managing children with acute non-traumatic limp: the utility of clinical findings, laboratory...

PAEDIATRIC EMERGENCY MEDICINE

Managing children with acute non-traumaticlimp: the utility of clinical findings, laboratoryinflammatory markers and X-raysLucy Reed,1 Abby Baskett2 and Nicholas Watkins3

1Bristol Royal Infirmary, University of Bristol NHS Trust, Bristol, UK; and 2Starship Children’s Hospital,Auckland, New Zealand; and 3Royal Hobart Hospital, Hobart, Tasmania, Australia

Abstract

Objectives: To examine the utility of clinical findings, laboratory markers and X-ray radiographs(X-ray) in the assessment of children presenting with an acute non-traumatic limp.

Methods: A retrospective review of all children who received hip X-rays over a 2 year period in theChildren’s Emergency Department, Starship Children’s Hospital, Auckland, New Zealand.Children were identified from the radiology database and clinical notes reviewed. Childrenaged 0–12 years old were included if the limp was acute (less than 2 weeks of duration) withno history of trauma. X-rays were reported by a consultant paediatric radiologist. Univari-ate and multivariate analysis was performed to determine predictors of osteomyelitis andseptic arthritis. Receiver operator curves were used to assess the optimum cut-off points forC reactive protein (CRP), erythrocyte sedimentation rate (ESR) and white cell count (WCC).

Results: A total of 350 patients were enrolled. There were 21 (6%) abnormal X-rays . Fever,non-weight bearing, raised white cell count, raised erythrocyte sedimentation rate andraised CRP were all associated with increased risk of septic hip or osteomyelitis. Theoptimum inflammatory marker cut-off was a CRP of 12 with a sensitivity of 87% andspecificity of 91%.

Conclusion: In acute non-traumatic limp, X-rays of the hips diagnose slipped upper femoral epiphysis,as such they should be routinely used from the age of 9 years upwards. Below this agethey are of little value. Inflammatory markers have utility in risk-stratifying children andselecting a group in whom to proceed with definitive tests to exclude osteomyelitis or septichip. Children with a short history and minimal symptoms can be managed with appropri-ate follow up and no investigations.

Key words: hip, inflammatory marker, limp, paediatric, X-ray.

Correspondence: Dr Nicholas Watkins, Royal Hobart Hospital, RHH, Liverpool Street, Hobart, Tas. 7000, Australia. Email:[email protected]

Lucy Reed, MB ChB, FCEM, Emergency Physician; Nicholas Watkins, FRACP, FACEM, Emergency Physician; Abby Baskett, MB BS, Fellow inPaediatric Emergency Medicine.

doi: 10.1111/j.1742-6723.2008.01144.xEmergency Medicine Australasia (2009) 21, 136–142

© 2008 The AuthorsJournal compilation © 2008 Australasian College for Emergency Medicine and Australasian Society for Emergency Medicine

mailto:[email protected]

Introduction

Acute non-traumatic limp in children is a frequent pre-sentation to the ED. The majority of children have anirritable hip: a simple, self-resolving disorder that wouldmost appropriately be managed with reassurance,analgesia and no investigations.

Unfortunately, the differential diagnosis of this pre-sentation is broad and includes a small number ofserious conditions that should not be missed. For thepurposes of the present study, the major differentialsconsidered were: septic arthritis, osteomyelitis, slippedupper femoral epiphysis (SUFE) and Perthes disease.It is, however, important to remember that infectionanywhere in the spine or leg can cause a limp, as canunwitnessed or non-accidental trauma. Acute haemato-logical, neurological and gastrointestinal conditionsmight also manifest as limp.

There have been several studies comparing differentalgorithms for assessing acute limp,1–9 based on combi-nations of clinical signs, blood inflammatory markersand radiological investigations. There remains consid-erable variation in recommendations and no consensusas to the best approach.

The purpose of the present study was to review ourcurrent assessment pathway; to further define the roleof clinical findings and inflammatory markers in theassessment of this population, and to determine whetherit is possible to reduce the number of plain radiographsof the hip and pelvis without missing importantdiagnoses.

Methods

This was a retrospective study reviewing patientspresenting between June 2003 and June 2005 to theChildren’s Emergency Department (CED) of the Star-ship Children’s Hospital, Auckland, New Zealand. TheCED has an annual census of approximately 32 000patients up to 15 years of age. The catchment ismulticultural with a large Maori and Pacific Islandpopulation.

Analysis of the hospital’s radiology database enabledidentification of all children, up to the age of 12 years,who had their pelvis or hips X-rayed between June 2003and June 2005.

This unusual method of case ascertainment was usedas it was departmental policy to X-ray the hips of allpatients presenting with a painful hip or a limp where

hip pathology was suspected. Patients X-rayed prior toarrival had their films digitalized onto the Starship radi-ology system.

To validate this methodology, the digital medicalrecords for every presentation to the CED over a10 week period were reviewed. This sampling con-firmed that 100% of 32 patients presenting with anacute limp or painful hip were X-rayed as part of theirassessment.

The clinical notes were reviewed on digital medicalrecords by the principal researcher.

Children were included if they had presented to theCED with a history of acute hip pain or an acute limpwith symptom duration of 2 weeks or less. The lowerage range was defined developmentally; any child whohad been walking and had stopped was included. Theyoungest child was 11 months. The upper age cut-off of12 years was chosen, as beyond this age irritable hip isan uncommon diagnosis with SUFE and sporting inju-ries being more prevalent. Two weeks was chosen asthe maximum duration for symptoms, as it was felt thatplain radiography might well reveal changes of Perthesor chronic osteomyelitis with more long-standing symp-toms. Furthermore, greater than 2 weeks of limp wouldeffectively exclude benign irritable hip.

Children were excluded if they had sustained signifi-cant trauma (e.g. pelvic films as part of a trauma series),if they had a history of hip pathology (e.g. congenitaldislocation), or a chronic illness that could affectjoints and bones, such as osteogenesis imperfecta, spinabifida, sickle cell disease or cerebral palsy.

Demographic and clinical data collected for eachchild included age, sex, ethnic origin, temperature atpresentation, duration of presenting symptoms andability to weight bear. X-ray reporting was by the con-sultant paediatric radiologist, all films were reportedwithin 24 h. The final diagnosis was either that at finaldischarge from CED or orthopaedic care whichever wasthe later. As Starship is the only children’s hospital inNew Zealand, it is reasonable to assume that childrendischarged with missed bone or joint infection wouldrepresent to our institution. Children were recorded ashaving irritable hip if the final diagnosis was irritablehip, transient synovitis, reactive arthritis or ‘no causefound’. Patients classified as osteomyelitis included twowith a psoas abscess and one with a hip adductorabscess.

Laboratory data collected were white cell count(WCC), erythrocyte sedimentation rate (ESR) and Creactive protein (CRP). Again, performing these testswas department policy and data were complete for all

Managing acute limp

137© 2008 The AuthorsJournal compilation © 2008 Australasian College for Emergency Medicine and Australasian Society for Emergency Medicine

patients with a final diagnosis of osteomyelitis or septichip. Fifty-nine patients with a final diagnosis of irritablehip had incomplete laboratory data and were notincluded in Table 3.

Univariate and multivariate analysis of clinical andinflammatory markers was performed to determinetheir utility in identifying bacterial bone and joint infec-tion. Receiver operator curves were used to determineoptimum cut-off values for the blood inflammatorymarkers.

Results

A total of 350 patients met the criteria for inclusion.All of these patients were X-rayed as part of theirclinical assessment. There were 21 (6%) abnormal films(Table 1). X-ray was 100% sensitive and 100% specificin the diagnosis of SUFE. The youngest child with thisdiagnosis was 9 years of age.

Hip X-ray added additional diagnoses, beyond thefive principal diagnoses considered, in only twopatients: an incidental finding of a bone cyst and asingle case of idiopathic chondrolysis. The case of apo-physitis was principally a clinical rather than a radio-logical diagnosis.

Inflammatory markers

Results of univariate and multivariate analysis aredisplayed in Table 2. Children of Maori and PacificIsland origin did carry a higher burden of bacterialbone and joint infection, but the association withethnic group was weak. CRP and temperature were thebest predictors in both univariate and multivariateanalysis.

Among the blood inflammatory markers, WCC wasthe weakest individual predictor and CRP the strongest(as per receiver operator curves, Fig. 1).

From a clinical perspective, it is disappointing to notethat even a modest rise in a single inflammatory markerhas poor sensitivity for identifying bone or jointinfection.

Given this observation, markers were analysedcumulatively to see if a useful ‘rule in’ or ‘rule out’combination could be identified. Five predictors wereused: non-weight bearing, temperature �38.5oC, WCC�12 ¥ 109/L, ESR �30 mm/h and CRP �12 mg/L.Results are displayed in Tables 3 and 4.

As can be seen, this combination appears useful:80% of those with �3 abnormal markers had a bone

or joint infection and none of those with 0 abnormalmarker. It would seem reasonable that their manage-ment strategies could be effectively decided on thisalone (Appendix 1). Unfortunately, for one-third of thepopulation the risk is intermediate (4% incidence withone abnormal marker, 30% incidence with two abnor-mal markers) and decisions must be made on a case-by-case basis.

Discussion

In this series, irritable hip was the most common diag-nosis in all age groups: it represented approximately85% of presentations during the first decade, afterwhich it began to become less common. Osteomyelitisoccurred at a similar rate, representing approximately8% of presentations, across all ages. Only sevenpatients (2%) had a septic hip and all of these wereunder the age of 5 years.

The prevalence of bone or joint sepsis in the presentstudy is noted to be higher than the 0–1% reported fromthe UK,1 which is in line with the authors’ expectationfor the Auckland region population.

In children aged 9 years and older, 11 of the 62 X-rayswere abnormal (18%). This included nine cases ofSUFE, one of idiopathic chondrolysis and one of irri-table hip. Plain X-ray had a sensitivity and specificity of100% in diagnosing SUFE. In a previous study fromAuckland,10 there was a higher incidence of SUFE inNew Zealand Maori and Pacific Island children. In theseethnic groups, the youngest age at diagnosis was8 years.

Although rare, SUFE should be considered at aslightly younger age in children with specific riskfactors (which would include ethnic predisposition,marked obesity, history of endocrinopathies or radia-tion therapy).

In children younger than 9 years of age, 10 of 288X-rays were abnormal (3%). This included one inciden-tal finding of a simple bone cyst in a patient with a finaldiagnosis of irritable hip. The other nine abnormalitiesrelated to more subjective comments on joint spacewidening or prominence of soft tissue shadows. Two ofthese patients did have bone infections, one with aseptic hip and one with osteomyelitis. The reports werenot specific for and not diagnostic of the condition.

There was one patient with a final diagnosis of Perthesdisease. They were seen on day six of symptoms and theX-ray was reported as normal at that stage.

L Reed et al.

138 © 2008 The AuthorsJournal compilation © 2008 Australasian College for Emergency Medicine and Australasian Society for Emergency Medicine

The major point of interest to be taken from the hipX-ray data is how little contribution they make outsidethe diagnosis of SUFE.

The present study completely supports the notionthat hip X-rays are not sensitive in diagnosing jointeffusion11 and that plain radiological changes seen inosteomyelitis and Perthes disease occur later, when thedisease is more established. Furthermore, the return of

only one case of idiopathic chondrolysis from 350 hipX-rays leads to the conclusion that hip X-ray cannot bejustified as a routine investigation looking for ‘other’pathology in this cohort of patients.

Interestingly, five patients less than 5 years oldwere diagnosed with lower limb fractures. None ofthese was seen on the hip X-rays. There were threetoddler’s fractures (spiral tibial fractures) from

Table 1. Details of X-ray data

Final diagnosis Reported normal Reported abnormal Abnormal report comment

Irritable hip 286 7 4 widened joint space or effusion1 possible early Perthes1 possible ischiopubic osteomyelitis1 soft tissue abnormality

Septic hip 6 1 Suggestive of joint effusionOsteomyelitis 30 1 Possible soft tissue collection lateral to hipSUFE 0 9 9 SUFEPerthes 1 0 Repeat film at follow up confirmed PerthesIrritable hip and simple bone cyst 0 1 Lytic lesion proximal femurIdiopathic Chondrolysis 0 1 Loss of joint spaceApophysitis 0 1 Widened joint space

SUFE, slipped upper femoral epiphysis.

Table 2. Statistical analysis of the data

Infective Non-infective Univariate relative risk ratio (95% CI) Multivariable relative risk ratio (95% CI)

Age group (years)<5 19 126 1.00 1.005+ 19 166 0.76 (0.39–1.49) 0.84 (0.23–3.05)

Ethnic originEuropean 16 189 1.00 1.00Maori/Pacific 14 55 3.01 (1.38–6.54) 1.04 (0.24–4.33)IslanderOther 8 49 1.93 (0.78–4.77) 1.75 (0.33–9.29)

Weight bearingNo 18 65 3.16 (1.58–6.32) 2.20 (0.62–7.81)Yes 20 228 1.00 1.00

Temperature (°C)�38.5 21 286 1.00 1.00>38.5 17 6 38.59 (13.77–108.17) 13.18 (2.45–70.86)

WCC (¥109/L)0–12 18 251 1.00 1.00>12 19 30 8.83 (4.18–18.65) 3.73 (1.10–12.65)

ESR (mm/h)0–30 15 206 1.00 1.00>30 22 18 16.79 (7.44–37.89) 4.85 (1.38–17.08)

CRP (mg/L)0–12 5 196 1.00 1.00>12 32 21 59.73 (21.02–169.75) 27.52 (7.00–108.13)

CRP, C reactive protein; ESR, erythrocyte sedimentation rate; WCC, white cell count.

Managing acute limp


unwitnessed falls and two cases of non-accidentalinjury: a child with an unexplained distal femoral frac-ture and a child with bilateral tibial fractures. Thishighlights that identifying the exact location of pain ina preschool-age child can be very challenging and thatX-ray for toddlers fracture is worth considerationwhen clinical signs are vague.

Ultrasound has been widely advocated as a valuabletool in the assessment of limping children, as it has highsensitivity for diagnosing hip effusion.8,9,11 However, itis important to remember that it cannot differentiate areactive effusion from septic arthritis.

Fink et al.8 studied 50 children with painful hips, per-formed ultrasound on all and aspirated all effusionsseen (72% aspirated). The authors comment that inflam-matory markers and fever are not adequate for predict-ing joint sepsis, as some will have normal parameters.Aspiration and culture of synovial fluid gives an earlydefinitive diagnosis, can relieve the child of pain andallow same-day discharge from hospital.

Even if Fink’s assertions regarding the utility ofinflammatory markers and fever are accepted, the lowincidence of joint sepsis (2%) and the high incidence ofirritable hip (~85%) would lead many to consider thisapproach excessively invasive.

Beach has advocated only ultrasounding childrenwith high inflammatory markers in whom a septic hip isbeing seriously considered and then aspirating if aneffusion is found.1

Certainly, given that ultrasound provides no insightas to the aetiology of a joint effusion, it would seemreasonable to only use this investigation in children

Cut-off value Sensitivity (%) Specificity (%)

WCCx109/L

Optimum 90% specificity 90% sensitivity

10.812†

6.5

67.65490

77.488.912

ESRmm/h

Optimum 90% specificity 90% sensitivity

16.530†

7.5

816590

79.590.648

CRPmg/L

Optimum 90% sensitivity

12.2†

6.886.590

90.781

Figure 1. Receiver operator curves for inflammatory markersand values for selected cut-offs. ( ) WCC, A = 0.71; ( )CRP, A = 0.94; ( ) ESR, A = 0.85. †Cut-off used in diagnosticalgorithm. CRP, C reactive protein; ESR, erythrocyte sedimenta-tion rate; WCC, white cell count; X, optimum point.

Table 3. Abnormal markers used were: non-weight bearing, temperature �38.5°C, WCC �12, ESR �30, CRP �12

No. of abnormal markers Irritable hip (n = 234) Osteomyelitis (n = 31) Septic hip (n = 7)

0 144 0 01 67 3 02 17 8 03 or more 6 20 7

CRP, C reactive protein; ESR, erythrocyte sedimentation rate; WCC, white cell count.

Table 4. Cumulative numbers of abnormal markers: likelihood ratio, sensitivity and specificity for osteomyelitis/septic arthritis

No. of abnormal markers Interval likelihood ratio Sensitivity (%) Specificity (%)

1 or more 2.5 (2.1–3.1) 100 60.32 or more 8.9 (5.6–14.5) 91.4 89.73 or more 26.4 (10.2–71.3) 71.4 97.3

L Reed et al.


where there is a particular suspicion of septic hip orosteomyelitis, rather than all children where the cause oflimp localizes to the hip.

Inflammatory markers

Previous studies have focused on a combination ofWCC, ESR and CRP as part of clinical prediction algo-rithms to differentiate between irritable hip and osteo-myelitis or septic hip.1–7,12,13

Kocher et al. examined four clinical predictors (fever,non-weight bearing, ESR �40 mm/h, WCC >12 ¥ 109

cells/L).2,3 With none present, there was a <0.2% chanceof septic arthritis being present. With one predictor itwas 3%, two predictors 40%, three predictors 93.1%and, if all four predictors were present, then there was a99.6% likelihood of septic arthritis.

Luhman et al. partially validated Kocher’s predictionalgorithm, but found that when all four predictors werepositive, the predicted probability of septic arthritis wasonly 59%.4

The current study supports the observation that themore clinical predictors present, the higher the likeli-hood of osteomyelitis or a septic hip.

In this cohort, all of the patients with septic hips wereunder 5 years old, were non-weight bearing and had atleast three markers abnormal out of the five analysed(Table 4). Those with osteomyelitis were more difficultto diagnose; they were frequently able to weight bearand approximately one-third had only one or two abnor-mal markers overall.

There are, however, a number of useful points thatcan be extracted from this data set to aid in manage-ment decisions:1. Forty-five per cent of children (n = 132) with an

irritable hip and only one case of osteomyelitis pre-sented afebrile, able to weight bear and with symp-toms for less than 48 h.

2. Greater than 80% of patients with three or moremarkers abnormal would have osteomyelitis orseptic hip, whereas only 1% of patients with one orfewer abnormal marker would be expected to haveosteomyelitis.

3. It appears useful to measure all the three of WCC,ESR and CRP as individually their predictive value ispoor. It was surprising how often one marker wouldbe very elevated and others normal, for example,ESR 140, CRP 4.4, or CRP 412, WCC 11.2, or CRP52.5, WCC 28.9, ESR 10.

The observation in point one above is interesting.This group, which represents more than one-third ofthe entire cohort, should be easy to identify clinicallyand has an incidence of bone or joint infection ofonly 0.75%, compared with 10% for the cohort as awhole. Given appropriate advice and follow up, itmight well be that many families from this low-riskgroup would opt for a strategy of observation onlythat did not, at least initially, entail their child havingblood tests. It appears safe and appropriate to offerthis option.

Limitations

The present study has the usual limitations of retro-spective design with the risks of missed cases andincomplete data. Our pilot study, however, suggeststhat the missed case rate was low and we were fortu-nate that there was such close adherence to depart-ment protocol.

The prevalence of bone and joint infection variesconsiderably between reported series and this mightaffect the applicability of results to different popula-tions. However, in the Auckland paediatric population,the burden of bone and joint infection appears to befairly high for a developed nation. It would thereforeseem that the major problem with translating data fromour site might be a tendency to over investigation inpopulations with comparatively lower pretest probabili-ties for these diagnoses.

Conclusion

In the context of an acute limp of less than 2 weeks ofduration, X-rays diagnose SUFE, rarely make ‘other’diagnoses and do not help differentiate between irritablehip, osteomyelitis and septic arthritis. They should beused routinely for children 9 years of age or older, butbe used only where specifically clinically indicated inyounger children.

The chance of missing osteomyelitis or septic hip in awell-looking, afebrile child who is mobile but limpingand has had symptoms for 48 h or less is very low. Itappears safe to manage this group with advice, followup and no investigations.

Increasing numbers of positive markers is stronglyassociated with an increased likelihood of bone or jointinfection. This can be used to select a group in whom toproceed with definitive testing.

Managing acute limp


Although it might be possible to identify almost allcases of septic hip, there is currently no algorithm thatwill easily identify all cases of osteomyelitis. Recogni-tion of this is important to enable planning of appropri-ate management and follow up.

Acknowledgement

Many thanks to John Thompson of the University ofAuckland for his statistical analysis.

Competing interests

None.

Accepted 21 November 2008

References

1. Beach R. Minimally invasive approach to management of irri-table hip in children. Lancet 2000; 355: 1202–3.

2. Kocher MS, Zurakowski D, Kasser JR. Differentiating betweenseptic arthritis and transient synovitis of the hip in children: anevidence-based clinical prediction algorithm. J. Bone Joint Surg.Am. 1999; 81: 1662–70.

3. Kocher MS, Mandiga R, Zurakowski D, Barnewolt C, Kasser JR.Validation of a clinical prediction rule for the differentiationbetween septic arthritis and transient synovitis of the hip inchildren. J. Bone Joint Surg. Am. 2004; 86-A (8): 1629–35.

4. Luhman SJ, Jones A, Schootman M, Gordon JE, Schoenecker PL,Luhmann JD. Differentiation between septic arthritis and tran-sient synovitis of the hip in children with clinical predictionalgorithms. J. Bone Joint Surg. Am. 2004; 86-A (5): 956–62.

5. Delaney RA, Lenehan B, O’sullivan L, McGuinness AJ, Street JT.The limping child: an algorithm to outrule musculoskeletalsepsis. Ir. J. Med. Sci. 2007; 176: 181–7.

6. Del Baccaro MA, Champoux AN, Bockers T, Mendelman PM.Septic arthritis versus transient synovitis of the hip: the value ofscreening laboratory tests. Ann. Emerg. Med. 1992; 21: 1418–22.

7. Eich GF, Superti-Furgon A, Umbricht FS, Willi UV. The painfulhip: evaluation of criteria for clinical decision making. Eur. J.Pediatr. 1999; 158 (11): 923.

8. Fink AM, Berman L, Edwards D, Jacobson SK. The irritable hip:immediate ultrasound guided aspiration and prevention of hos-pital admission. Arch. Dis. Child 1995; 72: 110–13.

9. Fischer SU, Beattie TF. The limping child: epidemiology, assess-ment and outcome. J. Bone Joint Surg. Br. 1999; 81: 1029–34.

10. Stott S, Bidwell T. Epidemiology of slipped capital femoral epi-physis in a population with a high proportion of New ZealandMaori and Pacific children. NZMJ. 2003; 116: 1184.

11. BestBets. Ultrasound is Better than X Ray at Detecting Hip Effu-sions in the Limping Child. [Cited 19 February 2003.] Availablefrom URL: http://www.bestbets.org.uk

12. BestBets. ESR in Childhood Septic Arthritis. [Cited 11 July 2001.]Available from URL: http://www.bestbets.org.uk

13. BestBets. CRP in Septic Arthritis in Children. [Cited 7 April 2004.]Available from URL: http://www.bestbets.org.uk

Appendix 1

Use of inflammatory markers in differentiating irritablehip from osteomyelitis/septic arthritis:(1) Child afebrile (temperature <38°C) and well, symp-toms for �48 h and able to walk with a limp.

If all of the above, then no investigations arenecessary.

Family should return if: symptoms worsen, childbecomes febrile or does not get better in the next4–5 days.

If not, then go to (2).(2) Take FBC, ESR and CRP

(a) 3 or more of:� non-weight wearing,� temperature �38.5°C,� WCC �12 ¥ 109/L,� ESR �30 mm/h,� CRP �12 mg/L

Refer for exclusion of osteomyelitis or septic hip.(b) One or two abnormal results as above: thesepatients are at intermediate risk of bone or jointinfection.

Factors to consider are: duration of symptoms,degree of limitation of hip movement, general con-dition of child, degree of abnormality of temperatureand laboratory markers.

Experienced doctor required to select mostappropriate management path on case-by-case basis.(c) No abnormal results as above: these children areat low risk. Discharge with advice as in (1).

L Reed et al.


http://www.bestbets.org.uk



managing children with acute non-traumatic limp: the utility of clinical findings, laboratory...

Documents