Breast-Feeding and the Risk of Life-Threatening Rotavirus Diarrhea: Prevention or Postponement?

(1)

Breast-Feeding

and

the

Risk

of Life-Threatening

Rotavirus

Diarrhea:

Prevention

or

Postponement?

J.

Clemens,

MD*;

M. Rao,

MEng*;

F. Ahmed,

MBBS,

MPH*; R.

Ward,

PhDII;

S. Huda*;

J.

Chakraborty;

M. Yunus,

MBBS,

MSc*;

M.R.

Khan*;

M. Ali,

MSc;

B. Kay,

DrPH*;

F. van

Loon,

MD,

PhD*1;

and

D. Sack,

MD*

ABSTRACT. Purpose.

To assess

the

relationship

be-tween

breast-feeding and

the

risk

of

life-threatening ro-tavirus

diarrhea

among

Bangladeshi

infants

and

children younger than 24 months of age.

Design. Case-control study.

Setting. A rural

Bangladesh

community.

Participants.

One

hundred two

cases

with clinically

severe rotavirus diarrhea detected in

a treatment

center-based survefflance system during 1985

and 1986, and 2587

controls

selected

in

three surveys

of the same

community

during

the

same

calendar

interval.

Outcomes. Cases

and controls

were

compared

for the

frequency

of antecedent

breast-feeding

patterns.

Results.

Compared

with

other

feeding

modes,

exclu-sive breast-feeding

of infants

was associated

with

signifi-cant protection

against

severe

rotavirus

diarrhea

(relative

risk (BR) = 0.10; 95% confidence

interval

[CI]

= 0.03, 0.34).

However

during

the second

year

of life,

the risk of this

outcome

was higher in breast-fed

than in non-breast-fed

children (RR =

2.85; 95% CI

= 0.37, 21.71),

and no overall

protection was

associated

with

breast-feeding

during

the

first

2 years

of

life (RR = 2.61; 95% CI = 0.62, 11.02).

Conclusions. Although

exclusive

breast-feeding

ap-peared

to protect

infants

against

severe

rotavirus

diar-rhea,

breast-feeding

per se conferred

no overall

protec-tion during

the

first 2 years of life, suggesting that

breast-feeding

temporarily

postponed

rather

than

prevented

this

outcome.

While

not detracting

from efforts

to

pro-mote

breast-feeding

to alleviate

the burden

of diarrhea

due

to

nonrotaviral enteropathogens, our findings

cast

doubt

on whether

such

efforts

will impact

on the

prob-lem

of

severe

rotavirus diarrhea.

Pediatrics

1993; 92:680-685; breast-feeding, rotavirus, diarrhea.

ABBREVIATIONS. ICDDR,B, International Centre for Diarrhoeal

Disease Research, Bangladesh; RR, relative risk; RRa, adjusted relative risk; CI, confidence interval.

While

there

is little

question

that breast-feeding

of

infants

is associated

with

a reduced

rate

of diarrhea

due

to

certain

bacterial

enteropathogens,

evidence

regarding

the

protective

role

of

breast-feeding

against rotavirus

diarrhea

is conflicting.1’

Moreover,

From the *lntlio Centre for Diarrhoeal Disease Research,

Bang-ladesh; Nafiona1 Institute of Child Health and Human Development, Bethesda, MD; §Johns Hopkins School of Public Health, Baltimore, MD; liJames Gamble Institute ofMedical Research, Cincinnati, OH; and ICenters for Disease Control, Atlanta, GA.

Received for publication Mar 12, 1993; accepted May 12, 1993.

Reprint requests to (J. Clemens) National Institute of Child Health and Human Development, Rin 7503, 6100 Executive Blvd. Bethesda, MD.

PEDIATRKS (ISSN 0031 4005). Copyright C 1993 by the American

Acad-emy of Pediatrim

because

properly

controlled

studies

of the

associa-tion

between

breast-feeding

and

rotavirus

diarrhea

have

relied

on community-based,

active

survefflance

of small

cohorts

of children,

in whom

few

episodes

of

clinically

severe

rotavirus

diarrhea

were

available

for

analysis,

their

findings pertain primarily

to clinically

mild rotavirus diarrhea.5

This

absence

of

informa-tion

about

the

impact

of breast-feeding

on dinically

severe

rotavirus

infections

represents

an important

gap,

because rotavirus

infections

are

the

most

com-mon

cause

of dehydrating

diarrhea

among

infants

and

young

children

in both

developing

and

devel-o_

settings.6

Findings

from rural

Bangladesh,

reported

prey-ously,

suggest

that

breast-feeding

may

be associated

with impressive levels

of protection

against

diarrhea

due

to

Vibrio

cholerae

01

and

Shigella.7’8

Compro-hensive survefflance

of treated

diarrheal

episodes

in

a population

of nearly

200 000 persons residing in rural

Bangladesh

offered

the

additional

opportunity

to

address

whether

breast-feeding

of

infants

and

young

children

was

associated

with protection

against

clinically

severe rotavirus diarrhea.

Here

we

report an analysis indicating that, although

exclusive

breast-feeding

was

associated

with protection during

infancy,

breast-feeding

was

not

associated

with

an

overall reduction

of the risk

of severe

rotavirus

diar-rhea

during

infancy

and

early

childhood.

METHODS

Strategy

for the Evaluation

The study employed acase-control design. “Cases” were

Bang-ladeshi children younger than 24 months of age who were treated

for rotavirus diarrhea of life-threatening severity; “controls” were

similarly aged children who were contemporaneously selected

during community surveys of the same population. Cases and

controls were contrasted for antecedent histories of breast-feeding to evaluate whether breast-feeding was associated with the risk of

clinically severe rotavirus diarrhea.

Setting

for the Study

The study was conducted in the Mathb field studies area of the

International Centre for Diarrhoeal Disease Research, Bangladesh

(ICDDR,B). This field area has approximately 200 000 residents

and is located about 40 km southeast of Dhaka. This population

has access to three diarrheal treatment centers, all of which are operated or assisted by the ICDDR,B. In conjunction with a field

trial of oral cholera vaccines,9 surveillance of all patients seeking

care for diarrhea at these three centers was instituted. In this

routine surveillance, patients received systematic clinical and

mi-crobiological assessments, and selected dietary and

sociodemo-graphic information was collected. Stool specimens were

(2)

well as for V cholerae 01, enterotoxigenic E coli, and Shigella with

conventional techniques.9 A sample of rotavirus isolates was also

serotyped, as previously described.10

Selection

of Cases

Cases were children with episodes of clinically severe rotavirus

diarrhea, diagnosed in the Matlab treatment centers. Clinically

severe disease was chosen as the target outcome for the study

because of its public health importance and because limitation of

the case group to episodes of severe rotavirus diarrhea focused the

study on children whose illnesses were so severe that solicitation

of care could be presumed to be nondiscretionary. Thus, selection

biases related to the tendency to use health services were likely to have been minimal.”

To assemble cases, several definitions were required. “Diar-rhea” was defined as an illness in which at least three loose or liquid stools were passed in any 24-hour period. Visits for treat-ment of diarrhea were grouped into “episodes” if the date of onset

of symptoms leading to a visit was within 7 days of the date of

discharge from the previous visit. The “onset” of an episode was

taken as the onset of symptoms leading to the initial visit of the

episode. An episode of “rotavirus diarrhea” denoted a diarrheal episode in which fecal rotavirus was detected during any

compo-nent treatment visit. Because rotavirus typically causes watery diarrhea,’2 only episodes described as liquid or watery, without

passage of visible blood, were included in the definition of

rota-virus diarrhea.

“Severe rotavirus diarrhea” referred to an episode of rotavirus

diarrhea in which (1) the patient died during hospitalization; (2)

the patient had significant dehydration, as manifested by at least

two objective signs (feeble or absent radial pulse, poor skin turgor,

sunken eyes, dry oral mucous membranes, depressed anterior fontanelle); or (3) the patient exhibited depressed mental status, as manifested by stupor or coma. All definitions were formulated

before conduct of the study, and all decisions about fulfillment of

these criteria were made with use of a computerized algorithm

performed without knowledge of breast-feeding status.

A child with severe rotavirus diarrhea was potentially eligible

to be a case if two criteria were met: (1) the onset of the episode

occurred betweenjanuary 1, 1985, and December 31, 1986; and (2)

the child was younger than 24 months of age at the onset of the

episode. The age range for the study had been posited to be 0

through 35 months before the analysis began, but was revised toO

through 23 months since no cases of severe rotavirus diarrhea were detected in the 24- through 35-month age group. A total of

364 episodes of rotavirus diarrhea were detected during the

inter-val for the study. Forty-three episodes were exduded because at

least one fecal copathogen was isolated, and 219 were exduded

because they were not clinically severe. One hundred two

epi-sodes of severe rotavirus diarrhea that displayed typical clinical

features but lacked enteric copathogens were eligible for study.

Each episode occurred in a different child.

Selection

of Controls

Control children were selected from three community surveys

of the Matlab population, conducted in conjunction with the chol-era vaccine trial, in August through September 1985; November through December 1985; and March through April 1986. In each survey 70 clusters of geographically contiguous families were

ran-domly selected from census records of the Matlab Demographic

Survefflance System and visited in their homes to collect sera and

other relevant information, after acquisition of informed consent.’3

Each of the resulting 210 dusters was unique, and each contained approximately 300 persons. In each cluster, 20 families having at least one child younger than 5 years of age were randomly

se-lected. We identified all 2652 children from these families who were younger than 24 months of age when visited during the three surveys; the 2587 for whom dietary histories were obtained

con-stituted the control group for this study.

Acquisition

of Breast-Feeding

Histories

and Other

Relevant

Data

Information about dietary histories was collected in a uniform

fashion from mothers or caretakers by interviewers who were not aware of the study hypothesis.’4 These data were collected for cases at the time of presentation for care and for controls at the time of visits to homes during the surveys. Before undertaking the

study, we defined a child as “breast-fed” if breast milk constituted

any portion of the child’s diet, as ascertained for the day before the

onset of the episode for cases and for the day before the interview for controls. “Exclusive breast-feeding” denoted a diet in which no

other solids or liquids were induded. Other breast-fed children

were classified as “partially breast-fed.” Data about

sociodemo-graphic variables were collected during home visits to the homes

of cases, shortly after the treatment visit, and during the surveys for controls. Cases and controls were also characterized anthropo-metrically with respect to height-for-age, in relation to National Center for Health Statistics standards.’5 Height was measured to the nearest 0.1 cm during treatment visits for cases and during the

surveys for controls, using identical supine length boards for cliii-dren too young to stand and standing height sticks for older

children.

Analyses

Comparisons of cases and controls for categorical variables were statistically appraised with the test, or with Fisher’s Exact Test for contrasts of binary variables in which the population was

sparsely distributed. Contrasts of dimensional variables were

evaluated with the Student ttest, or with the Mann-Whitney LItest

when parametric assumptions were not fulfilled. To appraise the

strength of associations between breast-feeding and the risk of severe rotavirus diarrhea, we calculated odds ratios. Because of the rarity of clinically severe rotavirus diarrhea in Matlab children younger than 24 months of age (<1% risk over the study interval), odds ratios relating breast-feeding to the occurrence of severe

rotavirus diarrhea estimated the relative risk (RR) of severe

rota-virus diarrhea in breast-fed vs non-breast-fed children.’6 Relative risks <1 corresponded to protective relationships. Ninety-five

per-cent confidence intervals (CIs) for these relative risks were esti-mated with test-based methods or with exact methods when

man-dated by sparse data.’7

To adjust the associations for the effects of potentially oon-founding variables, multiple logistic regression models were fit-ted, taking case-control status as the dependent variable and breast-feeding status, together with additional covariates, as inde-pendent variables. In these models, exponentiation of the coeffi-aent for the variable for breast-feeding status enabled estimation of the adjusted RR relating breast-feeding to rotavirus diarrhea,

and 95% confidence limits for these adjusted RRs were estimated

with standard errors of the coefficients.’7 All statistical tests were interpreted in a two-tailed fashion.

Features

of the Cases

RESULTS

The

peak

age

group

for

the

102

cases

was

6

through

11 months.

All

cases

fulfilled

the

study

ci-teria

for significant

dehydration,

and

I case

died

dur-ing

hospitalization.

Rotavirus

isolates

from

59 of the

cases were culture-adapted and serotyped. The dis-tribution of the four

major

serotypes

among

the cases

(serotype

1, 17%;

serotype

2, 19%;

serotype

3, 44%;

and

serotype

4, 20%)

was similar

to that

reported

for

all

rotavirus

isolates

in

Matlab

during

1985

and

1986.10

Comparability

of Cases

and

Controls

Table

1 compares

the

case

and

control

groups

for

features

that might

have

confounded

the relationship

between

breast-feeding

and

severe

rotavirus

diar-rhea.

Although

the mean

age

at selection

was

similar

for

cases

and

controls

(11.9

months

for

cases,

12.1

months for controls),

cases

had

proportionately

fewer

subjects

aged

0 through

5 months

and

18

through

23

months.

Cases

were

significantly

more

likely

to be male,

to have

a height-for-age

below

the

median

National

Center

for Health

Statistics

z score

for the control

group,

and

to come

from families

that

were

Muslim.

at Viet Nam:AAP Sponsored on September 1, 2020

www.aappublications.org/news

(3)

TABLE 1. Baseline Characteristics of

rus Diarrhea and Community Controls

Cases With 5ev ere

Rotavi-Feature Cases

(n=102)

Controls (n=2587)

Age, mot

0-5 8 (8)*** 620(24)

6.-Il 51 (50) 695(27)

12-17 30 (29) 631 (24)

18-23 13(13) 641(25)

Male 65 (64)** 1319 (51)

Median height-for-ages 62 (61)* 1127(50)

Muslim 99 (97)** 2230(86)

Residence >2.3 miles from TC 49 (48) 1284(50)

Family size >8(1 32 (31) 1080(42)

Annual family income >18 000 Tk’I 46 (46) 1249(48)

Family has educated HH# 32 (32) 1279(49)

Educated mother# 37 (36) 911 (35)

Older mother 54 (53) 1290(50)

Family owns landtt 86 (85) 2197(85)

Family owns tubewelt 56 (55) 1371 (53)

Family owns lathne. 20 (20) 680(26)

Values represent No. (%). P < .05; ** P < .01; *** I) < .001

(two-tailed) for cited comparison.

t Age at onset of episode for cases; age on date of visit for controls.

Refers to the median z score (-2.36) for height-for-age for the

oontrol population, estimated in relation to National Center for

Health Statistics reference values. Information was available for

102 cases and 2265 controls.

S

TC, diarrheal treatment center; 2.3 miles was the median for the

control population.

II

“Family” was defined as persons who shared the same cooking

pot.

‘I The median income for the control population was 18 000 taka

(Tk). At the time of this study, the exchange rate was

approxi-mately 30 taka to one US dollar. Information was available for 101

cases and 2583 controls.

#Refers toreceipt of any formal education. Information was

avail-able for 101 cases and 2585 controls for heads of households (HH),

and for 102 cases and 2581 controls for mothers.

** Age greater than 26.5 years, the median for the control

popula-tion. Information was available for 102 cases and 2580 controls.

ft Refers to ownership of land in addition to the land on which the

family’s house was built (eg, cultivated plot). Information was

available for 101 cases and 2585 controls.

if

Refers to presence of a tubewell in the residential compound.

Information was available for 101 cases and 2585 controls.

§

Refers to any structure built for defecation and used by family

members. Information was available for 101 cases and 2585 controls.

Association

between

Breast-Feeding

and

Severe

Rotavirus Diarrhea

Overall 100

(98%)

cases

and

2454

(95%)

controls

were

breast-fed

before

selection

(Table

2). The

RR for

this comparison

was

2.71

(95%

CI

=

0.72,

22.92),

reflecting

an elevation,

albeit

not

statistically

signifi-cant,

of the RR of severe

rotavirus

diarrhea

in

breast-fed

vs non-breast-fed

infants

and

children.

This

re-lationship

was further

evaluated

in several

ways.

It

seemed

possible

that

variables

unequally

distributed

in cases

and

controls

could

have

confounded

the

re-lationship.

However,

a multiple

logistic

regression

model that estimated

the

association

while

simulta-neously controlling for age, sex, height-for-age,

reli-gion,

family

size,

and

education

of

the

household

head

also yielded an

elevated

RR (adjusted

relative

risk [RRaI =

2.61;

95% CI

=

0.62,

11.02).

Important differences

were

evident

in infants

vs

older

subjects

(Table

3). Among

infants,

there

was

a

suggestion

of protection

associated

with

breast-feed-ing

(RRa

=

0.36;

95%

CI

=

0.04,

3.11).

In contrast,

TABLE 2. Overall Association Between Breast-Feeding and the Risk of Severe Rotavirus Diarrhea in Rural Bangladesh Children

Breast-Feeding Cases Controls

Status

Breast-fed 100 (98%)*t 244(95%)

Non-breast-fed 2 133

Total 102 2587

* Crude relative risk = 2.71 (95% confidence interval [CI] = 0.72,

22.92).

t Adjusted relative risk (controlling for age [in months], sex,

reli-gion [Muslim vs other], family size, education of household head [none vs any], and height-for-age z score) = 2.61 (95% CI = 0.62,

11.02). The wider confidence interval for the crude than for the

adjusted relative risk reflects the use of exact estimation

tech-niques for the former and parametric techniques for the latter (see Text).

TABLE 3. Association Between Breast-Feeding (BF) and Severe

Rotavirus Diarrhea in Rural Bangladeshi Children, Within Age

Groups

Age and Cases Controls Relative Risk Dietary

Status Crude Adjustedt

0-Il mo

Non-BF I 10 1.00” 1.O0

AnyBF 58 1305 0.44 0.36

(0.06, 19.62) (0.04, 3.11)

Partial 55 880 0.63 0.44

(0.09, 27.61) (0.05, 3.75)

Exclusive 3 425 0.07 O.06

(0.01, 4.07) (0.01, 1.46)

12-23 mo

Non-BF 1 123 1.00 1.00

AnyBF 42 1149 4.50 2.85

(0.75, 183.22) (0.37, 21.71)

*Reference category is assigned a value of 1.00.

t Adjusted for the oovariates cited in Table 2.

1:

P = .02 (two-tailed) for trend, on logistic scale, of increasing protection for exclusive vs partial vs no breast-feeding, after

ad-justment for oovariates cited in Table 2.

§

Relative risk, after adjustment for covariates cited inTable 2, was 0.10 (P < .001; 95% confidence interval = 0.03, 0.34) for contrast

between exdusive breast-feeding vs all other dietary categories.

during

the second

year

of life,

no such

suggestion

of

protection

was

evident

(RRa

= 2.85; 95% CI =

0.37,

21.71).

Because

of the

suggestion

of a protective

as-sociation between

breast-feeding

and

severe

rotavi-rus

diarrhea

during

the

first

year

of life,

it

was

of

interest

to evaluate

protective

associations

for more

specific

breast-feeding categories.

Table

3 shows

that

in the

first

year

of life,

protective

associations

were

more

pronounced

for exclusive

breast-feeding

(RRa

= 0.06;

95%

CI

0.01,

1.46)

than

for

partial

breast-feeding

(RRa

= 0.44;

95% CI

=

0.05,

3.75).

After

ad-justment

for

the

potential

confounding

variables

cited

earlier,

the

trend

for

increasing

protection

against

severe

rotavirus

diarrhea

by

feeding

mode

(exdusive

breast-feeding

> partial breast-feeding>

non-breast-feeding)

was

significant

(P

=

.02), as was

the

association

between

exclusive

breast-feeding

(vs

all

other

dietary

modes,

including

partial

breast-feeding) and severe rotavirus diarrhea (RRa =

0.10;

95%

CI

=

0.03,

0.34).

DISCUSSION

To

our knowledge, this is the

first

properly

(4)

be-tween

breast-feeding

and

the

risk

of rotavirus

diar-rhea

of life-threatening

severity.

Although

our

data

did

not

reveal

an overall

protective

effect

of

breast-feeding during

the

first

2 years

of

life,

there

was

evidence

of protection

during

the first

year,

particu-larly

among

infants

who

were

exclusively

breast-fed.

Before

discussing

the implications

of these

findings,

it is useful

to consider

several

potential

limitations

of

the

study.

Potential

Limitations

Processes

that

determine

whether

a child

is

breast-fed

are not

random,

and

it is possible

that

observed

relationships

between

breast-feeding

and

the

risk

of

severe

rotavirus

diarrhea

could

reflect

the

fact

that

breast-fed

and

non-breast-fed

children

may

have

had

inherently

different

risks

of

this

disease

out-come,

due

to factors

that

were

correlated

with

breast-feeding

practices.

Against

this

possibility

is

the

observation

that

control

for

several

potential

con-founding

variables

had

little

effect

on the magnitude

of the

observed

associations

(Table

3). Moreover,

the

pattern

of increasing

protection

in the

first

year

with

exclusivity

of breast-feeding,

but

the

absence

of any

protective

association

with

breast-feeding

during

the

second

year

of life,

is difficult

to explain

on the

basis

of any

plausible

confounding

bias.

Nevertheless,

bias

due

to

unmeasured

confounders

cannot

be

com-pletely

excluded.

It was

striking

that

breast-feeding

was

nearly

uni-versal

among

infants

and

2-year-olds

who

had

severe

rotavirus

diarrhea.

This

observation

might

indicate

that

breast-feeding

histories

may

have

been

ascer-tained

more

intensively

in

cases

than

in

controls.

Against

this

possibifity

are

the

facts

that

breast-feed-ing

histories

were

ascertained

with

identical

criteria

in

cases

and

controls

by

data

collectors

who

were

unaware

of the study

hypothesis;

that

breast-feeding

was

also

nearly

universal

among

controls,

in accord

with

the

ago-specific

prevalence

of breast-feeding

ro-ported

in other

studies

of this

population;18

that

his-tories

were

collected

in

diarrheal

cases

without

knowledge of

microbial

etiology;

and

that

selection

of cases

and

controls

from

the

same

diarrheal

sur-veillance

system

and

community

surveys

demon-strated

a

strong

protective

relationship

between

breast-feeding

and

the

risk

of cholera.7

Another

type

of bias

could

have

resulted

from

dif-ferential

receipt

of cointerventions

by breast-fed

vs

non-breast-fed

children.

For

example,

the

apparent

absence

of protection

against

clinically

severe

rotavi-nis

diarrhea

during

the

second

year

of

life

could

have

occurred

if non-breast-fed

children

had

been

more

intensively

treated

with

oral

rehydration

solu-tion

early

during

the

course

of diarrheal

episodes.

Apart

from

the fact that

there

is no empirical

basis

to

suggest such a

scenario,

the

heterogeneous

relation-ships between breast-feeding and severe rotavirus

diarrhea

during

the

first

vs the

second

years

of life,

alluded

to above,

makes

such

a bias

very

unlikely.

Detection

bias

also

fails

to provide

a plausible

ex-planation

for

our

findings.

Restriction

of

case

epi-sodes

to those

with

diarrhea

of life-threatening

so-verity

made

it unlikely

that the results

were

distorted

by bias due

to differential

use

of the

treatment

cen-ters

by

families

of breast-fed

vs non-breast-fed

chil-dren.

In addition,

stool

specimens

were

successfully

collected from a

high

(88%)

proportion

of children

younger

than

24 months

who

received

care

for

clini-cally severe watery diarrhea during

the

study

inter-val,

and

the

proportion

of children

who

were

breast-fed

was

nearly

identical

in patients

from

whom

stool

was

collected

(94%)

and

for

patients

from

whom

stool

could

not

be obtained

(93%).

Biased assembly of cases was made unlikely by the facts that the clinical definition of severe rotavirus

diarrhea

was

formulated

a priori,

that

fulfillment

of

these

criteria

was

assessed

by

a computerized

algo-rithm,

and

that

these

assessments

were

performed

without

knowledge

of breast-feeding

status.

More-over,

although

nonsevere

episodes

were

excluded

to

safeguard

against

such

a bias,

it is nevertheless

reas-suring that analyses of

all

detected

rotavirus

diar-rheal

episodes,

regardless

of severity,

demonstrated

associations not dissimilar to those for severe rotavi-fl’s diarrhea per

se (RRa

for

infants =

0.64,

95%

CI

= 0.12, 3.31;

RRa

for

2-year-olds

= 2.18, 95% CI =

0.77,

6.21).

Controls

were

randomly

selected

from

the

commu-mty

without

reference

to whether

they

had

experi-enced

severe

rotavirus

diarrhea.

Although

such

a

so-lection

strategy

has

been

demonstrated

to

be

appropriate

for

a case-control

study

designed

to

es-timate

the

relative

risk

of an outcome

in exposed

vs

nonexposed

individuals,16

it is nevertheless

reassur-ing that

the observed

associations

were

unaffected

by

exclusion

of the

19 controls

who

were

detected

as

having

severe

rotavirus

diarrhea

during

the

study

interval.

Finally,

because

of the

infrequency

of

non-breast-fed

subjects

in this

study,

it might

be suspected

that

the

study

lacked

adequate

power

to detect

a

signifi-cant protective overall relationship between

breast-feeding

and

the risk

of severe

rotavirus

diarrhea

dur-ing the first

2 years

of life.

However,

a posteriori

type

II errors

calculated

for the

study19

argue

against

in-sufficient

power

as an

explanation

for

our

findings.

For example,

the type

H

error

of missing

a true

level

of protection

of only

10%

(eg,

an RR of 0.9)

was

less

than

.01.

Relation

to Other

Studies

Passive,

oral

administration

of antirotavirus

anti-bodies

has,

under

some

experimental

circumstances,

led

to

a reduction

in

the

incidence

or

severity

of

rotavirus diarrhea in

both

animals

and

humans.

#{176}

Although

such

data

raise

the possibility

that

natural

patterns

of breast-feeding

may

also

protect

against

rotavirus

diarrhea,

studies

have

failed

to yield

clear-cut

evidence

of such

a relationship.

Among

several

studies

of

this

topic

in

industrialized

settings,5

none

demonstrated

significant

overall

protection.

However,

results

of one

study

hinted

at a lower

in-cidence

of

rotavirus

diarrhea

in breast-fed

than

in

non-breast-fed

infants

and

young

children,3

and

two

others

suggested

that

breast-feeding

may

be

associ-ated

with

reduction

of the severity

of rotavirus

diar-rhea.4’-

The

only

study

to address

this

relationship

in

www.aappublications.org/news

(5)

a developing

country

setting

was

conducted

in rural

Bangladesh.2

This

study,

which

found

that

breast-feeding

was

more

prevalent

among

infants

and

chil-dren

with

rotavirus

diarrhea

than

among

compara-bly aged subjects with diarrhea of other etiologies,

called

into

question

whether

breast-feeding

is

protec-tively related to the risk of rotavirus diarrhea.

Our

data

indicating

that

exclusive

breast-feeding

appeared

to protect

against

clinically

severe

rotavirus

diarrhea

in Bangladeshi

infants

are

consistent

with

studies from industrialized settings.4’5 The apparent

contradiction

between

our

findings

and

those

from

the

earlier

Bangladesh

study

may

have

two

explana-tions.

First,

the

earlier

study

relied

on

the

use

of

diarrhea!

subjects

as controls.

Because

breast-feeding

is associated with a lower risk of diarrhea of several

nonrotaviral etiologies,1’ the prevalence of

breast-feeding

could

have

been

underestimated

in this

con-trol

group,

obscuring

a protective

relationship.

Our

study

attempted

to safeguard

against

this

potential

bias by using community controls. Second, the earlier

Bangladesh

study

did

not

perform

separate

analyses

of the protective

effects

of partial

vs exclusive

breast-feeding,

a distinction

that

revealed

a protective

rela-tionship among infants in our study.

Biological

and

Pragmatic

Implications

Our

findings suggest that breast-feeding can

pro-vent

clinically

severe

rotavirus

diarrhea,

particularly

when

it constitutes

the

sole

source

of food

and

liq-uids in an infant’s diet. This observation does not,

however,

permit

assessment

of whether

such

protec-tion

might

result

from

immune

or nonimmune

pro-tective

properties

of breast

milk,

on the

one

hand,

or

from

exclusion

of

fecally

contaminated

foods

and

liquids from the diet, on the other hand. If rotavirus-inhibitory properties of breast milk do explain our findings,

however,

the

absence

of protection

associ-ated

with

partial

breast-feeding

in the

second

year

of

life

suggests

that

the

potency

of

these

inhibitory

properties

is very

limited.

That

the

apparent

protection

associated

with

breast-feeding during infancy was followed by an

elevation

of risk

in

breast-fed

children

during

the

second

year

of

life has both biological and pragmatic implications. Biologically, this pattern suggests that breast-feeding may merely postpone the occurrence

of severe

rotavirus

diarrhea

to a later

age,

perhaps

because

partially

breast-fed

2-year-olds

have

lower

levels

of natural

antirotavirus

immunity

induced

by

earlier

rotavirus

infections.24’25

Pragmatically,

these

findings indicate that breast-feeding per se had no

protective

impact

on

the

overall

problem

of

life-threatening rotavirus diarrhea in rural Bangladeshi infants and children. While not detracting from ef-forts to promote breast-feeding to alleviate the

bur-den

of diarrhea

due

to nonrotaviral

enteropathogens,

our findings cast doubt on whether such efforts will

impact

on

the

problem

of severe

rotavirus

diarrhea

and

underscore

the

need

to develop

alternative

pre-ventive

interventions

that

will

be

potentially

cost-effective,

such

as rotavirus

vaccines.26

ACKNOWLEDGMENTS

This research was supported by the International Centre for

Diarrhoeal Disease Research, Bangladesh, and funded by the

United States Agency for International Development (contract

282-90-0019). The International Centre for Diarrhoeal Disease Re-search, Bangladesh, is supported by countries and agencies which share its concern for the health problems of developing countries. Current donors include the governments of Australia, Bang-ladesh, Belgium, Canada, Denmark, France, Japan, the Nether-lands, Norway, Saudi Arabia, Sweden, Switzerland, the United Kingdom, and the United States; organizations including the United Nations Development Programme, the United Nations Children’s Fund, and the World Health Organization; and pri-vate foundations including the Ford Foundation and the Sasakawa Foundation.

We thank Arabinda Banik, M. Giassuddin, Moqbul Hossain, N.

Huda, Anwarul Huq, A. Nurul Islam, M.A.H. Miah, M.A. Satter

Miah, Uaquat Mondal, M. Noorullah, Badrul Pradhan, Anisur

Rahman, Siddiqur Rahman, A.M. Sarder, K. Shaikh, Charles

Si-mons, and the additional field, laboratory, and data management staff who worked so diligently in implementation of the oral

chol-era vaccine trial, which provided the basis for this project; Kenneth Bart,Jan Holmgren,Jane Menken, Michael Merson, Henry Mosley,

Nathaniel Pierce, Ann-Mari Svennerholm, Bogdan Wojtyniak, and

Andrew Foster for helpful advice; and William B. Greenough and

Roger Eeckels for support, guidance, and encouragement.

REFERENCES

1. JasonJ, Nieburg P, MarksJS. Mortality and infectious disease associated with infant-feeding practices in developing countries. Pediatrics.

1984;74(suppl):702-727

2. Glass R,Stoll B, Wyatt R, et al. Observations questioning a protective role for breast-feeding in severe rotavirus diarrhea. Acta Paediatr Scand.

1986;75:713-.718

3. Gurwith M, Wenman W, Hinde D,Ctal. Aprospective study of rotavirus infection in infants and young children. I Infect Dis. 1981;144:218-224 4.Dully L,Byers T, Reipenhoff-Talty M, et al.The effects of infant feeding

on rotavirus induced gastroenteritis: a prospective study. Am JPublic Health. 1986;76:259-263

5. Weinberg R, lipton G, Klish W, Brown M. Effect of breast feeding on morbidity in rotavirus gastroenteritis. Pediatrics. 1984;74:250-253 6. Kapikian A, Chanock R, Rotaviruses. In: Fields B, Knipe 0, Chanock R,

et al.Fields’s Virology. 2nd ed. New York, NY: Raven Press; 1990:1353-1404

7. Clemens J, Sack D, Harris J, et al. Breast-feeding and the risk of severe cholera in rural Bangladesh children. Am IEpidemiol. 1990;131:400-411

8. Ahmed F, Clemens JD, Rao MR. Sack DA, Khan MR. Haque E. Com-munity-based evaluation of the effect of breast-feeding on the risk of microbiologically confirmed or clinically presumptive shigellosis in Bangladesh children. Pediatrics. 199290:406-411

9. ClemensJ, Sack DA, HarrisJ, et aL Field trial of oral cholera vaccines in Bangladesh: results from long-term follow-up. Lancet. 1990335:270-273

10. Ward R, Clemens J, Sack D, et a!. Culture adaptation and characteriza-tion of group A rotaviruses causing diarrheal illnesses in Bangladesh during 1985-1986. 1 Clin Microbiol. 1991;29:1915-1923

11. Horwitz R, Feinstein A. Methodological standards and contradictory results in case-control research. Am JMed. 1979;66:556-564

12. Rodriquez W, Kim H, Arrobio J, et al. Clinical features of acute gastro-enteritis associated with human reovirus-like agent in infants and young children. JPediatr. 1977;91:188-193

13. Clemens JD, van Loon FPL, Sack DA, et al. Field trial of oral cholera vaccines in Bangladesh: serum vibriocidal and antitoxic antibodies as markers of the risk of cholera. JInfect Dis. 1991;163:1235-1242

14. Clemens J, Sack D, Harris J, et aL Breast feeding and the risk of severe cholera in rural Bangladeshi children. Am IEpidemiol. 1990;131:400-411

15. National Center for Health Statistics. NCHS Growth Curves for Children: Birth to 18 Years, United States. Washington, DC: National Center for Health Statistics; 1977. Vital and Health Statistics Series 2. US Dept. of Health, Education, and Welfare Publication PHS78-1650

16. Rodrigues L, Kirkwood B. Case-control designs in the study of common diseases: updates on the demise of the rare disease assumption and the choice of sampling frame for controls. mt I Epidemiol. 1990;19:205-213

17. Kleinbaum D, Kupper L, Morgenstern H. Epidemiologic Research: Prin-ciples and Quantitative Methods. Belmont, CA: Lifetime Learning Publi-cations; 1982

(6)

19. Feinstein AR. Clinical biostatistics, X)OUV: the other side of statistical significance: alpha, beta, delta, and the calculation of sample size. Clin Pharmacol Ther. 1975;18:491-505

20. Sail L, Redman D, Smith L, Theil K. Passive immunity to bovine rota-virus in newborn calves fed colostrum supplements from immunized or non-immunized cows. Infect Immun. 1983;41:1118-1131

21. Offit F,Clark F. Protection against rotavirus induced gastroenteritis in a murine model by passively acquired gastrointestinal but not circulating antibodies. IVirol. 19855:58-64

22. Davidson G, Whyte P, Daniels E, et al. Passive immunization of children with bovine colostrum containing antibodies to human rotavirus. Lan-cet. 19892:709-712

23. Brown K, Black R, Lopez de Romana G, Creed de Kanashiro H. Infant-feeding practices and their relationship with diarrheal and other dis-eases in Huascar (Lima), Peru. Pediatrics. 198983:31-40

24. Clemens J, Ward R, Rao M, et aL Seroepidemiological evaluation of antibodies to rotavirus as correlates of the risk of clinically significant rotavirus diarrhea in rural Bangladesh.

J

Infect Dis. 1992;165:161-165 25. Ward R, Clemens J,Knowlton D, et al. Evidence that protection against

rotavirus diarrhea following natural infection is not dependent on se-rotype-specific neutralizing antibody. IInfect Dis. In press

26. de Zoysa I, Feachem It Interventions for the control of diarrheal dis-eases among young children: rotavirus and cholera immunization. Bull WHO. 1985;63:569-583

MEAD JOHNSON ORDERED TO CHANGE NAME OF RICELYTE

A Federal

district

judge

ordered

Mead

Johnson

& Company

to change

the

name

of

its

Ricelyte

infant

nutritional

supplement

and

withdraw

many

of its

health

claims. The preliminary injunction, issued Tuesday, is the latest round in Mead

Johnson’s legal battle with Abbott Laboratories for the $45 million

market

for

supplements

that

treat

infant

diarrhea

and

vomiting.

For

24 years,

Chicago-based

Abbott’s Pedialyte has dominated the market.

Mead

Johnson,

a subsidiary

of the

Bristol-Myers

Squibb

Corporation,

introduced

Ricelyte

in October

1990,

contending

that

it is rico-based

and

is a major

departure

from

solutions

already

on

the

mar-ket...

In February

1991,

Abbott

sued

Mead

Johnson,

contending

the

Ricelyte

advertis-ing

was

deceptive

and

false.

While

Judge

William

E. Steckler

agreed

with

some

of

Abbott’s

claims,

he

denied

the

request

for

a halt

to the

advertising.

Instead,

he

ordered

Mead

Johnson

to stop

contending

that

Ricelyte

contains

rice

or

is

rico-based and gave the company 180 days to destroy promotional material touting the

rice content.

It has 90 days

to stop

using

the name

Ricelyte

but can continue

to ship

the

product

to fifi existing

orders.

Mead Johnson ordered to change name of a product. The New York Times. August 20, 1992.

Noted by J.F.L., MD

(7)

1993;92;680

Pediatrics

M. Ali, B. Kay, F. van Loon and D. Sack

J. Clemens, M. Rao, F. Ahmed, R. Ward, S. Huda, J. Chakraborty, M. Yunus, M. R. Khan,

Postponement?

Breast-Feeding and the Risk of Life-Threatening Rotavirus Diarrhea: Prevention or

Services

Updated Information &

http://pediatrics.aappublications.org/content/92/5/680

including high resolution figures, can be found at:

Permissions & Licensing

http://www.aappublications.org/site/misc/Permissions.xhtml

entirety can be found online at:

Information about reproducing this article in parts (figures, tables) or in its

Reprints

http://www.aappublications.org/site/misc/reprints.xhtml

(8)

1993;92;680

Pediatrics

M. Ali, B. Kay, F. van Loon and D. Sack

J. Clemens, M. Rao, F. Ahmed, R. Ward, S. Huda, J. Chakraborty, M. Yunus, M. R. Khan,

Postponement?

Breast-Feeding and the Risk of Life-Threatening Rotavirus Diarrhea: Prevention or

the World Wide Web at:

The online version of this article, along with updated information and services, is located on

been published continuously since 1948. Pediatrics is owned, published, and trademarked by the

Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it has

at Viet Nam:AAP Sponsored on September 1, 2020

http://www.aappublications.org/site/misc/Permissions.xhtml

http://www.aappublications.org/site/misc/reprints.xhtml