Rosaleen Peggy Cornish
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Rosaleen Peggy Cornish has a total of 90 co-authors. Here are the top co-authors by number of shared publications:
Kate Tilling
Population Health Sciences, Bristol Medical School, University of Bristol; MRC Integrative Epidemiology Unit at the University of Bristol
9 shared publications- Kate Tilling
MRC Integrative Epidemiology Unit, University of Bristol, Bristol, UK
9 shared publications - Andy Boyd
UK Longitudinal Linkage Collaboration Bristol University of Bristol
8 shared publications John Macleod
Population Health Sciences, Bristol Medical School, University of Bristol; National Institute for Health Research Applied Research Collaboration (NIHR ARC West), University Hospitals Bristol and Weston NHS Foundation Trust
8 shared publications- John Macleod
University of Bristol
8 shared publications - Alison Teyhan
Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, UK
5 shared publications
Latest Publications
- Open Access
- 5 November 2026
- Published by Wiley in Pharmacoepidemiology and Drug Safety
- Vol. 35 (11)
Monoclonal Antibody Dispensing Around Pregnancy: A Descriptive Analysis Using Electronic Health Records in Italy
Purpose: To describe trends in dispensing of monoclonal antibody biologics (mAbs) for autoimmune conditions during and around pregnancy. Methods: We conducted a population‐wide study of women of reproductive age (14–49 years) resident in Lombardy, Italy, between 2012 and 2024. We described trends over time in the prevalence of mAb dispensations among women of reproductive age, and during pregnancy. For each pregnancy, we assessed the prevalence of mAb dispensations before, during, and after pregnancy in trimesters. We described dispensing patterns and explored maternal factors associated with discontinuation. Results: We included 1 871 929 women of reproductive age and 855 865 pregnancies. Prevalence of mAb dispensing during pregnancy increased over 60‐fold over the study period, from 0.0041% in 2012 to 0.27% in 2024. Pregnancy affected mAb dispensing, with prevalence decreasing from 0.081% before pregnancy to 0.051% by the third trimester. Over half ( n = 385/722, 53.3%) of pre‐existing mAb users discontinued before or during pregnancy, while switching between mAbs during pregnancy was rare ( n = 24/722, 3.3%). Discontinuation decreased over time, and varied substantially between mAbs. We found limited evidence that sociodemographic factors were associated with discontinuation, but that some health factors may be, such as use of assisted reproductive technology (OR = 1.92, 95% CI: 0.98–3.77). Conclusions: Italian population‐wide data from 2012 to 2024 show an increase in mAbs dispensed during pregnancy, and fewer instances of discontinuing these drugs over time. These trends may reflect increasing drug availability, accumulating safety evidence, and recent changes in guidelines for mAb use in pregnancy.
- 5 November 2026
- Published by Wiley in Pharmacoepidemiology and Drug Safety
- Vol. 35 (11)
Open AccessMonoclonal Antibody Dispensing Around Pregnancy: A Descriptive Analysis Using Electronic Health Records in Italy
Purpose: To describe trends in dispensing of monoclonal antibody biologics (mAbs) for autoimmune conditions during and around pregnancy. Methods: We conducted a population‐wide study of women of reproductive age (14–49 years) resident in Lombardy, Italy, between 2012 and 2024. We described trends over time in the prevalence of mAb dispensations among women of reproductive age, and during pregnancy. For each pregnancy, we assessed the prevalence of mAb dispensations before, during, and after pregnancy in trimesters. We described dispensing patterns and explored maternal factors associated with discontinuation. Results: We included 1 871 929 women of reproductive age and 855 865 pregnancies. Prevalence of mAb dispensing during pregnancy increased over 60‐fold over the study period, from 0.0041% in 2012 to 0.27% in 2024. Pregnancy affected mAb dispensing, with prevalence decreasing from 0.081% before pregnancy to 0.051% by the third trimester. Over half ( n = 385/722, 53.3%) of pre‐existing mAb users discontinued before or during pregnancy, while switching between mAbs during pregnancy was rare ( n = 24/722, 3.3%). Discontinuation decreased over time, and varied substantially between mAbs. We found limited evidence that sociodemographic factors were associated with discontinuation, but that some health factors may be, such as use of assisted reproductive technology (OR = 1.92, 95% CI: 0.98–3.77). Conclusions: Italian population‐wide data from 2012 to 2024 show an increase in mAbs dispensed during pregnancy, and fewer instances of discontinuing these drugs over time. These trends may reflect increasing drug availability, accumulating safety evidence, and recent changes in guidelines for mAb use in pregnancy.
- Open Access
The Long-Term Outcomes of Neural Tube Defects in Eastern Africa: A Systematic Review and Meta-Analysis
Background: Neural tube defects (NTDs) are preventable congenital malformations that disproportionately affect low and middle-income countries and contribute to disability and mortality. Evidence on the long-term outcomes of children and adolescents with NTDs in Eastern Africa has not been comprehensively synthesised. We conducted a systematic review and meta-analysis to assess survival, complications, and functional outcomes among children with NTDs in this region. Methods: We searched PubMed, MEDLINE (Ovid), Cochrane Library, Web of Science, and Africa Index Medicus from the earliest records to April 2025. Two reviewers independently screened studies, assessed quality, and extracted data. We included studies reporting outcomes beyond one year of age, except mortality, which was assessed from birth onwards. Random-effects meta-analyses were undertaken when at least two sufficiently studies were available; otherwise, findings were synthesised narratively. Results: Of 597 articles screened, 16 studies involving 2,340 children with NTDs met the inclusion criteria. Pooled cumulative mortality was 23% (95% CI: 12–36%) in the neonatal period, 9% (95% CI: 2–33%) during infancy, 22% (95% CI: 16–28%) in toddlers, 37% (95% CI: 30–44%) in pre-school-aged children and 45% (95% CI: 36–54%) in school-aged children. The pooled prevalence of hydrocephalus was 41% (95% CI: 34–48%) with little variation by age. Neurogenic bladder increased from 53% (95% CI 49-81%) in toddlers to 83% (95 % CI 72-91%) in adolescents, while impaired mobility affected about 61% (95% CI 48-72%) of adolescents. School enrolment was 53% (95% CI: 39-67%), with 9% (95% CI: 4%–19%) in specialized education. Speech, hearing, and bowel dysfunction were understudied (≤ 2 studies each). Conclusion: Many children with NTDs in the Eastern African region survive beyond infancy but frequently experience hydrocephalus, neurogenic bladder and motor impairment. Longitudinal studies, context-specific guidelines, and follow-up systems are urgently needed to improve care and long-term outcomes. PROSPERO registered number: CRD420251019868 What is already known on this topic: NTDs carry a substantial burden of mortality and lifelong disability. Survival and long-term outcomes in high-income settings are well-documented, but evidence from the Eastern African region is limited and fragmented. Existing studies provide an incomplete picture of survival, complications, functional outcomes, and school participation beyond infancy What this study adds: This review demonstrates that mortality among children with NTDs increases substantially throughout childhood. Hydrocephalus, neurogenic bladder, impaired mobility, and low school enrolment are common, while bowel, speech, and hearing outcomes remain poorly reported. How this study might affect research practise or policy: These findings underscore the need for structured long-term follow-up and multidisciplinary care for children with NTDs encompassing rehabilitation, continence care, and inclusive education. Future research should prioritise prospective cohort designs with standardised outcome measurement to generate context-specific evidence
Open AccessThe Long-Term Outcomes of Neural Tube Defects in Eastern Africa: A Systematic Review and Meta-Analysis
Background: Neural tube defects (NTDs) are preventable congenital malformations that disproportionately affect low and middle-income countries and contribute to disability and mortality. Evidence on the long-term outcomes of children and adolescents with NTDs in Eastern Africa has not been comprehensively synthesised. We conducted a systematic review and meta-analysis to assess survival, complications, and functional outcomes among children with NTDs in this region. Methods: We searched PubMed, MEDLINE (Ovid), Cochrane Library, Web of Science, and Africa Index Medicus from the earliest records to April 2025. Two reviewers independently screened studies, assessed quality, and extracted data. We included studies reporting outcomes beyond one year of age, except mortality, which was assessed from birth onwards. Random-effects meta-analyses were undertaken when at least two sufficiently studies were available; otherwise, findings were synthesised narratively. Results: Of 597 articles screened, 16 studies involving 2,340 children with NTDs met the inclusion criteria. Pooled cumulative mortality was 23% (95% CI: 12–36%) in the neonatal period, 9% (95% CI: 2–33%) during infancy, 22% (95% CI: 16–28%) in toddlers, 37% (95% CI: 30–44%) in pre-school-aged children and 45% (95% CI: 36–54%) in school-aged children. The pooled prevalence of hydrocephalus was 41% (95% CI: 34–48%) with little variation by age. Neurogenic bladder increased from 53% (95% CI 49-81%) in toddlers to 83% (95 % CI 72-91%) in adolescents, while impaired mobility affected about 61% (95% CI 48-72%) of adolescents. School enrolment was 53% (95% CI: 39-67%), with 9% (95% CI: 4%–19%) in specialized education. Speech, hearing, and bowel dysfunction were understudied (≤ 2 studies each). Conclusion: Many children with NTDs in the Eastern African region survive beyond infancy but frequently experience hydrocephalus, neurogenic bladder and motor impairment. Longitudinal studies, context-specific guidelines, and follow-up systems are urgently needed to improve care and long-term outcomes. PROSPERO registered number: CRD420251019868 What is already known on this topic: NTDs carry a substantial burden of mortality and lifelong disability. Survival and long-term outcomes in high-income settings are well-documented, but evidence from the Eastern African region is limited and fragmented. Existing studies provide an incomplete picture of survival, complications, functional outcomes, and school participation beyond infancy What this study adds: This review demonstrates that mortality among children with NTDs increases substantially throughout childhood. Hydrocephalus, neurogenic bladder, impaired mobility, and low school enrolment are common, while bowel, speech, and hearing outcomes remain poorly reported. How this study might affect research practise or policy: These findings underscore the need for structured long-term follow-up and multidisciplinary care for children with NTDs encompassing rehabilitation, continence care, and inclusive education. Future research should prioritise prospective cohort designs with standardised outcome measurement to generate context-specific evidence
- 1 July 2025
- Published by Bristol University Press in Longitudinal and Life Course Studies
- Vol. 16 (3) , 355-372
A ‘double disadvantage’: neurodevelopmental profile and poverty confer synergistic risk of youth justice involvement
Background: It is known that children with lower neurodevelopmental abilities and children who live in poverty are at increased risk of contact with the criminal justice system, but whether these two risk factors interact is unknown. Methods: We used linked data from 519,920 children born in 2001/2002, who attended school in England. Prospective analysis tested whether Early Years Foundation Stage Profile (EYFSP) indexing functional development across six domains and pupil characteristics indexing poverty at age 4/5, predicted a caution or conviction by age 15/16. Findings: Higher scores on the EYFSP (indicating better functional development) were associated with decreased odds of a caution or conviction by the age of 15/16. Being eligible for Free School Meals (living in absolute poverty) at age 4/5 was associated with increased odds of having a caution or conviction by age 15/16. There was a statistical interaction indicating synergistic risk conferred by these two risk factors. Interpretation: Poverty and lower functional development independently confer risk of criminalisation, and together confer a ‘double disadvantage’. Structural public health interventions should reduce the criminalisation of children by boosting functional development through Special Educational Needs support in schools, and by reducing the number of children living in absolute poverty.
- 1 July 2025
- Published by Bristol University Press in Longitudinal and Life Course Studies
- Vol. 16 (3) , 355-372
A ‘double disadvantage’: neurodevelopmental profile and poverty confer synergistic risk of youth justice involvement
Background: It is known that children with lower neurodevelopmental abilities and children who live in poverty are at increased risk of contact with the criminal justice system, but whether these two risk factors interact is unknown. Methods: We used linked data from 519,920 children born in 2001/2002, who attended school in England. Prospective analysis tested whether Early Years Foundation Stage Profile (EYFSP) indexing functional development across six domains and pupil characteristics indexing poverty at age 4/5, predicted a caution or conviction by age 15/16. Findings: Higher scores on the EYFSP (indicating better functional development) were associated with decreased odds of a caution or conviction by the age of 15/16. Being eligible for Free School Meals (living in absolute poverty) at age 4/5 was associated with increased odds of having a caution or conviction by age 15/16. There was a statistical interaction indicating synergistic risk conferred by these two risk factors. Interpretation: Poverty and lower functional development independently confer risk of criminalisation, and together confer a ‘double disadvantage’. Structural public health interventions should reduce the criminalisation of children by boosting functional development through Special Educational Needs support in schools, and by reducing the number of children living in absolute poverty.
- Open Access
Selection bias due to omitting interactions from inverse probability weighting
Inverse probability weighting (IPW) is often used to adjust for selection bias, typically using a simple logit model without interactions as a missingness model. However, the size of the selection bias depends on the interaction between exposure and outcome in their effect on selection - implying that it may be important to include interactions in the IPW model. Via a simulation and a real-data application we compare the performance of IPW with and without interaction terms to estimate a regression coefficient. The simulation study shows that IPW including interactions gives less biased estimates than IPW without interactions in all scenarios studied. Importantly, IPW using a logit model with no interactions often gives estimates close to the complete case analysis (CCA) - perhaps giving false reassurance that results are robust to selection bias. The real-data application investigates the association between unemployment and sleep duration, using data from Understanding Society. IPW including interactions suggests that unemployment is associated with a reduction in sleep duration of around 23 (9, 38) minutes, compared to 27 (14, 40) minutes for IPW without interactions, and 31 (19, 43) minutes for CCA. We strongly recommend including interactions in missingness models to adjust for selection bias.
Open AccessSelection bias due to omitting interactions from inverse probability weighting
Inverse probability weighting (IPW) is often used to adjust for selection bias, typically using a simple logit model without interactions as a missingness model. However, the size of the selection bias depends on the interaction between exposure and outcome in their effect on selection - implying that it may be important to include interactions in the IPW model. Via a simulation and a real-data application we compare the performance of IPW with and without interaction terms to estimate a regression coefficient. The simulation study shows that IPW including interactions gives less biased estimates than IPW without interactions in all scenarios studied. Importantly, IPW using a logit model with no interactions often gives estimates close to the complete case analysis (CCA) - perhaps giving false reassurance that results are robust to selection bias. The real-data application investigates the association between unemployment and sleep duration, using data from Understanding Society. IPW including interactions suggests that unemployment is associated with a reduction in sleep duration of around 23 (9, 38) minutes, compared to 27 (14, 40) minutes for IPW without interactions, and 31 (19, 43) minutes for CCA. We strongly recommend including interactions in missingness models to adjust for selection bias.
- Open Access
- 23 January 2025
- Published by Swansea University in International Journal of Population Data Science
- Vol. 10 (1) , 2391
Measuring serious violence perpetration: comparison of police-recorded and self-reported data in a UK cohort
IntroductionDetermining risk factors and consequences of serious violence requires accurate measures of violence. Self-reported and police-recorded offending are subject to different sources of bias. ObjectivesTo compare risk of self-reported and police-recorded serious violence perpetration in late adolescence and early adulthood using linked UK birth cohort and police data, to examine the association between cohort participation and police-recorded violence, and to use police-records to impute missing self-reported data on violence MethodsWe included individuals in the Avon Longitudinal Study of Parents and Children (ALSPAC) who had been informed about the study's use of their linked data and had not opted out of linkage to police records (n = 12,662). We used descriptive statistics and logistic regression to address our objectives. Multiple imputation using chained equations was used to impute self-reported violence data to examine the likely impact of missing data on estimates of prevalence. ResultsSelf-reported violence perpetration in the past year ranged from 5.3% (at 25 years) to 12.9% (at 20 years) among males and 3.2% (at 17, 22, 24 and 25 years) to 6.4% (at 18 years) among females. Police-recorded serious violence was lower at all ages, peaking at 17-18 years (1.7% among males, 0.5% among females). Study participation was lower among people who had or went on to have a police record for serious violence; as a result, the prevalence of self-reported violence in the imputed data was higher (compared to observed data) at all ages. ConclusionsOverall, our study demonstrates the difficulties in measuring violence. While we have shown that a key advantage of linkage to police records is it enables outcomes to be measured irrespective of study participation, police data undercounts serious violence. Further, observational studies may also underestimate violence perpetration as individuals with police-recorded serious violence are less likely to participate in research. Therefore, while record linkage allows the advantages of both official police records and self-reported measures to be exploited, it does not negate their limitations.
- 23 January 2025
- Published by Swansea University in International Journal of Population Data Science
- Vol. 10 (1) , 2391
Open AccessMeasuring serious violence perpetration: comparison of police-recorded and self-reported data in a UK cohort
IntroductionDetermining risk factors and consequences of serious violence requires accurate measures of violence. Self-reported and police-recorded offending are subject to different sources of bias. ObjectivesTo compare risk of self-reported and police-recorded serious violence perpetration in late adolescence and early adulthood using linked UK birth cohort and police data, to examine the association between cohort participation and police-recorded violence, and to use police-records to impute missing self-reported data on violence MethodsWe included individuals in the Avon Longitudinal Study of Parents and Children (ALSPAC) who had been informed about the study's use of their linked data and had not opted out of linkage to police records (n = 12,662). We used descriptive statistics and logistic regression to address our objectives. Multiple imputation using chained equations was used to impute self-reported violence data to examine the likely impact of missing data on estimates of prevalence. ResultsSelf-reported violence perpetration in the past year ranged from 5.3% (at 25 years) to 12.9% (at 20 years) among males and 3.2% (at 17, 22, 24 and 25 years) to 6.4% (at 18 years) among females. Police-recorded serious violence was lower at all ages, peaking at 17-18 years (1.7% among males, 0.5% among females). Study participation was lower among people who had or went on to have a police record for serious violence; as a result, the prevalence of self-reported violence in the imputed data was higher (compared to observed data) at all ages. ConclusionsOverall, our study demonstrates the difficulties in measuring violence. While we have shown that a key advantage of linkage to police records is it enables outcomes to be measured irrespective of study participation, police data undercounts serious violence. Further, observational studies may also underestimate violence perpetration as individuals with police-recorded serious violence are less likely to participate in research. Therefore, while record linkage allows the advantages of both official police records and self-reported measures to be exploited, it does not negate their limitations.
- Open Access
- 27 August 2024
- Published by Oxford University Press (OUP) in American Journal of Epidemiology
- Vol. 194 (6) , 1756-1763
Analyses using multiple imputation need to consider missing data in auxiliary variables
Auxiliary variables are used in multiple imputation (MI) to reduce bias and increase efficiency. These variables may often themselves be incomplete. We explored how missing data in auxiliary variables influenced estimates obtained from MI. We implemented a simulation study with 3 different missing data mechanisms for the outcome. We then examined the impact of increasing proportions of missing data and different missingness mechanisms for the auxiliary variable on bias of an unadjusted linear regression coefficient and the fraction of missing information. We illustrate our findings with an applied example in the Avon Longitudinal Study of Parents and Children. We found that where complete records analyses were biased, increasing proportions of missing data in auxiliary variables, under any missing data mechanism, reduced the ability of MI including the auxiliary variable to mitigate this bias. Where there was no bias in the complete records analysis, inclusion of a missing not at random auxiliary variable in MI introduced bias of potentially important magnitude (up to 17% of the effect size in our simulation). Careful consideration of the quantity and nature of missing data in auxiliary variables needs to be made when selecting them for use in MI models.
- 27 August 2024
- Published by Oxford University Press (OUP) in American Journal of Epidemiology
- Vol. 194 (6) , 1756-1763
Open AccessAnalyses using multiple imputation need to consider missing data in auxiliary variables
Auxiliary variables are used in multiple imputation (MI) to reduce bias and increase efficiency. These variables may often themselves be incomplete. We explored how missing data in auxiliary variables influenced estimates obtained from MI. We implemented a simulation study with 3 different missing data mechanisms for the outcome. We then examined the impact of increasing proportions of missing data and different missingness mechanisms for the auxiliary variable on bias of an unadjusted linear regression coefficient and the fraction of missing information. We illustrate our findings with an applied example in the Avon Longitudinal Study of Parents and Children. We found that where complete records analyses were biased, increasing proportions of missing data in auxiliary variables, under any missing data mechanism, reduced the ability of MI including the auxiliary variable to mitigate this bias. Where there was no bias in the complete records analysis, inclusion of a missing not at random auxiliary variable in MI introduced bias of potentially important magnitude (up to 17% of the effect size in our simulation). Careful consideration of the quantity and nature of missing data in auxiliary variables needs to be made when selecting them for use in MI models.
- Open Access
- 24 August 2024
- Published by Center for Open Science in SocArXiv
Exclusion from school and risk of serious violence: a target trial emulation study
Expulsion or exclusion from school is a common event in the lives of people who have experience of the criminal justice system. Exclusion underpins the ‘school-to-prison pipeline’ and is, accordingly, one of the most contentious policies used in education systems. Critics of exclusion powers claim that it is a direct cause of later offending behaviour, among other adverse outcomes, while supporters of this power argue that it is necessary for maintaining school discipline and safety. Evidence for or against a causal effect of exclusion on offending is inhibited by random allocation of exclusion not being available on ethical grounds and the generally low internal validity of studies in this area. To advance understanding of the causal connection between school exclusion and offending – specifically, serious violent offending – this study leverages a recent national linkage of educational and criminal justice records in England to emulate a randomised controlled trial using a target trial framework. Across more than 20,000 matched pairs of excluded and not excluded children, within twelve months of target trial entry, the hazard ratio for serious violence was 2.04 (95% CI: 1.83, 2.28) and for homicide/near-miss homicide was 2.36 (95% CI: 1.07, 5.20). We discuss the implications of these findings for theory and policy in education and criminal justice as well as discussing the extent to which the observed relationships can be considered causal.
- 24 August 2024
- Published by Center for Open Science in SocArXiv
Open AccessExclusion from school and risk of serious violence: a target trial emulation study
Expulsion or exclusion from school is a common event in the lives of people who have experience of the criminal justice system. Exclusion underpins the ‘school-to-prison pipeline’ and is, accordingly, one of the most contentious policies used in education systems. Critics of exclusion powers claim that it is a direct cause of later offending behaviour, among other adverse outcomes, while supporters of this power argue that it is necessary for maintaining school discipline and safety. Evidence for or against a causal effect of exclusion on offending is inhibited by random allocation of exclusion not being available on ethical grounds and the generally low internal validity of studies in this area. To advance understanding of the causal connection between school exclusion and offending – specifically, serious violent offending – this study leverages a recent national linkage of educational and criminal justice records in England to emulate a randomised controlled trial using a target trial framework. Across more than 20,000 matched pairs of excluded and not excluded children, within twelve months of target trial entry, the hazard ratio for serious violence was 2.04 (95% CI: 1.83, 2.28) and for homicide/near-miss homicide was 2.36 (95% CI: 1.07, 5.20). We discuss the implications of these findings for theory and policy in education and criminal justice as well as discussing the extent to which the observed relationships can be considered causal.
- Open Access
- 1 March 2024
- Published by Center for Open Science in SocArXiv
Measuring serious violence: comparison of police-recorded and self-reported data
Determining risk factors and consequences of serious violence requires accurate measures of violence. Self-reported and police-recorded offending are subject to different sources of bias.Our objectives were to compare risk of self-reported and police-recorded serious violence in late adolescence and early adulthood using linked UK birth cohort and police data, and to examine the association between cohort participation at different ages and police-recorded violence.We included individuals in the Avon Longitudinal Study of Parents and Children (ALSPAC) who had received fair processing materials and had not opted out of linkage to police records (n=12,662). Self-reported violence in the past year ranged from 5.3% (25 years) to 12.9% (20 years) among males and 3.2% (17, 22, 24 and 25 years) to 6.4% (18 years) among females. Police-recorded serious violence was lower at all ages, peaking at 17-18 years (1.7% among males, 0.5% among females). Study participation was lower among people who went on to have or had a record for police-recorded serious violence. At younger ages (in this case participation by mothers) this association was explained by socio-demographic and other factors.A key advantage of linkage to police records is it enables outcomes to be measured irrespective of study participation. However, police data undercounts serious violence. Further, observational studies may also underestimate violence as individuals with police-recorded serious violence are less likely to participate in research. Combining the two sources of data may overcome some of these issues, but careful thought would need to be given to the potential sources of bias present in each.
- 1 March 2024
- Published by Center for Open Science in SocArXiv
Open AccessMeasuring serious violence: comparison of police-recorded and self-reported data
Determining risk factors and consequences of serious violence requires accurate measures of violence. Self-reported and police-recorded offending are subject to different sources of bias.Our objectives were to compare risk of self-reported and police-recorded serious violence in late adolescence and early adulthood using linked UK birth cohort and police data, and to examine the association between cohort participation at different ages and police-recorded violence.We included individuals in the Avon Longitudinal Study of Parents and Children (ALSPAC) who had received fair processing materials and had not opted out of linkage to police records (n=12,662). Self-reported violence in the past year ranged from 5.3% (25 years) to 12.9% (20 years) among males and 3.2% (17, 22, 24 and 25 years) to 6.4% (18 years) among females. Police-recorded serious violence was lower at all ages, peaking at 17-18 years (1.7% among males, 0.5% among females). Study participation was lower among people who went on to have or had a record for police-recorded serious violence. At younger ages (in this case participation by mothers) this association was explained by socio-demographic and other factors.A key advantage of linkage to police records is it enables outcomes to be measured irrespective of study participation. However, police data undercounts serious violence. Further, observational studies may also underestimate violence as individuals with police-recorded serious violence are less likely to participate in research. Combining the two sources of data may overcome some of these issues, but careful thought would need to be given to the potential sources of bias present in each.
- Top Cited PapersOpen Access
- 5 January 2024
- Published by Springer Nature in BMC Medicine
- Vol. 22 (1) , 1-13
Maternal pre-pregnancy body mass index and risk of preterm birth: a collaboration using large routine health datasets
Background: Preterm birth (PTB) is a leading cause of child morbidity and mortality. Evidence suggests an increased risk with both maternal underweight and obesity, with some studies suggesting underweight might be a greater factor in spontaneous PTB (SPTB) and that the relationship might vary by parity. Previous studies have largely explored established body mass index (BMI) categories. Our aim was to compare associations of maternal pre-pregnancy BMI with any PTB, SPTB and medically indicated PTB (MPTB) among nulliparous and parous women across populations with differing characteristics, and to identify the optimal BMI with lowest risk for these outcomes. Methods: We used three UK datasets, two USA datasets and one each from South Australia, Norway and Denmark, together including just under 29 million pregnancies resulting in a live birth or stillbirth after 24 completed weeks gestation. Fractional polynomial multivariable logistic regression was used to examine the relationship of maternal BMI with any PTB, SPTB and MPTB, among nulliparous and parous women separately. The results were combined using a random effects meta-analysis. The estimated BMI at which risk was lowest was calculated via differentiation and a 95% confidence interval (CI) obtained using bootstrapping. Results: We found non-linear associations between BMI and all three outcomes, across all datasets. The adjusted risk of any PTB and MPTB was elevated at both low and high BMIs, whereas the risk of SPTB was increased at lower levels of BMI but remained low or increased only slightly with higher BMI. In the meta-analysed data, the lowest risk of any PTB was at a BMI of 22.5 kg/m2 (95% CI 21.5, 23.5) among nulliparous women and 25.9 kg/m2 (95% CI 24.1, 31.7) among multiparous women, with values of 20.4 kg/m2 (20.0, 21.1) and 22.2 kg/m2 (21.1, 24.3), respectively, for MPTB; for SPTB, the risk remained roughly largely constant above a BMI of around 25–30 kg/m2 regardless of parity. Conclusions: Consistency of findings across different populations, despite differences between them in terms of the time period covered, the BMI distribution, missing data and control for key confounders, suggests that severe under- and overweight may play a role in PTB risk.
- 5 January 2024
- Published by Springer Nature in BMC Medicine
- Vol. 22 (1) , 1-13
Top Cited PapersOpen AccessMaternal pre-pregnancy body mass index and risk of preterm birth: a collaboration using large routine health datasets
Background: Preterm birth (PTB) is a leading cause of child morbidity and mortality. Evidence suggests an increased risk with both maternal underweight and obesity, with some studies suggesting underweight might be a greater factor in spontaneous PTB (SPTB) and that the relationship might vary by parity. Previous studies have largely explored established body mass index (BMI) categories. Our aim was to compare associations of maternal pre-pregnancy BMI with any PTB, SPTB and medically indicated PTB (MPTB) among nulliparous and parous women across populations with differing characteristics, and to identify the optimal BMI with lowest risk for these outcomes. Methods: We used three UK datasets, two USA datasets and one each from South Australia, Norway and Denmark, together including just under 29 million pregnancies resulting in a live birth or stillbirth after 24 completed weeks gestation. Fractional polynomial multivariable logistic regression was used to examine the relationship of maternal BMI with any PTB, SPTB and MPTB, among nulliparous and parous women separately. The results were combined using a random effects meta-analysis. The estimated BMI at which risk was lowest was calculated via differentiation and a 95% confidence interval (CI) obtained using bootstrapping. Results: We found non-linear associations between BMI and all three outcomes, across all datasets. The adjusted risk of any PTB and MPTB was elevated at both low and high BMIs, whereas the risk of SPTB was increased at lower levels of BMI but remained low or increased only slightly with higher BMI. In the meta-analysed data, the lowest risk of any PTB was at a BMI of 22.5 kg/m2 (95% CI 21.5, 23.5) among nulliparous women and 25.9 kg/m2 (95% CI 24.1, 31.7) among multiparous women, with values of 20.4 kg/m2 (20.0, 21.1) and 22.2 kg/m2 (21.1, 24.3), respectively, for MPTB; for SPTB, the risk remained roughly largely constant above a BMI of around 25–30 kg/m2 regardless of parity. Conclusions: Consistency of findings across different populations, despite differences between them in terms of the time period covered, the BMI distribution, missing data and control for key confounders, suggests that severe under- and overweight may play a role in PTB risk.
- Open Access
Analyses using multiple imputation need to consider missing data in auxiliary variables
Auxiliary variables are used in multiple imputation (MI) to reduce bias and increase efficiency. These variables may often themselves be incomplete. We explored how missing data in auxiliary variables influenced estimates obtained from MI. We implemented a simulation study with three different missing data mechanisms for the outcome. We then examined the impact of increasing proportions of missing data and different missingness mechanisms for the auxiliary variable on bias of an unadjusted linear regression coefficient and the fraction of missing information. We illustrate our findings with an applied example in the Avon Longitudinal Study of Parents and Children. We found that where complete records analyses were biased, increasing proportions of missing data in auxiliary variables, under any missing data mechanism, reduced the ability of MI including the auxiliary variable to mitigate this bias. Where there was no bias in the complete records analysis, inclusion of a missing not at random auxiliary variable in MI introduced bias of potentially important magnitude (up to 17% of the effect size in our simulation). Careful consideration of the quantity and nature of missing data in auxiliary variables needs to be made when selecting them for use in MI models.
Open AccessAnalyses using multiple imputation need to consider missing data in auxiliary variables
Auxiliary variables are used in multiple imputation (MI) to reduce bias and increase efficiency. These variables may often themselves be incomplete. We explored how missing data in auxiliary variables influenced estimates obtained from MI. We implemented a simulation study with three different missing data mechanisms for the outcome. We then examined the impact of increasing proportions of missing data and different missingness mechanisms for the auxiliary variable on bias of an unadjusted linear regression coefficient and the fraction of missing information. We illustrate our findings with an applied example in the Avon Longitudinal Study of Parents and Children. We found that where complete records analyses were biased, increasing proportions of missing data in auxiliary variables, under any missing data mechanism, reduced the ability of MI including the auxiliary variable to mitigate this bias. Where there was no bias in the complete records analysis, inclusion of a missing not at random auxiliary variable in MI introduced bias of potentially important magnitude (up to 17% of the effect size in our simulation). Careful consideration of the quantity and nature of missing data in auxiliary variables needs to be made when selecting them for use in MI models.