Growth-related conditions of the lower limb in children and adolescents

A limping child, an adolescent with knee pain: behind these two commonplace presentations hide four conditions of the growth plate, two of which are paid for in a lost hip when they are missed. This article is built on the approach, not on a list of diseases.

A synthesis article. Every figure cited carries its source where it is written, with its sample size and its method. Where the literature on an entity is thin, that is said rather than papered over: two of these four conditions rest, in rehabilitation, on no controlled study at all.

In brief: the clinical synthesis

What the available literature allows us to state, in twelve points. Each is taken up and sourced in the chapters that follow.

• Age is the first filter, and it is highly discriminating. Legg-Calvé-Perthes disease peaks between 2 and 5 years in the only study with a known denominator that has measured incidence by age band¹², slipped capital femoral epiphysis belongs to puberty³, and osteochondritis dissecans of the knee does not exist before 6 years: no lesion was found between 2 and 5 years in a cohort of more than a million children²⁶.
• The site of the pain, on the other hand, misleads. In a prospective study of 122 hips with a slipped epiphysis, 42.6 % had no hip pain at all and 26.2 % had knee pain¹. An adolescent's knee that hurts with no local cause is a hip until proven otherwise.
• Slipped capital femoral epiphysis is not an osteochondrosis. It is a mechanical slip of the epiphysis on its growth plate, which calls for surgical fixation. Classing it with the adolescent apophysitides is the reasoning error that costs most in this group.
• Continued weight-bearing makes that slip worse. The prognostic distinction is not radiological but functional: a slip in which weight-bearing is impossible, even with crutches, is called unstable, and 47 % of those hips developed avascular necrosis against 0 % of stable slips⁴.
• Delay in diagnosis is the rule, not the exception. The mean delay between the first symptoms and the diagnosis was 17 weeks across 481 patients, with no improvement at all over ten years⁶. Knee or distal thigh pain is an independent predictor of a longer delay⁵.
• The anteroposterior radiograph is not enough. The classic Klein line identified only 9 slips out of 23, against 20 out of 23 with its modified version and 23 out of 23 on the frog-leg lateral view⁷. Requesting an AP view alone means accepting that one slipped epiphysis in two will be missed.
• In Legg-Calvé-Perthes disease, containment surgery has not proved its overall superiority. In the British national prospective cohort (371 children, 396 hips), 36.4 % were operated on, and surgery did not change the radiological outcome at two years, odds ratio 1.03¹¹.
• Two prognostic factors dominate in this disease : age at onset and the height of the lateral pillar. Beyond 8 years with a pillar B or B/C, surgery does better than abstention; at 8 years or under, treatment of any kind does not change the outcome¹³.
• The physiotherapist's role in Perthes is to preserve range of motion, and it is the only point on which practitioners really agree. An international survey of 160 surgeons from 43 countries finds broad agreement on stretching and on the importance of range, and clear disagreement on strengthening¹⁷ ; in Sweden, none of the 44 practitioners surveyed recommended complete unloading¹⁸.
• Sinding-Larsen-Johansson disease rests on a very thin evidence base. The founding series has eight patients and ten knees²¹. It is benign and self-limiting, but its course is measured in months: five months in the case report best documented by imaging²².
• Osteochondritis dissecans is treated non-surgically first, with a highly variable healing rate. The reference systematic review, 27 studies and 908 knees, reports 61.4 % healing, within a range running from 10.4 to 95.8 %, and no high-level study²⁹.
• Clinical examination in osteochondritis has no reliable test. Wilson's sign was negative in 24 of the 32 patients with a lesion visible on radiographs³⁰. You do not rule it out by examination, you think of it from the setting and from the persistence.

Why does age guide almost the whole diagnosis?

Because each of these conditions strikes a structure that is only vulnerable at a precise moment of skeletal maturation. Age is not one guiding element among others: it is the only one that, on its own, rules out entire diagnoses.

The growth plate is not a homogeneous structure that would weaken uniformly throughout childhood. The blood supply of the femoral head, the mechanical strength of the proximal femoral physis, the ossification of the secondary centres of the knee and the maturation of subchondral bone each follow their own calendar. A condition can therefore only occur in the window when its target structure is both present and vulnerable.

That produces a distribution by age whose clinical use is direct. Legg-Calvé-Perthes disease is a necrosis of the femoral head linked to a failure of its blood supply, in the period when that supply is terminal and precarious: in a population of nearly 800,000 children in an integrated health system in southern California, the highest incidence was 3.05 per 100,000 in 2- to 5-year-olds, and the lowest 1.06 in 9- to 12-year-olds, that is an odds ratio of 3.13 in favour of the younger¹². Slipped capital femoral epiphysis is conversely a mechanical failure of the growth plate, occurring when it widens and becomes more horizontal under the effect of the hormones of puberty: the incidence reported in the United States from 1997 and 2000 hospital admissions was 10.80 per 100,000 children aged 9 to 16³.

These incidences have a practical corollary that is often forgotten when reading review articles: these diseases are rare. A physiotherapist with an ordinary paediatric caseload will see a great deal of anterior knee pain, a few apophysitides, and will meet a slipped epiphysis or a Perthes a few times in their career. That is precisely what makes the reasoning necessary: rarity forbids relying on pattern recognition, and demands an approach triggered by criteria, not by an impression.

Sex is added to age as a second filter, and it is almost as discriminating. The odds ratio for male sex is 12.44 for Legg-Calvé-Perthes disease in the Californian cohort¹², and the incidence of osteochondritis dissecans of the knee is 15.4 per 100,000 in boys against 3.3 in girls²⁶. For slipped epiphysis, the gap exists but stays more modest: 13.35 against 8.07 per 100,000³. These gaps never allow a diagnosis to be ruled out in a girl, but they shift the starting probability, which matters when the decision is about whether to image.

Why does a diseased hip hurt in the knee, and in what proportion?

This is the classic trap of hip conditions in children, and the figure that describes it is copied from review to review without its method being stated. Two primary studies exist, they do not measure the same thing, and the gap between their results is entirely explained by their definitions.

The hip joint and the knee joint share their sensory innervation. The obturator nerve, the femoral nerve and the nerve to quadratus femoris give articular branches to the hip; the cutaneous and articular territories of the femoral and obturator nerves run down to the medial side of the knee. A nociceptive afferent arising in the hip capsule therefore enters the cord through roots that also receive the knee, and the brain regularly attributes the pain to the more distal segment. That is not a curiosity: it is the usual mode of presentation of a substantial share of hip conditions in children.

What the two reference studies actually measure

The first is retrospective. Matava and colleagues reviewed 106 patients with a slipped capital femoral epiphysis and counted those whose main complaint at the first consultation was knee or distal thigh pain: there were 16, that is 15 %. In the subgroup of 65 patients who had received no treatment before arriving, the proportion rose to 15 out of 65, that is 23 %. The most important result of that work is not the percentage but what goes with it: those patients received significantly more wrong diagnoses, underwent significantly more unnecessary radiographs, and arrived with radiologically more severe slips².

The second is prospective, and that is what makes it valuable. Uvodich and colleagues had 107 children operated on for a slipped epiphysis complete a standardised body chart, giving 122 analysable hips. They did not ask for the main complaint but for the complete mapping of the painful areas. Result: 70 hips out of 122 had hip pain, so 52 hips, that is 42.6 %, had none. The knee was painful in 32 cases (26.2 %), the thigh or leg in 43 (35.2 %), the groin in 17 (13.9 %), the posterolateral region in 13 (10.7 %). A combination of several areas was present in 39.3 % of cases, and 49 % of patients had consulted more than once before the diagnosis¹.

That mechanism does not concern only slipped epiphysis. Legg-Calvé-Perthes disease also commonly presents with isolated knee or thigh pain, which, as a literature review written for paediatricians puts it, “obliges the clinician to keep the hip in mind during the examination”²⁰. The practical consequence is the same in both cases, and it comes down to one rule of examination.

The sign that costs thirty seconds

Loss of hip internal rotation is common to both hip conditions in this group. In Legg-Calvé-Perthes disease, deficits in abduction and internal rotation are the most consistent examination findings, the Trendelenburg limp appearing only late²⁰. In slipped epiphysis, the posterior slip of the epiphysis produces obligatory external rotation: the hip spontaneously moves into external rotation when the knee is passively flexed towards the abdomen, which is the most suggestive sign in the examination.

That examination is done supine, with the hip flexed to 90 degrees, and always compared with the opposite side. Asymmetry of internal rotation in a child with knee pain whose knee examination is normal is enough to request hip imaging. It is not enough to dispense with imaging when it is absent: internal rotation can be preserved in slips of small magnitude, and it is precisely those that need catching.

Faced with a limping child, what must be ruled out first?

Before reasoning about growth conditions, you have to have ruled out what will not wait. Two diagnoses dominate that step: septic arthritis of the hip, and acute leukaemia revealed by musculoskeletal pain. Neither belongs to physiotherapy, and both present like the conditions in this chapter.

A limping child is a presentation whose first task is triage by severity, before any aetiological reasoning³⁸. The most used prediction rule for separating septic arthritis of the hip from transient synovitis, the benign self-limiting form, rests on four variables: a history of fever, inability to bear weight, an erythrocyte sedimentation rate of at least 40 mm in the first hour, and a white cell count above 12,000 per cubic millimetre. The probability of septic arthritis goes from less than 0.2 % when no criterion is present to 3.0 % for one, 40.0 % for two, 93.1 % for three and 99.6 % for all four³³.

That rule was validated prospectively in a different population, where the area under the curve fell from 0.96 to 0.86: it remains highly accurate but loses precision outside its original population, which is the expected behaviour of any prediction rule³⁴. A later prospective study of 53 children in whom hip aspiration had been judged necessary identified fever above 38.5 degrees as the best single predictor, C-reactive protein being the only independent factor on multivariable analysis³⁵.

The differential table: age, site, entity

The table below crosses the two variables that do the sorting, and adds for each entity the action that must not be missed. It includes the common benign entities, without which differential reasoning makes no sense, with links to their own page where one exists.

One line of that table deserves reading twice. So-called growing pains are a diagnosis of exclusion whose usual criteria are themselves debated: an Australian twin study showed that those criteria take in a share of painful restless legs syndrome, and that the prevalence falls by a third as soon as the absence of motor restlessness is required⁴¹. What matters to the physiotherapist is simpler: these pains are bilateral, come in the evening, and never cause a limp. A limp, unilateral pain or mechanical daytime pain do not fit that frame, whatever label those around the child have applied.

Slipped capital femoral epiphysis: why must it be treated as an emergency?

Because it is not an osteochondrosis. It is a mechanical slip of the femoral epiphysis on its growth plate, which progresses as long as the hip is loaded, and whose treatment is surgical. An adolescent rehabilitated for knee pain while their epiphysis slips loses hip at every session.

The other three entities in this article are disorders of ossification or of blood supply: they follow their own natural history, and physiotherapy accompanies a process. Slipped capital femoral epiphysis belongs to another category. The proximal femoral growth plate, widened and straightened by the hormonal changes of puberty, stops resisting the shear forces imposed by weight-bearing. The epiphysis, held by the ligamentum teres and the periosteum, slips backwards and downwards while the femoral neck moves forwards and into external rotation.

The practical consequence is unmatched in the rest of this article: the factor that makes the lesion worse is exactly the one rehabilitation uses as its tool. Loading, walking, weight-bearing exercises: all of these, on a growth plate that is giving way, increase the displacement. That is why unloading precedes imaging, and not the other way round.

Who is affected, and how often

The incidence reported in the United States from the 1997 and 2000 paediatric admission databases is 10.80 per 100,000 children aged 9 to 16, with 13.35 in boys and 8.07 in girls. That same analysis finds a relative incidence 3.94 times higher in Black children and 2.53 times higher in Hispanic children than in white children, as well as geographical and seasonal variations³.

Overweight is the most consistently associated risk factor, to the point of being built into aetiological triage rules. It is not a diagnostic criterion and its absence rules nothing out: it is an element that shifts the probability, not a necessary condition.

The distinction that governs the whole prognosis

Loder and colleagues proposed in 1993 abandoning the traditional classification into acute, chronic and acute-on-chronic forms, based on symptom duration, in favour of a purely functional distinction. A slip is called unstable when the pain is such that weight-bearing is impossible, even with crutches; it is called stable when the patient can bear weight, with or without aids.

Across 55 hips with symptoms of less than three weeks' duration, 30 were unstable and 25 stable. The result was satisfactory for 14 of the 30 unstable hips (47 %) against 24 of the 25 stable hips (96 %). Above all, avascular necrosis occurred in 14 of the 30 unstable hips (47 %) and in none of the 25 stable hips⁴. That series does not allow a link to be established between early reduction and the risk of necrosis, which the authors state explicitly.

That definition is functional and not radiological, which makes it directly usable by a physiotherapist: it is made by observing whether the adolescent can put the foot down, with no investigation at all. A qualification has been added since, one team having shown that clinical stability does not perfectly predict the stability found intraoperatively¹⁰. That does not invalidate the rule, it is a reminder that a clinically stable slip is not a harmless slip.

Delay in diagnosis, measured and not estimated

The delay between the first symptoms and the diagnosis has been measured several times, with converging and poor results. Across 196 patients, the median delay was 8.0 weeks, it increased with the severity of the slip, and knee or distal thigh pain remained an independent multivariable predictor of a longer delay, alongside the type of health cover⁵.

Twelve years later, a study of 481 patients admitted to three paediatric hospitals between 2003 and 2012 showed no improvement at all : the mean delay was 17 weeks, with extremes of 0 to 169 weeks, and no significant difference between the two-year periods of the follow-up. The delay between the first professional consulted and the diagnosis was on average nil when that professional was an orthopaedic surgeon, 4 weeks for a primary care doctor and 6 weeks in the emergency department. Finally, 52 patients out of 481, that is 10.8 %, had a slip on the opposite side after treatment of the first⁶.

What imaging to request, and why the AP view is not enough

Klein's line is the classic landmark on the AP radiograph: a line drawn along the upper border of the femoral neck should cut across part of the epiphysis. In slips of small magnitude, it stays normal. One study compared the three methods across 23 hips with a slipped epiphysis: the classic Klein line allowed the diagnosis in only 9 cases (39 %), its modified version in 20 cases (87 %), and the frog-leg lateral view in all 23 cases (100 %)⁷. The modified version of that line, proposed to make up for the low sensitivity of the original, has been the subject of a reproducibility study that quantifies its limits: the minimum difference required between two measurements for them to be considered distinct with 95 % confidence reaches 8.80 degrees between observers for the head-neck angle⁴⁵.

Two practical points follow. The first is that the request must explicitly mention both views, failing which an isolated AP film is common. The second is that the examination covers the whole pelvis and not the painful hip alone : bilaterality of slipped epiphysis is reported within a wide range of 12 to 80 % depending on the series⁹, and the opposite side is frequently affected without yet being symptomatic.

When to suspect an endocrine cause

A slipped epiphysis occurring outside its usual window, or in a slim adolescent, should prompt a search for an underlying cause: hypothyroidism, growth hormone deficiency, hypogonadism, chronic kidney disease. Loder and Greenfield proposed a simple rule from 433 children, of whom 285 had idiopathic and 148 atypical slipped epiphyses, giving 612 hips. At equal weight, a child under 10 or over 16 was 4.2 times more likely to have an atypical form; at equal age, a child below the 50th percentile of weight was 8.4 times more likely.

The so-called age-weight test follows: it is negative when the child is under 16 and of a weight at least equal to the 50th percentile, positive beyond those bounds. The probability that a slipped epiphysis was idiopathic was 93 % when the test was negative, and the probability that it was atypical 52 % when it was positive⁸. A positive test therefore does not make the diagnosis of endocrinopathy, but it justifies investigation.

Legg-Calvé-Perthes disease: what can physiotherapy really do?

Far less than it is credited with, and one precise, defensible thing: preserve range of motion. It is the only point on which the literature and practitioners converge, and it is also the one that formal guidelines cannot support with a controlled trial, because none exists.

Legg-Calvé-Perthes disease is an idiopathic avascular necrosis of the femoral head in childhood. The head loses its blood supply, fragments, then rebuilds over a period measured in years, keeping the shape it had during its fragile phase. The whole prognosis rests on that final shape: a round, congruent head in a matching acetabulum ages well, a deformed head leads to impingement then osteoarthritis.

Frequency and prognostic factors

The most solid study on this disease is a British national prospective cohort, in which 143 of the 144 hospitals managing children's hips took part over 18 months. It recorded 371 children and 396 hips, that is an annual incidence of 2.48 per 100,000 children aged 0 to 14, 95 % confidence interval 2.20 to 2.76¹¹. An American study of nearly 800,000 children aged 2 to 12 gives a similar order of magnitude, 2.84 per 100,000, with a marked male predominance, odds ratio 12.44, and a peak between 2 and 5 years¹². The incidence of this disease also varies strongly with social deprivation and with region, a gradient documented in the United Kingdom on geographical and temporal trends¹⁹, which is a reminder that none of these incidences transfers directly from one country to another.

In that British cohort, the independent predictors of a worse radiological outcome at two years were female sex, odds ratio 2.27 (1.19 to 4.35), age over 6 years, odds ratio 2.62 (1.30 to 5.28), and radiological collapse of more than 50 % at inclusion, odds ratio 2.19 (0.99 to 4.83). And above all: 117 hips out of 396, that is 36.4 %, were operated on, without surgery changing the radiological outcome, odds ratio 1.03 (0.55 to 1.96)¹¹. The authors conclude that there is enough collective uncertainty to justify a randomised trial, which is an elegant way of saying that the reference treatment is not established.

That result must be read alongside Herring's prospective multicentre study, which remains the reference on the surgical indication. It followed 438 patients and 451 hips, all aged 6.0 to 12.0 years at disease onset, across five modalities: abstention, bracing, range-of-motion exercises, femoral osteotomy and innominate osteotomy. All hips were classified using the modified lateral pillar classification and the Stulberg classification. The first part of that work is devoted to their reproducibility and introduces the intermediate group known as the B/C border: agreement between six observers reached 81 then 85 % per radiograph, for a mean weighted kappa of 0.71 then 0.79¹⁴. Its conclusions are precise and nuanced¹³ :

• No difference between abstention, bracing and range-of-motion exercises.
• No difference between the two osteotomies.
• No effect of treatment, of any kind, in children aged 8.0 years or less at disease onset.
• A significant benefit of surgery in children over 8.0 years whose hip was classified lateral pillar B or B/C border.
• No difference between surgery and abstention for lateral pillar C hips, whose outcome was the least favourable in every case.
• A significantly worse result in girls over 8.0 years.

The French-language reference review draws the same line: age at onset and the lateral pillar classification are the two main prognostic factors, non-operative treatment is not effective, and surgery is useful only for pillars B and B/C starting after 8 years, abstention being recommended in the other situations¹⁶.

The long-term outcome

A prospective cohort re-examined 56 patients, that is 58 hips, at a mean of 20.4 years after inclusion, all treated without surgery by range-of-motion exercises or bracing. The mean non-arthritic hip score was 79 out of 100, and 44 % of hips showed moderate to severe osteoarthritis on radiographs. Three patients had needed arthroplasty¹⁵. The French-language review recalls for its part that osteoarthritis rarely develops before 50 years¹⁶. The two findings complement rather than contradict each other: the trouble appears early on the radiograph, disabling osteoarthritis much later.

What practitioners recommend, for want of trials

Two recent surveys document the real state of practice, and they must be read for what they are: opinion surveys, not evidence of effectiveness. The first surveyed 160 paediatric orthopaedic surgeons from 43 countries and seven continents. It finds general agreement that restricted range and pain are what prompt a physiotherapy prescription, broad agreement on the importance of stretching in the early stages, and considerable disagreement on strengthening exercises. As for activity, there is broad agreement to allow swimming and cycling, and to discourage impact activities, school sport included, at all stages¹⁷.

The second, Swedish, surveyed 25 surgeons and 19 physiotherapists on three clinical scenarios corresponding to the initial, fragmentation and reconstruction stages. Stretching is recommended at all stages; strengthening recommendations vary at the first two stages; none of the participants recommended complete unloading of the affected hip, at any stage. Most limited trampolining, running, ball sports and gymnastics at the first two stages, and allowed swimming, short walks, cycling and horse riding without restriction¹⁸.

Sinding-Larsen-Johansson disease: what do we really know about this apicitis?

Not much, and that must be said clearly. It is the entity in this article with the thinnest evidence base: a founding series of eight patients, a few case reports, no trial. Which does not prevent it from being recognised reliably, nor from knowing what becomes of it.

Sinding-Larsen-Johansson disease is distress of the proximal insertion of the patellar tendon on the inferior pole of the patella, in an adolescent in whom that pole is not yet ossified. It arises from the same mechanism as Osgood-Schlatter disease, at the other end of the same tendon: repeated traction on an ossification centre still maturing. It is the general rule for overuse injuries in the young athlete, which occur at the growth centres, epiphyses and apophyses, because those are the weakest zones of the immature skeleton²⁵.

The original description is a prospective study of children attending with knee pain, in which eight patients, that is ten knees, showed the clinical and radiographic signs of the disease. The authors attributed the picture to a traction tendinitis with de novo calcification in the partly avulsed proximal insertion of the patellar tendon, and followed these patients to resolution: the course proved self-limiting and benign, like that of Osgood-Schlatter disease²¹.

The diagnosis is clinical, and it is simple

The picture combines anterior knee pain of gradual onset, made worse by jumping, running and climbing stairs, with exquisite tenderness on palpation of the inferior pole of the patella, sometimes with local swelling. The pain increases when the patellar tendon is tensioned in flexion. A case report documented by imaging describes exactly that picture in a 13-year-old footballer, with ultrasound confirmation of the characteristic lesions, and complete recovery after five months away from sport, the follow-up ultrasound showing no remaining abnormality²².

Ultrasound is not essential to the diagnosis, but it is the most suitable investigation when a doubt exists: it shows the thickening of the tendon, the ossification fragments and any associated bursitis, with no radiation.

What separates this disease from Osgood-Schlatter

The question comes up constantly in practice, because both entities occur in the same profile of sporting adolescent and give anterior knee pain. The distinction is made on palpation, in a few seconds, and it has a limited practical consequence for treatment but a real one for what is said to the patient and for the duration announced.

The two can coexist in the same adolescent, which poses no management problem since the principle is identical: adapt the load on the tendon rather than impose strict rest.

The trap of this benign entity

It exists, and it is documented. A 12-year-old treated conservatively for Sinding-Larsen-Johansson disease sustained, during a high jump, a sleeve avulsion of the inferior pole of the patella, a rare injury of the immature skeleton which the authors recall is easily missed on plain radiographs, with well-documented poor results when management is delayed. The diagnosis required magnetic resonance imaging, and the lesion was fixed surgically²³.

That case does not call the benign nature of the apicitis into question, it defines the alarm signal: an abrupt change in the picture, pain that becomes acute and intense after an identified movement, loss of quadriceps function, inability to extend actively, fall outside the frame of a growth condition and call for an opinion. Finally, a few refractory forms have been the subject of published surgical treatment, as in a professional handball player treated arthroscopically²⁴ : these are exceptions, cited here as a reminder that pain lasting beyond several months despite well-conducted load adaptation deserves reassessment rather than prolongation.

Osteochondritis dissecans: how do you avoid confusing it with patellofemoral pain?

By relying neither on the location of the pain, which is vague, nor on the clinical examination, which has no reliable test, but on two simple elements: the presence of an effusion, and the persistence of symptoms despite well-conducted management.

Osteochondritis dissecans is a disorder of subchondral bone and of the articular cartilage covering it. A bone fragment gradually demarcates, may partly detach, then become loose in the joint. The outcome depends on a factor the physiotherapist does not control but which governs everything: the degree of skeletal maturity. In the child whose growth plates are open, the so-called juvenile form has a potential for spontaneous healing that the adult form no longer has.

Who is affected, and where the lesions sit

Two studies with known denominators frame the epidemiology. The first analysed a cohort of more than a million children in an integrated health system: 192 patients with 206 knee lesions, that is an incidence of 9.5 per 100,000 in 6- to 19-year-olds, with 15.4 in boys against 3.3 in girls, and 11.2 in 12- to 19-year-olds against 6.8 in 6- to 11-year-olds. The medial femoral condyle carried 63.6 % of lesions. No lesion was found in children aged 2 to 5²⁶. The second, in the general population of Olmsted County over nearly forty years, gives an incidence of 6.09 per 100,000 person-years, with a very clear peak between 11 and 15 years: 39.06 per 100,000 in boys and 16.15 in girls in that band²⁸.

The anatomical distribution is documented precisely by the North American prospective multicentre cohort dedicated to this condition, which included 1,004 knees in 903 patients, 68.9 % of them boys, of median age 13.1 years. The lesions sat in the medial femoral condyle in 66.2 % of cases, the lateral femoral condyle in 18.1 %, the trochlea in 9.5 %, the patella in 6.0 % and the tibial plateau in 0.2 %. A majority of these patients, 68.1 %, played several sports²⁷.

Why clinical examination is not enough

Wilson's sign is the test most often cited for this condition. Described in 1967, it consists of reproducing the pain by taking the tibia into internal rotation during knee extension, between 90 and 30 degrees of flexion, then making it disappear in external rotation. Its validity was re-examined in 32 patients with osteochondritis of the medial femoral condyle, 17 aged 9 to 12 and 15 aged 13 to 17: 24 of those 32 patients, that is 75 %, had a negative Wilson sign while the lesion was visible on radiographs. The authors conclude that the sign has minimal diagnostic value, and usefulness as a follow-up marker when it is initially positive³⁰.

There is no other validated clinical test for this condition. The diagnosis therefore rests on suspicion, triggered by the setting and by two elements: a joint effusion, and the absence of improvement in anterior knee pain that has been correctly managed.

Non-operative treatment, its results and its limits

The most complete systematic review retained 27 studies covering 908 knees. Its first result is methodological and deserves quoting as it stands: no high-level study was found, the whole consisting of 24 case series and 3 case reports. The overall healing rate is 61.4 %, with considerable variability from 10.4 to 95.8 % depending on the series. The poor prognostic factors identified are a large lesion, an advanced stage, older age and skeletal maturity reached, the presence of a discoid meniscus, and a clinical presentation combining swelling or locking²⁹.

The most useful point for physiotherapy practice is what that review does not support. The authors conclude that restricting intense sporting activity is the most favourable approach, possibly combined with physiotherapy, but that there is no evidence that instrumental physical therapy, immobilisation or restricted weight-bearing brings any benefit²⁹.

A systematic review devoted to return to sport after a stable lesion in adolescents, covering 13 studies, 710 patients and 783 knees aged 8 to 18, completes that picture. All patients had tried at least six weeks of non-operative treatment. The success rate of non-operative treatment ranged from 40.3 to 87.5 % depending on the series, and among those treated successfully without surgery, return to sport ranged from 84.7 to 100 %³¹.

And for the ankle

The talar location exists and follows similar logic, with a literature that is rarer still in the growing patient. The principle stays the same: in an adolescent, an ankle that stays persistently painful after what has been labelled a sprain, especially if it swells intermittently or locks, should raise an osteochondral lesion of the talar dome and lead to imaging. The practical point of vigilance is identical to that of the knee: it is persistence beyond the expected duration that triggers imaging, not the intensity of the pain.

What rehabilitation for these four conditions, and at what level of evidence?

The honest answer comes down to one sentence: none of these four entities has a randomised controlled trial evaluating physiotherapy. This chapter therefore sets out what the literature genuinely supports, graded by the solidity of what backs it, and not by habit.

The level of evidence in paediatric rehabilitation for these conditions is structurally low, for a reason that is not methodological laziness: these diseases are rare, their course is measured in years, and their relevant outcome is a joint shape at skeletal maturity. Mounting a randomised trial on an incidence of 2.5 to 10 per 100,000 with ten years of follow-up is an exercise that has succeeded only once, for the surgical question alone, and whose repetition the authors of the British cohort are now calling for¹¹.

What is left to the physiotherapist, and it is not nothing

A quick reading of this chapter might give the impression that rehabilitation has no place here. It is the opposite, provided the objective is shifted. These children and adolescents go through long, often painful diseases that keep them away from sport at an age when physical activity structures social life. Three missions are defensible and documented, at least by consensus.

• Maintain range, abduction and hip internal rotation in priority in Perthes: it is the only real point of international agreement^17,18.
• Maintain compatible physical activity. Swimming, cycling, short walks and horse riding are the subject of broad agreement in Perthes^17,18. A child who stays active keeps their fitness, their sleep and their social network.
• Reassess, and know when to refer back. It is the most specific mission of the physiotherapist in this group, because they are the professional who sees the child most often. The mean 17-week delay to diagnosis of slipped epiphysis⁶ plays out in exactly that window.

What do concrete clinical cases teach us?

Six published cases, all verifiable by their identifier. They are chosen because each illustrates a different turning point, and because none ends with immediate recognition of the diagnosis.

A normal knee, a diseased hip

A 7-year-old boy attends for recurrent left knee pain. The physical examination of the knee, the blood tests and the knee radiographs are all normal. It is the radiographs then requested on the hypothesis of hip involvement that show a focal cortical defect of the left femoral head and a larger effusion in the left hip than on the opposite side. The authors, paediatricians, publish this case precisely to recall that knee pain with no local cause must lead to examination of the hip⁴².

The same mechanism is the subject of a case published in a sports physiotherapy journal, in an 8-year-old athlete whose parapatellar pain was of hip origin. The author recalls that parapatellar pain in the active adolescent is usually attributed to patellofemoral pain syndrome, to Osgood-Schlatter disease, to Sinding-Larsen-Johansson disease or to a tendinopathy, and that every source of pain, primary as well as referred, must be assessed⁴⁴.

Two growth diseases in the same child

A boy treated at 7 for Legg-Calvé-Perthes disease of the left hip by femoral osteotomy sees his femoral head heal spherical and congruent. Three years later, he develops tibia vara on the right, treated by growth modulation. At 15, he presents with a painful limp on the left and the typical signs of a slipped capital femoral epiphysis, in the very hip that had healed. The authors' conclusion deserves quoting for what it implies: a slipped epiphysis must always be considered in an adolescent with hip pain or knee pain or a gait disturbance, even if they have had other conditions in the past such as Perthes disease⁴³.

That case is rare and the authors present it as such. Its value to the physiotherapist lies elsewhere: a known orthopaedic history is a powerful anchor. When an adolescent followed for the sequelae of Perthes starts limping again, the explanation that comes to mind is the sequelae. It can be wrong.

An apicitis that ends in a fracture

A 12-year-old athlete treated conservatively for Sinding-Larsen-Johansson disease is referred urgently for severe pain and loss of function after a high jump. Plain radiographs and magnetic resonance imaging confirm a sleeve avulsion of the inferior pole of the patella, fixed surgically with two transosseous bands. The authors stress that these injuries are easily missed on plain films, with well-documented poor results when there is a delay²³.

Conversely, the usual course of this disease is illustrated by a 13-year-old footballer whose anterior knee pain and swelling of the inferior pole of the patella were confirmed on ultrasound, and in whom stopping sport allowed complete recovery in five months, with a normal follow-up ultrasound²². The two cases frame the clinical reality: the rule is slow recovery, the exception is the acute accident, and it is the abrupt change in the picture that separates them.

Patellofemoral pain that was not

A 17-year-old elite footballer is spotted running with an antalgic gait, and then reveals knee pain that had been worsening for two months. The picture is one of patellofemoral pain with a small effusion and quadriceps wasting. After two weeks without running and classic patellofemoral treatment, there is no improvement. Magnetic resonance imaging and computed tomography reveal an osteochondritis dissecans of the trochlea, a location accounting for less than 1 % of cases. The player is operated on, follows a four-stage rehabilitation defined by objective criteria, and returns to competition 24 weeks after the operation³².

That case is the most directly transferable of all. The physiotherapist is the one who saw the antalgic gait, the one who raised the patellofemoral hypothesis, and the one who noted the absence of response. It is that last observation that produced the diagnosis. The authors conclude by recommending that osteochondritis dissecans be considered in the assessment of any persistent knee pain in the young athlete.

How do you apply all this concretely in practice?

An operational synthesis, organised in the real order of a consultation: what you rule out, what you examine, what you request, what you say.

At first contact, four questions

1. Is there fever, refusal to bear weight, night pain that wakes the child, or deterioration in general condition? If so, the child needs a medical opinion the same day, and not rehabilitation. The four criteria of the septic arthritis prediction rule are scored in a minute³³, and the combination of night pain plus subtle blood count abnormalities points to a blood disorder in three quarters of children who nevertheless have no circulating blasts³⁶.
2. How old is the child? Before 8 years, the hip is the first suspect. Between 9 and 16, slipped epiphysis must be ruled out before any weight-bearing. After 10 years, the knee enters the discussion with patellar apicitis and osteochondritis.
3. Is the pain unilateral and mechanical? Bilateral evening pain, in a child who runs normally the next day, with no limp or restriction, fits the frame of so-called growing pains, whose criteria are themselves debated⁴¹. A limp never fits it.
4. Is there a knee effusion? A knee that swells falls outside the frame of the apophysitides, which are extra-articular, and points to an intra-articular lesion, osteochondritis dissecans first of all²⁹.

On examination, three actions that take two minutes

• Comparative hip internal rotation, supine, hip and knee flexed to 90 degrees. Asymmetry in a child with knee pain is a warning sign for both hip conditions in this group²⁰.
• Passive hip flexion looking for obligatory external rotation, the most suggestive sign of slipped epiphysis.
• Level-by-level palpation of the extensor mechanism : inferior pole of the patella for Sinding-Larsen-Johansson disease, tibial tuberosity for Osgood-Schlatter disease, body of the tendon for a tendinopathy. Three points two to three centimetres apart, three different diagnoses.

What to request, and how to word it

When hip involvement is suspected, the request must specify a pelvic radiograph, AP and frog-leg lateral view. The word “pelvic” matters, because bilaterality is common and often asymptomatic⁹ ; mentioning the lateral view matters more still, since an AP view alone misses more than half of slipped epiphyses⁷. In the child under 5 who limps, the imaging appropriateness criteria published by the American College of Radiology are a useful reference for discussing the choice of investigations with the doctor³⁹.

For an adolescent's knee that is not improving, the most useful wording is not a presumed diagnosis but a fact: persistent exertional pain after six weeks of management, with or without effusion, in a growing patient. That is what triggers imaging in the published series^31,32.

What to say to the family

• In Perthes : the disease runs over years, physiotherapy maintains range and allows activity to be kept up, it does not change the shape the femoral head will take. That sentence is hard to say, and it prevents a far harder disappointment later.
• In slipped epiphysis : it is a relative emergency, walking must stop, and management is surgical. The opposite side will be monitored.
• In Sinding-Larsen-Johansson disease : it is benign, it heals, but it is measured in months and not in weeks. The case best documented by imaging took five months²². Announcing three weeks produces abandonment of treatment in the fourth.
• In osteochondritis : healing without surgery is possible and common, but it is not guaranteed, with an overall rate of 61.4 % and wide variability²⁹. Return to sport is the rule in those who heal without surgery³¹.