Pediatrics · Part Four · What Families Notice and How Care Works
Lesson 53 / 57
Safe by Design: What the Adverse Event Literature Actually Reports
Across the whole published literature, 15 serious adverse events in children have been reported under manual therapies. High-velocity rotation and missed pathology are what recur in them.
The published record of serious harm to children from manual therapies is small and specific. A review of 31 articles found 15 serious adverse events, with three deaths, none of which occurred under chiropractic care. High-velocity extension and rotational manipulation was reported in most cases, and underlying pathology was present in a majority. A separate systematic review adds 20 indirect harms through delayed diagnosis. The Unified Model of Tone reads safety as a design question.
Serious adverse events in the literature
15 across all providers
Indirect harms from delayed diagnosis
20 cases
Contact used on an infant
what rests on a closed eyelid
Spinal ossification
continues into the teen years
Direct and indirect harm
A direct adverse event is injury caused by the procedure itself. An indirect adverse event is harm caused by what the procedure displaced, most often a delayed diagnosis. Both belong in a safety discussion, and the second is easier to prevent than the first.
What the design consists of
A sustained, specific contact graded to the child, applied without rotation, without a high-velocity thrust and without forced end-range extension. The infant is read continuously through it. Heart rate, breathing, color, muscle tone and whether a baby stays asleep are the readouts a nervous system offers when a child cannot describe anything.
01What has actually been reported
Fifteen serious events across the whole published record
Safety claims are worth only as much as the record behind them. Todd and colleagues systematically searched peer-reviewed and non-peer-reviewed sources from database inception to March 2014 for adverse events in infants and children treated by chiropractors or other manual therapists. Thirty-one articles met the selection criteria, and 12 of them reported 15 serious adverse events (Todd 2015).
Three deaths were reported, under a physical therapist, an unknown practitioner and a craniosacral therapist. The authors state directly that no deaths associated with chiropractic care were found in the literature to date.
The full breakdown
Of the 12 serious injuries, 7 involved chiropractors. The rest involved a medical practitioner, an osteopath, two physical therapists and one unknown provider. That is the whole distribution as the reviewers reported it, and what the cases had in common turns out to matter more than the count.
02Findings
What the research shows
These figures come from the two systematic reviews of adverse events in children and from a practice survey.
03What recurs in those cases
Two specific failure modes, both avoidable
The useful part of that review is not the count. It is what the cases had in common, because that is what a practice can act on.
First, technique. High-velocity, extension and rotational spinal manipulation was reported in most of the serious cases (Todd 2015). One case involved forcibly applied craniosacral dural tension, and another an adjusting instrument.
Second, missed pathology
Underlying preexisting pathology was identified in a majority of the reported cases. The authors conclude that performing a thorough history and examination to exclude anatomical or neurologic anomalies before applying any manual therapy may further reduce adverse events across all manual therapy professions.
Those two findings together are the actual safety argument. The documented harm clusters around a specific class of technique applied to bodies with something else going on. Neither is a mystery, and both are addressable by design.
04The larger category
Delayed diagnosis is the more common documented harm
A second systematic review found something the first does not emphasize, and it deserves more attention than it usually gets. Vohra and colleagues reviewed 13 studies and identified 14 cases of direct adverse events, of which 9 were serious, including subarachnoid hemorrhage and paraplegia (Vohra 2007).
They also identified 20 cases of indirect adverse events. Those involved delayed diagnosis, including diabetes and neuroblastoma, and inappropriate provision of spinal manipulation for serious medical conditions such as meningitis and rhabdomyosarcoma.
Why the indirect number is the important one
Twenty indirect cases against 14 direct ones. The larger documented risk in that review is not what manual therapy did to a child. It is what was missed while it was being provided. That risk is entirely a function of how a clinician examines, what they recognize, and how readily they refer.
The same review states that neither causation nor incidence rates can be inferred from observational data, and calls for prospective population-based active surveillance. That limitation runs through the whole published record, in both directions.
05What the design does
No rotation, no thrust, and a contact you could rest on an eyelid
The care described in this section is defined by what it leaves out and by what it puts in its place. There is no high-velocity thrust in infants, no cervical rotation and no forced end-range extension. Those exclusions map directly onto the documented failure modes.
What is applied instead is a sustained, specific hold. The pressure is no more than a person could comfortably rest on a closed eyelid, held for a few seconds until the tissue softens, and graded to the age and size of the child. No audible release is sought at any age. Graded by age sets out the ladder, and what an adjustment is describes the contact itself.
Why the anatomy demands it
A young cervical spine fails differently from an adult one. Among 1,098 children with cervical spine injury drawn from 75,172 injured children, 52 percent of bony injury was upper cervical against 28 percent lower (Patel 2001). Half of those with cervical cord injury had no initial radiographic abnormality, and a national series found the cervical spine most commonly involved in cord injury without radiographic abnormality (Knox 2016). A child is not a small adult covers why.
06A body still forming
The tissue is young, and the baby is read the whole way through
An infant spine is a different structure from an adult one. At birth much of the vertebral column is still cartilage and the secondary curves of the neck and low back have not yet set. The joints carry a normal hypermobility that resolves as ossification proceeds through childhood and into the teen years.
The neuromuscular system is young alongside it. The asymmetric tonic neck reflex and the Moro are still present and still integrating, and primitive reflexes remain the standard instruments for reading brainstem integrity at this age (Zafeiriou 2004). The pathways carrying position sense up to the brain are still myelinating week by week. Matching the input to that body is the starting point of the whole method.
How an infant tells you
A baby cannot describe a sensation, so the nervous system speaks on its behalf and a clinician learns to listen. Heart rate, breathing, skin color, muscle tone and whether a baby stays settled are watched continuously through a contact. Those readouts report the state of the brainstem and the balance between vagal and sympathetic drive, which is the central integrative state of the child in front of you.
The vagus carries much of that regulation, and its output is still maturing across the first year. The vagus and the calm covers the nerve in detail. A baby who stays relaxed and asleep through a contact is giving real, readable feedback, and the model treats that undisturbed state as evidence the input was matched rather than as a pleasant coincidence.
07Examination is the safety measure
What is ruled out matters more than what is applied
If missed pathology is the recurring risk factor, then the examination is the intervention that matters most. This reframes what a first visit is for.
Primitive reflexes and postural reactions read against age remain the standard instruments for assessing central nervous system integrity in infants. Infants with five or more abnormal postural reactions went on to cerebral palsy or developmental delay in the studies reviewed (Zafeiriou 2004). Those findings are reasons to refer rather than reasons to adjust.
What a careful practice looks for
Fever in a young infant. A child who is difficult to rouse. Unexplained irritability with neurological signs, poor weight gain, focal neurological findings, or any history suggesting significant trauma. A fixed head preference carries a broad differential. Craniosynostosis, premature fusion of a skull suture, belongs with a craniofacial specialist. Head shape and torticollis covers that differential.
The profession has published its own consensus framework for this. A Delphi panel of 29 experts from five countries produced updated best-practice recommendations for chiropractic care of children (Hawk 2016). That is professional consensus on how care of a child should be conducted.
08After a demanding birth
Why families bring a newborn in, and what is examined
Birth is a mechanical event as well as everything else it is, and that is often what brings a family through the door. Passage through the pelvis, an assisted delivery using forceps or vacuum, and delivery by cesarean section all apply real forces to the head, neck and upper cervical spine of an infant. When birth needs help covers those forces directly.
The upper cervical region is where a gentle assessment concentrates afterward, and the anatomy is the reason. That region is densely packed with position receptors feeding the vestibular nuclei and the brainstem. Membrane runs continuously from inside the skull down into the spinal canal, and the lower cranial nerves leave the skull base alongside the great vessels.
What the assessment involves
Head rotation is checked to both sides and compared. Resting posture, the symmetry of spontaneous movement and the primitive reflexes are read against age. Palpation runs over the occiput, the upper cervical tissues and the cranial base, looking for tissue that stays guarded after the rest of the baby has let go. The upper neck in delivery covers the anatomy in full.
Any contact that follows is the sustained light hold, with no twist and no thrust. The model holds that tension held in that region is read by the nervous system as information, which is why so light an input is expected to count for anything at all. The first 12 months are when the brainstem centers receiving it are organizing.
09The model’s claim
What the Unified Model of Tone predicts about safety
Everything above is the published adverse event record, the anatomy, or a description of what is done. What follows is this model’s reading, stated as ours rather than drawn from the papers cited.
The model places every intervention on one continuous axis of magnitude and gives that axis a default. Begin with the least invasive input that can carry the message. A nervous system responds to what it can read rather than to what it is pushed by. Force past what it needs for a signal to register degrades the signal. On that reading, large force applied to an infant spine is not merely riskier. It is less informative, because magnitude is not what carries the message.
The prediction
That yields a testable claim about dose. The model predicts that outcomes track the specificity and fidelity of the contact rather than its magnitude, and therefore that reducing force while maintaining accuracy should not reduce whatever effect is present. A finding that effect held steady as force fell would confirm the model.
The default carries its own safety logic. When a small input fails, the system has lost little and the clinician has learned something. When a large input fails, the cost is high. The documented harms above cluster in the highest-magnitude class, high-velocity rotation and extension, a magnitude the model holds is rarely matched to what an infant's system can use.
Safety by design is not restraint for its own sake. A child is already doing developmental work of its own, and the model holds that the smallest input carrying the message is the one that leaves that work undisturbed.
Efficacy against a named condition is a separate question with its own trials. A randomized, blinded, placebo-controlled trial of 100 infants found no effect beyond placebo for infantile colic (Olafsdottir 2001), and the unsettled infant carries that result in full.
This is a claim about how development is organized rather than a claim about what treatment does. It holds that better-organized tone yields greater adaptive capacity, whichever appropriate input delivered the useful information. If autonomic variability, reflex responsiveness, postural symmetry and settling time after handling are shown to move together, the unification claim is confirmed.
10The tone reading
How safety expresses tone
Every topic in this library expresses all of tone. In safety three aspects carry the signature, because a record of 15 serious events clustering around one technique class is a statement about input rather than about risk.
Input quality
The model holds that information is what a nervous system responds to. Fidelity of contact matters, and magnitude is a separate axis.
Load
High-velocity rotation delivers a large mechanical dose to tissue that is more cartilage than bone in a young child.
Constraint
A young spine has different limits from an adult one, and its failure point sits high. Anatomy bounds what is reasonable.
The remaining foundations run through safety as well. Set point: a settled child is one whose baseline was not disturbed by the encounter. Prediction: a sustained contact is readable, where a sudden one arrives before it can be anticipated. Gain: a system that reads small signals accurately does not require large ones. Time course: the same contact means something different in a two-week-old than in a ten-year-old. Coupling: the upper neck reaches balance, gaze and autonomic circuits at once, which is why care there is taken seriously. Oscillation: a baby who stays asleep through a visit is one whose state was never disrupted. These are readings of one organization rather than separate systems, which is the core claim of the Unified Model of Tone.
11Across the library
How this page relates to the rest of the library
Safety connects the anatomy, the method, and the examination that comes before either.
The contact itself, described in terms of what it is and what it deliberately is not.
How the method changes with the child, and why the youngest receive the least.
Why a young cervical spine fails at the top, and why a normal first film means less.
The region a newborn assessment concentrates on, and what passes through it.
A differential worth ruling out, where one in four presentations is not muscular.
What actually happens, and where examination sits in the sequence.
The evidence library for this section, with the primary sources gathered.
12Frequently asked
Questions families ask about safety
Has a child ever been seriously hurt by this kind of care?
Yes, and the published record names what happened. A literature review found 31 relevant articles, with 12 reporting 15 serious adverse events across all manual therapy professions. Of 12 serious injuries, 7 involved chiropractors. Three deaths were reported, under a physical therapist, an unknown practitioner and a craniosacral therapist, and the authors state no deaths associated with chiropractic care were found in the literature to date.
What went wrong in those cases?
Two things recur. High-velocity, extension and rotational spinal manipulation was reported in most of the serious cases. And underlying preexisting pathology was identified in a majority of them. The review authors conclude that thorough history and examination to exclude anatomical or neurologic anomalies before applying any manual therapy may further reduce adverse events across all the professions involved. Both failure modes are addressable by design.
Is the risk mainly from the treatment itself?
Not according to the larger category in the literature. A systematic review identified 14 direct adverse events, 9 of them serious, but also 20 indirect cases involving delayed diagnosis or inappropriate provision of manipulation for serious conditions including meningitis and neuroblastoma. More documented harm came from what was missed than from what was done, which makes examination and referral the central safety measures at every age.
How much pressure is actually used on a newborn?
Far less than most parents expect. The contact is a sustained light hold, no more than you could comfortably rest on a closed eyelid, held for a few seconds until the tissue softens. There is no high-velocity thrust, no cervical rotation, no forced end-range extension and no audible release sought. The force is graded to a spine that is still largely cartilage, with joints that are naturally mobile at this age.
How will I know my baby is comfortable?
Your baby will tell you through the nervous system rather than in words. Heart rate, breathing, skin color, muscle tone and whether your child stays settled are watched continuously through any contact. Those readouts report the balance between vagal and sympathetic drive in a brainstem that is still maturing. A baby who stays relaxed and asleep is giving real feedback, and most babies do exactly that.
Is care worth considering after a hard delivery?
A demanding birth is the most common reason families come. Passage through the pelvis, forceps, vacuum and cesarean delivery all apply real force to a newborn head and neck. The upper cervical region is densely packed with position receptors feeding the vestibular nuclei and the brainstem, so assessment concentrates there. Some findings mean referral instead: fever in a young infant, a baby difficult to rouse, poor weight gain, focal neurological findings, or a history suggesting significant trauma.
How reliable are the reassuring numbers?
Less reliable than they look, and that cuts both ways. A practice-based survey reported 3 adverse events across 5,438 office visits and 2 across 1,735 visits, but self-report surveys under-capture harm. The systematic reviews are built from case reports, and their authors state directly that neither causation nor incidence rates can be inferred from observational data. Prospective surveillance has been called for and not done.
13The sources
References
8 primary sources, each linked to its PubMed record. Figures quoted on this page were checked against the published abstract.
Related evidence