dietary management, but the mean birthweight (3218 g) of the 166 liveborn infants born to .... asphyxia neonatorum among the recipients' infants. (6.8% v. 4.6%).
Nutritional services Original Articles
during pregnancy and birthweight: retrospective matched pair analysis a
DAVID RUSH,
MD
Since 1963, unselected prenatal patients at the Royal Victoria cours duquel les soins prenatals ont ete commences et le Hospital, Montreal, have been given nutritional counselling poids au moment de la conception. and, if it was judged necessary, dietary supplementation by Le pourcentage de bebes de faible poids de naissance (de the Montreal Diet Dispensary. From uniform data collected moins de 2500 g) fut de 5.7% pour les beneficiaires et de for all obstetric patients in 1963-74, 1213 recipients of the 6.8% pour les temoins; cette difference n'est pas significadispensary services (89.7% of those available and eligible tive, mais les enfants des beneficiaires etalent plus lourds a for matching) were paired with controls matched for date la naissance, de 40 g en moyenne (P < 0.05), que les enof delivery (within 12 months), religious affiliation, parity, fants temoins. La difference de poids de naissance etait la trimester of pregnancy during which prenatal care was plus grande pour les enfants des meres qui en etaient a begun and weight at the time of conception. leur premiere grossesse (61 g en moyenne) et la moindre The proportion of infants of low birthweight (less than pour les enfants des femmes qui avaient eu trois accouche2500 g) was 5.7% for the recipients and 6.8% for the con- ments ou plus dans le passe (moyenne de 9 g). L'augmentatrols; the difference was not significant, but the recipients' tion du poids de naissance (53 g en moyenne, P < 0.02) chez infants were heavier at birth than the controls' infants, by les bebes des beneficiaires fut limitee aux enfants des an average of 40 g (P < 0.05). The difference in birthweight femmes qui pesaient moins de 140 lb (63 kg) au moment de was greatest for the infants of women in their first preg- la conception; chez les femmes plus lourdes les temoins nancy (average 61 g) and least for the infants of women ont eu des enfants plus pesants, mals la difference n'est pas with three or more past deliveries (average 9 g). Increased significative. Les differences entre les groupes en ce qui birthweight (by an average of 53 g, P < 0.02) among the a trait a la duree de la grossesse et au gain de poids marecipients' infants was limited to those born to women ternel n'explique qu'une partie minime des differences de weighing less than 140 lb (63 kg) at the time of conception; poids de naissance. among the heavier women the controls had infants who Cette etude fait la preuve que le programme du Dispenwere heavier, but not significantly so. Differences between saire dietetique de Montreal a augment6 le poids de naisthe groups in duration of gestation and maternal weight gain sance de faron significative. II faut dorenavant consacrer accounted for only a small part of these differences in des efforts a evaluer les avantages a long terme de ces birthweight. changements. This study provides evidence that the Montreal Diet Dispensary program significantly increased birthweight. Further The relation between the nutritional status of the efforts must now be directed towards judging the long-term mother and fetal growth, studied as birthweight, has benefit of these changes. been of great interest because of the intense relation A l'h6pital Royal Victoria de Montreal, depuis 1963, des between birthweight and perinatal mortality.1-5 During famine, birthweight is depressed (for instance, patientes enceintes prises au hasard ont regu du Dispensaire dietetique de Montreal des conseils de nutrition et, in the Dutch famine of 1944-45 the mean birthweight lorsque cela fut juge necessaire, un supplement nutritionnel. was depressed as much as 300 g, or about 10% of A partir de donnees uniformes colligees pour l'ensemble des body weight3'), and with iatrogenic dietary manipulapatientes d'obstetrique pendant la periode de 1963-74, 1213 tion the birthweight can be lowered."'7 However, the beneficiaires des services du dispensaire (89.7% de celles effect of supplementation of the day-to-day maternal qui etaient disponibles et eligibles pour appareillement) ont diet on fetal growth has been unclear. Earlier studies ete appariees avec des sujets temoins comparables pour la of this issue have tended to suffer from several inherent date d'accouchement (a 12 mois pres), I'appartenance reli- faults. Observational studies have suffered because of gieuse, le nombre d'enfants, le trimestre de la grossesse au the difficulty of measuring nutritional status, as distinct From the developmental epidemiology research unit, division of epidemiology, school of public health, the department of pediatrics and the institute of human nutrition, all of the faculty of medicine, Columbia University, New York Reprint requests to: Dr. David Rush, 100 Haven Ave., SA, New York, NY 10032, USA
from concomitant problems associated with poverty. In studies of nutritional supplementation or counselling, frequently too few women have been studied to test the significance of a change in birthweight that might reasonably follow nutritional intervention. Often there have been no untreated controls or the controls have been self-selected - that is, they were women CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125 567
who chose not to participate or to participate at a of some women requiring special dietary management low level in the supplementation program. A new gen- by the hospital's obstetric service, there was no syseration of studies of dietary intervention during preg- tematic referral into or out of the clinics serviced by nancy is now available.45 The one reported here not the dispensary. From the middle of 1970 every other only is one of the largest of these but also has evaluated patient registering at three public prenatal clinics was a multifaceted nutritional program that is more im- assigned to the dispensary program. During 1973 one mediately applicable to the realities of clinical practice of these clinics became a high-risk clinic, but the disthan are many programs designed mainly to fulfil re- pensary continued to take alternate patients from it. The alternately allocated patients - that is, those research objectives. The Montreal Diet Dispensary, under the direction cruited since mid-1970 - accounted for about 15% of Agnes C. Higgins, has offered a nutritional inter- of the study population. vention program to public prenatal patients of the There were 177 mothers of singletons whom the Royal Victoria Hospital, Montreal since 1963. A de- dispensary excluded from their roster because these scription by Higgins of the method employed8 is in- women did not meet the dispensary's criteria for full cluded in this paper as an appendix. The program delivery of dietary services (Table 1). The reasons includes assessment of nutritional status, counselling for exclusion included early fetal death and special and, for about three quarters of the women, supple- dietary management, but the mean birthweight (3218 mentation of the diet with milk, eggs and oranges. g) of the 166 liveborn infants born to these women Higgins and others have described some outcomes did not differ markedly from that of the infants born among recipients of these services in several publica- to the recipients of the dispensary services or their tions.8-" In one study, siblings from untreated preg- controls. However, the rate of low birthweight (less nancies served as controls. The value of other analyses than 2500 g) for the 166 infants for whom birthweight was limited by the absence of concurrent controls. The was known, 11.4%, was considerably higher than that study described here involved 1213 recipients of the for the other two groups of infants. services who were retrospectively paired with controls. This study was limited to women who were judged white, were public patients at the Royal Victoria Hospital and delivered, at that hospital, singleton infants Methods whose birthweight was 1000 g or more and gestational During the study period, 1963-74, the Royal Vic- age 28 weeks or more. Of the 1892 such women who toria Hospital was a participant in the Obstetrical Stat- received dispensary services 1791 (95%) had identistical Cooperative, a consortium of over 20 obstetric ifiable records in the cooperative's file. The records departments in North American teaching hospitals that for 1464 of them had all the data necessary for matchcollected uniform data on all obstetric admissions. The ing and analysis. All the recipients of the dispensary data at the Royal Victoria Hospital were collected in services were registered as public patients, but the a standardized format for all obstetric patients. Using records indicated that 129 of the 1791 underwent these data we retrospectively matched the recipients delivery as private patients. We retained these patients of dispensary services with controls according to a in the analysis, and their inclusion did not affect the number of prior characteristics and compared the out- final results. We drew our matched controls from the come of pregnancy. The partially completed analysis has been published in abstract form."3 For the cooperative's file every prenatal and hospital record of women whose infants were delivered at the Royal Victoria Hospital was abstracted by specially trained and regularly supervised statistical clerks, who, on the basis of past training and a standardized format, filled out a precoded form developed for all hospitals participating in the cooperative. Although these data are limited by the adequacy of the records, since there was no difference between the records of the recipients of dispensary services and others the inaccuracy should not be systematic but should apply equally to both recipients and controls. Indeed, any measurement error would tend to underestimate differences between the two groups. Between 1963 and 1974, of the women the dispensary delivered services to and maintained statistics on at the Royal Victoria Hospital were 1892 who bore singletons with a birthweight of over 500 g. Until the middle of 1970 the dispensary provided services for all women who registered for care at certain of the hospital's public prenatal clinics and did not treat those registering at other clinics. Aside from the retention I 568 CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125
remaining 12 318 public patients delivered at the hospital who did not receive dispensary services during the study period and who met the same inclusion criteria as the recipients. While matching was used in part to achieve comparability between the recipient and control groups in factors known to affect birthweight, any differences between the recipient and control groups in the frequency of factors associated with birthweight, such as sex of the infant, could also be controlled in the analysis. Our central aim, however, was to match on characteristics hypothesized to possibly affect the maternal response to nutritional treatment, and thereby to be able to judge whether treatment effects might be conditional on those matching characteristics. We matched the patients according to the following five criteria: * Delivery within 12 months of each other. We thus meant to ensure that the comparability of recipients and controls was unaffected by change over time, whether in the population served or in medical services offered. If several women were eligible to be the control, the one delivering closest in time to the recipient was chosen. * The same religious affiliation (Catholic or Protestant). We would have preferred to control for language preference, since it is conceivable that the services of the dispensary, a predominantly Anglophone agency, might have been more effective among English-speaking recipients. However, language preference was not recorded in the cooperative's file, and religion was the closest surrogate available. While the Catholic group included both English- and French-speaking women, the Protestant group included predominantly English-speaking women. Thus, if the dispensary's services were more effective among English-speaking women, a greater effect among the Protestants would be expected, * The same number of previous deliveries of living infants weighing 2500 g or more (none, one to two, or three and over). (Women whose infants were of low birthweight were included in all analyses; however, only the count of infants that weighed 2500 g or more
was used for matching because of the way the variables were coded.) We controlled for the number of previous pregnancies and, therefore, indirectly for age, because multiparas and older women in general might be less flexible in their dietary habits, and because the quality of the available diet and the nutritional status might vary with parity and age. * Registration for prenatal care in the same trimester of pregnancy. We thus controlled for both duration of prenatal care and motivation to seek care at the same stage of pregnancy. This facilitated analysis of the relation between duration of exposure to the dispensary services and their effect. * The same reported body weight at the time of conception (less than 100 lb, 10-lb strata up to 179 lb, 180 to 199 lb, and 200 lb and over). Body weight is strongly related to both birthweight and nutrition.'4 In addition, not only would we expect dietitians to\respond differently to clients of different maternal weights, but also a woman's weight might affect how she participated in a program of dietary change. Because of the infrequency of very heavy women, relatively fewer control women were available in the higher weight strata, and matching was therefore less complete than at the lower weights. Of the 1464 recipients who were eligible for the study 89.1% (1305) could be matched with controls according to these five criteria. To refine the analysis of the effect of nutritional services on birthweight we further excluded, a priori, certain categories of women, without knowledge of the birthweight of their infants (Table II). The purposes of the further exclusions were to eliminate bias due to possible unequal distribution of these characteristics between recipient and control groups, such as with diabetes or altered glucose metabolism during pregnancy, and to exclude infants whose birthweights were likely to have been strongly influenced by factors that had little to do with maternal nutrition and, even if distributed comparably between the recipient and control groups, would dilute our estimate of the effect of nutritional services. Of the various reasons for further exclusion, only diabetes had a significantly different frequency in the recipient and control groups: 35 controls but only 10 recipients were identified as diabetic. Women whose infants died prior to labour were excluded because five such women had been excluded earlier by the dispensary (Table I) and because of the difficulty in judging when fetal growth stopped in such pregnancies. The rates of infant death during or after labour were similar in the recipient and control groups (0.7% and 0.9% respectively). There were more fetal deaths in the control group (16) than in the recipient group (7) (Table II); this difference was not statistically significant. It is difficult to interpret this as a treatment effect for two reasons. First, the dispensary excluded five early fetal deaths from their data. Second, 4 of the 16 control deaths were in fetuses of women who were diabetic or had abnormal results in glucose tolerance tests, and diabetic individuals were significantlv underrepresented in the dispensary population (Table III). CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125 569
made without knowledge of
to the difference in birthweight between the two groups
outcome; it is noteworthy that the difference in mean birthweight between the infants born to the recipient
of infants. The distribution of the birthweights in the two groups of infants is shown in Table IV. The rate of low birthweight (less than 2500 g) was 5.7% for the recipients' infants and 6.8% for the controls' infants; the difference was not significant. The rate for the controls' infants was nearly identical to that for the infants of all public patients at the hospital who would have been eligible for the study (7.0%). This rate was adjusted for the weight distribution of the study population before pregnancy. The mean birthweight of the infants born to the 1213 recipients was 40 g greater than that of the infants born to the matched controls; this difference was significant at the 0.05 level of probability (Tables V and VI). Of the five matching characteristics of the mothers, two were unrelated to the difference in birthweight of their children: religion and year of delivery. The relation of trimester of registration for prenatal care is unclear. (In our preliminary analysis of outcome through 1970 we found an inverse relation between trimester of registration and effect on birthweight.'3) Multiple regression analysis was performed to test whether there was interaction between dispensary services during pregnancy and the sex of the infant. There was no significant interaction, and the differences in
These refinements
were
and control groups before the refinements were made only 1 g different from that after refinement. After these additional exclusions 1213 matched pairs remained. (If a recipient had been matched to an ineligible control, we attempted to rematch her with an eligible control.) These 1213 pairs were studied for the effect of the dispensary's services on birthweight, maternal weight gain and duration of gestation. All statistical tests were for two-tailed probabilities and, except where specified, for matched pairs. was
Results
The recipient and control populations were nearly identical in age, marital status and frequency of past cesarean sections, delivery of infants of low birthweight and perinatal loss. However, significantly more of the recipients (73.6% v. 69.0%, P < 0.01) had been born in Canada. The experience of the two groups in the pregnancy under study was also similar (Table III). The only significant difference (P < 0.05) in the variables we were able to study was a higher rate of recorded asphyxia neonatorum among the recipients' infants (6.8% v. 4.6%). (The appearance of one such significant difference might be expected by chance alone.) The mean period of gestation was 0.4 days longer in the recipients' infants. This difference was not significant and can account for a difference of about 7 g in birthweight. Adjustment for the higher proportion of boys born to the controls would add less than 1 g Table Ill-Comparability of the recipient and matched control subjects used for the birthweight analysis Recipients Variable (n = 1213) During gestation Clinic visits (median no.) 7.2 6.4 Hemoglobin level < 10 g/dl (%) 0.7 Psychiatric problem (%) 2.4 Allergy (%) Venereal disease (%) 0.6 0.2 Hospitalized for hyperemesis (%) Toxemia (%) 6.4 At delivery Mean duration of gestation (d) 276.0 8.2 < 37 wk(%) 39-42 wk(%) 70.7 Entered hospital in true or 92.0 suspected labour (%) 7.4 Labour induced (%) Cesarean section (%) 5.0 Fever during labour or 0.2 delivery (%) Antimicrobial therapy before or during labour(%) 1.6 Breech delivery (%) 2.4 Newborn Male (%) 51.4 4.6 Congenital anomaly (%) "Placental insufficiency" (%) 2.9 6.8* Asphyxia neonatorum (%) Infection (%) 10.4 *Difference significant (X2 = 4.88, P < 0.05). 570
Controls (n = 1213) 6.9 6.9 0.8 3.2 0.2 0.7 6.9
Table IV-Distribution of the infants' birthweights %
of infants
Control mothers (n= 1213)
Recipient mothers (n= 1213)
Birthweight(g) < 2000 2000--2499 Total 140 Total < 140 (n 215) (n = 1213) (n = 998) Variable 276.0 278.6 275.4 Duration of gestation(d) (0.4) (0.6) (0.4) 25.3 23.1 25.8 Weight gain (lb) (0.3) (-1.7) (0.6) 3291 3488 3248 Birthweight (g)
(53)*
(-23)
(40)t
*Significant at P < 0.02. tSignificant at P < 0.05.
recipients gaining an average of 0.6 lb more with none, 0.3 lb more with one or two and 0.2 lb less with three or more) and with maternal weight at the time of conception (the recipients gaining an average of 0.6 lb more if they weighed under 140 lb at the time of conception and 1.7 lb less if they were heavier at the time of conception) were small but consistent in direction with the observed differences in birthweight. In Table VIII the birthweights of the recipients' infants and the differences between those weights and the weights of the controls' infants are arrayed by both the recorded trimester of registration for prenatal care and the trimester of initiation of dispensary services. The results are confusing since there is no gradient of birthweight difference by trimester of registration but there is a clear difference when the data are arrayed by the trimester in which dispensary services were initiated: there were differences between the recipient and control groups only for the pairs in which the recipient began receiving dispensary services before the third trimester. This pattern is due entirely to the lower birthweight (mean 119 g) of the infants born to the 203 recipients who registered at the Royal Victoria Hospital in a trimester before the one in which they began receiving dispensary services. We related the number of prenatal visits, as an index of intensity of dietary services, to outcome, and the results of this analysis did suggest that effects were greatest with greater exposure to dietary intervention. The number of prenatal clinic visits was known for both members of all but 49 recipient-control pairs. A seemingly straightforward analysis of differences between the infants of recipients and controls arrayed by the number of cliniic visits of the recipient would be severely confounded by the relation of the number of visits to the duration of gestation. This problem was solved with multiple regression analysis where the dependent variable was the difference in birthweight between the infants of the recipients and the controls, and the independent variables were the number of prenatal visits of the two groups of mothers. The birthTable Vil-Relation of trimester of registration for prenatal care and trimester of initiation of dispensary services to infant's birthweight Trimester of registration for prenatal care; mean birthweight (g) of recipients' infants (and mean difference between the two groups of infants)
Table VIl-Relation of number of past deliveries to duration of gestation, weight gain and infant's birthweight for the women weighing less than 140 lb (63 kg) at the time of conception No. of past deliveries;* mean for recipients (and mean difference between recipients and controls) 0 1-2 Total 3+ Variable (n = 439) (n= 367) (n = 192) (n= 998) 275.4 277.3 Duration of gestation (d) 274.7 275.3 (0.4) (0.9) (1.2) (-0.6) 25.8 24.8 26.3 25.6 Weight gain (lb)
(0.9) (0.5) (0.4) 3296 3276 3203 Birthweight (g) (49) (17) (73)t *Excluding births of infants weighing less than 2500 g. tSignificant at P < 0.02.
(0.6) 3248 (53)t
Trimester of initiation of dispensary services 1
(45)
3
Total
3
(63)
(244)
n= 18
Total 3354
3533
3336 n= 173
2
2
1
-
n= 191
3162
3282 n= 667
3294 (20) n= 104 3419
n
3283 (69) = 546 3235
n= 17
n= 13
n= 86
n= 256
n= 355
3324
3284
3271
3291
n= 290
n= 650
n= 273
(7)
(31)
(-155) (44)
(-37) 3279
(55)
(38)
(54)
3273
(0)
(40)
n= 1213
CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125
571
weight differences were estimated from the regression hemoglobin levels indicate greater expansion of the equation as -1.2 g for 3 clinic visits and 44.6 g for plasma volume. 12 visits (Table IX). Table X presents the differences in birthweight and Discussion and conclusions the frequency of low birthweight among the infants of This study has given us a better understanding of multiparas according to whether there was a past delivery of an infant of low birthweight. This analysis how much birthweight can be affected by nutritional is complicated because we did not match for past services in a Western public clinic population without delivery of infants of low birthweight. Since the pro- clinical evidence of nutritional deprivation. This work, tocol called for matching only by numbers of full- in conjunction with other recently reported studies,45 weight infants, deliveries of infants of low birthweight has also begun to define the role of nutrition in fetal growth. Nutritional services in the population we were included in the definition of parity in this table only. Among those with a history of delivering an studied were associated with a statistically significant infant of low birthweight in the past, fewer recipients increase in birthweight of 40 g. This increase was confined to infants of lighter of dispensary services than controls (14.7% v. 21.7%) had low-birthweight infants this time, but for the re- mothers (those weighing less than 140 lb at the time cipients the mean gestation was 3.9 days shorter, the of conception), who were presumably at greater nutritional risk than heavier women and able to benefit mean weight gain 4.0 lb less and the mean birthweight of their infants 86 g less. None of these differences more from increased dietary intake. The response to dietary services decreased with increasing parity; the were significant. The mean birthweight was 100 g higher for infants women of lower parity complied better with the treatwhose mothers' lowest hemoglobin level was under ment regimen or were at higher nutritional risk, or 10 g/dl, compared with infants whose mothers had both, and therefore were able to respond more to higher hemoglobin levels. This relation is probably not therapy. The relation between duration of dispensary services explained by "anemia" but by the fact that the lower and acceleration of fetal growth is not clear from our analyses: when the data were arrayed by trimester of registration for prenatal care there'was no relation to Table IX-Difference in birthweight between infants of recipients and controls and number of clinic visits: multiple regression analysis birthweight, but when they were arrayed by trimester for the 1164 recipient-control pairs for which the number of visits of initiation of dispensary services there was a relation. was known for both members We see two possible explanations: either those who delayed obtaining dispensary services after initially Regression equation for estimating difference (D) in birthweight (g): D = 20.46(no. of visits by recipient) - 15.35(no. of visits by control) appearing at the clinic were at inherently higher risk - 16.53 of producing infants of lower birthweight, or, given Estimated differences in birthweight the shorter exposure of these women to the dispensary for different numbers of clinic visits there was not enough time for dietary change services, (assuming the same number for recipients and controls) to affect fetal growth. Unfortunately, the data do not Estimated difference (g) No. of clinic visits allow a clear choice between these alternatives. If -1.2 3 the latter were true we would expect the difference 9.0 5 in birthweight between the infants of the recipients and 24.2 8 44.6 12 the controls for those pairs in which the recipient delayed obtaining dispensary services after registering for Table X-Relation of past deliveries of infants of low birthweight to difference between multiparous recipients and controls in duration of gestation, weight gain and infant's birthweight* and to current delivery of an infant of low birthweight Past delivery of infants of low birthweight; mean difference between recipients and controls
Variable Duration of gestation (d) Weight gain (lb) Birthweight (g)
Recipient and control (n= 13)
Recipient only (n= 82)
Control only (n= 70)
Weighted averaget
Neither (n= 617)
Total (n= 782)
-10.0 -4.6 - 292
-5.0 -2.9 -265
4.0 -0.8 270
-3.9 -4.0 -86
1.7 0.6 47
1.0 0.1 28
% of infants of current pregnancy weighing < 2500 g at bi rth 5.5 3.9 7.1 13.4 23.1 Recipient 6.4 4.9 21.4 2.4 23.1 Control 23 in pairs the *For this analysis all past pregnancies were considered to determine parity. In 2 pairs both women, in 22 pairs the recipient and control delivered only infants of low birthweight in the past; for the other analyses, therefore, they were considered nulliparous. tCalculated as follows: 182(recipient-only difference) + 13(recipient-and-control difference)j/95 - [70(control-only difference) + 13(recipientand-control difference)j/83. 572 CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125
prenatal care to be near zero. In fact, the mean birthweight of the infants of the 203 mothers who began receiving dispensary services in a later trimester than the one during which they registered for care was 119 g less than the control mean; this finding lends some support to the notion that these recipients were at inherently high risk. This 119-g difference was not significant, however, and the issue cannot be resolved by our data. While the relation of the duration of dispensary services to the outcome of pregnancy may be uncertain, there is a clear relation of birthweight to the number of visits to the clinic, an index of likely intensity of participation (Table IX). This relation was not confounded by the correlation of number of visits with duration of gestation, since we could control for the number of visits by controls. In another study, smaller than this one, a randomized controlled trial of nutritional supplementation in a poor black population in the United States,'5"6 there was no significant effect of the supplementation on birthweight, but the mean difference in birthweight between the offspring of women receiving a balanced protein and energy supplement and the offspring of controls was 41 g, nearly identical to the difference found in this study. (The women in the American study who received a high-protein supplement had infants whose birthweight was lower than that of infants born to the controls.) In the well designed study of Mora and colleagues'7 in a poor population living in Bogot'a, Colombia, nutritional supplementation was associated with a 51-g increase in mean birthweight (the increase was not significant by a two-tailed test). Another trial in a poor population in Taiwan raised the mean birthweight 51 g (calculated from the authors' data) by aggressive (and observed) supplementation.'8 The supplementation studies of Habicht and associates'9 in Guatemala were not controlled and thus were open to possible confounding. These authors contrasted women who chose to supplement their diets by 20 000 Cal (83.8 MJ) in the course of gestation with those not supplementing their diet that much: the birthweight advantage of the infants of women in the high-supplementation group was around 110 g. The effect on the average birthweight in the entire population of eligible women was about half that. The analysis is confounded by the lack of control for duration of gestation: with longer gestation, women had a greater opportunity to pass the threshold of 20000 Cal. Even so, the result in this study does not differ much from the results of the various controlled trials that did not depend on self-selection, and it probably represents the maximum change that might occur with nutritional supplementation in pregnancy, except in women who previously endured famine. It has been suggested that women with a history of delivering an infant of low birthweight in the past should be prime candidates for nutritional services, since their subsequent offspring would also be likely to have a low birthweight; such infants could therefore theoretically benefit more from a constant incremental gain in fetal weight (with a lower birthweight the risk
of death rises logarithmically).'9 Our results cannot lend support to this strategy, since the results among women with such a history were inconclusive; although the recipients' rates of delivery of low-birthweight infants were lower than the controls' rates, so were the mean birthweights of their infants. These differences were not significant. While ours was one of the largest studies of the effect of dietary intervention in pregnancy, it was too small to test this issue. When nutritional services are available, it is clearly better to offer them to women who are likely to be maximally responsive: the difference in mean birthweight between infants delivered by primiparas who weighed less than 140 lb at the time of conception and the infants of matched controls was 73 g, nearly double that for the total infant population and greater than that achieved in some frankly impoverished populations in developing countries.5 Indeed, in a review of all known studies of nutritional supplementation in pregnancy in which the choice of whether to supplement was not left to the women (that is, the confounding associated with selfselection was avoided) two patterns emerged. First, the use of high-density protein supplements (20% or more of the energy being supplied as protein) was uniformly associated with fetal growth retardation. This does not apply to the dispensary program, since it did not overly stress protein intake. Second, in all the reported studies in which the supplements used were of moderate or low protein density (less than 20% of the energy being supplied as protein) fetal growth acceleration was observed but was moderate, the differences in birthweight between the infants of recipients and controls ranging between 36 and 92 g in eight of the nine studies; the only exception was one very small Indian study, in which the difference was 540 g. These moderate effects were similar in poor populations in developing nations and in relatively affluent Western populations. We now have a better understanding of the changes in birthweight that can be achieved with the receipt of nutritional services during pregnancy. We must next decide which subgroups of women would most benefit from dietary services and whether the benefit to be derived from such services justifies their cost. Further research should be aimed at clarifying these benefits and costs. To this end, the school progress of the children in this study is now being evaluated. I am indebted to Agnes C. Higgins, director of the Montreal Diet Dispensary during the years under study, for both the initial invitation to study the results of the program of nutritional services during pregnancy that she developed, and for her continuing and open-hearted cooperation in this work; to Dr. John Strawbridge. department of psychology, Memorial University of Newfoundland, for critically reading this manuscript and suggesting several supplemental analyses, particularly those in Tables VIII and X; to Dr. Fred Naftolin, Dr. Robert Usher. Dr. Andre LaLonde, Miss Frances McLean and Mme Frances Prefontaine, all now or formerly of the Royal Victoria Hospital, Montreal; to Dr. John E. Moxley, MacDonald College, McGill University, Montreal; to Dr. Schuyler Kohl and Mr. Glenn Casey of the Obstetrical Statistical Cooperative, State University of New York Downstate CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125 573
Medical Center, Brooklyn; and to Susan Margolis, Mark Sadow, Hillel Cohen, Alan Mark, Heh Lee, Kiyomi Shimamoto, Patricia Cassano and Keith Callahan, of the division of epidemiology, Columbia University, New York. This research was supported by a grant from the National Foundation and by institutional funds from Columbia University.
Appendix: Method of Montreal Diet Dispensary, taken verbatim from Higgins8 I. Procedure for estimation of caloric and protein requirements Normal requirements: The normal caloric and protein requirements for mothers 20 years of age or more are determined on the basis of ideal body weight, physical activity, and week of gestation, according to the recommendations in the Dietary Standard for Canada (1948) prepared by the Canadian Council on Nutrition.t2. For mothers 19 years of age or less we use Recommended Dietary Allowances (1958) prepared by the Food and Nutrition Board, National Research Council, United States..21. For all mothers we add 500 calories and 25 g of protein after 20 weeks of gestation as recommended in the Canadian standard. Additional corrective allowances: Corrective caloric and protein allowances are given in addition to the normal requirements according to the degree of undernutrition, underweight, or for special high risk conditions which may be indicative of nutritional stress. A mother may have none or one or more of these conditions. II. Procedure for estimation of corrective allowances Undernutrition assessment and rehabilitation: Undernutrition is determined if a protein deficit is found between actual dietary intake and requirement. The method used is a 24-hour recall diet history, cross-checked with a food list and family market order compared with the appropriate standard. Undernutrition correction is equal to the amount of protein deficit allowing 10 calories for each gram of protein added to normal pregnancy requirements. Underweight assessment and rehabilitation: Underweight status is determined if the mother's pregravid weight is 5% or more under the weight recommended in the Table of Desirable Weights, prepared by the Metropolitan Life Insurance Company. Underweight correction should provide sufficient additional calories and protein to ensure that the mother gains during pregnancy the number of pounds she was underweight prior to conception. We allow 20 g of protein and 200 calories a day added to normal pregnancy requirements to permit a gain of one pound per week. Since 1971 the calorie correction has been changed from 200 to 500 calories. Nutritional stress assessment and rehabilitation: Nutritional stress is determined if any one of the following maternal conditions is present: pernicious vomiting, pregnancies spaced less than one year apart, previous poor obstetrical history, failure to gain 10 pounds by 20th week, serious emotional problems. Nutritional stress correction provides for the addition of 20 g of protein and 200 calories for each stress condition added to normal pregnancy requirements. III. Nutrition counselling When the mother's caloric and protein requirements have been determined, nutrition counselling begins. The 574 CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125
recommendations as to the kinds and amounts of food the mother should eat in order to meet her individual requirements for calories and protein are based on the mother's own dietary pattern and Canada's Food Guide. Changing eating patterns can be traumatic so only necessary minimum changes are made. Since we use milk principally to make up the deficiency in calories and protein between the mother's actual intake and her requirements, the major change our mothers have to make is to increase their milk consumption. Because our mothers are not accustomed to drinking milk, they find it easier to do when told that they are drinking the milk not for themselves but for the baby. It is very important that the nutritionist establish a good personal relationship with the mother. Mothers whose incomes are below the poverty level and therefore could not afford to buy the necessary food receive a food supplement of milk, eggs, and oranges, to make up the nutritional deficit. All mothers receive a vitamin and mineral supplement. We do not restrict salt or use diuretics. After the first interview, the mother is seen regularly and at each appointment a diet history is taken to determine if instructions are being followed. Constant encouragement and support are given by the nutritionist. At these interviews the weight of the mother is recorded and, if necessary, the food supplement is given. The supervision is continued until the baby is born. After the birth, a detailed report of the case is recorded on a summary sheet for subsequent analysis. Before the initial interview concerning assessment and counselling, a home visit is made to establish a friendly contact and to make arrangements for the first appointment. In our procedure it is necessary for the dietitian to practice both the science and the art of her profession. Scientific knowledge is needed so that the nutrient value of the kinds and amounts of foods recommended to each mother meet her nutrient requirements. However, the success of the counselling is determined by the dietitian's additional ability to influence the mother to follow the directions. The mother's compliance is dependent on the art of the dietitian to understand and motivate her patient. Understanding begins during the home visit and grows during the initial interview while gathering the information. From the process the dietitian becomes interested in the mother and develops a desire to help her. The mother responds to this concern and listens to the teaching and has faith in her dietitian's directions. IV. Level of nutrition During our service [in 1963-72], the mean daily intake for all cases was increased by 529 calories [from 2249 calories) and 33 g protein [from 68 g] but varied greatly according to nutritional status. For example, the average increase in intake in the undernourished group was 784 calories and 43 g of protein as compared with 210 calories and 16 g of protein for the good nutritional status group of mothers.
References 1. ABRAMOWICZ M, KASS EH: Pathogenesis and prognosis of prematurity. N Engi J Med 1966; 275: 868-885, 938-943, 1001-1007, 1053-1059 2. BERGNER L, SussaR MW: Low birth weight and prenatal nutrition: an interpretative review. Pediatrics 1970; 46: 946-966 3. Smn-t Z, SUSSER M, SAENGER G, MAROLLA F: Famine and Human
Development: the Dutch Hunger Winter oj' 1944-1945, Oxford U Pr, New York, 1975 4. STEIN Z, SUSSER M, RUSH D: Prenatal nutrition and birth weight:
experiments and quasi-experiments in the past decade. J Reprod Med 1978; 21: 287-299
5. RusI-i D: Effects of changes in protein and calorie intake during pregnancy on the growth of the human fetus. In CHALMERS I,
EN KIN M (eds): Effectiveness and Satisfaction in Antenatal Care
(Clinics in Developmental Medicine ser), Spastics International Medical Publications, London (in press) 6. SMITH CA: Effects of wartime starvation in Holland on pregnancy and its product. Am J Obstet Gynecol 1947; 53: 599-t08 7. KERR GRIEVE JF,
CAMPBELL BROWN
BM, JOHNSTONE FD:
Dieting
in pregnancy: a study of the effect of a high protein low carbohydrate diet on birthweight on an obstetric population. In SUTHER-
LAND HW, SrowERs JM (eds): Carbohydrate Metabolism in Pregnancy
and the Newborn: 1978, Springer-Verlag, New York, 1979: 518-534 8. HIGGINs AC: Nutritional status and the outcome of pregnancy. J Can Diet Assoc 1976; 37: 17-35 9. PRIMROSE T, HIGGINS A: A study of human antepartum nutrition. I Reprod Med 1971; 7: 257-264 10. HIGGINs AC, CRAMPTON EW, MOXLEY JE: Nutrition and the outcome of pregnancy. In Scow RO (ed), EBLING FiG, HENDERSON 1W
(co-eds): Endocrinology: Proceedings of the 4th International Congress of Endocrinology, Washington, D.C., June 18-24, 1972, American Elsevier, New York, 1973: 1071-1077
11. WYNN M, WYNN A: Nutrition Counselling in the Prevention 01
Low Birth-Weight, Foundation for Education and Research in ChildBearing. London, EngI, 1975 12. Idem: Nutrition and low birthweight. Action Mag Apr 1976; 16-20 13. RUSH D, HIGGINS AC, SADOW MD, MARGOLIs 5: Dietary services
during pregnancy, and birthweight: a retrospective matched pair analysis (abstr). Pedlair Res 1976; 10: 349 14. RUSH D, DAVIS H, SUSSER M: Antecedents of low birthweight in Harlem, New York City. mt J Epidemiol 1972; 1: 375-387 15. RUSH D, STEIN Z, SussER M: A randomized controlled trial of prenatal nutritional supplementation in New York City. Pediatrics 1980; 65: 683-697 16. Idem: Diet in Pregnancy: a Randomized Controlled Trial of Nutritional Supplements, AR Liss, New York, 1980 17. MORA JO, DE PAREDES B, WAGNER M, DE NAVARRO. L, SuascuN J,
CHRISTIAN SEN N, HERRERA MG: Nutritional supplementation and the outcome of pregnancy. 1. Birth weight. Am J Clin Nutr 1979; 32: 455-462 18. HERRIOTT RM, HSUEH AM, AITcHIsoN R: Influence of Maternal Diet on Offspring: Growtll, Behavior, Feed Efficiency and Susceptibility (Human); a Study in Suilin, Taiwan Initiated by Bacon F. Chow (final report on AID/CSD 2944 contract with The Johns
Hopkins University), 1978 19. HABICHT
JP,
LECHTIG
A,
YARBROUGH
C,
KLEIN
RE:
Maternal
nutrition, birthweight and infant mortality. In ELLIOTT K, KNIGHT
(eds): Size at Birth (Ciba Foundation Symposium, no 27), Elsevier, New York, 1974: 353-370 20. Canadian Council on Nutrition: A dietary standard for Canada. Can Bull Nutr 1950; 2: 6-34 21. Recommended Dietary Allowances, food and nutrition board, National Academy of Sciences-National Research Council, Washington, 1958
Hereditary pancreatitis: report of an affected Canadian kindred and review of the disease ROBERT M. GIRARD,* MD SERGE DUBI.,t MD ANDRII P. ARCHAMBAULT,t MD
Hereditary pancreatitis is an autosomal dominant disease with no other known cause. It usually begins in childhood and is characterized by recurrent attacks of abdominal pain of variable intensity and duration, followed by symptom-free periods. The diagnosis is usually made in early adult life, when pancreatic insufficiency and calcifications appear. Complications are less frequent than in nonhereditary chronic pancreatitis. There are also differences between the two forms of chronic pancreatitis in sex incidence, etiologic factors and life expectancy. In a Canadian kindred three generations are affected with hereditary pancreatitis; there are four definite and four suspected cases. More than 40 affected kindreds, including 195 proven cases and 190 suspected cases, have now been reported in the literature. Thus, hereditary pancreatitis should be considered in the differential diagnosis of chronic relapsing pancreatitis of unknown cause as well as recurrent abdominal pain in childhood.
le plus souvent durant l'enfance et se manifeste par une douleur abdominale recidivante d'intensite et de duree variables, entrecoupee de periodes asymptomatiques. La maladie est habituellement decouverte chez le jeune adulte, lorsqu'apparaissent les calcifications et I'insuffisance pancreatiques. Les complications sont moms frequentes que dans Ia pancreatite chronique non hereditaire. II existe aussi des differences entre Ia pancreatite chronique hereditaire et non hereditaire quant au sexe, aux facteurs etiologiques et a Ia duree de vie. Chez une famille canadienne quatre membres sont atteints et quatre autres sont soupconnes d'.tre porteurs de pancreatite hereditaire dans trois generations. Plus de 40 familIes atteintes, comprenant 195 cas prouves et 190 cas soup9onnes, ont ete decrites maintenant dans Ia litterature. Donc, Ia pancreatite hereditaire doit .tre envisagee en pr6sence d'une pancreatite chronique recidivante idiopathique et chez l'enfant presentant des douleurs abdominales inexpliquees.
La pancreatite chronique hereditaire est une maladie autosomique dominante sans autre cause connue. Elle debute
Hereditary pancreatitis, first described in 1952,' is an autosomal dominant disease characterized by recurrent episodes of epigastric pain that may radiate to the back; each episode lasts from a day to a week or longer. The diagnosis is often missed until pancreatic calcifications, pancreatic insufficiency and malabsorption occur, most frequently in early adult life. Since the critical review of 21 kindreds by Kattwinkel and colleagues in 1973,2 25 more kindreds have been reported in the literature.3'8 Apart from two Indian fain-
From the department of surgery and the division of gastroenterology, h6pital Maisonneuve-Rosemont, universit. de Montrial * Associate professor of surgery Assistant professor of surgery . Associate professor of medicine Reprint requests to: Dr. Robert M. Girard, 5415, boul. 1'Assomption, Montr.a1, PQ HiT 2M4 576
CMA JOURNAL/SEPTEMBER 15, 1981/VOL. 125
