added benefit to measure AGD is uncertain. We sought
to investigate the predictive ability of AGD of infertile
menind
isc
r
im
ina
t
ingspe
rmp
roduc
t
ionbycompa
r
ingi
t
to testis volume on an individual basis.
Methods
The methods of collection and cohort assembly have
been previously reported.[
8
] Briefly, after obtaining In-
stitutional Review Board approval from Baylor College
of Medicine (IRB No. H-27133), patients were recruited
from a urology clinic specializing in reproductive and
sexual medicine from August 2010 through October
2011. Men were evaluated for general urology, fertility,
sexual, and testosterone related concerns. Date of visit,
reason for visit, observer, anthropomorphic measure-
ments, and relevant laboratory data was recorded. All
men provided written consent for participation.
Genital measurements
The methods of genital measurement have been previ-
ously described.[
8
] In the supine, frog-legged position
with the legs abducted allowing the soles of the feet to
meet, the distance from the posterior aspect of the
scrotum to the anal verge was measured using a digital
caliper (Neiko USA, Model No. 01407A). As has pre-
v
i
o
u
s
l
yb
e
e
nr
e
p
o
r
t
e
d
,t
h
ecorrelation coefficient was
0.91 for AGD measurements suggesting good agreement
between investigators.[
8
] In all, eight investigators col-
lected data.
Testicular volume was estimated by palpation at the time of
the physical examination by a single investigator (LIL) at
approximately 25 to 27° Celsius. A Prader orchidometer was
used to calibrate size estimates.[
16
]
Statistical analysis
Abnormal semen parameters were defined based on the WHO
5th edition of the manual on semen analyses.[
17
] For sub-
group analyses, men were stratified into short/long AGD
compared to the median for the group (37.6 mm). Men were
also stratified into small/large total testis volume (<average of
16 cm
3
per testis). ANOVA was used to compare continuous
variables. Receiver Operating Characteristic (ROC) Curves
were generated using maximum likelihood estimation to fit a
binomial ROC curve to either continuously distributed data or
ordinal category data. Comparisons were made using the
technique of DeLong et al.[
18
] Multivariable logistic regres-
sion included age, race, body mass index, and date of collec-
tion. All p values were two sided. Analyses were performed
using Stata 10 (StataCorp LP, College Station, Texas).
Results
In all, 473 men with fatherhood data were included in
the analysis with a mean age of 43±13.0. Individual
characteristics are listed in Table
1
. Mean AGD was
40.8 mm for general urology patients, 42.5 mm for
erectile dysfunction patients, 40.1 mm for
hypogonadism patients, 36.8 mm for infertility patients,
and 42.3 mm for vasectomy patients. A total of 193
men had semen data available. Nine men not seen for
infertility had semen data available. 86 % of men seen
for infertility were childless. While comparisons be-
tween individuals diagnostic groups did not reach sta-
tistical significance due to limited patient numbers,
childless men seemed to have shorter AGD (Table
2
).
Table 1
Demographic, reproductive, and anthropomorphic
characteristics of the cohort
Characteristic
n
% or Mean (SD)
Age category
<30
54
11.4
30
–
40
194
41.0
40
–
50
97
20.5
50+
128
27.1
Race
White
398
84.1
Other
75
16.0
Father
No
246
52.0
Yes
227
48.0
Office visit
General Urology
38
8.1
Erectile Dysfunction
26
5.6
Hypogonadism
101
21.6
Infertlity
231
49.4
Vasectomy
72
15.4
Height (cm)
453
179.3 (7.4)
Weight (kg)
452
93.2 (18.1)
Body Mass Index
Normal
86
18.2
Overweight
219
46.3
Obese
167
35.3
Table 2
Comparison of AGD based on reason for office evaluation. P
value represents ANOVA comparison
Childless
Father
Office Visit
n
mean AGD
(cm)
n
mean AGD
(cm)
p
General Urology
11
35.6
27
42.9
0.18
Erectile Dysfunction
3
32
23
43.8
0.25
Hypogonadism
29
38.8
72
40.6
0.56
Infertility
199
36.3
32
40.4
0.09
Vasectomy
3
30.1
69
42.8
0.08
Total
243
36.4
223
41.9
<0.01
480
J Assist Reprod Genet (2015) 32:479
–
484