All perinatal adverse outcomes considered in our study occurred more commonly among infants of immigrants than Italians, independent of all confounders considered, except among mothers from developed countries that showed risk patterns similar to those observed among Italians. Being of Western or sub-Saharan African origin especially was positively associated with adverse perinatal outcomes. This finding may be due to a higher incidence of maternal diseases (e.g., hypertension, infectious diseases), and genetic differences . In addition, social exclusion for cultural reasons, poor living conditions, less frequent and sub-optimal prenatal care may explain these results [2, 6, 14, 45].
The recent increase in Romanian mothers, who had the lowest percentage of naturalised Italian citizens and therefore could be less integrated, and the presence of Roma people (about 7000 in Rome according to the Italian Red Cross in 2008) characterized by high birth rates and poor living conditions, may influence the results among Eastern Europeans. Suboptimal prenatal care due to a lack of information of women and poor training of health professionals  may also affect our results.
Over time, an overall increase in the odds was found only for late preterm live births and congenital malformations, but there were differences across regions, as found in Italian mothers.
Prematurity was associated with maternal immigrant status, especially among West and sub-Saharan Africans. This result may be explained by a higher incidence of maternal disease (e.g., pregnancy hypertension) or genetic differences [14, 45]. It is possible that immigrant mothers with pregnancy-associated diseases received sub-optimal prenatal care leading to an unfavourable perinatal outcome. Similar findings were observed in a northern Italian city in which Africans were at greater risk of prematurity associated with short permanence in the country independent of timely access to prenatal care . The higher prevalence rate of PBs among ethnic minority groups in a study conducted in Amsterdam was hypothetically due to earlier foetal maturation in black women . Also a meta-analysis of the literature published from 1995 to 2008 showed a higher risk of PBs in Asians and Africans, and lower risk in Latin Americans .
The Apgar score at 5 minutes was lower among West and sub-Saharan Africans and central and southern Americans. This finding may be linked to prematurity, observed in particular among Africans, and to sub-optimal care during pregnancy. Similar results were found in Washington state among infants born to Somali women compared to US-born blacks and whites, with prolonged gestation as a possible explanation accounted by the authors . In a Finnish study cited above the Apgar score at 1 minute was lowest among newborns of African- and Somali-origin; this result was attributed to variations in healthcare procedures during labour according to maternal origin . The absence of differences in health status at birth between Eastern European and Italian-origin infants may be partially due to residual confounding of young maternal age at delivery of Eastern European mothers. Interestingly, poorer health at birth was observed among infants born to East Asian compared to Italian women in the second period: the recent increase of Chinese immigration, characterised by young mothers from low socioeconomic backgrounds, linguistic and cultural barriers that may limit access to prenatal care may explain this result. Comparison with other Italian settings is limited by the scarcity of studies on this outcome.
The odds of having respiratory disease were the highest among infants born to West and sub-Saharan African women. This result might be partially mediated by the prematurity observed among infants born to Africans. The result may be also linked to a lower use of prenatal steroids as a consequence of poor prenatal care. A similar result was found in a study carried out in several Italian neonatal centres where infants born to nomadic and African parents had a higher incidence of neonatal asphyxia compared to those of native-born parents . Social disadvantage, difficult access to healthcare services, late or inadequate prenatal care were listed as possible explanations.
An association between maternal birth country and need for special or intensive neonatal care was observed, again particularly among West and sub-Saharan Africans. Very few studies have considered this outcome among immigrant mothers. In Spain a similar outcome was not more frequent among newborns of immigrant mothers compared to the host population . In addition to poor perinatal health, this outcome may reflect the use of invasive procedures that in some cases may have an iatrogenic effect (e.g. retinopathy of prematurity ).
Congenital malformations were higher among immigrants compared to Italians, especially among West and sub-Saharan Africans. Few studies are published regarding the association between birth defects and migrant status, therefore there have not been many hypotheses proposed. An Italian study  found a higher rate of deformities among migrants compared to Italians possibly due to poor prenatal care. In another study it was argued that socio-economic and cultural factors may explain differences in congenital malformations between migrant and non-migrants . In order to explain our result we hypothesised that these differences could be due to a higher occurrence of life threatening (anencephaly) or severe chromosomal anomalies (e.g., trisomy 13 or 18, Ebstein's anomaly) that could reflect less common use of legal termination of pregnancy among immigrant compared to Italian women. Point estimates confirmed this hypothesis (results not shown), although without statistical significance. Higher prevalence of risk factors for congenital malformations among immigrant mothers, such as lower periconceptional folic acid use, suboptimal control of preconceptional diabetes and epilepsy, obesity, maternal smoking, low vaccine coverage, higher alcohol consumption, and marriage between cousins in some migrant groups may also play a role. We cannot exclude that grouping birth defects may mask associations and that the different number of stillbirths in the two populations may indicate different prevalence of birth defects. Although the congenital anomalies data are not from a dedicated registry but from hospital records, we believe that any underestimate of birth defects made in the hospital environment was very unlikely.
The two time periods analysed allowed us to evaluate changes in the perinatal health in women of foreign origin. We observed a statistically significant decrease in the odds of very preterm live births only among Italian women; on the other hand, we found a slight increase of late preterm in both Italian and foreign women. This result may be in part explained by improved prenatal care in reducing the occurrence of very preterm births.
Improvement was also seen in decreases of other unfavourable outcomes (low Apgar score at 5 minutes, respiratory diseases, need for special or intensive neonatal care). The only worsening of perinatal outcomes regarded congenital malformations, which had about a fifty percent increase of odds ratio from the last to the first time period, in both groups. The high percentage of minor congenital anomalies observed may reflect an improvement in diagnostic assessment over the two periods in diagnostic assessment.
A limitation of our study is the lack of information in our administrative database of some risk factors such as direct measures of integration in the host country, socio-economic status, prenatal care, and life style (including tobacco use). However, we used proxy measures of acculturation (naturalized Italian citizens), socioeconomic status (educational level attained), and prenatal assistance (GA at the first visit). No reliable information was available on the fathers (e.g. country of origin, occupational status), which would have helped to interpret the results in relation to the families' social background and resources.
We could not analyse perinatal death as an outcome because our database does not list stillbirths or early neonatal death that occurred after transfer to another hospital. In most developed countries the etiology of perinatal death differs widely, and we expected this to be the case even more so when comparing immigrant and Italian birth experiences. For example, one might expect a higher prevalence of stillbirths among immigrant women if there are difficulties in prenatal access.
We could not identify illegal residents in the dataset. Since they may be those at greatest risk of unfavourable perinatal outcome, we cannot exclude outcome differences as a result.
It may be that some conditions may have been affected by changes in coding, hospital practices, or reporting over time. In addition, the association between maternal birthplace and outcomes may change over time if the composition of the region changes according to driving countries.
This study is one of the few to analyse the perinatal health status of immigrants in a wide region with significantly increasing immigration. Data were representative, as they covered all the births that occurred in the region; there were very low percentages of missing data except for GA (5% among Eastern European women). Out of hospital births were not available but this phenomenon is uncommon in our region (<0.1% in 2006) . Availability of routine data over time allowed a temporal comparison. Given the likelihood of cultural, socio-economic, and integration differences between mothers coming from different countries, the opportunity to split data by area of origin enabled us to identify subgroups at high risk of negative perinatal outcomes.