Results of a multicenter cluster-randomized controlled clinical trial testing the effectiveness of a bioinformatics-enabled pharmacist intervention in transplant recipients.

An ambulatory medication safety dashboard was developed to identify missing labs, concerning labs, drug interactions, nonadherence, and transitions in care. This system was tested in a 2-year, prospective, cluster-randomized, controlled multicenter study. Pharmacists at 5 intervention sites used the dashboard to address medication safety issues, compared with usual care provided at 5 control sites. A total of 2196 transplant events were included (1300 intervention vs 896 control). During the 2-year study, the intervention arm had a 11.3% (95% confidence interval, 7.1%-15.5%) absolute risk reduction of having ≥1 emergency department (ED) visit (44.2% vs 55.5%, respectively; P < .001, respectively) and a 12.3% (95% confidence interval, 8.2%-16.4%) absolute risk reduction of having ≥1 hospitalization (30.1% vs 42.4%, respectively; P < .001). In those with ≥1 event, the median ED visit rate (2 [interquartile range (IQR) 1, 5] vs 2 [IQR 1, 4]; P = .510) and hospitalization rate (2 [IQR 1, 3] vs 2 [IQR 1, 3]; P = .380) were similar. Treatment effect varied by comorbidity burden, previous ED visits or hospitalizations, and heart or lung recipients. A bioinformatics dashboard-enabled, pharmacist-led intervention reduced the risk of having at least one ED visit or hospitalization, predominantly demonstrated in lower risk patients.

Airflow limitation and mortality during cancer screening in the National Lung Screening Trial: why quantifying airflow limitation matters.

IMPORTANCE: Current eligibility criteria for lung cancer (LC) screening are derived from randomised controlled trials and primarily based on age and smoking history. However, the individual benefits of screening are highly variable and potentially attenuated by co-morbidities such as advanced airflow limitation (AL). OBJECTIVE: To examine the relationship between the presence and severity of AL and screening outcomes. METHODS: This was a secondary analysis of 18 463 high-risk smokers, a substudy from the National Lung Screening Trial, who underwent pre-bronchodilator spirometry at baseline and median follow-up of 6.1 years. We used descriptive statistics and a competing risk proportional hazards model to examine differences in screening outcomes by chronic obstructive pulmonary disease severity group. RESULTS: The risk of developing LC increased with worsening AL (effect size=0.34, p<0.0001), as did the risk of dying of LC (effect size=0.35, p<0.0001). While those with severe AL (Global Initiative for Obstructive Lung Disease, GOLD grade 3-4) had the highest risk of LC and the highest LC mortality, they also had fewer adenocarcinomas (effect size=-0.20, p=0.008) and a lower surgery rate (effect size=-0.16, p=0.014) despite comparable staging, and greater non-LC mortality relative to LC mortality (effect size=0.30, p<0.0001). In participants with no AL, screening with CT was associated with a significant reduction in LC deaths relative to chest X-ray (30.3%, 95% CI 4.5% to 49.2%, p<0.05). The clinically relevant but attenuated reduction in those with AL (18.5%, 95% CI -8.4% to 38.7%, p>0.05) could be attributed to GOLD 3-4, where no appreciable mortality reduction was observed. CONCLUSION: Despite a greater risk of LC, severe AL was not associated with any apparent reduction in LC mortality following screening.

Surgical prescription opioid trajectories among state Medicaid enrollees.

OBJECTIVE: The objective of this study was to evaluate opioid use trajectories among a sample of 10,138 Medicaid patients receiving one of six index surgeries: lumbar spine, total knee arthroplasty, cholecystectomy, appendectomy, colon resection, and tonsillectomy. DESIGN: Retrospective cohort. SETTING: Administrative claims data. PATIENTS AND PARTICIPANTS: Patients, aged 13 years and older, with 15-month continuous Medicaid eligibility surrounding index surgery, were selected from single-state Medicaid medical and pharmacy claims data for surgeries performed between 2014 and 2017. INTERVENTIONS: None. MAIN OUTCOME MEASURES: Baseline comorbidities and presurgery opioid use were assessed in the 6 months prior to admission, and patients' opioid use was followed for 9 months post-discharge. Generalized linear model with log link and Poisson distribution was used to determine risk of chronic opioid use for all risk factors. Group-based trajectory models identified groups of patients with similar opioid use trajectories over the 15-month study period. RESULTS: More than one in three (37.7 percent) patients were post-surgery chronic opioid users, defined as the dichotomous outcome of filling an opioid prescription 90 or more days after surgery. Key variables associated with chronic post-surgery opioid use include presurgery opioid use, 30-day post-surgery opioid use, and comorbidities. Latent trajectory modeling grouped patients into six distinct opioid use trajectories. Associates of trajectory group membership are reported. CONCLUSIONS: Findings support the importance of surgeons setting realistic patient expectations for post-surgical opioid use, as well as the importance of coordination of post-surgical care among patients failing to fully taper off opioids within 1-3 months of surgery.

Incidence and patterns of persistent opioid use in children following appendectomy.

BACKGROUND: The past 5 years have witnessed a concerted national effort to assuage the rising tide of the opioid misuse in our country. Surgical procedures often serve as the initial exposure of children to opioids, however the trajectory of use following these exposures remains unclear. We hypothesized that opioid exposure following appendectomy would increase the risk of persistent opioid use among publicly insured children. STUDY DESIGN: A retrospective longitudinal cohort study was conducted on South Carolina Medicaid enrollees who underwent appendectomy between January 2014 and December 2017 using administrative claims data. The primary outcome was chronic opioid use. Generalized linear models and finite mixture models were employed in analysis. RESULTS: 1789 Medicaid pediatric patients underwent appendectomy and met inclusion criteria. The mean age was 11.1 years and 40.6% were female. Most patients (94.6%) did not receive opioids prior to surgery. Opioid prescribing ≥90 days after surgery (chronic opioid use) occurred in 127 (7.1%) patients, of which 102 (80.3%) had no opioid use in the preexposure period. Risk factors for chronic opioid use included non-naïve opioid status, re-hospitalization more than 30 days following surgery, multiple opioid prescribers, age, and multiple antidepressants/antipsychotic prescriptions. Group-based trajectory analysis demonstrated four distinct post-surgical opioid use patterns: no opioid use (91.3%), later use (6.7%), slow wean (1.9%), and higher use throughout (0.4%). CONCLUSION: Opioid exposure after appendectomy may serve as a priming event for persistent opioid use in some children. Eighty percent of children who developed post-surgical persistent opioid use had not received opioids in the 90 days leading up to surgery. Several mutable and immutable factors were identified to target future efforts toward opioid minimization in this at-risk patient population. LEVEL OF EVIDENCE: III.

New chronic opioid use in Medicaid patients following cholecystectomy.

Background: Commercial insurance data show that chronic opioid use in opioid-naive patients occurs in 1.5% to 8% of patients undergoing surgical procedures, but little is known about patients with Medicaid. Methods: Opioid prescription data and medical coding data from 4,788 Medicaid patients who underwent cholecystectomy were analyzed to determine opioid use patterns. Results: A total of 54.4% of patients received opioids prior to surgery, and 38.8% continued to fill opioid prescriptions chronically; 27.1% of opioid-naive patients continued to get opioids chronically. Patients who received ≥ 50 MME/d had nearly 8 times the odds of chronic opioid use. Each additional opioid prescription filled within 30 days was associated with increased odds of chronic use (odds ratio: 1.71). Conclusion: Opioid prescriptions are common prior to cholecystectomy in Medicaid patients, and 38.8% of patients continue to receive opioid prescriptions well after surgical recovery. Even 27.1% of opioid-naive patients continued to receive opioid prescriptions chronically.

Patterns of dispensed opioids after tonsillectomy in children and adolescents in South Carolina, United States, 2010-2017.

OBJECTIVES: Tonsillectomy (with or without adenoidectomy) is a common pediatric surgical procedure requiring post-operative analgesia. Because of the respiratory depression effects of opioids, clinicians strive to limit the use of these drugs for analgesia post-tonsillectomy. The objective of this study was to identify demographic and medication use patterns predictive of persistent opioid dispensing (as a proxy for opioid use) to pediatric patients post-tonsillectomy. PATIENTS AND METHODS: Retrospective cohort of South Carolina (USA) Medicaid-insured children and adolescents 0-18 years old without malignancy who had tonsillectomy in 2014-2017. We evaluated opioid dispensing pre-surgery and in the 30 days exposure period after hospital discharge. The main outcome, persistent opioid dispensing, was defined as any subject dispensed ≥1 opioid prescription 90-270 days after discharge. Group-based trajectory analyses described post-procedure opioid dispensing trajectories. RESULTS: There were 11,578 subjects representing 12,063 tonsillectomy procedures. Few (3.5%) procedures were followed by persistent opioid dispensing. Any opioid dispensing during the exposure period was associated with an increased odds of persistent opioid dispensing status during the follow up period (OR 1.51 for 1-6 days of exposure and OR 1.65 for 7-30 days of opioid exposure), as was pre-procedure opioid dispensing, having >1 tonsillectomy procedure, and having complex chronic medical conditions. Group-based trajectory analyses identified 4 distinct patterns of post-discharge opioid dispensing. CONCLUSIONS: Any opioid dispensing during the 30 days after tonsillectomy increased the odds of persistent opioid dispensing by > 50%. Multivariable and group-based trajectory analyses identified patient and procedure variables that correlate with persistent opioid dispensing, primarily driven by groups receiving pre-tonsillectomy opioids and a second group who experienced multiple episodes of tonsillectomy.

Impact of Race and Location of Residence on Statin Treatment Among Veterans With Type 2 Diabetes Mellitus.

Rural residence and ethnic-minority status are individually associated with increased cardiovascular (CV) mortality. Statin therapy is known to reduce the risk of cardiovascular mortality. Although ethnic disparities in statin treatment exist, the joint impact of urban/rural residence and race/ethnicity on statin prescribing is unclear. Veterans Health Administration (VHA) and Centers for Medicare and Medicaid data were used to perform a longitudinal study of Veterans with Type 2 diabetes mellitus from 2007 to 2016. Mixed effects logistic regression with a random intercept was used to model the longitudinal association between the primary exposure (race/ethnicity and residence) and statin prescribing. After adjusting for covariates, non-Hispanic White (NHW)-Rural Veterans were 7% (odds ratio [OR] = 1.07; confidence interval [CI] 1.05 to 1.08), non-Hispanic Black (NHB)-Rural Veterans were 4% (OR 1.04; CI 1.00 to 1.08), and Hispanic-Urban Veterans were 20% (OR 1.20; CI 1.17 to 1.23) more likely to be prescribed statins versus NHW-Urban Veterans; whereas, NHB-Urban Veterans were 14% (OR 0.86; CI 0.85 to 0.55) and Hispanic-Rural Veterans were 10% (OR 0.90; CI 0.85 to 0.96) less likely. When disability and dual use were removed from the full model, compared with NHW-Urban, the odds of statin prescribing in NHW-Rural Veterans remained unchanged (OR 1.06; CI 1.04 to 1.07) whereas the odds of statin prescribing in all other groups were higher. In conclusion, NHB-Urban and Hispanic-Rural Veterans had lower odds of statin prescribing versus NHW-Urban Veterans; whereas NHW-Rural, NHB-Rural and Hispanic-Urban Veterans had higher odds. The findings in ethnic-minorities changed when we accounted for markers of VHA care (i.e., disability, dual use) showing that these individuals are more likely to receive statins when they receive more VHA care.

Approaches for missing covariate data in logistic regression with MNAR sensitivity analyses.

Data with missing covariate values but fully observed binary outcomes are an important subset of the missing data challenge. Common approaches are complete case analysis (CCA) and multiple imputation (MI). While CCA relies on missing completely at random (MCAR), MI usually relies on a missing at random (MAR) assumption to produce unbiased results. For MI involving logistic regression models, it is also important to consider several missing not at random (MNAR) conditions under which CCA is asymptotically unbiased and, as we show, MI is also valid in some cases. We use a data application and simulation study to compare the performance of several machine learning and parametric MI methods under a fully conditional specification framework (MI-FCS). Our simulation includes five scenarios involving MCAR, MAR, and MNAR under predictable and nonpredictable conditions, where "predictable" indicates missingness is not associated with the outcome. We build on previous results in the literature to show MI and CCA can both produce unbiased results under more conditions than some analysts may realize. When both approaches were valid, we found that MI-FCS was at least as good as CCA in terms of estimated bias and coverage, and was superior when missingness involved a categorical covariate. We also demonstrate how MNAR sensitivity analysis can build confidence that unbiased results were obtained, including under MNAR-predictable, when CCA and MI are both valid. Since the missingness mechanism cannot be identified from observed data, investigators should compare results from MI and CCA when both are plausibly valid, followed by MNAR sensitivity analysis.

Racial-Geographic Disparity in Lipid Management in Veterans with Type 2 Diabetes: A 10-Year Retrospective Cohort Study.

Purpose: The prevalence of diabetes in U.S. veterans (20.5%) is nearly three times that of the general population. Minority veterans have higher rates of diabetes compared with their counterparts and urban/rural residence is also associated with uncontrolled cholesterol. However, the interplay between urban/rural residence and race/ethnicity on cholesterol control is unclear. Methods: Veterans Health Administration Corporate Data Warehouse and Centers for Medicare and Medicaid data were used to create unique dataset and perform longitudinal study of veterans with type 2 diabetes from 2006 to 2016. Logistic regression was used to model the association between low-density lipoprotein (LDL) control and the primary exposures (race/ethnicity and location of residence) after adjusting for all measured covariates, including the interaction between location of residence and race/ethnicity. Results: There was a significant interaction between race/ethnicity and rural residence. Rural non-Hispanic Black (NHB) veterans had higher odds for LDL >100 mg/dL (odds ratio [OR]=1.70, 95% confidence interval [CI] 1.50-1.60) and for LDL >70 mg/dL (OR=1.59, 95% CI 1.53-1.64) compared with urban non-Hispanic White (NHW) veterans. Similarly, compared with urban NHW, urban NHB veterans had higher odds of LDL >100 mg/dL (OR=1.45, 95% CI 1.43-1.47) and LDL >70 mg/dL (OR=1.36, 95% CI 1.34-1.38). Conclusion: This study highlights health disparities for veterans with type 2 diabetes. Future research is needed to evaluate interventions for mitigating these disparities in cholesterol management among veterans with diabetes.

Impact of Tobacco Dependence in Risk Prediction Models for Lung Cancer Diagnoses and Deaths.

BACKGROUND: Stronger nicotine dependence is associated with greater lung cancer incidence and lung cancer death. This study investigates whether including nicotine dependence in risk prediction models for lung cancer incidence and mortality provides any important clinical benefits. METHODS: Smoking data were used from 14 123 participants in the American College of Radiology Imaging Network arm of the National Lung Screening trial. We added nicotine dependence as the primary exposure in two published lung cancer risk prediction models (Katki-Gu or PLCO-m2012) and compared four results: with no tobacco-dependence measure, with time to first cigarette, with heaviness of smoking index, and with Fagestrom test for nicotine dependence. We used a cross-validation method based on leave-one-out and compared performance using likelihood ratio tests (LRT), area under the curve, concordance, sensitivity and specificity for 1% and 2% risk thresholds, and net benefit statistics. Statistical tests were two-sided. RESULTS: All LRT results were statistically significant (P ≤ .0001), whereas other tests were not, except that specificity statistically significantly improved (P < .0001). Because the LRT is asymptotically more powerful for testing for prediction gain, we conclude that both models were improved on a statistical level by adding dependence measures. The other performance statistics generally indicated that such gains were likely very small. Net benefit analysis confirmed there was no apparent clinical benefit for including dependence measures. CONCLUSIONS: Although inclusion of dependence measures may not provide a clinical benefit when added to risk prediction models, nicotine-dependence measures should nonetheless be an integral tool for patient counseling and for encouraging tobacco cessation.

Creating a 13-year National Longitudinal Cohort of veterans with chronic kidney disease.

BACKGROUND: The development of large-scale chronic kidney disease (CKD) cohorts within the Veterans Affairs (VA) system has been limited by several factors, including the high proportion of missing race data etc. The goal of this study is to address the limitations of prior studies by creating a large cohort utilizing robust KDIGO recommendations for identifying and staging CKD. METHODS: Multiple patient and administrative files from the Veterans Health Administration (VHA) National Patient Care were linked to create a national cohort of Veterans with chronic kidney disease (CKD) between January 2000 - December 2012; patients identified during this period were followed until 2015. CKD was defined for stages 1 through 5 if markers of kidney damage, specifically proteinuria, were present for at least 3 months. Estimated glomerular filtration rate (eGFR) values were calculated based on serum creatinine levels and the patient's age, gender, and race using both the Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formulas. RESULTS: About 50 million observations were collected that supported a CKD diagnosis during the study period; these observations corresponded to 3,051,001 unique veterans; 80.9% were non-Hispanic white (NHW), 13.4% were non-Hispanic black (NHB), 3.6% were Hispanic, and 2.0% were in other groups. The mean age 76.7, about 97% were male and 50.2% died prior to January 2016. Among those with stage 3, 12.3% progressed to stage 4, 21.6% of those with stage 4 progressed to stage 5. We found that eGFR values calculated from serum creatinine levels identified about 98% of all patients, while about 11.4% of patients could be identified through ICD-9 codes; only 6.4% could be identified through both sources. CONCLUSION: This 13-year national cohort provides an important resource for answering numerous research questions in the future such as racial/ethnic disparities questions, tracking health service utilization, medication adherence, cost and health outcomes in veterans with CKD.

Investigating the Potential for Bias When Using a Widely Accepted Medication Adherence Measure to Predict Mortality.

STUDY OBJECTIVE: Summary measures of medication adherence, such as the proportion of days covered (PDC), are often used to analyze the association between medication adherence and various health outcomes. We hypothesized that PDC and similar measures may lead to biased results in some situations when used to estimate the association between adherence and the outcome event (e.g., mortality). Thus, the objective was to determine the conditions under which PDC and similar measures might produce biased estimates of the association between adherence and mortality and to review methods to avoid such bias. DESIGN: Simulation study and analysis of data from a large retrospective cohort study. MEASUREMENTS AND MAIN RESULTS: We conducted a comprehensive simulation to compare how adherence estimates varied-using prescription-based (end point was final date that medications were on hand or date of death) and interval-based (end point based on fixed calendar interval or date of death) PDC denominators-when deaths occurred either during or after the adherence exposure period. We then made similar comparisons using data from a retrospective study that included comprehensive measures of medication refill activity and clinical outcomes of 207,841 patients with diabetes mellitus who were prescribed one or more oral antidiabetic medications. When deaths occurred within the adherence exposure period, substantial bias in adherence estimates was possible regardless of the PDC denominator type. CONCLUSION: Investigators using PDC or similar proxy measures should carefully consider the temporal relationship between adherence exposure and clinical outcomes when the outcome event affects the adherence measurement.

Ethnic and geographic variations in multimorbidty: Evidence from three large cohorts.

A common characteristic of patients seen at the Veterans Health Administration (VHA) is a high number of concurrent comorbidities (i.e. multimorbidity). This study (i) examines the magnitude and patterns of multimorbidity by race/ethnicity and geography; (ii) compares the level of variation explained by these factors in three multimorbidity measures across three large cohorts. We created three national cohorts for Veterans with chronic kidney disease (CKD:n = 2,190,564), traumatic brain injury (TBI:n = 167,954) and diabetes-mellitus (DM:n = 1,263,906). Multimorbidity was measured by Charlson-Deyo, Elixhauser and Walraven-Elixhauser scores. Multimorbidity differences by race/ethnicity and geography were compared using generalized linear models (GLM). Latent class analysis (LCA) was used to identify groups of conditions that are highly associated with race/ethnic groups. Differences in age (CKD,74.5, TBI,49.7, DM, 66.9 years), race (CKD,80.9%, TBI,76.4%, DM, 63.8% NHW) and geography (CKD,64.4%, TBI,70%, DM, 70.9% urban) were observed among the three cohorts. Accounting for these differences, GLM results showed that risk of multimorbidity in non-Hispanic blacks (NHB) with CKD were 1.16 times higher in urban areas and 1.10 times higher in rural areas compared to non-Hispanic whites (NHW) with CKD. DM and TBI showed similar results with risk for NHB, 1.05 higher in urban areas and 0.97 lower in rural areas for both diseases. Overall, our results show that (i) multimorbidity risk was higher for NHB in urban areas compared to rural areas in all three cohorts; (ii) multimorbidity risk was higher for Hispanics in urban areas compared to rural areas in the DM and CKD cohorts; and (iii) the highest overall multimorbidity risk of any race group or location exists for Hispanics in insular islands for all three disease cohorts. These findings are consistent among the three multimorbidity measures. In fact, our LCA also showed that a three class LC model based on Elixhauser or Charlson provides good discrimination by type and extent of multimorbidity.

Conscious Perception of Facial Asymmetry in a Unilateral Cleft Lip Model.

OBJECTIVE: Lip asymmetry after a unilateral cleft lip repair can be perceived as an unsatisfactory result. The objective of this study is to determine the degree of upper lip asymmetry and/or nasal alar hooding required for recognition of asymmetry in a simulated model of unilateral cleft lip. DESIGN: A model of unilateral cleft lip was created using digital morphing software to simulate asymmetries in vermilion height and nasal hooding in photographs of children. Volunteers were shown photographs for different time intervals and with varying degrees of asymmetry. Ability to detect facial asymmetry was recorded and analyzed. SETTING: This study was conducted by surveying layperson volunteers in public community settings. PARTICIPANTS: 108 layperson volunteers were randomly surveyed. MAIN OUTCOME MEASURES: The primary outcome measure was a reported lip or nose asymmetry by the volunteers. Proportions and corresponding 95% confidence intervals were obtained to estimate the probability of reporting an asymmetry at 3- and 10-second intervals. RESULTS: After 3- and 10-second exposure, labial asymmetry was perceived by ≥50% of subjects at 2 mm (62%, P = .001) and 1 mm (89%, P < .0001), respectively. Nasal asymmetry was detected by <50% of subjects at 3 seconds, but ≥50% perceived a 3-mm alteration at 10 seconds (64%, P < .0001). Photographs with combined nasal and labial modification did not lower the threshold for asymmetry perception compared to either deformity alone. CONCLUSIONS: This study is the first to determine a predictable millimeter threshold for perceived asymmetry in cleft lip deformity using a digital model.

Genes associated with SLE are targets of recent positive selection.

The reasons for the ethnic disparities in the prevalence of systemic lupus erythematosus (SLE) and the relative high frequency of SLE risk alleles in the population are not fully understood. Population genetic factors such as natural selection alter allele frequencies over generations and may help explain the persistence of such common risk variants in the population and the differential risk of SLE. In order to better understand the genetic basis of SLE that might be due to natural selection, a total of 74 genomic regions with compelling evidence for association with SLE were tested for evidence of recent positive selection in the HapMap and HGDP populations, using population differentiation, allele frequency, and haplotype-based tests. Consistent signs of positive selection across different studies and statistical methods were observed at several SLE-associated loci, including PTPN22, TNFSF4, TET3-DGUOK, TNIP1, UHRF1BP1, BLK, and ITGAM genes. This study is the first to evaluate and report that several SLE-associated regions show signs of positive natural selection. These results provide corroborating evidence in support of recent positive selection as one mechanism underlying the elevated population frequency of SLE risk loci and supports future research that integrates signals of natural selection to help identify functional SLE risk alleles.