{
  "abstract": "Objective To investigate the magnitude and evolution of inequalities in neonatal mortality rates by using area based socioeconomic indices in France.Design National population based study.Setting For 2015-20, data from the French National Health Data System (Système National des Données de Santé, SNDS). For 2001-08, neonatal death certificates and aggregate vital statistics data by municipality of residence.Participants Live births with a gestational age ≥22 completed weeks to a mother residing in metropolitan France, 2015-20 (4 293 403 live births and 10 869 neonatal deaths), compared with a 2001-08 study (6 202 918 live births and 14 851 neonatal deaths).Main outcome measures Differences in neonatal mortality rate (death before day 28 of life) according to the socioeconomic characteristics of the mother's municipality of residence. Comparison with data from a 2001-08 study to assess changes in socioeconomic inequalities and their contribution to the increase in neonatal mortality rate.Results The neonatal mortality rate was 2.53 per 1000 live births in 2015-20. Five indicators, previously associated with perinatal mortality, were combined into a perinatal French deprivation index (P-FDep) for the main analysis. P-FDep was categorised into five equal groups (deprivation groups 1-5) for comparison with other research and into 10 equal groups (deprivation groups 1-10) for more granular analyses, with group 1 being the least and group 5 (or group 10) the most deprived group. The rate in the most deprived compared with the least deprived group for P-FDep was 1.71 (95% confidence interval 1.60 to 1.83) times higher, based on the analysis of deprivation groups 1-5. A mortality gradient existed across the groups, translating into 2496 excess deaths (23.3%) when the rate in the least deprived group was applied to all areas. The gradient was more marked when deprivation groups 1-10 were used (relative risk 1.88, 95% CI 1.71 to 2.07 for the highest to the lowest deprived group). Compared with 2001-08 (neonatal mortality rate 2.39 per 1000), the rate remained constant in the least deprived areas, but worsened in the most deprived areas (+10.1% and +11.7% for groups 4 and 5, respectively), increasing the relative risks between the highest and lowest groups, which were 1.54 (95% CI 1.46 to 1.62) for deprivation groups 1-5 and 1.67 (1.55 to 1.79) for deprivation groups 1-10, in 2001-08.Conclusions In this study, the socioeconomic level of the mother's place of residence was strongly associated with the neonatal mortality rate. The data showed that inequalities have widened, contributing to the increase in the neonatal mortality rate.",
  "authors": [
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France",
        "Department of Neonatal Intensive Care, AP-HP, Hôpital Necker Enfants-Malades, Paris, France"
      ],
      "name": "Victor Sartorius"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France",
        "Department of Neonatal Intensive Care, AP-HP, Hôpital Cochin-Port Royal, Paris, France"
      ],
      "name": "Héloïse Torchin"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France"
      ],
      "name": "Luc Gaulard"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France"
      ],
      "name": "Marianne Philibert"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France",
        "Department of Neonatal Intensive Care, AP-HP, Hôpital Cochin-Port Royal, Paris, France"
      ],
      "name": "Victoria Butler"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France"
      ],
      "name": "Monica Saucedo"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France"
      ],
      "name": "Catherine Deneux-Tharaux"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France",
        "Directorate for Research, Studies, Evaluation, and Statistics (DREES), French Ministry for Health and Solidarity, Paris, France"
      ],
      "name": "Jeanne Fresson"
    },
    {
      "affiliations": [
        "Centre for Research for Epidemiology and Statistics (CRESS-UMR1153), Obstetrical, Perinatal and Pediatric Lifecourse Epidemiology (OPPaLE), Université Paris Cité, Inserm, INRAE, Paris, France"
      ],
      "name": "Jennifer Zeitlin"
    }
  ],
  "full_text": "WHAT IS ALREADY KNOWN ON THIS TOPIC Neonatal mortality is strongly linked to socioeconomic factors at both the individual and area levelsThe risk of neonatal death is substantially higher in more deprived areas, making these inequalities a preventable contributor to neonatal mortalityAlthough neonatal mortality rates continue to decrease in most countries, progress has slowed, and in France, a slight upward trend in neonatal mortality has been seen since 2012WHAT THIS STUDY ADDS Between 2015 and 2020, an index combining area based measures of socioeconomic status (perinatal French deprivation index) showed a linear rise in neonatal mortality rates with increasing deprivation in the mother's municipality of residenceThe risk of neonatal death was 1.71 times higher among the 20% of births from the most deprived areas than the 20% from the least deprived areasNeonatal mortality rates in France were higher in 2015-20 than in 2001-08 and increased exclusively in deprived areas, whereas rates remained stable in the least deprived areas, leading to widening of socioeconomic inequalitiesHOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY Research should focus on understanding the mechanisms underlying area based inequalities by investigating differences in population characteristics, healthcare access, environmental factors, and quality of care between the least and most deprived areasThe findings suggest that socioeconomic inequalities contributed to the recent worsening of neonatal mortality rates in France and call for public health actions targeting the highest risk areas, which have been identified in this studyIntroduction The neonatal mortality rate, defined as the number of deaths occurring in 0-27 completed days of life per 1000 live births, 1 is a critical indicator of the health status of newborns and provides insights into overall population health.2 Neonatal mortality is closely linked with population socioeconomic characteristics, both at the individual and area levels.3 In many countries, the risk of neonatal death is higher in more deprived than in less deprived areas, and these socioeconomic inequalities constitute a substantial and preventable portion of neonatal mortality.4 5 Consequently, understanding the association between area level socioeconomic characteristics and neonatal mortality is important for countries aiming to reduce their neonatal mortality rates.In France, recent trends in neonatal mortality rates have caused concern nationwide. The neonatal mortality rate has remained the same since the mid-2000s, with a slight increase since 2012.6 In 2015, the neonatal mortality rate was 2.5 per 1000 live births, with France ranked 22 out of 33 European countries for neonatal mortality.7 Most countries continue to report decreases in neonatal mortality, but these decreases are slowing, and concerns have also been raised in the UK and US about no change or increasing mortality rates.8 9 One hypothesis for France's poor performance on this indicator is high and growing social inequalities. A study on data from 2001-08 found that the risk of neonatal death was 1.2-1.5 times higher in the most deprived than in the less deprived areas in France.10 Since then, no national study has examined the effect of socioeconomic level on neonatal mortality, revealing a substantial knowledge gap about recent patterns and impeding the development of effective preventive action.The aim of this study was to measure the magnitude and evolution of area based socioeconomic inequalities in neonatal mortality based on recent data in France. We first investigated differences in the neonatal mortality rate according to the socioeconomic characteristics of the mother's municipality of residence, by using individual area based indicators available from French census data and a composite index created from indicators associated with perinatal risks in previous research. We then compared these data with a 2001-08 study to assess changes in socioeconomic inequalities and their contribution to the increase in neonatal mortality rate.Methods Data sources We conducted a national population based study. For the 2015-20 period, data were from the French National Health Data System (Système National des Données de Santé, SNDS) which includes hospital discharge data, providing standardised medical information for all deliveries and births occurring in hospital as well as unintentional out-of-hospital births when the mother and baby are admitted to hospital after the birth (<0.2% of births are planned home births in France). 11 Hospital discharge data have minimal missing data, and good concordance exists when perinatal variables are compared with data from surveys based on medical records.12 We linked births to neonatal death certificates (deaths from day 0 to day 27), also included in the French National Health Data System, with a three step process that sequentially uses anonymous identification numbers followed by deterministic and probabilistic linkage (linkage rate 96.3%).13 When death certificates could not be linked to births in the hospital discharge data, we added a live birth corresponding to the death. To avoid double counting, we then excluded live births in hospital discharge data that ended in an unlinked neonatal death. Online supplemental table 1 provides annual linkage rates and numbers of unlinked deaths in both sources. For 2001-08, neonatal data were from neonatal death certificates, and live births were taken from aggregate vital statistics data by municipality of residence.SP210.1136/bmjmed-2024-000954.supp2Supplementary dataStudy population The study population included all live births with a gestational age ≥22 completed weeks to a mother residing in metropolitan France from 2015 to 2020. The study's start year corresponds to the year when neonatal death certificate data could be reliably linked with hospital data, and 2020 was the most recent year at the time of the study. For the analysis of changes over time, we selected live births from 2001 to 2008, consistent with the previous national study on perinatal outcomes and socioeconomic status. 10Characteristics of births and deaths Data on the mother's place of residence at birth and other characteristics of the population used in this study (gestational age, birth weight, sex, maternal age, and type of pregnancy) were from hospital discharge data. When deaths could not be linked, this information was taken from the neonatal death certificate. The municipality of residence, the smallest geographical unit in the French National Health Data System (comprising about 36 000 units in France), was accessible for all deaths and 97.0% of live births. If the municipality was unavailable (eg, if the mother moved between the time of birth and when the information was updated in the national system), the municipality of residence was derived from the mother's geographic code at birth (about 6000 units) from the data system's equivalence table. For the 2001-08 period, we used the municipality of residence from the death certificate for neonatal deaths and vital statistics for live births that did not result in a neonatal death.Area based socioeconomic characteristics Area based deprivation was defined at the geographical scale of the municipality of residence. We investigated seven indicators of deprivation that have consistently been part of the census surveys in France over time and have been included in previous studies of area based deprivation in France. 10 14 These indicators are: unemployment rate, proportion of immigrants, proportion of non-homeowners, proportion of single parent families, median income for each household, proportion of women aged 15-45 years with a low educational level, and proportion of manual workers. We also created a composite area indicator from the five variables associated with adverse perinatal outcomes in previous research (unemployment rate, proportion of immigrants, proportion of non-homeowners, proportion of single parent families, and median income for each household).10 14 We used the first component of a principal component analysis to synthesise this information into one index, the perinatal French deprivation index (P-FDep), following the methodology used for deriving the existing French deprivation score included in the French National Data System.15 This general French deprivation index was not used for our analysis because studies have shown a low correlation with perinatal risk, including perinatal mortality and preterm birth.10 16 P-FDep was categorised into five equal groups, for comparison with other research, and into 10 equal groups for more granular analyses. Socioeconomic data on the municipalities were from the 2015 population census for the 2015-20 period and from the 2006 census for the 2001-08 period. Comprehensive data were not available for municipalities of <50 households and some of those that changed during the study periods (representing <1% of live births, online supplemental table 1).Analysis strategy We first described the characteristics of live births and neonatal deaths and then divided live births into five equal groups according to the distribution of each of the socioeconomic indicators based on the mother's place of residence. We selected five equal groups for these descriptive analyses for comparison with previous studies on area based inequalities in perinatal mortality. 17–20 The neonatal mortality rate per 1000 live births was calculated for each of the five groups and the Cochrane-Armitage test was used to test for linear trends across groups.Poisson regression models were used to compute the relative risk between the five equal groups and 10 equal groups based on the number of neonatal deaths, with the least deprived group for both categories being the reference group. To take into consideration the clustering of the data at the municipal level, robust clustered standard errors were calculated. These models were used to derive relative risk estimates adjusted for maternal age and multiplicity. These two population characteristics, available in the hospital data, are related to risks of adverse perinatal outcomes and socioeconomic status.21 22 Proportions of missing data for maternal age (3.8% and 0.5% for neonatal deaths and other live births, respectively) and type of pregnancy (3.0% and 0.1% for neonatal deaths and other live births, respectively) were low. These data were imputed with the missForest method, a method based on iteratively fitting random forests to predict missing values for continuous and categorical variables.23 All variables in table 1, including outcome, were included as predictors.Table 1Characteristics of live births and neonatal deaths in metropolitan France, 2015-20CharacteristicsLive birthsNeonatal deathsNeonatal mortality rate (‰)Whole population4 293 40310 8692.53Mean±SD maternal age (years)30.4±5.330.4±5.8—Maternal age (years): <2075 254 (1.8)260 (2.4)3.45 20-2449 8006 (11.6)1413 (13.0)2.84 25-342 747 687 (64.0)6205 (57.1)2.26 35-39762 713 (17.8)1922 (17.7)2.52 ≥40190 170 (4.4)652 (6.0)3.43 Missing data19 573 (0.5)417 (3.8)—Type of pregnancy: Singleton4 146 015 (96.6)8592 (79.1)2.07 Twin139 779 (3.3)1789 (16.5)12.80 Triplet or higher order3085 (0.1)162 (1.5)52.51 Missing data4524 (0.1)326 (3.0)—Sex: Male2 205 798 (51.4)6089 (56.0)2.76 Female2 108 854 (49.1)4780 (44.0)2.27Birth weight (g): <100017 447 (0.4)5320 (48.9)304.92 1000-149926 735 (0.6)1005 (9.2)37.59 1500-2499267 128 (6.2)1497 (13.8)5.60 2500-34992 565 607 (59.8)2208 (20.3)0.86 3500-39991 123 240 (26.2)527 (4.8)0.47 ≥4000293 362 (6.8)158 (1.5)0.54 Missing data—Gestational age (weeks): 22-232094 (<0.1)1895 (17.4)904.97 24-2713 634 (0.3)3389 (31.2)248.57 28-3129 549 (0.7)1147 (10.6)38.82 32-36254 868 (5.9)1423 (13.1)5.58 37-413 955 900 (92.1)2856 (26.3)0.72 ≥4236 923 (0.9)25 (0.2)0.68 Missing data435 (<0.1)134 (1.2)—Year of birth: 2015745 459 (17.4)1845 (17)2.48 2016731 909 (17.0)1797 (16.5)2.44 2017719 892 (16.8)1905 (17.5)2.65 2018707 035 (16.5)1817 (16.7)2.57 2019702 347 (16.4)1794 (16.5)2.55 2020686 761 (16.0)1711 (15.7)2.49Data are number (%) unless indicated otherwise.To compare with the 2001-08 data, we derived rates and univariate relative risks of neonatal death (with the least deprived groups as the references) for the earlier period and calculated the absolute and relative changes in these estimates from 2001-08 to 2015-20. We also conducted a sensitivity analysis by excluding 2020 because the covid-19 pandemic could have influenced the results. Linkage of the cause of death certificates with hospital discharge data was conducted with SAS software (version 8); all other analyses were conducted with R 4.2.2 (www.r-project.org).Patient and public involvement Given the use of pseudonymised national data, patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Findings will be communicated to the public through multiple media channels.Results Among the 4 293 403 live births in 2015-20 in metropolitan France, 10 869 infants died before day 28 of life, resulting in a neonatal mortality rate of 2.53 per 1000 live births. Table 1 shows the individual characteristics of births in the population. Neonatal mortality was higher among babies born to younger and older mothers, in multiple births, and in male infants. Table 1 also shows the strong association between gestational age and weight at birth, and risk of death.The neonatal mortality rate was higher in areas with greater socioeconomic deprivation (table 2). Mortality increased across the five deprivation groups (group 1 being the least and group 5 the most deprived) and for all indicators, with the exception of the proportion of manual workers and to a lesser degree for educational level. Table 2 shows the results for the composite P-FDep. This indicator, derived from the first component of a principal component analysis, combined five variables (proportion of non-homeowners, proportion of immigrants, proportion of single parent families, unemployment rate, and median income for each household) into one index. This first component accounted for 41.5% of the total variance, with high loading values for included variables. Online supplemental figure 1 provides more detail, including the correlation circle plot and loading values.Figure 1 shows the unadjusted and adjusted relative risks of neonatal death in the most deprived compared with the least deprived areas, defined by deprivation groups 1-5 and deprivation groups 1-10, for the composite indicator P-FDep. For deprivation groups 1-5, infants born to mothers residing in the most deprived group (group 5), comprising 20% of live births, had a 1.71-fold (95% confidence interval (CI) 1.60 to 1.83) higher risk of neonatal mortality compared with the 20% born to mothers in the least deprived municipalities. This gap was more marked when deprivation groups 1-10 were used (1.88, 95% CI 1.71 to 2.07). We found a strong linear gradient across deprivation groups 1-5 and deprivation groups 1-10. Deprived municipalities according to P-FDep were more often concentrated in urban areas (online supplemental figure 2).Figure 1Adjusted and unadjusted (crude) relative risks of neonatal mortality by area based deprivation, measured by the perinatal French deprivation index, in 2015-20. Top=relative risk by deprivation groups 1-5; bottom=relative risk by deprivation groups 1-10. Adjusted relative risk is relative risk adjusted for maternal age and multiple birth. Relative risks (crude and adjusted) are presented with 95% confidence intervals. Deprivation groups are presented from the least (group 1) to the most (group 5 or group 10) deprived groupsTable 2Number of neonatal deaths and mortality rates by deprivation groups 1-5, defined by socioeconomic characteristics of the municipality of residence in France, 2015-20Place of residence socioeconomic indicatorsDeprivation groupTrend testGroup 1 (least deprived)Group 2Group 3Group 4Group 5 (most deprived)Z†P valueProportion of non-homeowners: No of deaths16791883213824192750 Mortality rate (‰)1.962.192.492.823.2018.1<0.001Proportion of migrants: No of deaths17721807210823732809 Mortality rate (‰)2.062.102.442.763.2918.1<0.001Proportion of single parent families: No of deaths17321888207224482729 Mortality rate (‰)2.022.202.372.893.1817.6<0.001Unemployment rate: No of deaths17311918218724532580 Mortality rate (‰)2.022.232.552.863.0115.2<0.001Median income: No of deaths18341978208823092660 Mortality rate (‰)2.142.322.412.693.0913.3<0.001Proportion without high school diploma: No of deaths19022053215823062450 Mortality rate (‰)2.222.392.512.692.859.1<0.001Proportion of manual workers: No of deaths20882141223722612142 Mortality rate (‰)2.432.492.612.632.501.50.120P-FDep* No of deaths16651812206024592873 Mortality rate (‰)1.952.122.402.843.3420.4<0.001*P-FDep, derived from the first component of a principal component analysis, combines the five variables proportion of non-homeowners, proportion of immigrants, proportion of single parent families, unemployment rate, and median income for each household into one index.†Estimate for the Cochrane-Armitage trend test.P-FDep, perinatal French deprivation index.Adjusting for maternal age and multiple births did not significantly affect the estimates, although the risks were slightly lower. In absolute terms, if the mortality rates of the population in group 1 of P-FDep, for deprivation groups 1-5, were applied to all births in France, 2496 fewer deaths would be found, corresponding to 23.0% of the total. The same calculation, based on mortality rates of group 1 for deprivation groups 1-10, would result in 2803 fewer deaths (25.8% of the total).For the period 2001-08, 14 851 neonatal deaths occurred among the 6 202 918 live births recorded. Online supplemental table 2 shows the characteristics of neonatal deaths from 2001 to 2008. The neonatal mortality rate for this period was 2.39‰, or 0.14‰ lower than in the 2015-20 period. Figure 2 shows the unchanged mortality rates in the least deprived areas and the deterioration in rates in the most deprived areas. Relative risks of neonatal mortality in 2015-20 between the lowest and highest deprived areas, for both deprivation groups 1-5 and deprivation groups 1-10, were higher than in 2001-08, illustrating the widening of inequalities in the most deprived areas (figure 3). In sensitivity analyses, mortality rates by P-FDep for deprivation groups 1-5 were almost identical when excluding 2020 (online supplemental figure 3).Figure 2Neonatal mortality rates by area based deprivation, measured by perinatal French deprivation index, in 2015-20 compared with 2001-08. Top=mortality rate by deprivation groups 1-5; bottom=mortality rate by deprivation groups 1-10. Neonatal mortality rates are presented with 95% confidence intervals. Deprivation groups are presented from the least (group 1) to the most (group 5 or group 10) deprived groupsFigure 3Relative risks of neonatal mortality by area based deprivation, measured by perinatal French deprivation index, in 2015-2020 compared with 2001-2008. Top=relative risk by deprivation groups 1-5; bottom=relative risk by deprivation groups 1-10. Relative risks are presented with 95% confidence intervals. Deprivation groups are presented from the least (group 1) to the most (group 5 or group 10) deprived groupsDiscussion Principal findings In this study, we found that during the period 2015-20, infants were more likely to die in the neonatal period when their mothers resided in more deprived French municipalities, based on an index of deprivation constructed from area based indicators associated with perinatal risks. Overall, neonatal mortality rates were slightly higher in 2015-20 compared with 2001-08, reflecting a more substantial rise in mortality rates in the socioeconomically most deprived areas, whereas rates remained stable or even decreased in the least deprived areas. The results of the study also show the margin for improvement by dealing with socioeconomic inequalities; over a period of six years, 2472 or 2803 fewer neonatal deaths would occur if the rates of neonatal mortality of deprivation group 1 could be applied to all areas, based on our analysis for deprivation groups 1-5 and deprivation groups 1-10, respectively.Comparison with other studies Our results corroborate those from multiple studies showing that area based socioeconomic deprivation is strongly associated with the risk of neonatal death. Previous studies have found similar gradients of 1.5-2 between the lowest and highest deprivation groups. 5 17 Our results show that trends over time might differ within these groups. In France, our findings suggest that socioeconomic inequalities in neonatal health have a role in explaining the recent decline in neonatal mortality performance in France.Multiple inter-related factors can explain the association between socioeconomic status and adverse perinatal health outcomes.24 These factors include obesity, smoking, and other behavioural factors (eg, use of healthcare services, adherence to treatment, and substance misuse).25 26 Area based socioeconomic indices are a proxy for these individual level effects,27 but they can also reflect the characteristics of the physical and policy environment. Examples are poor healthcare organisation and access, air pollution, and climate risks, such as extreme heat, that increase the risk of unfavourable perinatal and neonatal outcomes.28–30 Research has also suggested that, in addition to socioeconomic and environmental factors, worse quality of care might affect neonatal outcomes in hospitals that care for populations from ethnic minority groups and from deprived areas.31 These environmental factors are amenable to prevention and public health measures, and studying them could help identify those associated with the deterioration in neonatal mortality in France.The use of a tailored deprivation index allowed us to identify geographic areas that could be prioritised in public health policies to improve perinatal health. As found in other French studies, these risks are highest in urban areas.32 This finding contrasts with other countries, such as the US, where the highest risks are found in rural areas.33 Identification of the municipalities at highest risk (online supplemental figure 2) provides an opportunity to explore areas for improvement at the local level following, for example, the approach used in the French district of Seine-Saint-Denis, where an audit to identify preventable factors was carried out.34In our population level analysis, we found no association between mortality rate and the proportion of manual workers in the municipality, one of the components of the general French deprivation score included in the French National Data System. Although the occupational category is associated with general and perinatal health in other contexts, the proportion of manual workers, as an area based indicator, likely does not reflect the individual and environmental factors relevant to neonatal health in France. This finding could be related to other sociodemographic characteristics recorded by this indicator, such as the age of the residents in these municipalities, reflecting decreasing proportions of manual workers (from 30% in the 1980s to 20% after 2010) and changes in the proportions of unskilled versus skilled workers.35 36 Our study shows the importance of reflecting on the individual components of deprivation scores to assess whether they relate to pregnancy outcomes. Further work should assess whether a new indicator is needed for surveillance and research on mothers and children in the French National Health Data System, as is done in research in the UK, for example, with the Children in Low Income Families Local Measure.5 We should also investigate whether integrating other dimensions of vulnerability could make research more useful for developing and monitoring policy initiatives.37Strengths and limitations of this study The strengths of our study were the use of national, population based data, with a complete record of neonatal deaths. This study also benefited from the improvement in the quality and availability of perinatal health data from the French National Health Data System. This improvement enabled the creation of a cohort of all live births, with the pooling of hospital discharge and death certificate data, allowing for adjusted analyses of maternal characteristics.The absence of data on maternal and birth characteristics for the period 2001-08 prevented us from exploring how these factors contributed to the rise in neonatal mortality between the two periods. Gestational age at birth is closely linked to neonatal mortality, with evidence indicating that, over the past decade, mortality rates have changed the most at the viability threshold (ie, babies born at 22 or 23 weeks' gestation).38 A further limitation of this study was the absence of individual level socioeconomic data in the French National Health Data System, which precluded analyses of individual and contextual effects.39 In accordance with international recommendations, we only included births starting at 22 weeks' gestation, so caution is needed when comparing with other studies that include all live births, regardless of gestational age.40 41 Also, the most granular geographic zone in the health data system is the municipality, which covers varying population sizes with differences in social status, especially in urban areas. This heterogeneity might result in an underestimation of the social gradient.42Conclusions In this study, the socioeconomic characteristics of the mother's municipality of residence were strongly associated with the risk of neonatal death in France. Furthermore, births in deprived municipalities have been the most affected by the recent worsening in neonatal mortality outcomes in France. These geographical contrasts provide an opportunity to investigate underlying contributing social and policy factors and to promote the development of public health interventions targeted to the populations most in need.SP110.1136/bmjmed-2024-000954.supp1Supplementary data",
  "title": "Evaluation of area based socioeconomic inequalities and neonatal mortality rates in France: national population based study",
  "uid": "7a0c74c3-4df9-528c-8b49-ec58567a145a"
}