Immortal Time Bias in the Study of Stillbirth Risk Factors

Hutcheon, Jennifer A.; Kuret, Verena; Joseph, K. S.; Sabr, Yasser; Lim, Kenneth

doi:10.1097/ede.0b013e3182a6d9aa

Cited by 33 publications

(26 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most recent studies have indicated the absence of increased risk [5,8], but these studies did not take immortality bias into account. In the study by Hutcheon et al [14], the RR for stillbirth among women with GDM was 1.25 in analyses limited to births after 28 weeks. In our study, the risk of perinatal mortality was not increased among deliveries after 28 weeks: surprisingly, it was increased only among deliveries after 37 weeks, whether or not women were insulin-treated.…”

Section: Discussionmentioning

confidence: 99%

“…Immortal time refers to the follow-up time during which, because of the exposure definition, the outcome under study could not occur, corresponding to the time prior to the diagnosis of GDM [13]. As highlighted by Hutcheon et al, differences at the start of follow-up between women with and without GDM can lead to overestimation of the perinatal mortality rate in the population without diabetes [14]. This bias can also affect risk estimation for other outcomes that may occur before the start of the GDM screening period.…”

Section: Outcomes For Neonatesmentioning

confidence: 99%

See 1 more Smart Citation

Gestational diabetes and adverse perinatal outcomes from 716,152 births in France in 2012

Billionnet¹,

Mitanchez

Weill³

et al. 2017

Diabetologia

416

282

View full text Add to dashboard Cite

Aims/hypothesis The aim of this study was to assess the risk of adverse perinatal outcomes in gestational diabetes mellitus (GDM) in a large national cohort. Methods All deliveries taking place after 22 weeks in France in 2012 were included by extracting data from the hospital discharge database and the national health insurance system. The diabetic status of mothers was determined by the use of glucose-lowering agents and by hospital diagnosis. Outcomes were analysed according to the type of diabetes and, in the GDM group, whether or not diabetes was insulin-treated. Results The cohort of 796,346 deliveries involved 57,629 (7.24%) mothers with GDM. Mother-infant linkage was obtained for 705,198 deliveries. The risks of adverse outcomes were much lower with GDM than with pregestational diabetes. After limiting the analysis to deliveries after 28 weeks to reduce immortal time bias, the risks of preterm birth (OR 1. were increased in women with GDM compared with the non-diabetic population. Higher risks were observed in women with insulin-treated GDM than those with diet-treated GDM. After limiting the analysis to term deliveries, an increased risk of perinatal mortality was observed. After excluding women suspected to have undiagnosed pregestational diabetes, the risk remained moderately increased only for those with diet-treated GDM (OR 1.3 [95% CI 1.0, 1.6]). Conclusions/interpretation GDM is associated with a moderately increased risk of adverse perinatal outcomes, which is higher in insulin-treated GDM than in non-insulintreated GDM for most outcomes.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Outcomes For Neonatesmentioning

confidence: 99%

Gestational diabetes and adverse perinatal outcomes from 716,152 births in France in 2012

Billionnet¹,

Mitanchez

Weill³

et al. 2017

Diabetologia

416

282

View full text Add to dashboard Cite

show abstract

“…Similarly, this study found increased risk of stillbirth due to gestational diabetes from 28 weeks onwards, in concordance with Hutcheon and colleagues [233]. It has been argued that the 'protective' effect observed in relation to gestational diabetes at earlier gestational ages may be due to routine screening for GDM around 24-28 weeks [233]. There is evidence of higher rates of increase in gestational diabetes among non-Indigenous women, despite currently lower absolute rates of GDM compared with Indigenous women [87].…”

Section: Interpretation Of Findings Diabetessupporting

confidence: 80%

“…Furthermore, there was disparity in the population attributable fractions associated with pre-existing diabetes among Indigenous and non-Indigenous women (7.4% vs 1.6%); however, these were similar to those reported for other high income countries (3-5%) [32]. Similarly, this study found increased risk of stillbirth due to gestational diabetes from 28 weeks onwards, in concordance with Hutcheon and colleagues [233]. It has been argued that the 'protective' effect observed in relation to gestational diabetes at earlier gestational ages may be due to routine screening for GDM around 24-28 weeks [233].…”

Section: Interpretation Of Findings Diabetessupporting

confidence: 80%

Preventing stillbirths: Better data to inform interventions in Australia

Ibiebele¹

View full text Add to dashboard Cite

BackgroundIn Australia, one in every 139 women reaching 20 weeks gestation will have a stillborn baby. There is significant disparity in stillbirth rates within subgroups of the Australian population. Among Aboriginal and Torres Strait Islander (Indigenous) women, the stillbirth rate is one in every 93 women reaching 20 weeks gestation. Unfortunately, there has been no reduction in the national stillbirth rates over the past two decades and nearly a third are 'unexplained'. The ability to identify women at increased risk of stillbirth is an important and challenging priority. At present in Australia, there is a paucity of high quality data on causes and contributing factors to stillbirth. Furthermore, inconsistent approaches to investigation and classification affect the quality of data on causes of death and hamper the development of effective interventions to prevent stillbirth. AimsThe primary aim of this Thesis is to describe the epidemiology of stillbirth within the Australian context. This aim was addressed by: Examining trends in stillbirth by clinical classification of cause of death, Indigenous status and gestational age, to identify focal areas for preventive efforts  Assessing gestational age specific risk of stillbirth associated with four important contributors (diabetes, hypertension, antepartum haemorrhage and small-forgestational age) to higher stillbirth rates among Indigenous women in order to identify periods of increased risk  Developing and validating a statistical model to predict the risk of antepartum stillbirth at term (≥37 weeks) using maternal and pregnancy factors as a potential decision-making aid for clinicians and women  Assessing consistency in application of the Perinatal Society of Australia and New Zealand Perinatal Death Classification system between hospital committees and an independent expert panel, to identify areas for quality improvement  Determining maternal and pregnancy factors associated with parental consent to autopsy following stillbirth and explore parents' views and experiences of the autopsy consent process to inform clinical practice iii Methods and Results Data for 1995-2011 (n=881,211 singleton births) from the Queensland Perinatal DataCollection was analysed to address the first three aims. Stillbirth trends analysis found consistently higher rates of stillbirth among Indigenous women, however, the gap in stillbirth rates had narrowed. Gestational age specific stillbirth risk analysis found disparity in the magnitude of stillbirth risk for pre-existing diabetes and small-for-gestational age between Indigenous and non-Indigenous women. Despite strong association between maternal clinical factors and antepartum stillbirth risk, the prediction model had a poor ability to predict stillbirth risk at term.Consistency in application of the perinatal death classification system was assessed by calculating agreement. A substantial level of agreement was found between hospital committee and expert panel review of a cohort of 217 stillbirth cases, howeve...

show abstract

“…One problem with the example (acknowledged by the authors as a limitation) relates to a temporal issue involving the determinant and the outcome. Most women who develop preeclampsia do so after 37 weeks' gestation and therefore an immortal time bias plagues the preeclampsia‐preterm birth association . Unfortunately, as Bandoli et al mention, absence of information on gestational age at which the diagnosis of preeclampsia occurred precluded restriction of preeclampsia cases to those with a diagnosis prior to 37 weeks.…”

Section: Critique Of the Preeclampsia‐preterm Birth Examplementioning

confidence: 99%