Continuous perinatal health monitoring and analysis in Europe based on a federated analytic approach: a proof of concept study

Abstract Background International comparisons of population indicators of maternal and newborn health are valuable for guiding health policy and practice. The Covid-19 pandemic revealed the difficulties of compiling comparable, timely data in Europe. As part of the PHIRI (Population Health Information Research Infrastructure) project, we developed a protocol to facilitate the exchange and analysis of population birth data in Europe. Methods The Euro-Peristat network, which includes experts from 31 European countries, developed a common data model and R scripts to facilitate rapid exchange of anonymised aggregate tables (https://zenodo.org/record/5148032#.YmlUttpBxPY). These tables were used to compile comparable perinatal health indicators from routine population-based sources for the years 2015 to 2020. We assessed the feasibility of this approach and the availability, quality and comparability of the data. Results Building on previous Euro-Peristat recommendations and a structured consensus process, the network defined a common data model including 22 variables for the testing phase. 17 additional variables were considered important and feasible for a second phase. 25 countries created patient-level data files. Most countries had 20 or more of the data items, whereas 1 had 18, 3 had 16 and 2 had 15 variables. Limiting factors included not having all data in a single database, most often the case for neonatal and infant mortality or vital statistics versus healthcare data, and the diversity or absence of data on socioeconomic status. Setting up the model was time consuming, but once established, running the R scripts was easy and quick (<15 min). The protocol requires the active participation of each country to ensure it is correctly applied. Conclusions We illustrated the feasibility of using a common data model with open source scripts to facilitate rapid production of data and analysis on key perinatal health indicators in European countries


Background:
International comparisons of population indicators of maternal and newborn health are valuable for guiding health policy and practice. The Covid-19 pandemic revealed the difficulties of compiling comparable, timely data in Europe. As part of the PHIRI (Population Health Information Research Infrastructure) project, we developed a protocol to facilitate the exchange and analysis of population birth data in Europe.

Methods:
The Euro-Peristat network, which includes experts from 31 European countries, developed a common data model and R scripts to facilitate rapid exchange of anonymised aggregate tables (https://zenodo.org/record/5148032#.YmlUttpBxPY). These tables were used to compile comparable perinatal health indicators from routine population-based sources for the years 2015 to 2020. We assessed the feasibility of this approach and the availability, quality and comparability of the data.

Results:
Building on previous Euro-Peristat recommendations and a structured consensus process, the network defined a common data model including 22 variables for the testing phase. 17 additional variables were considered important and feasible for a second phase. 25 countries created patient-level data files. Most countries had 20 or more of the data items, whereas 1 had 18, 3 had 16 and 2 had 15 variables. Limiting factors included not having all data in a single database, most often the case for neonatal and infant mortality or vital statistics versus healthcare data, and the diversity or absence of data on socioeconomic status. Setting up the model was time consuming, but once established, running the R scripts was easy and quick (<15 min). The protocol requires the active participation of each country to ensure it is correctly applied.

Conclusions:
We illustrated the feasibility of using a common data model with open source scripts to facilitate rapid production of data and analysis on key perinatal health indicators in European countries

Background:
The Euro-Peristat network documented disparities in perinatal outcomes between countries in Europe in its reports published every 5 years, but trend analyses were limited because data were not collected annually.

Methods:
Using the Euro-Peristat PHIRI protocol, we estimated rates and assessed trends between 2015 and 2019 for preterm birth, stillbirth, neonatal mortality and caesarean delivery. Countryspecific relative risks (RR) for year, modelled as a continuous variable, were estimated and random effects meta-analysis used to generate pooled RRs. Heterogeneity was measured with the I2 statistic (percentage of variability in estimates due to heterogeneity rather than sampling error).

Results:
Stillbirth rates 24 weeks of gestational age (GA) varied in 2019 from <2.5 per 1000 births in Denmark, Estonia, Finland and Slovenia to over 4 per 1000 in Belgium, Cyprus, UK Wales and Lithuania. Preterm birth rates ranged from <6% in Lithuania, Finland, Latvia, Estonia and Denmark to 8% or more in Portugal, Belgium, UK Scotland and Cyprus. Fewer than 20% of births were by caesarean in Norway, the Netherlands, Finland, Estonia in comparison to one-third in Cyprus, Ireland, Italy, UK Scotland. Trends over time differed between countries and were not related to the level of the indicator: the pooled RR by year for preterm birth was 0.99 [0.99; 1.00] with five countries having significant decreases and three countries having increases. Caesarean section rates were stable overall (RR: 1.00 [0.99; 1.01]RR:1.00, 95% CI: 0.99-1.01), but with high heterogeneity (I2 = 99%); in six countries rates increased significantly, whereas in nine rates decreased between 2015 and 2019.

Conclusions:
European countries have varying rates and trends of the principal perinatal health indicators. Investigation of policies in high-performing countries could provide guidance for improvement elsewhere.