrete to work on. Researchers have developed low-cost video setups specifically adapted for lower-income settings, arguing that the technology needed to record and review resuscitations does not have to be expensive or complex to be useful.10PubMed Central. Neonatal resuscitation monitoring: A low-cost video recording setup for quality improvement in the delivery room at the resuscitation table
Artificial Intelligence Reading the Footage
A growing body of research is exploring whether AI can watch neonatal resuscitation videos and automatically identify what is happening, reducing the need for human reviewers to manually code every action. One approach uses a two-step deep learning system: first, convolutional neural networks detect and track relevant objects in the frame (the baby, the mask, the clinician’s hands), and then a second network classifies what activity is taking place based on those tracked objects.11IEEE Journal of Biomedical and Health Informatics. Activity Recognition From Newborn Resuscitation Videos
Another system goes further by combining video data with physiological signals from ECG and accelerometer sensors. For detecting stimulation activity, this multimodal approach achieved a sensitivity of about 82% and a specificity of about 82%, performing best when all available data streams were combined rather than relying on video alone.12Biomedical Signal Processing and Control. Automatic prediction of therapeutic activities during newborn resuscitation combining video and signal data The practical promise here is large. If automated systems can reliably flag deviations from protocol in real time or shortly after the event, feedback loops could tighten dramatically, and video review could scale to every resuscitation rather than a selected sample.
Related work has applied video-based monitoring beyond resuscitation to ongoing NICU care. One system for estimating infant respiratory rate from video achieved an intraclass correlation of 0.91 compared to manually counted rates, indicating excellent agreement, and outperformed traditional impedance-based sensors in some comparisons.13PubMed Central. Video-Based Respiratory Rate Estimation for Infants in the NICU Non-contact heart rate monitoring using computer vision has also been tested in preterm infants, comparing favorably against ECG measurements.14Pediatric Research. Non-contact heart and respiratory rate monitoring of preterm infants based on a computer vision system: a method comparison study These technologies are not yet standard equipment, but they point toward a future where the camera is not just a passive recorder but an active participant in monitoring infant wellbeing.
The Ethics of Filming a Crisis
Recording a neonatal resuscitation raises real ethical questions, and the field has grappled with them openly. The central tension is consent. A resuscitation is an emergency. There is no time for a careful informed consent discussion, and a parent’s mental state during an urgent delivery is not ideal for making research participation decisions. Reviewing the ethical and legal landscape, one analysis noted that informed consent for research involving emergency procedures is often not feasible, because the time pressure and emotional intensity of the moment undermine the conditions that make consent meaningful.15Journal of Medical Ethics. Ethical and legal considerations in video recording neonatal resuscitations
Most institutions that record resuscitations have navigated this by classifying the recording as a quality improvement activity rather than research, which changes the consent requirements, or by obtaining consent afterward with a clear option for parents to have the footage deleted. Staff concerns run in a different direction: clinicians worry about accountability if the video captures errors, and some are reluctant to participate if the footage could be used in legal proceedings. These concerns are not hypothetical, and program designers have had to address them head-on to build enough trust for video review to function.
Healthcare professionals and parents who participated in focus groups about live video recording in neonatal units identified five main themes: potential benefits for newborn care, the impact on parents during periods of separation from their baby, the importance of informed consent and guarantees about how footage will be used, worries about the camera disrupting clinical work, and data protection.16PubMed Central. Parents’ and healthcare professionals’ perceptions of the use of live video recording in neonatal units: a focus group study The fear of disruption is worth noting: some clinicians worry that knowing they are being filmed will make them self-conscious or change their behavior. In practice, most programs report that staff forget about the camera within minutes once a resuscitation is underway, though the evidence on this is largely anecdotal.
What the Footage Means to Parents
For parents, neonatal resuscitation is often a terrifying blur. If the birth is complicated, the baby may be whisked to a warming table surrounded by a team of clinicians while the parents watch from across the room, or the baby may be taken to an entirely different area. Parents frequently describe having fragmented, incomplete memories of what happened, sometimes filled in by fear and worst-case imagining.
When researchers offered parents the chance to review the recording of their infant’s resuscitation, the response was overwhelmingly positive. Parents reported that watching the footage helped them understand the start of their child’s life, make sense of what the medical team had done, and process the trauma of the event. Reviewing the recording frequently produced feelings of appreciation, both for the infant and for the clinical team. Parents valued screenshots or copies of the video as keepsakes, particularly when a child did not survive. In those cases, the recording sometimes became one of the few tangible records of the child’s brief life, and parents reported that having it brought them comfort.17Archives of Disease in Childhood – Fetal and Neonatal Edition. Reviewing recordings of neonatal resuscitation with parents
The timing and setting of the review mattered. Parents who were guided through the footage by a clinician who could explain what was happening at each step had a better experience than those who might have encountered the recording cold. This suggests that the footage itself is not inherently helpful or harmful; the context in which parents see it determines the experience. Programs that offer video review to families typically build in a structured session with a neonatologist or nurse who walks through the recording, pausing to explain and answer questions.
Thermal Care and What Happens Before the First Breath
Not everything that video captures relates to dramatic interventions. Some of the most practically important footage shows the mundane early steps: drying the baby, wrapping in plastic for very premature infants, positioning the head, and keeping the environment warm. A large randomized trial examined whether drying very preterm infants before plastic wrapping made a difference in temperature at NICU admission. It found essentially no difference: about 46% of dried infants and 46% of undried infants achieved normal temperature at admission, with mean temperatures of 36.4°C and 36.5°C respectively.18JAMA Network Open. Drying Very Preterm Infants Before Plastic Wrapping at Birth: A Randomized Clinical Trial This kind of finding shifts protocol. If the extra step of drying before wrapping does not help, it can be skipped, saving precious seconds.
Video has been instrumental in revealing how much time these seemingly minor steps actually consume. When reviewers watch footage and time each action, they often discover that preparation steps take longer than anyone estimated. The camera does not lie about the clock, and that objectivity has made it possible to shave seconds from protocols in ways that add up to meaningful improvements when a baby is waiting for ventilation to begin.
What Video Cannot Show
For all its value, video has blind spots. The camera captures what people do with their hands but not what they are thinking, deciding, or communicating sotto voce. A clinician who hesitates before intubating might be second-guessing their skill, or they might be doing a rapid mental assessment that ultimately leads to a better decision. Video reviewers can see the pause but cannot always interpret it. Some programs address this by pairing video review with structured interviews where the clinician narrates their thought process while watching the footage, adding a layer of reasoning that the recording alone cannot provide.
Camera angles also limit what is visible. A single overhead camera may not capture the face mask seal clearly, or the view may be obscured by a clinician’s hand or head. Multiple camera setups improve coverage but add cost and complexity. And in chaotic resuscitations with large teams, bodies crowd together around a small infant, and even well-positioned cameras can miss critical moments. The technology is getting better, with wider-angle lenses and additional sensors, but a single video feed will never capture everything that matters in the room.
Audio is another underexplored dimension. Most video setups capture sound, but few studies have systematically analyzed the communication patterns during resuscitation. Team communication breakdowns, someone calling out a heart rate that another team member does not hear, or two people simultaneously reaching for the same piece of equipment, are the kinds of failures that audio could reveal but that current analysis methods largely overlook. As AI-based analysis matures, parsing both the visual and auditory channels of resuscitation footage may become the next frontier.