ARTIGO ARTICLE 773

Risk and prognostic factors for diarrheal

disease in Brazilian infants:
a special case-control design application

Fatores de risco e prognsticos para diarria

entre crianas brasileiras: uma aplicao especial
do delineamento de casos e controles

Sandra Costa Fuchs 1

Cesar Gomes Victora 2

1 Departamento de Abstract The aim of this study was to examine the effect of demographic, socioeconomic, envi-
Medicina Social,
ronmental, maternal reproductive, dietary, and nutritional variables on diarrhea risk and prog-
Faculdade de Medicina,
Universidade Federal nosis using a hierarchical framework. A case-control study of children aged 0-23 months in
do Rio Grande do Sul. Greater Metropolitan Porto Alegre was conducted during the peak season for diarrhea in 1987-
Rua Ramiro Barcelos 2600,
1988. Three groups were investigated, with 192 children each. The first group included hospital-
sala 415, Porto Alegre, RS
90035-003, Brasil. ized children with an episode of acute diarrhea complicated by moderate to severe dehydration.
scfuchs@zaz.com.br The second group included children with acute mild diarrhea without signs of dehydration who
2 Departamento de Medicina
were identified in the same neighborhood as hospitalized cases. The third group consisted of
Social, Faculdade de
Medicina, Universidade controls without diarrhea. Mothers were interviewed by trained interviewers using a standard-
Federal de Pelotas. ized questionnaire. Data analysis included a hierarchical approach to control for confounding,
C. P. 464, Pelotas, RS
96001, Brasil.
using conditional logistic regression. Comparison of the three groups aimed to identify risk fac-
tors for diarrhea complicated by dehydration, prognostic factors for dehydration, and risk fac-
tors for mild diarrhea. Low birth weight, stunting, and lack or breastfeeding acted simultane-
ously as risk and prognostic factors for diarrhea.
Key words Diarrhea; Child Health; Risk Factors; Case-Control Studies; Epidemiology

Resumo O objetivo deste estudo foi examinar o efeito de variveis demogrficas, scio-econ-
micas, ambientais, reprodutivas maternas, dietticas e nutricionais sobre o risco e o prognstico
de diarria usando anlise hierarquizada. Um estudo de caso-controle incluindo crianas entre
0-23 meses de idade, residentes na rea metropolitana de Porto Alegre, foi realizado durante os
meses de vero de 1987-88. Investigaram-se trs grupos de 192 crianas, incluindo as hospitali-
zadas com um episdio agudo de diarria e desidratao moderada a grave; as crianas com
diarria aguda leve e sem desidratao, identificadas na vizinhana dos casos, assim como
crianas sem diarria. Entrevistas padronizadas foram realizadas por entrevistadoras treinadas.
Utilizou-se anlise hierarquizada no controle de fatores de confuso. Calcularam-se razes de
odds e intervalos de confiana de 95% atravs de regresso logstica condicional. O emparelha-
mento de diferentes combinaes de casos e controles permitiu invetigar fatores de risco para
diarria grave, fatores prognsticos e fatores de risco para diarria leve. Baixo peso de nascimen-
to, dficit altura-idade e ausncia de aleitamento materno foram simultaneamente fatores de
risco e prognstico para diarria.
Palavras-chave Diarria; Sade Infantil; Fatores de Risco; Estudos de Casos e Controles; Epi-
demiologia

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

774 FUCHS, S. C. & VICTORA, C. G.

Introduction Study definitions

Diarrheal diseases are still associated with a Dehydrating diarrhea cases included children
high mortality rate among children (Murray & aged 0-23 months residing in Greater Metro-
Lopez, 1997), causing an estimated 1.5 million politan Porto Alegre, southern Brazil, enrolled
deaths in the year 2000 ( Victora et al., 2000). from December 1987 to March 1988 from the
Diarrhea is a worldwide problem, but the im- citys two largest pediatric hospitals. Dehydrat-
plications of diarrheal diseases are particularly ing diarrhea was defined as an episode of acute
evident in developing countries. diarrhea (less than eight days duration) and
Although oral rehydration therapy has been presence of a persistent skinfold plus at least
available for approximately thirty years and its one of the following signs: sunken fontanel, dry
use has contributed to a major decline in the mouth and tongue, sunken eyes, reduced uri-
diarrheal mortality rate ( Victora et al., 2000), nary output, weak pulse, drowsiness, or irri-
diarrheal morbidity is still a major problem tability. Diagnosis of diarrhea was based on
among under-fives, mostly in the first year of three or more loose or watery bowel move-
life. Pooled data published in 1992 from 22 lon- ments within 24 hours for children older than 3
gitudinal studies conducted in 12 countries months or according to the mothers report of
showed that infants aged 6-11 months had a more frequent and poorly formed stools (as
median incidence of five episodes of diarrhea compared to normal) for younger children. All
per year (Bern et al., 1992). Several reports have children meeting these criteria were enrolled
identified socioeconomic, environmental, ma- from the largest pediatric hospital, and a sys-
ternal, nutritional, and other characteristics as tematic random sample of eligible children
risk factors for diarrheal morbidity or mortality from the second hospital was included. Cases
(Awasthi et al., 1996; Clemens et al., 1999; Mirza of mild diarrhea were defined as children with
et al., 1997). Poverty, low parental schooling, diarrhea in the seven days preceding the inter-
poor sanitation, lack of water supply, crowding, view and without signs of dehydration. They
early childbirth, short birth intervals, lack of were individually matched to the severe diar-
breastfeeding, and malnutrition are factors as- rhea cases by age (0-11, 12-23 months) and
sociated with diarrhea (Awasthi et al., 1996; neighborhood. When a hospital case was iden-
Brattacharya et al., 1995; Howie et al., 1990; tified, an interviewer visited this childs home
Mirza et al., 1997; Raisler et al., 1999; Scariati et to interview the mother. Starting at this site,
al., 1997). the interviewer used a standard procedure to
Most studies have used cross-sectional or move around the neighborhood, visiting every
cohort designs as opposed to a case-control house until a child with mild diarrhea was
methodology. We found no case-control stud- identified.
ies designed to simultaneously compare risk Non-diarrhea controls were children iden-
and prognostic factors for dehydrating diar- tified in the same neighborhood and from the
rhea, using data from the same population. same age bracket as dehydrating diarrhea cas-
Identification of factors related simultaneously es, who had not presented diarrhea in the pre-
to the risk of acquiring diarrhea and its prog- ceding seven days.
nosis would be particularly relevant, potential- Figure 1 shows the framework for enroll-
ly bolstering the effect of a particular interven- ment of cases and controls. Risk factors for de-
tion. We therefore examined the effects of so- hydrating diarrhea were investigated by com-
cioeconomic, environmental, maternal repro- paring cases of dehydrating diarrhea and non-
ductive, dietary, and nutritional variables on diarrhea controls. Prognostic factors for dehy-
the risk and prognosis of diarrhea, using data dration were established by comparing cases of
from a 1987-1988 case-control study. dehydrating and mild diarrhea. Finally, risk
factors for mild diarrhea were ascertained by
comparing cases of mild diarrhea with non-di-
Participants and methods arrhea controls.

Design Study variables and data collection

This was a case-control study including three Standardized interviews with mothers or care-
groups: dehydrating diarrhea cases, mild diar- takers provided data on socioeconomic, mater-
rhea cases, and non-diarrhea controls. nal reproductive, demographic, nutritional,
and health-care characteristics. The mothers
skin color and key environmental characteris-

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

 RISK AND PROGNOSTIC FOR DIARRHEAL DISEASE 775

tics were assessed through observation by six Figure 1

trained interviewers. The variables presented
in this paper included childrens age (0-1, 2-3, Framework for studying risk and prognostic factors for diarrheal diseases.
4-5, 6-8, 9-11, 12-17, or 18-23 months), family
income (measured as times the prevailing
minimum wage; 3.6 or > 3.6), fathers pres-
Healthy children
ence in household and schooling (absent and
illiterate or 1 or more years of schooling),
mothers schooling (in years, 8 or > 8), type of Risk factors for mild diarrhea
housing (masonry construction versus shacks), v
water supply (indoor running water, outdoor Risk factors for Children with
running water on property, outdoor running dehydrating diarrhea mild diarrhea
water in neighborhood, public well, or river),
flush toilet (yes or no), number of under-fives Prognostic factors for diarrhea
living in the house (1-2 or 3-6), home cleanli- v
ness (index based on observation of availabili- Children with

v
ty of soap for washing hands and towels for dehydrating diarrhea
drying, food scraps in uncovered pans, pans
kept covered on the stove, presence and num-
ber of flies in the kitchen or living room, feces
or standing water in the yard, presence of pets),
maternal age ( 20 years old versus > 20), twin
birth (yes or no), birth order (1-2, 3, > 3), birth
weight (reported by the mother or recorded on case-control combination through conditional
the birth certificate, categorized as < 2,500, logistic regression using the Egret statistical
2,500-2,999, 3,000g), height-for-age, weight- package (Breslow & Day, 1980).
for-height, and weight-for-age (all three in z-
scores categorized as, 2, -1 to -1,9, > -1), pre-
vious hospitalization for any reason (yes or no), Results
type of milk consumed (breast, breast + non-
breast, non-breast), and current breastfeeding All but eight children out of 200 identified with
status (still breastfeeding, if breastmilk consti- dehydrating diarrhea were studied. Seven chil-
tuted any portion of the childs diet; stopped dren with mild diarrhea could not be inter-
breastfeeding; never breastfed). Since breast- viewed and were replaced by their next-nearest
feeding may have been interrupted as a result eligible neighbor. All eligible children without
of the diarrhea, children weaned during the diarrhea agreed to participate. Table 1 shows
episode were classified as still breastfed. the distribution of target variables according to
the diarrhea outcomes. Most of the character-
Sample size and strategies for data analysis istics of children with mild diarrhea showed an
intermediate distribution between those from
The sample size was calculated to detect an odds dehydrating diarrhea cases and non-diarrhea
ratio of at least 1.5, with 80% power and 5% sig- controls.
nificance level (two-tailed) for a prevalence of Table 2 shows that age was closely related
exposure among controls ranging from 20% to to all diarrhea outcomes, but the groups at
65%. Independent variables were grouped into highest risk varied. Children in the first two
different hierarchical levels of determination, months of life were protected from developing
ranging from distal determinants such socioe- diarrhea, but once they acquired the disease,
conomic characteristics, including environ- they were at about 23 times the risk of dehydra-
mental, maternal reproductive, and demo- tion as compared to those in the 9-11 month
graphic factors, and leading to proximal deter- age bracket. Infants aged two to three months
minants such as nutrition (anthropometry and had seven times the likelihood of developing
diet). We included variables at each level based dehydrating diarrhea as compared to those 9-
on the strength of association in the crude 11 months old. Due to matching, children aged
analyses (a p level < 0.1 was required). 12-23 months had to be analyzed separately
One regression equation was fitted for each from those under 12 months. In the second
hierarchical level, also including variables from year of life an increased risk of dehydrating
higher levels of determination ( Victora et al., diarrhea appeared in children aged 12 to 17
1997). We conducted separate analyses for each months. Our data analysis did not indicate any

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

776 FUCHS, S. C. & VICTORA, C. G.

Table 1 gender-related differences in risk or prognostic

factors for diarrhea.
Distribution of variables for healthy children and according to diarrhea outcomes. Low socioeconomic status was one deter-
minant of dehydrating diarrhea, although
Variables Dehydrating Mild Healthy there was no significant association with mild
diarrhea diarrhea children diarrhea (Table 3). Low family income doubled
% (n = 192) % (n = 192) % (n = 192)
the risk of dehydrating diarrhea, independent-
Age (months) ly of the fathers presence in the household or
0-1 16 4 14 educational level or the mothers skin color. Pa-
2-3 23 15 9 ternal schooling was the main socioeconomic
4-5 17 11 13 risk and prognostic factor for dehydrating diar-
6-8 20 28 24 rhea, while maternal education had no inde-
9-11 9 29 25 pendent effect. Mixed skin color classification
12-17 9 7 5 of mothers, although adjusted for other socioe-
18-23 5 6 9 conomic variables, was associated with a two-
fold increase in the risk and in a poor progno-
Gender
sis for dehydration. Children of working moth-
Male 53 52 51
ers were less likely to have a poor prognosis
Female 47 48 49
than children of non-working mothers.
Family income (times Table 4 shows that living in the same house-
minimum wage)
hold with three to six other under-fives or lack
3,6 19 24 29
of home cleanliness (in the kitchen, living
< 3,6 81 76 71
room, and yard) increased the risk of dehydrat-
Fathers presence ing diarrhea, since both variables were associ-
and schooling (years) ated with the risk of mild diarrhea as well as
1 61 82 83 with a poor prognosis. Lack of a refrigerator
absent or 0 39 18 17 was associated with a two-fold increase in the
Maternal schooling (years) risk of developing mild diarrhea.
8 11 15 18 Children of mothers under 20 years of age
<8 89 85 82 had approximately four times the risk of dehy-
dration, since this exposure was independently
Mothers skin color
associated with risk and prognosis for dehy-
White 53 68 66
drating diarrhea (Table 5). Among the remain-
Black 20 18 17
ing maternal reproductive characteristics, high-
Mixed 27 14 16
er birth order was associated with a poor prog-
Maternal work nosis, while twinship was associated with in-
No 59 53 62 creased risk of dehydrating diarrhea.
Yes 41 47 38 Table 6 shows the strength of independent
Type of housing associations between nutritional status/feed-
Masonry 54 66 72 ing practices and diarrhea outcomes. Most nu-
Shack 46 34 28 tritional variables were significantly associated
with both the risk of dehydrating diarrhea and
Water supply
a poor prognosis, but not with risk of mild di-
Indoor running water 52 58 68
arrhea. Low birth-weight children were at a sig-
Outdoor running water (on property) 31 28 20
nificantly higher risk of developing dehydra-
Outdoor running (public) 11 10 9
tion as compared to normal-weight children.
Well or river 6 4 2
Poor nutritional status as measured by height-
Use of refrigerator for-age resulted in a 4.5-fold increase in the
Yes 42 49 59 risk of dehydrating diarrhea. Poor feeding prac-
No 58 51 41 tices accounted for both an increased risk and
Number of under-fives in household a worse prognosis. During the first two years of
1-2 77 86 89 life, interrupting breastfeeding just prior to the
3-6 23 14 11 onset of the episode resulted in a 6-fold in-
crease in the risk of developing dehydrating di-
Home cleanliness
arrhea.
Yes 21 37 45
No 79 63 55

(to be continued on the next page)

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

 RISK AND PROGNOSTIC FOR DIARRHEAL DISEASE 777

Discussion Table 1 (continued from previous page)

Variables Dehydrating Mild Healthy

Table 7 summarizes both risk and prognostic diarrhea diarrhea children
factors. Factors that increased the risk of dehy- % (n = 192) % (n = 192) % (n = 192)
drating diarrhea were associated with a poor
Maternal age (years)
prognosis for a given episode of diarrhea rather
20 76 80 88
than increasing the risk of mild diarrhea. Inso-
< 20 24 20 12
far as the study was able to ascertain, socioeco-
nomic variables did not distinguish children at Birth order
risk of developing mild diarrhea. 1-2 41 52 54
Children in low-income areas often have 3 22 20 13
similar exposures, such as low parental educa- >3 37 28 33
tion. The neighborhood-matched design did Twin
not allow the identification of these variables No 94 97 99
as risk factors for developing mild diarrhea. Al- Yes 6 3 1
though conditional logistic regression was per-
Birth weight (grams)
formed and the analyses were based on discor-
> 3,000 54 67 73
dant pairs, overmatching may still have affect-
2,9992,500 22 22 18
ed the precision of odds ratios estimates (Roth-
< 2,500 24 11 9
man & Greenland, 1998).
Children were matched in two broad age Height-for-age (z score)
categories to the dehydrating diarrhea cases. > -1 30 59 64
However, since there were residual differences -1 to -1.9 37 27 25
and analyses were conducted within small in- -2 33 14 11
tervals, age was both a risk and prognostic fac- Weight-for-age (z score)
tor for dehydrating diarrhea. Children under > -1 46 63 73
two months were largely protected from mild -1 to -1.9 29 26 20
diarrhea, but they were also at increased risk of -2 25 11 7
dehydration, since the prognosis of diarrhea
improved sharply with age. Furthermore, chil- Type of milk consumed

dren aged two to three months were equally ex- Breast 8 15 23

posed to a high risk and poor prognosis for de- Breast + non-breast 17 23 23

hydrating diarrhea. These associations may be Non-breast 75 62 54

linked to feeding practices, particularly the in- Breastfeeding history

terruption of breastfeeding. Still breastfeeding 25 38 46
The effects of each socioeconomic risk fac- Stopped breastfeeding 70 59 50
tor were adjusted for those of all other such Never breastfed 5 3 4
variables. Both very low household income and
low paternal education were independent de-
terminants of dehydrating diarrhea. Although
most studies detected maternal schooling as a
risk factor for childhood diarrhea (Hussain & 1996). However, once their children had devel-
Smith, 1999; Yoon et al., 1996), our study did oped diarrhea, working mothers reinforced their
not detect an association with mothers educa- care and were thus able to prevent dehydration.
tion. This may be explained by the inclusion of Although several environmental factors were
paternal schooling in the same model, attenu- significantly associated with diarrheal morbid-
ating the effect of the former (Desai & Alva, ity according to bivariate analysis, after adjust-
1998) and suggesting that the effect of mater- ing for confounding only the lack of a refrigera-
nal education is not independent of socioeco- tor in the household remained as a risk factor
nomic level (Dargent-Molina et al., 1994). Ma- for mild diarrhea (Hussain & Smith, 1999).
ternal work had a paradoxical effect. Children Families living in this city seldom used open
of working mothers tended to be at increased wells or rivers as their source of drinking and
risk of mild diarrhea (although statistically not cooking water. Accordingly, the odds ratios were
quite significant) (Hussain & Smith, 1999) but approximately equal to three but the confi-
at the same time had a better prognosis. Mater- dence intervals were wide due to the low preva-
nal work may keep women outside the home lence of exposure. Children of families who had
without simultaneously ensuring adequate outdoor running water had a two-fold risk of
child care (Lamontagne et al., 1998; Reed et al., dehydrating diarrhea as compared to those with

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

778 FUCHS, S. C. & VICTORA, C. G.

Table 2

Association between demographic variables and diarrhea outcomes.

Variables Dehydrating diarrhea Prognosis of diarrhea Mild diarrhea

Crude odds ratio (95% CI) Crude odds ratio (95% CI) Crude odds ratio (95% CI)

Age (months)
0-1 2.6 (1.3-5.5) 23.1 (6.9-77.8) 0.2 (0.1-0.6)
2-3 7.1 (3.0-16.5) 7.1 (2.9-17.4) 1.2 (0.6-2.5)
4-5 3.5 (1.6-7.5) 6.8 (2.7-17.3) 0.9 (0.4-1.7)
6-8 2.4 (1.2-4.8) 3.2 (1.4-7.4) 0.9 (0.5-1.5)
9-11 1.0 1.0 1.0
p level* < 0.001 < 0.001 0.05
12-17 3.7 (1.0-13.1) 0.8 (0.3-2.5) 0.9 (0.6-1.3)
18-23 1.0 1.0 1.0
p level* 0.03 0.80 0.60

Gender
Male 1.0 1.0 1.0
Female 0.9 (0.6-1.4) 0.9 (0.6-1.5) 0.9 (0.6-1.3)
p level* 0.70 0.90 0.60

* p level assessed by likelihood ratio test.

Table 3

Association between socioeconomic variables and diarrhea outcomes.

Variables Dehydrating diarrhea Prognosis of diarrhea Mild diarrhea

Adjusted odds ratio1 Adjusted odds ratio2 Adjusted odds ratio3
(95% CI) (95% CI) (95% CI)

Family income (times minimum wage)

3.6 1.0 1.0 1.0
< 3.6 1.8 (1.0-3.1) 1.2 (0.6-2.1) 1.4 (0.9-2.2)
p level* 0.04 0.60 0.15

Fathers presence and schooling (years)

1 1.0 1.0 1.0
absent or 0 2.7 (1.6-4.8) 2.7 (1.6-4.6) 0.9 (0.5-1.7)
p level* < 0.001 < 0.001 0.90

Maternal schooling (years)

8 1.0 1.0 1.0
<8 1.5 (0.8-2.7) 1.4 (0.7-2.8) 1.5 (0.8-2.6)
p level* 0.20 0.30 0.20

Mothers skin color

White 1.0 1.0 1.0
Black 1.7 (0.9-3.1) 1.4 (0.8-2.5) 0.9 (0.5-1.6)
Mixed 2.2 (1.2-4.1) 2.3 (1.2-4.2) 1.1 (0.7-1.9)
p level* 0.02 0.03 0.90

Maternal work
No 1.0 1.0 1.0
Yes 1.1 (0.7-1.7) 0.6 (0.3-0.9) 1.4 (0.9-2.1)
p level* 0.80 0.02 0.09

* p level assessed by likelihood ratio test.

Odds ratios were adjusted for:
1 fathers schooling and presence, mothers skin color, family income;
2 fathers schooling and presence, mothers skin color, maternal work;
3 maternal work.

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

 RISK AND PROGNOSTIC FOR DIARRHEAL DISEASE 779

Table 4

Association between environmental variables and diarrhea outcomes.

Variables Dehydrating diarrhea Prognosis of diarrhea Mild diarrhea

Adjusted odds ratio1 Adjusted odds ratio2 Adjusted odds ratio3
(95% CI) (95% CI) (95% CI)

Type of housing
Masonry 1.0 1.0 1.0
Shack 2.3 (1.1-5.0) 1.1 (0.6-1.9) 1.0 (0.6-1.7)
p level* 0.03 0.80 1.00

Water supply
Indoor running water 1.0 1.0 1.0
Outdoor running water (on property) 2.1 (1.1-4.3) 0.9 (0.5-1.6) 1.5 (0.9-2.7)
Outdoor running (public) 1.0 (0.4-2.6) 1.0 (0.4-2.5) 0.8 (0.3-2.1)
Well or river 3.0 (0.7-12.5) 1.8 (0.5-6.0) 3.3 (0.9-12.9)
p level* 0.08 0.70 0.09

Use of refrigerator
Yes 1.0 1.0 1.0
No 1.1 (0.6-2.2) 0.7 (0.4-1.2) 1.8 (1.2-2.9)
p level* 0.70 0.16 0.009

Number of under-fives in household

1-2 1.0 1.0 1.0
3-6 2.4 (1.0-5.6) 1.9 (1.0-3.6) 1.5 (0.8-2.9)
p level* 0.03 0.05 0.20

Home cleanliness
Yes 1.0 1.0 1.0
No 2.9 (1.5-5.6) 2.0 (1.2-3.4) 0.9 (0.6-1.4)
p level* < 0.001 0.007 0.60

* p level assessed by likelihood ratio test.

Odds ratios were adjusted for variables described in Table 2 plus:
1 number of under-fives, type of housing, water supply, home cleanliness;
2 number of under-fives, home cleanliness;
3 use of refrigerator, water supply.

Table 5

Association between maternal reproductive variables and diarrhea outcomes.

Variables Dehydrating diarrhea Prognosis of diarrhea Mild diarrhea

Adjusted odds ratio1 Adjusted odds ratio2 Adjusted odds ratio3
(95% CI) (95% CI) (95% CI)

Maternal age (years)

20 1.0 1.0 1.0
< 20 3.8 (1.6-8.9) 2.4 (1.2-4.6) 1.6 (0.9-3.0)
p level* < 0.001 0.01 0.10

Birth order
1-2 1.0 1.0 1.0
3 2.1 (0.9-4.8) 1.8 (0.9-3.3) 1.7 (0.9-3.1)
>3 1.2 (0.5-2.4) 2.3 (1.2-4.5) 0.8 (0.5-1.4)
p level* 0.20 0.03 0.08

Twin
No 1.0 1.0 1.0
Yes 11.6 (0.9-143) 1.4 (0.4-5.3) 2.0 (0.4-9.2)
p level* 0.03 0.60 0.40

* p level assessed by likelihood ratio test.

Odds ratios were adjusted for variables described in Table 3 plus:
1 twin, maternal age;
2 maternal age, birth order;
3 birth order.

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

780 FUCHS, S. C. & VICTORA, C. G.

Table 6

Association between nutritional variables and diarrhea outcomes.

Variables Dehydrating diarrhea Prognosis of diarrhea Mild diarrhea

Adjusted odds ratio1 Adjusted odds ratio2 Adjusted odds ratio3
(95% CI) (95% CI) (95% CI)

Birth weight (grams) 4

> 3,000 1.0 1.0 1.0
2,9992,500 2.4 (1.2-4.9) 1.3 (0.7-2.3) 1.0 (0.6-1.7)
< 2,500 3.6 (1.5-8.8) 3.2 (1.6-6.3) 0.9 (0.5-1.8)
p level* 0.002 0.002 0.90

Height-for-age (z score)
> -1 1.0 1.0 1.0
-1 to -1.9 1.4 (0.6-3.4) 2.5 (1.2-5.1) 0.9 (0.5-1.5)
-2 4.5 (1.3-15.5) 4.2 (1.8-9.9) 1.5 (0.8-3.0)
p level* 0.04 < 0.001 0.30

Weight-for-age (z score)
> -1 1.0 1.0 1.0
-1 to -1.9 1.5 (0.5-4.2) 1.7 (0.9-3.1) 1.3 (0.7-2.3)
-2 1.9 (0.2-16.9) 2.2 (0.9-5.0) 1.5 (0.7-3.4)
p level* 0.70 0.09 0.50

Type of milk consumed

Breast 1.0 1.0 1.0
Breast + non-breast 1.5 (0.4-5.1) 0.8 (0.3-2.5) 1.2 (0.6-2.3)
Non-breast 6.0 (1.9-14.6) 2.3 (0.9-5.9) 1.2 (0.7-2.2)
p level* < 0.001 0.03 0.80

Breastfeeding history
Still breastfeeding 1.0 1.0 1.0
Stopped breastfeeding 6.4 (2.3-17.3) 2.5 (1.2-5.0) 1.1 (0.7-1.7)
Never breastfed 0.7 (0.1-3.7) 3.8 (0.5-31.0) 0.5 (0.2-1.8)
p level* < 0.001 0.02 0.50

* p level assessed by likelihood ratio test.

Odds ratios were adjusted for variables described in Table 4 plus:
1 birth weight, age of the children, weight-for-age, type of milk, previous hospitalization;
2 birth weight, age of the children, height-for-age, type of milk;
3 age of the children, weight-for-age, type of milk;
4 not adjusted for age of the children.

indoor running water. Families tended to use the ease burden and increased mortality (DSouza,
same sink for washing hands, diapers, clothes, 1997; Hussain & Smith, 1999). Particularly for
and kitchen utensils. In addition, the number diarrheal diseases, most studies have analyzed
of under-fives in the home and home cleanli- risk and prognostic factors without proper con-
ness were associated with the severity but not trol for potential confounding variables (Dar-
with the incidence of diarrhea (Gorter et al., gent-Molina et al., 1994; Teka et al., 1996).
1998; Vasquez et al., 1999). One might expect Adjustment for confounding variables is
environmental factors to be primarily associat- necessary because higher socioeconomic sta-
ed with incidence and rather than severity of tus is known to be associated both with lower
diarrhea. However, children in sub-standard rates of diarrhea and breastfeeding practices
environments may have been subjected to during infancy. Even children selected from the
greater microbial loads and therefore present- same neighborhood had different exposure
ed more severe disease (al-Mazrou et al., 1995). levels, as shown in this study for severity of the
Children of teenage mothers were at greater diarrheal episode. Lack of breastfeeding had a
risk of dehydration, suggesting that these moth- greater effect on risk of dehydration than on
ers were less prepared to deal with the disease prognosis of diarrhea, but unlike other studies
(al-Mazrou et al., 1995). High birth order was it was not associated with mild diarrhea. Since
associated with increased severity of diarrhea, mixed-fed and weaned infants consume greater
and twins were at greater risk of dehydrating amounts of supplemental liquids, the protec-
diarrhea, independently of maternal age. tive effect of breastmilk was shown by compar-
Low birth weight, lack of breastfeeding, and ing total breastfed with non-breastfed children
malnutrition have been associated with dis- (Raisler et al., 1999). Due to the small number

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

 RISK AND PROGNOSTIC FOR DIARRHEAL DISEASE 781

Table 7

Summary of p values for findings on risk and prognostic factors for diarrhea outcomes.

Variables Dehydrating diarrhea Prognosis of diarrhea Mild diarrhea

Age (first year of life) < 0.001 < 0.001 0.05

Age (second year of life) 0.03 0.80 0.60
Gender 0.70 0.90 0.60
Family income 0.04 0.60 0.15
Fathers presence and schooling < 0.001 < 0.001 0.90
Maternal schooling 0.20 0.30 0.20
Mothers skin color 0.02 0.03 0.90
Maternal work 0.80 0.02 0.09
Type of housing 0.03 0.80 1.00
Water supply 0.08 0.70 0.09
Use of refrigerator 0.70 0.16 0.009
Number of under-fives in household 0.03 0.05 0.20
Home cleanliness < 0.001 0.007 0.60
Maternal age < 0.001 0.01 0.10
Birth order 0.20 0.03 0.08
Twin 0.030 0.60 0.40
Birth weight 0.002 0.002 0.90
Height-for-age 0.04 < 0.001 0.30
Weight-for-age 0.70 0.09 0.50
Type of milk consumed < 0.001 0.03 0.80
Breastfeeding history < 0.001 0.02 0.50

of children in some categories, confidence in- der to make valid comparisons about exposure
tervals were wide and included unity (prognos- conditions. To assure the validity of this case-
tic factor). control study, community controls were select-
This study discussed the application of a ed from the same neighborhood as the cases
case-control design to address an issue usually treated in the reference hospitals. These controls
examined by cohort studies. In a cohort study a were expected to have been identified as cases
group of healthy children needs to be identi- if they had developed dehydrating diarrhea.
fied and followed up for a certain period of time However, individual matching led to poten-
to detect diarrheal episodes. Although most tial similarities regarding exposure. For exam-
children have self-limiting diarrhea, a few ple, matching controls to cases according to
progress to dehydration. At this point it would age was particularly relevant to allow the iden-
be theoretically possible to assess the three tification of breastfeeding status and other age-
outcomes: risk factors for mild diarrhea, prog- dependent exposures. Although this approach
nosis of the diarrheal episode, and risk factors is advantageous, overmatching could be a po-
for dehydrating diarrhea. However, since a di- tential problem since case and controls were
arrheal episode mandates the use of oral rehy- likely to share the same environmental condi-
dration therapy, such an approach would be tions, therefore leading some exposures to not
ethically unfeasible. Second, dehydration com- be identified as risk or prognostic factors.
plicates less than 5% of all diarrheal episodes, Finally, this study discussed a unique appli-
and a large number of children would have to cation of a case-control design to assess risk
be followed up in order to obtain a sufficient and prognostic factors simultaneously; an ap-
sample. A case-control design overcame both proach that would usually require a cohort
the potential ethical issue (since children were study. Besides, the introduction of explanatory
selected after the outcome occurred) and logis- variables in the analysis according to a concep-
tic problems related to the sample size. tual framework of determination demonstrat-
Some additional methodological issues need ed that in the same setting, different variables
to be highlighted. Ideally, cases and controls might act as risk, prognostic, and/or confound-
would come from the same population in or- ing factors, depending on the outcomes.

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002

782 FUCHS, S. C. & VICTORA, C. G.

Acknowledgments

This study was supported by the World Health Orga-

nization, Maternal and Child Health Division.

References

al-MAZROU, Y. Y.; KHAN, M. U.; AZIZ, K. M.; FARAG, tritional status of 12-18-month-old children in
M. K. & al-JEFRY, M., 1995. Role of social factors Managua, Nicaragua. Social Science and Medi-
in the prevalence of diarrhoeal diseases in under- cine, 46:403-414.
five Saudi children. Journal of Tropical Pediatrics, MIRZA, N. M.; CAULFIELD, L. E.; BLACK, R. E. &
41(Sup. 1):45-52. MACHARIA, W. M., 1997. Risk factors for diarrheal
AWASTHI, S.; PANDE, V. K. & GLICK, H., 1996. Under duration. American Journal of Epidemiology, 146:
fives mortality in the urban slums of Lucknow. In- 776-785.
dian Journal of Pediatrics, 63:363-368. MURRAY, C. J. & LOPEZ, A. D., 1997. Mortality by
BERN, C.; MARTINES, J.; ZOYSA, I. & GLASS, R. I., cause for eight regions of the world: Global bur-
1992. The magnitude of the global problem of di- den of disease study. Lancet, 349:1269-1276.
arrhoeal disease: A ten-year update. Bulletin of RAISLER, J.; ALEXANDER, C. & OCAMPO, P., 1999.
the World Health Organization, 70:705-714. Breast-feeding and infant illness: A dose-re-
BRATTACHARYA, S. K.; BRATTACHARYA, M. K.; MAN- sponse relationship? American Journal of Public
NA, B.; DUTTA, D.; DEB, A.; DUTTA, P.; GOSWA- Health, 89:25-30.
MI, A. G.; DUTTA, A.; SARKAR, S.; MUKHOPAD- REED, B. A.; HABICHT, J. P. & NIAMEOGO, C., 1996.
HAYA, A.; KRISHNAN, T.; NAIK, T. N. & NAIR, G. The effects of maternal education on child nutri-
B., 1995. Risk factors for development of dehydra- tional status depend on socio-environmental con-
tion in young children with acute watery diar- ditions. International Journal of Epidemiology,
rhoea: A case-control study. Acta Paediatrica, 84: 25:585-592.
160-164. ROTHMAN, K. J. & GREENLAND, S., 1998. Modern
BRESLOW, N. E. & DAY, N. E., 1980. Statistical Meth- Epidemiology. Philadelphia: Lippincott- Raven.
ods in Cancer Research. v. 1. The Analysis of Case- SCARIATI, P. D.; GRUMMER-STRAWN, L. M. & FEIN,
Control Studies. Lyon: International Agency for S. B., 1997. A longitudinal analysis of infant mor-
Research on Cancer. bidity and the extent of breastfeeding in the Unit-
DARGENT-MOLINA, P.; JAMES, A. S.; STROGATZ, D. ed States. Pediatrics, 99:E5.
S. & SAVITZ, D., 1994. Association between ma- TEKA, T.; FARUQUE, A. S. & FUCHS, G. J., 1996. Risk
ternal education and infant diarrhoea in different factors for deaths in under-age-five children at-
households and community environments of Ce- tending a diarrhoea treatment centre. Acta Paedi-
bu, Philippines. Social Science and Medicine, 38: atrica, 85:1070-1075.
343-350. VASQUEZ, M. L.; MOSQUERA, M.; CUEVAS, L. E.;
DESAI, S. & ALVA, S., 1998. Maternal education and GONZALEZ, E. S.; VERAS, I. C.; LUZ, E. O.; BA-
child health: Is there a strong causal relationship? TISTA FILHO, M. & GURGEL, R. Q., 1999. Inci-
Demography, 35:71-81. dncia e fatores de risco de diarria e infeces
DSOUZA, R. M., 1997. Housing and environmental respiratrias agudas em comunidades urbanas
factors and their effects on the health of children de Pernambuco, Brasil. Cadernos de Sade Pbli-
in the slums of Karachi, Pakistan. Journal of ca, 15:163-171.
Biosocial Science, 29:271-281. VICTORA, C. G.; BRYCE, J.; FONTAINE, O. &
GORTER, A. C.; SANDIFORD, P.; PAUW, J.; MORALES, MONASCH, R., 2000. Reducing deaths from diar-
P.; PEREZ, R. M. & ALBERTS, H., 1998. Hygiene rhoea through oral rehydration therapy. Bulletin
behaviour in rural Nicaragua in relation to diar- of the World Health Organization, 78:1246-1255.
rhoea. International Journal of Epidemiology, VICTORA, C. G.; HUTTLY, S. R.; FUCHS, S. C. & OLIN-
27:1090-1100. TO, M. T. A., 1997. The role of conceptual frame-
HOWIE, P. W.; FORSYTH, J. S.; OGSTON, S. A.; CLARK, works in epidemiological analysis: A hierarchical
A. & FLOREY, C., 1990. Protective effect of breast- approach. International Journal of Epidemiology,
feeding against infection. BMJ, 300:11-16. 26:224-247.
HUSSAIN, T. M. & SMITH, J. F., 1999. The relationship YOON, P. W.; BLACK, R. E.; MOULTON, L. H. & BECK-
between maternal work and other socio-eco- ER, S., 1996. Effect of not breastfeeding on the
nomic factors and child health in Bangladesh. risk of diarrheal and respiratory mortality in chil-
Public Health, 113:299-302. dren under 2 years of age in Metro Cebu, The
CLEMENS, J.; ABU-ELYAZEED, R.; RAO, M.; SAVARI- Philippines. American Journal of Epidemiology,
NO, S.; MORSY, B. Z.; KIM, Y.; WIERZBA, T.; NAFI- 143:1142-1148.
CY, A. & LEE, J., 1999. Early initiation of breast-
feeding and the risk of infant diarrhea in rural Submitted on 22 December 2000
Egypt. Pediatrics, 104:E3. Final version resubmitted on 10 September 2001
LAMONTAGNE, J. F.; ENGLE, P. L. & ZEITLIN, M. F., Approved on 28 December 2001
1998. Maternal employment, child care, and nu-

Cad. Sade Pblica, Rio de Janeiro, 18(3):773-782, mai-jun, 2002