Low Birth Weight Trends in OECD Countries, 2000-2015: Economic and Health System Conditionings

Background: Low birth weight rates are increasing in developed countries. However, little is known of economic or organization factors inuencing this increase. This study aims to ascertain the 21st century relationships between the contextual country factors and low birth weight rates. Methods: We analyse trends of low birth weight rates in OECD countries. Data from 2000 to 2015 were obtained from the OECD data base. Their relationships with demographic and economic variables, health habits, woman-related preventive measures, health care system organization and funding, health care work force and obstetric care were analysed using random-effects linear regression. Results: Low birth weight rates are higher in Southern Europe and lower in Northern Europe. Low birth weight rates escalated about 20% in Southern Europe and to less extent in Easter Europe and Asian/Oceanian countries, while remained stable in America, Central Europe and Northern Europe. Investment in health care, private health system coverage, ratios of paediatricians and obstetricians, average length of admission due to pregnancy or birth and Caesarean section rate were associated with higher low birth weight rates. Factors associated with lower low birth weight rates were health care coverage, public health system coverage, hospitals per million inhabitants, and ratios of health care workers, physicians, midwives and nurses. Conclusions: Low birth weight rates are higher and are stepping faster in Southern Europe than in other developed countries. Many factors on health care system organization and funding as well as rations of health care workers were associated with low birth weight rates.


Background
Low birth weight (LBW) is de ned as birth weight ≤ 2500 g by the World Health Organization, not considering gestational age [1]. LBW is a key factor because it affects about 15.5% of all newborns [2]. LBW babies are at higher risk of neonatal morbidity and mortality [2], including higher risk for developing diseases such as childhood obesity, hypertension, adverse cardiovascular and metabolic outcomes or impaired neurodevelopment [2,3]. LBW is associated with enhanced infant mortality and long-term morbidity during adulthood [4], which makes weight at birth a relevant indicator for the fetus' development quality and a predictor of health throughout its life course [5].
Little is known, however, on the relationship between LBW and contextual (i.e. general) country factors.
Factors as, for instance, the evolution of per capita GDP, budget dedicated to the health system, number of physicians, nurses, obstetricians or other health care professionals, or health system performance have been little studied. Most of recent studies analyze the impact of the global nancial crisis of 2008 on LBW [17,[20][21][22], con rming the signi cant increase in LBW during that crisis. Countries where the most severe austerity measures were implemented have experienced the highest increase in LBW rate [17,[20][21][22].
The Organisation for Economic Co-operation and Development (OECD) conducts annually an evaluation of health variables entitled "Heath at a glance", presenting an analysis with aggregated data on different health variables. "Health at a glance" dedicates a speci c chapter to children's health, where it collects data on LBW [23].
The purpose of this study was to ascertain the 21st century relationships between the contextual country factors and low birth weight rates in OECD countries.

Source of information and data gathered
We gathered data for all 35 OECD countries (Supplementary Table 1 Information obtained included LBW rate (per 100 newborns), demographic and economic variables (percentage of women in the general population, fertility and natality rates, per capita GDP), health habits and woman-related preventive measures (smoking, ethanol consumption, people reporting good health, breast and cervical cancer screening coverages), health care system organization and funding (investment in health care, health care coverage, public and private system coverage, number of hospitals), health care work force (health care workers, physicians, general practitioners, paediatricians, obstetricians, midwives, nurses) and obstetric care (average length of admission due to pregnancy, single birth, other births and puerperium-associated complications, and Caesarean rate).

Statistical analysis
Data were arranged as panels according to country and year. Relationships between explainable variables and LBW rates were analysed using random-effects linear regression. Results are presented as linear regression coe cient with 95% con dence interval (CI). All analyses were carried out with the Stata 16/SE statistical package (Stata Corp., College Station, Texas, US).

Results
Mean and standard deviations for all variables recorded are displayed in Table 1, strati ed by 5-year periods. LBW average rate increased from 6.2% in 2005 to 6.5% in 2015. Fertility and natality rate showed inconsistent trends, while per capita GDP about doubled its average from 2000 to 2010 but slightly decreased towards 2015. Average number of cigarettes and ethanol consumption declined throughout the period, whereas the percentage of people reporting good health and breast cancer screening coverage increased. Investment in health care escalated from 7.2% GDP to 8.9% GDP, but trends in health care coverage were inconsistent. Physician ratios increased all over the studied period as well as ratios of GPs, paediatricians, obstetricians and nurses. Average length of admission due to obstetrics procedures decreased, but Caesarean section rates escalated from 181 to 252 per 1000 newborns.  Figure 1 shows LBW rate trends in each country. In American countries, Chile and Canada reported similar gures and trends, while the US had higher rates and Mexico displayed a somewhat erratic trend (Fig. 1a). All seven Central European countries had steady trends and similar rates (Fig. 1b), being the most homogeneous region as shown by the small standard deviations in Table 2. Trends in Eastern Europe were somewhat divergent, with Estonia and Latvia having lower rates and some decreasing trends, and Hungary, Czech Republic and Slovakia having higher rates and raising trends (Fig. 1c). All countries in Northern Europe save for the United Kingdom had low rates of LBW; the United Kingdom, however, is the only one displaying a decreasing trend (Fig. 1d). Countries grouped in "Other" category were highly heterogenous; Korea and Japan had the lowest and the highest LBW rates respectively and both exhibit a stepping trend (Fig. 1e). Finally, all ve countries in Southern Europe had high LBW rates, France being the only one not showing a raising trend (Fig. 1f). The associations between explainable variables and LBW rates appear in Table 3. Fertility and natality rates were not associated with LBW rate. Per capita GDP was inversely associated with LBW, although a 1000$ increase in "per capita" GDP only accounted for a 0.012% decrease in LBW (95% CI: -0.018, -0.007). Countries with higher number of cigarettes per smoker reported higher LBW rates, while countries with higher breast or cervical cancer screening coverages reported lower LBW. All variables on health care system organization and funding were associated with LBW rates: the higher the investment in health care or the private health system coverage, the higher the LBW rate. Health care coverage, public health system coverage and hospitals per million inhabitants were negatively associated with LBW rates. Ratios of health care workers, physicians, midwives and nurses were negatively associated with LBW rates, while higher ratios of paediatricians and obstetricians were associated with higher LBW rates. Finally, most obstetric care indicators (average length of admission due to pregnancy-related factors, single birth or puerperium-associated complications, as well as the Caesarean rate) were related with higher LBW rates.

Discussion
An increase in low birth weight rates has been observed between 2000 and 2015 in some OECD countries, and most particularly the Southern European countries, especially Greece, Portugal, Spain, together with the UK at the opposite geographical extreme (Fig. 1). Some authors [17,20,21] considered that the increase in LBW rate during the 2008-2014 period cannot be explained only by maternal factors. These authors relate the nancial crisis starting in 2008 with the perinatal problems. So, Zografaki et al. [21] associated the fact that from 2008 to 2014 GDP per capita was reduced by 32% in Greece with the increase observed in their country of the LBW rate [21]. In the same way, Portugal [20] and Spain [5,17] also have associated the nancial crisis and its ensuing decrease in per capita GDP with the observed increment of LBW rates during that period. Our ndings support this idea, given that we observe an association between some economic and health system conditionings and the LBW rate. In particular, from our data, per capita GDP was inversely associated with LBW, although a 1000$ increase in GDP only accounted for a 0.012% decrease in LBW, as already mentioned above.
Along this same line, Varea et al. [5] mentioned how the impact of the 2008 economic crisis introduced changes into the healthcare systems of some European countries, affecting per capita GDP spending on healthcare systems, among other factors [5]. This, in its turn, appears to have had a negative impact on pregnant women´s health and foetal development, as registered in some European countries such as Ireland or Greece. Varea et al. [5] therefore concluded that per capita GDP decrease is, by extension, a signi cant risk factor for LBW. Similarly, Sidebotham et al. [4] estimated that women who do not have adequate access to antenatal care can have problems in their own wellbeing and that of their foetus, which can result in LBW. Our results also support this idea, with the observed inverse relationship between health care system organization such as health coverage, public health system coverage or the number of hospitals per million inhabitants and LBW rate emanating from our data (Table3).
Additionally, some authors did not consider the nancial crisis as the sole responsible for the increase of LBW, but also the measures adopted by governments to deal with the crisis. As Rajmil et al. [22] pointed out, some governments decided to preserve the public systems while others chose to make cuts in health and other public services. Rajmil  Our nding that the investment in private health care is directly associated with LBW could be explained by the exponential increase in assisted reproduction procedures, which are mainly in the hands of the private sector [25]. Rich regions tend to spend more on assisted reproductive techniques [26]. This fact is closely related whit LBW, mainly as a result of multiple pregnancies and an increase in preterm birth, as reported by Goisis et al. [25]. They observed that 13% of children conceived by means of assisted reproduction showed LBW, as compared to 3% resulting from natural pregnancies. Also, women residing in rich regions tend to have their rst pregnancy later than women belonging to poorer regions [27], and high age at rst pregnancy is also intimately associated with LBW [6][7][8].
Our study has some limitations. Firstly, our analysis is ecological in nature, so causality cannot be invoked. Both collinearity and reverse causality could also be sensible explanations in some associations; for instance, higher LBW rates could be both the cause and the result of higher obstetrician ratios. Secondly, statistics were recorded at country level, allowing the use of different de nitions between countries despite OECD standardization. In this regard, even de nition changes within individual countries are possible throughout the studied period, leading to abrupt departures from the background trend (as shown for Mexico in Fig. 1a). Figure 1 Trends in low weight at birth rates from 2000 to 2015 in OECD countries, by region. 1a: America, 1b:

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryMaterial.docx