Long-term Effects of Neonatal Sodium Restriction on Blood Pressure
Abstract In 1980, a randomized trial was conducted among 476 Dutch newborn infants to study the effect of a low or normal sodium diet on blood pressure during the first 6 months of life. At the end of the trial, systolic blood pressure in the low sodium group (n=231) was 2.1 mm Hg lower than in the control group (n=245). To investigate whether contrasting levels of sodium intake in infancy are associated with blood pressure differences in adolescence, we measured blood pressure in 167 children from the original cohort (35%) after 15 years of follow-up. We assessed the differences in systolic and diastolic blood pressure levels between the diet groups using a multivariate regression model with adjustment for potential confounders. The adjusted systolic blood pressure at follow-up was 3.6 mm Hg lower (95% confidence interval, −6.6 to −0.5) and the diastolic pressure was 2.2 mm Hg lower (95% confidence interval, −4.5 to 0.2) in children who had been assigned to the low sodium group (n=71) compared with the control group (n=96). These findings suggest that sodium intake in infancy may be important in relation to blood pressure later in life.
Despite extensive research, the causes of the epidemic of high BP in Westernized societies remain largely unknown. Population studies indicate that primary hypertension has its roots in childhood.1 2 3 There is evidence that BP levels in infancy and childhood are predictive for BP later in life.4 5 6 Consequently, determinants of BP in early life may be important in relation to future hypertension.
In a randomized, double-blind trial among 476 newborn Dutch infants conducted in 1980, sodium intake was positively related to BP during the first 6 months of life.7 In the present analysis, we investigated whether contrasting levels of sodium intake in infancy are still associated with BP differences later in life. For this purpose, BP in 167 youngsters who had participated in the original trial was measured again after 15 years of follow-up. In addition, possible modification of the effect of sodium on BP by sex, BMI, and resting HR was investigated.
A group of 476 newborn infants participated in a randomized, double-blind trial of the effect of sodium intake on BP. All infants were born in 1980 to healthy women living in Zoetermeer, Netherlands. Two hundred and thirty-one infants were randomly assigned to a low sodium diet and 245 to a normal sodium diet during the first 6 months of life. Systolic BP was measured in the first week of life and every 4 weeks thereafter with a Doppler ultrasound device connected to a random-zero sphygmomanometer. Diastolic BP could not be recorded accurately by this procedure. The study protocol has been described in detail previously.7
The average daily amount of sodium (±SD) consumed during the trial amounted to 0.89±0.26 mol in the low sodium group and 2.50±0.95 mol in the normal sodium group. After 25 weeks of intervention, systolic BP in the low sodium group was 2.1 mm Hg lower (90% CI, −3.7 to −0.5, P1=.01) than in the normal sodium group after adjustment for weight and length at birth, systolic BP in the first week, and BP observers.
Subjects for the Follow-up Study
A total of 466 infants (98%) completed the trial and were eligible for the present follow-up study. The current addresses of 185 children could not be traced. Seventy-one children had moved out of Zoetermeer, and they were not invited for logistic reasons. Eleven children could not be contacted by telephone. A total of 199 children was contacted (45% of those eligible). Twelve children refused participation. Twenty children could not participate because of illness, holiday, or other activities. A total of 167 children (84% of those contacted) agreed to participate, and they visited the study center for BP measurements. Of those, 71 children had initially been assigned to the low sodium group and 96 to the normal sodium group. All subjects gave informed consent.
BP and HR were measured on the right arm by two investigators using an automatic device (Dinamap model 8100, Critikon Inc) while the participant was seated. After at least 5 minutes of rest, four measurements of BP and HR were made; the last three were averaged and used in the present analysis. The observers did not know the original group assignment of the participants. Body weight and height were measured with participants not wearing heavy clothing and shoes. BMI was computed as weight divided by height squared.
During the center visit, the participant filled in a questionnaire regarding the use of medication, frequency of smoking and alcohol consumption, and amount of physical activity (sports, biking, and walking) and education at the time of follow-up. Girls additionally filled in the age at menarche. We also asked about antihypertensive treatment in parents and whether a salt-restricted diet was consumed at home. If the parents joined their child at the center visit, they provided this information themselves. When items on the questionnaire were left blank because the participant did not know the information requested, attempts were made to obtain this information from a parent by means of a telephone call. The educational level of the participants was rated on a five-point scale according to the Dutch school system for secondary education, ranging from lower vocational education to preuniversity education. The highest achieved education of both parents was assessed on a nine-point scale, ranging from primary school to university. The current educational level of the participant and the highest achieved educational level of the parents were recoded at three levels (low, average, and high) for data presentation. In the statistical analyses, the original rates were used.
The subjects collected an overnight urine sample after the center visit. Instructions were given both orally and in writing. The subjects were not informed about the purpose of the urine collection. Labeled containers were provided on which the subjects recorded the times the collection began and ended. When data were missing, they were obtained afterwards by means of a telephone call. The sample volume was measured at the study center. Urinary concentrations of sodium and potassium were determined by flame photometry with a Beckman KLiNa-flame device (Beckman Instruments Inc). Urinary creatinine concentration was measured by an automated semi-kinetic method based on a direct Jaffé procedure without deproteinization using a Kone Specific Analyzer and Kone reagents (Kone Corp Instrument Group). Electrolyte and creatinine excretions were standardized to 24-hour values.
The present study population was regarded as a cohort with two different levels of (experimentally manipulated) sodium intake early in life. Exposure was based on the initial randomization code, irrespective of subsequent protocol deviations or compliance. The difference in BP between the study groups after 15 years of follow-up was tested for statistical significance by means of a two-tailed unpaired t test.
The loss to follow-up of subjects for the present study, although for reasons unrelated to the object of the research, introduced unequal distributions of baseline and follow-up characteristics between the study groups. Therefore, we performed a second analysis, taking potential confounders into account. We used a multiple linear regression model, with systolic and diastolic BPs as the outcome variables. We made adjustments for sex, birth length and weight, maternal educational level, maternal systolic BP at baseline, educational level of the participant, presence of treated hypertension in parents, and BP observers. An indicator for the sodium group (0=low, 1=normal) was entered as an independent variable. The regression coefficient for the sodium group yielded by this model estimates the adjusted BP difference between the sodium groups.
The analyses were repeated for boys and girls separately and in BMI and resting HR strata. The strata were formed by dividing the participants into two groups according to the sex-specific median of the BMI and HR distribution at follow-up, respectively. The differences in BP are given with 95% CI and two-sided P values.
Baseline characteristics of the participants in the follow-up study are given in Table 1⇓. The low sodium group included more boys than the normal sodium group, but the difference was not statistically significant. Birth length and weight were significantly higher in the low sodium group than the control group. Furthermore, mothers had significantly lower education, and maternal systolic BP tended to be higher in the low sodium group. The other baseline characteristics were similar in the two groups.
The characteristics of the participants after 15 years of follow-up are given in Table 2⇓. No significant differences in body height and weight, BMI, and physical activity were observed between the study groups. The frequencies of smoking and alcohol consumption were similarly distributed over the groups. Among girls, the age at menarche was not significantly different between the groups. Significantly more participants in the low sodium group than the normal sodium group reported having a parent currently treated for hypertension. None of the participants reported antihypertensive treatment in both parents. The subjects in the low sodium group had significantly lower education than the control subjects. Urinary excretions of sodium and potassium in both groups were low. Urinary electrolyte values were lowest in the low sodium group but not significantly different from the values in the normal sodium group. Urinary sodium-potassium ratio and creatinine excretion were similar in both groups. One participant in each study group reported being on a salt-restricted diet (data not presented in table).
Table 3⇓ shows the differences in BP between the study groups after 15 years of follow-up. Systolic and diastolic BP levels were slightly lower in the low sodium group than the control group, but the differences were not statistically significant. After adjustment for potential confounders (indicated in Table 3⇓), systolic BP was 3.6 mm Hg lower (95% CI, −6.6 to −0.5, P2=.02) and diastolic BP was 2.2 mm Hg lower (95% CI, −4.5 to 0.2, P2=.08) in the low sodium group than the control group. The change in regression coefficients after adjustment in the multivariate model was attributable to the inclusion of birth length and weight, maternal educational level, maternal systolic BP, and presence of treated hypertension in parents. A small additional change was observed after further adjustment for the educational level of the participant, sex, and BP observers. When sodium and potassium excretions were additionally included in the model (with or without correction for creatinine excretion), the results did not change. The regression coefficient for sodium group did not differ significantly between boys and girls or between subjects with a low or high BMI.
The Figure⇓ presents the results within resting HR strata at follow-up. When the analysis was restricted to subjects with an HR above the sex-specific median (76 beats per minute for boys and 83 for girls), the adjusted differences in BP between the sodium groups were −6.0 mm Hg (95% CI, −10.5 to −1.5, P2=.01) for systolic and −4.8 mm Hg (95% CI, −8.7 to −0.9, P2=.02) for diastolic BP. In subjects with an HR lower than the median, the differences in BP were small (0.8 and 1.7 mm Hg for systolic and diastolic BPs, respectively) and not statistically significant.
This article presents the results of a long-term follow-up of a cohort of children who had been exposed to contrasting levels of sodium intake during the first 6 months of life. The study shows an association of sodium intake in infancy with systolic and diastolic BPs 15 years later in life, after adjustment for confounders. The observed BP effect was observed only in children with a relatively high HR.
The children remeasured for the present analysis form a subgroup of the original population that participated in a double-blind, randomized trial of low versus normal sodium intake.7 As such, the present study is a cohort study in which the contrast in baseline exposure, ie, neonatal sodium intake, has been experimentally achieved. Forty percent of the subjects initially assigned to the normal sodium group and 31% of the subjects in the low sodium group participated in the follow-up study. Whereas the study groups in the original trial were similar with regard to birth length and weight, maternal systolic BP, and maternal educational level, these variables differed significantly between the sodium groups in the follow-up study. Furthermore, the subjects in the low sodium group appeared to have less education and more often reported antihypertensive treatment in their parents. These observed differences could be related to the change in BP during childhood. The mother’s systolic BP level at baseline, for example, appeared to be positively associated with the child’s BP 15 years later in our study. Other researchers8 have also reported a significant correlation of maternal BP with BP in the adolescent child. Because the maternal BP level at baseline was higher in our low sodium group than the control group, the intervention-related difference in BP levels at follow-up was reduced. To reliably estimate the true effect of neonatal sodium intake on future BP, we repeated the analysis using a multivariate model that included all potential confounders. The conclusions of the present study are based on the adjusted BP differences obtained from the multivariate analysis.
We cannot infer from our findings which mechanism has led to higher BP in the normal sodium group. Possibly, differences in electrolyte intake between the study groups subsequent to the trial may have contributed to the differences in BP level at follow-up. However, there are several factors that do not favor a behavioral change. All babies had to adapt to a diet with a regular amount of salt after the trial, as the low sodium baby foods were not commercially available. Furthermore, the daily diet of the youngsters in the follow-up study to a large extent depended on their mothers’ food choices and cooking habits. Although it is possible that some mothers changed the family’s diet because of the trial, it is unlikely that this occurred differentially according to the study group to which the child had belonged. To get an indication of dietary electrolyte intake at the time of follow-up, the participants collected an overnight urine sample. No significant differences in sodium and potassium excretions as well as sodium-potassium ratio were observed. The mean sodium-potassium ratios in both groups were around 3.3, which is comparable to the average urinary sodium-potassium ratios in a Dutch cohort of children with more than 7 yearly measurements.9 Sodium and potassium excretions in both groups appeared to be relatively low, for which we have no explanation. No irregularities in the collection procedure or laboratory methods could be detected. Twenty-four-hour collections would probably have yielded a better estimate of habitual sodium and potassium intakes. Unfortunately, this appeared not to be feasible in these youngsters because of frequent outdoor activities. The urinary sodium-potassium ratio, which we consider the most important determinant of BP in this age group,9 is not likely to be affected largely by the relatively low values for urinary sodium and potassium or by the overnight sampling. We therefore think that we can validly infer from these data that the diets in both groups at the time of follow-up did not differ to a great extent with regard to their sodium-potassium balance. It would have been interesting to study possible differences in salt taste threshold or preference level between the low sodium and control groups after the trial. This, however, requires careful chemosensoric investigations and assessment of electrolyte intakes early and later in life. We did not obtain these data in our study.
Possibly, salt intake interacts with the sympathetic nervous system in determining BP level.10 DiBona11 suggested an interaction between intake of dietary sodium, environmental stress, and sympathetic neural control of renal function as part of the early hypertensive process. Staessen et al,12 in a population study, found a significant positive association between 24-hour urinary sodium and BP in subjects with high HRs, whereas an opposite tendency was observed in people with lower HRs. This interaction may also underlie the observed BP difference in our study, as the BP effect appeared to be restricted to participants with higher resting HR. It should be noted, however, that the stratification in our study was performed on HR measured at the time of follow-up and not on baseline values. We considered the HR at 15 years of age a better indicator of sympathetic nervous system activity than the HR measured in the early neonatal period. Because we did not stratify on HR before the intervention, it could be hypothesized that salt intake had an effect on BP through influencing HR. If this were true, however, the BP difference should no longer be observed within HR strata. Our findings show the opposite and therefore do not support an intermediate role for HR in the salt-BP relation. Alternatively, if HR is correlated with BP, the differences in HR between the groups at follow-up could be a marker for the BP response to the sodium intervention. This may explain why the BP effect was present only in the high HR stratum. Because we cannot exclude this alternative explanation, the results from the stratified analysis should be interpreted with caution.
Several alternative explanations for the observed BP difference in our study could be given. Exposure to high sodium levels in infancy may damage the immature kidney tissue, which in turn could adversely affect the development of BP during growth and maturation.13 The higher BP levels in the normal sodium group may also be explained by an alteration in renal sodium handling, resulting from early changes in renal hemodynamics. In line with this hypothesis, an increased renal vasoconstriction and decreased renin and aldosterone secretions have been observed in young offspring of hypertensive parents.14
Environmental exposures in utero and in infancy may be more important in relation to cardiovascular disease than exposures in adulthood, as “programming” of different systems and organs in the body occurs very early in life.15 Evidence is accumulating that birth weight is inversely related to the risk of hypertension later in life.16 17 18 The present study suggests that neonatal sodium intake could be related to BP level in adolescence. Data on the effect of nutrient intake in infancy on future BP are scarce. Studies in this field may provide clues for the prevention of hypertension in Westernized societies and a further understanding of its causes.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
The study in infants was supported by the Netherlands Heart Foundation (grant 78,041). We are grateful to Joke Jansen and Ria Rijneveldshoek for assistance in the follow-up study. We thank Robbert den Bleker for careful reading of the manuscript.
Reprint requests to Johanna M. Geleijnse, Department of Epidemiology and Biostatistics, Erasmus University Medical School, PO Box 1738, 3000 DR Rotterdam, Netherlands.
- Received June 6, 1996.
- Revision received June 27, 1996.
- Accepted October 10, 1996.
Szklo M. Epidemiologic patterns of blood pressure in children. Epidemiol Rev. 1979;1:143-169.
Labarthe DR. Blood pressure studies in children throughout the world. In: Gross F, Strasser T, eds. Mild Hypertension. New York, NY: Raven Press Publishers; 1983:85-96.
Gillman MW, Rosner B, Evans DA, Keough ME, Smith LA, Taylor JO, Hennekens CH. Use of multiple visits to increase blood pressure tracking correlations in childhood. Pediatrics. 1991;87:708-711.
Prineas RJ, Sinaiko AR. Hypertension in children. In: Swales JD, ed. Textbook of Hypertension.Oxford, UK: Blackwell Scientific Publications; 1994:750-766.
Miller JZ, Weinberger MH, Christian JC, Daugherty SA. Familial resemblance in the blood pressure response to sodium restriction. Am J Epidemiol. 1987;126:822-830.
Geleijnse JM, Grobbee DE, Hofman A. Sodium and potassium intake and blood pressure change in childhood. BMJ. 1990;300:899-902.
Poulter NR, Shipley MJ, Bulpitt CJ, Markowe HL, Marmot MG. Pulse rate and twenty-four hour urinary sodium content interact to determine blood pressure levels of male London civil servants. J Hypertens Suppl. 1988;6:611-613.
DiBona GF. Sympathetic neural control of the kidney in hypertension. Hypertension. 1992;19(suppl I):I-28-I-35.
Barker DJP. The fetal origins of adult hypertension. J Hypertens. 1992;10(suppl 7):39-44.
Whincup P, Cook D, Papacosta O, Walker M. Birth weight and blood pressure: cross sectional and longitudinal relations in childhood. BMJ. 1995;311:773-776.