A Propensity Score Approach and a Partitioned Approach for the Self-Controlled Case Series Design to Evaluate Safety of a 2-dose Vaccine Series: Application to myocarditis/pericarditis following mRNA COVID-19 vaccination.

The assumption that serious adverse events (SAEs) do not affect subsequent exposure might not hold when evaluating 2-dose vaccine safety through a self-controlled case series (SCCS) design. To address this, we developed: 1) propensity score SCCS (PS-SCCS) using a propensity score model involving SAEs during the risk interval after dose 1 (${R}_1\Big)$, and 2) partitioned SCCS (P-SCCS) estimating relative incidence (RI) separately for doses 1 and 2. In simulations, both provided unbiased RIs. Conversely, standard SCCS overestimated RI after dose 2. We applied these approaches to assess myocarditis/pericarditis risks after 2-dose mRNA COVID-19 vaccination in 12-39-year-olds. For BNT162b2, PS-SCCS yielded RIs of 1.85 (95% CI, 0.75-4.59) and 11.05 (95% CI, 6.53-18.68) 14 days after doses 1 and 2 respectively; standard SCCS provided similar RI after dose 1 and RI of 12.92 (95% CI, 7.56-22.09) after dose 2. For mRNA-1273, standard SCCS showed RIs of 1.96 (95% CI, 0.56-6.91) after dose 1 and 7.87 (95% CI, 3.33-18.57) after dose 2. As no mRNA-1273 recipients with SAEs during ${R}_1$ received dose 2, P-SCCS was used, yielding similar RI after dose 1 and RI of 6.48 (95% CI, 2.83-14.83) after dose 2. mRNA vaccines were associated with elevated myocarditis/pericarditis risks following dose 2 in 12-39-year-olds.

An individual segmented trajectory approach for identifying opioid use patterns using longitudinal dispensing data.

PURPOSE: The aim of this study is to use electronic opioid dispensing data to develop an individual segmented trajectory approach for identifying opioid use patterns. The resulting opioid use patterns can be used for examining the association between opioid use and drug overdose. METHODS: We retrospectively assembled a cohort of members on long-term opioid therapy (LTOT) between January 1, 2006 and June 30, 2019 who were 18 years and older and enrolled in one of three health care systems in the US. We have developed an individual segmented trajectory analysis for identifying various opioid use patterns by scanning over the follow-up and finding distinct opioid use patterns based on variability measured with coefficient of variation and trends of milligram morphine equivalents levels. RESULTS: Among 31, 865 members who were on LTOT between January 1, 2006 and June 30, 2019, 58.3% were female, and the average age was 55.4 years (STD = 15.4). The study population had 152 557 person-years of follow-up, with an average follow-up of 4.4 years per enrollment per person (STD = 3.4). This novel approach identified up to 13 distinct patterns including 88 756 episodes of "stable" pattern (42.1%) with an average follow-up of 11.2 months, 29 140 episodes of "increasing" pattern (13.8%) with an average follow-up of 6.0 months, 13 201 episodes of ≤10% dose reduction (6.3%) with an average follow-up of 10.4 months, 7286 episodes of 11%-20% dose reduction (3.5%) with an average follow-up of 5.3 months, 4457 episodes of 21%-30% dose reduction (2.1%) with an average follow-up of 4.0 months, and 9903 episodes of >30% dose reduction (4.7%) with an average follow-up of 2.6 months. CONCLUSIONS: A novel approach was developed to identify 13 distinct opioid use patterns using each individual's longitudinal dispensing data and these patterns can be used in examining overdose risk during the time that these patterns are ongoing.

Opioid Overdose Risk Following Hospital Discharge Among Individuals Prescribed Long-Term Opioid Therapy: a Risk Interval Analysis.

BACKGROUND: Individuals prescribed long-term opioid therapy (LTOT) have increased risk of readmission and death after hospital discharge. The risk of opioid overdose during the immediate post-discharge time period is unknown. OBJECTIVE: To examine the association between time since hospital discharge and opioid overdose among individuals prescribed LTOT. DESIGN: Self-controlled risk interval analysis. PARTICIPANTS: Adults prescribed LTOT with at least one hospital discharge at a safety-net health system and a non-profit healthcare organization in Colorado. MAIN MEASURES: We identified individuals prescribed LTOT who were discharged from January 2006 through June 2019. The outcome was a composite of fatal and non-fatal opioid overdoses during a 90-day post-discharge observation period, identified using electronic health record (EHR) and vital statistics data. Risk intervals included days 0-6 after index and subsequent hospital discharges. Control intervals ranged from days 7 to 89 after index discharge and included all other time during the observation period that did not fall within a risk interval or time readmitted during a subsequent hospitalization, which was excluded. Poisson regression was used to estimate incidence rate ratios (IRR) and 95% confidence intervals (CI) for overdose events during risk in comparison to control intervals. KEY RESULTS: We identified 7695 adults (63.3% over 55 years, 59.4% female, 20.3% Hispanic) who experienced 9499 total discharges during the study period. Twenty-one overdoses occurred during their observation periods (1174 per 100,000 person-years [9 in risk, 12 in control]). Overdose risk was significantly higher during the risk interval in comparison to the control interval (IRR 6.92; 95% CI 2.92-16.43). CONCLUSION: During the first 7 days after hospital discharge, individuals prescribed LTOT appear to be at elevated risk for opioid overdose. Clarifying mechanisms of overdose risk may help inform in-hospital and post-discharge prevention strategies.

Association Between Opioid Dose Reduction Rates and Overdose Among Patients Prescribed Long-Term Opioid Therapy.

BACKGROUND: Tapering long-term opioid therapy is an increasingly common practice, yet rapid opioid dose reductions may increase the risk of overdose. The objective of this study was to compare overdose risk following opioid dose reduction rates of ≤10%, 11% to 20%, 21% to 30%, and >30% per month to stable dosing. METHODS: We conducted a retrospective cohort study in three health systems in Colorado and Wisconsin. Participants were patients ≥18 years of age prescribed long-term opioid therapy between January 1, 2006, and June 30, 2019. Five opioid dosing patterns and drug overdoses (fatal and nonfatal) were identified using electronic health records, pharmacy records, and the National Death Index. Cox proportional hazard regression was conducted on a propensity score-weighted cohort to estimate adjusted hazard ratios (aHRs) for follow-up periods of 1, 3, 6, 9, and 12 months after a dose reduction. RESULTS: In a cohort of 17 540 patients receiving long-term opioid therapy, 42.7% of patients experienced a dose reduction. Relative to stable dosing, a dose reduction rate of >30% was associated with an increased risk of overdose and the aHR estimates decreased as the follow-up increased; the aHRs for the 1-, 6- and 12-month follow-ups were 5.33 (95% CI, 1.98-14.34), 1.81 (95% CI,1.08-3.03), and 1.49 (95% CI, 0.97-2.27), respectively. The slower tapering rates were not associated with overdose risk. CONCLUSIONS: Patients receiving long-term opioid therapy exposed to dose reduction rates of >30% per month had increased overdose risk relative to patients exposed to stable dosing. Results support the use of slow dose reductions to minimize the risk of overdose.

Identifying Relative Changes in Social Risk Factors: An Analytic Approach.

BACKGROUND: Individuals often report concurrent social risk factors such as food insecurity, unstable housing, and transportation barriers. Comparing relative changes between pairs of social risk factors may identify those that are more resistant to change. OBJECTIVE: The objective of this study was to develop a method to describe relative changes in pairs of social risk factors. RESEARCH DESIGN: This was a prospective cohort study. SUBJECTS: Participants in a randomized controlled trial of hypertension care in an Urban Indian Health Organization. MEASURES: We measured 7 social risk factors (housing, transportation, food, clothing, health care, utilities, and debts) at enrollment, 6, and 12 months among 295 participants in the trial. We hypothesized that pairwise comparisons could identify social risk factors that were less likely to change over time. We used conditional odds ratios (ORs) with 95% confidence intervals (CIs) to rank each pair. RESULTS: Food, clothing, health care, utilities, and debts had more changes between 0 and 6 months relative to housing (OR=2.3, 3.4, 4.7, 3.5, and 3.4, respectively; all 95% CI excluded 1.0). These same social risk factors also had more changes between baseline and 6 months relative to transportation (OR=2.8, 3.4, 4.9, 4.7, and 4.1, respectively; all 95% CI excluded 1.0). Changes in housing and transportation risk factors were comparable (OR=0.7, 95% CI: 0.4-1.4). Relative changes between 6 and 12 months were similar. CONCLUSIONS: Housing and transportation exhibited fewer relative changes than other social risk factors and might be more resistant to change. Awareness of the relationships between social risk factors can help define priorities for intervention.

COVID-19 Vaccination and Non-COVID-19 Mortality Risk - Seven Integrated Health Care Organizations, United States, December 14, 2020-July 31, 2021.

By September 21, 2021, an estimated 182 million persons in the United States were fully vaccinated against COVID-19.* Clinical trials indicate that Pfizer-BioNTech (BNT162b2), Moderna (mRNA-1273), and Janssen (Johnson & Johnson; Ad.26.COV2.S) vaccines are effective and generally well tolerated (1-3). However, daily vaccination rates have declined approximately 78% since April 13, 2021†; vaccine safety concerns have contributed to vaccine hesitancy (4). A cohort study of 19,625 nursing home residents found that those who received an mRNA vaccine (Pfizer-BioNTech or Moderna) had lower all-cause mortality than did unvaccinated residents (5), but no studies comparing mortality rates within the general population of vaccinated and unvaccinated persons have been conducted. To assess mortality not associated with COVID-19 (non-COVID-19 mortality) after COVID-19 vaccination in a general population setting, a cohort study was conducted during December 2020-July 2021 among approximately 11 million persons enrolled in seven Vaccine Safety Datalink (VSD) sites.§ After standardizing mortality rates by age and sex, this study found that COVID-19 vaccine recipients had lower non-COVID-19 mortality than did unvaccinated persons. After adjusting for demographic characteristics and VSD site, this study found that adjusted relative risk (aRR) of non-COVID-19 mortality for the Pfizer-BioNTech vaccine was 0.41 (95% confidence interval [CI] = 0.38-0.44) after dose 1 and 0.34 (95% CI = 0.33-0.36) after dose 2. The aRRs of non-COVID-19 mortality for the Moderna vaccine were 0.34 (95% CI = 0.32-0.37) after dose 1 and 0.31 (95% CI = 0.30-0.33) after dose 2. The aRR after receipt of the Janssen vaccine was 0.54 (95% CI = 0.49-0.59). There is no increased risk for mortality among COVID-19 vaccine recipients. This finding reinforces the safety profile of currently approved COVID-19 vaccines in the United States.

Sensitivity analyses of unmeasured and partially-measured confounders using multiple imputation in a vaccine safety study.

PURPOSE: Sensitivity analyses have played an important role in pharmacoepidemiology studies using electronic health records data. Despite the existence of quantitative bias analysis in pharmacoepidemiologic studies, simultaneously adjusting for unmeasured and partially measured confounders is challenging in vaccine safety studies. Our objective was to develop a flexible approach for conducting sensitivity analyses of unmeasured and partially-measured confounders concurrently for a vaccine safety study. METHODS: We derived conditional probabilities for an unmeasured confounder based on bias parameters, used these conditional probabilities and Monte Carlo simulations to impute the unmeasured confounder, and re-constructed the analytic datasets as if the unmeasured confounder had been observed. We simultaneously imputed a partially measured confounder using a prediction model. We considered unmeasured breastfeeding and partially measured family history of Type 1 diabetes (T1DM) in a study examining the association between exposure to rotavirus vaccination and T1DM. RESULTS: Before sensitivity analyses, the hazard ratios (HR) were 1.50 (95% CI, 0.81-2.77) for those partially exposed and 1.03 (95% CI, 0.62-1.72) for those fully exposed with unexposed children as the referent group. When breastfeeding and family history of T1DM were adjusted, the HR was 1.55 (95% CI, 0.84-2.87) for the partially exposed group; the HR was 0.98 (95% CI, 0.58-1.63) for the fully exposed group. CONCLUSIONS: We conclude that adjusting for unmeasured breastfeeding and partially measured family history of T1DM did not alter the conclusion that there was no evidence of association between rotavirus vaccination and developing T1DM. This novel approach allows for simultaneous adjustment for multiple unmeasured and partially-measured confounders.

Correction: Impact of the COVID-19 Pandemic on Health Care Utilization in a Large Integrated Health Care System: Retrospective Cohort Study.

[This corrects the article DOI: 10.2196/26558.].

Impact of the COVID-19 Pandemic on Health Care Utilization in a Large Integrated Health Care System: Retrospective Cohort Study.

BACKGROUND: The COVID-19 pandemic has caused an abrupt reduction in the use of in-person health care, accompanied by a corresponding surge in the use of telehealth services. However, the extent and nature of changes in health care utilization during the pandemic may differ by care setting. Knowledge of the impact of the pandemic on health care utilization is important to health care organizations and policy makers. OBJECTIVE: The aims of this study are (1) to evaluate changes in in-person health care utilization and telehealth visits during the COVID-19 pandemic and (2) to assess the difference in changes in health care utilization between the pandemic year 2020 and the prepandemic year 2019. METHODS: We retrospectively assembled a cohort consisting of members of a large integrated health care organization, who were enrolled between January 6 and November 2, 2019 (prepandemic year), and between January 5 and October 31, 2020 (pandemic year). The rates of visits were calculated weekly for four settings: inpatient, emergency department (ED), outpatient, and telehealth. Using Poisson models, we assessed the impact of the pandemic on health care utilization during the early days of the pandemic and conducted difference-in-deference (DID) analyses to measure the changes in health care utilization, adjusting for the trend of health care utilization in the prepandemic year. RESULTS: In the early days of the pandemic, we observed significant reductions in inpatient, ED, and outpatient utilization (by 30.2%, 37.0%, and 80.9%, respectively). By contrast, there was a 4-fold increase in telehealth visits between weeks 8 (February 23) and 12 (March 22) in 2020. DID analyses revealed that after adjusting for prepandemic secular trends, the reductions in inpatient, ED, and outpatient visit rates in the early days of the pandemic were 1.6, 8.9, and 367.2 visits per 100 person-years (P<.001), respectively, while the increase in telehealth visits was 272.9 visits per 100 person-years (P<.001). Further analyses suggested that the increase in telehealth visits offset the reduction in outpatient visits by week 26 (June 28, 2020). CONCLUSIONS: In-person health care utilization decreased drastically during the early period of the pandemic, but there was a corresponding increase in telehealth visits during the same period. By end-June 2020, the combined outpatient and telehealth visits had recovered to prepandemic levels.

Disparities in Outpatient and Telehealth Visits During the COVID-19 Pandemic in a Large Integrated Health Care Organization: Retrospective Cohort Study.

BACKGROUND: Dramatic decreases in outpatient visits and sudden increases in telehealth visits were observed during the COVID-19 pandemic, but it was unclear whether these changes differed by patient demographics and socioeconomic status. OBJECTIVE: This study aimed to assess the impact of the pandemic on in-person outpatient and telehealth visits (telephone and video) by demographic characteristics and household income in a diverse population. METHODS: We calculated weekly rates of outpatient and telehealth visits by age, sex, race/ethnicity, and neighborhood-level median household income among members of Kaiser Permanente Southern California (KPSC) from January 5, 2020, to October 31, 2020, and the corresponding period in 2019. We estimated the percentage change in visit rates during the early pandemic period (March 22 to April 25, 2020) and the late pandemic period (October 4 to October 31, 2020) from the prepandemic period (January 5 to March 7, 2020) in Poisson regression models for each subgroup while adjusting for seasonality using 2019 data. We examined if the changes in visit rates differed by subgroups statistically by comparing their 95% CIs. RESULTS: Among 4.56 million KPSC members enrolled in January 2020, 15.0% (n=682,947) were ≥65 years old, 51.5% (n=2,345,020) were female, 39.4% (n=1,795,994) were Hispanic, and 7.7% (n=350,721) lived in an area of median household income

Prediction Model for Two-Year Risk of Opioid Overdose Among Patients Prescribed Chronic Opioid Therapy.

BACKGROUND: Naloxone is a life-saving opioid antagonist. Chronic pain guidelines recommend that physicians co-prescribe naloxone to patients at high risk for opioid overdose. However, clinical tools to efficiently identify patients who could benefit from naloxone are lacking. OBJECTIVE: To develop and validate an overdose predictive model which could be used in primary care settings to assess the need for naloxone. DESIGN: Retrospective cohort. SETTING: Derivation site was an integrated health system in Colorado; validation site was a safety-net health system in Colorado. PARTICIPANTS: We developed a predictive model in a cohort of 42,828 patients taking chronic opioid therapy and externally validated the model in 10,708 patients. MAIN MEASURES: Potential predictors and outcomes (nonfatal pharmaceutical and heroin overdoses) were extracted from electronic health records. Fatal overdose outcomes were identified from state vital records. To match the approximate shelf-life of naloxone, we used Cox proportional hazards regression to model the 2-year risk of overdose. Calibration and discrimination were assessed. KEY RESULTS: A five-variable predictive model showed good calibration and discrimination (bootstrap-corrected c-statistic = 0.73, 95% confidence interval [CI] 0.69-0.78) in the derivation site, with sensitivity of 66.1% and specificity of 66.6%. In the validation site, the model showed good discrimination (c-statistic = 0.75, 95% CI 0.70-0.80) and less than ideal calibration, with sensitivity and specificity of 82.2% and 49.5%, respectively. CONCLUSIONS: Among patients on chronic opioid therapy, the predictive model identified 66-82% of all subsequent opioid overdoses. This model is an efficient screening tool to identify patients who could benefit from naloxone to prevent overdose deaths. Population differences across the two sites limited calibration in the validation site.

Meningococcal conjugate vaccine safety surveillance in the Vaccine Safety Datalink using a tree-temporal scan data mining method.

PURPOSE: The objective of our study was to conduct a data mining analysis to identify potential adverse events (AEs) following MENACWY-D using the tree-temporal scan statistic in the Vaccine Safety Datalink population and demonstrate the feasibility of this method in a large distributed safety data setting. METHODS: Traditional pharmacovigilance techniques used in vaccine safety are generally geared to detecting AEs based on pre-defined sets of conditions or diagnoses. Using a newly developed tree-temporal scan statistic data mining method, we performed a pilot study to evaluate the safety profile of the meningococcal conjugate vaccine Menactra® (MenACWY-D), screening thousands of potential AE diagnoses and diagnosis groupings. The study cohort included enrolled participants in the Vaccine Safety Datalink aged 11 to 18 years who had received MenACWY-D vaccination(s) between 2005 and 2014. The tree-temporal scan statistic was employed to identify statistical associations (signals) of AEs following MENACWY-D at a 0.05 level of significance, adjusted for multiple testing. RESULTS: We detected signals for 2 groups of outcomes: diseases of the skin and subcutaneous tissue, fever, and urticaria. Both groups are known AEs following MENACWY-D vaccination. We also identified a statistical signal for pleurisy, but further examination suggested it was likely a false signal. No new MENACWY-D safety concerns were raised. CONCLUSIONS: As a pilot study, we demonstrated that the tree-temporal scan statistic data mining method can be successfully applied to screen broadly for a wide range of vaccine-AE associations within a large health care data network.

The safety of live attenuated influenza vaccine in children and adolescents 2 through 17 years of age: A Vaccine Safety Datalink study.

PURPOSE: To evaluate the safety of live attenuated influenza vaccine (LAIV) in children 2 through 17 years of age. METHODS: The study was conducted in 6 large integrated health care organizations participating in the Vaccine Safety Datalink (VSD). Trivalent LAIV safety was assessed in children who received LAIV between September 1, 2003 and March 31, 2013. Eighteen pre-specified adverse event groups were studied, including allergic, autoimmune, neurologic, respiratory, and infectious conditions. Incident rate ratios (IRRs) were calculated for each adverse event, using self-controlled case series analyses. For adverse events with a statistically significant increase in risk, or an IRR > 2.0 regardless of statistical significance, manual medical record review was performed to confirm case status. RESULTS: During the study period, 396 173 children received 590 018 doses of LAIV. For 13 adverse event groups, there was no significant increased risk of adverse events following LAIV. Five adverse event groups (anaphylaxis, syncope, Stevens-Johnson syndrome, adverse effect of drug, and respiratory failure) met criteria for manual medical record review. After review to confirm cases, 2 adverse event groups remained significantly associated with LAIV: anaphylaxis and syncope. One confirmed case of anaphylaxis was observed following LAIV, a rate of 1.7 per million LAIV doses. Five confirmed cases of syncope were observed, a rate of 8.5 per million doses. CONCLUSIONS: In a study of trivalent LAIV safety in a large cohort of children, few serious adverse events were detected. Anaphylaxis and syncope occurred following LAIV, although rarely. These data provide reassurance regarding continued LAIV use.

Analyzing self-controlled case series data when case confirmation rates are estimated from an internal validation sample.

Vaccine safety studies are often electronic health record (EHR)-based observational studies. These studies often face significant methodological challenges, including confounding and misclassification of adverse event. Vaccine safety researchers use self-controlled case series (SCCS) study design to handle confounding effect and employ medical chart review to ascertain cases that are identified using EHR data. However, for common adverse events, limited resources often make it impossible to adjudicate all adverse events observed in electronic data. In this paper, we considered four approaches for analyzing SCCS data with confirmation rates estimated from an internal validation sample: (1) observed cases, (2) confirmed cases only, (3) known confirmation rate, and (4) multiple imputation (MI). We conducted a simulation study to evaluate these four approaches using type I error rates, percent bias, and empirical power. Our simulation results suggest that when misclassification of adverse events is present, approaches such as observed cases, confirmed case only, and known confirmation rate may inflate the type I error, yield biased point estimates, and affect statistical power. The multiple imputation approach considers the uncertainty of estimated confirmation rates from an internal validation sample, yields a proper type I error rate, largely unbiased point estimate, proper variance estimate, and statistical power.

Monitoring Depression Rates in an Urban Community: Use of Electronic Health Records.

OBJECTIVES: Depression is the most common mental health disorder and mediates outcomes for many chronic diseases. Ability to accurately identify and monitor this condition, at the local level, is often limited to estimates from national surveys. This study sought to compare and validate electronic health record (EHR)-based depression surveillance with multiple data sources for more granular demographic subgroup and subcounty measurements. DESIGN/SETTING: A survey compared data sources for the ability to provide subcounty (eg, census tract [CT]) depression prevalence estimates. Using 2011-2012 EHR data from 2 large health care providers, and American Community Survey data, depression rates were estimated by CT for Denver County, Colorado. Sociodemographic and geographic (residence) attributes were analyzed and described. Spatial analysis assessed for clusters of higher or lower depression prevalence. MAIN OUTCOME MEASURE(S): Depression prevalence estimates by CT. RESULTS: National and local survey-based depression prevalence estimates ranged from 7% to 17% but were limited to county level. Electronic health record data provided subcounty depression prevalence estimates by sociodemographic and geographic groups (CT range: 5%-20%). Overall depression prevalence was 13%; rates were higher for women (16% vs men 9%), whites (16%), and increased with age and homeless patients (18%). Areas of higher and lower EHR-based, depression prevalence were identified. CONCLUSIONS: Electronic health record-based depression prevalence varied by CT, gender, race/ethnicity, age, and living status. Electronic health record-based surveillance complements traditional methods with greater timeliness and granularity. Validation through subcounty-level qualitative or survey approaches should assess accuracy and address concerns about EHR selection bias. Public health agencies should consider the opportunity and evaluate EHR system data as a surveillance tool to estimate subcounty chronic disease prevalence.

Risk of Breast Cancer With Long-Term Use of Calcium Channel Blockers or Angiotensin-Converting Enzyme Inhibitors Among Older Women.

Controversy exists about breast cancer risk associated with long-term use of calcium channel blockers (CCBs) or angiotensin-converting enzyme inhibitors (ACEis), respectively. Our objective in this study was to separately evaluate associations between duration of CCB or ACEi use and breast cancer in hypertensive women aged ≥55 years at 3 sites in the Kaiser Permanente health-care system (1997­2012). Exposures included CCB or ACEi use of 1­12 years' duration, determined from pharmacy dispensings. Outcomes included invasive lobular or ductal carcinoma. Statistical methods included discrete-time survival analyses. The cohort included 19,674 (17.9%) CCB users and 90,078 (82.1%) ACEi users. Two percent (n = 397) of CCB users and 1.9% (n = 1,733) of ACEi users developed breast cancer. Compared with 1­<2 years of use, in adjusted analysis, there was no association between CCB use for 2­<12 years and breast cancer: All 95% confidence intervals included 1. Increasing duration of ACEi use was associated with reduced breast cancer risk: Compared with 1­<2 years of use, the adjusted hazard ratio was 0.76 (95% confidence interval: 0.63, 0.92) for 5­<6 years of use and 0.63 (95% confidence interval: 0.43, 0.93) for 9­<10 years of use. We conclude that among older women with hypertension, long-term CCB use does not increase breast cancer risk and long-term treatment with ACEis may confer protection against breast cancer.

A Multilevel Analysis of Individual, Health System, and Neighborhood Factors Associated with Depression within a Large Metropolitan Area.

Depression prevalence is known to vary by individual factors (gender, age, race, medical comorbidities) and by neighborhood factors (neighborhood deprivation). However, the combination of individual- and neighborhood-level data is rarely available to assess their relative contribution to variation in depression across neighborhoods. We geocoded depression diagnosis and demographic data from electronic health records for 165,600 patients seen in two large health systems serving the Denver population (Kaiser Permanente and Denver Health) to Denver's 144 census tracts, and combined these data with indices of neighborhood deprivation obtained from the American Community Survey. Non-linear mixed models examined the relationships between depression rates and individual and census tract variables, stratified by health system. We found higher depression rates associated with greater age, female gender, white race, medical comorbidities, and with lower rates of home owner occupancy, residential stability, and higher educational attainment, but not with economic disadvantage. Among the Denver Health cohort, higher depression rates were associated with higher crime rates and a lower percent of foreign born residents and single mother households. Our findings suggest that individual factors had the strongest associations with depression. Neighborhood risk factors associated with depression point to low community cohesion, while the role of education is more complex. Among the Denver Health cohort, language and cultural barriers and competing priorities may attenuate the recognition and treatment of depression.

Risk of Preterm or Small-for-Gestational-Age Birth After Influenza Vaccination During Pregnancy: Caveats When Conducting Retrospective Observational Studies.

Vaccines are increasingly targeted toward women of reproductive age, and vaccines to prevent influenza and pertussis are recommended during pregnancy. Prelicensure clinical trials typically have not included pregnant women, and when they are included, trials cannot detect rare events. Thus, postmarketing vaccine safety assessments are necessary. However, analysis of observational data requires detailed assessment of potential biases. Using data from 8 Vaccine Safety Datalink sites in the United States, we analyzed the association of monovalent H1N1 influenza vaccine (MIV) during pregnancy with preterm birth (<37 weeks) and small-for-gestational-age birth (birth weight < 10th percentile). The cohort included 46,549 pregnancies during 2009-2010 (40% of participants received the MIV). We found potential biases in the vaccine-birth outcome association that might occur due to variable access to vaccines, the time-dependent nature of exposure to vaccination within pregnancy (immortal time bias), and confounding from baseline differences between vaccinated and unvaccinated women. We found a strong protective effect of vaccination on preterm birth (relative risk = 0.79, 95% confidence interval: 0.74, 0.85) when we ignored potential biases and no effect when accounted for them (relative risk = 0.91; 95% confidence interval: 0.83, 1.0). In contrast, we found no important biases in the association of MIV with small-for-gestational-age birth. Investigators conducting studies to evaluate birth outcomes after maternal vaccination should use statistical approaches to minimize potential biases.

Evaluation of propensity scores, disease risk scores, and regression in confounder adjustment for the safety of emerging treatment with group sequential monitoring.

PURPOSE: The objective of this study was to evaluate regression, matching, and stratification on propensity score (PS) or disease risk score (DRS) in a setting of sequential analyses where statistical hypotheses are tested multiple times. METHODS: In a setting of sequential analyses, we simulated incident users and binary outcomes with different confounding strength, outcome incidence, and the adoption rate of treatment. We compared Type I error rate, empirical power, and time to signal using the following confounder adjustments: (i) regression; (ii) treatment matching (1:1 or 1:4) on PS or DRS; and (iii) stratification on PS or DRS. We estimated PS and DRS using lookwise and cumulative methods (all data up to the current look). We applied these confounder adjustments in examining the association between non-steroidal anti-inflammatory drugs and bleeding. RESULTS: Propensity score and DRS methods had similar empirical power and time to signal. However, DRS methods yielded Type I error rates up to 17% for 1:4 matching and 15.3% for stratification methods when treatment and outcome were common and confounding strength with treatment was stronger. When treatment and outcome were not common, stratification on PS and DRS and regression yielded 8-10% Type I error rates and inflated empirical power. However, when outcome and treatment were common, both regression and stratification on PS outperformed other matching methods with Type I error rates close to 5%. CONCLUSIONS: We suggest regression and stratification on PS when the outcomes and/or treatment is common and use of matching on PS with higher ratios when outcome or treatment is rare or moderately rare.

Trends in diabetes incidence among 7 million insured adults, 2006-2011: the SUPREME-DM project.

An observational cohort analysis was conducted within the Surveillance, Prevention, and Management of Diabetes Mellitus (SUPREME-DM) DataLink, a consortium of 11 integrated health-care delivery systems with electronic health records in 10 US states. Among nearly 7 million adults aged 20 years or older, we estimated annual diabetes incidence per 1,000 persons overall and by age, sex, race/ethnicity, and body mass index. We identified 289,050 incident cases of diabetes. Age- and sex-adjusted population incidence was stable between 2006 and 2010, ranging from 10.3 per 1,000 adults (95% confidence interval (CI): 9.8, 10.7) to 11.3 per 1,000 adults (95% CI: 11.0, 11.7). Adjusted incidence was significantly higher in 2011 (11.5, 95% CI: 10.9, 12.0) than in the 2 years with the lowest incidence. A similar pattern was observed in most prespecified subgroups, but only the differences for persons who were not white were significant. In 2006, 56% of incident cases had a glycated hemoglobin (hemoglobin A1c) test as one of the pair of events identifying diabetes. By 2011, that number was 74%. In conclusion, overall diabetes incidence in this population did not significantly increase between 2006 and 2010, but increases in hemoglobin A1c testing may have contributed to rising diabetes incidence among nonwhites in 2011.