S.L. Sood, G.S.
Saiprasad and C.G. Wilson
From the Department of
Pediatrics and Department of Preventive and Social Medicine, Armed
Forces Medical College, Pune 411 040, Maharastra, India.
Correspondence to: Col.
G.S. Saiprasad, Epidemiologist/Professor and Head, Department of
Preventive and Social Medicine, Armed Forces Medical College, Pune 411
040, Maharastra, India.
E-mail:
[email protected]
Manuscript received:
July 8, 2001, Initial review completed: August 30, 2001;
Revision accepted: March
6, 2002.
The present study is an
attempt at evaluating the utility of mid-arm circumference(MAC) to
detect low birth weight in the neonatal period. This cross sectional
study was conducted in 1272 newborns. The mid arm circumference ranged
from 8.2 cm to 9.7 cm (mean 9.2 cm, 95% CI 9.15 cm, 9.19 cm). A
positive correlation existed between MAC and birth weight (r = 0.76,
95% CI:0.73, 0.78; P<0.001). A mid-arm circumference of 8.7 cm
predicts a birth weight of 2580 gm (95% PI 2071g, 3099g) and it
definitely excludes newborns with birth weight <2000 g. Measurement
of MAC is simple and valid method of screening for low birth weight
babies.
Keywords:
Low birth weight, Mid-arm circumference, Neonate, Screening test
Birth weight is one of the most
sensitives and reliable predictors of the health of any community(1).
Detection of low birth weight immediately at birth is of paramount
importance to ensure
infant’s survival(2). Recording of accurate birth weight by primary
health care workers has been a problem in rural areas leading to a
search for an alternate, inexpensive, age independent and non invasive
method to predict neonatal well being(3,4).
Many studies were carried
out to evaluate the mid-arm circumference (MAC) as an indicator of low
birth weight (LBW) and different values of MAC were advocated by various
authors(5-7), but most of these studies did not provide a critical
evaluation of MAC as a screening test. The present study attempts a
detailed scrutiny of the concept of using MAC as a screening test.
Subjects and Methods
The present study was
conducted at Command Hospital (Southern Command), Pune, a tertiary care
hospital affiliated to the Armed Forces Medical College, Pune. Weight
and MAC of 1272 newborns were recorded between 24 and 48 hours of life
using standard techniques. Birth weight was the gold standard against
which MAC was evaluated as surrogate in this screening marker evaluation
study to detect low birth weight (LBW), i.e., < 2500g.
Sensitivity, specificity,
positive and negative predictive values for various values of MAC were
calculated and a receiver operating characteristic (ROC) curve was
constructed to arrive at the optimum MAC cut off level(8). A simple
linear regression model was fitted to predict birth weight from MAC
values and 95% Prediction Intervals (95% PI) constructed to examine the
range of error in prediction on account of sampling variation(9).
Results
Table I shows
the distribution of birth weight and MAC by estimated gestational age (EGA).
The MAC ranged from 8.2 cm to 9.7 cm with a mean of 9.2 cm (95% CI 9.15
cm, 9.19 cm). Statistically highly significant positive linear
correlations were observed between MAC and birth weight (r = 0.76, 95%
CI 0.73, 0.78; P<0.001), between EGA & MAC (r = 0.93, 95% CI
0.92, 0.93; P<0.001), and between EGA & birth weight (r = 0.67,
95% CI 0.62, 0.68; P<0.001).
Table II
shows distribution of the study subjects by different MAC values and
also by birth weight categorized as low birth weight (<2500g) or
normal weight (2500g+). Sensitivity, specificity and positive and
negative predictive values at these MAC values are shown. As cut off
value of MAC is raised the sensitivity to detect low birth weight is
seen to increase and the specificity is seen to decrease. It can also be
seen from Table II that as the MAC value increases, the positive
predictive value decreases and the negative predictive value increases.
The ROC curve was plotted to identify the optimal cut off point of MAC (Fig.
1). The value of MAC as indicated by the ROC curve is 8.7 cm, and is
considered as the optimal cut off point in this study.
Table I-Correlation of Birth Weight, Mid-arm Circumference and Estimated Gestational
Age (EGA) in Newborns
Expected
gestational
age (weeks)
|
Infants
Studied
|
Birth weight (g)
|
Mid-arm Circumference (cm)
|
|
n
|
%
|
Mean (SD)
|
95% CI
|
Mean (SD)
|
95% CI
|
37
|
147
|
11.2
|
2230(327)
|
2177,2283
|
8.5(0.14)
|
8.49, 8.54
|
38
|
208
|
16.4
|
2273 (266)
|
2703,2773
|
8.8(0.10)
|
8.79, 8.82
|
39
|
251
|
19.7
|
2971 (272)
|
2938,3004
|
9.0(0.14)
|
8.99, 9.03
|
40
|
548
|
43.4
|
3135(282)
|
3111,3159
|
9.4(0.14)
|
9.38,9.41
|
41
|
71
|
5.6
|
3109(266)
|
3046,3172
|
9.6(0.13)
|
9.57,9.63
|
42
|
47
|
3.7
|
3324(368)
|
3214,3430
|
9.8(0.11)
|
9.79,9.85
|
Total
|
1272
|
100.0
|
2950(400)
|
2928,2972
|
9.2(0.37)
|
9.15,9.19
|
Table II-Sensitivity, Specificity, Positive and Negative Predictive Values for Different
Cut off Values of Mid-arm Circumference
Mid-arm
Circumference
(cm)
|
Number(%) of newborns
by birth weight(g)
|
Sensitivity
(%)
|
Specificity
(%)
|
Positive
predictive
value (%)
|
Negative
predictive
value (%)
|
|
< 2500
|
2500+
|
|
|
|
|
|
n(%)
|
n(%)
|
|
|
|
|
8.3
|
15 (1.2)
|
3 (0.2)
|
10.2
|
99.7
|
83.3
|
89.5
|
8.4
|
12 (0.9)
|
4 (0.3)
|
18.4
|
99.4
|
79.4
|
90.3
|
8.5
|
55 (4.3)
|
11 (0.9)
|
55.8
|
98.4
|
82.0
|
94.5
|
8.6
|
20 (1.6)
|
8 (0.6)
|
69.4
|
97.7
|
79.7
|
96.1
|
8.7
|
26 (2.0)
|
32 (2.5)
|
87.1
|
94.8
|
68.8
|
98.3
|
8.8
|
7 (0.6)
|
116 (9.1)
|
91.8
|
84.5
|
43.7
|
98.8
|
8.9
|
2 (0.2)
|
75 (5.9)
|
93.2
|
77.9
|
35.5
|
98.9
|
9.0
|
4 (0.3)
|
163 (12.8)
|
95.9
|
63.4
|
25.5
|
99.2
|
9.1
|
3 (0.2)
|
33 (2.6)
|
98.0
|
60.4
|
24.4
|
99.6
|
9.2 & above
|
3 (0.2)
|
680 (53.5)
|
100.0
|
0.0
|
11.6
|
100.0
|
A simple linear
regression was fitted with birth weight as independent variable and MAC
as dependent variable. The regression coefficient ‘b’ (slope) was
0.8218 (95% CI of b = 0.7828, 0.8607; P < 0.001). A MAC value of 8.6
cm was seen to predict a birth weight of 2500 g (95% PI 1990 g, 3010 g).
MAC of 8.7 cm was seen to predict birth weight as 2580 g (95% PI 2071 g,
3099 g).
Discussion
To detect a birth weight of < 2500 g,
various studies have proposed different values of MAC as a cut off
point. These, as studied in different settings, were 8.6 cm (7,12), less
than 9.0 cm(11) and 9.0 cm(10). At any cut off level some false
negatives as well as false positives are bound to occur. However, the
cut off point
Fig.
1. Receiver operating characteristics curve to choose the optimal
cut-off point of mid-arm circumference. The points represent the
sensitivity and 1-specificity for each vale of MAC. The MAC point which
is nearest to upper and outer corner in the chart i.e., 8.7 cm is the
optimal cut off point.
that gives the optimal
level of false negatives and false positives can be easily found out
from the ROC curve. This point happens to be the MAC of 8.7 cm with a
sensitivity of 87.1% and a specificity of 98.5%. Further lowering of the
cut off point to 8.6 cm results in missing out of 26 LBW cases. Since
the condition being screened is a life threatening one it may not be
desirable to miss as many low birth weight cases. Raising the cut off
point to 8.8 cm increases the sensitivity to 97.9% and also lowers the
specificity to 60.4% adding 116 false positives requiring re-screening.
The rationale of using
MAC to predict birth weight is further evaluated by fitting a regression
line based on the sample data. A MAC value of 8.6 cm is seen to predict
a birth weight of 2500 g (95% PI 1990 g, 3010 g). It also means that
just by sampling variation a value of birth weight could be as low as
less than 2000 g. MAC value of 8.7 cm definitely excludes newborns with
birth weight < 2000g. A birth weight of 2000 g is considered as a
more appropriate criteria for defining low birth weight among Indian
population(13). Based on the present study a cut off point of MAC of 8.7
cm appears to be appropriate as a general guideline.
Contributors:
SLS conceptualized research question for the study and collected the
data. GSS provided statistical and methodological advice and carried out
statistical analysis. CGW helped in conceptualization and provided
guidance and critical analysis. GSS shall act as guarantor for the
study.
Funding:
None.
Competing interests:
None stated.
Key
Messages |
• Measurement of Mid
Arm Circumference (MAC) is simple and valid method of screening for low
birth weight in the newborns.
• MAC value of 8.7 cm predicts a birth
weight of 2580 g (95% PI 2071 g, 3099 g) and it definitely excludes
newborns with birth weight < 2000 g.
|