Fractures
WHAT’S KNOWN ON THIS SUBJECT: Patient assessment by AUTHORS: Nisanthini Ravichandiran,a Suzanne Schuh,
physicians of children who are at risk for abuse is suboptimal, MD,a Marta Bejuk, MD,a Nesrin Al-Harthy, MD,a
and, therefore, abusive fractures are at risk for escaping Michelle Shouldice, MD,b Hosanna Au, MD,b and
detection or delayed recognition. It is unknown, however, how Kathy Boutis, MD, MSca
often this occurs. Divisions of aPediatric Emergency Medicine and bPediatric
Medicine and Suspected Child Abuse and Neglect, Hospital for
Sick Children, University of Toronto, Toronto, Ontario, Canada
WHAT THIS STUDY ADDS: Approximately 20% of abusive
fractures were missed at initial physician visits. Boys who KEY WORDS
pediatrics, child abuse, bone fractures, diagnosis
present to a nonpediatric ED with an extremity fracture seem to
be at highest risk of the abusive etiology of the fracture escaping ABBREVIATIONS
ED— emergency department
of detection by a physician. SCAN—Suspected Child Abuse and Neglect
HSC—Hospital for Sick Children
OR— odds ratio
CI— confidence interval
www.pediatrics.org/cgi/doi/10.1542/peds.2008-3794
abstract doi:10.1542/peds.2008-3794
OBJECTIVES: Because physicians may have difficulty distinguishing Accepted for publication Jul 28, 2009
accidental fractures from those that are caused by abuse, abusive Address correspondence to Kathy Boutis, MD, MSc, 555
fractures may be at risk for delayed recognition; therefore, the primary University Ave, Toronto, ON, M5G 1X8, Canada. E-mail:
boutis@pol.net
objective of this study was to determine how frequently abusive frac-
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
tures were missed by physicians during previous examinations. A sec-
Copyright © 2009 by the American Academy of Pediatrics
ondary objective was to determine clinical predictors that are associ-
ated with unrecognized abuse. FINANCIAL DISCLOSURE: The authors have indicated they have
no financial relationships relevant to this article to disclose.
METHODS: Children who were younger than 3 years and presented to
a large academic children’s hospital from January 1993 to December
2007 and received a diagnosis of abusive fractures by a multidisci-
plinary child protective team were included in this retrospective re-
view. The main outcome measures included the proportion of children
who had abusive fractures and had at least 1 previous physician visit
with diagnosis of abuse not identified and predictors that were inde-
pendently associated with missed abuse.
RESULTS: Of 258 patients with abusive fractures, 54 (20.9%) had at
least 1 previous physician visit at which abuse was missed. The median
time to correct diagnosis from the first visit was 8 days (minimum: 1;
maximum: 160). Independent predictors of missed abuse were male
gender, extremity versus axially located fracture, and presentation to a
primary care setting versus pediatric emergency department or to a
general versus pediatric emergency department.
CONCLUSIONS: One fifth of children with abuse-related fractures are
missed during the initial medical visit. In particular, boys who present
to a primary care or a general emergency department setting with an
extremity fracture are at a particularly high risk for delayed diagnosis.
Pediatrics 2010;125:60–66
60 RAVICHANDIRAN et al
Downloaded from by guest on August 24, 2017
ARTICLES
Although fractures are a common pre- METHODS A case was considered “recognized”
senting finding in child abuse,1,2 clini- when a referral to local child protec-
Patient Population
cians may have difficulty differentiat- tion authorities was made the first
ing abuse-related fractures from Children who were younger than 3 time the child presented to a physician
those that are caused by accidental years,2,9–11 had abusive fractures that with the index fracture(s). This is in
trauma2–4; however, this distinction is occurred from January 1993 to Decem-
contrast to “missed” when the child
ber 2007, and were referred to a
crucial because of its impact on con- had at least 1 physician encounter for
multidisciplinary hospital-based Sus-
sequences for the child.5,6 Whereas the index fracture(s) before the visit
pected Child Abuse and Neglect (SCAN)
accidental injuries carry only their when the abuse was confirmed. In all
team at the Toronto Hospital for Sick
inherent risks, repeat injury occurs missed cases, the signs and symptoms
Children (HSC) were included. HSC
in 35% of all abuse cases, and 5% to compatible with a fracture and/or a
SCAN consists of specialty pediatri-
10% of patients will die if there is no radiograph diagnosis were present at
cians, psychologists, social workers,
intervention.7 and nurse practitioners. Members of the initial visit, but the possibility of
Despite the serious risks associated HSC SCAN team are the only child abuse was not raised. Thereafter, ⱖ1
with delayed recognition of abusive abuse specialists in the Greater To- of the following occurred: (1) the child
fractures, patient assessment for this ronto Area and are involved in the as- improved clinically but experienced re-
diagnosis is often suboptimal.1–4,8 One sessment of most cases of suspected peat trauma and the HSC SCAN team
study found that of 100 children who abuse in that area. The HSC SCAN found the previous fracture(s) abu-
were younger than 3 years and pre- team’s assessment results in a classi- sive; (2) recognition of red flags and
sented to an emergency department fication of these fractures as abusive, referral to the SCAN team at a routine
(ED) with long bone fractures, 31 had indeterminate, or accidental. The study follow-up for the index fracture(s) led
indicators suggestive of abuse but only sample included only cases for which to recognition of abuse; (3) the child’s
1 was referred to child protection ser- the first physician visit was primarily continued symptoms resulted in re-
vices for additional assessment.2 Ban- for an isolated fracture. Cases were peat unscheduled visits and a referral
askiewitz et al3 demonstrated that in excluded when the child’s clinical pre- to the SCAN team with recognition of
infants who were younger than 1 year, sentation was predominantly consis- the index fracture(s) as abuse-related;
the possibility of abuse was underesti- tent with some other type of trauma, (4) the index radiographs initially read
mated by ED clinicians in ⬃28% of medical records were inaccessible, as normal by the primary treating phy-
cases when compared with a retro- only metaphyseal corner chip frac- sician were found by a radiologist to
spective diagnosis by a child protec- tures (usually asymptomatic) were have a fracture that required a repeat
present, or the cause of the fracture visit, when the suspicion for abuse was
tion team pediatrician. Moreover, re-
was indeterminate or accidental. raised; (5) the perpetrator later con-
search conducted in a pediatric ED
demonstrated that 42% of charts re- fessed or a witness came forward;
Definitions
viewed did not have adequate docu- and/or (6) abuse was suspected in a
mentation to explain the cause of the Fractures were determined to be abu- sibling and review of the patient’s frac-
fractures, and inflicted injuries were sive when at least 1 of the following tures yielded abuse as the cause. The
therefore not adequately ruled out.1 criteria was met2,6,12: (1) confession of determination of missed versus recog-
intentional injury by an adult caregiver; nized cases was made independent of
This evidence suggests that abusive (2) inconsistent/inadequate history
fractures are likely at risk for escaping the knowledge of potential predictors.
provided; (3) inappropriate delay in
detection or delayed recognition; how- seeking medical care; (4) associated Because specific income of the fam-
ever, the frequency with which this oc- inadequately explained injuries; (5) ily was not available, this was esti-
curs remains unknown. The primary in the absence of bone disease, pres- mated on the basis of median income
objective of this study was to deter- ence of fractures uncommon for of families in a given postal code.15
mine the proportion of abuse-related accidental injury and frequently re- On the basis of the 2006 Ontario me-
fractures that were missed at previ- ported in abusive injury (eg, meta- dian household income of $60 455,
ous physician encounters. The clinical physeal limb fractures, posterior rib median income was then additionally
factors that may have contributed to fractures not caused by birth trau- classified as low (⬍$45 341.25), mid-
the reasons for the diagnostic delay ma)6,13,14; and (6) witness to abuse dle ($45 341.25—$90 682.50), or high
were also examined. came forward. (⬎$90 682.50).16 Income classifica-
62 RAVICHANDIRAN et al
Downloaded from by guest on August 24, 2017
ARTICLES
TABLE 1 Characteristics of Missed and Recognized Abuse Cases 1 (minimum: 1; maximum: 3). Of the
Characteristic Recognized Missed P for Univariate children who re-presented for medical
Cases Cases Analysis of care after the abusive cause of the
(n ⫽ 204) (n ⫽ 54) Independent
Variables fracture was missed, 9 (16.7%) pre-
Potential predictors independently associated with sented with new abusive injuries. In 7
missed abuse of these cases, there was a different
Age, mean ⫾ SD, mo 8.28 ⫾ 7.05 9.24 ⫾ 8.31 .3910 fracture; 1 child had serious abdomi-
Male gender, % 44.4 60.8 .0250a
Pediatric ED setting at initial visit, % 89.9 10.1 ⬍.0001a nal injuries, and another had serious
Injury event reported, % 41.5 38.9 .8840 head trauma that resulted in death.
Extremity fracture, % 51.0 75.9 .0010a
Parents living apart, % 26.5 31.5 .4410
Incorrect interpretation of the radio-
Low socioeconomic status, % 27.4 22.4 .6100 graph findings by the physician re-
Additional baseline characteristics (not independently sulted in 18 (33.3%) missed cases
associated with missed abuse)
Nonambulatory, % 71.6 66.7
(Fig 2), 7 of which were skull fractures.
No. of fractures on initial radiograph, median (range) 1 (2) 1.0 (2) In 7 (13.0%), the initial imaging series
Positive skeletal survey, n (%) 82 (40.1) 34 (63.0) was incomplete and the abuse-related
No. of fractures on skeletal survey, median (range) 1 (25) 2 (26)
fracture was therefore not seen. This
Lack of plausible mechanism, % 98.5 94.4
Delay in seeking care, % 29.5 38.9 subgroup returned to an ED because of
Single caregiver, % 26.0 29.8 persistence of symptoms, more exten-
Social concerns, % 43.6 50.9
sive imaging was performed, the frac-
a Statistically significant.
ture was detected, and a referral to
the SCAN team was made. The exact
reasons that the remaining 29 cases
t test for continuous variables. Inde- to iteratively remove noncontributory
pendent variables with P ⱕ .20 and any variables from the model.24 Goodness were missed are not certain because
relevant interaction and confounding of fit of the final model to the data was of a lack of available data; however,
terms were entered into a multivariate tested by using the Hosmer-Lemeshow inadequate screening or accepting im-
logistic regression model using the test. A receiver operating characteris- plausible mechanisms may have con-
forward selection method (Table 2).24 tic curve was plotted to check the pre- tributed to missing these cases. SCAN
Approximately 14 missed cases per dictive ability of the model. Odds of a documentation revealed that these
variable were entered into the model, case being missed for a given variable children had risk factors for abuse: 25
meeting the minimal criteria of 10 were reported with respective 95% (86.2%) of 29 were nonambulatory; in
events per variable to minimize over- confidence intervals (CIs). All analyses 26 (90.0%) of 29, parental report of
fitting of the data.25–27 Wald and Like- were performed by using SPSS 13 for mechanism did not explain injuries;
lihood ratio testing were then used Windows (SAS Institute, Cary, NC). and 14 (48.3%) of 29 had social con-
cerns. Furthermore, review of the ini-
RESULTS tial visit records demonstrated that 13
TABLE 2 Fracture Locations of Recognized This study included 258 eligible pa- (50.0%) of 26 had incomplete docu-
Versus Missed Cases mentation of the preceding events or
Fracture Recognized Missed
tients with abusive fractures (Fig 1).
A comparison of characteristics of possible related risk factors for abuse.
Location Abuse Abuse
Cases Cases missed and recognized cases is de- The univariate analysis demonstrated
(n ⫽ 204) (n ⫽ 54) that 3 variables were found to be sig-
tailed in Tables 1 and 2. Of the 258 pa-
Clavicle, n (%) 8 (3.92) 2 (3.70) nificantly associated with a missed di-
tients, 54 (20.9% [95% CI: 15.8 –26.0])
Humerus, n (%) 32 (15.70) 13 (24.10)
Forearm, n (%) 19 (9.30) 7 (13.00) had at least 1 previous physician visit agnosis of abuse: male gender, initial
Wrist, n (%) 0 (0.00) 1 (1.90) at which abuse was missed. Of the 145 presentation to a nonpediatric ED, and
Digits, n (%) 0 (0.00) 1 (1.90) children with an abusive extremity an extremity fracture (Table 1). The
Femur, n (%) 48 (23.50) 9 (16.70)
Tibia/fibula, n (%) 24 (11.80) 11 (20.40) fracture, 41 (28.3% [95% CI 20.8 –35.8]) probability of missing this diagnosis
Scapula, n (%) 0 (0.00) 2 (3.70) were “missed.” From the initial visit for for each predictor after adjustment
Skull, n (%) 73 (35.80) 15 (27.80) the index fracture(s), the median delay for all significant predictors is summa-
Sternum, n (%) 4 (2.00) 0 (0.00)
of the diagnosis of abuse was 8 days rized in Table 3. No statistically signifi-
Totala 208 61
a Numbers exceed total number of patients because some (minimum: 1; maximum: 160), and the cant interaction terms or confounding
patients had ⬎1 fracture. median number of physician visits was variables were identified in this analy-
TABLE 3 Predictor Variables That Were visit documents were not available for an extremity fracture seem to be at the
Independently Associated With
Missed Abuse
detailed review. In 2 (0.8%) of the 258 highest risk of the abusive etiology of
Predictor OR 95% CI
cases, the initial clinical setting could the fracture escaping of detection by a
Male vs female gender 2.00 1.03–3.80
not be determined. In 16 (6.2%) cases, physician at an initial visit.
Setting it was uncertain whether an injury was In ⬃17% of missed abuse cases, chil-
Primary care office vs 5.20 1.77–15.39 reported. Forty (15.5%) had living sta- dren sustained repeat injuries be-
pediatric ED
tus of parents unavailable. Finally, a tween their initial visit and their even-
General ED vs pediatric ED 7.20 3.00–17.30
Extremity vs axial skeleton 2.30 1.10–4.77 postal code was not recorded for 49 tual diagnosis of abuse; previously
fracture (19.0%) cases. Sensitivity analyses missed fractures that led to serious
with and without imputed data for
abusive injuries were also found by
these missing variables were per-
sis. In the resultant model, the Hosmer- Oral et al.28 The skeletal survey that
formed and did not reveal any signifi-
Lemeshow goodness-of-fit test did not was performed during subsequent vis-
cant differences; therefore, only unim-
reject the null hypothesis of good fit its may have a major impact on the
puted results are presented.
(P ⫽ .718), and the predictive ability of correct diagnosis. In this study, two
the model is good (area under the DISCUSSION thirds of patients had healing frac-
curve: 0.841). Applying this model pre- tures identified on the survey, and this
This study is the first to report the fre-
dicts that if all 3 factors were present, is higher than that reported previous-
quency of delayed recognition of abu-
then the probability that an abusive ly.5,22 This highlights the importance of
sive fractures in children. One fifth of
fracture would be missed is 50%. children with abusive fractures were having a low threshold to consider a
Sixteen charts were missing and un- missed at initial physician visits, which skeletal survey for children who may
available for review, and if we assume is comparable to that reported for be at risk for abuse5,14,22 before dis-
that all were recognized abuse, then other types of abuse12,19; however, we missing the fractures as accidental.
the proportion of missed abuse would do not know how many cases of abu- In the 54 missed cases, approximately
decrease only to 54 (19.1%) of 274. Of sive fractures are never detected. We one third of the fractures were not
the 139 initial visits that occurred out- also found that boys, children who detected on the initial radiographs
side HSC, 3 of 45 of the missed and 15 present to a nonpediatric ED or a pri- by front-line physicians in a country
of 94 of the recognized first physician mary care setting, and/or those with where immediate radiology interpre-
64 RAVICHANDIRAN et al
Downloaded from by guest on August 24, 2017
ARTICLES
tations are not routine practice in the nonambulatory child and when the tial for referral bias, and it may result
ED or office setting. Pediatricians have fracture type includes the femur or hu- in an overestimation of the proportion
limited skills in the recognition of frac- merus in infants who are younger than of cases that are missed at an initial
tures on radiographs.29 This is true 18 months.31,33,34 Indeed, in this study, physician visit; however, child abuse is
particularly of skull fractures,30 and these types of fractures in nonambula- underrecognized,12 and there is also
identification of this type of fracture, tory children were commonly seen in the possibility that we are underesti-
especially in very young infants, may cases for which abuse was missed. In mating the proportion of cases
prompt the physician to assess for addition, many of the missed extremity missed.
other maltreatment risk factors.31 This fractures had associated risk factors
study suggests that front-line physi- for abuse that were not adequately CONCLUSIONS
cians should strongly consider con- screened for at the initial physician Our results suggest that a consider-
sulting a radiologist when the pres- visit; therefore, the possibility of abuse able proportion of abuse-related pedi-
ence of a fracture may lead to should be carefully considered for atric fractures are missed during the
increased suspicion of abuse. children with extremity fractures, and
initial visit. We can make the following
In our study, abuse was more likely to associated risk factors should be
suggestions that may facilitate the di-
be missed when a child presented to a excluded.
agnoses of abusive fractures. A de-
general ED or primary care setting. An abuse-related fracture was almost tailed review of the mechanism and
These results support those by Trokel twice as likely to be missed in a boy screening for other risk factors of
et al,23 who found lower rates of abuse versus a girl. Although the reason for abuse should be included in the initial
in patients who had traumatic brain this is unclear, injuries in general oc- assessment of a young child with frac-
injury or femur fracture and pre- cur more often in boys,35 which may tures. Children who are nonambula-
sented to general hospitals compared bias a clinician in assuming that the tory are at especially high risk, and
with their pediatric counterparts. This cause of a fracture is accidental. consultation with the child protection
could suggest that abuse may be This study has limitations that warrant team in these cases is often appropri-
missed in these settings. Clinicians consideration. This was a retrospec- ate. Clinicians should have a low
who work in these areas may lack ex- tive study with its inherent limitations, threshold to perform a skeletal survey
pertise in the recognition of abuse- such as missing data, and thus absent in potentially vulnerable populations,
related fractures despite the presence data may have biased predictor vari- and a radiologist’s review of any imag-
of indicators for abuse.1,2,32 This re- able results. Although our case classi- ing that may change suspicion for
search supports the need for quality fication was based on current avail- abuse is recommended. Finally, appro-
improvement programs at general able standards for the diagnosis of priate targeted education or practice
hospitals and primary care settings. abuse, there may have been ascertain- guidelines may help in achieving bet-
Children with extremity shaft frac- ment errors. Children with abusive ter outcomes in clinical settings that
tures caused by abuse were also found fractures that were never referred to are susceptible to missing abusive
to be at increased risk for having the SCAN team and were assumed to fractures.
physicians attribute their injuries to be accidental were not included in this
accidental causes. Although extremity review; however, given that ED records
fractures are the most common skele- are often incomplete,1 a retrospective ACKNOWLEDGMENTS
tal injuries that occur in abused chil- assessment by the child protection This research was supported by a
dren,2 radiology literature demon- team of all of the nonreferred cases grant from the Canadian Hospitals In-
strates that these injuries also have would have resulted in only specula- jury Reporting and Prevention Pro-
the lowest specificity for abuse.14 No tive assignments of cause. Finally, al- gram (CHIRPP).
fracture on its own can distinguish an though most cases of abusive frac- We acknowledge the efforts of Dr S.
accidental from a nonaccidental trau- tures are seen by our SCAN team, some Walter and Mr A. O. Odueyungbo for
ma,31 but the likelihood of abuse in- of the less complex cases may not have statistical expertise and critical review
creases when there is a fracture in a been seen. This introduces the poten- of the analysis.
66 RAVICHANDIRAN et al
Downloaded from by guest on August 24, 2017
Delayed Identification of Pediatric Abuse-Related Fractures
Nisanthini Ravichandiran, Suzanne Schuh, Marta Bejuk, Nesrin Al-Harthy, Michelle
Shouldice, Hosanna Au and Kathy Boutis
Pediatrics 2010;125;60; originally published online November 30, 2009;
DOI: 10.1542/peds.2008-3794
Updated Information & including high resolution figures, can be found at:
Services /content/125/1/60.full.html
References This article cites 28 articles, 5 of which can be accessed free
at:
/content/125/1/60.full.html#ref-list-1
Citations This article has been cited by 21 HighWire-hosted articles:
/content/125/1/60.full.html#related-urls
Subspecialty Collections This article, along with others on similar topics, appears in
the following collection(s):
Child Abuse and Neglect
/cgi/collection/child_abuse_neglect_sub
Permissions & Licensing Information about reproducing this article in parts (figures,
tables) or in its entirety can be found online at:
/site/misc/Permissions.xhtml
Reprints Information about ordering reprints can be found online:
/site/misc/reprints.xhtml
The online version of this article, along with updated information and services, is
located on the World Wide Web at:
/content/125/1/60.full.html