The milestone tab answers one question: what would most children this age be doing? That is not the question that decides a referral. This section is about the second question — what is late enough to act on, and what does the pattern of lateness mean? It is written for the clinician who sees children occasionally and has one consultation, not for the child development service.
NICE quoted or closely paraphrased from a named NICE recommendation Cohort observational or descriptive study — see Evidence Review systematic review or meta-analysis Teaching framing or standard practice, not a cited result
A 20-month-old boy is brought by his mother, who is worried that he still is not walking. He cruises around furniture and walks a few steps holding a hand. He has about twelve words. He was born at term, and there is nothing in the pregnancy or neonatal history. He is a happy, sociable child who plays well.
His mother first mentioned the walking at his one-year review and was told that boys are often later, which is true. She has mentioned it twice since. Today she says, in passing, that he finds it hard to get up off the floor and pushes on his thighs to stand.
What has to happen at this consultation, and what is the one blood test?
Every milestone chart, including the one on the first tab of this page, is a list of typical ages. A typical age is the middle of a distribution. Roughly half of normal children are later than it. Using a typical age as a threshold to act would refer an enormous number of entirely normal children, and — more dangerously — it invites the opposite error: a child who is inside the normal range for one item gets called normal overall. Teaching
A limit age is the other end of the same distribution: the age by which almost all normal children have achieved the skill, so that failing to achieve it is unusual enough to justify looking further. Referral thresholds are limit ages. The distinction is the single most useful idea in this whole page. Teaching
The best measured picture of how wide that distribution really is comes from the WHO Motor Development Study, part of the Multicentre Growth Reference Study. It followed 816 healthy children in Ghana, India, Norway, Oman and the USA from 4 to 24 months and published, for six gross motor milestones, a window of achievement bounded by the 1st and 99th percentiles [Evidence 01]. Cohort
Two things in that figure are worth more than any milestone list. The first is how wide the windows are: walking alone spans 8.2 to 17.6 months, a 9.4-month window, and standing alone is wider still at 10.0 months. Walking at fifteen months is not an abnormality; it is ordinary. The second is where the referral thresholds fall. The 18-month threshold for boys sits essentially at the far edge of the measured window, so a boy who crosses it is genuinely unusual. The 15-month threshold for girls sits inside the window, which means it will pick up more normal children in exchange for identifying the abnormal ones earlier. Teaching
You cannot generate a referral decision by comparing a child to the milestone tab. Compare the child to the threshold, and use the milestone tab for what it is good for: working out which domains to ask about, and giving a parent an honest account of the normal range.
Five things in NG127 §1.25 send a child onwards: three thresholds for delay, all inside recommendation 1.25.2, and two findings that are not delays at all. The three thresholds are short enough to be worth knowing verbatim rather than approximately. NICE
To those three thresholds, add the two that are not about delay at all: new-onset gait abnormality — refer immediately to acute paediatric services (NG127 1.25.1), and motor development regression — refer to a paediatric neurodevelopmental service or paediatric neurology (NG127 1.25.4). A child who was walking and now is not is in a different clinical category from a child who has not started. NICE
Early asymmetry of hand function before one year is the least intuitive of the three, because it does not present as a delay. The child is doing things — with one hand. A strong hand preference in an infant is not precocity; the normal developmental sequence uses both hands, and typical handedness emerges later. NICE’s cerebral palsy guideline lists the same feature, and lists asymmetry or paucity of movement among the early motor features of cerebral palsy generally (NG62 1.3.3, 1.3.5). NICE
This is a real discrepancy and it is better to know it than to discover it in a referral letter. NG127 1.25.2 splits the walking threshold by sex — 15 months for girls, 18 months for boys. NG62 1.3.5, describing the most common delayed motor milestones in children with cerebral palsy, gives “not walking by 18 months (corrected for gestational age)” with no sex split at all. Both are current NICE recommendations. NICE
The safe reading is to treat the earlier of the two as the action point for the child in front of you, because neither guideline describes a harm from referring at 15 months, and both describe the harm of late identification. A girl not walking at 15 months meets NG127; a child of either sex not walking at 18 months meets both.
Every threshold in chapter 02 carries the phrase corrected for gestational age. It is in the guideline three times in three lines, which suggests the committee expected it to be ignored. A baby born at 28 weeks is twelve weeks preterm; at nine months of life she is six months corrected, and a six-month-old who is not sitting is not late. Using chronological age here manufactures delay. NICE
NICE’s preterm follow-up guideline states the convention directly: parents should be told that the child’s developmental (corrected) age, calculated from the original due date and not the date of birth, will be used for the first 2 years when assessing functional and developmental skills such as walking and talking (NG72 1.1.13). NICE
Two years is the cut-off, and NG72 is explicit about the switch: for a preterm child who is having enhanced developmental surveillance the structured assessment happens at 2 years corrected age, and all children born before 28+0 weeks get a further face-to-face developmental assessment at 4 years uncorrected age (NG72 1.3.11, 1.3.13). After two years, stop correcting. NICE
Correcting the age does not neutralise the prematurity. NG72 records that children born preterm are at increased risk of cerebral palsy, of motor function problems, of developmental coordination disorder, of learning disability, of special educational needs, and of speech, language and communication problems, with prevalence rising as gestational age falls (NG72 1.2.3–1.2.10). Preterm birth is also an independent risk factor for cerebral palsy in NG62 1.1.1, alongside chorioamnionitis, low birth weight, neonatal encephalopathy, neonatal sepsis and postnatal meningitis. A corrected age tells you where to set the threshold; it does not tell you to relax. NICE
Delay means a skill has not arrived. Regression means a skill arrived and left. They are not points on one scale, and the guidelines route them differently, urgently, and to different services. If you take one thing from this page, take the habit of asking explicitly: is there anything he used to do that he no longer does? Parents do not always volunteer it, because a lost skill is frightening to say out loud. Teaching
| What is lost | Age | Where NICE sends it |
|---|---|---|
| Motor skills | Any age | Paediatric neurodevelopmental service or paediatric neurology, per local pathways (NG127 1.25.4). Also: refer first to a paediatrician or paediatric neurologist, who can refer on to the autism team (CG128 1.3.2) |
| Language or social skills | Under 3 years | Autism team (CG128 1.3.1) |
| Language | Over 3 years | Paediatrician or paediatric neurologist first (CG128 1.3.2) |
| Any attained cognitive or developmental ability, in a child carrying a diagnosis of cerebral palsy | Any age | A red flag against that diagnosis — refer to a specialist in paediatric neurology (NG62 1.4.2) |
That last row is easy to skim past and it matters. Cerebral palsy is a non-progressive disorder. A child with cerebral palsy who is losing abilities is telling you the label may be wrong, and NG62 1.4.2 lists loss of already attained cognitive or developmental abilities among the red flags for a neurological disorder other than cerebral palsy. NICE
In autism specifically, regression is neither rare nor universal. A 2021 systematic review and meta-analytical update pooled 75 studies and 33,014 participants and found a pooled prevalence of 30% (95% CI 27–32%), with a weighted average age of onset of 19.8 months [Evidence 04]. The pooled figure conceals a great deal: heterogeneity was very high, and the prevalence moved with the definition used — 20% for language regression alone, 40% for language and social, 30% mixed. Loss typically involved language and social skills with motor skills unaffected. Review
Not as a test. The value of “about a third, around 20 months” is that it tells you regression around the second birthday is common enough to ask about at every developmental consultation in that age band, and that language and social loss is the pattern to ask about specifically. It does not tell you that a child who regressed has autism, and the absence of regression does not argue against it.
NG127 says it twice, once for delay and once for regression, in identical words: “If the child is a boy, consider measuring creatinine kinase level to exclude Duchenne muscular dystrophy before the boy has had a specialist review” (NG127 1.25.3 and 1.25.5). The guideline’s spelling is creatinine kinase; the enzyme is creatine kinase, and the test you order is CK. Quoted here as written so that the recommendation is findable. NICE
The phrase that carries the clinical weight is before the boy has had a specialist review. The test is not the specialist’s job to think of. It is a serum sample, it costs very little, and the point of doing it now is that it can be grossly abnormal months before anyone reaches a clinic.
Because the delay in diagnosing Duchenne muscular dystrophy is a delay in testing, not in presenting. A retrospective case note review of 20 boys diagnosed without a family history at an English tertiary neuromuscular centre mapped the pathway [Evidence 05]: Cohort
Read the third and fourth figures together. The gap between them is 1.6 months — once the CK was sent, the diagnosis followed. (That interval is derived by subtracting the two means the paper reports; it is not a figure the authors state.) The delay from parental concern to diagnosis was 19.2 months on average, and most of it was spent between a parent raising a concern and someone taking blood. Cohort
A separate record review of 179 boys with Duchenne muscular dystrophy found that delayed walking (defined as at or after 16 months) and cognitive impairment were strongly associated (P ≤ 0.0001), with cognitively delayed boys about three times more likely to have walked late; the association was independent of the rate of motor degeneration and of the age at diagnosis [Evidence 06]. The authors’ interpretation is the teaching point: when a boy has global developmental delay, clinicians stop attributing the motor delay to a muscle problem and attribute the whole picture to the global delay. Duchenne belongs on the list of causes of global developmental delay, and a boy with global delay should have a CK. Cohort
Ask how he gets up off the floor, and whether he climbs stairs one foot at a time or pulls on the banister. “Pushes on his thighs to stand” is a parent describing Gowers’ manoeuvre. Ask about calf size, about toe walking, and about whether he tires or falls more than other children. Then send the CK. Teaching
A milestone tab that lists four domains side by side is making an argument: that the useful question is not which item is late but how the domains relate to each other. Three patterns carry different meanings. Teaching
| Pattern | What it looks like | What it suggests |
|---|---|---|
| Global delay | All four domains behind, roughly evenly | A whole-child problem: genetic, metabolic, syndromic, a significant early insult, or severe deprivation. Dysmorphic features alongside it make this urgent (NG127 1.22.1) |
| Isolated delay | One domain behind, the others on track | A problem of that system — motor delay in cerebral palsy or a neuromuscular disorder, isolated speech delay, isolated hearing loss. Check the sensory channel that domain depends on before anything else |
| Dissociated / atypical | Domains out of step, or skills present but used oddly | Often social communication. A child can have “normal language milestones” and still have autism — CG128 1.2.7 says in terms that reported pretend play or normal language milestones do not rule it out |
One corollary is worth stating plainly, because it is where a milestone chart becomes actively misleading: do not check hearing and vision only when the language card looks late. An unresponsive infant can look like a social problem; a visually impaired infant can look like a motor problem. The red-flag list on the first tab for the youngest band includes not responding to loud noises and not following objects with the eyes for exactly this reason — those two items come from the standard developmental references rather than from a NICE recommendation, and they are none the less worth acting on. Teaching
A child can be on target in all four domains and still need referring today. These are findings, not delays, and none of them appear on a milestone chart. NICE
| Finding | Action | Source |
|---|---|---|
| Dysmorphic features with developmental delay | Refer urgently to paediatric services | NG127 1.22.1 |
| New-onset gait abnormality | Refer immediately to acute paediatric services | NG127 1.25.1 |
| Acute-onset hypotonia in a baby under 1 year | Examine for cardiac failure, liver or kidney enlargement, pyrexia, altered level of consciousness — and refer immediately | NG127 1.23.1 |
| Hypotonia present for weeks or months with weakness (feeding or breathing difficulty) | Refer urgently to paediatric services. If not weak and no intercurrent illness, consider referring per the cerebral palsy guideline | NG127 1.23.2 |
| Head circumference crossing 2 or more centile lines from a previous measurement | Refer to paediatric services for assessment and cranial imaging, to exclude progressive hydrocephalus or microcephaly. Refer immediately if there are also signs of raised intracranial pressure: tense fontanelle, sixth nerve palsy, failure of upward gaze (‘sunsetting’), vomiting, unsteadiness (ataxia), headache | NG127 1.22.3 |
| Persistent toe walking | Refer to a child development service | NG62 1.3.7 |
NG127 1.22.2 is unusually procedural, and the procedure is the point: in any child under 4 with a suspected abnormal head shape or size, take three consecutive measurements at the same appointment with a disposable paper tape, plot the longest of the three on a standardised growth chart corrected for gestational age, and refer if it is below the 2nd centile. A single measurement is not a data point in this decision — the recommendation is written the way it is because head circumference is easy to measure badly. NICE
Separately from the developmental thresholds: refer immediately children with sudden-onset or rapidly progressive (hours to days) limb or facial weakness, and refer urgently children with progressive limb weakness (NG127 1.24.1, 1.24.2). For boys with limb weakness, NG127 1.24.4 points straight back to the motor-delay recommendations — which is to say, to the CK. NICE
Autism is the developmental concern most likely to be missed by a page laid out like the first tab of this one, because its early features are mostly not milestones. They are the quality of communication rather than its timing. CG128’s preschool box lists them in detail; the ones that do most work in a short consultation cluster around joint attention. NICE
Alongside those, the same box lists language delay “in babble or words, for example less than ten words by the age of 2 years”, regression in or loss of use of speech, absent or delayed response to name being called despite normal hearing, reduced or absent responsive social smiling, reduced or absent imagination and variety of pretend play, and repetitive stereotyped movements such as hand flapping, body rocking, spinning or finger flicking. NICE
CG128 1.2.7 is the most quietly useful recommendation in the guideline, because it names the four things that most often close down a consultation prematurely. Do not rule out autism because of: good eye contact, smiling and showing affection to family members; reported pretend play or normal language milestones; difficulties appearing to resolve after a needs-based intervention such as a supportive structured learning environment; or a previous assessment that concluded there was no autism, if new information has become available. NICE
CG128 1.2.5 adds the epidemiological version of the same warning: autism may be under-recognised in girls, leading to underdiagnosis; may be missed in children with a learning disability and in children who are verbally able; may have been masked in an older child by coping mechanisms or a supportive environment; and — a specific trap — do not assume that language delay is accounted for by English not being the family’s first language, or by early hearing difficulties. NICE
The M-CHAT-R/F is the tool most likely to arrive in a referral letter. Its validation study screened 16,071 toddlers at 18- and 24-month well-child visits: children scoring ≥3 initially and ≥2 after the structured follow-up interview had a 47.5% risk of an autism spectrum diagnosis (95% CI 0.41–0.54) and a 94.6% risk of any developmental delay or concern (95% CI 0.92–0.98) [Evidence 02]. Cohort
Those two numbers are the whole story of what a positive screen means, and they point in a useful direction: a fully positive M-CHAT-R/F is a coin-flip for autism specifically, but it is close to definitive that something developmental needs assessing. A later systematic review of 50 studies put the pooled sensitivity at 0.83 (95% CI 0.77–0.88) and specificity at 0.94 (0.89–0.97), and found accuracy varied markedly with how and where the instrument was used — better in low-likelihood than high-likelihood samples, and better when the follow-up stage was actually completed [Evidence 03]. Review
Regression in language or social skills under 3 years goes to the autism team (CG128 1.3.1). Regression in language over 3 years, or regression in motor skills at any age, goes first to a paediatrician or paediatric neurologist, who can refer on (CG128 1.3.2). Concern about possible autism without regression is a “consider referring” to the autism team, weighing severity, duration, whether features appear in more than one setting, impact, parental concern, and the likelihood of an alternative diagnosis (CG128 1.3.3).
CG128 also lists factors associated with increased prevalence, worth knowing because several are things you would otherwise not connect: a sibling with autism, gestational age under 35 weeks, a learning disability, ADHD, cerebral palsy and other CNS malformations, neonatal or epileptic encephalopathy including infantile spasms, Down’s syndrome, fragile X, muscular dystrophy, neurofibromatosis, tuberous sclerosis, parental schizophrenia-like psychosis or affective disorder, and maternal valproate or topiramate in pregnancy or paternal valproate in the 3 months before conception (CG128 Box 1). NICE
There are three recommendations on speech in children in NG127 and they are worth reading as they stand. NICE
1.29.2 is often repeated as a rule against referring under-2s. The guideline does not say that. It makes a “consider referring” recommendation about children over 2, and it is silent on the younger group rather than prohibitive — and 1.29.1 and 1.29.3, which have no lower age limit, both apply underneath it. A 20-month-old who has lost words is covered by 1.29.3 and CG128 1.3.1, not excluded by 1.29.2. Teaching
A second silence is worth noticing: the first thing to establish in a child with delayed speech is whether they can hear. That is not in NG127’s speech section, which addresses referral rather than assessment; CG128 1.2.5 warns specifically against the reverse error of writing off language delay as explained by “early hearing difficulties” once a hearing problem is known about. Both things are true at once — check the hearing, and do not let a hearing result end the developmental thinking. Teaching
CG128 1.2.2 is unqualified: “Always take parents’ or carers’ concerns and, if appropriate, the child’s or young person’s concerns, about behaviour or development seriously, even if these are not shared by others.” The last clause is doing the work. It covers the situation where the health visitor is not worried, the nursery is not worried, and the examination in front of you looks unremarkable. NICE
The reason this is a recommendation rather than a courtesy is that the parent has a sample size you cannot match. They have watched the child for months across every setting and mood; you have ten minutes with a child who is tired and in an unfamiliar room. A long-standing review of the literature on parental report in primary care concluded that parents’ concerns, and good-quality standardised parent-report measures, are the formats that make best use of what parents know, and that the two together are an effective method of early detection in primary care [Evidence 07]. That is a narrative review of 78 articles and tests, not a diagnostic accuracy study — it supports taking the history seriously, not a sensitivity figure. Review
CG128 1.2.9 and 1.2.10 add the other half: raising the possibility of a developmental condition may cause distress even when you are being careful, it may take time for a family to come to terms with the concern, and they may not share it. Take time to listen, discuss sensitively, and agree the actions including referral. And CG128 1.2.3 asks something uncomfortable and useful — be critical about your own professional competence here, and ask a colleague if you are unsure of the next step. NICE
A concern raised three times and answered three times with reassurance is not three reassurances; it is a repeated finding. The case at the top of this page turns on exactly that. Reassurance given without a threshold check and without a CK is not a clinical decision, it is a deferral.
Referring everything is not a safe default; it costs the family anxiety and it costs the child development service the capacity to see the children who need it. Four common findings are usually variants — with named caveats. Teaching
The WHO Motor Development Study found that 4.3% of healthy children did not exhibit hands-and-knees crawling at all, and that about 90% achieved five of the six milestones in a common sequence [Evidence 01]. Crawling is not on the NICE trigger list. Bottom-shuffling as a locomotor variant is standard teaching rather than a finding of that study, which reports only that a proportion of healthy children never crawled on hands and knees. Teaching The caveat that keeps it safe: the child must still be meeting the sitting and walking thresholds, and must be moving symmetrically — NG62 1.3.3 lists asymmetry or paucity of movement, and abnormal motor development including late rolling and crawling, among the early motor features of cerebral palsy. Bottom-shuffling symmetrically at nine months with good sitting is a variant. Dragging one side is not. Cohort
Most young children with idiopathic toe walking are expected to resolve spontaneously, and early management for a young child with adequate range of movement is generally reassurance [Evidence 08]. But NG62 1.3.7 says to refer persistent toe walking to a child development service, and the same review notes that motion analysis may help distinguish mild cerebral palsy. Toe walking that persists, is asymmetrical, comes with tight heel cords or reduced ankle movement, or arrives in a child who previously walked flat, is not the idiopathic kind. Review
NG127 1.22.4 gives explicit permission not to refer: for a head circumference above the 98th centile that has not changed by more than 2 centile lines from the previous measurement, in a child developing normally with no symptoms of raised intracranial pressure, note the head size of the biological parents and — if familial macrocephaly is likely — do not routinely refer in the absence of any other problem. Every clause in that sentence is load-bearing: it is the stability across centile lines, not the absolute centile, that makes it safe. NICE
For babies under 1 year with a head flattened on one side, NG127 1.22.5 asks for a specific measurement: the distance from the outer canthus of the eye to the tragus of the ear on each side. If the two measurements differ, that confirms positional plagiocephaly and there is no need to refer routinely provided the baby is developing normally. If they are the same, suspect unilateral premature closure of the lambdoid suture and refer. Positional plagiocephaly is the commonest cause of an asymmetric head shape, and 1.22.6 gives the advice to pass on: change the baby’s position when lying, encourage sitting up when awake, and give time on the tummy. NICE
The first tab is a reference for typical ages and a checklist of red flags by age band. It is worth being precise about what that is not. Teaching
4.3% of the healthy WHO cohort never crawled on hands and knees, and crawling is on no NICE trigger list. What matters is whether sitting and walking meet their thresholds, and whether movement is symmetrical.
NG127 1.22.4 says not to refer routinely for a stable head circumference above the 98th centile in a normally developing child with likely familial macrocephaly. It is movement across centile lines, not the centile, that refers.
Most young idiopathic toe walkers resolve spontaneously. Persistence, asymmetry, tight heel cords, or a child who previously walked flat change the meaning — and NG62 1.3.7 refers persistent toe walking.
A fully positive M-CHAT-R/F carried a 47.5% risk of autism — and a 94.6% risk of some developmental delay or concern. It is close to definitive that something needs assessing, and a coin-flip as to what.
Pooled sensitivity across 50 studies was 0.83, with accuracy varying by setting and by whether the follow-up stage was completed. A negative screen does not overrule a concerned parent or an abnormal examination.
CG128 1.2.7 lists a previous negative assessment as something not to rule autism out on, if new information has become available. New information includes a new concern.
The outcome that matters is not the diagnosis; it is that the consultation ends with a referral letter and a blood form rather than a fourth reassurance. If the CK is normal, the referral still stands on the threshold alone.
Eight sources, each read as an abstract of record from PubMed and summarised with its design, its population and its limits. Every figure quoted in the Teaching section appears here with the study it came from. Developmental paediatrics has very few randomised trials of recognition, so most of what follows is descriptive or diagnostic-accuracy evidence, and it is labelled as such rather than dressed up.
These studies were done in children. The referral thresholds they sit beside are for children under 16 in a non-specialist UK setting, with motor thresholds corrected for gestational age to 2 years. None of this material is about adults, and none of it is a substitute for a structured developmental assessment by someone trained to do one.
| Study | WHO Multicentre Growth Reference Study Group. WHO Motor Development Study: windows of achievement for six gross motor development milestones. Acta Paediatr Suppl 2006;450:86–95. PMID 16817682. (Corporate authorship — no individual bylines.) |
| Design | Longitudinal data collected within the WHO Multicentre Growth Reference Study. 816 children in Ghana, India, Norway, Oman and the USA, assessed by trained fieldworkers at scheduled visits — monthly in year 1, bimonthly in year 2 — from 4 to 24 months. Caretakers independently recorded ages of achievement. Failure time models constructed windows bounded by the 1st and 99th percentiles, with no internal demarcations. |
| Findings | About 90% of children achieved five of the milestones in a common sequence; 4.3% did not exhibit hands-and-knees crawling. Windows overlap but vary in width: narrowest sitting without support (5.4 months), widest walking alone (9.4) and standing alone (10.0). 1st and 99th percentiles in months: sitting without support 3.8–9.2; standing with assistance 4.8–11.4; hands-and-knees crawling 5.2–13.5; walking with assistance 5.9–13.7; standing alone 6.9–16.9; walking alone 8.2–17.6. 95% CI widths 0.2–0.4 months at the 1st percentile and 0.5–1.0 at the 99th. |
| Authors’ conclusion | The windows represent normal variation among healthy children, recommended for descriptive comparisons between populations, to signal the need for appropriate screening when individual children appear late, and to raise awareness of overall development in child health. |
| Study | Robins DL, Casagrande K, Barton M, Chen CM, Dumont-Mathieu T, Fein D. Validation of the Modified Checklist for Autism in Toddlers, Revised with Follow-up (M-CHAT-R/F). Pediatrics 2014;133(1):37–45. PMID 24366990. |
| Design | 16,071 toddlers screened at 18- and 24-month well-child care visits in metropolitan Atlanta and Connecticut. Parents of toddlers screening at risk completed the structured follow-up interview; those who continued to show risk were evaluated. Scoring optimised using receiver operating characteristic curves. |
| Findings | Children scoring ≥3 initially and ≥2 after follow-up had a 47.5% risk of an autism spectrum diagnosis (95% CI 0.41–0.54) and a 94.6% risk of any developmental delay or concern (95% CI 0.92–0.98). Total score outperformed alternative scores; a 3-level risk algorithm is recommended. The revised instrument detected autism at a higher rate than the original M-CHAT while reducing the number of children needing follow-up. Children in this study were diagnosed 2 years younger than the national median age of diagnosis. |
| Study | Wieckowski AT, Williams LN, Rando J, Lyall K, Robins DL. Sensitivity and specificity of the Modified Checklist for Autism in Toddlers (original and revised): a systematic review and meta-analysis. JAMA Pediatr 2023;177(4):373–383. PMID 36804771. |
| Design | PubMed, Web of Science and Scopus searched for studies published January 2001 to August 2022. Included if the M-CHAT or M-CHAT-R/F was used to identify new autism cases, ≥10 autism cases, false-negative identification procedures reported, children screened by 48 months. PRISMA-reported; QUADAS-2 for selection bias; hierarchic summary ROC model. 50 studies, 51 samples. |
| Findings | Pooled sensitivity 0.83 (95% CI 0.77–0.88); pooled specificity 0.94 (95% CI 0.89–0.97). Accuracy was greater in low- versus high-likelihood samples, with concurrent versus prospective case confirmation, in larger samples, when the M-CHAT Follow-up was used, and in non-English versus English-only studies. |
| Declared interest | The senior author declares co-ownership of M-CHAT LLC with royalty income and holds the copyright to the instruments. Declared in the paper; recorded here because it is relevant to how an instrument’s own accuracy literature should be read. |
| Study | Tan C, Frewer V, Cox G, Williams K, Ure A. Prevalence and age of onset of regression in children with autism spectrum disorder: a systematic review and meta-analytical update. Autism Res 2021;14(3):582–598. PMID 33491292. |
| Design | Update of a 2013 review that had reported 32%. Medline, Embase, PsycINFO and Cochrane Library searched for studies from 2010 onward; risk-of-bias assessment performed; random effects model. 97 studies in the systematic review, 75 studies and 33,014 participants with sufficient data for meta-analysis. |
| Findings | Pooled proportion of autistic regression 30% (95% CI 27–32%). Heterogeneity very high (I² = 96.91) and not reduced by sensitivity or subgroup analyses. Prevalence varied by risk of bias (low-risk studies: 27%) and by definition of regression — language 20%, language/social 40%, mixed 30%, unspecified 27%. Weighted average age of onset 19.8 months. Loss typically involves language and/or social skills, with motor skills typically unaffected. |
| Authors’ conclusion | The findings highlight the need for a standardised definition of autistic regression and tools to measure it at multiple time points in early childhood. |
| Study | van Ruiten HJ, Straub V, Bushby K, Guglieri M. Improving recognition of Duchenne muscular dystrophy: a retrospective case note review. Arch Dis Child 2014;99(12):1074–7. PMID 25187493. |
| Design | Retrospective case note review at a tertiary centre for neuromuscular diseases in England. All 20 boys without a family history diagnosed with Duchenne muscular dystrophy in the preceding 10 years. Outcome: mean age at four key steps of the diagnostic pathway. |
| Findings | First reported symptoms 32.5 months (range 8–72). First engagement of a healthcare professional 42.9 months (10–90). CK checked 50.1 months (14–91). Diagnosis confirmed 51.7 months (16–91), i.e. 4.3 years. Total delay from parental concern to diagnosis 19.2 months (4–50). Background: over the preceding 30 years there had been little improvement in the age of diagnosis (mean 4.5–4.11 years). |
| Authors’ conclusion | An improvement in age at diagnosis, but continuing delay both in presentation to a health professional and in obtaining a CK test. The authors propose screening for Duchenne within the Child Health Surveillance Programme and lowering the threshold for CK testing in primary care. |
| Study | Mirski KT, Crawford TO. Motor and cognitive delay in Duchenne muscular dystrophy: implication for early diagnosis. J Pediatr 2014;165(5):1008–10. PMID 25149498. |
| Design | Records of 179 boys with Duchenne muscular dystrophy evaluated by the senior author between 1989 and June 2012. Delayed walking defined as at or after 16 months. Cognitive impairment identified by special education placement or lower-than-expected grade level in the elementary school years. |
| Findings | Delayed walking and cognitive impairment were highly correlated (P ≤ 0.0001). Cognitively delayed boys were about three times more likely to have delayed walking. The association was independent of the rate of motor degeneration (P = 0.9) and of the age at diagnosis (P = 0.6; combined average age at diagnosis 5.1 ± 2 years). |
| Authors’ conclusion | Primary care givers may overlook Duchenne as a cause of delayed independent walking when early cognitive delay is also apparent. Duchenne should be included among the disorders causing global developmental delay, and the standard evaluation of boys with global developmental delay should include a serum CK. |
| Study | Glascoe FP, Dworkin PH. The role of parents in the detection of developmental and behavioral problems. Pediatrics 1995;95(6):829–36. PMID 7539122. |
| Design | Review of 78 research articles and tests relying on parent information, drawn from the paediatric, psychological and education literature. The stated purpose was to help paediatricians make optimal use of clinical information from parents. |
| Findings | Some formats for eliciting parental information are superior in accuracy and ease: specifically parents’ concerns, and good-quality standardised parent-report measures such as the Child Development Inventories. The authors conclude that the two types of parental information in combination offer an effective method for early detection of developmental and behavioural problems in primary care settings. |
| Study | Bauer JP, Sienko S, Davids JR. Idiopathic toe walking: an update on natural history, diagnosis, and treatment. J Am Acad Orthop Surg 2022;30(22):e1419–e1430. PMID 36084329. |
| Design | Narrative review from a paediatric orthopaedic perspective. No pooled data; no risk-of-bias assessment. |
| Findings | Toe walking is a common presentation and idiopathic toe walking is a diagnosis reached when no other cause is suspected; the nomenclature is itself contested. Most young children with idiopathic toe walking are expected to resolve spontaneously, and early treatment for a young child with adequate range of movement is generally reassurance. Motion analysis, including EMG and multisegment foot modelling, may help differentiate children with mild forms of cerebral palsy. When toe walking persists, options run from stretching, serial casting, gait retraining and bracing to surgical lengthening at the Achilles or gastrocnemius level. |
Milestone ages on the first tab additionally draw on standard developmental references (Sheridan; Lissauer). Those are textbook typical ages, not referral thresholds — see Teaching chapter 01.