A negative impact of altitude on birth weight has been demonstrated for medium to high-altitude countries.
The present study aims to show a similar effect for a lower altitude range up to 1600 m in the country of Austria and to adjust for several further risk factors related to mother and infant. In addition, we analyzed whether the effect of altitude influenced birth weight independently or interacted with other factors. For the purpose of this study, almost 1.9 million individual birth certificates of Austrian newborns born between 1984 and 2013 were analyzed. In a multivariate linear regression model, birth weight was associated with altitude of living address and following variables: sex, birth length, gestational age, level of education, maternal age, year of birth, parity, time to previous birth and marital status.
Birth weight decreased by 150 g/1000 m altitude, demonstrating a clear effect of altitude on birth weight also in a low to medium altitude level. Additionally, we could show that this effect also depends on other factors, namely gestational age, education of the mother, sex, birth length, year of birth and time to previous delivery. All variables interacted significantly (p<0.0001) with altitude.
We observed a negative effect of altitude in an altitude range up to 1600 m. Furthermore, this effect also depends on other risk factors. Therefore, unadjusted estimates as described in many studies may be biased. This population-based study describes the effect of low-to-medium altitude on birth weight in central Europe over a period of 30 years.
The negative impact of increasing altitude on birth weight has been described in numerous studies.1, 2, 3, 4, 5, 6, 7, 8 The observed decrease in birth weight is in the range of 50 to 200 g per 1000 m increase in altitude1, 3, 4, 5, 6, 7, 8, 9, 10 and depends strongly on the study design as well as on the range of altitude covered.
This effect on birth weight of infants born at higher altitudes is generally attributed to a lack of oxygen rather than an increase in other risk factors possibly contributing to low birth weight,6, 11 although the mechanisms by which altitude is retarding fetal growth are not completely understood yet.7 One of the underlying hypothesis is that the reduction in birth weight results from failure of the maternal oxygen transport system to meet the increased fetal demands at higher altitudes.11 The literature suggests that the systemic circulatory response to pregnancy is altered in high altitude—increment of cardiac output is lower and the mean blood pressure is higher in high-altitude pregnant women—contributing to a reduced uteroplacental delivery of oxygen and other nutrients.11, 12 This is in part compensated by an increased vascularization and proliferation of the cytotrophoblast13 and by hemodynamic changes (such as cerebral vasodilation and pulmonary vasoconstriction) in the pregnant woman and her fetus at high altitude.14 Birth weight seems to depend on availability of both oxygen and glucose. Maternal fasting glucose levels are shown to be lower at higher altitude in the presence of similar insulin secretion.7, 15
The altitude-associated decrease in birth weight has been noted worldwide, being greatest in shorter staying high-altitude residents and lowest in the longest-staying high-altitude residents (for example, Tibetans), suggesting compensatory genetic or epigenetic processes.11
Epidemiological studies revealed that low birth weight significantly contributes to perinatal mortality and morbidity.7 Whether altitude-associated growth retardation also impairs infant mortality and morbidity is controversially discussed.11
In the Andean6, 13 and Tibetan mountains,11 studies cover altitude ranges up to over 4000 m and some authors note that the effect seems to start at altitudes above 1000 to 2000 m.1, 10, 16 It is proposed that birth weight decreases only after reaching a threshold for a critical barometric pressure reaching a hypoxic effect, which exists approximately at 2000 m.7
Many maternal, fetal and environmental factors can influence birth weight.17 Demographic differences between lowland and highland populations, such as income and availability of health care among others, interfere when correlating altitude and birth weight. Hypertension, preeclampsia, primiparity, low maternal age, socioeconomic level, lack of prenatal care, maternal nutrition, weight gain during pregnancy and maternal smoking are well-known predictors of low birth weight.7 Altitude is only one of many factors influencing birth weight18 but may be considerably larger than any other risk factor known for low birth weight.1, 7, 10
Despite the surprisingly large effect of altitude on birth weight, most studies do not question whether this effect depends on other interacting factors related to mother and infant. Only very few studies tried to answer this question2, 7, 10, 16 and conclude that, apart from population group effects, altitude acts independently from other variables. In a recent medium-sized study10 using data from seven countries in South America, the authors compare two altitude samples and suggest interactions between infant health inputs and altitude, but do not further quantify these results.
In another pilot study19 the effect of altitude on birth weight has been analyzed for Austrian newborns born between 1999 and 2008. This study showed concordance with former studies in showing a negative influence of increasing altitude on birth weight, although the Austrian population resides in a low-to-medium altitude range.
The aim of the present study was, first, to assess whether altitude influences birth weight in a low-to-medium altitude range as can be found in the country of Austria, and, second, to quantify whether the effect of altitude depends on other factors related to mother and infant using almost 1.9 million individual birth certificates in the period from 1984 to 2013 in Austria.
All data are retrieved from individual birth certificates collected between 1984 and 2013. These are routinely collected by the Austrian statistical office (Statistics Austria20) and are provided in an anonymized form for scientific research.
Parameters and definitions
The following demographic details defined inclusion criteria and were collected to build the data set:
Austrian citizenship of mother, maternal age 15 to 45 years, live birth, singleton, gestational age 36 to 44 weeks, birth length 42 to 60 cm, birth weight >1000 g, parity 1 to 3. From 1984 to 1998, birth weight was measured in steps of 100 g, from 1999 to 2010 in steps of 10 g and since 2011 in steps of 1 g.
Altitude was calculated as the altitude in meter of the centroid of the municipality of the mother's living address.
The effect of altitude on birth weight was modeled by means of a multivariable linear model in SAS21 including 10 independent variables with following groups of variables:
(i) Infant-related: birth weight (kg, dependent variable), sex, birth length (cm), altitude (m) and gestational age (weeks).
(ii) Mother-related: level of education (compulsory schooling, apprenticeship, vocational education, secondary school, college, university and unknown), age (years), year of birth of newborn (grouped into 3-year groups beginning with 1984–1986 to 2011–2013), parity (1, 2 and 3), time to previous birth (years) and marital status.
All variables were treated as categorical variables (n=8), except altitude and age of mother in order to provide sufficient fit of the model and to allow estimation of possible nonlinear associations. The hypothesis whether altitude acted additively or interactively was tested by an interaction term altitude × variable leading to nine possible interaction terms for the analysis. For each value of the categorical variable as well as for each combination with altitude (that is, interaction) one parameter had to be estimated leading to a total of n=114 parameters in the model. The effect of age of mother and altitude was modeled by a linear and squared term. In addition, a cubic term was added for altitude in order to improve fit of the model.
A very large number of observations lead to a very large power of the test statistics, allowing detection of very small effects. Bauer et al.22 showed that for that kind of analyses the P-values for variable selection must approach zero in order to ensure consistent model selection. Therefore, selection of variables containing interactions terms was carried out by choosing a low significance level of P=0.0001. Main effects were kept in the model irrespectively of their P-value.
Because of the large number of estimated parameters (n=114), the full regression model results were not shown but are available from the authors on request.
Inclusion criteria described above were applied in order to provide sufficiently large numbers of observations for the regression model for each subgroup formed by the categorical variables and therefore to provide robust results for the interaction terms.
The estimated interaction effect of altitude on the mean birth weight is described by showing the effect of virtually changing the living address from 200 to 1200 m, with all other mother- and infant-related factors being unchanged. Depending on the interacting variable the interaction effect is calculated for a child with the following arbitrarily set characteristics: male, gestational age=40 weeks, year of birth 1984 to 1986, length=50 cm, mother-related characteristics: parity=1, time to previous birth=2 years, age =27 years, education=vocational, marital status=married.
Descriptive statistics for birth weight in dependence on altitude are given by means and 99.9% confidence intervals.
From 2 515 242 births in the period 1984 to 2013, 1 881 370 fulfilled the selection criteria. A total of 264 observations had one or more missing values, so that the number of birth certificates used in the regression analysis dropped to 1 881 104.
Altitude of living addresses varied between 117 and 1628 m with a median of 380 m; half of all mothers had a living address between 200 and 500 m and roughly 2% of all mothers had a living address higher than 1000 m (Table 1).
Effect of altitude from univariate regression model
The unadjusted mean birth weight fell from 3386 g at 150 m altitude to 3147 g at maximum altitude of 1650 m corresponding to a decrease in the mean weight of roughly 150 g per 1000 m (Figure 1). We used the following regression model to describe birth weight:
Weight (kg)=3.37267+0.145297 × Altitude−0.317762 altitude2+0.123457 × altitude3 (in km).
In the range 150 to 300 m, the observed weight begins to decrease linearly to the maximum altitude of ~1600 m. Confidence intervals increase with altitude as sample sizes decrease.
Interaction effects from multivariate regression model
All variables included in the linear regression model were significant at the P<0.0001 level with the exception of the interaction terms altitude × marital status, altitude × age of mother and altitude × parity (all P>0.05). Therefore, these interaction terms were dropped, leaving 10 significant main effect and six interaction terms in the regression model.
In the following we describe significant interaction effects of altitude on the mean birth weight and show the effect of fictitiously changing the living address from 200 to 1200 m.
Effect of gestational age
In Figure 2 the interaction effect of altitude and gestational age is shown. A change from 200 to 1200 m results in a decrease in the mean birth weight of ~200 g for children born at 36–37 weeks. The birth weight difference decreases to ~150 g for full-term children.
Effect of education of mother
In Figure 3 the combined effect of altitude and education of mother is shown. Mothers with missing information concerning education show a much smaller decrease (39 g) between low and high altitude than the rest of the educational-level groups (121 to 154 g).
Effect of sex
For boys the estimated effect on birth weight between low and high altitude is 155 g, for girls it is 187 g.
Effect of length of newborn
The interaction effect of altitude and length of newborn is calculated for a boy with a gestational age of 40 weeks (Figure 4). At a length of 47 cm the difference between low and high altitude is 68 g and increases linearly to 291 g for newborns with a length of 54 cm.
Effect of year of birth
The combined effect of altitude and year of birth (Figure 5) decreases from 155 g in the period 1984 to 1986 to 92 g in the last period 2009 to 2013.
Effect of time to previous birth
The interactive effects of altitude and time to previous birth (Figure 6) consists of a decrease in birth weight of 148 g for children born 1 to 1.5 years after their previously born sibling and increases to 175 g for children born 10 to 13 years after their sibling.
The main result of our study showing that birth weight is influenced by altitude is in accordance to numerous studies.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 16 We found an almost linear negative influence on birth weight in a low-to-moderate altitude residence level as existing in Austria. In our study, as described by many other studies1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 16 an unadjusted decline in birth weight of 150 g per 1000 m of altitude could be demonstrated.
Many of the so-far published studies are comparing two different altitude levels and mainly data from very high-altitude levels (2500 to 4500 m). Therefore, it is not possible to clarify whether the reduction in fetal growth occurs only above a critical altitude.16 It is proposed that birth weight decreases only after reaching a certain threshold of a critical barometric pressure reaching a hypoxic effect, which is supposed to start at ~2000 m.7 In healthy adults, significant arterial oxygen desaturation starts at an altitude range of 1000 to 1500 m;1 therefore, the mechanisms of the demonstrated reduction of birth weight in a low-to-moderate altitude level cannot be attributed only to a decrease in fetal oxygenation. A second explanation for the association of birth weight and altitude is referring to a lower glucose delivery to the fetus, as pregnant women show a lower fasting glucose at higher altitude in the presence of similar insulin secretion.15 Notzon et al.23 are reporting that an increase in altitude of as little as 500 m affects not only the proportion of births under 2500 g, but also causes the entire birth weight distribution to shift leftwards. Residence in high altitudes results in different compensatory mechanism. Increment of cardiac output is lower and the mean blood pressure is higher in high-altitude pregnant woman contributing to a reduced uteroplacental delivery of oxygen and other nutrients,11, 12 which in part is compensated by an increase in hematocrit and erythropoietin,14 increased vascularization of the placenta13 and by hemodynamic changes (such as cerebral vasodilation and pulmonary vasoconstriction) occurring in the pregnant woman and her fetus at high altitude.14 The altitude-associated decrease in birth weight has been noted worldwide, being lowest in the longest-staying high-altitude residents, suggesting ‘acute’ and ‘chronic’ adaptive processes. As we had no access to prenatal or genetic data, we can only speculate that the decline in birth weight from low-to-medium altitudes as shown in this study, is based on the same factors as when occurring in very high altitude.
In addition, we wanted to analyze whether the effect of altitude on birth weight acted additively or interactively to other factors. Our data show several significant as well as relevant in size interactions between the effect of altitude on birth weight and other factors related to mother and infant.
First of all, a significant interaction of gestational age on the effect of altitude on birth weight could be demonstrated. A change from 200 to 1200 m resulted in a decrease in the mean birth weight of ~200 g for children born at 36 to 37 weeks and only 150 g for full-term infants. Generally, the effect of altitude on birth weight is not due to a shorter gestation or higher risk of preterm birth as it also was not the case in our cohort; however, analysis showed a greater weight difference in birth weight (~200 g) between 200 and 1200 m in younger (=37 weeks of gestation) infants when compared with infants born at 40 weeks of gestation with a difference of only ~150 g. Krampl et al.6 have shown that the trajectory of fetal growth began to slow at weeks 25 to 29 in babies born in high altitude, and it is reported that the greatest reduction in fetal growth rate is seen during the third trimester of pregnancy.2 Intrauterine growth charts demonstrate a significant growth spurt between 30 to 32 and 37 to 38 weeks, slowing down again between 38 and 40 weeks, which explains greater variability in birth weight being born earlier and greater influence of any factor attributing to birth weight at younger gestational ages24 without reducing the length of gestation in general.
Furthermore, a significant interaction of maternal education on the effect of altitude on birth weight could be demonstrated in our study, showing a much smaller difference in birth weight at different altitudes in mothers where information is missing when compared with the rest. Within the group of mothers with non-missing maternal education no relevant interaction effect could be observed. Interpretation of the striking difference between observations with missing and non-missing maternal education is difficult. The percentage of missing maternal education has increased from ~1 to 2% at the beginning to ~11% at the end of the study period. Inquiries at Statistics Austria showed that in the last years midwives tended to ignore the question concerning maternal education because of time restrictions. Since paternal education as a proxy is missing also in most cases of missing maternal education, we do not have any information which group of mothers may be responsible for the observed interaction effect. Nevertheless, the inclusion of maternal age in our regression model partially adjusted for maternal education so that the effect of missing maternal education should have been explained at least to some extent by maternal age.
Moreover, a significant interaction of the infant’s sex on the effect of altitude on birth weight could be demonstrated in our study, showing a larger difference in birth weight at different altitudes for girls than for boys. This is in contrast to published data, showing that male fetuses are gaining weight faster than girls under similar conditions and, therefore, should be more prone to any influence on birth weight than girls.17 In a study by Roland et al.25 it was shown that boys were heavier and had a higher fetal:placental ratio than girls, indicating that boys have more efficient placentas than girls. Ericsson et al.26 also observed that boys tended to be longer and have larger head circumferences at birth for any placental weight. They suggested that boys invest in brain growth rather than placental growth, which leaves them with less reserve capacity and makes them more vulnerable to undernutrition. However, Roland et al.25 could also show that fasting glucose was associated with birth weight only in girls, but not in boys. Thus, glucose has an effect on both placental weight and birth weight only among girls, which might at least in part explain the results found in our study.
Another significant interaction on the effect of altitude on birth weight was found for birth length of the infant, resulting in a higher birth weight difference between the altitudes with increasing birth length of the infant. This might also be explained by the human growth spurt in the last trimester resulting in a higher weight gain than length gain during late pregnancy and therewith a higher variability of birth weight compared with birth length. Intrauterine growth retardation reflecting maternoplacental insufficiency is usually affecting mostly weight and length. In contrast, especially head circumference is preserved resulting in an asymmetric growth retardation sparing brain growth at the expense of the fetal trunk.2 Galan et al.27 also demonstrated that the reduction in birth weight in higher altitude is mainly due to a reduction in fetal subcutaneous fat tissue and not lean mass, resembling an early or mild effect of fetal undernutrition, which does not influence birth length. This might best resemble the situation at medium altitude as occurring in Austria and therefore might explain the results of our study.
As a further result, we found a joint effect of altitude and year of birth with an almost linear decrease in birth weight variance at different altitudes from the late 1980s to 2013. This might be explained by a generally better prenatal care with more awareness to intrauterine deficits over the last three decades.
At last, we found a significant interaction of altitude on birth weight and time to previous birth showing an increasing birth weight difference between the altitudes with increasing time to previous delivery. Zhu et al.28 demonstrated a significant effect of the interval between pregnancies on perinatal outcomes including low birth weight. Infants conceived at 18 to 23 months after a previous life birth had the lowest risk of adverse perinatal outcomes. Shorter and longer interpregnancy intervals were associated with higher risk for small size for gestational age.
The strength of our population-based study is the very large number of births including almost all term live births in Austria within the study period of 30 years and the analysis of the effect of altitude in interaction to other factors in a statistically adjusted model based on individual data. The limitation of our study is the non-availability of other known risk factors influencing birth weight as, for example, maternal and paternal biometric measures, gestational weight gain and nutrition, maternal hypertension, preeclampsia, cigarette smoking, alcohol consumption and general morbidity of the mother.
In summary, our data demonstrate a clear effect of altitude on birth weight in a low-to-medium altitude level in a European country. In addition, we could show that this effect is influenced by other factors related to mother and infant, namely gestational age, education of the mother, sex, birth length, year of birth and time to previous delivery.
Yip R . Altitude and birth weight. J Pediatrics 1987; 111 (6 Pt 1): 869–876.
Giussani DA, Phillips PS, Anstee S, Barker DJ . Effects of altitude versus economic status on birth weight and body shape at birth. Pediatric Res 2001; 49 (4): 490–494.
Gonzales GF, Tapia V . Birth weight charts for gestational age in 63,620 healthy infants born in Peruvian public hospitals at low and at high altitude. Acta Paediatr 2009; 98 (3): 454–458.
Jensen GM, Moore LG . The effect of high altitude and other risk factors on birthweight: independent or interactive effects? Am J Public Health 1997; 87 (6): 1003–1007.
Khalid ME, Ali ME, Ali KZ . Full-term birth weight and placental morphology at high and low altitude. Int J Gynaecol Obstetr 1997; 57 (3): 259–265.
Krampl E, Lees C, Bland JM, Espinoza Dorado J, Moscoso G, Campbell S . Fetal biometry at 4300m compared to sea level in Peru. Ultrasound Obstetr Gynecol 2000; 16 (1): 9–18.
Lopez Camelo JS, Campana H, Santos R, Poletta FA . Effect of the interaction between high altitude and socioeconomic factors on birth weight in a large sample from South America. Am J Phys Anthropol 2006; 129 (2): 305–310.
Zahran S, Breunig IM, Link BG, Snodgrass JG, Weiler S . A quasi-experimental analysis of maternal altitude exposure and infant birth weight. Am J Public Health 2014; 104 (Suppl 1): S166–S174.
Niermeyer S, Andrade Mollinedo P, Huicho L . Child health and living at high altitude. Arch Dis Child 2009; 94 (10): 806–811.
Wehby GL, Castilla EE, Lopez-Camelo J . The impact of altitude on infant health in South America. Econ Hum Biol 2010; 8 (2): 197–211.
Moore LG . Fetal growth restriction and maternal oxygen transport during high altitude pregnancy. High Alt Med Biol 2003; 4 (2): 141–156.
Zamudio S, Postigo L, Illsley NP, Rodriguez C, Heredia G, Brimacombe M et al. Maternal oxygen delivery is not related to altitude- and ancestry-associated differences in human fetal growth. J Physiol 2007; 582 (Pt 2): 883–895.
Zamudio S . The placenta at high altitude. High Alt Med Biol 2003; 4 (2): 171–191.
Galan HL, Rigano S, Chyu J, Beaty B, Bozzo M, Hobbins JC et al. Comparison of low- and high-altitude Doppler velocimetry in the peripheral and central circulations of normal fetuses. Am J Obstetr Gynecol 2000; 183 (5): 1158–1161.
Krampl E, Kametas NA, Nowotny P, Roden M, Nicolaides KH . Glucose metabolism in pregnancy at high altitude. Diabetes Care 2001; 24 (5): 817–822.
Mortola JP, Frappell PB, Aguero L, Armstrong K . Birth weight and altitude: a study in Peruvian communities. J Pediatrics 2000; 136 (3): 324–329.
Lehre AC, Laake P, Sexton JA . Differences in birth weight by sex using adjusted quantile distance functions. Stat Med 2013; 32 (17): 2962–2970.
Kramer MS . Determinants of low birth weight: methodological assessment and meta-analysis. Bull World Health Org 1987; 65 (5): 663–737.
Rottensteiner S . Zusammenhang Geburtsgewicht von Neugeborenen und Seehöhe der Wohngemeinde in Österreich zwischen 1998 und 2008 adjustiert nach wichtigen geburtsbezogenen Parametern. Dr. thesis, Medical University of Vienna: Vienna, 2012.
Statistik Austria Demographisches Jahrbuch. Statistik Austria: Austria, Vienna, 2012.
SAS Institute Inc. SAS Institute Inc., 2002-2012. Cary, NC, USA, 2012.
Bauer P, Pötscher B, Hackl P . Model Selection by multiple test procedures. Statistics 1988; 19 (1): 39–44.
Notzon FC, Bobadilla JL, Coria I . Birthweight distributions in Mexico City and among US Southwest Mexican Americans: the effect of altitude. Am J Public Health 1992; 82 (7): 1014–1017.
Fenton TR . A new growth chart for preterm babies: Babson and Benda's chart updated with recent data and a new format. BMC Pediatrics 2003; 3: 13.
Roland MC, Friis CM, Godang K, Bollerslev J, Haugen G, Henriksen T . Maternal factors associated with fetal growth and birthweight are independent determinants of placental weight and exhibit differential effects by fetal sex. PLoS ONE 2014; 9 (2): e87303.
Eriksson JG, Kajantie E, Osmond C, Thornburg K, Barker DJ . Boys live dangerously in the womb. Am J Hum Biol 2010; 22 (3): 330–335.
Galan HL, Rigano S, Radaelli T, Cetin I, Bozzo M, Chyu J et al. Reduction of subcutaneous mass, but not lean mass, in normal fetuses in Denver, Colorado. Am J Obstetr Gynecol 2001; 185 (4): 839–844.
Zhu BP, Rolfs RT, Nangle BE, Horan JM . Effect of the interval between pregnancies on perinatal outcomes. New Engl J Med 1999; 340 (8): 589–594.
The authors declare no conflict of interest.
About this article
Cite this article
Waldhoer, T., Klebermass-Schrehof, K. The impact of altitude on birth weight depends on further mother- and infant-related factors: a population-based study in an altitude range up to 1600 m in Austria between 1984 and 2013. J Perinatol 35, 689–694 (2015). https://doi.org/10.1038/jp.2015.30