Risk for Perinatal HIV-1 Transmission

According to Maternal Immunologic,

Virologic, and Placental Factors
Michael E. St. Louis, MD; Munkolenkole Kamenga, MD; Christopher Brown, MD; Ann Marie Nelson, MD;
Tarande Manzila, MD; Veronique Batter, MS; Frieda Behets, MPH; Uwa Kabagabo, MD; Robert W. Ryder, MD;
Margaret Oxtoby, MD; Thomas C. Quinn, MD; William L. Heyward, MD

Objective.\p=m-\Toevaluate how maternal and obstetric factors interact to influence

mother-to-child human immunodeficiency virus type 1 (HIV-1) transmission.
Design.\p=m-\Prospective,observational cohort study of children born to HIV\x=req-\
infected women to determine child's HIV infection status. The analysis then compared peripartum maternal, placental, and obstetric variables between HIV-1
transmitter and nontransmitter women.
Setting.\p=m-\Twolarge maternity wards in Kinshasa, Zaire.
Participants.\p=m-\Consecutivesample of 324 HIV-1\p=m-\infectedwomen at delivery,
with 254 HIV-seronegative women followed up as control subjects.
Principal Outcome Measures.\p=m-\HIVinfection status of children, to classify
each woman as an HIV-1 transmitter or nontransmitter.
Results.\p=m-\Thehighest transmission risk (TR) was associated with maternal p24
antigenemia (TR, 71%; relative risk [RR], 3.0; 95% confidence interval [CI], 1.7 to
5.2) and maternal CD8+ lymphocyte counts of at least 1.80\m=x\109/L (1800/\g=m\L)(TR,
50%; RR, 2.2; 95% CI, 1.2 to 4.2). Among women with CD8+ lymphocyte counts
of less than 1.80\m=x\109/L, CD4+ lymphocyte counts of less than 0.60\m=x\109/L were a
risk factor (TR, 29%; RR, 2.2; 95% CI, 1.2 to 4.2). In women with neither high CD8+
nor low CD4+ lymphocyte counts, placental membrane inflammation was associated with perinatal transmission (TR, 40%; RR, 4.2; 95% CI, 1.3 to 13.7). In women with neither p24 antigenemia, high CD8+ or low CD4+ lymphocyte counts, nor
placental membrane inflammation, the transmission risk was only 7%. Additional
correlates of transmission included maternal anemia and fever, but not maternal
sexually transmitted diseases.
Conclusions.\p=m-\Identifiablesubgroups of HIV-1\p=m-\infectedwomen based on maternal and placental characteristics had between a 7% and 71% risk of perinatal
HIV-1 transmission. Not only the overall rate of transmission but the impact of different risk factors for transmission appear to vary over the course of HIV infection.
(JAMA. 1993;269:2853-2859)

From Projet SIDA, Kinshasa, Zaire (Drs St. Louis,

Kamenga, Brown, Nelson, Manzila, Kabagabo, Ryder,
and Heyward and Mss Batter and Behets); Division of

HIV/AIDS, National Center for Infectious Diseases,

Centers for Disease Control and Prevention, Atlanta,
Ga (Drs St. Louis, Oxtoby, and Heyward and Ms
Batter); National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Md (Drs
Brown and Quinn); and Armed Forces Institute of Pathology, Washington, DC (Dr Nelson).
The opinions and assertions herein are those of the
authors and do not necessarily reflect the view of the
Departments of Health and Human Services or Defense.
Use of trade names is for identification only and does
not imply endorsement by the Public Health Service or
by the US Department of Health and Human Services.
Reprint requests to Centers for Disease Control and
Prevention, 1600Clifton Rd, Mailstop E-50, Atlanta, GA
30333 (Dr St. Louis).

PERINATAL transmission of human

immunodeficiency virus type 1 (HIV-1)
occurs in 13% to 40% of children born to
HIV-1-infected women.1'8 Understand
ing why some HIV-infected women
transmit HIV to their children, while
most do not, is critically important. First,
identifying risk factors for transmission
may reveal opportunities for prevention
through the medical or obstetric man
agement of HIV-infected women. Sec
ond, these risk factors will be important
variables to control for in observational
studies or intervention trials to prevent

perinatal transmission. Third, a better

understanding of each woman's profile
of risk factors should improve the coun
seling of HIV-1-infected women, be
cause women may vary markedly in risk

for HIV-1 transmission to their children.

Many potential risk factors for trans
mission have been evaluated in earlier
studies, often with conflicting results.
Low maternal CD4+ lymphocyte counts
and more advanced clinical HIV disease
have been linked to higher transmission
risk in several studies23,68 but not in oth
ers.910 Histological chorioamnionitis was
associated with increased risk for peri
natal transmission in a previous study in
Kinshasa,3 but that analysis did not con
trol for the potential confounding effect
of maternal immune status.11 Sexually
transmitted diseases (STDs) have been
identified as cofactors in the female-tomale sexual transmission of HIV,12 but
despite the infant's prolonged exposure
to the mother's genital tract during vag
inal delivery, little is known about STDs
as cofactors for perinatal HIV transmis
sion beyond one report of maternal syph
ilis linked to a higher transmission risk.13
Premature delivery is associated with
higher perinatal transmission risk,29 but
cause and effect remain unclear.
The investigation of these and other
risk factors for perinatal HIV-1 trans
mission is complicated by the high de
gree of correlation between these po
tential risk factors. For example, de
pressed cellular immunity may be as
sociated with an increased risk for
chorioamnionitis1415 and with more se
vere and protracted genital ulcer dis
eases.1617 Ascending infection by vagi
nal microbes, including sexually trans
missible agents, may cause chorioam
nionitis.1820 Chorioamnionitis and STDs
each may contribute to preterm deliv
ery.1821 Therefore, we conducted this
study to evaluate the independent role
of each of these potential risk factors in

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015

population sufficiently large to control

multiple simultaneous effects.

for

METHODS

Study Population
Between October 1989 and April 1990,
pregnant women at two large hospitals
in Kinshasa, Zaire, were offered HIV-1

testing and counseling, according to a

protocol approved by the Ethical Com
mittee ofthe Ministry of Health of Zaire.
Consenting women who tested positive
for HIV-1 antibody by rapid test (Du
Pont, Wilmington, Del), and enzyme im-

munoassay (Ortho Diagnostics, Raritan,

NJ) and were confirmed by Western
blot (Du Pont) were considered HIVinfected. All HIV-infected women and
their children were recruited for followup. For 4 months, HIV-seronegative
women matched by age (2 years) and
parity were also enrolled; these women
were retested for HIV seroconversion
12 months later. Children were moni
tored until September 1992 (mean age,
18 months; range, 13 to 24 months).

Studies in Mothers
At delivery, all enrolled mothers un
derwent the following: a physical exam
ination with scoring according to a mod
ified World Health Organization clinical
case definition for the acquired immu

nodeficiency syndrome (AIDS)22; auto

complete blood cell count
(Coulter Electronics, Luton, United
Kingdom); manual white blood cell dif
ferential by trained technologists; de

mated

termination
CD3VCD8+

commercial,

site of CD3+/CD4+ and

counts using
dual-label monoclonal anti
on

lymphocyte

bodies (Becton-Dickinson Immunocytometry, San Jose, Calif) and standard

flow cytometric analysis (FACScan,
Becton-Dickinson) following whole
blood lysis; determination of HIV-1 p24
antigen by a commercial assay (Coulter,
Hialeah, Fla) after immune complex dis
sociation by acid hydrolysis according to
manufacturer's instructions23; thick
smear for malaria of both maternal pe
ripheral blood and placenta; syphilis serology (rapid plasma reagin and Tre
ponema pallidum hemagglutination);
cervical swab for chlamydia antigen
(Abbott, North Chicago, 111); culture for
Neisseria gonorrhoeae and Trichomo
nas vaginalis; and examination for gen
ital ulcers, with culture for Haemophilus ducreyi and dark-field microscopy
for women with clinical ulcers. Blood for
lymphocyte and antigen studies was col
lected 24 to 72 hours after delivery. His
tory of the pregnancy, labor, delivery,
and neonatal course was recorded; pr
terai delivery was defined as 37 weeks'
gestation or less by dates, and postterm

delivery as 42 weeks' gestation or more.

Placentas were processed immediate
ly after delivery. A 2- to 3-cm strip of

membrane was cut with scissors and

rolled from the margin of rupture to the
placenta and secured. The whole pla
centa was immersed in 10% neutral-buff
ered formalin and fixed for at least 24
hours; then five representative samples
were taken from the membrane roll, fullthickness sections of midplacenta and
periumbilical placenta, and proximal and
distal umbilical cord. Sections were
stained by hematoxylin-eosin and ex
amined for the presence and severity of
inflammation ofthe fetal membrane and
cord and for malarial pigment and par
asites. Placental villous inflammation
(villitis) was noted if observed, but was
not a principal focus of histopathologic
review. Chorioamnionitis was diagnosed
if 10 or more polymorphonuclear leuko
cytes were present in 10 nonadjacent 400power fields, with migration of maternal
leukocytes from the decidual bed (free
membrane) or intervillous space (chorionic plate) into or through the chorion,
and was graded as mild, moderate, or
severe by the number of polymorphonu
clear leukocytes and the level of migra
tion of maternal leukocytes.24 Inflamma
tion of the umbilical cord was diagnosed
if inflammatory cells were found in any
of the umbilical vessels or in Wharton's
jelly (funisitis).25 The presence of at least
some degree of both histological chorio
amnionitis and funisitis was used for a
specific definition of placental membrane
inflammation (PMI).

Diagnosis of HIV Infection

in Children
Children were seen monthly. Blood
was drawn at 2 days of age and every 3
months thereafter. For the 2-day and

3-month specimens, polymerase chain

reaction (PCR) was performed on ap
proximately 1 g of DNA, representing
DNA from 150000 peripheral blood
mononuclear cells with the SK38/39
primer pair specific for a sequence of
p24.26 After the DNA was amplified for
35 cycles in a DNA thermal cycler (Perkin Elmer Cetus, Norwalk, Conn), HIV
proviral DNA was detected using acridinium-ester-labeledchemiluminescent
probes27; only samples for which the PCR
reaction was repeatedly positive were
reported as positive. For HIV culture,
performed every 3 months when suffi
cient peripheral blood mononuclear cells
were available, 2 to 4xl06 peripheral
blood mononuclear cells were cocultured
with 4 to 8X106 fresh donor cells. Both
cell populations were stimulated with
phytohemagglutinin and cultures were
maintained in RPMI to which 10% fetal
calf serum and interleukin 2 were add-

ed. Coculture supernatants were tested

every 3 days for 21 days for p24 anti
genemia (Coulter); cultures repeatedly
positive by the manufacturers' criteria
were considered positive. The same as
says used to detect HIV-1 antibody for
mothers were used for children.
Human immunodeficiency virus infec
tion in children was defined as a positive
virologie assay (PCR or viral culture) on
two or more specimens drawn on dif
ferent days or a persistently positive
HIV antibody assay at 15 months or
later. Uninfected children were defined
as those who were HIV antibody-neg
ative and whose virologie assays were
all negative. Other children were clas
sified as HIV-indeterminate.
Statistical Analysis
For children whose HIV infection sta
tus was known, univariate and strati
fied analyses of potential maternal and
obstetric risk factors were undertaken
and relative risks (RRs) calculated.28 The
associations between transmission prob
abilities and continuous variables (such
as CD8+ or CD4+ lymphocytes) were
evaluated first with nonparametric sta
tistics for continuous variables (Wilcoxon's rank sum test) to minimize biases
resulting from interval selection and nonnormal distributions. Selected continu
ous variables were then categorized into
quartiles and deciles of the distribution
to permit inspection of transmission risks
across these distributions. Finally,
smoothed transmission probabilities
were reviewed to help select the inter
val boundaries that best corresponded
to the relationship observed in the data
between values of the independent, con
tinuous variables and behavior of the
response variabletransmission risk.
The purpose of establishing these cut
offs was to allow calculation of measures
of association (RRs for univariate anal
yses and odds ratios [ORs] for multi-

variable, logistic regression models) and,

most important, to permit examination
of confounding and effect modification
among different risk factors.29

Multivariable logistic regression mod

els were developed using BMDP soft
ware and tested with the Hosmer-Lemeshow test.30
RESULTS
Characteristics of
Childbearing Women
Of 7364 women screened, 396 (5.4%)
tested positive for HIV antibody. Of these
396 women, 22 left the hospital without
delivering and subsequently delivered
elsewhere, 19 declined follow-up, three
died in childbirth, and 28 were not en
rolled because they delivered when study

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015

Table 1.Characteristics of

HIV-Seroposltive and HIV-Seronegative Women at Delivery, Kinshasa, Zaire*

HIV-Seropositive

Characteristics

Women

Mean age, y
Mean parity
Clinical HIV status, % of
AIDS

(n=324)

HIV-Seronegative
Women

CD8+ Count >1.80x109/L

(n=253)

26.5

26.7

3.0

3.4

women

Yes
TR=50%

No

(n=24)

RR=3.7(95%CI, 1.8-7.6)

ARC

Asymptomatic
CD4+ lymphocytes, ' -,
a0.80, % of

CD4+Count <0.60x109/L

92
0.59

mean

women

0.20-0.59

23.9

72.5

16.8

13.1

47.3

Yes
TR=29% (n=96)
RR=2.2 (95% CI,

1.2-4.2)

No
TR=13% (n=67)

<0.20
% CD4*

lymphocytes,

^=35, % of

24

mean

75.4

women

29.7
15-24

36.1

<15

1.7

CD8+ lymphocytes, cellsx109/L,

a1.80, % of women

1.1E

mean

5.5

Fig 1.Classification tree for maternal CD8+ lym

phocytes and CD4+ lymphocytes as risk factors for
perinatal human immunodeficiency virus transmis
sion. The transmission risk (TR) and the relative risk
(RR) for transmission compared with women with
neither high CD8+ nor low CD4+ lymphocyte counts
are shown for each subset of women classified by
maternal immunophenotype. CI indicates confi
dence interval.

52.8
<0.80

33.2

p24 antigenemla,

% of

Sexually transmitted diseases,

Syphilis (RPR-positive)

53.4

0t

women

positive

3.0
1.8

Gonorrhea
Genital ulcers

1.3

Trichomoniasis

Chlamydia infection
Any sexually transmitted disease
Placental histology, % of women

8.0

26.2

25.6

severe)

33.2

Chorioamnionitis and funisitis

Anemia

(hemoglobin level <0.85 g/L),

Prolonged fever, % of women

3.9
14.0

Chorioamnionitis

Chorioamnionitis (moderate to
Funisitis

0.8

10.0

13.1
% of

women

Gestational age, wk, mean

Preterm (<37 wk), % of deliveries
Postterm

(>42 wks), % of deliveries

Hours of ruptured membranes, mean
Breast-feeding
Ever, % ever breast-fed
Mean duration, mo

13.4

0.4
38.7

39.5

20.5

11.2

9.7
5.6
99.6

11.2

13.0

*HIV indicates human immunodeficiency virus; AIDS, acquired immunodeficiency syndrome; ARC, AIDS-related
complex; and RPR, rapid plasma reagin.
tp24 antigen was assayed in a sample of 10 HIV-seronegative women.

not present; the women

failed to enroll did not differ
from those enrolled in age, parity, or zone
of residence in Kinshasa, but no system
atic measure of disease severity was avail
able to compare with that of enrollees.
The remaining 324 HIV-seropositive
women enrolled differed from sero
negative women in numerous ways
(Table 1). No women received antiretroviral therapy, and only 16 had cesarean sections; 10 of these women were
HIV-seropositive. The one initially HIVseronegative woman who seroconverted to HIV-seropositive by 12 months
post partum was excluded from this
teams
whom

were

we

analysis.

Outcomes of Pregnancy and

Diagnosis of Perinatal HIV Infection
The 324 HIV-infected women gave
birth to 323 live-born infants and eight
(2.4%) stillbirths; the 254 HIV-seronegative mothers had 255 live-born in
fants and three (1.2%) stillbirths. A de
finitive laboratory diagnosis of perina
tal HIV-1 infection was established in
261 children of HIV-seropositive wom
en, among whom 69 (26%) were HIVinfected. Sixty-two (19%) live-born chil
dren of HIV-infected women were in
determinate for HIV infection; 37 (60%)
of these died before 12 months and 16
died before 1 month of age. The 27 chil-

dren whose indeterminate HIV status

was not due to known infant death be
came unavailable for follow-up princi
pally (56%) because the family moved
away from Kinshasa; in these 27 chil
dren, the last contact was at a median of
6 months of age, and 20 (74%) had at
least one negative HIV culture or PCR
before contact was lost. The HIV-in
fected children had a mean of 2.4 posi
tive viral cultures; 82% were PCR-positive by 3 months, and 85% were viral
culture-positive at least once by 12
months of age.
Risk Factors for Perinatal
HIV Transmission
Although clinical AIDS was uncom
mon and was not associated with per
inatal HIV transmission, persistent ma
ternal fever during pregnancy was as
sociated (Table 2). A high maternal CD8+
lymphocyte count was strongly associ
ated with increased risk for perinatal HIV
transmission (Table 2); this association
was not eliminated or substantially re
duced by stratification by CD4+ lympho
cyte counts, percentage of CD4+ lym
phocytes, CD4+/CD8+ ratio, p24 antigenemia, or PMI, or by multivariable models
that included any or all of these vari
ables. Evaluated alone, the CD4+ lym
phocyte count did not show a linear as
sociation with risk for perinatal trans
mission (Table 2) and was not signifi
cantly associated with transmission risk

(P=.3). However, stratifying for CD8+

lymphocyte counts of at least 1.80X 109/L
revealed a trend toward higher trans
mission risk with decreasing CD4+ lym
phocytes (Fig 1) among women with a
CD8+ lymphocyte count of less than 1.80
X109/L. In contrast, among 24 HIV-in
fected women with a CD8+ count of at
least 1.80xl09/L, for whom the overall

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015

Table 2.Characteristics of HIV-Transmitting and Nontransmitting Mothers

Pregnant Women by Univariate Analysis, Kinshasa, Zaire*
No. of Women

Factors

With Factor

Transmission
Risk, %

Among

HIV-1-lnfected

RR

(95% CI)

1.2

(0.2-6.9)
(0.5-2.5)

All woment

Clinical HIV status

AIDS
29

ARC

Asymptomatic^
Prolonged fever during pregnancy
Yes

Not
CD4* lymphocytes, cellsx109/L

241

26

12

58

a0.80

50

0.60-0.79$

33

15

0.20-0.59

84

<0.20

16

% CD4*

a35t

lymphocytes

(0.6-4.0)

32

2.1

31

2.1

(0.95-4.7)
(0.7-6.1)

32

1.9(0.7-5.1)
2.8(1.1-6.7)
3.8(1.5-9.3)

35

15-24

73

<15

30

lymphocytes, cellsx10g/L

1.80

24

0.80-1.79

103

23

58

22

<0.80t
Maternal

Yes
No

p24 antigenemia

Sexually transmitted diseases

Syphilis

14
23

10

30

Chlamydia infection
Any sexually transmitted disease
No sexually transmitted diseaset
Placental membrane inflammation
Yes

22

1.0(0.4-2.7)
1.2(0.2-6.6)
1.0 (0.2-5.4)
0.9(0.5-1.7)
0.3(0.1-4.1)
0.8(0.4-1.4)

39

1.8(1.1-2.9)

33

Genital ulcers

25

34
46

2.2(1.2-4.2)
1.04(0.57-1.9)

3.0(1.7-5.1)

219

Gonorrhea
Trichomoniasis

2.3(1.2-4.5)
1.6

25-34

CD8*

1.1

157
49

Maternal anemia

Yes

22

Not

164

Gestational age, wk
37

1.9(1.04-3.3)
24
31

1.6(0.9-4.0)

21

38

2.0(1.0-6.!

Yes

62

27

1.1

Not

157

25

38-411

149

2:42

Ruptured

membranes a10 h

(0.7-1.8)

*HIV indicates human immunodeficiency virus; RR, relative risk; CI, confidence interval; AIDS, acquired immu
nodeficiency syndrome; and ARC, AIDS-related complex.
tAII HIV-infected women for whom definitive laboratory diagnosis of HIV infection in the child was available.
tReference category for calculation of odds ratios and tests of significance.
Placental membrane inflammation is defined as chorioamnionitis and funisitis.
IIRelative risk is calculated with respect to normal gestational age (38 to 41 weeks).

transmission risk was very high (50%),

the CD4+ count was generally high (mean,
0.92xl09/L) and was unrelated to the
transmission risk. A simple classification
tree distinguished a woman's transmis
sion risk using an algorithm that first
considered the mother's CD8+ lympho
cyte count, then, second, the mother's
CD4+ lymphocyte count (Fig 1).
The percentage of CD4+ lymphocytes
was inversely associated with transmis
sion risk in a clear, stepwise fashion (Ta-

ble 2); in all statistical analyses and mod

els, the percentage of CD4+ lymphocytes
was a stronger correlate of transmission
than was the CD4+ lymphocyte count.
Nonetheless, because a high CD8+ count
was a leading risk factor for transmis
sion and because an increasing number
of CD8+ lymphocytes directly changes
(lowers) the calculated percentage of
CD4+ lymphocytes, the absolute CD4+
lymphocyte count rather than the per
centage of CD4+ was principally used in

further analyses. Also, the percentage of

CD8+ lymphocytes and the CD4+/CD8+
ratio were each associated with perina
tal transmission risk in univariate anal
yses. However, these latter two indices
are arithmetically related to, and highly
correlated with, the percentage of CD4+
lymphocytes, and neither showed any as
sociation with transmission after strati
fication for the percentage of CD4+ lym
phocytes or inclusion in a multivariable
logistic regression model along with the
percentage of CD4+ lymphocytes.
The p24 antigen was detected in only
14 (6.0%) serum samples of 233 HIVinfected women with data for both ma
ternal antigenemia and HIV transmis
sion outcome, but 10 (71%) of these antigenemic women transmitted HIV-1 to
their children (RR, 3.0; 95% confidence
interval [CI], 1.7 to 5.2). The p24 antigen
was detected in none of the 24 women
with CD8+ lymphocyte counts of 1.80
X109/L or greater. The p24 antigenemia
was associated with transmission in wom
en with CD4+ lymphocyte counts of less
than 0.60 Xl09/L (RR, 3.9; 95% CI, 1.8 to
8.3) and in women with CD4+ lympho
cyte counts of 0.60 Xl09/L or greater
(RR, 2.7; 95% CI, 0.4 to 17.3).
Chorioamnionitis and funisitis were
associated with higher risk of perinatal
transmission (Table 2); villitis was
noted in only one placenta. Chorioam
nionitis was more prevalent among wom
en with active syphilis (RR, 1.6; 95% CI,
0.7 to 3.4), but only 10 HIV-1-infected
women had syphilis and also had chil
dren with known outcome of HIV in
fection. Malaria pigment or parasites in
histological sections were not associ
ated with PMI. The association between
PMI and perinatal transmission risk was
stronger among women with neither an
elevated CD8+ nor a low CD4+ count
(RR, 4.2; 95% CI, 1.3 to 13.7) than among
women with either a high CD8+ or low
CD4+ lymphocyte count (RR, 1.5; 95%
CI, 0.8 to 2.6) (Fig 2); similarly, the trans
mission risk was elevated to approxi
mately 40% in women with either an
abnormal lymphocyte immunophenotype (high CD8+ or low CD4+ lympho

cyte count)

or

placental inflammation,

but the combination of both an immu

nologie abnormality and PMI did not
increase the transmission risk substan
tially beyond the risk for either immu
nologie abnormality or PMI alone (Fig
2). The duration of membrane rupture
before delivery was correlated with PMI,
but was not associated independently
with perinatal HIV transmission.
Severe maternal anemia (hemoglobin
level <0.85 g/L) was associated with
transmission (Table 2). There was a weak
association between maternal anemia and
placental malaria parasites or pigment,

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015

tors that

usually function effectively to

prevent maternal-child HIV transmis

sion. In this study, high CD8+ as well as
low CD4+

lymphocyte counts distin

guished two different groups of women
with elevated risk for perinatal HIV

Fig 2.Interaction between maternal immune sta

tus and placental membrane inflammation (PMI) as
risk factors for perinatal human immunodeficiency
virus transmission. Unshaded bars indicate PMI in
fection; shaded bars, no PMI infection.

but not between malaria and perinatal

HIV-1 transmission. In contrast to the
association between PMI and transmis
sion risk, the association between ane
mia and transmission risk was strongest
among women with low CD4+ lympho
cyte counts (data not shown). There was
an apparent higher risk of transmission
in both prterai (RR, 1.6; 95% CI, 0.9 to
4.0) and postterm (RR, 2.0; 95% CI, 1.0
to 6.8) infants (Table 2). None of the ma
ternal STDs evaluated were associated
with perinatal HIV transmission (Table
2), including when considered together,
but the prevalence of most STD patho
gens was low (Tables 1 and 2).
In a multivariable logistic regression
model, maternal p24 antigenemia, CD8+
lymphocyte count, PMI, and maternal
fever remained independently associ
ated with risk for perinatal transmis
sion (Table 3). In a separate multivari
able model (not shown) in which mater
nal percentage of CD4+ lymphocytes
coded as the log transform of the
continuous variablereplaced the CD4+
count, the percentage of CD4+ lympho
cytes remained significantly associated
with transmission (P=.03) without sub
stantially changing the OR and CIs for
the CD8+ lymphocytes or PMI. In 40
children born to women with none of the
four principal risk factors (CD8+ lym
phocyte count >1.80xl09/L, CD4+ lym
phocyte count <0.60X109/L, p24 anti
genemia, and PMI), only three (7%) of
the children became HIV-infected, com
pared with 38% born to women with at
least one of these risk factors.
COMMENT
Most HIV-1-infected women do not
transmit HIV to their infants. Perinatal
infections therefore reflect deficits in
maternal immunologie or placental fac-

transmission. Placental inflammation in

creased the transmission risk principal
ly among women with neither high CD8+
nor low CD4+ lymphocyte counts. There
fore, while immunologie and placental
risk factors were both associated with
increased risk for transmission, those
factors were neither synergistic nor ap
parently even additive in effect.
It is not clear why a high CD8+ lym
phocyte count identifies women with a
high risk for perinatal HIV transmis
sion. CD8+ lymphocytes are a diverse
cell population ofboth cytotoxic and sup
pressor cells with functional diversity
that might plausibly either limit3132 or
promote3233 the risk for transmission.
Furthermore, CD8+ lymphocytes vary
in additional cell surface markers and in
function over the course of HIV infec
tion34 in a manner that remains poorly
understood. The number of CD8+ lym
phocytes increases dramatically in ear
ly HIV infection as in other acute viral
infections,35 declines to the normal-tohigh range after acute HIV infection is
brought under immunologie control, and
then drops sharply before onset of
AIDS.36 During early HIV infection,3738
HIV replicates at high levels, and one
previous report of higher risk of homo
sexual HIV transmission associated with
high CD8+ lymphocyte counts hypoth
esized that early HIV infection was a
period of higher risk for sexual trans
mission.39 Acute HIV infection charac
terized by elevated CD8+ lymphocytes
is, moreover, clearly a risk factor for

breast-feeding-associated

postnatal

HIV transmission.40 While we do not

know the duration of HIV infection in
women in this study, the high mean
CD4+ lymphocyte count (0.92X107L)
among HIV-infected women with CD8+
counts of 1.80xl09/L or greater is con
sistent with the hypothesis that a high
CD8+ lymphocyte count occurred prin
cipally during relatively early HIV in
fection in this population.
These data also add to an emerging
consensus that a low maternal CD4+ lym
phocyte count increases the risk for peri
natal HIV transmission. They may also
help to resolve apparently conflicting
reports of the association between Tlymphocyte subsets and risk for HIV
transmission. The U-shaped relationship
we found between CD4+ lymphocyte
counts and transmission risk (Table 2)
suggests that detecting an association
between CD4+ lymphocyte counts and
transmission risk using standard linear

statistics will depend largely on the dis

tribution of maternal CD4+ counts in
the study population. In our setting,
these interactions were best addressed
by stratifying for high CD8+ lympho
cyte counts. Different study populations,
each with its own mean stage of HIV
disease and immunologie characteristics,
may offer different opportunities to de
tect epidemiologie associations. There
fore, future studies will need to consider
CD8+ as well as CD4+ lymphocytes and
to show the distribution of at least these
two immunologie markers in the popu
lations studied, so that apparently dis
crepant findings that are due instead to
the composition ofthe study populations
may be assessed.
In this study, the percentage of CD4+
lymphocytes was a better independent
predictor of risk for transmission than
the absolute CD4+ count. This differ
ence may be due to the lower measure
ment error of the percentage of CD4+
lymphocytes than ofthe absolute count,41
an interpretation that implies that the
percentage of CD4+ lymphocytes is sim
ply a better measure than the absolute
CD4+ lymphocyte count of worsening
immunologie status associated with pro
gressive HIV disease.4243 However, the
better statistical performance ofthe per
centage of CD4+ lymphocytes in pre
dicting transmission risk is also due at
least in part to its correlation with two
apparently distinct immunologie states
in women that are each independently
associated with elevated transmission
risk: high CD8+ lymphocyte counts and
low CD4+ lymphocyte counts. Because
understanding of the biology of HIV
transmission is still limited, it seems un
wise to obscure such possibly distinct
influences on transmission through the
exclusive use of either the percentage of
CD4+ lymphocytes or the CD4+/CD8+
ratio, each of which index does not dis
tinguish the effect of high CD8+ from
low CD4+ counts. Researchers and cli
nicians may need to consider a patient's
numbers of CD8+ and CD4+ lympho
cytes as well as the percentage of CD4+
lymphocytes, rather than a single mea
sure, to best interpret that patient's
overall immune status. Also, we note
that while the relationships between spe
cific changes in lymphocyte subsets and
transmission risk are likely to be appli
cable to other populations, the specific
cutoffs cannot be applied to other pop
ulations without evaluation. This is both
because of differences in normal immu
nologie values observed between Afri
can and North American populations4445
and because CD4+ and CD8+ numbers
are recognized to change in predictable
ways over the course of pregnancy and

delivery.4346

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015

Table 3.Multlvarlable
Kinshasa, Zaire*

Logistic Regression

Model of Risk Factors for Perinatal HIV Transmission in

Odds Ratio

Risk Factors

95% CI

Maternal

lymphocyte mmunophenotype
CD8* lymphocytes 2l.80x109/L
CD4* lymphocytes <0.60x109/Lf
p24 antigenemia

2.2

Placental membrane inflammation

2.5

hemoglobin level
Prolonged fever (a30 d)

Maternal

Gestation
s37 wk
42 wk

0.85

g/L

(pretermit
(postterm)

2.4-22.0

.001

0.9-4.9

.06

2.6-33

.001

1.2-5.2

.02
.07

2.5
4.2

1.2-15

1.3

0.6-3.0

2.2

0.8-6.3

*HIV indicates human immunodeficiency virus; and CI, confidence interval.

tVariables reflect classification tree shown in Fig 1, so that the variable for low CD4*
women with the CD8* lymphocyte count <1.80x109/L.
JNot included in final model.

Maternal p24 antigenemia was strong

ly associated with transmission risk in
dependently of lymphocyte immunophenotype, suggesting that virologie fac
tors influence transmission independent
ly of host immune factors. However,
antigen could be detected in only 6% of
women despite sensitive methods using
immune complex dissociation. Low rates
of p24 antigen detection have been pre
viously noted in serum samples of HIVinfected Africans, perhaps due to longterm immune stimulation and general
ized hypergammaglobulinemia in trop
ical countries.4445 It will be important to
develop more sensitive techniques for
measuring viral burden and activation
to better understand the relationship
between immunologie and virologie fac
tors as they influence the risk of HIV-1

transmission.
Inflammation of the placental mem
branes, as a risk factor for transmission,
was not only independent of maternal
CD4+ and CD8+ lymphocyte levels but
had a greater impact on risk of trans
mission in women who might be consid
ered "immunologically quiescent"that
is, women with neither high CD8+ nor
low CD4+ lymphocyte counts. The pla
centa is normally a highly efficient bar
rier to maternal cellular "traffic,"4748 but
PMI is by definition a circumstance of
maternal-fetal cellular exchange. Histologically detectable chorioamnionitis
may sufficiently disrupt the placental
barrier to allow maternal-fetal transfer
of cell-associated virus even in women
with low or undetectable cell-free virus.
Late gestational age at delivery was also
associated with higher risk for perinatal
transmission. A higher risk among postterm infants is biologically plausible: pla
cental senescence may increase the per
meability of the placental barrier,4950 the
skin of postterm infants is often cracked
and peeling, and the delivery itself is
often obstetrically more complicated and
traumatic. Although this association was
not noted in previous studies that re-

lymphocytes

.12

is 1

only for

ported associations between preterm de

livery and perinatal infection,29 few postterm infants were reported in those se

ries. In industrialized countries, the prac

tice of inducing labor with advanced
gestational age may have prevented ear
lier detection of this association while
perhaps inadvertently decreasing the
perinatal transmission risk compared
with some African countries. Severe ma
ternal anemia was associated with HIV
transmission in this study, but malaria
was not. Since the effect of anemia on
transmission risk was restricted to the
most immune deficient women, severe
anemia likely serves as a marker sup
plementary to the CD4+ lymphocyte
count of advanced HIV disease in Af
rican women.
Nearly universal breast-feeding and
vaginal delivery in this observational
study made evaluation of these poten
tial risk factors impossible. Also, in the
future, it will be important to try to
distinguish those HIV-infected infants
who were infected during the intrau

terine, intrapartum, or postpartum pe

breast-feeding on the basis of

laboratory data5153 and clinical progres
sion of HIV disease,54 to evaluate the

riods via

relative contribution of different risk fac

tors for transmission that occurred at
different times between conception and
the second year of life. For interpreting
the present analysis, we note that any
risk factors demonstrated in this study
either must be associated with risk for
transmission throughout all or several
of the possible stages in the maternalfetal and maternal-child relationship, or
must exert a particularly strong influ
ence on risk of transmission during at
least one specific stage.
The interactions observed between the
principal risk factors for perinatal HIV
transmission in this studyhigh CD8+
lymphocyte counts, low CD4+ lympho
cyte counts, p24 antigenemia, and inflam
mation of placental membraneslend
themselves to an interpretation based on

the natural history of HI V infection. Ear

ly HIV infection is characterized by high
HIV replication and likelihood of circu
lating, cell-free virus and was marked
best in this study by high CD8+ lympho
cyte counts and secondarily by a rela
tively high CD4+ lymphocyte count.
Viremic women in this early stage of
HIV infection may transmit HIV to the
child by infection of placental trophoblast
or Hofbauer cell,5556 by contamination of
the birth canal,5758 or by breast milk.40 A
virologically quiescent middle stage of
HIV infection is probably a low-risk pe
riod for perinatal HIV transmission; as a
result, placental barrier defects that
might lead to maternal-fetal cellular ex
change, including inflammation of fetal
membranes, postmaturity of the placen
ta, or syphilis, will have the most prom
inent impact at this stage. Finally, as
HIV disease and CD4+ lymphocyte de
pletion advance, viral replication and viremia-associated mechanisms for transmis
sion similar to those proposed for early
HIV infection may again predominate.
These findings suggest opportunities
for prevention of perinatal HIV trans
mission, but also signal that the effec
tiveness of different interventions may
vary according to the stage of maternal
HIV infection at the time of pregnancy.
Antiretroviral therapy may be more ef
fective in blocking viremia-associated
perinatal HIV transmission both very
early and late in the course of HIV in
fection. If the pathogenic agents of PMI
in HIV-infected women can be identi
fied, antimicrobial prophylaxis for chorioamnionitis59 might be considered as a po
tential strategy to prevent HIV trans
mission, which could be more effective
during the intervening years when HIV
viremia is less common and placental bar
rier defects are more prominent. Finally,
we note that women in this study had
several-fold variation in risk for perina
tal transmission, even based on maternal
CD8+ and CD4+ lymphocytes alone. If
corroborated, these findings need to be
incorporated into the counseling of preg
nant and nonpregnant HIV-infected
women and into the design of clinical
trials of interventions to block motherto-child transmission of HIV.
We gratefully acknowledge the Perinatal Study
Teams at Mama Yemo Hospital and Ngaliema
Clinics for their excellent care of patients, partic
ularly Mandala Kol, MD, Kakanda Kanjinga, MD,
Malulu Makizayi, MD, and Mayala Mabasi, MD,
under the clinical supervision of Farzin Davachi,
MD; Ndongala Lubaki and Mwamba Kasali for flow
cytometry; Chin-Yih Ou, PhD, for the PCR analy
ses; Linda Carr for expert data management;
Nathan Shaffer, MD, Robert Horsburgh, MD,
Steven McDougal, MD, Adolfo Firpo, MD, and
Meade Morgan, PhD, for thoughtful insights about
data analysis; and Martha Rogers, MD, and James
Curran, MD, MPH, for supporting the initiation of
this research.

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015

References

Oxtoby MJ. Perinatally acquired human immunodeficiency virus infection. Pediatr Infect Dis J.
1.

1990;9:609-619.

2. The European Collaborative Study. Risk factors

for mother-to-child transmission of HIV-1. Lancet.

1992;339:1007-1012.

3. Ryder RW, Wato N, Hassig SE, et al. Perinatal

transmission of the human immunodeficiency virus
type 1 to infants of seropositive women in Zaire.
N Engl J Med. 1989;320:1637-1642.
4. Blanche SB, Rouzioux C, Moscato MG, et al. A
prospective study of infants born to women seropositive for human immunodeficiency virus type 1.
N Engl J Med. 1989;320:1649-1654.
5. Halsey NA, Boiulos R, Holt E, et al. Transmission of HIV-1 infections from mothers to infants in
Haiti. JAMA. 1990;264:2088-2092.
6. Hira SK, Mangrola UG, Mwale C, et al. Apparent vertical transmission of human immunodeficiency virus type 1 by breast-feeding in Zambia.
J Pediatr. 1990;117:421-424.
7. Hutto C, Parks W, Lai S, et al. A hospital-based
prospective study of perinatal infection with human immunodeficiency virus type 1. J Pediatr. 1991;
118:347-353.
8. Mayers MM, Davenny K, Schoenbaum EE, et al.
A prospective study of infants of human immunodeficiency virus seropositive and seronegative women with a history of intravenous drug use or of
intravenous drug\p=m-\usingsex partners, in the Bronx,
New York City. Pediatrics. 1991;88:1248-1256.
9. Goedert JJ, Mendez H, Drummond JE, et al.
Mother-to-infant transmission of human immunodeficiency virus type 1: association with prematurity or low anti-gp120. Lancet. 1989;2:1351-1354.
10. Rossi P, Moschese V, Broliden P, et al. Presence of maternal antibodies to human immunodeficiency virus 1 envelope glycoprotein gp120 epitopes
correlates with the uninfected status of children
born to seropositive mothers. Proc Natl Acad Sci
U SA. 1989;86:8055-8058.
11. Chandwani S, Greco MA, Mittal K, Antoine C,
Krasinski K, Borkowsky W. Pathology and human
immunodeficiency virus expression in placentas of
seropositive women. J Infect Dis. 1991;163:1134\x=req-\
1138.
12. Cameron DW, Simonsen JN, D'Costa LJ, et al.
Female to male transmission of human immunodeficiency virus type 1: risk factors for seroconversion in men. Lancet. 1989;2:403-407.
13. Pollack H, Borkowsky W, Krasinski K. Maternal syphilis is associated with enhanced perinatal
HIV transmission. Read before the 30th Interscience
Conference on Antimicrobial Agents and Chemotherapy; October 24, 1990; Atlanta, Ga. Abstract
1274A.
14. Nelson AM, Firpo A, Kamenga M. HIV infection and chorioamnionitis: a case-control study, Kinshasa, Zaire. Read before the Annual Meeting of
the United States and Canadian Academy of Pa-

thology; March 19, 1991; Chicago, Ill.

15. Jauniaux E, Nessmann C, Imbert MC, Meuris

S, Puissant F, Hustin J. Morphological aspects of
the placenta in HIV pregnancies. Placenta. 1988;
9:633-642.
16. Musher DM, Hamill RJ, Baughn RE. Effect of
human immunodeficiency virus (HIV) on the course
of syphilis and on the response to treatment. Ann
Intern Med. 1990;113:872-881.
17. Siegel FP, Lopez C, Hammer GS, et al. Severe
acquired immunodeficiency in male homosexuals,
manifested by chronic perianal ulcerative herpes
simplex lesions. N Engl J Med. 1981;305:1439-1444.
18. Hillier SL, Martius J, Krohn M, Kiviat N,
Holmes KK, Eschenbach DE. A case-control study
of chorioamnionic infection and histologic chorioamnionitis in prematurity. N Engl J Med. 1988;
319:972-978.
19. Handsfield HH, Hodson WA, Holmes KK. Neonatal gonococcal infection, I: orogastric contamination with Neisseria gonorrhoeae. JAMA. 1973;225:
697-701.

20. Blanco JD, Gibbs RS, Malherbe H, et al. A

controlled study of genital mycoplasmas in amniotic fluid from patients with intra-amniotic infection. J Infect Dis. 1983;147:650-653.
21. Guzick DS, Winn K. The association of chorioamnionitis with preterm delivery. Obstet Gynecol.

1985;65:11-16.

22. Colebunders RL, Mann JM, Francis H, et al.

Evaluation of a clinical case definition of AIDS in
Africa. Lancet. 1987;1:492-494.
23. Quinn TC, Kline R, Moss MW, Livingston RA,
Hutton N. Acid dissociation of immune complexes
improves diagnostic utility of p24 antigen detection
in perinatally acquired HIV-1 infection. J Infect
Dis. 1993;167:1193-1196.
24. Blanc WA. Pathology of the placenta. In:
Naeye RL, Kissane JM, Kaufman N, eds. Perinatal
Diseases. Baltimore, Md: Williams & Wilkins; 1981:
67-132. International Academy of Pathology mono-

graph.

H, Elston CW. Infection of the placenta.

H, ed. Pathology of the Placenta. Philadelphia, Pa: WB Saunders; 1978:286-325.
26. Rogers MF, Ou CY, Rayfield M, et al. Use of
polymerase chain reaction for early detection of the
proviral sequence of human immunodeficiency virus in infants born to seropositive mothers. N Engl
J Med. 1989;320:1649-1654.
27. Ou CY, McDonough SH, Cabanas D, et al. Rapid and quantitative detection of enzymatically amplified HIV-1 DNA using chemiluminescent oligonucleotide probes. AIDS Res Hum Retroviruses.
1990;6:1323-1329.
28. SAS Procedures Guide for Personal Computers, Version 6 Edition. Cary, NC: SAS Institute
Inc; 1985.
29. Kleinbaum DG, Kupper LL, Morgenstern H.
Epidemiologic Research: Principles and Quantitative Methods. New York, NY: Van Nostrand Reinhold Co; 1982.
30. BMDP Statistical Software Manual, 1990 Release. Los Angeles, Calif: BMDP Statistical Software Inc; 1990.
31. Walker CM, Moody DJ, Stites DP, Levy JA.
CD8+ lymphocytes can control HIV infection in
vitro by suppressing virus replication. Science. 1986;
25. Fox
In: Fox

234:1563-1566.
32. Bell SJD, Cooper DA, Kemp BE, Doherty RR,
Penny R. CD8+ T-cells from HIV-9 infected patients can either augment or abrogate HIV-specific
lymphoproliferation. Clin Immunol Immunopathol.

1992;64:254-260.

33. Clerici M, Roilides E, Via CS, Pizzo PA, Shearer GM. A factor from CD8 cells of human immunodeficiency virus\p=m-\infectedpatients suppresses
HLA self-restricted T helper cell responses. Proc
Natl Acad Sci U S A. 1992;89:8424-8428.
34. Prince HE, Jensen ER. Three-color cytofluorometric analysis of CD8 cell subsets in HIV-1 infection. J Acquir Immune Defic Syndr. 1991;4:1227\x=req-\
1232.
35. Lang W, Perkins H, Anderson RE, Royce R,
Jewell N, Winkelstein W. Patterns of T lymphocyte
changes with human immunodeficiency virus infection: from seroconversion to the development of
AIDS. J Acquir Immune Defic Syndr. 1989;2:63\x=req-\
69.
36. Goedert JJ, Kessler CM, Aledort LM, et al. A
prospective study of human immunodeficiency virus type 1 infection and the development of AIDS
in subjects with hemophilia. N Engl J Med. 1989;
321:1141-1148.
37. Ho DD, Moudgil T, Alam M. Quantitation of
human immunodeficiency virus type 1 in the blood
of infected persons. N Engl J Med. 1989;321:1621\x=req-\
1625.
38. Coombs RW, Collier AC, Allain J-P, et al. Plasma viremia in human immunodeficiency virus infection. N Engl J Med. 1989;321:1626-1631.
39. Seage GR, Horsburgh CR, Hardy AM, et al.
Increased suppressor T cells in probable transmitters of human immunodeficiency virus infection.

Am J Public Health. 1989;79:1638-1642.

40. Van de Perre P, Simonon A, Msellati P, et al.
Postnatal transmission of human immunodeficiency virus type 1 from mother to infant. N Engl
J Med. 1991;325:593-598.
41. Stein DS, Korvick JA, Vermund SH. CD4+ lymphocyte cell enumeration for prediction of clinical
course of human immunodeficiency virus disease: a
review. J Infect Dis. 1992;165:352-363.
42. Taylor JM, Fahey JL, Detels R, Giorgi JV. CD4
percentage, CD4 number, and CD4:CD8 ratio in
HIV infection: which to choose and how to use.
J Acquir Immune Defic Syndr. 1989;2:114-124.
43. Miotti PG, Liomba G, Dallabetta GA, Hoover
DR, Chiphangwi JD, Saah AJ. T lymphocyte subsets during and after pregnancy: analysis in human
immunodeficiency virus type 1-infected and \p=m-\un
infected Malawian mothers. J Infect Dis. 1992;165:
1116-1119.
44. Brown C, Kline C, Atibu L, Francis H, Ryder
RW, Quinn TC. Prevalence of HIV-1 p24 antigenemia in Africa and North American populations and
correlation with clinical status. AIDS. 1991;4:89-92.
45. Quinn TC, Piot P, McCormick JB, et al. Serologic and immunologic studies in patients with AIDS
in North America and Africa. JAMA. 1987;257:2617\x=req-\
2621.
46. Dinsmoor MJ, Christmas JT. Changes in T\x=req-\
lymphocyte subpopulations during pregnancy complicated by human immunodeficiency virus infection. Am J Obstet Gynecol. 1992;167:1575-1579.
47. Herzenberg LA, Bianchi DW, Schroder J, Cann
HM, Iverson GM. Fetal cells in the blood of pregnant women: detection and enrichment by fluorescence-activated cell sorting. Proc Natl Acad Sci
U SA. 1979;76:1453-1457.
48. Hunziker RD, Gambel P, Wegmann TG. Placenta as a selective barrier to cellular traffic.
J Immunol. 1984;133:667-671.
49. Clifford SH. Postmaturity with placental dysfunction. J Pediatr. 1954;44:1-8.
50. Vorherr H. Placental insufficiency in relation
to post-term pregnancy and fetal postmaturity. Am
J Obstet Gynecol. 1975;123:67-92.
51. Burgard M, Mayaux MJ, Blanche S, et al. The
use of viral culture and p24 antigen testing to diagnose human immunodeficiency virus infection in
neonates. N Engl J Med. 1992;327:1192-1197.
52. Alimenti A, O'Neill M, Sullivan JL, Luzuriaga
K. Diagnosis of vertical human immunodeficiency
virus type 1 infection by whole blood culture.
J Infect Dis. 1992;166:1146-1148.
53. Bryson YJ, Luzuriaga K, Sullivan JL, Wara
DW. Proposed definitions for in utero versus intrapartum transmission of HIV-1. N Engl J Med.

1992;327:1246-1247.

S, Tardieu M, Duliege A, et al. Longitudinal study of 94 symptomatic infants with perinatally acquired human immunodeficiency virus infection: evidence for a bimodal expression of clinical
and biological symptoms. AJDC. 1990;144:1210-1215.
55. Lewis SH, Reynolds-Kohler C, Fox HE, Nelson JA. HIV-1 in trophoblastic and villous Hofbauer cells, and haematological precursors in eight\x=req-\
week fetuses. Lancet. 1990;335:565-568.
56. Martin AW, Brady K, Smith SI, et al. Immu54. Blanche

nohistochemical localization of human immunodeficiency virus p24 antigen in placental tissue. Hum
Pathol. 1992;23:411-414.
57. Ehrnst A, Lindgre S, Dictor M, et al. HIV in
pregnant women and their offspring: evidence for
late transmission. Lancet. 1991;338:203-207.
58. Goedert JJ, Duliege AM, Amos CI, Felton S,
Biggar RJ. International Registry of HIV-Exposed
Twins: high risk of HIV-1 infection for first-born
twins. Lancet. 1991;338:1471-1475.
59. Mercer BM, Moretti ML, Prevost RR, Sibai
BM. Erythromycin therapy in preterm premature
rupture of the membranes: a prospective, randomized trial of 220 patients. Am J Obstet Gynecol.

1992;166:794-802.

Downloaded From: http://jama.jamanetwork.com/ by a Karolinska Institutet University Library User on 01/29/2015