Screening is frequently used for developmental dysplasia of the
hip (DDH) as there is a preclinical period when diagnosis is possible.
An appropriate therapeutic intervention during this period1 can change the natural history of the
disorder positively.
In the early 1980s, Graf8,9 introduced the diagnosis of DDH
diagnosis by ultrasonography. It avoids radiology, an invasive procedure that is not sensitive in the diagnosis of DDH in the neonate.
In addition, ultrasonography reveals characteristics undetectable with
the Ortolani-Barlow maneuver or radiography.
Most ultrasound screening studies have been population studies carried
out on samples selected on the basis of risk factors and/or clinical
examinations.10 The rare studies on unselected populations have involved insufficient numbers of cases to assess the
validity and accuracy of sonographic screening
definitively.13 The objective of our study was to
evaluate an organizational model for neonatal population screening.
METHODS 
All subjects born between December 1989 and November 30, 1990 at
six hospitals of the Lombardy region where routine ultrasonography DDH
screening was available were considered for inclusion in the study.
Excluded were those with multiple malformations or syndromes, those
requiring intensive therapy, and those transferred to other hospitals.
Two doctors from each hospital (expert doctors) took preliminary 5-day
training course to improve their performance of the
Ortolani17-Barlow18 maneuver. All other doctors
participated in a DDH training seminar with practical exercises using a
dummy. For each baby, the following data were obtained by interview: gender, twin birth, parity, oligohydramnios, breech position at the
third trimester of pregnancy, breech presentation, family history of
DDH, and associated malformations (twisted foot, talipes calcaneovalgus
due to position or structural defect, torticollis).
Information about DDH and the routine diagnostic program
(Ortolani-Barlow maneuver and ultrasonography) were recorded on the recovery card.
During the first 4 or 5 days of life, each baby underwent the
Ortolani-Barlow maneuver two or three times in the hospital. This was
almost always done by different doctors, each of whom was blinded with
respect to the results obtained by the other(s). The doctor who did the
last Ortolani-Barlow maneuver recorded the result on the clinical
record form. The maneuver was repeated by an expert doctor when
positive results were obtained at one or more maneuvers.
All subjects underwent sonography. An ultrasound screening examination
using linear 5 and 7.5 Mhz probes was performed on all the babies
within the first month of life. The examination was done by a doctor
blinded to both the Ortolani-Barlow maneuver results and to any risk
factors present. The doctors involved in DDH ultrasound examination had
a mean of 27 months experience with it before the beginning of this
study. The results were recorded according to Graf's system (Fig
1). Angles 
and 
were measured only when an
anomaly was found (type IIa or worse). Sonography was repeated monthly
on all babies with one or two type IIa hip(s) at the initial screening
until their condition was found normal or appropriate therapy was
decided.
Fig. 1.
Angles according to Graf classification.
[View Larger Version of this Image (32K GIF file)]
The relation between clinical and ultrasonographic results was
analyzed. Clinical agreement was analyzed by using the 
statistic; the accuracy (sensitivity, specificity, and predictive value) and the
relative risk for the subjects with a positive Ortolani-Barlow maneuver
were calculated. Data were stored and analyzed using a specific DB3
plus computer program.
RESULTS
Screened Population
A total of 4648 neonates, 2312 female (49.7%) and 2336 male
(50.3%), were recruited at the six hospitals participating in the
study. The mean gestational age was 40 weeks (range 27-42), the mean
neonatal weight was 3260 grams (range 730-4850), and the mean maternal
age was 29 years (range 15-53). Significant differences were not
observed between the populations recruited from the different
hospitals.
Risk Factors
The prevalence of risk factors is reported in Table
1. Analysis of the risk-factor distribution at
individual hospitals showed similar percentages at all six.
|
Table 1.
Prevalence of Risk Factors in Recruited Population
[View Table]
|
Ortolani-Barlow Maneuver Results
The Ortolani-Barlow maneuver was done for the same baby by one or
more pediatricians. In 134 cases (2.9%), the maneuver was carried out
by one pediatrician; in 1762 cases (37.9%) by two; in 2706 cases
(58.2%) by three; and in 46 cases (1.0%) by four. A positive test
result was reported by one or more pediatricians in 233 subjects
(5.1%). The Ortolani-Barlow maneuver gave a positive result in 128 (19.1%) of the 671 newborns in hospital four and in 105 (2.6%) of
3977 newborns in the other hospitals. When the positive results were
expressed as a percentage of the total maneuvers carried out, values
between 0.4 and 1.2% were obtained in the different hospitals
except
in hospital four (4.9%). Thus, hospital four was excluded from the
subsequent analyses of the Ortolani-Barlow maneuver.
In the subjects for whom the Ortolani-Barlow maneuver gave a negative
result, clicks were identified in 2.2 to 13.4% of the maneuvers
performed at the various hospitals.
Most of the positive maneuvers were not confirmed by the other
pediatricians who carried out the test on the same baby. Of a total of
277 positive maneuvers, 227 (81.9%) were detected by one pediatrician
alone, and in only 17 cases (6.1%) was the positivity confirmed by all
the pediatricians.
Sonography Screening Results
In 3509 patients (75.4%) the ultrasonography was performed within
the first week of life, whereas in 1139 subjects (Lecco hospital) it
was done at 22 days of life as a mean. The distribution of subjects by
sonographic results is reported in Table 2. The subjects
with type I bilateral hip constituted 50.1% of the population studied;
type IIa, 44.8%; type IIc or IId, 4.5%; and type III or worse, 0.6%.
|
Table 2.
Distribution of Subjects by Sonographic Results
[View Table]
|
The distribution of the sonographic results varied considerably in the
different hospitals; the frequency of a normal finding (type I) varied
from 37.3 to 72.1%; that of type IIa, from 23.6 to 57.6%; that of
type IIc-IId, from 0.8 to 7.0%; and that of type III-IV, from 0.0 to
1.1%.
Risk Factors and Sonographic Screening Results
The ratio between the presence of risk factors and the sonographic
results is analyzed in Table 3. The relative risk (RR) is reported for subjects presenting with risk factors, calculated for
three classes of sonographically demonstrated abnormalities: IIa, IIc
to IId, and III to IV.
|
Table 3.
Relative Risk (RR) of the Different Risk Conditions
[View Table]
|
Comparison Between the Two Tests Used
Clinical agreement was evaluated considering the two test results
as independent variables. The overall agreement of the tests in
identifying worse than type IIa hip was 0.96, positive agreement 0.12, and the -statistic 0.11.
The accuracy of the Ortolani-Barlow maneuver was then calculated, with
the sonographic result the gold standard. The sensitivity of the test
in identifying subjects with type IIc-IId hip or worse was 0.07, the
specificity 0.99, and the positive predictive value 0.35.
The relation between results of the Ortolani-Barlow maneuver and
ultrasonography was also analyzed. The former result was considered a
simple risk factor and the RR calculated (type IIa is deemed a degree
of normality in this analysis). Subjects with a positive
Ortolani-Barlow maneuver had a relative risk of 7.2 (confidence
interval [CI] 4.9-9.6) for type IIc-IId hip and 27.1 (CI 18.5-38.7)
for type III-IV hip when compared to subjects with a negative maneuver.
In subjects with clicks at clinical examination the relative risk was
2.1 (CI 1.4-2.5) for type IIc-IId hip and 1.8 (CI .5-6,2) for type
III-IV hip.
Even when all the pediatricians agreed on a positive or negative
clinical examination, discrepancies arose between clinical and
sonographic results. In 15 cases of Ortolani-Barlow maneuvers were
positive for all pediatricians, only in 6 did the sonographic examination reveal a pathologic picture (type III or IV); in 6 others
the sonography showed an intermediate type (IIc or IId), and in 3 cases
it did not detect a pathologic condition. In 7742 cases of agreement
between all the pediatricians on a negative clinical examination, type
III or IV hip was found by ultrasonography in 22 subjects and type
IIc-IId in 206. A review of 22 false-negatives has demonstrated that
the presence of asymmetric limbs or reduced abduction of the hip was
not given due consideration during the examination. These conditions
may occur in subjects with irreducible hip dislocations.
Follow-up of Subjects With Type IIa Hip
The screening examinations showed 2206 subjects (44.8% of the
population) with a type IIa hip (bilateral or unilateral). A follow-up
study was possible in 1927 (87.4%) subjects (range 73.3 to 95.8% for
the individual hospitals). The mean age of the subjects at the first
follow-up examination was 69 days.
Considering the number of hips (instead of subjects), 83.1% of type
IIa hips at screening were normal at follow-up and 16% were confirmed
as type IIa, whereas in 0.9% the sonographic result worsened.
Follow-up of the subjects with monolateral type IIa hip demonstrated a
worsening of the controlateral hip diagnosed at screening as type I in
3.4% of the cases.
DISCUSSION
Although the importance of DDH screening is widely
recognized,1,19 there is no definite agreement on the
organizational model or the diagnostic screening tests that should be
used. No population studies are yet available that meet the minimum
requirements for correct execution and interpretation of the diagnostic
(clinical and sonographic) tests and allow evaluation of clinical
screening results done in different hospitals. Our study, performed at
six hospitals in which clinical and sonographic DDH screening
procedures were already under way, represents a concrete example of
screening results carried out at hospitals typical of the Italian
national health system.
We also studied some previously suggested risk
factors6,12,20 and found an association between type III or
IV hip and female gender (RR = 2.8), family history of DDH
(RR = 3.4), and breech presentation (RR = 2.8). A weaker
association with type IIc and IId hip was also detected.
Analysis of the Ortolani-Barlow maneuver results was complex because,
according to the design study, the maneuvers were carried out on the
same subject by several pediatricians. Excluding the hospital in which
the frequency of positive maneuver was particularly high, the ratio of
subjects with positive Ortolani-Barlow tests was 15/1000, a figure
similar to that reported by other authors.7,21,22
The reproducibility of the Ortolani-Barlow maneuver was the main
problem to emerge from our study. Wide variations in the frequency of
detection of positive maneuvers and clicks were also observed among the
different hospitals; this could be explained as a different prevalence
of the disorder, but in view of the homogeneous recruited population,
the variations were probably attributable to incorrect execution of the
clinical examination.
The inconsistent results obtained by the pediatricians who made the
clinical investigations show clearly that the Ortolani-Barlow maneuver,
as carried out by the six hospitals, cannot be considered a reliable
screening test. We believed that clinical training of two doctors from
each hospital in a specialized center and clinical practice with a
simulator for the other doctors would make the test sufficiently
reliable, but this was not the case. In our opinion the low
reproducibility of the clinical investigation was due not to
unreliability of the maneuver itself but to insufficient clinical
training of the investigators. This was confirmed by the admission of
most of the physicians who attended the specialized center that they
were not sufficiently expert to carry out an accurate clinical
examination of the hip. Other studies have shown that highly skilled
personnel are needed to obtain good results from clinical screening,
and this is possible when only two or three physicians of the staff
perform the clinical test.23 Good results can be obtained
also by nonmedical investigators, provided they are thoroughly trained
to carry out the clinical test.24
The reliability and interobserver variability between ultrasound
examination at birth (very unreliable) and at 4 weeks (more reliable)
vary considerably. In our study ultrasonography was performed in 75.4%
of cases within the first week of life.
According to the results of the neonatal ultrasound screening
examination, only 50.1% of the studied population could be described as normal; of the subjects examined, 44.8% had a type IIa hip (unilateral or bilateral), which inevitably requires a repeat sonography and, according to several authors, postural
treatment.16,25 Type IIc or IId hip(s) were found in
43/1000 subjects, and type III or IV in 6/1000 of our population. The
latter figure does not differ significantly from those reported in the
literature.16 Considerable variations were also seen in
the data from different hospitals relative to both sonographic and
clinical examination.
Analysis of the relationship between the results of the Ortolani-Barlow
maneuver and the sonographic examination is difficult due to the lack
of a valid gold standard. If sonography is assumed as a gold standard,
as is done to some extent in our study, this brings with it the risk of
an excessive number of diagnoses. Analysis of DDH prevalence reported
in the literature shows a constant increase each time a new test is
used.2,12,26 However, the fact that, for ethical reasons,
all patients with a positive result undergo treatment independently of
the test used impedes evaluation of the specificity of the test. The
detection of false-negatives was difficult, and it cannot be excluded
that arthrosis of the hip in adulthood is not due to an undetected
abnormality present in neonatal age.27
Bearing in mind these methodologic limits, the results of our study
demonstrate that if sonography is assumed to be the gold standard and
if sonographic results worse than type IIa are considered pathologic,
the sensitivity and positive predictive value of the Ortolani-Barlow
test are extremely low. Positive Ortolani-Barlow maneuver was
associated significantly with a sonography showing a type worse than
IIa; the RR was statistically significant (P < .00001).
The ultrasound detection of a type IIa hip raises screening costs
considerably, in that half of the population must undergo a further
sonographic examination, which, as shown by our follow-up data, very
often shows a normalization of the clinical picture. Screening done at
60 days of life would reduce the number of subjects with type IIa
hip(s) (requiring repeat sonography) by 78 and 64%, respectively, in
the case of unilateral and bilateral involvement. This would result in
a considerable reduction in the screening costs.
An improvement in screening efficiency could be achieved by screening
subjects with risk factors (family history, breech presentation, positive Ortolani-Barlow maneuver) in the neonatal
period.12 All other subjects could be screened at 60 days
of life. About 50% of the subjects with dislocation (type III or IV
hip) and 34% of those with type IIc-IId hip could be detected by
screening 15% of the population in the neonatal period, and therapy
could be anticipated. Obviously a strategy that delays screening to a
later age than the neonatal period could in theory increase the number
of babies lost to screening. Based on our experience, such losses can
be minimized by a careful information program addressed to the
families.
To lower DDH screening costs, some authors10 have
proposed that only subjects with risk factors or positive clinical
examination should undergo sonographic screening. To evaluate this
proposal, we applied these restrictions on the examined population. Had we screened all females and the males with a risk factor (family history, breech or transverse presentation, positive clinical examination), there would have been a population of 2730 subjects (instead of 4648) and a loss of 7% in diagnosis of subjects with sonography showing type III or IV hip and of 15% for type IIc or IId
hip.
In conclusion, our results demonstrate that some aspects of DDH
screening need further detailed study; the reduced reproducibility of
the diagnostic tests used in screening is the main problem. The study
was conducted in a relatively small geographic area that was ethnically
homogeneous. This makes it improbable that the observed differences are
due to a different prevalence of DDH. The insufficient agreement
between pediatricians who carried out the Ortolani-Barlow maneuver
indicates that this examination, as now performed, cannot be proposed
as a screening test. Specific training given to the doctors
participating in the study was not sufficient to warrant acceptable
reproducibility of the test. It was not possible to evaluate the
reproducibility of the ultrasound examination in this study. The
substantial differences observed among the different hospitals lead us
to hypothesize that important problems of reproducibility exist even in
these cases. A correct DDH screening model for Italian hospitals cannot
therefore be defined at this time.
APPENDIX
The Collaborative Group DDH Project consists of PediatriciansD.
Baronciani, MD, M. Petrone, MD, and R. Zanini, MD (Lecco); C. Di
Pietro, MD and F. Gaboardi, MD (Cernusco); E. Bianchi, MD, G. C. Calligari, MD, and C. Scaravelli, MD (Erba); C. Belloni, MD and F. Paolillo, MD (Lodi); A. Avanzini, MD and S. Santucci, MD (Magenta); P. L. Patriarca, MD and M. Branchi, MD (Sondrio); radiologistsF.
Andiloro, MD and P. Minola, MD (Lecco); A. Nicolini, MD
(Cernusco); C. Passamonti, MD (Lodi); G. Ballerini, MD (Magenta); paediatric orthopedic consultantsG. Atti, MD and C. Vullo, MD (Centre
for Congenital Hip Dislocation, Ferrara); orthopedic consultantA. Bartesaghi, MD (Department of Orthopedics, Lecco); and statistical consultantsP. G. Duca, MD and L. Gagliardi, MD (Institute of Biometry
and Medical Statistics, Milan).
Reprint requests to (D.B.) Divisione Patologia
Neonatale-Ospedale di Lecco, Via Ghislanzoni 22-22053 Lecco, Italy.
ABSTRACT
