Prediction of large-for-gestational age infants in relation to hyperglycemia in pregnancy - A comparison of statistical models.

AIMS: Using data from a large multi-centre cohort, we aimed to create a risk prediction model for large-for-gestational age (LGA) infants, using both logistic regression and naïve Bayes approaches, and compare the utility of these two approaches. METHODS: We have compared the two techniques underpinning machine learning: logistic regression (LR) and naïve Bayes (NB) in terms of their ability to predict large-for-gestational age (LGA) infants. Using data from five centres involved in the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study, we developed LR and NB models and compared the predictive ability and stability between the models. Models were developed combining the risks of hyperglycaemia (assessed in three forms: IADPSG GDM yes/no, GDM subtype, OGTT z-score quintiles), demographic and clinical variables as potential predictors. RESULTS: The two approaches resulted in similar estimates of LGA risk (intraclass correlation coefficient 0.955, 95% CI 0.952, 0.958) however the AUROC for the LR model was significantly higher (0.698 vs 0.682; p < 0.001). When comparing the three LR models, use of individual OGTT z-score quintiles resulted in statistically higher AUROCs than the other two models. CONCLUSIONS: Logistic regression can be used with confidence to assess the relationship between clinical and biochemical variables and outcome.

Development of the Birthweight Appropriateness Quotient: A New Measure of Infant's Size.

INTRODUCTION: The customised birthweight model can be used to improve detection of babies that may be at risk of adverse outcomes associated with abnormal growth, however it is currently used in conjunction with either an intrauterine growth standard or the individualised birthweight ratio (IBR), both of which have significant methodological flaws. Our aim was to investigate the statistical validity of the IBR and attempt to develop a new measurement to represent the appropriateness of an infant's size at birth that will support clinicians in identifying infants requiring further attention. METHODS: Routinely collected hospital maternity and neonatal data on singleton, term births from a tertiary Australian hospital were extracted for the time period 1998-2009. The relationships between birthweight, customised birthweight and IBR are investigated using correlation, regression analysis and division of births into groups of < 2500 g, 2500-4000 g and > 4000 g. A new measure, the Birthweight Appropriateness Quotient (BAQ), is developed. The utility of the BAQ is compared with IBR and birthweight to identify infants with a composite neonatal morbidity outcome. RESULTS: Statistical flaws with the IBR due to significant correlation between birthweight and customised birthweight and a heterogenous relationship between these two measurements across the range of birthweight are present. BAQ is uncorrelated with birthweight. Comparison of BAQ and IBR as indicators of adverse neonatal outcome demonstrates that BAQ identifies babies at risk due to their small size and those babies at risk due to inappropriate size. CONCLUSIONS FOR PRACTICE: BAQ is a customised measurement of an infant's size free of the statistical flaws experienced by the IBR with the ability to identify at-risk infants.

A test of agreement of customised birthweight models.

BACKGROUND: The objective of this study was to determine whether the physiological effects on birthweight as described by customised birthweight models (CBMs) from various populations and locations are consistent when applied to a single sample. METHODS: The predicted birthweight was calculated for 52 826 White-European singleton term births between 1997 and 2008 from a large Australian hospital using the same set of variables from 12 published CBMs. The accuracy of prediction was tested against both the actual birthweight and a reference model. Intraclass correlation coefficients (ICCs) along with 95% confidence intervals of the measurements, paired differences (predicted-actual birthweight) and absolute values of the paired differences are reported. RESULTS: The average difference in predicted and actual birthweight was <200 g for all CBMs, with ICCs for all but one model indicating fair agreement (between 0.3 and 0.5). When compared with the reference model, eight of the 11 models had a difference in predicted birthweight of <220 g, and the ICCs indicated that the majority of models had strong agreement. CONCLUSION: All published CBMs demonstrated ability to predict birthweight with reasonable accuracy. The effects of maternal and fetal characteristics on birthweight appear to be consistent across birthweight models. This finding is a further step in validating the CBM, and provides greater evidence for the creation of a global model.

Prepartum and intrapartum caesarean section rates at Mater Mothers' Hospital Brisbane 1997-2005.

AIM: To document the rise in prepartum and intrapartum caesarean section and the demographic and medical factors contributing to the rise. METHODS: Data from 52,423 deliveries between January 1997 to May 2005 were analysed for yearly change in caesarean section rates and multiple demographic and medical factors. RESULTS: The prepartum caesarean section rate increased by 1.6% per year and the intrapartum caesarean section rate by 0.8% per year. There was no increase in the overall prevalence of obesity, short stature, advanced maternal age, medical complications or previous caesarean section. There were significant increases in nulliparity, private care, induction of labour and the use of electronic monitoring, but these were insufficient to explain the magnitude of the rise. CONCLUSION: The increase in prepartum and intrapartum caesarean section displayed was not fully explained by medical and demographic changes in the population.

Path analysis modeling indicates free transport increases ambulance use for minor indications.

INTRODUCTION: Clinically unnecessary ambulance transport is increasing, diverting limited resources from patients needing ambulance transport. It was anecdotally observed that inappropriate ambulance use increased after abolition of a direct patient cost for ambulance transport. HYPOTHESIS: In July 2003, direct patient fees were abolished in favor of a universally applied ambulance levy, potentially leading to increased ambulance use by patients with low illness acuity and admission rates. METHODS: The influence of age, illness acuity, and need for admission on ambulance use was assessed for 55,397 emergency department attendances in 2002 and 2004. Ambulance users were compared with nonusers in both years and attendances for 2002 compared with 2004 using chi-square test for two groups. Logistic regression provided a multivariate model leading to ambulance use. Path analysis modeling to assess interrelationships between factors associated with ambulance use was developed. RESULTS: Ambulance users in both years were older, had more acute illness, and had greater need for admission compared with nonusers. The odds ratio (OR) of arrival by ambulance in 2004 compared with 2002 was 1.14 (95% confidence interval, [CI], 1.12 to 1.17). In 2002, ambulance users were older (OR, 1.42; 95% CI, 1.40 to 1.43), were more likely to need admission (OR, 2.28; 95% CI, 2.16 to 2.4) and had higher illness acuity (OR, 2.02; 95% CI, 1.94 to 2.09). There was a negative correlation between 2004 and illness acuity. CONCLUSIONS: Ambulance use increased in 2004 after patient transport fees were abolished. Increased use was associated with decreased age, clinical acuity, and admission need. Abolishing direct patient cost stimulates ambulance use, potentially including inappropriate transport. Path analysis to assess the effect of changed funding on ambulance use could be used to the influence of other locally relevant factors contributing to ambulance use.

Customised birthweight: coefficients for an Australian population and validation of the model.

BACKGROUND: Published birthweight references in Australia do not fully take into account constitutional factors that influence birthweight and therefore may not provide an accurate reference to identify the infant with abnormal growth. Furthermore, studies in other regions that have derived adjusted (customised) birthweight references have applied untested assumptions in the statistical modelling. AIMS: To validate the customised birthweight model and to produce a reference set of coefficients for estimating a customised birthweight that may be useful for maternity care in Australia and for future research. METHODS: De-identified data were extracted from the clinical database for all births at the Mater Mother's Hospital, Brisbane, Australia, between January 1997 and June 2005. Births with missing data for the variables under study were excluded. In addition the following were excluded: multiple pregnancies, births less than 37 completed week's gestation, stillbirths, and major congenital abnormalities. Multivariate analysis was undertaken. A double cross-validation procedure was used to validate the model. RESULTS: The study of 42,206 births demonstrated that, for statistical purposes, birthweight is normally distributed. Coefficients for the derivation of customised birthweight in an Australian population were developed and the statistical model is demonstrably robust. CONCLUSIONS: This study provides empirical data as to the robustness of the model to determine customised birthweight. Further research is required to define where normal physiology ends and pathology begins, and which segments of the population should be included in the construction of a customised birthweight standard.

Four-year follow-up of women who were diagnosed to have postpartum urinary retention.

OBJECTIVE: The purpose of this study was to evaluate the long-term prevalence of urinary incontinence in women with postpartum urinary retention. STUDY DESIGN: A telephone interview was conducted by contacting a cohort of 691 women who delivered vaginally 4 years ago, of which 101 women had been diagnosed as having postpartum urinary retention. A structured telephone interview consisted of 9 questions on the possible outcomes of postpartum urinary retention. RESULTS: Of the original cohort of 691 women, 394 women were contacted. Seventy-three women had had postpartum urinary retention, and 321 women had not. In women who had had postpartum urinary retention, the prevalence of the outcome variables were urinary stress incontinence (28.8%), fecal incontinence (2.7%), frequency (39.1%), nocturia (65.2%), urgency (26.1%), urge incontinence (26.1%), and coital incontinence (13%). Analyses showed that there was no significant difference between women with and without urinary retention. CONCLUSION: Women who had had postpartum urinary retention did not have a higher prevalence of urinary stress incontinence.