We propose a structural framework for population-based cancer epidemiology and evaluate the performance of double-robust estimators for a binary exposure in cancer mortality. We performed numerical analyses to study the bias and efficiency of these estimators. Furthermore, we compared two different model selection strategies based on 1) the Akaike and Bayesian Information Criteria and 2) machine-learning algorithms, and illustrated double-robust estimators’ performance in a real setting. In simulations with correctly specified models and near-positivity violations, all but the naïve estimators presented relatively good performance. However, the augmented inverse-probability treatment weighting estimator showed the largest relative bias. Under dual model misspecification and near-positivity violations, all double-robust estimators were biased. Nevertheless, the targeted maximum likelihood estimator showed the best bias-variance trade-off, more precise estimates, and appropriate 95% confidence interval coverage, supporting the use of the data-adaptive model selection strategies based on machine-learning algorithms. We applied these methods to estimate adjusted one-year mortality risk differences in 183,426 lung cancer patients diagnosed after admittance to an emergency department versus non-emergency cancer diagnosis in England, 2006–2013. The adjusted mortality risk (for patients diagnosed with lung cancer after admittance to an emergency department) was 16% higher in men and 18% higher in women, suggesting the importance of interventions targeting early detection of lung cancer signs and symptoms.
BACKGROUND: Variation in colon cancer mortality occurring shortly after diagnosis is widely reported between socio-economic status (SES) groups: we investigated the role of different prognostic factors in explaining variation in 90-day mortality. METHODS: National cancer registry data were linked with national clinical audit data and Hospital Episode Statistics records for 69 769 adults diagnosed with colon cancer in England between January 2010 and March 2013. By gender, logistic regression was used to estimate the effects of SES, age and stage at diagnosis, comorbidity and surgical treatment on probability of death within 90 days from diagnosis. Multiple imputations accounted for missing stage. We predicted conditional probabilities by prognostic factor patterns and estimated the effect of SES (deprivation) from the difference between deprivation-specific average predicted probabilities. RESULTS: Ninety-day probability of death rose with increasing deprivation, even after accounting for the main prognostic factors. When setting the deprivation level to the least deprived group for all patients and keeping all other prognostic factors as observed, the differences between deprivation-specific averaged predicted probabilities of death were greatly reduced but persisted. Additional analysis suggested stage and treatment as potential contributors towards some of these inequalities. CONCLUSIONS: Further examination of delayed diagnosis, access to treatment and post-operative care by deprivation group may provide additional insights into understanding deprivation disparities in mortality.British Journal of Cancer advance online publication, 31 August 2017; doi:10.1038/bjc.2017.295 www.bjcancer.com.
Background Patients with comorbidities do not receive optimal treatment for their cancer, leading to lower cancer survival. Information on individual comorbidities is not straightforward to derive from population-based administrative health datasets. We described the development of a reproducible algorithm to extract the individual Charlson index comorbidities from such data. We illustrated the algorithm with 1,789 laryngeal cancer patients diagnosed in England in 2013. We aimed to clearly set out and advocate the time-related assumptions specified in the algorithm by providing empirical evidence for them. Methods Comorbidities were assessed from hospital records in the ten years preceding cancer diagnosis and internal reliability of the hospital records was checked. Data were right-truncated 6 or 12 months prior to cancer diagnosis to avoid inclusion of potentially cancer-related comorbidities. We tested for collider bias using Cox regression. Results Our administrative data showed weak to moderate internal reliability to identify comorbidities (ICC ranging between 0.1 and 0.6) but a notably high external validity (86.3%). We showed a reverse protective effect of non-cancer related Chronic Obstructive Pulmonary Disease (COPD) when the effect is split into cancer and non-cancer related COPD (Age-adjusted HR: 0.95, 95% CI:0.7–1.28 for non-cancer related comorbidities). Furthermore, we showed that a window of 6 years before diagnosis is an optimal period for the assessment of comorbidities. Conclusion To formulate a robust approach for assessing common comorbidities, it is important that assumptions made are explicitly stated and empirically proven. We provide a transparent and consistent approach useful to researchers looking to assess comorbidities for cancer patients using administrative health data.
To evaluate the independent and combined associations of maternal self-reported poor sleep quality and antepartum depression with suicidal ideation during the third trimester METHODS: A cross-sectional study was conducted among 1298 pregnant women (between 24 and 28 gestational weeks) attending prenatal clinics in Lima, Peru. Antepartum depression and suicidal ideation were assessed using the Patient Health Questionnaire-9 (PHQ-9). The Pittsburgh Sleep Quality Index (PSQI) questionnaire was used to assess sleep quality. Multivariate logistical regression procedures were used to estimate odds ratios (OR) and 95% confidence intervals (95% CI) after adjusting for putative confounders.
While obesity is an indicated risk factor for hypertensive disorders of pregnancy, smoking during pregnancy has been shown to be inversely associated with the development of preeclampsia and gestational hypertension. The purpose of this study was to investigate the combined effects of high body mass index and smoking on hypertensive disorders during pregnancy. This was a case-control study based on national registers, nested within all pregnancies in Iceland 1989-2004, resulting in birth at the Landspitali University Hospital. Cases (n = 500) were matched 1:2 with women without a hypertensive diagnosis who gave birth in the same year. Body mass index (kg/m2) was based on height and weight at 10-15 weeks of pregnancy. We used logistic regression models to calculate odds ratios and corresponding 95% confidence intervals as measures of association, adjusting for potential confounders and tested for additive and multiplicative interactions of body mass index and smoking. Women's body mass index during early pregnancy was positively associated with each hypertensive outcome. Compared with normal weight women, the multivariable adjusted odds ratio for any hypertensive disorder was 1.8 (95% confidence interval, 1.3-2.3) for overweight women and 3.1 (95% confidence interval, 2.2-4.3) for obese women. The odds ratio for any hypertensive disorder with obesity was 3.9 (95% confidence interval 1.8-8.6) among smokers and 3.0 (95% confidence interval 2.1-4.3) among non-smokers. The effect estimates for hypertensive disorders with high body mass index appeared more pronounced among smokers than non-smokers, although the observed difference was not statistically significant. Our findings may help elucidate the complicated interplay of these lifestyle-related factors with the hypertensive disorders during pregnancy.
BACKGROUND: There is a scarcity of data on the association of sexual violence and women's subsequent obstetric outcomes. Our aim was to investigate whether women exposed to sexual violence as teenagers (12-19 years of age) or adults present with different obstetric outcomes than women with no record of such violence.
METHODS: We linked detailed prospectively collected information on women attending a Rape Trauma Service (RTS) to the Icelandic Medical Birth Registry (IBR). Women who attended the RTS in 1993-2010 and delivered (on average 5.8 years later) at least one singleton infant in Iceland through 2012 formed our exposed cohort (n = 1068). For each exposed woman's delivery, nine deliveries by women with no RTS attendance were randomly selected from the IBR (n = 9126) matched on age, parity, and year and season of delivery. Information on smoking and Body mass index (BMI) was available for a sub-sample (n = 792 exposed and n = 1416 non-exposed women). Poisson regression models were used to estimate Relative Risks (RR) with 95% confidence intervals (CI).
RESULTS: Compared with non-exposed women, exposed women presented with increased risks of maternal distress during labor and delivery (RR 1.68, 95% CI 1.01-2.79), prolonged first stage of labor (RR 1.40, 95% CI 1.03-1.88), antepartum bleeding (RR 1.95, 95% CI 1.22-3.07) and emergency instrumental delivery (RR 1.16, 95% CI 1.00-1.34). Slightly higher risks were seen for women assaulted as teenagers. Overall, we did not observe differences between the groups regarding the risk of elective cesarean section (RR 0.86, 95% CI 0.61-1.21), except for a reduced risk among those assaulted as teenagers (RR 0.56, 95% CI 0.34-0.93). Adjusting for maternal smoking and BMI in a sub-sample did not substantially affect point estimates.
CONCLUSION: Our prospective data suggest that women with a history of sexual assault, particularly as teenagers, are at increased risks of some adverse obstetric outcomes.
Although smoking during pregnancy may lead to many adverse outcomes, numerous studies have reported a paradoxical inverse association between maternal cigarette smoking during pregnancy and preeclampsia. Using a counterfactual framework we aimed to explore the structure of this paradox as being a consequence of selection bias. Using a case-control study nested in the Icelandic Birth Registry (1309 women), we show how this selection bias can be explored and corrected for. Cases were defined as any case of pregnancy induced hypertension or preeclampsia occurring after 20 weeks' gestation and controls as normotensive mothers who gave birth in the same year. First, we used directed acyclic graphs to illustrate the common bias structure. Second, we used classical logistic regression and mediation analytic methods for dichotomous outcomes to explore the structure of the bias. Lastly, we performed both deterministic and probabilistic sensitivity analysis to estimate the amount of bias due to an uncontrolled confounder and corrected for it. The biased effect of smoking was estimated to reduce the odds of preeclampsia by 28 % (OR 0.72, 95 %CI 0.52, 0.99) and after stratification by gestational age at delivery (<37 vs. ≥37 gestation weeks) by 75 % (OR 0.25, 95 %CI 0.10, 0.68). In a mediation analysis, the natural indirect effect showed and OR > 1, revealing the structure of the paradox. The bias-adjusted estimation of the smoking effect on preeclampsia showed an OR of 1.22 (95 %CI 0.41, 6.53). The smoking-preeclampsia paradox appears to be an example of (1) selection bias most likely caused by studying cases prevalent at birth rather than all incident cases from conception in a pregnancy cohort, (2) omitting important confounders associated with both smoking and preeclampsia (preventing the outcome to develop) and (3) controlling for a collider (gestation weeks at delivery). Future studies need to consider these aspects when studying and interpreting the association between smoking and pregnancy outcomes.
AbstractIntroduction The circadian clock plays an important role in several aspects of female reproductive biology. Evidence linking circadian clock-related genes to pregnancy outcomes has been inconsistent. We sought to examine whether variations in single nucleotide polymorphisms (SNPs) of circadian clock genes are associated with \PA\ risk. Methods Maternal blood samples were collected from 470 \PA\ case and 473 controls. Genotyping was performed using the Illumina Cardio-MetaboChip platform. We examined 119 \SNPs\ in 13 candidate genes known to control circadian rhythms (e.g., CRY2, ARNTL, and RORA). Univariate and penalized logistic regression models were fit to estimate odds ratios (ORs); and the combined effect of multiple \SNPs\ on \PA\ risk was estimated using a weighted genetic risk score (wGRS). Results A common \SNP\ in the \RORA\ gene (rs2899663) was associated with a 21% reduced odds of \PA\ (P < 0.05). The odds of \PA\ increased with increasing wGRS (Ptrend < 0.001). The corresponding \ORs\ were 1.00, 1.83, 2.81 and 5.13 across wGRS quartiles. Participants in the highest wGRS quartile had a 5.13-fold (95% confidence interval: 3.21–8.21) higher odds of \PA\ compared to those in the lowest quartile. Although the test for interaction was not significant, the odds of \PA\ was substantially elevated for preeclamptics with the highest wGRS quartile (OR = 14.44, 95%CI: 6.62–31.53) compared to normotensive women in the lowest wGRS quartile. Discussion Genetic variants in circadian rhythm genes may be associated with \PA\ risk. Larger studies are needed to corroborate these findings and to further elucidate the pathogenesis of this important obstetrical complication.
Obstructive sleep apnea (OSA), a common and serious disorder in which breathing repeatedly stops during sleep, is associated with excess weight and obesity. Little is known about the co-occurrence of OSA among pregnant women from low and middle-income countries.
BACKGROUND: Although rare, placental abruption is implicated in disproportionately high rates of perinatal morbidity and mortality. Understanding geographic and temporal variations may provide insights into possible amenable factors of abruption. We examined abruption frequencies by maternal age, delivery year, and maternal birth cohorts over three decades across seven countries.
METHODS: Women that delivered in the US (n = 863,879; 1979-10), Canada (4 provinces, n = 5,407,463; 1982-11), Sweden (n = 3,266,742; 1978-10), Denmark (n = 1,773,895; 1978-08), Norway (n = 1,780,271, 1978-09), Finland (n = 1,411,867; 1987-10), and Spain (n = 6,151,508; 1999-12) were analyzed. Abruption diagnosis was based on ICD coding. Rates were modeled using Poisson regression within the framework of an age-period-cohort analysis, and multi-level models to examine the contribution of smoking in four countries.
RESULTS: Abruption rates varied across the seven countries (3-10 per 1000), Maternal age showed a consistent J-shaped pattern with increased rates at the extremes of the age distribution. In comparison to births in 2000, births after 2000 in European countries had lower abruption rates; in the US there was an increase in rate up to 2000 and a plateau thereafter. No birth cohort effects were evident. Changes in smoking prevalence partially explained the period effect in the US (P = 0.01) and Sweden (P<0.01).
CONCLUSIONS: There is a strong maternal age effect on abruption. While the abruption rate has plateaued since 2000 in the US, all other countries show declining rates. These findings suggest considerable variation in abruption frequencies across countries; differences in the distribution of risk factors, especially smoking, may help guide policy to reduce abruption rates.
Background Migraine is associated with a number of cardiometabolic risk factors including abnormalities in lipid metabolism. However, little is known about these associations among pregnant migraineurs. We conducted the present study to evaluate the extent to which altered lipid profiles are associated with history of migraine among pregnant women. Methods A cohort of 1062 Peruvian women were interviewed at 24-28 weeks of gestation. Migraine status was classified based on the International Classification of Headache Disorders-II diagnostic criteria. Serum lipid concentrations were measured enzymatically using standardized assays. Multivariable logistic regression was used to estimate adjusted odds ratios (AORs) and 95% confidence intervals (CIs) as measures of associations of migraine status with varying concentrations of lipids and lipoproteins during pregnancy. Results Approximately 18.5% of the study participants were identified as migraineurs (196 of 1062). Maternal serum total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides, and total cholesterol : HDL ratio were all statistically significantly elevated among pregnant migraineurs compared with pregnant non-migraineurs. In multivariate adjusted models, pregnant women with migraine had higher odds of elevated total cholesterol, LDL, and total cholesterol : HDL ratio as compared with pregnant women without migraine. For instance, the AOR and 95% CI for successive quartiles of the total cholesterol associated with history of migraine were Q2 (219-247 mg/dL): 1.05 (0.64-1.70), Q3 (248-281 mg/dL): 1.16 (0.72-1.86), and Q4 (≥282 mg/dL): 1.87 (1.20-2.91) with the lowest quartile (<219 mg/dL) as the referent group (P value for trend = .003). Obese women with elevated total cholesterol (≥282 mg/dL) were more likely to be migraineurs (OR = 3.71; 95% CI 1.58-8.71) as compared with non-obese women with lower total cholesterol (<219 mg/dL). Similar elevated odds of migraine were observed for obese women with elevated LDL cholesterol, elevated triglycerides and high total cholesterol : HDL ratio. Conclusion Pregnant migraineurs had elevated odds of dyslipidemia, particularly hypercholesterolemia, elevated LDL, and total cholesterol : HDL ratio as compared with pregnant non-migraineurs. The observed associations were more pronounced among obese migraineurs. Our findings add to the accumulating evidence of adverse cardiometabolic risk profiles among migraineurs and extend these associations to pregnant women.
Higher placental weight relative to birthweight has been described as an adaptive mechanism to fetal hypoxia in small for gestational age (SGA) infants. However, placental weight alone may not be a good marker reflecting intrauterine growth restriction. We hypothesized that fetoplacental ratio (FPR)-the ratio between birthweight and placental weight-may serve as a good marker of SGA after adjustment for surrogates of fetal hypoxemia (maternal iron deficiency anemia, smoking and choriodecidual necrosis). We conducted a within-sibling analysis using data from the US National Collaborative Perinatal Project (1959-1966) of 1,803 women who delivered their first two (or more) consecutive infants at term (n = 3,494). We used variance-component fixed-effect linear regression models to explore the effect of observed time-varying factors on placental weight and conditional logistic regression to estimate the effects of the tertiles of FPRs (1st small, 2nd normal and 3rd large) on the odds of SGA infants. We found placental weights to be 15 g [95 % confidence interval (CI) 8, 23] higher and -7 g (95 % CI -13, -2) lower among women that had anemia and choriodecidual necrosis, respectively. After multivariable adjustment, newborns with a small FPR (1st-tertile ≤7) had twofold higher odds of being SGA (OR 2.0, 95 % CI 1.2, 3.5) than their siblings with a large FPR (3nd-tertile ≥9). A small FPR was associated with higher odds of SGA, suggesting that small FPR may serve as an indicator suggestive of adverse intrauterine environment. This observation may help to distinguish pathological from constitutional SGA.
The tuberculosis surveillance system in the Balearic Islands was assessed from 2005 to 2007. Applying the capture-recapture method the completeness of this system was evaluated to be 58.4%. When a new electronic recorded data was included in Primary Health Care, up to 66.5% was obtained. This new source of data increased the detected cases of pulmonary tuberculosis from 572 to 681. As a result, the estimated annual incidence rate increases from 18.9 cases/10(5) to 22.6 cases/10(5) [95% CI, 20.9-24.3], similar to figures issued by WHO.
PURPOSE: Data regarding circadian rhythm in the onset of spontaneous preterm premature rupture of membranes (PROM) and placental abruption (PA) cases are conflicting. We modeled the time of onset of preterm PROM and PA cases and examined if the circadian profiles varied based on the gestational age at delivery.
METHODS: We used parametric and nonparametric methods, including trigonometric regression in the framework of generalized linear models, to test the presence of circadian rhythms in the time of onset of preterm PROM and PA cases among 395 women who delivered a singleton between 2009 and 2010 in Lima, Peru.
RESULTS: We found a diurnal circadian pattern, with a morning peak at 07:32 AM (95% confidence interval, 05:46 AM-09:18 AM) among moderate preterm PROM cases (P value < .001), and some evidence of a diurnal circadian periodicity among PA cases in term infants (P value = .067). However, we did not find evidence of circadian rhythms in the time of onset of extremely or very preterm PROM (P value = .259) and preterm PA (P value = .224).
CONCLUSIONS: The circadian rhythms of the time of onset of preterm PROM and PA cases varied based on gestational weeks at delivery. Although circadian rhythms were presented among moderate preterm PROM and term PA cases, there was no evidence of circadian rhythms among preterm PA and very or extremely preterm PROM cases, underlying other mechanisms associated with the time of onset.
Aims Alterations in organic acid biomarkers from fatty acid and carbohydrate metabolism have been documented in type 2 diabetes patients. However, their association with gestational diabetes mellitus (GDM) is largely unknown. Methods Participants were 25 GDM cases and 25 non-GDM controls. Biomarkers of fatty acid (adipate, suberate and ethylmalonate) and carbohydrate (pyruvate, L-lactate and β-hydroxybutyrate) metabolism were measured in maternal urine samples collected in early pregnancy (17 weeks) using liquid chromatography-mass spectrometry methods. Logistic regression were used to calculate odds ratios (OR) and 95% confidence intervals (CI). Results GDM cases and controls differed in median urinary concentrations of ethylmalonate (3.0 vs. 2.3 μg/mg creatinine), pyruvate (7.4 vs. 2.1 μg/mg creatinine), and adipate (4.6 vs. 7.3 μg/mg creatinine) (all p-values <0.05). Women in the highest tertile for ethylmalonate or pyruvate concentrations had 11.4-fold (95%CI 1.10-117.48) and 3.27-fold (95%CI 0.72-14.79) increased risk of GDM compared with women in the lowest tertile for ethylmalonate and pyruvate concentrations, respectively. Women in the highest tertile for adipate concentrations, compared with women in the lowest tertile, had an 86% reduction in GDM risk (95%CI 0.02-0.97). Conclusions These preliminary findings underscore the importance of altered fatty acid and carbohydrate metabolism in the pathogenesis of GDM.
Background: Vitamin D deficiency during pregnancy has been associated with increased risk of complications and adverse perinatal outcomes. We evaluated seasonal variation of 25-hydroxyvitamin D [25(OH)D] among pregnant women, focusing on patterns and determinants of variation. Methods: Data came from three cohort studies in the US that included 2583 non-Hispanic Black and White women having prenatal 25(OH)D concentrations determined. Fourier time series and generalised linear models were used to estimate the magnitude of 25(OH)D seasonality. We modelled seasonal variability using a stationary cosinor model to estimate the phase shift, peak–trough difference, and annual mean of 25(OH)D. Results: We observed a peak for 25(OH)D in summer, a nadir in winter, and a phase of 8 months, which resulted from fluctuations in 25(OH)D3 rather than 25(OH)D2. After adjustment for covariates, the annual mean concentrations and estimated peak–trough difference of 25(OH)D among Black women were 19.8 ng/mL [95% confidence interval (CI) 18.9, 20.5] and 5.8 ng/mL [95% CI 4.7, 6.7], and for non-Hispanic White women were 33.0 ng/mL [95% CI 32.6, 33.4] and 7.4 ng/mL [95% CI 6.0, 8.9]. Conclusions: Non-Hispanic Black women had lower average 25(OH)D concentrations throughout the year and smaller seasonal variation levels than non-Hispanic White women. This study’s confirmation of 25(OH)D seasonality over a calendar year has the potential to enhance public health interventions targeted to improve maternal and perinatal outcomes.
We describe stillbirth and unemployment rates by autonomous region in Spain and analyse whether women who gave birth in regions with high unemployment rates were more likely to have a stillborn. We designed a multilevel population-based observational study of births from 2007 to 2010. We defined stillbirth as the outcome, individual maternal socioeconomic and pregnancy-related characteristics as covariates, and maternal autonomous region of residence as the contextual covariate. We used mixed-logistic regression models to account for differences across regions. In total, 1,920,235 singleton births and 5,560 stillbirths were included in the study. Women residing in autonomous regions with the highest rates of unemployment had a two-times-greater chance of delivering a stillborn (adjusted OR 2.60; 95 % CI 2.08-3.21). The region where women resided explained 14 % of the total individual differences in the risk of delivering a stillborn. The odds of stillbirth were 1.82 (95 % CI 1.62-2.05) times higher for African-born women than for Spanish-born women and 1.90 (95 % CI 1.68-2.15) times higher for women with low educational attainment than for women with higher education. In conclusion, regional disparities in stillbirth rates in Spain in the period 2007-2010 were mainly associated with mothers who had low levels of education, were African-born, and lived in regions with higher unemployment.