Ž .International Journal of Gynecology & Obstetrics 63 1998 15]20

Article

Fracture of the clavicle in the newborn following normallabor and delivery

B. Kaplana,d,U , D. Rabinersona, O.M. Avrecha,d, N. Carmia, D.M. Steinbergb,P. Merlobc,d

aDepartment of Obstetrics and Gynecology, Rabin Medical Center, Petah Tiq¨a, IsraelbDepartment of Statistics and Operation Research, Tel A¨ i Uni ersity, Tel A¨ i , Israel

cDepartment of Neonatology, Rabin Medical Center, Petah Tiq¨a, IsraeldSackler Faculty of Medicine, Tel A¨ i Uni ersity, Tel A¨ i , Israel

Received 16 April 1998; received in revised form 8 June 1998; accepted 10 June 1998

Abstract

Ž .Objecti e: Earlier works have associated neonatal clavicular fracture 0.2]3.5% of all deliveries with a range ofprocedural, fetal and maternal risk factors; more recent studies, however, have failed to firmly identify any one or acombination of them. In the present work we sought to determine possible anterintra-partum risk factors for thiscondition. Study design: Using a retrospective case-controlled approach, we examined a series of maternal, fetal and

Ž .pregnancy or delivery-related variables in 87 cases out of 403 of fractured clavicle of the newborn diagnosed in ourŽ .department from 1986 to 1994. All infants were delivered vaginally in the occipito-anterior position at term by a

specialist obstetrician and underwent peripartum sonographic fetal weight estimation. All variables were comparedwith those of an equal number of infants born immediately before or after each affected infant and delivered by thesame obstetrical team. Results: Fractured clavicles were found in 1.65% of the total number of deliveries during thestudy period. Neonatal clavicular fracture was significantly and directly related to the duration of the second stage oflabor, peripartum sonographic fetal weight estimation, and neonatal length, and inversely related to maternal height;nevertheless, all values in both the study and control groups were within normal range. Logistic regression analysisshowed that these antenatal variables significantly affect the chances of clavicular fracture, but due to the highfalse-positive rate they cannot serve clinically as a comprehensive antenatal prediction index. Conclusions: Themajority of clavicular fractures occur in normal newborns following normal labor and delivery. The risk factors weidentified statistically do not offer a method for clinical prenatal prediction. This work provides statistical evidence ofthe nature of this complication of early newborn life. Q 1998 International Federation of Gynecology andObstetrics

Keywords: Fracture; Clavicle; Delivery; Complication; Neonate

U Corresponding author. Tel.: q 972 3 5355828; fax: q972 3 6354254.

0020-7292r98r$19.00 Q 1998 International Federation of Gynecology and ObstetricsŽ .P I I S 0 0 2 0 - 7 2 9 2 9 8 0 0 1 2 7 - 1

( )B. Kaplan et al. r International Journal of Gynecology & Obstetrics 63 1998 15]2016

1. Introduction

Newborn bone fractures are a known complica-w xtion of parturition 1,2 , with neonatal clavicular

fracture being the most frequent, at a rate ofw x0.2]3.5% of all deliveries 3 . Though early re-

Žports implicated a range of fetal high birth weight,. w x Žshoulder dystocia 4,5 , procedural induction of

labor, post-term pregnancy, prolonged labor and. Ž .instrumental vaginal delivery 6]13 and mater-

Žnal variables older age, obesity, diabetes and. Ž . w xshort stature 6]9 , more recent studies 1]3,14

have suggested that no one or a combination ofrisk factors can be firmly identified. Today, clavic-ular fracture is generally considered an unpre-dictable and unavoidable complication of normalbirth.

To test this assumption, we conducted a retro-spective case-controlled study of newborns withclavicular fracture.

2. Material and methods

The computerized database of the Departmentof Neonatology was reviewed for all cases ofneonatal clavicular fracture in singleton-term de-liveries from August 1986 to July 1994. Duringthis period, there were no changes in the institu-tional policy regarding management of delivery orthe criteria for diagnosis of neonatal clavicularfracture. Of the 403 cases identified, 87 wereincluded in the study. We included only thosedelivered by spontaneous vaginal delivery in theoccipito-anterior position by an experienced spe-cialist in obstetrics. Furthermore, only those who

Žalso underwent peripartum during the last 3 days.of pregnancy to immediately before delivery

Žsonographic fetal weight estimation all estima-tions were performed by an experienced techni-cian and when compared to the actual birth weightwere found to be accurate within a 5% range in

.average were analyzed. For each infant with aclavicular injury, the records of the baby bornimmediately before or after and delivered by thesame obstetrical team were examined as wellŽ .provided it, too, had all the inclusion criteria .

Every newborn in our department undergoes

routine physical examination twice: during thefirst 24 h of birth and later, before dischargeŽ .56]72 h postpartum . Diagnostic criteria for neo-natal clavicular fracture are: by inspection }asymmetry of the clavicular bones, absence of thesupraclavicular notch and local edema or hema-toma; by palpation } crepitations and local ten-

Ž .derness reflected by the baby’s cry . X-ray filmsare not performed unless Erb’s palsy or otherbrachial plexus injury is suspected.

The variables evaluated for the study and con-trol groups were:

Ž1. Maternal: age, ethnic origin Asian, African,Russian, Ethiopian, American, European or

.Arabic , parity, height, time elapsed sinceprevious delivery, previous maximal birthweight, presence of gestational diabetes ordiabetes in a first-degree relative, and weightgain during current pregnancy.

2. Fetal: peripartum sonographic fetal weightŽ .estimation already specified , birth weight,

birth length, head circumference, sex, intra-Žpartum fetal well being fetal heart rate mon-

itor, cord blood pH-7.20, Apgar score or the.need for intubation , presence of neonatal

paralysis and the side of the fracture.3. Pregnancy and delivery: high or low risk preg-

Žnancy course normal or significant complica-tions, such as hypertension, major maternalinfection and premature rupture of the mem-

. Ž .branes , gestational age and reason s forŽadmission to the delivery room rupture of

the membranes, contractions, bleeding or in-.duction of labor , duration and progress ofŽlabor by stages duration of second stage de-

fined as the time from full dilatation to deliv-.ery , the need for oxytocic augmentation of

uterine contractions, use and type of analge-sia or anesthesia, presence of meconium, per-formance of episiotomy.

Statistical analysis was performed with Pearsonx 2-tests for categorical variables and one-wayanalysis of variance for continuous variables, us-

w xing the BMDP statistical software 15 . Variableswith P-values F0.05 were considered significant.

( )B. Kaplan et al. r International Journal of Gynecology & Obstetrics 63 1998 15]20 17

Logistic regression analysis was performed onvariables proven to be statistically significant fordiscriminating between the study and controlgroups.

3. Results

During the period under study, there were27 386 singleton-term deliveries in our center. The403 cases of neonatal clavicular fracture repre-sented 1.65% of all vaginal deliveries and 0.2%Ž .7r3391 of all cesarean deliveries. Erb’srbrachialpalsy was noted in 4% of the clavicular fracture

Ž .group 16r403 ; all recovered within 1 year. Ma-Žternal age average 30.7"5.5 years; range 20]46

. Ž .years and parity average 2.1"1.8 deliverieswere similar in both groups.

Four of the analyzed variables were demon-strated to be significantly different for the twogroups: maternal height; the duration of secondstage of labor; peripartum sonographic fetal

Žweight estimation; and neonatal birth length Ta-.ble 1 . Nevertheless, all values were still within

the normal range.The first three variables, being documented

antepartum data and having a significant relation-ship to the occurrence of neonatal clavicular frac-ture, were further analyzed in a stepwise logisticregression analysis. All three were significant atthe 0.01 level. On this basis, we formulated acombined index, which we believed would be use-ful for anterintra-partum identification of indi-vidual parturients predisposed to deliver an infantwith a fractured clavicle, as follows:

IndexseŽ5.9y0.082 Hq1.765Wq1.22T .

Ž Ž5.9y0.082 Hq1.765Wq1.22T . .r 1qe

Žwhere the index is the probability between 0 and. Ž .1 of a fractured clavicle, H is the height cm of

the mother, W is the peripartum sonographicŽ .fetal weight estimation in kg and T is the base

Ž .10 logarithm of the duration min of the secondstage of labor.

Ideally, this formula could serve as a clinicaldiagnostic tool. Mothers whose index values ex-ceed a predetermined threshold would be advisednot to proceed with a vaginal delivery. For giventhreshold values, clinicians could also use thedata to estimate the proportion of mothers amongthose whose infants suffered a fracture who would

Ž .be referred for a cesarean section sensitivity andthe proportion of mothers among those whoseinfants did not suffer a fracture who would beadvised to continue with a vaginal deliveryŽ .specificity . Table 2 lists the estimated sensitivityand specificity for a variety of thresholds.

Unfortunately, the clinical usefulness of ourformula is undermined by the high false-positiveand false-negative rates for the different thresh-

Ž .old values Table 3 . The false-positive rate is theproportion of mothers recommended for cesareansection whose infants would not have had a clav-

Žicular fracture i.e. mothers incorrectly classified.as high risk , and the false-negative rate is the

proportion of mothers advised to have a vaginaldelivery whose infants would be born with a clav-icular fracture. Our calculation assumes that the

Žoverall fracture rate is 2% of all births average. w xof the range of prevalences in the literature 3 .

Table 1Peripartum characteristics of the clavicular fracture and control groups

Ž . Ž .Variable SignificanceStudy group ns87 Control group ns87of differenceMean"S.D. Range Mean"S.D. Range

Ž .Maternal height cm 162.4"6.6 140]175 164.5"5.5 150]175 0.025Ž .Second stage duration min 32.8"34.6 3]160 21.9"26.4 5]180 0.025a Ž .Estimated fetal weight g 3620"394 2500]4575 3311"49 2210]4750 0.0001

Ž .Neonatal length cm 50.5"1.7 45.0]54.5 49.3"1.9 45]54 0.0001

a Ž .Peripartum 3 days before to the delivery day sonographic fetal weight estimation.

( )B. Kaplan et al. r International Journal of Gynecology & Obstetrics 63 1998 15]2018

Table 2Estimated sensitivity and specificity for different calculatedthresholds of the proposed neonatal clavicular fracture index

Threshold Sensitivity Specificity

0.3 0.931 0.2690.4 0.816 0.5380.5 0.724 0.6540.6 0.609 0.7560.7 0.402 0.8850.8 0.195 0.949

The false-positive rate may be high either be-cause fractured clavicle is not a common outcomeor because the logistic regression equation doesnot sharply discriminate between mothers whoseinfants will have a fracture and mothers whoseinfants will not. As the great majority of births donot result in a clavicular fracture regardless of thethreshold chosen, most of the mothers with indexvalues exceeding the threshold will come from thelarge no-fracture group. For example, for every 10mothers whose neonates have a fracture, thereare 500 whose infants do not. Thus, even if thesensitivity and specificity are both 0.7, sevenmothers from the first group will exceed thethreshold vs. 150 from the second. The low-falsenegative rate also reflects the low prevalence ofclavicular fractures. That is, if no attempt wasmade to classify mothers on the basis of our data,the false-negative rate would be 0.02 } the ac-cepted overall rate of clavicular fracture. Using athreshold of F0.6 would screen out at least halfthe mothers whose infants would have a fracture

Table 3Estimated false-positive and false-negative rates for differentcalculated thresholds of the proposed neonatal clavicularfracture index

Threshold False-positive rate False-negativerate

0.3 0.97 0.00520.4 0.97 0.00690.5 0.96 0.00850.6 0.96 0.01040.7 0.93 0.01360.8 0.93 0.0170

} as the false-negative rate is approx. F0.01 }but at the expense of performing unnecessarycesarean sections for many mothers } as thefalse-positive rate is )0.95.

4. Discussion

Newborn infants with clavicular fracture rarelyw xhave symptoms 16 and most have no long-term

w xproblem 17 . Nevertheless, this complication isimportant because of the concern it raises inparents and the occasional associated neurologic

w xtrauma 8,18 .In our study, the neonatal clavicular fracture

rate was 1.65%, about mid-range of the reportedw xfigures in the literature 16,19,20 . Although we

found that several variables } duration of sec-ond stage of labor, peripartum sonographic fetalweight estimation and neonatal birth length }were positively correlated with neonatal clavicularfracture, and maternal height was inversely corre-lated with neonatal clavicular fracture, all thesevalues were still within normal range. In addition,statistical analysis proved that despite their sig-nificance, the application of these criteria for theprenatal identification of mothers predisposed toa delivery complicated by clavicular fracture isimpractical owing to the high false-positive rate,which indeed, approached the prevalence rate ofclavicular fracture in the general population.

How can a relationship be statistically signifi-cant but the diagnostic value so weak? Statisticalsignificance depends on two factors: strength ofthe relationship between the outcome and theexplanatory variables, and sample size; diagnosticability depends only on the strength of the rela-tionship. The relationship in this study was notsufficiently strong to yield a sharp discriminationbetween the two groups of mothers, as shown bythe considerable overlap in the predicted valuesfrom the logistic regression equation. However,coupled with sample size, it was strong enough toenable us to conclude beyond a reasonable doubtŽ .i.e. with a low P-value that the explanatoryvariables are related to the outcome.

Thus, our results bridge the earlier studiesshowing associations between clavicular fracture

( )B. Kaplan et al. r International Journal of Gynecology & Obstetrics 63 1998 15]20 19

w xin the newborn and different risk factors 4]13and the newer ones which failed to pinpoint any

w xrisk factors 1]3,14 and which emphasized theelusive nature of this obstetrical complication. Asour statistical conclusion that neonatal clavicularfracture cannot be predicted antenatally is inde-pendent of the sample size, we must accept thatthe fractured clavicle of the newborn is mostprobably unpredictable and unavoidable.

Additional findings in our study were theabsence of sex predominance among affectedneonates and the elimination of traumatic deliv-ery, Apgar score, need for postpartum resuscita-tion or for instrumental delivery as importantvariables. Our data also did not support the re-ported relationship between clavicular fracture

w xand shoulder dystocia 5 . Finally, despite therelatively high proportion of diabetic mothers inboth groups } because our department serves asa referral center for diabetic pregnancies } wedetected no association between gestational dia-betes and macrosomia, which may lead to shoulder

w xdystocia 21 , and clavicular fracture. The rate ofneurologic sequelae such as Erb’srbrachial palsyin the neonates with fractured clavicles was only4%, less than the reported 7]13%, and all theaffected newborns were found to be healthy at1-year follow-up. No other neonatal or maternal

Žproblems meconium, episiotomy, and perinatal.morbidity were noted. Therefore it seems that

clavicular fracture is a transient event that nor-mally leaves no sequelae.

Finally, our findings preclude the human factorin neonatal clavicular fracture, not mentioned inearlier works, as all our deliveries were managedby expert obstetricians.

In conclusion, the majority of clavicular frac-tures in our series were noted in normal new-borns following normal labor and delivery. Thethree antenatal risk factors identified statisticallydo not offer a practical method for clinical prena-tal prediction because of high false-positive rate,due to the low prevalence of this complication inthe obstetrical population. Yet, it seems justifiedto determine the two statistically significant ante-

Žnatal variables i.e. peripartum sonographic fetalweight estimation and the duration of the second

.stage of delivery in women of short stature in thecontext of a trial to prevent clavicular fracture ofthe newborn. This work offers statistical evidenceof the ‘stealthy’ nature of fractured clavicle in theneonate, as proposed in several recent studies.

References

w x1 Roberts SW, Hernandez C, Mabbery MC, Adams MD,Leveno KJ, Wendel GD, Jr. Obstetric clavicular frac-ture: the enigma of normal birth. Obstet Gynecol1995;86:978]981.

w x2 Al-Qattan MM, Clarke HM, Curtis CG. The prognosticvalue of concurrent clavicular fractures in newbornswith obstetric brachial plexus palsy. J Hand Surg1994;6:729]730.

w x3 Ohel G, Haddad S, Fischer O, Levit A. Clavicular frac-ture of the neonate: can it be predicted before birth?Am J Perinatol 1993;6:441]443.

w x4 Walle T, Hartikainen-Sorri AL. Obstetric shoulder in-jury. Associated risk factors, prediction and prognosis.Acta Obstet Gynecol Scand 1993;72:450]454.

w x5 Baskett TF, Allen AC. Perinatal implications of shoulderdystocia. Obstet Gynecol 1995;86:14]17.

w x6 Sandmire HF, O’Halloin TJ. Shoulder dystocia; its inci-dence and associated risk factors. Int J Gynaecol Obstet1988;26:65]73.

w x7 Wikstrom I, Axelsson O, Bergstrom R, Meirik O. Trau-matic injury in large-for-date infants. Acta Obstet Gy-necol Scand 1988;67:359]364.

w x8 Levine MG, Holroyde J, Woods JR, Siddiqi TA, ScottM, Myodovnik M. Birth trauma: incidence and predis-posing factors. Obstet Gynecol 1984;63:792]795.

w x9 O’Leari JA, Leonetti HB. Shoulder dystocia: preventionand treatment. Am J Obstet Gynecol 1990;162:5]9.

w x10 Balanta A, Olzai MG, Porcu A et al. Clavicular frac-tures in the newborn infant. Pediatr Med Chir 1984;6:125]129.

w x11 McFarland LV, Raskin M, Daling JR, Benedetti TJ. ErbDuchenne’s palsy: a consequence of fetal macrosomiaand method of delivery. Obstet Gynecol 1986;68:784]788.

w x12 Gordon M, Rich H, Deutschberger J, Green M. Theimmediate and long-term outcome of obstetric birthtrauma: brachial plexus paralysis. Am J Obstet Gynecol1973;117:51]56.

w x13 Cohen AW, Otto SR. Obstetric clavicular fractures: athree-year analysis. J Reprod Med 1980;25:119]122.

w x14 Chez RA, Carlan S, Greenberg SL, Spellacy WN. Frac-tured clavicle is an unavoidable event. Am J ObstetGynecol 1994;171:797]798.

w x15 Dixon WJ, editor. BMDP Statistical software. Universityof California Press, 1990.

w x16 Joseph PR, Rosenfeld W. Clavicular fractures in neo-nates. Am J Dis Child 1990;144:165]167.

( )B. Kaplan et al. r International Journal of Gynecology & Obstetrics 63 1998 15]2020

w x17 Gilbert WM, Tchabo JG. Fractured clavicle in new-borns. Int Surg 1988;73:123]125.

w x18 Oppenheim WL, Davis A, Growdon WA, Dorey FJ,Davlin LB. Clavicle fractures in the newborn. Clin Or-thop 1990;250:176]180.

w x19 Turnpenny PD, Nimmo A. Fractured clavicle of thenewborn in a population with a high prevalence of grand

multiparity: analysis of 78 consecutive cases. Br J ObstetGynaecol 1993;100:338]341.

w x20 Madsen ET. Fractures of the extremities in the new-born. Acta Obstet Gynecol Scand 1955;34:41]74.

w x21 Hod M, Rabinerson D, Peled Y. Gestational diabetesmellitus: is it a clinical entity? Diabet Rev 1995;3:602]612.