Beyond birth-weight: early growth and adolescent blood pressure in a Peruvian population

Introduction

The prevalence of hypertension has increased dramatically in the past two decades among adults in developing nations (Kearney et al., 2004) and globally among adolescents and children (McNiece et al., 2007; Stranges & Cappuccio, 2007; Jackson, Thalange & Cole, 2007). Currently, high blood pressure is the leading risk factor for death globally and is the cause of an estimated 57 million disability adjusted life years (WHO, 2009).

Longitudinal investigations into the origins of adult essential hypertension have found elevated blood pressure in children to accurately track into adulthood (Bao et al., 1995), however the direct causes of essential hypertension in adolescence and adulthood remains unclear. The developmental origins of disease theory posits that characteristic growth patterns during the prenatal period (often proxied as birth weight or length), infancy (birth to two years), and childhood (2–9 years) predict subsequent risk for elevated adult blood pressure and essential hypertension (Eriksson et al., 2000; Adair & Dahly, 2005; Barker, 2006).

To date, the association of these three developmental periods on adolescent and adult blood pressure has varied in both reproducibility and magnitude. Low birth weight has been well established to predict higher adolescent and adult blood pressure (Eriksson et al., 2000; Adair & Dahly, 2005; Adair & Cole, 2003; Law et al., 2002; Adair et al., 2009; Huxley, Shiell & Law, 2000; Horta et al., 2003). However, the degree of association has differed across populations and some investigations were unable to replicate these findings (Hemachandra et al., 2006; Huxley, Neil & Collins, 2002; Menezes et al., 2007). Evidence regarding the influence of weight gain in infancy on adolescent and adult blood pressure remains inconclusive, with studies showing positive (Singhal et al., 2007; Kark, Tynelius & Rasmussen, 2009; Jones et al., 2012) and negative (Adair & Cole, 2003) associations, or no association (Law et al., 2002). Likewise, increased rates of childhood weight gain have also been linked to later blood pressure (Eriksson et al., 2000; Adair & Cole, 2003; Law et al., 2002; Huxley, Shiell & Law, 2000; Horta et al., 2003). Yet the clinical importance of this association remains difficult to determine, as increased weight gain is also highly predictive of obesity (Adair et al., 2009), a confounding factor, as BMI has been found to positively associate with blood pressure (Ford, Nonnemaker & Wirth, 2008).

We sought to address the following research question: how does growth rate in infancy and early childhood, independently, affect blood pressure and central adiposity in adolescence in a middle-income country? To answer this, we took advantage of a longitudinal cohort of Peruvian infants tracked in infancy via monthly growth measurements and revisited them at adolescence.

Materials and Methods

The study setting was Pampas de San Juan de Miraflores, a peri-urban shanty-town (pueblo joven) located 25 km south of Lima, Peru. Initially settled in the early 1990s, with temporary structures lacking water or sewage lines, the community has undergone many economic and social developments over the last two decades. By 2008, over 75% of homes were constructed from brick or cement with in-home water and sewage. Improved transportation within this community has changed patterns of exercise as “moto-taxis” have decreased the need to walk from main byways to the residential regions. The diet, while maintaining many typical aspects, now is more similar to an urban setting, an energy dense diet. The region continues its transition from a communicable to a chronic disease society, with current rates of hypertension ranging from 12.6% and 14.4 in private urban clinics (Schargrodsky et al., 2008), to 19.5 and 11.4 in residents of Pampas de San Juan de Miraflores, for men and women respectively (Davies et al., 2008). Further details of this community are described elsewhere (Gilman et al., 1993; Berkman et al., 2002; Checkley et al., 2003).

Results

Cohort characteristics

In the 2008–2009 follow-up visit, we completed measurement on 75% (147 out of 196) of the eligible participants (Fig. 1). We found no differences in early change in height (p = 0.387) or weight (p = 0.834), maternal height (p = 0.172), the percentage of households without sewage connection (p = 0.077) or in the type of house flooring (p = 0.860) between the 147 participants who completed follow-up and the remaining 49 (25%) participants of the original cohort that met inclusion criteria but did not participate in the follow-up assessments.

Table 1 shows summarized characteristics of the participants. Average age at enrollment for the group was 13.4 days and duration of diarrheal surveillance was 24.7 months. During the initial study in 1995, 46.9% of participants were stunted at some point in infancy. Rates of underweight and overweight were 5.5 and 57.2%. The rate of stunting at follow-up was 13.6%, while 1.4% and 10.2% were underweight vs. overweight, respectively.

Table 1:

Participant characteristics at baseline and follow-up.

	Total	Male	Female	p-value
Sample size	147	81	66
Baseline 1995–1998
Age of entry into study (days), mean ± SD	13.4 ± 14.4	14.6 ± 15.2	11.9 ± 13.4	0.01
Follow-up time (months), mean ± SD	24.7 ± 5.7	23.7 ± 6.4	26.1 ± 4.5	0.033
Birth weight for length z-score (WLZ), mean ± SD	0.79 ± 1.0	0.58 ± 1.0	1.1 ± 1.0	0.066
Birth length for age z-score (LAZ), mean ± SD	−0.14 ± 0.9	−0.21 ± 0.9	−0.05 ± 1.0	0.313
Stunted from 0 to 30 months (LAZ < −2 SD), %	46.9 ± 0.5	51.9 ± 0.5	40.9 ± 0.5	0.189
Overweight from 0 to 30 months (WLZ > 2 SD), %	57.2 ± 0.5	53.7 ± 0.5	61.5 ± 0.5	0.349
Diarrhea prevalence, %	2.39 ± 2.57	2.41 ± 2.40	2.36 ± 2.79	0.33
Diarrhea episodes per month, mean ± SD	0.27 ± 0.22	0.27 ± 0.21	0.26 ± 0.23	0.622
Follow-up 2008–2009
Age (years), mean ± SD	13.4 ± 0.76	13.3 ± 0.70	13.4 ± 0.70	0.379
Height (cm), mean ± SD	150.4 ± 7.4	151.1 ± 8.2	149.5 ± 6.2	0.188
Stunted (LAZ < −2 SD), %	13.6 ± 0.3	8.6 ± 0.3	19.7 ± 0.4	0.052
Overweight (BAZ > 2 SD), %	10.2 ± 0.3	12.3 ± 0.3	7.6 ± 0.3	0.345
Systolic blood pressure (mm Hg), mean ± SD	103.3 ± 8.8	104.0 ± 8.1	102.4 ± 9.6	0.249
Diastolic blood pressure (mm Hg), mean ± SD	63.0 ± 7.2	63.0 ± 6.9	63.0 ± 7.6	0.987
Waist circumference (cm), mean ± SD	71.1 ± 7.9	70.8 ± 7.9	71.4 ± 8.0	0.679
Waist to hip ratio, mean ± SD	0.84 ± 0.05	0.86 ± 0.05	0.82 ± 0.05	<0.01
Socioeconomic traits 2008
Educationally-deprived mother, % (n)	68.5% (102)	71.1% (59)	65.2% (43)	0.405
Child repeated a grade, % (n)	26.2% (39)	21.7% (18)	31.8% (21)	0.164
People per room ≥3, % (n)	6.0% (9)	3.6% (3)	9.1% (6)	0.166
In home sewage, % (n)	87.3% (130)	88.0% (73)	86.4% (57)	0.842

DOI: 10.7717/peerj.381/table-1

Discussion

Our longitudinal study aimed to assess the effects of early growth on adolescent blood pressure and central adiposity by taking advantage of serial measurement of anthropometric indicators, with data accruing monthly for a period of up to 30 months, and thus overcoming limitations posed by single ascertainment of birth weight or modelling techniques that approximate growth patterns. We found increased rates of weight gain in infancy were strongly associated with decreased systolic blood pressure and waist circumference but not diastolic blood pressure, 10 years later, at adolescence. Weight gain in early and late infancy, respectively, predicted decreases in systolic blood pressure of 1.3 and 2.5 mm Hg, and decreases in waist circumference of 0.6 and 1.2 cm. This is consistent with findings observed in a population of Filipino adolescent males (Adair & Cole, 2003) where increased weight gain from 0 to 2 years of age resulted in decreased odds of high adult blood pressure in males. Notably, no association was found between weight gain and diastolic blood pressure, which recent evidence suggests may be a stronger correlate to later mortality than systolic blood pressure (Sundstrom et al., 2011). The magnitude of the decrease in systolic blood pressure observed in this study, via improved weight gain during infancy, is equivalent to that achieved via resource-intensive community-based lifestyle interventions in Pakistan (Jafar et al., 2010).

The public health implications of these findings are substantial given that blood pressure in teens has been found to track into adulthood. The magnitude of reductions in blood pressure observed in our cohort could be linked to significant gains in terms of cardiovascular disease outcomes. In a meta-analysis of 61 prospective observational studies of blood pressure and vascular disease in adults, for each 2 mm Hg decrease in systolic blood pressure, to a minimum of 115 mm Hg, stroke mortality and cardiovascular mortality decreased by 10% and 7%, respectively (Lewington et al., 2002).

Our findings contrast with recent reports from Belarus (Tilling et al., 2011) and the United Kingdom (Jones et al., 2012) in which growth from birth was associated with increases in blood pressure in childhood and at early adolescence, respectively. Their findings from settings with advanced child health outcomes, e.g., low rates of stunting and child mortality, may indicate that, in healthy infants, increased rates of early growth pose increased cardiovascular risk as there is less need for compensatory growth following growth delays and stunting, both commonly encountered in the developing country settings. On the contrary, additional evidence from other similar resource-poor settings aligns with our observations. Albeit weak, hospital admissions due to diarrhea during the first year of life have been linked to high blood pressure in childhood (Miranda et al., 2009), and undernutrition has been linked to increased cardiovascular risk (Sawaya et al., 2003).

Much as growth during development has been correlated with later blood pressure, it has also been found to predict an increased risk of overweight and obesity. We found increased weight at birth and rate of weight gain in early and late infancy to be associated with decreased adolescent waist circumference, a measure of central adiposity and cardiovascular risk (Huxley et al., 2010). Birth length and growth in length in infancy were not associated with the degree of central adiposity in adolescence. These findings indicate that in this population, of generally undernourished children, rapid weight gain during infancy may help decrease the risk for central adiposity in adolescence, possibly by preventing compensatory weight gain later in childhood when the effects on cardiovascular risk are greater. This is consistent with findings in Brazil, in which rapid growth during infancy in males was associated with increased adolescent lean mass, while rapid growth in childhood was associated with fat mass (Wells et al., 2005). This appears to indicate that increased growth rates early in development have a protective effect on adult cardiovascular risk factors such as stunting (Sterling et al., 2012), obesity (Wells et al., 2005) and hypertension (Adair & Cole, 2003).

Our study has a number of strengths. While other studies looking at the developmental origins of disease relied, at most, on yearly anthropometric measurements, frequent serial measurement of our cohort afforded us enough power to find effects of early growth on systolic blood pressure. The use of short interval serial measurements early in life and a relatively novel methodology offered an accurate characterization of growth early in development. This characterization of growth benefited uniquely from the longitudinal diarrheal data, which was incorporated in our modeling to account for growth variations due to infection. Thus, our work provides further advancements on the relationships of early life exposures to adolescent outcomes in countries with a high burden of childhood chronic under nutrition.

We also considered various potential limitations to our analysis. First, we did not directly measure birth weight or length, however there was a mean age of entry into the study of 14 days, with many individuals entering in the first days of life. Second, our estimations were adjusted for maternal education and overcrowding as proxies of long-term and current SES respectively. We considered that since the time of the initial study, maternal education was unlikely to have changed and thus a good proxy of SES at baseline (Howe et al., 2012). To account for socioeconomic variation over time, since the community was settled, we used number of people per room, a measure of SES at follow-up. Despite this approach, we did not, however, address socioeconomic change over time and this may have introduced some bias. As described earlier, most of the community setting progressed to meet basic needs by the time of follow-up, yet in general this community remains a resource-deprived area compared to Lima (Instituto Nacional de Estadística e Informática, 2007). Third, we did not include breastfeeding as an explanatory variable in our model; however, since both type and duration of breastfeeding are strongly associated with growth patterns in early childhood (Baker et al., 2004; Singhal & Lanigan, 2007), breastfeeding is likely in the causal pathway of the relationship growth trajectories in early childhood and blood pressure in later life. Hence, we opted not to include breastfeeding in our analysis. Moreover, a large, long-term follow-up of 13,879 mother–infant pairs who were randomized to receive a breastfeeding promotion intervention vs. usual care did not prevent overweight or obesity, nor did it affect IGF-I levels at 11.5 years of age (Martin et al., 2013). We also did not formally assess degree of pubertal development, which may have altered our outcomes in regards to growth as well as blood pressure. Finally, our sample size did not enable sex stratification that could disentangle potential differences as with Filipino adolescents (Adair & Cole, 2003).

As has been described by Lucas and colleagues (Lucas, Fewtrell & Cole, 1999), the inclusion of current size in the regression model is an adjustment for growth between the time of exposure and time of follow-up. Consistent with findings in other populations, adjustment for current size caused the beta coefficient for WLZ at birth to shift sign from positive to negative (Lucas, Fewtrell & Cole, 1999). This positive to negative shift indicates that, holding childhood growth constant, increased weight growth in the first 30 months of life may have a protective effect on adolescent blood pressure levels.

In low- and middle-income countries undergoing rapid economic development there is an increased risk of concurrent stunting and overweight, due to infectious disease associated growth retardation early in development (Victora et al., 2008) followed by overnutrition in adolescence and adulthood (Popkin, 2001). Rapid compensatory growth in underweight children has been posited to increase the risk of long-term cardiovascular morbidities such that nutritional intervention may do more harm than good (Singhal et al., 2007). However, we found increased weight growth during infancy and early childhood to predict decreases in adolescent systolic blood pressure and central adiposity. The impact of such findings at the population level, as indicated by the magnitude of the change in systolic blood pressure observed in this study, aligns with significant decreases in vascular mortality in the future (Lewington et al., 2002). Thus, in addition to the known short term (Victora et al., 2001) and long term (Martorell et al., 2010a; Martorell et al., 2010b) benefits of increased growth in underweight children, increased weight growth early in development may also decrease rates of risk factors for cardiovascular disease and its associated health outcomes.

Novelty and Significance