Comparing methods of analysing datasets with small clusters: case studies using four paediatric datasets

Accounting for Twins and Other Multiple Births in Perinatal Studies Conducted Using Healthcare Administration Data

The analysis of perinatal studies is complicated by twins and other multiple births even when they are not the exposure, outcome, or a confounder of interest. Common approaches to handling multiples in studies of infant outcomes include restriction to singletons, counting outcomes at the pregnancy-level (i.e., by counting if at least one twin experienced a binary outcome), or infant-level analysis including all infants and, typically, accounting for clustering of outcomes by using generalised estimating equations or mixed effects models. Several healthcare administration databases only support restriction to singletons or pregnancy-level approaches. For example, in MarketScan insurance claims data, diagnoses in twins are often assigned to a single infant identifier, thereby preventing ascertainment of infant-level outcomes among multiples. Different approaches correspond to different causal questions, produce different estimands, and often rely on different assumptions. We demonstrate the differences that can arise from these different approaches using Monte Carlo simulations, algebraic formulas, and an applied example. Furthermore, we provide guidance on the handling of multiples in perinatal studies when using healthcare administration data.

A Smartphone App for Promoting Mental Well-being and Awareness of Anxious Symptoms in Adolescents: A Pilot Cluster Randomized Controlled Trial

Objective: To design a smartphone app and evaluate its effect on promoting mental well-being and awareness of anxious symptoms in adolescents. Materials and Methods: A pilot cluster randomized controlled trial was conducted with two secondary schools (390 students, mean age 13.1 years) randomized to the intervention and control groups. An app was designed to encourage "Sharing, Mind, and Enjoyment (SME)" between students and their parents (e.g., express gratitude), including interaction and game elements. The app was used daily over 1 month. The waitlist control group was offered the app after completing all assessments. The primary outcomes were changes in SME behaviors measured at 1 month and 3 months. Secondary outcomes included subjective happiness, well-being, personal health and happiness, family health, happiness and harmony, self-perceived knowledge, and understanding of anxious symptoms. Two focus groups of students and three individual in-depth interviews of community partners were conducted. Results: Seventeen of 152 students (11.2%) in the intervention group used the app together with parents (adherent subgroup) and 69 (45.4%) used it without involving parents. The intervention group did not show significant difference in the change of SME behaviors at 1- or 3-month follow-up compared with the control group. However, the intervention group reported greater increase in the awareness of anxious symptoms at follow-ups than the control group (d = 0.52 at 1 month and d = 0.43 at 3 month, both P < 0.001). Post hoc analysis showed a significantly greater increase in SME-related behaviors in the adherent subgroup than the control group at 3 months (d = 0.46, P = 0.04). The interviews found favorable changes in app users, but motivation to use the app was low in general. Both students and community partners suggested primary school students would be more receptive users. Conclusions: The app did not show effectiveness in increasing SME behaviors of students, but increased awareness of anxious symptoms. Further improvements and tests are warranted. Trial Registration: ClinicalTrials.gov NCT03361475.

The effect of missing levels of nesting in multilevel analysis

Multilevel analysis is an appropriate and powerful tool for analyzing hierarchical structure data widely applied from public health to genomic data. In practice, however, we may lose the information on multiple nesting levels in the multilevel analysis since data may fail to capture all levels of hierarchy, or the top or intermediate levels of hierarchy are ignored in the analysis. In this study, we consider a multilevel linear mixed effect model (LMM) with single imputation that can involve all data hierarchy levels in the presence of missing top or intermediate-level clusters. We evaluate and compare the performance of a multilevel LMM with single imputation with other models ignoring the data hierarchy or missing intermediate-level clusters. To this end, we applied a multilevel LMM with single imputation and other models to hierarchically structured cohort data with some intermediate levels missing and to simulated data with various cluster sizes and missing rates of intermediate-level clusters. A thorough simulation study demonstrated that an LMM with single imputation estimates fixed coefficients and variance components of a multilevel model more accurately than other models ignoring data hierarchy or missing clusters in terms of mean squared error and coverage probability. In particular, when models ignoring data hierarchy or missing clusters were applied, the variance components of random effects were overestimated. We observed similar results from the analysis of hierarchically structured cohort data.

Very preterm birth and trajectories of domain-specific self-concept from childhood into adulthood

Self-concept refers to individuals' perceptions of themselves in specific domains and is closely related with their overall self-esteem. Lower self-esteem has been reported in those born preterm (<37 weeks gestation), but the development of self-concept has not been studied in this population. This study investigates whether differences in trajectories of domain-specific self-concepts are explained by premature birth or other risk factors, using the Bavarian Longitudinal Study (N = 460), a population-based study of very preterm (VP; <32 weeks gestation)/very low birth weight (VLBW; <1500 g) cohort and term-born controls. Trajectories of body and social self-concept from 6 to 26 years of age were estimated using latent class growth analysis. Regression models examined the effects of VP/VLBW and other individual, social, and family factors. Two trajectories - one stable and one decreasing - were identified for both self-concepts. VP/VLBW birth was associated with decreasing self-concept in both domains, although the effect of VP/VLBW on social self-concept was weakened in the adjusted analysis. Furthermore, mediated pathways were found from VP/VLBW to decreasing social self-concept via chronic bullying (β = 0.05, 95% CI [0.002, 0.12]) and motor impairments (β = 0.04, 95% CI [0.01, 0.07]), suggesting that negative self-concept in the VP/VLBW population is partially modifiable through improving peer relationships and motor impairments in childhood.

Computerised interpretation of the fetal heart rate during labour: a randomised controlled trial (INFANT)

BACKGROUND

Continuous electronic fetal monitoring (EFM) in labour is widely used and computerised interpretation has the potential to increase its utility.

OBJECTIVES

This trial aimed to find out whether or not the addition of decision support software to assist in the interpretation of the cardiotocograph (CTG) reduced the number of poor neonatal outcomes, and whether or not it was cost-effective.

DESIGN

Two-arm individually randomised controlled trial. The allocations were computer generated using stratified block randomisation employing variable block sizes. The trial was not masked.

SETTING

Labour wards in England, Scotland and the Republic of Ireland.

PARTICIPANTS

Women in labour having EFM, with a singleton or twin pregnancy, at ≥ 35 weeks' gestation.

INTERVENTIONS

Decision support or no decision support.

MAIN OUTCOME MEASURES

The primary outcomes were (1) a composite of poor neonatal outcome {intrapartum stillbirth or early neonatal death (excluding lethal congenital anomalies), or neonatal morbidity [defined as neonatal encephalopathy (NNE)], or admission to a neonatal unit within 48 hours for ≥ 48 hours (with evidence of feeding difficulties, respiratory illness or NNE when there was evidence of compromise at birth)}; and (2) developmental assessment at the age of 2 years in a subset of surviving children.

RESULTS

Between 6 January 2010 and 31 August 2013, 47,062 women were randomised and 46,042 were included in the primary analysis (22,987 in the decision support group and 23,055 in the no decision support group). The short-term primary outcome event rate was higher than anticipated. There was no evidence of a difference in the incidence of poor neonatal outcome between the groups: 0.7% (n = 172) of babies in the decision support group compared with 0.7% (n = 171) of babies in the no decision support group [adjusted risk ratio 1.01, 95% confidence interval (CI) 0.82 to 1.25]. There was no evidence of a difference in the long-term primary outcome of the Parent Report of Children's Abilities-Revised with a mean score of 98.0 points [standard deviation (SD) 33.8 points] in the decision support group and 97.2 points (SD 33.4 points) in the no decision support group (mean difference 0.63 points, 95% CI -0.98 to 2.25 points). No evidence of a difference was found for health resource use and total costs. There was evidence that decision support did change practice (with increased fetal blood sampling and a lower rate of repeated alerts).

LIMITATIONS

Staff in the control group may learn from exposure to the decision support arm of the trial, resulting in improved outcomes in the control arm. This was identified in the planning stage and felt to be unlikely to have a significant effect on the results. As this was a pragmatic trial, the response to CTG alerts was left to the attending clinicians.

CONCLUSIONS

This trial does not support the hypothesis that the use of computerised interpretation of the CTG in women who have EFM in labour improves the clinical outcomes for mothers or babies.

FUTURE WORK

There continues to be an urgent need to improve knowledge and training about the appropriate response to CTG abnormalities, including timely intervention.

TRIAL REGISTRATION

Current Controlled Trials ISRCTN98680152.

FUNDING

This project was funded by the National Institute for Health Research (NIHR) HTA programme and will be published in full in Health Technology Assessment; Vol. 22, No. 9. See the NIHR Journals Library website for further project information. Sara Kenyon was part funded by the NIHR Collaboration for Leadership in Applied Health Research and Care West Midlands.

Accounting for twin births in sample size calculations for randomised trials

BACKGROUND

Including twins in randomised trials leads to non-independence or clustering in the data. Clustering has important implications for sample size calculations, yet few trials take this into account. Estimates of the intracluster correlation coefficient (ICC), or the correlation between outcomes of twins, are needed to assist with sample size planning. Our aims were to provide ICC estimates for infant outcomes, describe the information that must be specified in order to account for clustering due to twins in sample size calculations, and develop a simple tool for performing sample size calculations for trials including twins.

METHODS

ICCs were estimated for infant outcomes collected in four randomised trials that included twins. The information required to account for clustering due to twins in sample size calculations is described. A tool that calculates the sample size based on this information was developed in Microsoft Excel and in R as a Shiny web app.

RESULTS

ICC estimates ranged between -0.12, indicating a weak negative relationship, and 0.98, indicating a strong positive relationship between outcomes of twins. Example calculations illustrate how the ICC estimates and sample size calculator can be used to determine the target sample size for trials including twins.

CONCLUSIONS

Clustering among outcomes measured on twins should be taken into account in sample size calculations to obtain the desired power. Our ICC estimates and sample size calculator will be useful for designing future trials that include twins. Publication of additional ICCs is needed to further assist with sample size planning for future trials.

Correlation between neonatal outcomes of twins depends on the outcome: secondary analysis of twelve randomised controlled trials

OBJECTIVE

To estimate the magnitude of the correlation between neonatal outcomes of twins and demonstrate how this information can be used in the design of randomised controlled trials (RCTs) in women with twin pregnancies.

DESIGN

Secondary analysis of data from 12 RCTs.

SETTING

Obstetric care in multiple countries, 2004-2012.

POPULATION OR SAMPLE

4504 twin pairs born to women who participated in RCTs to assess treatments given during pregnancy.

METHODS

Intraclass correlation coefficients (ICCs) were estimated using log-binomial and linear models.

MAIN OUTCOME MEASURES

Perinatal death, respiratory distress syndrome, bronchopulmonary dysplasia, intraventricular haemorrhage, necrotising enterocolitis, sepsis, neonatal intensive care unit admission, birthweight, low birthweight and two composite measures of adverse neonatal outcome.

RESULTS

ICCs for the composite measures of adverse neonatal outcome were all above 0.5, indicating moderate to strong correlation between adverse outcomes of twins. For individual neonatal outcomes, median ICCs across trials ranged from 0.13 to 0.79 depending on the outcome. An example illustrates how ICCs can be used in sample size calculations for RCTs in women with twin pregnancies.

CONCLUSIONS

The correlation between neonatal outcomes of twins varies considerably between outcomes and may be lower than expected. Our ICC estimates can be used for designing and analysing RCTs that recruit women with twin pregnancies and for performing meta-analyses that include such RCTs. Researchers are encouraged to report ICCs for neonatal outcomes in twins in their own RCTs.

TWEETABLE ABSTRACT

Correlation between neonatal outcomes of twins depends on the outcome and may be lower than expected.

Analysis of Randomised Trials Including Multiple Births When Birth Size Is Informative

BACKGROUND

Informative birth size occurs when the average outcome depends on the number of infants per birth. Although analysis methods have been proposed for handling informative birth size, their performance is not well understood. Our aim was to evaluate the performance of these methods and to provide recommendations for their application in randomised trials including infants from single and multiple births.

METHODS

Three generalised estimating equation (GEE) approaches were considered for estimating the effect of treatment on a continuous or binary outcome: cluster weighted GEEs, which produce treatment effects with a mother-level interpretation when birth size is informative; standard GEEs with an independence working correlation structure, which produce treatment effects with an infant-level interpretation when birth size is informative; and standard GEEs with an exchangeable working correlation structure, which do not account for informative birth size. The methods were compared through simulation and analysis of an example dataset.

RESULTS

Treatment effect estimates were affected by informative birth size in the simulation study when the effect of treatment in singletons differed from that in multiples (i.e. in the presence of a treatment group by multiple birth interaction). The strength of evidence supporting the effectiveness of treatment varied between methods in the example dataset.

CONCLUSIONS

Informative birth size is always a possibility in randomised trials including infants from both single and multiple births, and analysis methods should be pre-specified with this in mind. We recommend estimating treatment effects using standard GEEs with an independence working correlation structure to give an infant-level interpretation.

Severe sepsis in women with group B Streptococcus in pregnancy: an exploratory UK national case-control study

OBJECTIVE

To estimate the incidence of severe maternal sepsis due to group B Streptococcus (GBS) in the UK, and to investigate the associated outcomes for mother and infant.

DESIGN

National case-control study.

SETTING

All UK consultant-led maternity units.

PARTICIPANTS

30 women with confirmed or suspected severe GBS sepsis, and 757 control women.

MAIN OUTCOME MEASURES

Disease incidence, additional maternal morbidity, critical care admission, length of stay, infant infection, mortality.

RESULTS

The incidences of confirmed and presumed severe maternal GBS sepsis were 1.00 and 2.75 per 100,000 maternities, respectively, giving an overall incidence of 3.75 per 100,000. Compared with controls, severe GBS sepsis was associated with higher odds of additional maternal morbidity (OR 12.35, 95% CI 3.96 to 35.0), requiring level 2 (OR 39.3, 95% CI 16.0 to 99.3) or level 3 (OR 182, 95% CI 21.0 to 8701) care and longer hospital stay (median stay in cases and controls was 7 days (range 3-29 days) and 2 days (range 0-16 days), respectively, p<0.001). None of the women died. Severe maternal GBS sepsis was associated with higher odds of infant sepsis (OR 32.7, 95% CI 8.99 to 119.0); 79% of infants, however, did not develop sepsis. There were no associated stillbirths or neonatal deaths.

CONCLUSIONS

Severe maternal GBS sepsis is a rare occurrence in the UK. It is associated with adverse maternal and neonatal outcomes.

Survival, morbidity, growth and developmental delay for babies born preterm in low and middle income countries - a systematic review of outcomes measured

BACKGROUND

Premature birth is the leading cause of neonatal death and second leading in children under 5. Information on outcomes of preterm babies surviving the early neonatal period is sparse although it is considered a major determinant of immediate and long-term morbidity.

METHODS

Systematic review of studies reporting outcomes for preterm babies in low and middle income settings was conducted using electronic databases, citation tracking, expert recommendations and "grey literature". Reviewers screened titles, abstracts and articles. Data was extracted using inclusion and exclusion criteria, study site and facilities, assessment methods and outcomes of mortality, morbidity, growth and development. The Child Health Epidemiology Reference Group criteria (CHERG) were used to assess quality.

FINDINGS

Of 197 eligible publications, few (10.7%) were high quality (CHERG). The majority (83.3%) report on the outcome of a sample of preterm babies at time of birth or admission. Only 16.0% studies report population-based data using standardised mortality definitions. In 50.5% of studies, gestational age assessment method was unclear. Only 15.8% followed-up infants for 2 years or more. Growth was reported using standardised definitions but recommended morbidity definitions were rarely used. The criteria for assessment of neurodevelopmental outcomes was variable with few standardised tools - Bayley II was used in approximately 33% of studies, few studies undertook sensory assessments.

CONCLUSIONS

To determine the relative contribution of preterm birth to the burden of disease in children and to inform the planning of healthcare interventions to address this burden, a renewed understanding of the assessment and documentation of outcomes for babies born preterm is needed. More studies assessing outcomes for preterm babies who survive the immediate newborn period are needed. More consistent use of data is vital with clear and aligned definitions of health outcomes in newborn (preterm or term) and intervention packages aimed to save lives and improve health.

Multiples and parents of multiples prefer same arm randomization of siblings in neonatal trials

OBJECTIVE

Although common among Neonatal Intensive Care Units, multiples births are randomized inconsistently within trials, which can impact enrollment, analytical approach and trial outcomes. It is not known what randomization approach (same arm, different arm and independent randomization) is preferred by multiples and their families.

STUDY DESIGN

Surveys distributed to parents of multiples and adult multiples addressed the preferences on randomization by eliciting the most desired method and likelihood of enrolling twins for each randomization approach.

RESULT

Populations included 209 parents and 321 adult multiples. Seventy-eight percent of parents and 59% of multiples prefer same arm placement of multiples over other methods (both P<0.001), which also had highest likelihood of enrollment among both the groups.

CONCLUSION

Parents of multiples and adult multiples prefer placement of multiples into same treatment arm in randomized trials, making such methodology a potential way to optimize consent rates while ethically approaching human subject research.

Accounting for multiple births in randomised trials: a systematic review

OBJECTIVES

Multiple births are an important subgroup to consider in trials aimed at reducing preterm birth or its consequences. Including multiples results in a unique mixture of independent and clustered data, which has implications for the design, analysis and reporting of the trial. We aimed to determine how multiple births were taken into account in the design and analysis of recent trials involving preterm infants, and whether key information relevant to multiple births was reported.

DESIGN

We conducted a systematic review of multicentre randomised trials involving preterm infants published between 2008 and 2013. Information relevant to multiple births was extracted.

RESULTS

Of the 56 trials included in the review, 6 (11%) excluded multiples and 24 (43%) failed to indicate whether multiples were included. Among the 26 trials that reported multiples were included, only one (4%) accounted for clustering in the sample size calculations and eight (31%) took the clustering into account in the analysis of the primary outcome. Of the 20 trials that randomised infants, 12 (60%) failed to report how infants from the same birth were randomised.

CONCLUSIONS

Information on multiple births is often poorly reported in trials involving preterm infants, and clustering due to multiple births is rarely taken into account. Since ignoring clustering could result in inappropriate recommendations for clinical practice, clustering should be taken into account in the design and analysis of future neonatal and perinatal trials including infants from a multiple birth.

Clustering in Northern Territory Perinatal Data for 2003–2005: Implications for Analysis and Interpretation

Clustering in perinatal data can violate assumptions of independence, an important consideration for data analysis. Few published studies report on the extent of repeat births in routinely collected Australian perinatal data and the implications thereof for analysis and interpretation. This paper reports on a case study that examined the extent and implications of clustering in the Northern Territory Midwives Collection (NTMC) for the period 2003–2005. Data were obtained on 7,741 individual mothers giving birth to 8,707 babies in public hospitals during 2003–2005. Clusters of multiple pregnancies and repeat births were identified and the design effects for birth weight of Aboriginal and non-Aboriginal newborns were calculated. Of the mothers, 46.1% were Aboriginal. Of these, 13.2% had repeat singleton births; 0.4% had multiple pregnancies, and 0.3% had both. Of non-Aboriginal mothers, 8.7% had repeat singleton births; 1.2% had multiple pregnancies; and 0.3% had both. The design effect was 1.07 for Aboriginal newborns and 1.04 for non-Aboriginal newborns. The design effects indicate that the correct variance accounting for clustering is 4–7% larger than the incorrect variance ignoring clustering when three consecutive years of NT data are considered and an intracluster correlation coefficient of 0.48 is assumed for birth weight between twin and non-twin siblings. Depending on the outcome of interest, the impact of clustering should be considered in multivariate analysis of perinatal data, especially when such analyses involve more than one year's data, include large proportions of Aboriginal mothers and newborns, and groups with different rates of repeat births.

Using the 7-point checklist as a diagnostic aid for pigmented skin lesions in general practice: a diagnostic validation study

BACKGROUND

GPs need to recognise significant pigmented skin lesions, given rising UK incidence rates for malignant melanoma. The 7-point checklist (7PCL) has been recommended by NICE (2005) for routine use in UK general practice to identify clinically significant lesions which require urgent referral.

AIM

To validate the Original and Weighted versions of the 7PCL in the primary care setting.

DESIGN AND SETTING

Diagnostic validation study, using data from a SIAscopic diagnostic aid randomised controlled trial in eastern England.

METHOD

Adults presenting in general practice with a pigmented skin lesion that could not be immediately diagnosed as benign were recruited into the trial. Reference standard diagnoses were histology or dermatology expert opinion; 7PCL scores were calculated blinded to the reference diagnosis. A case was defined as a clinically significant lesion for primary care referral to secondary care (total 1436 lesions: 225 cases, 1211 controls); or melanoma (36).

RESULTS

For diagnosing clinically significant lesions there was a difference between the performance of the Original and Weighted 7PCLs (respectively, area under curve: 0.66, 0.69, difference = 0.03, P<0.001). For the identification of melanoma, similar differences were found. Increasing the Weighted 7PCL's cut-off score from recommended 3 to 4 improved detection of clinically significant lesions in primary care: sensitivity 73.3%, specificity 57.1%, positive predictive value 24.1%, negative predictive value 92.0%, while maintaining high sensitivity of 91.7% and moderate specificity of 53.4% for melanoma.

CONCLUSION

The Original and Weighted 7PCLs both performed well in a primary care setting to identify clinically significant lesions as well as melanoma. The Weighted 7PCL, with a revised cut-off score of 4 from 3, performs slightly better and could be applied in general practice to support the recognition of clinically significant lesions and therefore the early identification of melanoma.

Respiratory morbidity at follow-up of small-for-gestational-age infants born very prematurely

BACKGROUND

The aim of this study was to determine whether small-for-gestational-age (SGA) infants born very prematurely had increased respiratory morbidity in the neonatal period and at follow-up.

METHODS

Data were examined from infants recruited into the United Kingdom Oscillation Study (UKOS). Of the 797 infants who were born at <29 wk of gestational age, 174 infants were SGA. Overall, 92% were exposed to antenatal corticosteroids and 97% received surfactant; follow-up data at 22-28 mo were available for 367 infants.

RESULTS

After adjustment for gestational age and sex, SGA infants had higher rates of supplementary oxygen dependency at 36 wk postmenstrual age (odds ratio (OR): 3.23; 95% confidence interval: 2.03, 5.13), pulmonary hemorrhage (OR: 3.07; 95% CI: 1.82, 5.18), death (OR: 3.32; 95% CI: 2.13, 5.17), and postnatal corticosteroid requirement (OR: 2.09; 95% CI: 1.35, 3.23). After adjustment for infant and respiratory morbidity risk factors, a lower mean birth weight z-score was associated with a higher prevalence of respiratory admissions (OR: 1.40; 95% CI: 1.03, 1.88 for 1 SD change in z-score), cough (OR: 1.28; 95% CI: 1.00, 1.65), and use of chest medicines (OR: 1.32; 95% CI: 1.01, 1.73).

CONCLUSION

SGA infants who were born very prematurely, despite routine use of antenatal corticosteroids and postnatal surfactant, had increased respiratory morbidity at follow-up, which was not due to poor neonatal outcome.

Factors influencing place of delivery for women in Kenya: an analysis of the Kenya demographic and health survey, 2008/2009

BACKGROUND

Maternal mortality in Kenya increased from 380/100000 live births to 530/100000 live births between 1990 and 2008. Skilled assistance during childbirth is central to reducing maternal mortality yet the proportion of deliveries taking place in health facilities where such assistance can reliably be provided has remained below 50% since the early 1990s. We use the 2008/2009 Kenya Demographic and Health Survey data to describe the factors that determine where women deliver in Kenya and to explore reasons given for home delivery.

METHODS

Data on place of delivery, reasons for home delivery, and a range of potential explanatory factors were collected by interviewer-led questionnaire on 3977 women and augmented with distance from the nearest health facility estimated using health facility Global Positioning System (GPS) co-ordinates. Predictors of whether the woman's most recent delivery was in a health facility were explored in an exploratory risk factor analysis using multiple logistic regression. The main reasons given by the woman for home delivery were also examined.

RESULTS

Living in urban areas, being wealthy, more educated, using antenatal care services optimally and lower parity strongly predicted where women delivered, and so did region, ethnicity, and type of facilities used. Wealth and rural/urban residence were independently related. The effect of distance from a health facility was not significant after controlling for other variables. Women most commonly cited distance and/or lack of transport as reasons for not delivering in a health facility but over 60% gave other reasons including 20.5% who considered health facility delivery unnecessary, 18% who cited abrupt delivery as the main reason and 11% who cited high cost.

CONCLUSION

Physical access to health facilities through distance and/or lack of transport, and economic considerations are important barriers for women to delivering in a health facility in Kenya. Some women do not perceive a need to deliver in a health facility and may value health facility delivery less with subsequent deliveries. Access to appropriate transport for mothers in labour and improving the experiences and outcomes for mothers using health facilities at childbirth augmented by health education may increase uptake of health facility delivery in Kenya.

Neonatal and infant outcome in boys and girls born very prematurely

INTRODUCTION

Although important new strategies have improved outcomes for very preterm infants, males have greater mortality/morbidity than females. We investigated whether the excess of adverse later effects in males operated through poorer neonatal profile or if there was an intrinsic male effect.

RESULTS

Male sex was significantly associated with higher birth weight, death or oxygen dependency (72% vs. 61%, boys vs. girls), hospital stay (97 vs. 86 days), pulmonary hemorrhage (15% vs. 10%), postnatal steroids (37% vs. 21%), and major cranial ultrasound abnormality (20% vs. 12%). Differences remained significant after adjusting for birth weight and gestation. At follow-up, disability, cognitive delay, and use of inhalers remained significant after further adjustment.

DISCUSSION

We conclude that in very preterm infants, male sex is an important risk factor for poor neonatal outcome and poor neurological and respiratory outcome at follow-up. The increased risks at follow-up are not explained by neonatal factors and lend support to the concept of male vulnerability following preterm birth.

METHODS

Data came from the United Kingdom Oscillation Study, with 797 infants (428 boys) born at 23-28 wk gestational age. Thirteen maternal factors, 8 infant factors, 11 acute outcomes, and neurological and respiratory outcomes at follow-up were analyzed. Follow-up outcomes were adjusted for birth and neonatal factors sequentially to explore mechanisms for differences by sex.

Analysis of binary outcomes from randomised trials including multiple births: when should clustering be taken into account?

Randomised trials involving infants from both single and multiple births present unique statistical challenges. A range of methods have been used to analyse such data, including standard methods which treat all infants as independent, and more complex methods which account for the dependence between outcomes of infants from the same pregnancy. Conflicting recommendations have been made regarding if and when this dependence, or clustering, should be taken into account in the analysis. We studied the performance of ordinary logistic regression, which ignores the clustering, compared with logistic generalised estimating equations (GEEs) and mixed effects models (MEMs), which account for the clustering, using real and simulated datasets. Ordinary logistic regression produced appropriate type I error and coverage rates, provided the dependence between outcomes of infants from the same pregnancy was small and the multiple birth rate was low, but performed poorly otherwise. The type I error rate increased and the coverage rate decreased as either the strength of the dependence or the multiple birth rate increased. In contrast, logistic GEEs maintained appropriate type I error and coverage rates across a wide range of settings. The performance of logistic MEMs varied depending on the setting and the estimation procedure used but was often similar to or better than ordinary logistic regression. We recommend using a method which takes the clustering into account when analysing datasets including infants from multiple births.