Understanding the impact of non-shared unmeasured confounding on the sibling comparison analysis

BACKGROUND

The sibling comparison analysis is used to deal with unmeasured confounding. It has previously been shown that in the presence of non-shared unmeasured confounding, the sibling comparison analysis may introduce substantial bias depending on the sharedness of the unmeasured confounder and the sharedness of the exposure. We aimed to improve the awareness of this challenge of the sibling comparison analysis.

METHODS

First, we simulated sibling pairs with an exposure, a confounder and an outcome. We simulated sibling pairs with no effect of the exposure on the outcome and with positive confounding. For varying degrees of sharedness of the confounder and the exposure and for varying prevalence of the exposure, we calculated the sibling comparison odds ratio (OR). Second, we provided measures for sharedness of selected treatments based on Danish health data.

RESULTS

The confounded sibling comparison OR was visualized for varying degrees of sharedness of the confounder and the exposure and for varying prevalence of the exposure. The confounded sibling comparison OR was seen to increase with increasing sharedness of the exposure and the confounded sibling comparison OR decreased with an increasing prevalence of exposure. Measures for sharedness of treatments based on Danish health data showed that treatments of chronic diseases have the highest sharedness and treatments of non-chronic diseases have the lowest sharedness.

CONCLUSIONS

Researchers should be aware of the challenge regarding non-shared unmeasured confounding in the sibling comparison analysis, before applying the analysis in non-randomized studies. Otherwise, the sibling comparison analysis may lead to substantial bias.

High-intensity versus moderate-intensity statin treatment for patients with ischemic stroke: Nationwide cohort study

PURPOSE

Guidelines recommend high-intensity statin treatment after ischemic stroke, but evidence is sparse on the effectiveness and safety of different statin treatment intensities. We examined effectiveness and safety outcomes among patients initiating high-intensity versus moderate-intensity statins after ischemic stroke.

METHODS

In this population-based new-user active-comparator cohort study, we used the Danish Stroke Registry, covering all Danish hospitals, to identify patients with a first-time ischemic stroke during 2012-2021. Using multiple Danish registries, patients who redeemed a statin prescription within 21 days after stroke admission were classified as high-intensity statin initiators or moderate-intensity statin initiators. Propensity score inverse probability of treatment weighting was used to balance patient characteristics. We used competing risk methods to compute 5 year risk differences (RDs) and Cox proportional hazards regression to compute 5 year hazard ratios (HRs) of stroke recurrence, myocardial infarction, heart failure, venous thromboembolism, and all-cause mortality (effectiveness outcomes) and diabetes, liver disease, and kidney disease (safety outcomes).

RESULTS

High-intensity (n = 13,032) and moderate-intensity (n = 14,355) statin initiators were identified. Risks of most examined effectiveness outcomes were comparable between statin intensities. There was no clear association between statin intensity and stroke recurrence (RD: 0.8% [95% CI: 0.1, 1.4], HR: 1.08 [95% CI: 0.96, 1.22]). All-cause mortality was slightly reduced among high-intensity statin initiators (RD: -1.1% [95% CI: -0.1, -2.1], HR: 0.93 [95% CI: 0.85, 1.01]. Risks of most safety outcomes were comparable between statin intensities, but high-intensity statin use was associated with an increased risk of diabetes (RD: 1.2% [95% CI: 0.4, 1.9], HR: 1.10 [95% CI: 1.00, 1.21]).

DISCUSSION AND CONCLUSION

Compared with initiation of moderate-intensity statins, initiation of high-intensity statins after ischemic stroke was associated with similar risks of most effectiveness and safety outcomes. However, mortality risk was reduced, and diabetes risk was increased.

Prenatal Antidepressant Exposure and the Risk of Attention-Deficit/Hyperactivity Disorder in Childhood: Accounting for Misclassification of Exposure

BACKGROUND

Many studies of prenatal antidepressant exposure and the risk of attention-deficit/hyperactivity disorder (ADHD) have done little to reduce bias from exposure misclassification. We assessed the prenatal antidepressant-ADHD effect by incorporating information on repeatedly redeemed prescriptions and redemptions of drug classes commonly used in pregnancy in the analyses to reduce bias from exposure misclassification.

METHODS

Using population-based registries, we conducted a nationwide cohort study of all children born in Denmark from 1997 to 2017. In a former-user analysis, we compared children prenatally exposed, defined by a redeemed prescription by the mother during pregnancy, to a comparison cohort consisting of prenatally unexposed children whose mothers had redeemed a prescription before pregnancy. We incorporated information on repeatedly redeemed prescriptions and redemptions of drug classes commonly used in pregnancy in the analyses to reduce bias from exposure misclassification. We used incidence rate ratios (IRRs) and incidence rate differences (IRDs) as effect measures.

RESULTS

The cohort included 1,253,362 children, among whom 24,937 were prenatally exposed to antidepressants. The comparison cohort consisted of 25,698 children. During follow-up, 1,183 exposed children and 1,291 children in the comparison cohort developed ADHD yielding an IRR of 1.05 (95% confidence interval [CI] = 0.96, 1.15) and an IRD of 0.28 (95% CI = -0.20, 0.80) pr. 1,000 person-years. IRRs from analyses attempting to reduce exposure misclassification varied from 1.03 to 1.07.

CONCLUSIONS

Our results were not consistent with the hypothesized effect of prenatal antidepressant exposure on the risk of ADHD. Attempts to reduce exposure misclassification did not alter this finding.

Prenatal opioid exposure and risk of asthma in childhood: a population-based study from Denmark, Norway, and Sweden

Background: Opioids may modulate the immune function through opioid receptors on immune cells. Long-term consequences of prenatal opioid exposure on the immune system, such as childhood asthma, are unknown. Objectives: To investigate whether prenatal opioid exposure is associated with the risk of childhood asthma. Methods: Cohort study using linked nationwide registers in Denmark (1996-2015), Norway (2005-2015), and Sweden (2006-2013). Children born by mothers who were chronic opioid analgesics users before pregnancy (n = 14,764) or who were receiving opioid maintenance therapy (OMT) before or during pregnancy (n = 1,595) were identified based on information from each of the medical birth registers and prescription registers. Long-term opioid analgesics exposed children were compared to short-term exposed or unexposed, whereas OMT exposed children were compared to OMT unexposed. Asthma among children ≥1 years of age was defined as two or more filled prescriptions of antiasthmatic medication within 365 days, or a diagnosis of asthma. Hazard ratios (HRs) were calculated using Cox proportional hazards regression with attained age as the time scale. Inverse probability of treatment weights based on propensity scores were applied to adjust for measured confounders. Individual level data from Norway and Sweden were pooled, whereas individual level data from Denmark were analyzed separately. For the opioid analgesics comparisons, adjusted HRs (aHR) from the combined Norwegian/Swedish data and the Danish data were pooled in a fixed-effects meta-analysis. Results: For the opioid analgesics cohort, no increased risk of asthma was observed in long-term exposed children neither compared with unexposed [aHR = 0.99 (95% CI 0.87-1.12)], nor compared with short-term exposed [aHR = 0.97 (0.86-1.10)]. No increased risk of asthma was observed in OMT exposed compared with OMT unexposed children [Norway/Sweden: aHR = 1.07 (0.60-1.92), Denmark: aHR = 1.25 (0.87-1.81)]. Results from sensitivity analyses, where potential misclassification of the outcome and misclassification of OMT exposure were assessed, as well as starting follow-up at 6 years of age, showed that the estimates of association were generally robust. Conclusion: We found no association between prenatal exposure to opioids and risk of childhood asthma. Results were consistent across two different opioid exposure groups with different confounder distributions.

Medication-Associated Phthalate Exposure and Childhood Cancer Incidence

BACKGROUND

Human phthalate exposure is widespread through contact with myriad consumer products. Exposure is particularly high through medications formulated with phthalates. Phthalates disrupt normal endocrine signaling and are associated with reproductive outcomes and incidence of some cancers. We measured associations between gestational and childhood medication-associated phthalate exposures and the incidence of childhood cancers.

METHODS

We identified all live births in Denmark between 1997 and 2017, including both children and birth mothers. Using drug ingredient data merged with the Danish National Prescription Registry, we measured phthalate exposure through filled prescriptions for mothers during pregnancy (gestational exposure) and for children from birth until age 19 years (childhood exposure). Incident childhood cancers were ascertained from the Danish Cancer Registry, and associations were estimated with Cox regression models.

RESULTS

Among 1 278 685 children, there were 2027 childhood cancer cases diagnosed over 13.1 million person-years of follow-up. Childhood phthalate exposure was strongly associated with incidence of osteosarcoma (hazard ratio [HR] = 2.78, 95% confidence interval [CI] = 1.63 to 4.75). We also observed a positive association with incidence of lymphoma (HR = 2.07, 95% CI = 1.36 to 3.14), driven by associations with Hodgkin and non-Hodgkin lymphoma but not Burkitt lymphoma. Associations were apparent only for exposure to low-molecular phthalates, which have purportedly greater biological activity.

CONCLUSIONS

Childhood phthalate exposure was associated with incidence of osteosarcoma and lymphoma before age 19 years. Lingering questions include which specific phthalate(s) are responsible for these associations, by what mechanisms they occur, and to what extent childhood cancer cases could be avoided by reducing or eliminating the phthalate content of medications and other consumer products.

The Incidence of Breast Cancer Recurrence 10-32 Years after Primary Diagnosis

Background

Extended, more effective breast cancer treatments have increased the prevalence of long-term survivors. We investigated the risk of late breast cancer recurrence (BCR), 10 years or more after primary diagnosis, and associations between patient and tumor characteristics at primary diagnosis and late BCR up to 32 years after primary breast cancer diagnosis.

Methods

Using the Danish Breast Cancer Group clinical database, we identified all women with an incident early breast cancer diagnosed during 1987-2004. We restricted to women who survived 10 years without a recurrence or second cancer (10-year disease-free survivors) and followed them from 10 years after breast cancer diagnosis date until late recurrence, death, emigration, second cancer, or December 31, 2018. We calculated incidence rates per 1000 person-years and cumulative incidences for late BCR, stratifying by patient and tumor characteristics. Using Cox regression, we calculated adjusted hazard ratios for late BCR accounting for competing risks.

Results

Among 36 924 women with breast cancer, 20 315 became 10-year disease-free survivors. Of these, 2595 developed late BCR (incidence rate = 15.53 per 1000 person-years, 95% confidence interval = 14.94 to 16.14; cumulative incidence = 16.6%, 95% confidence interval = 15.8% to 17.5%) from year 10 to 32 after primary diagnosis. Tumor size larger than 20 mm, lymph node–positive disease, and estrogen receptor–positive tumors were associated with increased cumulative incidences and hazards for late BCR.

Conclusions

Recurrences continued to occur up to 32 years after primary diagnosis. Women with high lymph node burden, large tumor size, and estrogen receptor–positive tumors had increased risk of late recurrence. Such patients may warrant extended surveillance, more aggressive treatment, or new therapy approaches.

Prognostic and discriminative accuracy of the quick Sepsis-related Organ Failure Assessment compared with an early warning score: a Danish cohort study

BACKGROUND

The clinical benefit of implementing the quick Sepsis-related Organ Failure Assessment (qSOFA) instead of early warning scores (EWS) to screen all hospitalised patients for critical illness has yet to be investigated in a large, multicentre study.

METHODS

We conducted a cohort study including all hospitalised patients ≥18 years with EWS recorded at hospitals in the Central Denmark Region during the year 2016. The primary outcome was intensive care unit (ICU) admission and/or death within 2 days following an initial EWS. Prognostic accuracy was examined using sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV). Discriminative accuracy was examined by the area under the receiver operating characteristic curve (AUROC).

RESULTS

Among 97 332 evaluated patients, 1714 (1.8%) experienced the primary outcome. The qSOFA ≥2 was less sensitive (11.7% (95% CI: 10.2% to 13.3%) vs 25.1% (95% CI: 23.1% to 27.3%)) and more specific (99.3% (95% CI: 99.2% to 99.3%) vs 97.5% (95% CI: 97.4% to 97.6%)) than EWS ≥5. The NPV was similar for the two scores (EWS ≥5, 98.6% (95% CI: 98.6% to 98.7%) and qSOFA ≥2, 98.4% (95% CI: 98.3% to 98.5%)), while the PPV was 15.1% (95% CI: 13.8% to 16.5%) for EWS ≥5 and 22.4% (95% CI: 19.7% to 25.3%) for qSOFA ≥2. The AUROC was 0.72 (95% CI: 0.70 to 0.73) for EWS and 0.66 (95% CI: 0.65 to 0.67) for qSOFA.

CONCLUSION

The qSOFA was less sensitive (qSOFA ≥2 vs EWS ≥5) and discriminatively accurate than the EWS for predicting ICU admission and/or death within 2 days after an initial EWS. This study did not support replacing EWS with qSOFA in all hospitalised patients.

Mortality of older acutely admitted medical patients after early discharge from emergency departments: a nationwide cohort study

Background

The mortality of older patients after early discharge from hospitals is sparsely described. Information on factors associated with mortality can help identify high-risk patients who may benefit from preventive interventions. The aim of this study was to examine whether demographic factors, comorbidity and admission diagnoses are predictors of 30-day mortality among acutely admitted older patients discharged within 24 h after admission.

Methods

All medical patients aged ≥65 years admitted acutely to Danish hospitals between 1 January 2013 and 30 June 2014 surviving a hospital stay of ≤24 h were included. Demographic factors, comorbidity, discharge diagnoses and mortality within 30 days were described using data from the Danish National Patient Registry and the Civil Registration System. Cox regression was used to estimate adjusted hazard ratios (aHR) with 95% confidence intervals (CI) for all-cause mortality.

Results

A total of 93,295 patients (49.4% men) with a median age of 75 years (interquartile range: 69–82 years), were included. Out of these, 2775 patients (3.0%; 95% CI 2.9–3.1%) died within 30 days after discharge. The 30-day mortality was increased in patients with age 76–85 years (aHR 1.59; 1.45–1.75) and 86+ years (aHR 3.35; 3.04–3.70), male gender (aHR 1.22; 1.11–1.33), a Charlson Comorbidity Index of 1–2 (aHR 2.15; 1.92–2.40) and 3+ (aHR 4.07; 3.65–4.54), and unmarried status (aHR 1.17; 1.08–1.27). Discharge diagnoses associated with 30-day mortality were heart failure (aHR 1.52; 1.17–1.95), respiratory failure (aHR 3.18; 2.46–4.11), dehydration (aHR 2.87; 2.51–3.29), constipation (aHR 1.31; 1.02–1.67), anemia (aHR 1.45; 1.27–1.66), pneumonia (aHR 2.24; 1.94–2.59), urinary tract infection (aHR 1.33; 1.14–1.55), dyspnea (aHR 1.57; 1.32–1.87) and suspicion of malignancy (aHR 2.06; 1.64–2.59).

Conclusions

Three percent had died within 30 days. High age, male gender, the comorbidity burden, unmarried status and several primary discharge diagnoses were identified as independent prognostic factors of 30-day all-cause mortality.

Early-Life Exposure to Antibiotics and Risk for Crohn's Disease: A Nationwide Danish Birth Cohort Study

BACKGROUND

Early-life antibiotic use can alter the intestinal flora and modify the risk of developing Crohn disease (CD), but rigorous epidemiological evidence is limited, with inconsistent results.

METHODS

We identified all children born in Denmark from 1995 to 2009 and followed them from birth until death, emigration, a diagnosis of CD, or January 1, 2013. Using Cox regression, we assessed the association between antibiotic exposure in the first year of life and subsequent risk for CD, adjusting for sex, degree of urbanization, birth order, birth year, route of delivery, gestational age, smoking during pregnancy, intake of nonsteroidal anti-inflammatory drugs in the first year of life, and family history of CD.

RESULTS

During a median 9.5 years (9.3 million total person-years), CD was diagnosed in 208 of 979,039 children. Antibiotic use in the first year of life was associated with a higher risk of CD (adjusted hazard ratio, 1.4; 95% confidence interval [CI], 1.1-1.8), with the highest risk with ≥6 courses of antibiotics (adjusted hazard ratio, 4.1; 95% CI, 2.0-8.5). A family history of CD did not modify these risk associations. The cumulative risk of CD at the 11th birthday for children exposed to antibiotics in their first year of life was 0.16‰ (95% CI, 0.11‰-0.22‰) compared to 0.11‰ (95% CI, 0.08‰-0.15‰) for children unexposed to antibiotics in their first year of life.

CONCLUSIONS

Antibiotic use in the first year of life is associated with a modestly increased risk for CD, although the absolute risk is very low.