Separable effects for adherence.

Comparing different medications is complicated when adherence to these medications differs. We can overcome the adherence issue by assessing effectiveness under sustained use, as in usual causal 'per-protocol' estimands. However, when sustained use is challenging to satisfy in practice, the usefulness of these estimands can be limited. Here we propose a different class of estimands: separable effects for adherence. These estimands compare modified medications, holding fixed a component responsible for non-adherence. Under assumptions about treatment components' mechanisms of effect, a separable effects estimand can quantify the effectiveness of medication initiation strategies on an outcome of interest under the adherence mechanism of one of the medications. These assumptions are amenable to interrogation by subject-matter experts and can be evaluated using causal graphs. We describe an algorithm for constructing causal graphs for separable effects, illustrate how these graphs can be used to reason about assumptions required for identification, and provide semi-parametric weighted estimators.

Estimating the Effect of Bariatric Surgery on Cardiovascular Events Using Observational Data?

BACKGROUND: Observational studies have reported strongly protective effects of bariatric surgery on cardiovascular disease, but with oversimplified definitions of the intervention, eligibility criteria, and follow-up, which deviate from those in a randomized trial. We describe an attempt to estimate the effect of bariatric surgery on cardiovascular disease without introducing these sources of bias, which may not be entirely possible with existing observational data. METHODS: We propose two target trials among persons with diabetes: (1) bariatric operation (vs. no operation) among individuals who have undergone preoperative preparation (lifestyle modifications and screening) and (2) preoperative preparation and a bariatric operation (vs. neither preoperative nor operative component). We emulated both target trials using observational data of US veterans. RESULTS: Comparing bariatric surgery with no surgery (target trial #1; 8,087 individuals), the 7-year cardiovascular risk was 18.0% (95% CI = 6.9, 32.7) in the surgery group and 18.9% (95% CI = 17.7, 20.1) in the no-surgery group (risk difference -0.9, 95% CI = -12.0, 14.0). Comparing preoperative components plus surgery vs. neither (target trial #2; 10,065 individuals), the 7-year cardiovascular risk was 17.4% (95% CI = 13.6, 22.0) in the surgery group and 18.8% (95% CI = 17.8, 19.9) in the no-surgery group (risk difference -1.4, 95% CI = -5.1, 3.2). Body mass index and hemoglobin A1c were reduced with bariatric interventions in both emulations. CONCLUSIONS: Within limitations of available observational data, our estimates do not provide evidence that bariatric surgery reduces cardiovascular disease and support equipoise for a randomized trial of bariatric surgery for cardiovascular disease prevention.

The Prevalence of Sentinel Lymph Node Positivity and Implications for the Utility of Frozen Section Diagnosis Following Neoadjuvant Systemic Therapy in Patients with Clinically Node-Negative HER2-Positive or Triple-Negative Breast Cancer.

BACKGROUND: Axillary dissection is the standard of care for patients with positive sentinel lymph nodes (SLNs) following neoadjuvant systemic therapy. Frozen section can provide intraoperative information regarding the need for axillary dissection during the index operation. However, there are limited data on the utility of frozen section in patients with clinically node-negative (cN0) HER2-positive or triple-negative breast cancer. METHODS: We conducted a single-institution observational cohort study including patients with non-inflammatory, cN0, HER2-positive or triple-negative breast cancer treated with neoadjuvant systemic therapy between 2015 and 2019. We estimated the prevalence of SLN positivity and the diagnostic test characteristics of SLN frozen section. RESULTS: Overall, 662 patients were eligible for inclusion, and 44 patients had one or more positive SLNs (prevalence: 6.6%, 95% confidence interval [CI] 4.9-8.8). There were 490 (74.0%) patients who had intraoperative frozen section, and 19 (3.9%) tested positive among 33 (6.7%) with positive final pathology. Frozen section sensitivity was 57.6% (95% CI 39.2-74.5), specificity was 100% (95% CI 99.2-100), positive predictive value was 100% (95% CI 82.4-100), and negative predictive value was 97.0% (95% CI 95.1-98.4). The sensitivity of frozen section for detection of micrometastases or isolated tumor cells was 35.3% (95% CI 14.2-61.7). CONCLUSION: In patients with cN0 HER2-positive or triple-negative breast cancer who have been treated with neoadjuvant therapy, positive SLNs are uncommon and frozen section sensitivity is modest. Decisions to defer SLN evaluation to final pathology, which may be reasonable in many settings, can be informed, in part, by these findings.

Self-Reported Management of Inflammatory Breast Cancer Among the American Society of Breast Surgeons Membership: Consensus and Opportunities.

BACKGROUND: Inflammatory breast cancer (IBC) is rare and biologically aggressive. We sought to assess diagnostic and management strategies among the American Society of Breast Surgeons (ASBrS) membership. PATIENTS AND METHODS: An anonymous survey was distributed to ASBrS members from March to May 2023. The survey included questions about respondents' demographics and information related to stage III and IV IBC management. Agreement was defined as a shared response by >80% of respondents. In areas of disagreement, responses were stratified by years in practice, fellowship training, and annual IBC patient volume. RESULTS: The survey was administered to 2337 members with 399 (17.1%) completing all questions and defining the study cohort. Distribution of years in practice was 26.0% 0-10 years, 26.6% 11-20 years and 47.4% > 20 years. Overall, 51.2% reported surgical oncology or breast fellowship training, 69.2% maintain a breast-only practice, and 73.5% treat < 5 IBC cases/year. Agreement was identified in diagnostic imaging, trimodal therapy, and mastectomy with wide skin excision for stage III IBC. Lack of agreement was identified in surgical management of the axilla; respondents with < 10 years in practice or fellowship training were more likely to perform axillary dissection for cN0-N2 stage III IBC. Locoregional management of stage IV IBC was variable. CONCLUSIONS: Among ASBrS members, there is consensus in diagnostic evaluation, treatment sequencing and surgical approach to the breast in stage III IBC. Differences exist in surgical management of the cN0-2 axilla with uptake of de-escalation strategies. Clinical trials are needed to evaluate oncologic safety of de-escalation in this high-risk population.

Effect of immediate initiation of invasive ventilation on mortality in acute hypoxemic respiratory failure: a target trial emulation.

PURPOSE: Invasive ventilation is a fundamental treatment in intensive care but its precise timing is difficult to determine. This study aims at assessing the effect of initiating invasive ventilation versus waiting, in patients with hypoxemic respiratory failure without immediate reason for intubation on one-year mortality. METHODS: Emulation of a target trial to estimate the benefit of immediately initiating invasive ventilation in hypoxemic respiratory failure, versus waiting, among patients within the first 48-h of hypoxemia. The eligible population included non-intubated patients with SpO2/FiO2 ≤ 200 and SpO2 ≤ 97%. The target trial was emulated using a single-center database (MIMIC-IV) which contains granular information about clinical status. The hourly probability to receive mechanical ventilation was continuously estimated. The hazard ratios for the primary outcome, one-year mortality, and the secondary outcome, 30-day mortality, were estimated using weighted Cox models with stabilized inverse probability weights used to adjust for measured confounding. RESULTS: 2996 Patients fulfilled the inclusion criteria of whom 792 were intubated within 48 h. Among the non-invasive support devices, the use of oxygen through facemask was the most common (75%). Compared to patients with the same probability of intubation but who were not intubated, intubation decreased the hazard of dying for the first year after ICU admission HR 0.81 (95% CI 0.68-0.96, p = 0.018). Intubation was associated with a 30-day mortality HR of 0.80 (95% CI 0.64-0.99, p = 0.046). CONCLUSION: The initiation of mechanical ventilation in patients with acute hypoxemic respiratory failure reduced the hazard of dying in this emulation of a target trial.

Longitudinal incremental propensity score interventions for limited resource settings.

Many real-life treatments are of limited supply and cannot be provided to all individuals in the population. For example, patients on the liver transplant waiting list usually cannot be assigned a liver transplant immediately at the time they reach highest priority because a suitable organ is not immediately available. In settings with limited supply, investigators are often interested in the effects of treatment strategies in which a limited proportion of patients receive an organ at a given time, that is, treatment regimes satisfying resource constraints. Here, we describe an estimand that allows us to define causal effects of treatment strategies that satisfy resource constraints: incremental propensity score interventions (IPSIs) for limited resources. IPSIs flexibly constrain time-varying resource utilization through proportional scaling of patients' natural propensities for treatment, thereby preserving existing propensity rank ordering compared to the status quo. We derive a simple class of inverse-probability-weighted estimators, and we apply one such estimator to evaluate the effect of restricting or expanding utilization of "increased risk" liver organs to treat patients with end-stage liver disease.

Addressing the longitudinal components of surgical treatments.

Despite their pre- and postoperative components, surgical treatment strategies have typically been cast as point interventions in causal inference research. When longitudinal perioperative components affect outcomes of interest, leaving them unspecified or failing to measure adherence to them complicates the interpretation of effect estimates. Inspired by two recent landmark trials that assessed the risk of stroke or death after transcatheter aortic valve replacement (TAVR) compared with surgical aortic valve replacement (SAVR), the PARTNER 3 trial and the Evolut Low Risk trial, we discuss possible ways that different postoperative therapies in treatment arms and incomplete adherence to those therapies can impact the interpretation of intention-to-treat effect estimates in surgical trials. We argue that surgical treatments are not necessarily point interventions, and make recommendations for improving the design and analysis of trials involving surgical interventions. Central to these recommendations is the need for investigators to specify and report adherence to longitudinal perioperative treatment components.

From Formation to Closure: Aggregate Morbidity and Mortality Associated With Defunctioning Loop Ileostomies.

BACKGROUND: Defunctioning loop ileostomies are used commonly, but there are significant morbidities. OBJECTIVE: This study aimed to describe the morbidity and mortality associated with the formation and closure of defunctioning loop ileostomies. DESIGN: This descriptive study is based on electronic health records and claims data. SETTINGS: This study was conducted at academic and community hospitals in Ontario, Canada. PATIENTS: Adult patients who had a low anterior resection with concurrent defunctioning loop ileostomy from 2002 to 2014 were included. MAIN OUTCOME MEASURES: Outcomes of interest included 30-day major complications, acute kidney injury, transfusion, and deep space infection. The rate of ileostomy reversal and the percentage of permanent ostomies were also collected. RESULTS: The cohort consists of 4658 patients who underwent low anterior resection with concurrent defunctioning loop ileostomy. The 30-day, 90-day, and 1-year mortality rates of these patients were 1.2%, 2.2%, and 5.1%. The rate of reoperation was 5.5%, the rate of hospital readmission was 13.4%, the rate of major complications was 28.5%, the rate of deep organ/space infection requiring percutaneous intervention was 5.2%, and the rate of acute kidney injury requiring hospitalization was 10.4%. Eighty-six percent had their ileostomy reversed, leaving 13.2% with a permanent ostomy. After ileostomy reversal, 30-day and 90-day mortality rates were 0.6% and 0.9%. The rate of major complications was 10.3%, bowel obstruction 7%, ventral hernia 10.5%, deep space infection 1.7%, and repeat operation 2.3%. LIMITATIONS: This study is based on electronic health records and claims data and, thus, the accuracy of results depends on the accuracy of data administration' which can be variable across institutions. CONCLUSIONS: Morbidity and mortality of defunctioning loop ileostomies are significant. One in 8 patients will have a permanent ostomy. See Video Abstract at http://links.lww.com/DCR/B810 . DESDE LA FORMACIN HASTA EL CIERRE AGREGADA MORBILIDAD Y MORTALIDAD ASOCIADA CON LAS ILEOSTOMAS EN ASA DERIVATIVA: ANTECEDENTES:Las ileostomías en asa derivativa se utilizan con frecuencia, pero existen morbilidades importantes.OBJETIVO:Describir la morbilidad y mortalidad asociadas con la formación y cierre de ileostomías en asa derivativa.DISEÑO:Estudio descriptivo basado en historias clínicas electrónicas y datos de reclamaciones.ENTORNO CLINICO:Hospitales académicos y comunitarios en Ontario, Canadá.PACIENTES:Pacientes adultos sometidos a resección anterior baja con concurrente ileostomía en asa derivativa de 2002 a 2014.PRINCIPALES MEDIDAS DE VALORACION:Los resultados de interés incluyeron complicaciones mayores a los 30 días, lesión renal aguda, transfusión e infección del espacio profundo. También se recolectó la tasa de reversión de la ileostomía y el porcentaje de ostomías permanentes.RESULTADOS:La cohorte consistió de 4658 pacientes sometidos a resección anterior baja con concurrente ileostomía en asa derivativa. La mortalidad de estos pacientes, a treinta días, 90 días y un año, fue del 1,2%, 2,2% y 5,1%, respectivamente. La tasa de reintervención fue del 5,5%, el reingreso hospitalario fue del 13,4%, la complicación mayor fue del 28,5%, la infección profunda de órganos / espacios que requirieron intervención percutánea fue del 5,2%, y la lesión renal aguda que requirió hospitalización fue del 10,4%. Ochenta y seis por ciento tuvieron reversión de su ileostomía, dejando al 13.2% con una ostomía permanente. Después de la reversión de la ileostomía, la mortalidad a los 30 días y 90 días fue de 0,6% y 0,9%, respectivamente. La tasa de complicaciones mayores fue del 10,3%, obstrucción intestinal del 7%, hernia ventral del 10,5%, infección del espacio profundo del 1,7% y reintervención del 2,3%.LIMITACIONES:El estudio se basa en registros médicos electrónicos y datos de reclamos y, por lo tanto, la precisión de los resultados depende de la precisión en la administración de datos, que pueden variar entre instituciones.CONCLUSIONES:La morbilidad y la mortalidad de las ileostomías en asa derivativa son significativas. Uno de cada 8 pacientes tendrá una ostomía permanente. Consulte Video Resumen en http://links.lww.com/DCR/B810 . (Traducción-Dr. Fidel Ruiz Healy ).

Colonic displacement as a marker of endoscopic skill: development of a novel tool for endoscopy training.

BACKGROUND: Colonoscopy is a technically challenging procedure. The colonoscope is prone to forming loops in the colon, which can lead patient discomfort and even perforation. We hypothesized that expert endoscopists use techniques to avoid loop formation, identify and straighten loops earlier, and thus exert less force. METHODS: Using a commercially available physical colon simulator model (Kyoto Kagaku), electromagnetic tracking markers (NDI Medical) were placed along the mobile segments of the colon (sigmoid, transverse) to measure the degree of displacement of the colon as the scope was advanced to the cecum. The colon model was set for each participant to simulate a redundant alpha loop in the sigmoid colon. Gastroenterology and surgical trainees and attendings were assessed. Demographic data were collected for each participant. RESULTS: Seventy-five participants were enrolled in the study. There were 17 (22.7%) attending physicians, and 58 (77.3%) trainees. Attending physicians advanced the scope to the cecum faster. The mean time required for procedure completion was 360.5 s compared to 178.4 s for the trainee and attending groups respectively (mean difference: 182.1 s, 95% CI: 93.0, 269.7; p = 0.0002). Attending physicians exerted significantly lower mean colonic displacement than trainees. The mean colonic displacement was 79.8 mm for the trainee group and 57.9 mm for the attending group (mean difference: 21.9 mm, 95% CI: 2.6, 41.2; p = 0.04). Those who used torque steering caused lower maximum colonic displacement than those who used knob steering. CONCLUSION: Attending physicians advance the scope during colonoscopy in a manner that results in significantly less colonic displacement than resident trainees. Although prior studies have shown a difference in force application between endoscopists and inexperienced students, ours is the first to differentiate across varying degrees of endoscopic skill. Future studies will define metrics for incorporation into endoscopic training curricula, focusing on techniques that encourage safety and comfort for patients.

Colonoscopy Needs for Implementation of a Colorectal Cancer Screening Program in Ukraine.

BACKGROUND: In Ukraine, there is no established colorectal cancer screening program. We aimed to project the number of screening colonoscopies needed for implementation of various CRC screening strategies in Ukraine. METHODS: We modified a previously developed Markov microsimulation model to reflect the natural history of adenoma and CRC progression among average-risk 50-74-year-olds. We simulated colonoscopies needed for the following screening strategies: no screening, fecal occult blood test yearly, FOBT yearly with flexible sigmoidoscopy every 5 years, FS every 5 years, fecal immunohistochemistry test (FIT) yearly, or colonoscopy every 10 years. Assuming 80% screening adherence, we estimated colonoscopies required at 1 and 5 years depending on the implementation rate. In one-way sensitivity analyses, we varied implementation rate, screening adherence, sensitivity, and specificity. RESULTS: Assuming an 80% screening adherence and complete implementation (100%), besides a no screening strategy, the fewest screening colonoscopies are needed with an FOBT program, requiring on average 6,600 and 26,800 colonoscopies per 100,000 persons at 1 and 5 years post-implementation, respectively. The most screening colonoscopies are required with a colonoscopy program, requiring on average 76,600 and 101,000 colonoscopies per 100,000 persons at 1 and 5 years post-implementation, respectively. In sensitivity analyses, the biggest driver of number of colonoscopies needed was screening adherence. CONCLUSIONS: The number of colonoscopies needed and therefore the potential strain on the healthcare system vary substantially by screening test. These findings can provide valuable information for stakeholders on equipment needs when implementing a national screening program in Ukraine.

Strengthening Health Services Research Using Target Trial Emulation: An Application to Volume-Outcomes Studies.

The number of operations that surgeons have previously performed is associated with their patients' outcomes. However, this association may not be causal, because previous studies have often been cross-sectional and their analyses have not considered time-varying confounding or positivity violations. In this paper, using the example of surgeons who perform coronary artery bypass grafting, we describe (hypothetical) target trials for estimation of the causal effect of the surgeons' operative volumes on patient mortality. We then demonstrate how to emulate these target trials using data from US Medicare claims and provide effect estimates. Our target trial emulations suggest that interventions on physicians' volume of coronary artery bypass grafting operations have little effect on patient mortality. The target trial framework highlights key assumptions and draws attention to areas of bias in previous observational analyses that deviated from their implicit target trials. The principles of the presented methodology may be adapted to other scenarios of substantive interest in health services research.

Association and interplay of surgeon and hospital volume with mortality after open abdominal aortic aneurysm repair in the modern era.

OBJECTIVE: Operative volume has been used as a marker of quality. Research from previous decades has suggested minimum open abdominal aortic aneurysm (AAA) repair volume requirements for surgeons of 9 to 13 open AAA repairs annually and for hospitals of 18 open AAA repairs annually to purportedly achieve acceptable results. Given concerns regarding the decreased frequency of open repairs in the endovascular era, we examined the association of surgeon and hospital volume with the 30- and 90-day mortality in the Vascular Quality Initiative (VQI) registry. METHODS: Patients who had undergone elective open AAA repair from 2013 to 2018 were identified in the VQI registry. We performed a cross-sectional evaluation of the association between the average hospital and surgeon volume and 30-day postoperative mortality using a hierarchical Bayesian model. Cross-level interactions were permitted, and random surgeon- and hospital-level intercepts were used to account for clustering. The mortality results were adjusted by standardizing to the observed distribution of relevant covariates in the overall cohort. The outcomes were compared to the Society for Vascular Surgery guidelines recommended criteria of <5% perioperative mortality. RESULTS: A total of 3078 patients had undergone elective open AAA repair by 520 surgeons at 128 hospitals. The 30- and 90-day risks of postoperative mortality were 4.1% (n = 126) and 5.4% (n = 166), respectively. The mean surgeon volume and hospital volume both correlated inversely with the 30-day mortality. Averaged across all patients and hospitals, we found a 96% probability that surgeons who performed an average of four or more repairs per year achieved <5% 30-day mortality. Substantial interplay was present between surgeon volume and hospital volume. For example, at lower volume hospitals performing an average of five repairs annually, <5% 30-day mortality would be expected 69% of the time for surgeons performing an average of three operations annually. In contrast, at higher volume hospitals performing an average of 40 repairs annually, a <5% 30-day mortality would be expected 96% of the time for surgeons performing an average of three operations annually. As hospital volume increased, a diminishing difference occurred in 30-day mortality between lower and higher volume surgeons. Likewise, as surgeon volume increased, a diminishing difference was found in 30-day mortality between the lower and higher volume hospitals. CONCLUSIONS: Surgeons and hospitals in the VQI registry achieved mortality outcomes of <5% (Society for Vascular Surgery guidelines), with an average surgeon volume that was substantially lower compared with previous reports. Furthermore, when considering the development of minimal surgeon volume guidelines, it is important to contextualize the outcomes within the hospital volumes.

Estimating the effect of increasing utilization of living donor liver transplantation using observational data.

There has been a recent increase in enthusiasm for expansion of living donor liver transplantation (LDLT) programmes. Using all adults initially placed on the waiting list in the United States, we estimated the risk of overall mortality under national strategies which differed in their utilization of LDLT. We used a generalization of inverse probability weighting which can estimate the effect of interventions in the setting of finite resources. From 2005 to 2015, 93 812 eligible individuals were added to the waitlist: 51 322 received deceased donor grafts while 1970 underwent LDLT. Individuals who underwent LDLT had more favourable prognostic factors, including lower mean MELD score at transplant (14.6 vs. 20.5). The 1-year, 5-year and 10-year cumulative incidence of death under the current level of LDLT utilization were 18.0% (95% CI: 17.8, 18.3%), 41.2% (95% CI: 40.8, 41.5%) and 57.4% (95% CI: 56.9, 57.9%) compared to 17.9% (95% CI: 17.7, 18.2%), 40.6% (95% CI: 40.2, 40.9%) and 56.4% (95% CI: 55.8, 56.9%) under a strategy which doubles LDLT utilization. Expansion of LDLT utilization would have a measurable, modest effect on the risk of mortality for the entire cohort of individuals who begin on the transplant waiting list.

Variation in complications and mortality following ALPPS at early-adopting centers.

BACKGROUND: Various, often conflicting, estimates for post-operative morbidity and mortality following ALPPS have been reported in the literature, suggesting that considerable center-level variation exists. Some of this variation may be related to center volume and experience. METHODS: Using data from seventeen centers who were early adopters of the ALPPS technique, we estimated the variation, by center, in standardized 90-day mortality and comprehensive complication index (CCI) for patients treated between 2012 and 2018. RESULTS: We estimated that center-specific 90-day mortality following treatment with ALPPS varied from 4.2% (95% CI: 0.8, 9.9) to 29.1% (95% CI: 13.9, 50.9), and that center-specific CCI following treatment with ALPPS varied from 17.0 (95% CI: 7.5, 26.5) to 49.8 (95% CI: 38.1, 61.8). Declines in estimated 90-day mortality and CCI were observed over time, and almost all individual centers followed this trend. Patients treated at centers with a higher number of ALPPS cases performed over the prior year had a lower risk of post-operative mortality. CONCLUSION: Despite considerable center-level variation in ALPPS outcomes, perioperative outcomes following ALPPS have improved over time and treatment at higher volume centers results in a lower risk of 90-day mortality. Morbidity and mortality remain concerningly high at some centers.

Opioid Prescription Patterns for Children Following Laparoscopic Appendectomy.

OBJECTIVE: To describe variability in and consequences of opioid prescriptions following pediatric laparoscopic appendectomy. SUMMARY BACKGROUND DATA: Postoperative opioid prescribing patterns may contribute to persistent opioid use in both adults and children. METHODS: We included children <18 years enrolled as dependents in the Military Health System Data Repository who underwent uncomplicated laparoscopic appendectomy (2006-2014). For the primary outcome of days of opioids prescribed, we evaluated associations with discharging service, standardized to the distribution of baseline covariates. Secondary outcomes included refill, Emergency Department (ED) visit for constipation, and ED visit for pain. RESULTS: Among 6732 children, 68% were prescribed opioids (range = 1-65 d, median = 4 d, IQR = 3-5 d). Patients discharged by general surgery services were prescribed 1.23 (95% CI = 1.06-1.42) excess days of opioids, compared with those discharged by pediatric surgery services. Risk of ED visit for constipation (n = 61, 1%) was increased with opioid prescription [1-3 d, risk ratio (RR) = 2.46, 95% CI = 1.31-5.78; 4-6 d, RR = 1.89, 95% CI = 0.83-4.67; 7-14 d, RR = 3.75, 95% CI = 1.38-9.44; >14 d, RR = 6.27, 95% CI = 1.23-19.68], compared with no opioid prescription. There was similar or increased risk of ED visit for pain (n = 319, 5%) with opioid prescription [1-3 d, RR = 1.00, 95% confidence interval (CI) = 0.74-1.32; 4-6 d, RR = 1.31, 95% CI = 0.99-1.73; 7-14 d, RR = 1.52, 95% CI = 1.00-2.18], compared with no opioid prescription. Likewise, need for refill (n = 157, 3%) was not associated with initial days of opioid prescribed (reference 1-3 d; 4-6 d, RR = 0.96, 95% CI = 0.68-1.35; 7-14 d, RR = 0.91, 95% CI = 0.49-1.46; and >14 d, RR = 1.22, 95% CI = 0.59-2.07). CONCLUSIONS: There was substantial variation in opioid prescribing patterns. Opioid prescription duration increased risk of ED visits for constipation, but not for pain or refill.

A Graphical Description of Partial Exchangeability.

Partial exchangeability is sufficient for the identification of some causal effects of interest. Here we review the use of common graphical tools and the sufficient component cause model in the context of partial exchangeability. We illustrate the utility of single world intervention graphs (SWIGs) in depicting partial exchangeability and provide an illustrative example of when partial exchangeability might hold in the absence of complete exchangeability.

Adjusting for adherence in randomized trials when adherence is measured as a continuous variable: An application to the Lipid Research Clinics Coronary Primary Prevention Trial.

BACKGROUND: Clinicians and patients may be more interested in per-protocol effect estimates than intention-to-treat effect estimates from randomized trials. However, per-protocol effect estimates may be biased due to insufficient adjustment for prognostic factors that predict adherence. Adjustment for this bias is possible when appropriate methods, such as inverse probability weighting, are used. But, when adherence is measured as a continuous variable, constructing these weights can be challenging. METHODS: In the placebo arm of the Lipid Research Clinics Coronary Primary Prevention Trial, we estimated the 7-year cumulative incidence of coronary heart disease under 100% adherence and 0% adherence to placebo. We used dose-response discrete-hazards models with inverse probability weighting to adjust for pre- and post-randomization covariates. We considered several continuous distributions for constructing the inverse probability weights. RESULTS: The risk difference estimate for 100% adherence compared with 0% adherence ranged from -7.7 to -6.1 percentage points without adjustment for baseline and post-baseline covariates, and ranged from -1.8 to 2.2 percentage points with adjustment using inverse probability weights, depending on the dose-response model and inverse probability weight distribution used. CONCLUSIONS: Methods which appropriately adjust for time-varying post-randomization variables can explain away much of the bias in the "effect" of adherence to placebo. When considering continuous adherence, investigators should consider several models as estimates may be sensitive to the model chosen.