DA COSTA LIMA ET AL. RESPOND

We would like to thank London and Promislow for their thoughtful review of our article. They raise the possibility that our results might have been affected by confounding, selection bias, and misclassification.

Regarding confounding, we investigated all variables for which we had information in early life and which, according to the literature, might affect either asthma or respiratory illnesses in general. These variables included family income, maternal education, assets index, number of persons sharing a bedroom, number of other children in the home, maternal age, parental smoking, birthweight, gestational age, intrauterine growth retardation, parity, and type of delivery.

London and Promislow rightly point out that the number of persons sharing a bedroom was not included as a possible covariate in the results presented in Table 3 ▶ of our article. In this table, we adjusted only for socioeconomic variables; however—as stated in the text (p1859)—after further adjustment for the other potential confounders listed above, including both variables related to crowding (number of persons per bedroom and number of children in the home), there were no further changes in the results (Table 1 ▶). Thus, lack of adjustment for variables related to crowding does not explain our results.

TABLE 3—

Prevalence Ratios (PRs) and 95% Confidence Intervals (CIs) for Asthma, by Breastfeeding Status and Monthly Family Income in 1982: Pelotas, Brazil

	Adjusted PR (95%CI)b
Breastfeeding Status, 1983–1986	Income <$150	Income ≥$150	Overall
Feeding pattern at age 3 mo
Exclusive	1.26 (0.86, 1.85)	1.25 (0.99, 1.51)	1.27 (0.97, 1.66)
Partial	1.46 (1.02, 2.09)	1.18 (0.95, 1.47)	1.27 (0.97, 1.66)
Not Breastfed	1.00	1.00	1.00
P a	.11	.13	.12
Type of milk received at age 3 mo
Breast	1.37 (0.97, 1.95)	1.22 (1.00, 1.50)	1.28 (0.99, 1.65)
Mixed	1.33 (0.89, 1.99)	1.17 (0.93, 1.47)	1.25 (0.93, 1.66)
Non-breast	1.00	1.00	1.00
P a	.16	.13	.13
Type of milk received at age 6 mo
Breast	1.21 (0.78, 1.87)	1.26 (1.00, 1.58)	1.29 (0.98, 1.71)
Mixed	1.60 (1.08, 2.37)	1.19 (0.91, 1.54)	1.33(0.96, 1.83)
Non-breast	1.00	1.00	1.00
P a	.06	.09	.08
Type of milk received at age 9 mo
Breast	1.22 (0.74, 2.02)	1.34 (1.02, 1.75)	1.43 (1.04, 1.97)
Mixed	1.48 (0.95, 2.31)	1.41 (1.08, 1.83)	1.42 (1.01, 2.00)
Non-breast	1.00	1.00	1.00
P a	.20	.009	.02
Type of milk received at age 12 mo
Breast	1.19 (0.67, 2.12)	1.26 (0.90, 1.77)	1.35 (0.90, 2.03)
Mixed	1.28 (0.79, 2.09)	1.39 (1.06, 1.82)	1.44 (1.01, 2.03)
Non-breast	1.00	1.00	1.00
P a	.55	.03	.05

^aχ² test for trend.

^bAdjusted for family income, maternal education, assets index, number of persons sharing a bedroom, number of other children in the home, maternal age, parental smoking, birthweight, gestational age, intrauterine growth retardation, parity, and type of delivery.

TABLE 1—

Prevalence Ratios (PRs) and 95% Confidence Intervals (CIs) for Asthma, by Breastfeeding Status: Pelotas, Brazil

Breastfeeding Status, 1983–1986	Crude PR (95%CI)	Adjusted PRa (95%CI)	Adjusted PRb (95%CI)
Feeding pattern at age 3 mo
Exclusive	1.26 (1.02, 1.54)	1.22 (0.99, 1.51)	1.27 (0.97, 1.66)
Partial	1.20 (0.97, 1.48)	1.18 (0.95, 1.47)	1.27 (0.97, 1.66)
Not breastfed	1.00	1.00	1.00
P c	.06	.13	.12
Type of milk received at age 3 mo
Breast	1.25 (1.03, 1.53)	1.22 (1.00, 1.50)	1.28 (0.99, 1.65)
Mixed	1.19 (0.94, 1.49)	1.17 (0.93, 1.47)	1.25 (0.93, 1.66)
Non-breast	1.00	1.00	1.00
P c	.06	.13	.13
Type of milk received at age 6 mo
Breast	1.27 (1.02, 1.58)	1.26 (1.00, 1.58)	1.29 (0.98, 1.71)
Mixed	1.20 (0.92, 1.55)	1.19 (0.91, 1.54)	1.33 (0.96, 1.83)
Non-breast	1.00	1.00	1.00
P c	.07	.09	.08
Type of milk received at age 9 mo
Breast	1.30 (0.99, 1.71)	1.34 (1.02, 1.75)	1.43 (1.04, 1.97)
Mixed	1.41 (1.08, 1.83)	1.41 (1.08, 1.83)	1.42 (1.01, 2.00)
Non-breast	1.00	1.00	1.00
P c	.01	.009	.02
Type of milk received at age 12 mo
Breast	1.18 (0.84, 1.65)	1.26 (0.90, 1.77)	1.35 (0.90, 2.03)
Mixed	1.36 (1.04, 1.79)	1.39 (1.06, 1.82)	1.44 (1.01, 2.03)
Non-breast	1.00	1.00	1.00
P c	.06	.03	.05

^a Adjusted for family income, maternal education, and assets index.

^b Adjusted for family income, maternal education, assets index, number of persons sharing a bedroom, number of other children in the home, maternal age, parental smoking, birthweight, gestational age, intrauterine growth retardation, parity, and type of delivery.

^Cχ² test for trend.

The writers mention the possibility of selective loss of asthma-prone children (e.g., those with childhood lung diseases) and ask us to show follow-up rates for different subgroups of children. In Table 2 ▶ here, we show the follow-up rates at age 18 years according to family income at birth and reported history of asthma/bronchitis or pneumonia in the first 2 years of life (this information was collected in 1984). The table also shows follow-up rates according to hospital admissions in the first 4 years of life. There is no evidence that morbidity in early life was associated with follow-up rates. For a general discussion of follow-up rates in the study we refer to Victora et al.1 More than 70% of subjects in each socioeconomic category were traced in 2000.

TABLE 2—

Characteristics of the Original Cohort (n = 3037) and Percentage Located in 2000: Pelotas, Brazil

	Original Cohort, No.	Deceased, No.	Interviewed, No.	Located in 2000, %a
Family income, $/mo (1982)
<50	666	63	420	72.5
50–149	1463	65	1105	80.0
150–299	544	8	448	83.8
≥300	351	5	269	78.1
Asthma or bronchitis (1984)
No	2041	38	1656	83.0
Yes	513	11	414	82.8
Pneumonia (1984)
No	2084	34	1701	83.3
Yes	467	14	369	82.0
Hospital admissions (1982–1986)
No	1791	21	1466	83.0
Yes	910	40	699	81.2

^aIncludes subjects interviewed as well as those known to have died.

As suggested, we investigated effect modification by income, after adjustment for the confounding variables listed in Table 1 ▶. For low-income families (up to 3 times minimum wage per month), mixed feeding at 9 months was associated with a prevalence ratio (PR) for asthma of 1.48 (95% confidence interval [CI] = 0.95, 2.31) and breastfeeding was associated with a PR of 1.22 (95% CI = 0.74, 2.02), relative to children who did not receive any breast milk (Table 3 ▶). For upper-income families the corresponding PRs were 1.41 (95% CI = 1.08, 1.83) and 1.34 (95% CI = 1.02, 1.75). Therefore there was no apparent effect modification.

London and Promislow suggest that some mothers may delay weaning because of early respiratory infection, and they ask that we present information on early wheezing and asthma according to feeding patterns. In fact, our data suggest the opposite. Infants who at 9 months of age received breastfeeding (PR = 0.80; 95% CI = 0.53, 1.23) or mixed feeding (PR = 0.55; 95% CI = 0.35, 0.87) were less likely to have reported asthma or wheezing at the age of 2 years than those who did not receive any breast milk. It should be noted, however, that reported wheezing or asthma at this age is often due to infectious rather than atopic conditions. The literature, as reviewed by Sears,2 shows that breastfeed-ing tends to protect against wheezing conditions in early life, but not later on.

Finally, we share the concern expressed by the writers about the public health message of an article suggesting that breastfeeding may have some detrimental effects, despite the wealth of literature showing its benefits. However, our article was not the first to show such an association. Studies of the 1958 and 1970 United Kingdom birth cohorts3,4 and studies in Arizona,5 Italy,6 and, more recently, New Zealand2 all show increased risk of atopy, asthma, or both among breastfed children.