- Email: [email protected]
Periconceptional Multivitamin Use and Infant Birth Weight Disparities
Periconceptional Multivitamin Use and Infant Birth Weight Disparities HEATHER H. BURRIS, MD, MPH, ALLEN A. MITCHELL, MD, AND MARTHA M. WERLER, SCD
PURPOSE: In the United States, African American women deliver preterm and low birth weight infants two to three times more frequently than their white counterparts. Our objective was to determine whether maternal periconceptional multivitamin (MVI) use is associated with this disparity. METHODS: As a secondary analysis of previously collected data from mothers of non-malformed infants from the Slone Epidemiology Center Birth Defects Study, we conducted a retrospective cohort study of 2331 non-Hispanic white and 133 non-Hispanic black mother/infant pairs from 1998 through 2007. To estimate the effect of MVI use on birth outcomes, linear regression models were used. RESULTS: In white subjects, MVI use was not associated with birth weight, gestational age, or weightfor-gestational-age. However, in black subjects, MVI use was associated with a 536-gram increased birth weight (p Z 0.001). Black MVI users also had longer gestations (although not statistically significant). When birth weights were adjusted for gestational age using z scores, MVI use was associated with increased fetal growth in black infants (þ0.86 z score units, 95% confidence interval: 0.35–1.36). CONCLUSIONS: The present findings suggest MVI use may improve fetal growth and possibly gestational age in the offspring of African American women. Ann Epidemiol 2010;20:233–240. Ó 2010 Elsevier Inc. All rights reserved. KEY WORDS:
Pregnancy, Preconception Care, Birth Weight, Vitamins, Premature Birth, Ethnic Groups, African Continental Ancestry Group.
INTRODUCTION African American women in the United States deliver preterm and low birth weight infants two to three times more frequently than their white counterparts (1). Studies of socioeconomic, genetic, and infectious factors to date have failed to entirely account for these disparities (2–8). Furthermore, racial disparities in these outcomes remain among women with equal access to health care and with college educations (9, 10). Discovering the etiology of racial disparities in birth outcomes and possible modifiable risk factors is crucial to resolving this immense public health problem. Recent evidence suggests that the processes of implantation and placentation are related to pregnancy outcomes, shifting the focus of potential risk factors to the periconceptional period (11–13). A study from the Netherlands recently found that folic acid supplementation during the preconceptional and periconceptional periods was associated with a decreased risk of low birth weight and smallfor-gestational-age (SGA) births (14). Another recent
From the Department of Neonatology, Beth Israel Deaconess Medical Center, Harvard University (H.H.B.), and Slone Epidemiology Center at Boston University (A.A.M, M.M.W.). Address correspondence to: Heather H. Burris, MD, MPH, Beth Israel Deaconess Medical Center, Neonatology–RO 318, 330 Brookline Ave., Boston, MA 02215. Tel: (617) 667-3276; Fax: (617) 667-7040. E-mail: [email protected]. Received July 29, 2009; accepted December 7, 2009. Ó 2010 Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010
observational study from Pittsburgh found that periconceptional vitamin use was associated with decreased rates of preterm and SGA births (15). However, in this study, changes in birth weight-for-gestational-age were not reported and race/ethnicity differences were not found. In our study, we examined the relationships between periconceptional multivitamin (MVI) use and birth weight, gestational age, and weight-for-gestational age as a measure of fetal growth. We also considered effect differences according to race/ ethnicity. Such differences could have important implications. If we found that MVI use improved neonatal outcomes among black women, then the documented lower rates of MVI use in black than in white women (16) may represent a targeted area for a public health intervention.
METHODS Study Population and Design The Slone Epidemiology Center Birth Defects Study (17, 18) includes a random sample of Massachusetts births who serve as control subjects. Mothers of malformed infants (cases) and controls are interviewed by telephone within 6 months of delivery about pregnancy events and exposures. For the present secondary analysis, control subjects interviewed between 1998 and 2007 were used to conduct a population-based retrospective cohort study. Of the eligible mothers, 80% were successfully contacted and 76% agreed to participate (60% overall participation), yielding a final 1047-2797/10/$–see front matter doi:10.1016/j.annepidem.2009.12.003
234
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
Selected Abbreviations and Acronyms MVI Z periconceptional multivitamin BMI Z body mass index LMP Z last menstrual period SGA Z small-for-gestational-age LGA Z large-for-gestational-age
sample size of 3,047 mother-infant pairs. This analysis was restricted to non-Hispanic black and white subjects; there were 2,374 white and 135 black mother-infant pairs. The Boston University institutional review board approved this study, and verbal consent was obtained before the start of the interview. Periconceptional Multivitamin Use A nurse-interviewer contacted mothers via telephone and conducted a standardized interview on demographic, reproductive, medical, dietary, and birth histories. As part of their responses to detailed questions about medication use, mothers provided the names, start and stop dates, and frequency of use of medications and vitamins, including MVIs. For this study, a supplement was considered an MVI if it contained at least two water-soluble and two fat-soluble vitamins. We defined the periconceptional period as 28 days prior to the last menstrual period (LMP) to 28 days after the LMP (lunar months –1 to 1). We categorized women as MVI users if they took an MVI four or more times per week in the periconceptional period. We also considered exposure to MVI in the second lunar month (LM2) of pregnancy. We categorized women as LM2 initiators of MVI if their use was at least four times per week. Measurement of Birth Weight, Gestational Age, and Fetal Growth Birth weight reported by subjects in pounds and ounces was converted to grams. Gestational age was determined by comparing the birth date to the reported due date. The onset of gestation was calculated as the due date – 280 days. Estimates of fetal growth or ‘‘weight-for-gestational-age’’ were based on the median birth weights for each completed week of gestation for a U.S. natality data set of more than six million neonates born in 1999 and 2000 (19). Weights for gestational age were converted to z scores where each z score unit represented the distance of the birth weight from the mean for any given gestational age (20). In this way, the effect of gestational age on birth weight was separated from fetal growth. Measurement of Covariates Covariates entered into the multivariable linear regression models included known risk factors for growth restriction
AEP Vol. 20, No. 3 March 2010: 233–240
and preterm birth. Because maternal age and pre-pregnancy body mass index (BMI) may not be linearly associated with outcomes, terms representing categories were included. Maternal age and BMI were each divided into three categories: !20, 20–34, and O34 years and !19.5, 19.5–25, and O25 kg/m2, respectively. Models also included terms for maternal education (!high school graduation vs. Ohigh school graduation), annual household income (!$45, 000 vs. more), cigarette smoking during pregnancy (vs. not), unmarried status (vs. married), and primiparity, defined as no prior viable pregnancies, (vs. multiparity). Self-reported maternal race/ethnicity was categorized as non-Hispanic black or non-Hispanic white. Women whoself-reported other race/ethnicities were excluded because of small numbers as well as the recognition of the long-standing black/white disparity in birth outcomes. The Willett Semiquantitative Food Frequency Questionnaire was used to ascertain dietary folic acid, iron, and caloric intakes (21). Intakes of each were categorized into quartiles. Analysis Bivariate analyses (using chi-square tests) were performed to determine whether the proportions of MVI use differed between the two race/ethnicity groups. Within each race/ ethnicity category, bivariate analyses using chi-square or Fisher’s exact tests were performed for white and black subjects separately to determine the characteristics of MVI users versus non-users in this cohort. With two-sided p values, student t tests were performed to obtain unadjusted associations of MVI use with birth weight, gestational age, and weight-for-gestational-age within the two strata. A t test was also performed to compare the mean birth weights in the two race/ethnicity categories. Stratified by race/ethnicity, multivariable linear regression models included terms for MVI use and the previously mentioned covariates. In addition, combined models including black and white subjects and interaction terms for MVI use and race/ethnicity were performed to formally test for effect measure modification of MVI use by race/ethnicity and its association with birth outcomes. To assess how our findings might have been influenced by incomplete participation among eligible mother-infant pairs, we performed sensitivity analyses in two ways. First, we assumed that black non-responders and black responders had similar MVI usage rates. We then assigned the infants of the non-responders the mean birth weight of the black responders’ exposed infants. Our second sensitivity analysis considered a different set of conditions: that black nonresponders had a higher rate of MVI use (50%) than black responders. All of the infants of black non-responders were assigned the mean birth weight of unexposed black infants. In these scenarios, we considered the most extreme examples
AEP Vol. 20, No. 3 March 2010: 233–240
of selection bias by assuming no exposure-outcome relationship among the non-responders. Lastly, to confirm that our findings were not secondary to an abnormal birth weight distribution (e.g., a disproportionate number of large-for-gestational-age [LGA] infants), we analyzed the data with and without infants at the tails of the distribution. All analyses were performed using the SAS version 9.1 (Cary, NC).
RESULTS Characteristics of Periconceptional Multivitamin Users The cohort for this analysis comprised 2,374 non-Hispanic white and 135 non-Hispanic black mother-infant pairs. Among women who reported any MVI use, all but 29 white women took MVIs at least four times per week. The less frequent supplementers were therefore excluded from subsequent analyses. Among white women considered exposed (at least four times per week), 1,192 reported daily use, 89 reported use four to six times per week, and two reported ‘‘regular’’ use. Among the 34 exposed black women, 32 reported daily use and two reported use four to six times per week. There were 720 white women and 49 black women who started MVI use in the second lunar month after their last menstrual period (LM2). Among them, 14 white and two black women reported initiating MVI use in LM2, but less than four times per week and were excluded from the cohort. The final cohort included 2,331 white and 133 black mother-infant pairs. MVI use at least four times per week in the periconceptional period (lunar months –1 to 1) was reported by 55.0% of white women (1,283/2,331) and by 25.6% of black women (34/133) (p ! 0.001). Among white women, users tended to be non-smokers, married, older, wealthier, and better educated. Patterns were similar for black women (Table 1).
Race/Ethnicity Disparities of Birth Outcomes The mean birth weight for black infants was 215 grams less than for white infants (3,291 vs. 3,506 respectively; p ! 0.001). Black infants’ average gestational age was 3.0 days less than white infants’ (273.3 vs. 276.3, respectively; p Z 0.03). Black infants’ weight-for-gestational-age was an average of 0.35 z score units less than for white infants (p ! 0.001) (mean z scores for black and white infants were –0.13 and 0.22, respectively). For white infants, birth weights (Table 2) and weight-for-gestational-age z-scores (data not shown) were lower in infants born to women who were smokers, unmarried, younger than 20 years, and primiparous. Associations were similar for black women, although they did not reach statistical significance. Neither
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
235
mean birth weight nor mean gestational age was statistically associated with mothers’ initiation of MVI use in LM2, although birth weights were lower in black infants of LM2 initiators. LM2 initiators were excluded from analyses comparing MVI users and non-users. Periconceptional Multivitamin Use and Associations with Birth Outcomes Among white women, there were no appreciable differences in birth weight (Fig. 1), gestational age (Fig. 2), or weight-forgestational-age (z score) (Fig 3) between users and non-users of MVI. Adjustment for the covariates listed in Table 1 made no substantial changes to these findings. However, among black mother-infant pairs, MVI users delivered substantially larger infants; after adjustment for multiple covariates, the difference persisted, with black MVI-exposed infants weighing an average of 536 g more than black unexposed infants (p Z 0.002) (see Fig. 1). Furthermore, weight-for-gestational age was substantially increased among black women who took MVIs; exposed black infants had z scores 0.86 greater than unexposed black infants (p Z 0.001) (see Fig. 3). Gestational age differed by 1.5 days between the two groups of black women (p Z 0.71) (see Fig. 2). Combined models with race/ ethnicity and MVI use included as interaction terms revealed that the interaction of non-Hispanic black race/ethnicity with MVI use was highly significantly associated with higher birth weights (p ! 0.0001) and increased weight for gestational age z scores (p Z 0.0004). The interaction was not significant for gestational age (p Z 0.3). Sensitivity Analysis to Account for Selection Bias We conducted sensitivity analyses to determine whether selection bias due to incomplete participation might account for our findings among black women. First, we assumed that the black women who did not respond to the survey had approximately the same frequency of MVI use as black responders (25%). We then assumed that the infants of black non-responders had the same mean birth weight (3,598 grams) as black MVI–exposed infants. After this adjustment, the unexposed infants’ mean birth weight increased from 3,165 to 3,412 grams, but a substantial difference of 186 grams (3598–3412 g) remained. Our second analysis considered a different set of conditions. We assumed that MVI users were underrepresented in the study and assigned a 50% rate of MVI use among non-responding black women. (This is twice the rate among black responders and closer to the 55% observed among white responders.) We assigned the mean birth weight of the unexposed black infants in the study (3,165 g) to all of the infants of the black non-responders. After this adjustment, the mean birth weight for exposed infants decreased to 3,353 grams from 3,598 grams whereas
236
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
AEP Vol. 20, No. 3 March 2010: 233–240
TABLE 1. Characteristics of mothers, by race/ethnicity, and periconceptional vitamin use, Slone Epidemiology Center Birth Defects Study Control Subjects, 1998–2007 NH white mothers (n Z 1,611)
Smoked during pregnancy* Yes No Educationz,x !High school High school or more Maternal age !20 yr 20 to !35 yr >35 yr Maternal pre-pregnancy BMI (kg/m2)k,{ <19.5 19.5 to !25 O25 Marital status Unmarried Married Parity Primiparous Multiparous Annual household income#,** !$45,000 >$45,000 Dietary intakeyy Folic acid intake (mg/d) Iron intake (mg/d) Total daily calorie intake
NH black mothers (n Z 84)
Peri-MVI use n Z 1,283 (%)
No peri-MVI use or LM2 MVI use n Z 328 (%)
73 (5.7)y 1209 (94.2)y
46 (14.1)y 281 (85.9)y
2 (5.9) 32 (94.1)
5 (10.0) 45 (90.0)
18 (1.5)y 1227 (98.6)y
24 (7.3)y 303 (92.7)y
2 (6.3) 30 (93.8)
8 (16.0) 42 (84.0)
20 (1.6)y 946 (73.7)y 317 (24.7)y
42 (12.8)y 234 (71.3)y 52 (15.9)y
5 (14.7) 20 (58.8) 9 (26.5)
6 (12.0) 37 (74.0) 7 (14.0)
126 (9.9) 778 (61.0) 372 (29.2)
41 (12.6) 187 (57.4) 98 (30.1)
2 (5.9) 16 (47.1) 16 (47.1)
5 (10.4) 20 (41.7) 23 (47.9)
121 (9.4)y 1162 (90.6)y
131 (39.9)y 197 (60.1)y
13 (38.2) 21 (61.8)
24 (48.0) 26 (52.0)
576 (44.9) 707 (55.1)
123 (37.5) 205 (62.5)
15 (44.1) 19 (55.9)
14 (28.0) 36 (72.0)
177 (14.2)y 1071 (85.8)y
130 (43.8)y 167 (56.2)y
13 (41.9) 18 (58.1)
26 (65.0) 14 (35.0)
391 (156)zz 12.9 (5.6)zz 1528 (479)
350 (167)zz 11.7 (5.5)zz 1553 (576)
401 (162) 12.7 (6.0) 1594 (700)
405 (188) 13.1 (5.3) 1679 (565)
Peri-MVI use n Z 34 (%)
No peri-MVI use or LM2 MVI use n Z 50 (%)
NH Z non-Hispanic; Peri Z periconceptional; MVI Z multivitamin use; LM2 Z second lunar month since last menstrual period; BMI Z body mass index. NOTE: Dietary intake data expressed as means (standard deviations). Periconceptional use defined as >4 times per week. *Two white women had missing smoking data. y p ! 0.05; chi-square test or Fisher’s exact test within race/ethnicity subgroups. z Thirty-nine white women had missing education data. x Two black women had missing education data. k Nine white women had missing BMI data. { Two black women had missing BMI data. # Sixty-six white women had missing income data. **Thirteen black women had missing income data. yy Twelve black women and 188 white women had missing dietary intake data. zz p ! 0.05; t test within race/ethnicity subgroups.
that for the unexposed infants remained at 3,165 grams. Even under these extreme assumptions, a 188-gram difference (3,417 g – 3,165 g) persisted between exposed and unexposed infants. We also considered the possibility that the MVI effect in black subjects was actually due to an increased number of large-for-gestational-age (LGA) births (O90th percentile). Indeed, the frequency of LGA was unusually high among our black MVI-exposed infants; nine of the 34 exposed infants were LGA., compared with three of the 50 MVI-unexposed infants. After eliminating LGA infants from the sample, we found that weight-for-gestational-age was still increased among infants exposed to MVIs; exposed black infants had
average z scores of 0.40 (95% CI: –0.06 to 0.86) greater than non-exposed infants. DISCUSSION In this study of 2,464 mother-infant pairs, MVI use was not associated with indices of fetal growth among non-Hispanic white subjects, but did appear to be associated with increased birth weight and fetal growth in non-Hispanic black subjects. The positive associations with MVI use and both birth weight and gestational age in black subjects raised the question as to whether the birth weight finding might be due to longer gestations and not increased fetal growth.
AEP Vol. 20, No. 3 March 2010: 233–240
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
237
TABLE 2. Mean birth weight and gestational age by maternal characteristics, stratified by race/ethnicity, Slone Epidemiology Center Birth Defects Study Control Participants, 1998–2007 NH white infants (n Z 2331)
All subjects Maternal characteristic Smoked during pregnancyy Yes No Educationx,k !High school High school or more Maternal age !20 yr >20 to !35 yr (ref) >35 yr Maternal pre-pregnancy BMI{,# (kg/m2) <19.5 19.5 to !25 (ref) >25 Marital status Unmarried Married Parity Primiparous Multiparous Annual household income**,yy !$45,000 >$45,000 MVI use (>4 times per week) In LM –1 or LM1 Initiation in LM2 No use in LM –1, LM1, or LM2 (ref) Folic acid daily intake,zz mg (quartiles) !272 (ref) 272–354 355–452 O452 Iron daily intake, zz mg (quartiles) !8.5 (ref) 8.5 to 11.5 11.6 to 15.4 O15.4
NH black infants (n Z 133)
Birth weight, grams (SD)
Gestational age, days (SD)
Birth weight, grams (SD)
Gestational age, days (SD)
3506 (522)*
276.3 (11.3)*
3291 (706)*
273.3 (16.9)*
3277 (508)z 3531 (519)
275.3 (12.4) 276.4 (11.2)
2967 (607) 3315 (709)
271.9 (17.2) 273.4 (16.9)
3372 (572)z 3511 (522)
277.6 (10.8) 276.3 (11.4)
2959 (1058) 3340 (651)
261.3 (37.6) 274.8 (12.6)
3335 (497)z 3508 (519) 3537 (534)
277.2 (11.2) 276.5 (11.0) 275.4 (12.5)
3103 (776) 3336 (665) 3274 (812)
272.7 (27.7) 274.3 (12.9) 269.1 (19.4)
3359 (497)z 3486 (487) 3591 (578)z
275.8 (11.5) 276.7 (10.3) 276.0 (12.2)
3382 (583) 3297 (780) 3301 (621)
277.5 (5.1) 272.2 (19.3) 274.5 (13.8)
3450 (516)z 3520 (523)
276.9 (12.2) 276.2 (11.1)
3240 (610) 3348 (800)
275.6 (16.5) 270.7 (17.1)
3424 (520)z 3569 (515)
277.1 (11.8)z 275.7 (10.9)
3143 (683) 3375 (709)
273.3 (19.9) 273.3 (15.1)
3486 (548) 3525 (510)
276.7 (12.8) 276.3 (10.8)
3380 (547) 3252 (673)
276.9 (8.4)z 272.1 (15.2)
3512 (503) 3495 (535) 3506 (566)
276.0 (11.0) 276.6 (11.3) 277.0 (12.6)
3598 (768)z 3207 (643) 3165 (673)
274.6 (9.5) 273.5 (14.1) 272.1 (22.6)
3451 (515) 3539 (482)z 3513 (523) 3511 (506)
275.6 (12.5) 277.5 (9.8)z 275.9 (10.8) 276.5 (11.5)
3295 (375) 3343 (580) 3218 (637) 3444 (592)
276.0 (6.8) 276.4 (11.1) 272.6 (11.9) 277.2 (8.5)
3454 (485) 3535 (518)z 3466 (516) 3562 (504)z
276.1 (11.7) 276.6 (10.9) 276.1 (10.8) 276.7 (11.3)
3273 (514) 3419 (502) 3239 (573) 3394 (593)
275.2 (9.8) 277.6 (8.5) 273.4 (11.1) 276.5 (8.6)
NH Z non-Hispanic; SD Z standard deviation; ref Z reverence group; BMI Z body mass index; MVI Z multivitamin; LM –1 Z lunar month prior to last menstrual period; LM1 Z first lunar month after last menstrual period; LM2 Z second lunar month since last menstrual period. *p ! 0.05; t test comparing between race/ethnic groups. y Two white women had missing smoking data. z p ! 0.05; t test compared with reference group within race/ethnic group. x Thirty-nine white women had missing education data. k Two black women had missing education data. { Nine white women had missing BMI data. # Two black women had missing BMI data. **Sixty-six white women had missing income data. yy Thirteen black women had missing income data. zz Twelve black women and 188 white women had missing dietary intake data.
However, the effect on birth weight remained after weightfor-gestational-age z scores were considered. Our finding of no effect in white women differs from results reported for women in the Netherlands and Pittsburgh (14, 15). The Pittsburgh study (15) found a beneficial effect of periconceptional vitamin use on preterm delivery
and fetal growth when race was controlled as a confounder, and the authors stated that there was no evidence of effect modification by race. Their rate of MVI use was similar to that of white women but higher than that of black women in our study. The definition of the periconceptional period for their study was much broader than ours, covering the
238
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
AEP Vol. 20, No. 3 March 2010: 233–240
b
900 800 700 600 Difference in 500 Birth Weight 400 300 (grams) 200 100 0 -100
Black Unadjusted Adjusted †
b
Black Unadjusted Adjusted †
10 8
412
536
White Unadjusted Adjusted †
13
-13
FIGURE 1. Periconceptional MVIs and birth weight differences. Vertical lines represent 95% confidence intervals; yAdjusted for smoking, income, education, age, body mass index, parity, marital status, folic acid intake, iron intake and caloric intake.
6 months leading up to their initial interview (average 9.9 weeks’ gestation), but did not include supplementation that began after pregnancy was detected. In the Dutch study (14) of primarily white women, the rate of supplementation was lower than in the white subgroup of our study population and was positively associated with fetal growth. The Dutch investigators defined the periconceptional period differently. They compared women with no MVI use at all before and during pregnancy with women who started MVI use before conception and women who started after the recognition of pregnancy but before the eighth week of gestation. In contrast, we defined the periconceptional period narrowly, as the month prior to the last menstrual period and the month after the menstrual period, thus capturing the moment of conception. In addition, it is possible that the white subjects in our study had higher levels of micronutrient(s) than the subjects in the other studies, thus exceeding a threshold below which MVIs could have an effect. Although there were no material differences in total iron or folic acid intake between white and black women, in our cohort we detected a significant difference in total caloric intake, with black women consuming an average of 146 more calories per day than white women (p Z 0.02). Thus, overall dietary patterns, and potentially quality, may differ between white and black women in our study, accounting for the MVI effect in black, but not in white women. We considered several potential limitations that could have affected our findings. Because our study was retrospective, we were concerned about recall error. Women were asked to report their MVI intake that occurred approximately 1 year earlier. We think that recall error in MVI reporting was minimal since the rates reported in our study are consistent with prospectively collected data from other studies (16, 22). Furthermore, while the collection of neonatal data was also retrospective, it was more recent (within 6 months), and maternal recall of birth weight has been demonstrated to be reliable years after delivery (23). For gestational age, we were reassured about accuracy
6 4
Gestational 2 Age 0 Difference -2 (days) -4
White Unadjusted Adjusted †
-0.8
1.8
1.5
-1.1
-6 -8
FIGURE 2. Periconceptional MVIs and gestational age differences. Vertical lines represent 95% confidence intervals; yAdjusted for smoking, income, education, age, body mass index, parity, marital status, folic acid intake, iron intake and caloric intake.
because due dates were used to determine gestational age and the onset of pregnancy. Indeed, more than 99% of women in our Birth Defects Study provided a due date. Furthermore, recall of due dates has been shown to be accurate (24), and it represents clinicians’ best estimate of gestational timing. Recall bias might also affect our findings, where the accuracy of maternal reporting of MVI use varies according to infants’ birth weights or gestational ages. We think this is unlikely because we would not have expected recall bias to differ between two race/ethnicity subgroups. Furthermore, mothers were not aware of whether they were cases or controls when they were asked to participate in the study. However, it remains possible that women who had smaller babies could have differentially recalled their MVI use. Participation bias might also affect our findings because approximately 40% of eligible subjects did not participate. Furthermore, some characteristics of the study sample differed from coincident Massachusetts births. For example, the proportion of women who considered themselves to be non-Hispanic white was 78% in the study sample and 68% in statewide data for 2006 (25). Despite these differences, when we considered extreme theoretical scenarios of selection bias, birth weights among non-Hispanic black infants remained higher in the MVI-exposed infants compared with the unexposed. While we cannot rule out the possibility that our findings are confounded by unmeasured dietary, lifestyle, and other demographic factors, we believe it unlikely that there would be differential unmeasured confounding between race/ ethnicity subgroups. However, the possibility remains that MVI use is simply a proxy for a healthier lifestyle and that this lifestyle could differ by race/ethnicity. Our findings are consistent with a plausible role played by micronutrients in fetal growth. It is not known which nutrient or combination of nutrients in MVIs might affect
AEP Vol. 20, No. 3 March 2010: 233–240
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
b
Black Unadjusted Adjusted †
1.5 Weight for Gestational Age Differences (Z score)
1 0.5
White Unadjusted Adjusted †
0.65
0.86
0 0.04
0.02
-0.5
FIGURE 3. Periconceptional MVIs and fetal growth differences. Vertical lines represent 95% confidence intervals; yAdjusted for smoking, income, education, age, body mass index, parity, marital status, folic acid intake, iron intake and caloric intake.
fetal growth, and our data do not contribute to this question. As other authors have suggested, micronutrients may affect implantation, placentation, and adequate vascular development of the placenta (15). Many micronutrients may play a role, and differential micronutrient status has been observed between black and white populations. For example, vitamin D levels are lower in black subjects because of less efficient sunlight absorption and vitamin D synthesis by darker pigmented skin as well as decreased intake of vitamin D–fortified milk (26). Whether vitamin D status affects fetal growth remains unknown. Folic acid is another micronutrient that might account for our findings. Recently a secondary analysis of a large cohort study found that preconceptional folic acid supplementation was associated with a decreased risk of early preterm birth (27). This is consistent with an earlier smaller study in New Jersey (28). Furthermore, there is evidence that, on average, black women consume less folic acid (29) and have lower circulating red blood cell folate levels even in the post-dietary fortification period (30). Without knowing which micronutrient may be responsible for our results, our findings raise the question as to whether the threshold for micronutrient deficiencies differs by disease state. The difference between clinical folate deficiency and disease-specific folate deficiency for neural tube defects is an example of one such discrepancy. To prevent megaloblastic anemia (the classic presentation of folate deficiency), one requires only 50–200 mg per day of folic acid (31). However, this level of intake among pregnant women is insufficient to prevent neural tube defects in the fetus (32). Public health recommendations recognize the different thresholds for folate deficiency for the general population and pregnant women (33–37). Furthermore, folic acid has been shown to decrease the risk of other anomalies in addition to neural tube defects including urinary anomalies and cardiac anomalies (38). The same phenomenon may apply for the micronutrient(s) in MVIs and the birth outcomes in our study.
239
In summary, we found that among non-Hispanic black women, those who took an MVI delivered infants with increased birth weights and fetal growth and possibly increased durations of gestation. Similar effects were not observed among non-Hispanic white women. These results are based on secondary analyses of an existing data set and therefore need to be replicated. If our findings were confirmed and subsequently shown to be causal, then increasing periconceptional MVI use among African American women in the United States could help to eliminate longstanding disparities in birth weight, gestational age, and fetal growth. We thank Katherine Ahrens, MPH, Dawn Jacobs, RN, MPH, Fiona Rice, MPH, Rita Krolak, RN, Kathleen Sheehan, RN, Claire Coughlin, RN, Moira Quinn, RN, Nancy Rodriguez, Carolina Tejedor Meyers, and Nastia Dynkin, for their assistance in data collection and computer programming; and the staff of the Massachusetts Department of Public Health Center for Birth Defects Research and Prevention. We also thank all the mothers who participated in the study. Dr. Burris thanks the Harvard Pediatric Health Services Research Fellowship and its funders, including the Agency for Healthcare and Quality. The data for this project are held at the Slone Epidemiology Center at Boston University. Dr. Burris from Beth Israel Deaconess Medical Center led the project.
REFERENCES 1. Behrman R, Butler A. Preterm birth: Causes, consequences, and prevention. Washington (DC): National Academies Press; 2007. 2. David RJ, Collins JW Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks, and U.S.-born whites. N Engl J Med. 1997;337:1209–1214. 3. Wang H, Parry S, Macones G, Sammel MD, Kuivaniemi H, Tromp G, et al. A functional SNP in the promoter of the SERPINH1 gene increases risk of preterm premature rupture of membranes in African Americans. Proc Natl Acad Sci U S A. 2006;103:13463–13467. 4. Collins JW Jr, Wambach J, David RJ, Rankin KM. Women’s lifelong exposure to neighborhood poverty and low birth weight: a population-based study. Matern Child Health J. 2009;13:326–333. 5. Collins JW Jr, David RJ, Handler A, Wall S, Andes S. Very low birthweight in African American infants: the role of maternal exposure to interpersonal racial discrimination. Am J Public Health. 2004;94:2132– 2138. 6. Collins JW Jr, David RJ. Racial disparity in low birth weight and infant mortality. Clin Perinatol. 2009;36:63–73. 7. Pallotto EK, Collins JW Jr, David RJ. Enigma of maternal race and infant birth weight: a population-based study of US-born Black and Caribbeanborn Black women. Am J Epidemiol. 2000;151:1080–1085. 8. Hitti J, Nugent R, Boutain D, Gardella C, Hillier SL, Eschenbach DA. Racial disparity in risk of preterm birth associated with lower genital tract infection. Paediatr Perinat Epidemiol. 2007;21:330–337. 9. Barfield WD, Wise PH, Rust FP, Rust KJ, Gould JB, Gortmaker SL. Racial disparities in outcomes of military and civilian births in California. Arch Pediatr Adolesc Med. 1996;150:1062–1067. 10. Schoendorf KC, Hogue CJ, Kleinman JC, Rowley D. Mortality among infants of black as compared with white college-educated parents. N Engl J Med. 1992;326:1522–1526. 11. Steegers E. Begin at the beginning: some reflections on future periconceptional and obstetric care and research. Eur Clin Obstet Gynaecol. 2005;1:203–214.
240
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
12. Cetin I, Berti C, Calabrese S. Role of micronutrients in the periconceptional period. Hum Reprod Update. 2010;16:80–95. 13. Catov JM, Nohr EA, Bodnar LM, Knudson VK, Olsen SF, Olsen J. Association of periconceptional multivitamin use with reduced risk of preeclampsia among normal-weight women in the Danish National Birth Cohort. Am J Epidemiol. 2009;169:1304–1311. 14. Timmermans S, Jaddoe VW, Hofman A, Steegers-Theunissen RP, Steegers EA. Periconception folic acid supplementation, fetal growth and the risks of low birth weight and preterm birth: the Generation R Study. Br J Nutr. 2009;102:777–785. 15. Catov JM, Bodnar LM, Ness RB, Markovic N, Roberts JM. Association of periconceptional multivitamin use and risk of preterm or small-for-gestational-age births. Am J Epidemiol. 2007;166:296–303. 16. Vahratian A, Siega-Riz AM, Savitz DA, Thorp JM Jr. Multivitamin use and the risk of preterm birth. Am J Epidemiol. 2004;160:886–892. 17. Mitchell AA, Rosenberg L, Shapiro S, Slone D. Birth defects related to bendectin use in pregnancy. I. Oral clefts and cardiac defects. JAMA. 1981;245:2311–2314. 18. Werler MM, Hayes C, Louik C, Shapiro S, Mitchell AA. Multivitamin supplementation and risk of birth defects. Am J Epidemiol. 1999;150:675–682. 19. Oken E, Kleinman KP, Rich-Edwards J, Gillman MW. A nearly continuous measure of birth weight for gestational age using a United States national reference. BMC Pediatr. 2003;3:6. 20. Oken E, Kleinman KP, Olsen SF, Rich-Edwards JW, Gillman MW. Associations of seafood and elongated n-3 fatty acid intake with fetal growth and length of gestation: results from a US pregnancy cohort. Am J Epidemiol. 2004;160:774–783. 21. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122:51–65. 22. Use of dietary supplements containing folic acid among women of childbearing age–United States, 2005. MMWR Morb Mortal Wkly Rep. 2005;54:955–958. 23. Yawn BP, Suman VJ, Jacobsen SJ. Maternal recall of distant pregnancy events. J Clin Epidemiol. 1998;51:399–405. 24. Pearl M, Wier ML, Kharrazi M. Assessing the quality of last menstrual period date on California birth records. Paediatr Perinat Epidemiol. 2007;21(Suppl. 2):50–61. 25. Massachusetts Births 2006. Boston, MA: Division of Research and Epidemiology, Bureau of Health Information, Statistics, Research, and Evaluation, Massachusetts Department of Public Health; 2008. 26. Fulgoni V 3rd, Nicholls J, Reed A, Buckley R, Kafer K, Huth P, et al. Dairy consumption and related nutrient intake in African-American adults and
AEP Vol. 20, No. 3 March 2010: 233–240
children in the United States: continuing survey of food intakes by individuals 1994-1996, 1998, and the National Health and Nutrition Examination Survey 1999-2000. J Am Diet Assoc.. 2007;107:256–264. 27. Bukowski R, Malone FD, Porter FT, Nyberg DA, Comstock CH, Hankins GD, et al. Preconceptional folate supplementation and the risk of spontaneous preterm birth: a cohort study. PLoS Med. 2009;6 e1000061. 28. Scholl TO, Hediger ML, Schall JI, Khoo CS, Fischer RL. Dietary and serum folate: their influence on the outcome of pregnancy. Am J Clin Nutr. 1996;63:520–525. 29. Yang QH, Carter HK, Mulinare J, Berry RJ, Friedman JM, Erickson JD. Raceethnicity differences in folic acid intake in women of childbearing age in the United States after folic acid fortification: findings from the National Health and Nutrition Examination Survey, 2001-2002. Am J Clin Nutr. 2007;85:1409–1416. 30. Dowd JB, Aiello AE. Did national folic acid fortification reduce socioeconomic and racial disparities in folate status in the US? Int J Epidemiol. 2008;37:1059–1066. 31. Cotran RS, Kumar V, Collins T. Robbins Pathologic Basis of Disease. 6th ed. Philadelphia, PA: WB Saunders; 1999. 32. Werler MM, Shapiro S, Mitchell AA. Periconceptional folic acid exposure and risk of occurrent neural tube defects. JAMA. 1993;269:1257–1261. 33. American College of Obstetricians and Gynecologists; Nutrition During Pregnancy. 2008. Available at: http://www.acog.org/publications/patient_ education/bp001.cfm. Accessed April 22, 2009. 34. Recommendations for the use of folic acid to reduce the number of cases of spina bifida and other neural tube defects. MMWR Recomm Rep. 1992;41(RR-14):1–7. 35. Food and Nutrition Board, Institute of Medicine, Subcommittee on Upper Reference Levels of Nutrients. Dietary reference intakes for thiamin, riboflavin, niacin, vitamin B6, folate, vitamin B12, pantothenic acid, biotine, and coholine; A Report of the Standing Committee on the Scientific Evaluation of Dietary Reference Intakes and its Panel on Folate, Other B Vitamins, and Choline. Washington (DC): National Academy Press; 2000. 36. Dietary Supplementation Fact Sheet: Folate. National Institutes of Health, Office of Dietary Supplements. Available at: http://ods.od.nih.gov/ factsheets/folate.asp. Accessed July 8, 2009. 37. Folic acid for the prevention of neural tube defects: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2009;150:626–631. 38. Czeizel AE. Periconceptional folic acid and multivitamin supplementation for the prevention of neural tube defects and other congenital abnormalities. Birth Defects Res A Clin Mol Teratol. 2009;85:260–268.
PURPOSE: In the United States, African American women deliver preterm and low birth weight infants two to three times more frequently than their white counterparts. Our objective was to determine whether maternal periconceptional multivitamin (MVI) use is associated with this disparity. METHODS: As a secondary analysis of previously collected data from mothers of non-malformed infants from the Slone Epidemiology Center Birth Defects Study, we conducted a retrospective cohort study of 2331 non-Hispanic white and 133 non-Hispanic black mother/infant pairs from 1998 through 2007. To estimate the effect of MVI use on birth outcomes, linear regression models were used. RESULTS: In white subjects, MVI use was not associated with birth weight, gestational age, or weightfor-gestational-age. However, in black subjects, MVI use was associated with a 536-gram increased birth weight (p Z 0.001). Black MVI users also had longer gestations (although not statistically significant). When birth weights were adjusted for gestational age using z scores, MVI use was associated with increased fetal growth in black infants (þ0.86 z score units, 95% confidence interval: 0.35–1.36). CONCLUSIONS: The present findings suggest MVI use may improve fetal growth and possibly gestational age in the offspring of African American women. Ann Epidemiol 2010;20:233–240. Ó 2010 Elsevier Inc. All rights reserved. KEY WORDS:
Pregnancy, Preconception Care, Birth Weight, Vitamins, Premature Birth, Ethnic Groups, African Continental Ancestry Group.
INTRODUCTION African American women in the United States deliver preterm and low birth weight infants two to three times more frequently than their white counterparts (1). Studies of socioeconomic, genetic, and infectious factors to date have failed to entirely account for these disparities (2–8). Furthermore, racial disparities in these outcomes remain among women with equal access to health care and with college educations (9, 10). Discovering the etiology of racial disparities in birth outcomes and possible modifiable risk factors is crucial to resolving this immense public health problem. Recent evidence suggests that the processes of implantation and placentation are related to pregnancy outcomes, shifting the focus of potential risk factors to the periconceptional period (11–13). A study from the Netherlands recently found that folic acid supplementation during the preconceptional and periconceptional periods was associated with a decreased risk of low birth weight and smallfor-gestational-age (SGA) births (14). Another recent
From the Department of Neonatology, Beth Israel Deaconess Medical Center, Harvard University (H.H.B.), and Slone Epidemiology Center at Boston University (A.A.M, M.M.W.). Address correspondence to: Heather H. Burris, MD, MPH, Beth Israel Deaconess Medical Center, Neonatology–RO 318, 330 Brookline Ave., Boston, MA 02215. Tel: (617) 667-3276; Fax: (617) 667-7040. E-mail: [email protected]. Received July 29, 2009; accepted December 7, 2009. Ó 2010 Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010
observational study from Pittsburgh found that periconceptional vitamin use was associated with decreased rates of preterm and SGA births (15). However, in this study, changes in birth weight-for-gestational-age were not reported and race/ethnicity differences were not found. In our study, we examined the relationships between periconceptional multivitamin (MVI) use and birth weight, gestational age, and weight-for-gestational age as a measure of fetal growth. We also considered effect differences according to race/ ethnicity. Such differences could have important implications. If we found that MVI use improved neonatal outcomes among black women, then the documented lower rates of MVI use in black than in white women (16) may represent a targeted area for a public health intervention.
METHODS Study Population and Design The Slone Epidemiology Center Birth Defects Study (17, 18) includes a random sample of Massachusetts births who serve as control subjects. Mothers of malformed infants (cases) and controls are interviewed by telephone within 6 months of delivery about pregnancy events and exposures. For the present secondary analysis, control subjects interviewed between 1998 and 2007 were used to conduct a population-based retrospective cohort study. Of the eligible mothers, 80% were successfully contacted and 76% agreed to participate (60% overall participation), yielding a final 1047-2797/10/$–see front matter doi:10.1016/j.annepidem.2009.12.003
234
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
Selected Abbreviations and Acronyms MVI Z periconceptional multivitamin BMI Z body mass index LMP Z last menstrual period SGA Z small-for-gestational-age LGA Z large-for-gestational-age
sample size of 3,047 mother-infant pairs. This analysis was restricted to non-Hispanic black and white subjects; there were 2,374 white and 135 black mother-infant pairs. The Boston University institutional review board approved this study, and verbal consent was obtained before the start of the interview. Periconceptional Multivitamin Use A nurse-interviewer contacted mothers via telephone and conducted a standardized interview on demographic, reproductive, medical, dietary, and birth histories. As part of their responses to detailed questions about medication use, mothers provided the names, start and stop dates, and frequency of use of medications and vitamins, including MVIs. For this study, a supplement was considered an MVI if it contained at least two water-soluble and two fat-soluble vitamins. We defined the periconceptional period as 28 days prior to the last menstrual period (LMP) to 28 days after the LMP (lunar months –1 to 1). We categorized women as MVI users if they took an MVI four or more times per week in the periconceptional period. We also considered exposure to MVI in the second lunar month (LM2) of pregnancy. We categorized women as LM2 initiators of MVI if their use was at least four times per week. Measurement of Birth Weight, Gestational Age, and Fetal Growth Birth weight reported by subjects in pounds and ounces was converted to grams. Gestational age was determined by comparing the birth date to the reported due date. The onset of gestation was calculated as the due date – 280 days. Estimates of fetal growth or ‘‘weight-for-gestational-age’’ were based on the median birth weights for each completed week of gestation for a U.S. natality data set of more than six million neonates born in 1999 and 2000 (19). Weights for gestational age were converted to z scores where each z score unit represented the distance of the birth weight from the mean for any given gestational age (20). In this way, the effect of gestational age on birth weight was separated from fetal growth. Measurement of Covariates Covariates entered into the multivariable linear regression models included known risk factors for growth restriction
AEP Vol. 20, No. 3 March 2010: 233–240
and preterm birth. Because maternal age and pre-pregnancy body mass index (BMI) may not be linearly associated with outcomes, terms representing categories were included. Maternal age and BMI were each divided into three categories: !20, 20–34, and O34 years and !19.5, 19.5–25, and O25 kg/m2, respectively. Models also included terms for maternal education (!high school graduation vs. Ohigh school graduation), annual household income (!$45, 000 vs. more), cigarette smoking during pregnancy (vs. not), unmarried status (vs. married), and primiparity, defined as no prior viable pregnancies, (vs. multiparity). Self-reported maternal race/ethnicity was categorized as non-Hispanic black or non-Hispanic white. Women whoself-reported other race/ethnicities were excluded because of small numbers as well as the recognition of the long-standing black/white disparity in birth outcomes. The Willett Semiquantitative Food Frequency Questionnaire was used to ascertain dietary folic acid, iron, and caloric intakes (21). Intakes of each were categorized into quartiles. Analysis Bivariate analyses (using chi-square tests) were performed to determine whether the proportions of MVI use differed between the two race/ethnicity groups. Within each race/ ethnicity category, bivariate analyses using chi-square or Fisher’s exact tests were performed for white and black subjects separately to determine the characteristics of MVI users versus non-users in this cohort. With two-sided p values, student t tests were performed to obtain unadjusted associations of MVI use with birth weight, gestational age, and weight-for-gestational-age within the two strata. A t test was also performed to compare the mean birth weights in the two race/ethnicity categories. Stratified by race/ethnicity, multivariable linear regression models included terms for MVI use and the previously mentioned covariates. In addition, combined models including black and white subjects and interaction terms for MVI use and race/ethnicity were performed to formally test for effect measure modification of MVI use by race/ethnicity and its association with birth outcomes. To assess how our findings might have been influenced by incomplete participation among eligible mother-infant pairs, we performed sensitivity analyses in two ways. First, we assumed that black non-responders and black responders had similar MVI usage rates. We then assigned the infants of the non-responders the mean birth weight of the black responders’ exposed infants. Our second sensitivity analysis considered a different set of conditions: that black nonresponders had a higher rate of MVI use (50%) than black responders. All of the infants of black non-responders were assigned the mean birth weight of unexposed black infants. In these scenarios, we considered the most extreme examples
AEP Vol. 20, No. 3 March 2010: 233–240
of selection bias by assuming no exposure-outcome relationship among the non-responders. Lastly, to confirm that our findings were not secondary to an abnormal birth weight distribution (e.g., a disproportionate number of large-for-gestational-age [LGA] infants), we analyzed the data with and without infants at the tails of the distribution. All analyses were performed using the SAS version 9.1 (Cary, NC).
RESULTS Characteristics of Periconceptional Multivitamin Users The cohort for this analysis comprised 2,374 non-Hispanic white and 135 non-Hispanic black mother-infant pairs. Among women who reported any MVI use, all but 29 white women took MVIs at least four times per week. The less frequent supplementers were therefore excluded from subsequent analyses. Among white women considered exposed (at least four times per week), 1,192 reported daily use, 89 reported use four to six times per week, and two reported ‘‘regular’’ use. Among the 34 exposed black women, 32 reported daily use and two reported use four to six times per week. There were 720 white women and 49 black women who started MVI use in the second lunar month after their last menstrual period (LM2). Among them, 14 white and two black women reported initiating MVI use in LM2, but less than four times per week and were excluded from the cohort. The final cohort included 2,331 white and 133 black mother-infant pairs. MVI use at least four times per week in the periconceptional period (lunar months –1 to 1) was reported by 55.0% of white women (1,283/2,331) and by 25.6% of black women (34/133) (p ! 0.001). Among white women, users tended to be non-smokers, married, older, wealthier, and better educated. Patterns were similar for black women (Table 1).
Race/Ethnicity Disparities of Birth Outcomes The mean birth weight for black infants was 215 grams less than for white infants (3,291 vs. 3,506 respectively; p ! 0.001). Black infants’ average gestational age was 3.0 days less than white infants’ (273.3 vs. 276.3, respectively; p Z 0.03). Black infants’ weight-for-gestational-age was an average of 0.35 z score units less than for white infants (p ! 0.001) (mean z scores for black and white infants were –0.13 and 0.22, respectively). For white infants, birth weights (Table 2) and weight-for-gestational-age z-scores (data not shown) were lower in infants born to women who were smokers, unmarried, younger than 20 years, and primiparous. Associations were similar for black women, although they did not reach statistical significance. Neither
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
235
mean birth weight nor mean gestational age was statistically associated with mothers’ initiation of MVI use in LM2, although birth weights were lower in black infants of LM2 initiators. LM2 initiators were excluded from analyses comparing MVI users and non-users. Periconceptional Multivitamin Use and Associations with Birth Outcomes Among white women, there were no appreciable differences in birth weight (Fig. 1), gestational age (Fig. 2), or weight-forgestational-age (z score) (Fig 3) between users and non-users of MVI. Adjustment for the covariates listed in Table 1 made no substantial changes to these findings. However, among black mother-infant pairs, MVI users delivered substantially larger infants; after adjustment for multiple covariates, the difference persisted, with black MVI-exposed infants weighing an average of 536 g more than black unexposed infants (p Z 0.002) (see Fig. 1). Furthermore, weight-for-gestational age was substantially increased among black women who took MVIs; exposed black infants had z scores 0.86 greater than unexposed black infants (p Z 0.001) (see Fig. 3). Gestational age differed by 1.5 days between the two groups of black women (p Z 0.71) (see Fig. 2). Combined models with race/ ethnicity and MVI use included as interaction terms revealed that the interaction of non-Hispanic black race/ethnicity with MVI use was highly significantly associated with higher birth weights (p ! 0.0001) and increased weight for gestational age z scores (p Z 0.0004). The interaction was not significant for gestational age (p Z 0.3). Sensitivity Analysis to Account for Selection Bias We conducted sensitivity analyses to determine whether selection bias due to incomplete participation might account for our findings among black women. First, we assumed that the black women who did not respond to the survey had approximately the same frequency of MVI use as black responders (25%). We then assumed that the infants of black non-responders had the same mean birth weight (3,598 grams) as black MVI–exposed infants. After this adjustment, the unexposed infants’ mean birth weight increased from 3,165 to 3,412 grams, but a substantial difference of 186 grams (3598–3412 g) remained. Our second analysis considered a different set of conditions. We assumed that MVI users were underrepresented in the study and assigned a 50% rate of MVI use among non-responding black women. (This is twice the rate among black responders and closer to the 55% observed among white responders.) We assigned the mean birth weight of the unexposed black infants in the study (3,165 g) to all of the infants of the black non-responders. After this adjustment, the mean birth weight for exposed infants decreased to 3,353 grams from 3,598 grams whereas
236
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
AEP Vol. 20, No. 3 March 2010: 233–240
TABLE 1. Characteristics of mothers, by race/ethnicity, and periconceptional vitamin use, Slone Epidemiology Center Birth Defects Study Control Subjects, 1998–2007 NH white mothers (n Z 1,611)
Smoked during pregnancy* Yes No Educationz,x !High school High school or more Maternal age !20 yr 20 to !35 yr >35 yr Maternal pre-pregnancy BMI (kg/m2)k,{ <19.5 19.5 to !25 O25 Marital status Unmarried Married Parity Primiparous Multiparous Annual household income#,** !$45,000 >$45,000 Dietary intakeyy Folic acid intake (mg/d) Iron intake (mg/d) Total daily calorie intake
NH black mothers (n Z 84)
Peri-MVI use n Z 1,283 (%)
No peri-MVI use or LM2 MVI use n Z 328 (%)
73 (5.7)y 1209 (94.2)y
46 (14.1)y 281 (85.9)y
2 (5.9) 32 (94.1)
5 (10.0) 45 (90.0)
18 (1.5)y 1227 (98.6)y
24 (7.3)y 303 (92.7)y
2 (6.3) 30 (93.8)
8 (16.0) 42 (84.0)
20 (1.6)y 946 (73.7)y 317 (24.7)y
42 (12.8)y 234 (71.3)y 52 (15.9)y
5 (14.7) 20 (58.8) 9 (26.5)
6 (12.0) 37 (74.0) 7 (14.0)
126 (9.9) 778 (61.0) 372 (29.2)
41 (12.6) 187 (57.4) 98 (30.1)
2 (5.9) 16 (47.1) 16 (47.1)
5 (10.4) 20 (41.7) 23 (47.9)
121 (9.4)y 1162 (90.6)y
131 (39.9)y 197 (60.1)y
13 (38.2) 21 (61.8)
24 (48.0) 26 (52.0)
576 (44.9) 707 (55.1)
123 (37.5) 205 (62.5)
15 (44.1) 19 (55.9)
14 (28.0) 36 (72.0)
177 (14.2)y 1071 (85.8)y
130 (43.8)y 167 (56.2)y
13 (41.9) 18 (58.1)
26 (65.0) 14 (35.0)
391 (156)zz 12.9 (5.6)zz 1528 (479)
350 (167)zz 11.7 (5.5)zz 1553 (576)
401 (162) 12.7 (6.0) 1594 (700)
405 (188) 13.1 (5.3) 1679 (565)
Peri-MVI use n Z 34 (%)
No peri-MVI use or LM2 MVI use n Z 50 (%)
NH Z non-Hispanic; Peri Z periconceptional; MVI Z multivitamin use; LM2 Z second lunar month since last menstrual period; BMI Z body mass index. NOTE: Dietary intake data expressed as means (standard deviations). Periconceptional use defined as >4 times per week. *Two white women had missing smoking data. y p ! 0.05; chi-square test or Fisher’s exact test within race/ethnicity subgroups. z Thirty-nine white women had missing education data. x Two black women had missing education data. k Nine white women had missing BMI data. { Two black women had missing BMI data. # Sixty-six white women had missing income data. **Thirteen black women had missing income data. yy Twelve black women and 188 white women had missing dietary intake data. zz p ! 0.05; t test within race/ethnicity subgroups.
that for the unexposed infants remained at 3,165 grams. Even under these extreme assumptions, a 188-gram difference (3,417 g – 3,165 g) persisted between exposed and unexposed infants. We also considered the possibility that the MVI effect in black subjects was actually due to an increased number of large-for-gestational-age (LGA) births (O90th percentile). Indeed, the frequency of LGA was unusually high among our black MVI-exposed infants; nine of the 34 exposed infants were LGA., compared with three of the 50 MVI-unexposed infants. After eliminating LGA infants from the sample, we found that weight-for-gestational-age was still increased among infants exposed to MVIs; exposed black infants had
average z scores of 0.40 (95% CI: –0.06 to 0.86) greater than non-exposed infants. DISCUSSION In this study of 2,464 mother-infant pairs, MVI use was not associated with indices of fetal growth among non-Hispanic white subjects, but did appear to be associated with increased birth weight and fetal growth in non-Hispanic black subjects. The positive associations with MVI use and both birth weight and gestational age in black subjects raised the question as to whether the birth weight finding might be due to longer gestations and not increased fetal growth.
AEP Vol. 20, No. 3 March 2010: 233–240
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
237
TABLE 2. Mean birth weight and gestational age by maternal characteristics, stratified by race/ethnicity, Slone Epidemiology Center Birth Defects Study Control Participants, 1998–2007 NH white infants (n Z 2331)
All subjects Maternal characteristic Smoked during pregnancyy Yes No Educationx,k !High school High school or more Maternal age !20 yr >20 to !35 yr (ref) >35 yr Maternal pre-pregnancy BMI{,# (kg/m2) <19.5 19.5 to !25 (ref) >25 Marital status Unmarried Married Parity Primiparous Multiparous Annual household income**,yy !$45,000 >$45,000 MVI use (>4 times per week) In LM –1 or LM1 Initiation in LM2 No use in LM –1, LM1, or LM2 (ref) Folic acid daily intake,zz mg (quartiles) !272 (ref) 272–354 355–452 O452 Iron daily intake, zz mg (quartiles) !8.5 (ref) 8.5 to 11.5 11.6 to 15.4 O15.4
NH black infants (n Z 133)
Birth weight, grams (SD)
Gestational age, days (SD)
Birth weight, grams (SD)
Gestational age, days (SD)
3506 (522)*
276.3 (11.3)*
3291 (706)*
273.3 (16.9)*
3277 (508)z 3531 (519)
275.3 (12.4) 276.4 (11.2)
2967 (607) 3315 (709)
271.9 (17.2) 273.4 (16.9)
3372 (572)z 3511 (522)
277.6 (10.8) 276.3 (11.4)
2959 (1058) 3340 (651)
261.3 (37.6) 274.8 (12.6)
3335 (497)z 3508 (519) 3537 (534)
277.2 (11.2) 276.5 (11.0) 275.4 (12.5)
3103 (776) 3336 (665) 3274 (812)
272.7 (27.7) 274.3 (12.9) 269.1 (19.4)
3359 (497)z 3486 (487) 3591 (578)z
275.8 (11.5) 276.7 (10.3) 276.0 (12.2)
3382 (583) 3297 (780) 3301 (621)
277.5 (5.1) 272.2 (19.3) 274.5 (13.8)
3450 (516)z 3520 (523)
276.9 (12.2) 276.2 (11.1)
3240 (610) 3348 (800)
275.6 (16.5) 270.7 (17.1)
3424 (520)z 3569 (515)
277.1 (11.8)z 275.7 (10.9)
3143 (683) 3375 (709)
273.3 (19.9) 273.3 (15.1)
3486 (548) 3525 (510)
276.7 (12.8) 276.3 (10.8)
3380 (547) 3252 (673)
276.9 (8.4)z 272.1 (15.2)
3512 (503) 3495 (535) 3506 (566)
276.0 (11.0) 276.6 (11.3) 277.0 (12.6)
3598 (768)z 3207 (643) 3165 (673)
274.6 (9.5) 273.5 (14.1) 272.1 (22.6)
3451 (515) 3539 (482)z 3513 (523) 3511 (506)
275.6 (12.5) 277.5 (9.8)z 275.9 (10.8) 276.5 (11.5)
3295 (375) 3343 (580) 3218 (637) 3444 (592)
276.0 (6.8) 276.4 (11.1) 272.6 (11.9) 277.2 (8.5)
3454 (485) 3535 (518)z 3466 (516) 3562 (504)z
276.1 (11.7) 276.6 (10.9) 276.1 (10.8) 276.7 (11.3)
3273 (514) 3419 (502) 3239 (573) 3394 (593)
275.2 (9.8) 277.6 (8.5) 273.4 (11.1) 276.5 (8.6)
NH Z non-Hispanic; SD Z standard deviation; ref Z reverence group; BMI Z body mass index; MVI Z multivitamin; LM –1 Z lunar month prior to last menstrual period; LM1 Z first lunar month after last menstrual period; LM2 Z second lunar month since last menstrual period. *p ! 0.05; t test comparing between race/ethnic groups. y Two white women had missing smoking data. z p ! 0.05; t test compared with reference group within race/ethnic group. x Thirty-nine white women had missing education data. k Two black women had missing education data. { Nine white women had missing BMI data. # Two black women had missing BMI data. **Sixty-six white women had missing income data. yy Thirteen black women had missing income data. zz Twelve black women and 188 white women had missing dietary intake data.
However, the effect on birth weight remained after weightfor-gestational-age z scores were considered. Our finding of no effect in white women differs from results reported for women in the Netherlands and Pittsburgh (14, 15). The Pittsburgh study (15) found a beneficial effect of periconceptional vitamin use on preterm delivery
and fetal growth when race was controlled as a confounder, and the authors stated that there was no evidence of effect modification by race. Their rate of MVI use was similar to that of white women but higher than that of black women in our study. The definition of the periconceptional period for their study was much broader than ours, covering the
238
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
AEP Vol. 20, No. 3 March 2010: 233–240
b
900 800 700 600 Difference in 500 Birth Weight 400 300 (grams) 200 100 0 -100
Black Unadjusted Adjusted †
b
Black Unadjusted Adjusted †
10 8
412
536
White Unadjusted Adjusted †
13
-13
FIGURE 1. Periconceptional MVIs and birth weight differences. Vertical lines represent 95% confidence intervals; yAdjusted for smoking, income, education, age, body mass index, parity, marital status, folic acid intake, iron intake and caloric intake.
6 months leading up to their initial interview (average 9.9 weeks’ gestation), but did not include supplementation that began after pregnancy was detected. In the Dutch study (14) of primarily white women, the rate of supplementation was lower than in the white subgroup of our study population and was positively associated with fetal growth. The Dutch investigators defined the periconceptional period differently. They compared women with no MVI use at all before and during pregnancy with women who started MVI use before conception and women who started after the recognition of pregnancy but before the eighth week of gestation. In contrast, we defined the periconceptional period narrowly, as the month prior to the last menstrual period and the month after the menstrual period, thus capturing the moment of conception. In addition, it is possible that the white subjects in our study had higher levels of micronutrient(s) than the subjects in the other studies, thus exceeding a threshold below which MVIs could have an effect. Although there were no material differences in total iron or folic acid intake between white and black women, in our cohort we detected a significant difference in total caloric intake, with black women consuming an average of 146 more calories per day than white women (p Z 0.02). Thus, overall dietary patterns, and potentially quality, may differ between white and black women in our study, accounting for the MVI effect in black, but not in white women. We considered several potential limitations that could have affected our findings. Because our study was retrospective, we were concerned about recall error. Women were asked to report their MVI intake that occurred approximately 1 year earlier. We think that recall error in MVI reporting was minimal since the rates reported in our study are consistent with prospectively collected data from other studies (16, 22). Furthermore, while the collection of neonatal data was also retrospective, it was more recent (within 6 months), and maternal recall of birth weight has been demonstrated to be reliable years after delivery (23). For gestational age, we were reassured about accuracy
6 4
Gestational 2 Age 0 Difference -2 (days) -4
White Unadjusted Adjusted †
-0.8
1.8
1.5
-1.1
-6 -8
FIGURE 2. Periconceptional MVIs and gestational age differences. Vertical lines represent 95% confidence intervals; yAdjusted for smoking, income, education, age, body mass index, parity, marital status, folic acid intake, iron intake and caloric intake.
because due dates were used to determine gestational age and the onset of pregnancy. Indeed, more than 99% of women in our Birth Defects Study provided a due date. Furthermore, recall of due dates has been shown to be accurate (24), and it represents clinicians’ best estimate of gestational timing. Recall bias might also affect our findings, where the accuracy of maternal reporting of MVI use varies according to infants’ birth weights or gestational ages. We think this is unlikely because we would not have expected recall bias to differ between two race/ethnicity subgroups. Furthermore, mothers were not aware of whether they were cases or controls when they were asked to participate in the study. However, it remains possible that women who had smaller babies could have differentially recalled their MVI use. Participation bias might also affect our findings because approximately 40% of eligible subjects did not participate. Furthermore, some characteristics of the study sample differed from coincident Massachusetts births. For example, the proportion of women who considered themselves to be non-Hispanic white was 78% in the study sample and 68% in statewide data for 2006 (25). Despite these differences, when we considered extreme theoretical scenarios of selection bias, birth weights among non-Hispanic black infants remained higher in the MVI-exposed infants compared with the unexposed. While we cannot rule out the possibility that our findings are confounded by unmeasured dietary, lifestyle, and other demographic factors, we believe it unlikely that there would be differential unmeasured confounding between race/ ethnicity subgroups. However, the possibility remains that MVI use is simply a proxy for a healthier lifestyle and that this lifestyle could differ by race/ethnicity. Our findings are consistent with a plausible role played by micronutrients in fetal growth. It is not known which nutrient or combination of nutrients in MVIs might affect
AEP Vol. 20, No. 3 March 2010: 233–240
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
b
Black Unadjusted Adjusted †
1.5 Weight for Gestational Age Differences (Z score)
1 0.5
White Unadjusted Adjusted †
0.65
0.86
0 0.04
0.02
-0.5
FIGURE 3. Periconceptional MVIs and fetal growth differences. Vertical lines represent 95% confidence intervals; yAdjusted for smoking, income, education, age, body mass index, parity, marital status, folic acid intake, iron intake and caloric intake.
fetal growth, and our data do not contribute to this question. As other authors have suggested, micronutrients may affect implantation, placentation, and adequate vascular development of the placenta (15). Many micronutrients may play a role, and differential micronutrient status has been observed between black and white populations. For example, vitamin D levels are lower in black subjects because of less efficient sunlight absorption and vitamin D synthesis by darker pigmented skin as well as decreased intake of vitamin D–fortified milk (26). Whether vitamin D status affects fetal growth remains unknown. Folic acid is another micronutrient that might account for our findings. Recently a secondary analysis of a large cohort study found that preconceptional folic acid supplementation was associated with a decreased risk of early preterm birth (27). This is consistent with an earlier smaller study in New Jersey (28). Furthermore, there is evidence that, on average, black women consume less folic acid (29) and have lower circulating red blood cell folate levels even in the post-dietary fortification period (30). Without knowing which micronutrient may be responsible for our results, our findings raise the question as to whether the threshold for micronutrient deficiencies differs by disease state. The difference between clinical folate deficiency and disease-specific folate deficiency for neural tube defects is an example of one such discrepancy. To prevent megaloblastic anemia (the classic presentation of folate deficiency), one requires only 50–200 mg per day of folic acid (31). However, this level of intake among pregnant women is insufficient to prevent neural tube defects in the fetus (32). Public health recommendations recognize the different thresholds for folate deficiency for the general population and pregnant women (33–37). Furthermore, folic acid has been shown to decrease the risk of other anomalies in addition to neural tube defects including urinary anomalies and cardiac anomalies (38). The same phenomenon may apply for the micronutrient(s) in MVIs and the birth outcomes in our study.
239
In summary, we found that among non-Hispanic black women, those who took an MVI delivered infants with increased birth weights and fetal growth and possibly increased durations of gestation. Similar effects were not observed among non-Hispanic white women. These results are based on secondary analyses of an existing data set and therefore need to be replicated. If our findings were confirmed and subsequently shown to be causal, then increasing periconceptional MVI use among African American women in the United States could help to eliminate longstanding disparities in birth weight, gestational age, and fetal growth. We thank Katherine Ahrens, MPH, Dawn Jacobs, RN, MPH, Fiona Rice, MPH, Rita Krolak, RN, Kathleen Sheehan, RN, Claire Coughlin, RN, Moira Quinn, RN, Nancy Rodriguez, Carolina Tejedor Meyers, and Nastia Dynkin, for their assistance in data collection and computer programming; and the staff of the Massachusetts Department of Public Health Center for Birth Defects Research and Prevention. We also thank all the mothers who participated in the study. Dr. Burris thanks the Harvard Pediatric Health Services Research Fellowship and its funders, including the Agency for Healthcare and Quality. The data for this project are held at the Slone Epidemiology Center at Boston University. Dr. Burris from Beth Israel Deaconess Medical Center led the project.
REFERENCES 1. Behrman R, Butler A. Preterm birth: Causes, consequences, and prevention. Washington (DC): National Academies Press; 2007. 2. David RJ, Collins JW Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks, and U.S.-born whites. N Engl J Med. 1997;337:1209–1214. 3. Wang H, Parry S, Macones G, Sammel MD, Kuivaniemi H, Tromp G, et al. A functional SNP in the promoter of the SERPINH1 gene increases risk of preterm premature rupture of membranes in African Americans. Proc Natl Acad Sci U S A. 2006;103:13463–13467. 4. Collins JW Jr, Wambach J, David RJ, Rankin KM. Women’s lifelong exposure to neighborhood poverty and low birth weight: a population-based study. Matern Child Health J. 2009;13:326–333. 5. Collins JW Jr, David RJ, Handler A, Wall S, Andes S. Very low birthweight in African American infants: the role of maternal exposure to interpersonal racial discrimination. Am J Public Health. 2004;94:2132– 2138. 6. Collins JW Jr, David RJ. Racial disparity in low birth weight and infant mortality. Clin Perinatol. 2009;36:63–73. 7. Pallotto EK, Collins JW Jr, David RJ. Enigma of maternal race and infant birth weight: a population-based study of US-born Black and Caribbeanborn Black women. Am J Epidemiol. 2000;151:1080–1085. 8. Hitti J, Nugent R, Boutain D, Gardella C, Hillier SL, Eschenbach DA. Racial disparity in risk of preterm birth associated with lower genital tract infection. Paediatr Perinat Epidemiol. 2007;21:330–337. 9. Barfield WD, Wise PH, Rust FP, Rust KJ, Gould JB, Gortmaker SL. Racial disparities in outcomes of military and civilian births in California. Arch Pediatr Adolesc Med. 1996;150:1062–1067. 10. Schoendorf KC, Hogue CJ, Kleinman JC, Rowley D. Mortality among infants of black as compared with white college-educated parents. N Engl J Med. 1992;326:1522–1526. 11. Steegers E. Begin at the beginning: some reflections on future periconceptional and obstetric care and research. Eur Clin Obstet Gynaecol. 2005;1:203–214.
240
Burris et al. MULTIVITAMINS AND BIRTH WEIGHT DISPARITIES
12. Cetin I, Berti C, Calabrese S. Role of micronutrients in the periconceptional period. Hum Reprod Update. 2010;16:80–95. 13. Catov JM, Nohr EA, Bodnar LM, Knudson VK, Olsen SF, Olsen J. Association of periconceptional multivitamin use with reduced risk of preeclampsia among normal-weight women in the Danish National Birth Cohort. Am J Epidemiol. 2009;169:1304–1311. 14. Timmermans S, Jaddoe VW, Hofman A, Steegers-Theunissen RP, Steegers EA. Periconception folic acid supplementation, fetal growth and the risks of low birth weight and preterm birth: the Generation R Study. Br J Nutr. 2009;102:777–785. 15. Catov JM, Bodnar LM, Ness RB, Markovic N, Roberts JM. Association of periconceptional multivitamin use and risk of preterm or small-for-gestational-age births. Am J Epidemiol. 2007;166:296–303. 16. Vahratian A, Siega-Riz AM, Savitz DA, Thorp JM Jr. Multivitamin use and the risk of preterm birth. Am J Epidemiol. 2004;160:886–892. 17. Mitchell AA, Rosenberg L, Shapiro S, Slone D. Birth defects related to bendectin use in pregnancy. I. Oral clefts and cardiac defects. JAMA. 1981;245:2311–2314. 18. Werler MM, Hayes C, Louik C, Shapiro S, Mitchell AA. Multivitamin supplementation and risk of birth defects. Am J Epidemiol. 1999;150:675–682. 19. Oken E, Kleinman KP, Rich-Edwards J, Gillman MW. A nearly continuous measure of birth weight for gestational age using a United States national reference. BMC Pediatr. 2003;3:6. 20. Oken E, Kleinman KP, Olsen SF, Rich-Edwards JW, Gillman MW. Associations of seafood and elongated n-3 fatty acid intake with fetal growth and length of gestation: results from a US pregnancy cohort. Am J Epidemiol. 2004;160:774–783. 21. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122:51–65. 22. Use of dietary supplements containing folic acid among women of childbearing age–United States, 2005. MMWR Morb Mortal Wkly Rep. 2005;54:955–958. 23. Yawn BP, Suman VJ, Jacobsen SJ. Maternal recall of distant pregnancy events. J Clin Epidemiol. 1998;51:399–405. 24. Pearl M, Wier ML, Kharrazi M. Assessing the quality of last menstrual period date on California birth records. Paediatr Perinat Epidemiol. 2007;21(Suppl. 2):50–61. 25. Massachusetts Births 2006. Boston, MA: Division of Research and Epidemiology, Bureau of Health Information, Statistics, Research, and Evaluation, Massachusetts Department of Public Health; 2008. 26. Fulgoni V 3rd, Nicholls J, Reed A, Buckley R, Kafer K, Huth P, et al. Dairy consumption and related nutrient intake in African-American adults and
AEP Vol. 20, No. 3 March 2010: 233–240
children in the United States: continuing survey of food intakes by individuals 1994-1996, 1998, and the National Health and Nutrition Examination Survey 1999-2000. J Am Diet Assoc.. 2007;107:256–264. 27. Bukowski R, Malone FD, Porter FT, Nyberg DA, Comstock CH, Hankins GD, et al. Preconceptional folate supplementation and the risk of spontaneous preterm birth: a cohort study. PLoS Med. 2009;6 e1000061. 28. Scholl TO, Hediger ML, Schall JI, Khoo CS, Fischer RL. Dietary and serum folate: their influence on the outcome of pregnancy. Am J Clin Nutr. 1996;63:520–525. 29. Yang QH, Carter HK, Mulinare J, Berry RJ, Friedman JM, Erickson JD. Raceethnicity differences in folic acid intake in women of childbearing age in the United States after folic acid fortification: findings from the National Health and Nutrition Examination Survey, 2001-2002. Am J Clin Nutr. 2007;85:1409–1416. 30. Dowd JB, Aiello AE. Did national folic acid fortification reduce socioeconomic and racial disparities in folate status in the US? Int J Epidemiol. 2008;37:1059–1066. 31. Cotran RS, Kumar V, Collins T. Robbins Pathologic Basis of Disease. 6th ed. Philadelphia, PA: WB Saunders; 1999. 32. Werler MM, Shapiro S, Mitchell AA. Periconceptional folic acid exposure and risk of occurrent neural tube defects. JAMA. 1993;269:1257–1261. 33. American College of Obstetricians and Gynecologists; Nutrition During Pregnancy. 2008. Available at: http://www.acog.org/publications/patient_ education/bp001.cfm. Accessed April 22, 2009. 34. Recommendations for the use of folic acid to reduce the number of cases of spina bifida and other neural tube defects. MMWR Recomm Rep. 1992;41(RR-14):1–7. 35. Food and Nutrition Board, Institute of Medicine, Subcommittee on Upper Reference Levels of Nutrients. Dietary reference intakes for thiamin, riboflavin, niacin, vitamin B6, folate, vitamin B12, pantothenic acid, biotine, and coholine; A Report of the Standing Committee on the Scientific Evaluation of Dietary Reference Intakes and its Panel on Folate, Other B Vitamins, and Choline. Washington (DC): National Academy Press; 2000. 36. Dietary Supplementation Fact Sheet: Folate. National Institutes of Health, Office of Dietary Supplements. Available at: http://ods.od.nih.gov/ factsheets/folate.asp. Accessed July 8, 2009. 37. Folic acid for the prevention of neural tube defects: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2009;150:626–631. 38. Czeizel AE. Periconceptional folic acid and multivitamin supplementation for the prevention of neural tube defects and other congenital abnormalities. Birth Defects Res A Clin Mol Teratol. 2009;85:260–268.