ABSTRACT
BACKGROUND: Instrumental variable (IV) analysis provides an alternative set of identification assumptions in the presence of uncontrolled confounding when attempting to estimate causal effects. Our objective was to evaluate the suitability of measures of prescriber preference and calendar time as potential IVs to evaluate the comparative effectiveness of buprenorphine/naloxone versus methadone for treatment of opioid use disorder (OUD). METHODS: Using linked population-level health administrative data, we constructed five IVs: prescribing preference at the individual, facility, and region levels (continuous and categorical variables), calendar time, and a binary prescriber's preference IV in analyzing the treatment assignment-treatment discontinuation association using both incident-user and prevalent-new-user designs. Using published guidelines, we assessed and compared each IV according to the four assumptions for IVs, employing both empirical assessment and content expertise. We evaluated the robustness of results using sensitivity analyses. RESULTS: The study sample included 35,904 incident users (43.3% on buprenorphine/naloxone) initiated on opioid agonist treatment by 1585 prescribers during the study period. While all candidate IVs were strong (A1) according to conventional criteria, by expert opinion, we found no evidence against assumptions of exclusion (A2), independence (A3), monotonicity (A4a), and homogeneity (A4b) for prescribing preference-based IV. Some criteria were violated for the calendar time-based IV. We determined that preference in provider-level prescribing, measured on a continuous scale, was the most suitable IV for comparative effectiveness of buprenorphine/naloxone and methadone for the treatment of OUD. CONCLUSIONS: Our results suggest that prescriber's preference measures are suitable IVs in comparative effectiveness studies of treatment for OUD.
Subject(s)
Methadone , Opioid-Related Disorders , Humans , Methadone/therapeutic use , Opioid-Related Disorders/drug therapy , Buprenorphine, Naloxone Drug Combination/therapeutic use , Opiate Substitution Treatment/methods , Health Status , Analgesics, Opioid/therapeutic useABSTRACT
PURPOSE: Propensity score matching is vital in epidemiological studies using observational data, yet its estimates relies on correct model-specification. This study assesses supervised deep learning models and unsupervised autoencoders for propensity score estimation, comparing them with traditional methods for bias and variance accuracy in treatment effect estimations. METHODS: Utilizing a plasmode simulation based on the Right Heart Catheterization dataset, under a variety of settings, we evaluated (1) a supervised deep learning architecture and (2) an unsupervised autoencoder, alongside two traditional methods: logistic regression and a spline-based method in estimating propensity scores for matching. Performance metrics included bias, standard errors, and coverage probability. The analysis was also extended to real-world data, with estimates compared to those obtained via a double robust approach. RESULTS: The analysis revealed that supervised deep learning models outperformed unsupervised autoencoders in variance estimation while maintaining comparable levels of bias. These results were supported by analyses of real-world data, where the supervised model's estimates closely matched those derived from conventional methods. Additionally, deep learning models performed well compared to traditional methods in settings where exposure was rare. CONCLUSION: Supervised deep learning models hold promise in refining propensity score estimations in epidemiological research, offering nuanced confounder adjustment, especially in complex datasets. We endorse integrating supervised deep learning into epidemiological research and share reproducible codes for widespread use and methodological transparency.
Subject(s)
Deep Learning , Propensity Score , Humans , Supervised Machine Learning , Logistic Models , Cardiac Catheterization/methods , Cardiac Catheterization/statistics & numerical data , Algorithms , Computer SimulationABSTRACT
ObjectiveIdentifying the effects of comorbidity on healthcare utilization is critical for understanding the benefits of improved comorbidity management. Asthma is a common respiratory condition, associated with gastrointestinal, metabolic, psychiatric, and other respiratory conditions. Adults with asthma represent a key population in understanding comorbidity and its consequences. The objective was to explore the relationship between comorbidity and overnight hospitalizations in U.S. adults with asthma.Study Design and MethodsA cross-sectional sample of 3,887 subjects aged 20-79 was aggregated from seven cycles (2005-2018) of the National Health and Nutrition Examination Survey (NHANES). The survey design was created using the full seven cycles, then a subpopulation was used for the analysis. Design-based modified Poisson regression with robust standard errors compared the prevalence of overnight hospitalizations in subjects with and without comorbidities. Comorbidity was defined as the presence of one or more additional chronic conditions.ResultsOver half (61.6%) of patients with asthma reported having comorbidities. The overnight hospitalization prevalence was higher in those with comorbidities (21.6%) than those without (7.4%). The adjusted prevalence ratio of overnight hospitalizations in those with comorbidities versus those without was 2.02 (95% CI: 1.54-2.66). Conclusions from sensitivity analyses remained the same.ConclusionsComorbidity in U.S. adult asthma patients is associated with increased overnight hospitalizations. Study results concur with examinations of other healthcare utilization outcomes, revealing how comorbidity influences healthcare utilization patterns in patients with asthma. The reduction of overnight hospitalizations should be a targeted goal when developing and evaluating interventions to manage comorbidities in patients with asthma.
ABSTRACT
Recent research has highlighted the Omicron variant's capacity to evade immune protection conferred by wild-type (WT) mRNA vaccines. Despite this observation, the potential involvement of antigenic sin phenomena remains unclear. Our hypothesis posited that a greater number of prior WT vaccine doses might lead to reduced anti-Omicron neutralization Abs following Omicron infection. To investigate this, we analyzed blood samples from human participants in the COVID-19 Occupational Risk, Seroprevalence, and Immunity among Paramedics (CORSIP) study who had received at least one WT mRNA vaccine before contracting Omicron. The exposure variable was the number of WT mRNA vaccines administered, and the outcome was the angiotensin-converting enzyme 2 (ACE-2) percent inhibition specific to the BA.4/BA.5 Omicron Ag. Contrary to expectations, our findings revealed that more WT-based vaccines were associated with an enhanced Omicron-specific immune response.
Subject(s)
COVID-19 , SARS-CoV-2 , Humans , Paramedics , Seroepidemiologic Studies , mRNA Vaccines , Canada/epidemiology , AntibodiesABSTRACT
PURPOSE: Household air pollution is one of the leading causes of death and disease globally. Emerging evidence of elevated risk of neonatal death has been reported in Africa and South Asia. However, evidence on the extent of the problem in Latin America is limited despite the persistent use of highly polluting cooking fuels. We assessed whether the use of high-polluting household cooking fuels increases the risk of neonatal death compared to low-polluting fuels in Colombia. METHODS: We used cross-sectional data from the 2005-2015 Colombian Demographic Health Survey and performed a survey-featured multivariate logistic regression. We selected adjustment covariates based on a causal diagram, addressed missing data through multiple imputation, and conducted several sensitivity analysis, such as propensity score matching. RESULT: We found evidence suggesting an increased risk of neonatal death in households using high-polluting fuels (OR: 1.48; 95% CI: 0.91, 2.39). The sensitivity analyses were consistent with the main analysis. CONCLUSION: We observed increased odds of neonatal death associated with using high-polluting household cooking fuels compared to low-polluting fuels, although this association was not statistically significant. This study contributes evidence to a region where the issue is not yet a priority and should be included in national-level discussions and interventions that impact cooking fuel use patterns.
Subject(s)
Air Pollution, Indoor , Cooking , Humans , Colombia/epidemiology , Air Pollution, Indoor/adverse effects , Infant, Newborn , Infant , Female , Cross-Sectional Studies , Male , Infant Mortality , Infant Death/etiology , Adult , Air Pollutants/analysis , Air Pollutants/adverse effectsABSTRACT
Objectives: Poverty exposes children to adverse conditions that negatively impact development. However, there is limited understanding on how different types of poverty may affect children of various immigration backgrounds differently in outcomes such as school readiness. This study examined these relationships between household and/or neighbourhood poverty, poverty timing, and immigration background with school readiness outcomes at kindergarten. Methods: This study utilized a retrospective, population-based cohort of administrative records linked with surveys completed by kindergarten teachers for 15 369 children born in British Columbia, Canada. The exposures investigated were neighbourhood poverty (residing in a neighbourhood in the lowest income-quintile) and/or household poverty (receiving a health insurance subsidy due to low household income). Experiencing both neighbourhood and household poverty simultaneously was defined as "combined" household and neighbourhood poverty. The outcome of vulnerability on school readiness domains was assessed at kindergarten (47.8% female; mean age = 6.01 years) using teacher ratings on the Early Development Instrument (EDI). Results: Children exposed to combined poverty between age 0 and 2 had greater odds of being vulnerable in two or more domains of school readiness than children not exposed to any poverty during this period (adjusted odds ratio (aOR) = 2.07, 95% CI: [1.74; 2.47], p < 0.001). The effect of combined poverty was larger than household poverty only (aOR = 1.54, 95% CI: [1.31; 1.82], p < 0.001) or neighbourhood poverty only (aOR = 1.49, 95% CI: [1.30; 1.70], p < 0.001). Combined poverty was associated with negative outcomes regardless of timing. Both non-immigrants (aOR = 2.40, 95% CI: [1.92; 3.00], p < 0.001) and second-generation immigrants (aOR = 1.63, 95% CI: [1.22; 2.17], p < 0.001) experiencing combined poverty scored lower on school readiness. Conclusions: Children who experienced combined poverty had lower levels of school readiness at kindergarten, regardless of timing and immigration background.
ABSTRACT
BACKGROUND: In many jurisdictions, policies restrict access to Opioid Agonist Treatment (OAT) in correctional facilities. Receipt of OAT during incarceration is associated with reduced risk of fatal overdose after release but little is known about the effect on nonfatal overdose. This study aimed to examine the association between OAT use during incarceration and nonfatal overdose in the 30 days following release. METHODS AND FINDINGS: Using linked administrative healthcare and corrections data for a random sample of 20% of residents of British Columbia, Canada we examined releases from provincial correctional facilities between January 1, 2015 -December 1, 2018, among adults (aged 18 or older at the time of release) with Opioid Use Disorder. We fit Andersen-Gill models to examine the association between receipt of OAT in custody and the hazard of nonfatal following release. We conducted secondary analyses to examine the association among people continuing treatment initiated prior to their arrest and people who initiated a new episode of OAT in custody separately. We also conducted sex-based subgroup analyses. In this study there were 4,738 releases of 1,535 people with Opioid Use Disorder. In adjusted analysis, receipt of OAT in custody was associated with a reduced hazard of nonfatal overdose (aHR 0.55, 95% CI 0.41, 0.74). This was found for prescriptions continued from community (aHR 0.49, 95%CI 0.36, 0.67) and for episodes of OAT initiated in custody (aHR 0.58, 95%CI 0.41, 0.82). The effect was greater among women than men. CONCLUSIONS: OAT receipt during incarceration is associated with a reduced hazard of nonfatal overdose after release. Policies to expand access to OAT in correctional facilities, including initiating treatment, may help reduce harms related to nonfatal overdose in the weeks following release. Differences in the effect seen among women and men indicate a need for gender-responsive policies and programming.
Subject(s)
Drug Overdose , Opioid-Related Disorders , Humans , Male , Female , British Columbia/epidemiology , Adult , Drug Overdose/drug therapy , Drug Overdose/epidemiology , Drug Overdose/prevention & control , Opioid-Related Disorders/drug therapy , Opioid-Related Disorders/epidemiology , Middle Aged , Young Adult , Analgesics, Opioid/therapeutic use , Prisons/statistics & numerical data , Adolescent , Opiate Substitution Treatment/methods , Prisoners/statistics & numerical data , Opiate Overdose/drug therapy , Opiate Overdose/epidemiologyABSTRACT
Background: HCV infection is associated with mortality due to extrahepatic manifestations (EHM). Sustained virologic response (SVR) following direct-acting antiviral (DAA) therapy has been linked to decreased all-cause and liver-related mortality. However, evidence regarding the impact of DAA on EHM-related deaths is lacking. This study aimed to assess the impact of DAA and SVR on EHM-related mortality. Methods: The British Columbia Hepatitis Testers Cohort comprises â¼1.7 million people tested for HCV between 1990 and 2015 and is linked with administrative health data. Among individuals diagnosed with HCV by 12/31/2020, those who received at least one DAA treatment were matched to those who never received treatment by the year of their first HCV RNA positive date. We compared three groups: treated & SVR, treated & no-SVR, and untreated; and generated EHM mortality rates and incidence curves. To account for differences in baseline characteristics, we used inverse probability of treatment weights (IPTW). IPTW-weighted multivariable cause-specific Cox regression models were adjusted for competing risk and confounders. Findings: Study population included 12,815 treated (12,287 SVR, 528 no-SVR) and 12,815 untreated individuals (median follow-up 3.4 years, IQR 2.9). The untreated group had the highest EHM mortality rate (30.9 per 1000 person-years [PY], 95% CI 29.2-32.8), followed by the treated & no-SVR group (21.2 per 1000 PY, 95% CI 14.9-30.1), while the treated & SVR group had the lowest EHM mortality rate (7.9 per 1000 PY, 95% CI 7.1-8.7). In the multivariable model, EHM mortality in the treated & SVR group was significantly decreased (adjusted cause-specific hazard ratio [acsHR] 0.20, 95% CI 0.18-0.23). The treated & SVR group had significant reductions in mortality related to each of the EHMs (78-84%). Interpretation: Treatment of HCV with DAA was associated with significant reductions in EHM-related mortality. These findings emphasize the critical importance of timely diagnosis and treatment of HCV to prevent deaths associated with EHM, and have important implications for clinical practice and public health. Funding: This work was supported by the BC Centre for Disease Control and the Canadian Institutes of Health Research (CIHR) [Grant # NHC-348216, PJT-156066, and PHE-337680]. DJ has received Doctoral Research Award (#201910DF1-435705-64343) from the Canadian Institutes of Health Research (CIHR) and Doctoral fellowship from the Canadian Network on Hepatitis C (CanHepC). CanHepC is funded by a joint initiative of the Canadian Institutes of Health Research (CIHR) (NHC-142832) and the Public Health Agency of Canada (PHAC).
ABSTRACT
BACKGROUND: In Italy, surgeons continue to drain the abdominal cavity in more than 50 per cent of patients after colorectal resection. The aim of this study was to evaluate the impact of abdominal drain placement on early adverse events in patients undergoing elective colorectal surgery. METHODS: A database was retrospectively analysed through a 1:1 propensity score-matching model including 21 covariates. The primary endpoint was the postoperative duration of stay, and the secondary endpoints were surgical site infections, infectious morbidity rate defined as surgical site infections plus pulmonary infections plus urinary infections, anastomotic leakage, overall morbidity rate, major morbidity rate, reoperation and mortality rates. The results of multiple logistic regression analyses were presented as odds ratios (OR) and 95 per cent c.i. RESULTS: A total of 6157 patients were analysed to produce two well-balanced groups of 1802 patients: group (A), no abdominal drain(s) and group (B), abdominal drain(s). Group A versus group B showed a significantly lower risk of postoperative duration of stay >6 days (OR 0.60; 95 per cent c.i. 0.51-0.70; P < 0.001). A mean postoperative duration of stay difference of 0.86 days was detected between groups. No difference was recorded between the two groups for all the other endpoints. CONCLUSION: This study confirms that placement of abdominal drain(s) after elective colorectal surgery is associated with a non-clinically significant longer (0.86 days) postoperative duration of stay but has no impact on any other secondary outcomes, confirming that abdominal drains should not be used routinely in colorectal surgery.