Pragmatic considerations for negative control outcome studies to guide non-randomized comparative analyses: A narrative review

This narrative review describes the application of negative control outcome (NCO) methods to assess potential bias due to unmeasured or mismeasured confounders in non-randomized comparisons of drug effectiveness and safety. An NCO is assumed to have no causal relationship with a treatment under study while subject to the same confounding structure as the treatment and outcome of interest; an association between treatment and NCO then reflects the potential for uncontrolled confounding between treatment and outcome. We focus on two recently completed NCO studies that assessed the comparability of outcome risk for patients initiating different osteoporosis medications and lipid-lowering therapies, illustrating several ways in which confounding may result. In these studies, NCO methods were implemented in claims-based data sources, with the results used to guide the decision to proceed with comparative effectiveness or safety analyses. Based on this research, we provide recommendations for future NCO studies, including considerations for the identification of confounding mechanisms in the target patient population, the selection of NCOs expected to satisfy required assumptions, the interpretation of NCO effect estimates, and the mitigation of uncontrolled confounding detected in NCO analyses. We propose the use of NCO studies prior to initiating comparative effectiveness or safety research, providing information on the potential presence of uncontrolled confounding in those comparative analyses. Given the increasing use of non-randomized designs for regulatory decision-making, the application of NCO methods will strengthen study design, analysis, and interpretation of real-world data and the credibility of the resulting real-world evidence. This article is protected by copyright. All rights reserved.

SARS-CoV-2 Mitigation Strategies, Testing, and Cases at 254 Jails in the US Southeast, October 2020 to May 2021

Objectives. To characterize severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mitigation strategies, testing, and cases across county jails in the Southeastern United States, examining variability by jail characteristics. Methods. We administered a 1-time telephone survey to personnel of 254 jails in Alabama, Georgia, North Carolina, and South Carolina between October 2020 and May 2021. Results. Some SARS-CoV-2 mitigation strategies (e.g., screening at intake, isolation and masking for symptomatic persons) were commonly reported (≥ 75% of jails). Other measures, such as masking regardless of symptoms (52%) and screening at release (26%), were less common and varied by jail state or population size. Overall, 41% of jails reported no SARS-CoV-2 testing in the past 30 days. Jails with testing (59%) tested a median of 6 per 100 incarcerated persons; of those jails, one third reported 1 or more cases of positive tests. Although most jails detected no cases, in the 20% of all jails with 1 or more case in the past 30 days, 1 in 5 tests was positive. Conclusions. There was low testing coverage and variable implementation of SARS-CoV-2 mitigation strategies in Southeastern US jails during the first year of the COVID-19 pandemic. (Am J Public Health. 2022;112(11):1589-1598. https://doi.org/10.2105/AJPH.2022.307012).

Lipid Testing Trends Before and After Hospitalization for Myocardial Infarction Among Adults in the United States, 2008-2019

Background

Routine monitoring of low-density lipoprotein cholesterol (LDL-C) identifies patients who may benefit from modifying lipid-lowering therapies (LLT). However, the extent to which LDL-C testing is occurring in clinical practice is unclear, specifically among patients hospitalized for a myocardial infarction (MI).

Methods

Using US commercial claims data, we identified patients with an incident MI hospitalization between 01/01/2008-03/31/2019. LDL-C testing was assessed in the year before admission (pre-MI) and the year after discharge (post-MI). Changes in LDL-C testing were evaluated using a Poisson model fit to pre-MI rates and extrapolated to the post-MI period. We predicted LDL-C testing rates if no MI had occurred (ie, based on pre-MI trends) and estimated rate differences and ratios (contrasting observed vs predicted rates).

Results

Overall, 389,367 patients were hospitalized for their first MI during the study period. In the month following discharge, 9% received LDL-C testing, increasing to 27% at 3 months and 52% at 12 months. Mean rates (tests per 1000 patients per month) in the pre- and post-MI periods were 51.9 (95% CI: 51.7, 52.1) and 84.4 (95% CI: 84.1, 84.6), respectively. Over 12 months post-MI, observed rates were higher than predicted rates; the maximum rate difference was 66 tests per 1000 patients in month 2 (rate ratio 2.2), stabilizing at a difference of 15-20 (ratio 1.2-1.3) for months 6-12.

Conclusion

Although LDL-C testing increased following MI hospitalization, rates remained lower than recommended by clinical guidelines. This highlights a potential gap in care, where increased LDL-C testing after MI may provide opportunities for LLT modification and decrease risk of subsequent cardiovascular events.

Use of negative control outcomes to assess the comparability of patients initiating lipid-lowering therapies

PURPOSE

Clinical trials have demonstrated efficacy of proprotein convertase subtilisin/kexin type 9 inhibitors (PCSK9i) in reducing risk of cardiovascular disease events, but effectiveness in routine clinical care has not been well-studied. We used negative control outcomes to assess potential confounding in an observational study of PCSK9i versus ezetimibe or high-intensity statin.

METHODS

Using commercial claims, we identified U.S. adults initiating PCSK9i, ezetimibe, or high-intensity statin in 2015-2018, with other lipid-lowering therapy (LLT) use in the year prior (LLT cohort) or atherosclerotic cardiovascular disease (ASCVD) in the past 90 days (ASCVD cohort). We compared initiators of PCSK9i to ezetimibe and high-intensity statin by estimating one-year risks of negative control outcomes influenced by frailty or health-seeking behaviors. Inverse probability of treatment and censoring weighted estimators of risk differences (RDs) were used to evaluate residual confounding after controlling for covariates.

RESULTS

PCSK9i initiators had lower one-year risks of negative control outcomes associated with frailty, such as decubitus ulcer in the ASCVD cohort (PCSK9i vs. high-intensity statin RD = -3.5%, 95% confidence interval (CI): -4.6%, -2.5%; PCSK9i vs. ezetimibe RD = -1.3%, 95% CI: -2.1%, -0.6%), with similar but attenuated associations in the LLT cohort. Lower risks of accidents and fractures were also observed for PCSK9i, varying by cohort. Risks were similar for outcomes associated with health-seeking behaviors, although trended higher for PCSK9i in the ASCVD cohort.

CONCLUSIONS

Observed associations suggest lower frailty and potentially greater health-seeking behaviors among PCSK9i initiators, particularly those with a recent ASCVD diagnosis, with the potential to bias real-world analyses of treatment effectiveness.

HIV and STI prevalence and testing history among men who have sex with men in Hanoi, Vietnam

BACKGROUND

HIV and other sexually transmitted infections (STIs) have disproportionately affected communities of men who have sex with men (MSM). We describe HIV and STI prevalence and testing patterns among urban Vietnamese MSM.

METHODS

We conducted a cross-sectional community-based study of MSM in Hanoi, Vietnam in 2016. Participants self-reported experiences of social stigma in healthcare settings and previous HIV and STI testing. STI testing included HIV, herpes simplex virus-2 (HSV-2), syphilis, gonorrhea, and chlamydia.

RESULTS

205 MSM participated in the study. STI prevalence was HIV (10%), HSV-2 (4%), syphilis (13%), gonorrhea (34%), and chlamydia (19%). More than half (55%) of participants tested positive for at least one STI. Most participants had been previously tested for HIV or another STI (72%), with 24% previously receiving a positive result. Perceived and enacted social stigma in healthcare contexts was negatively associated with previous HIV or STI testing (adjusted prevalence odds ratio (aPOR): 0.22; 95% confidence interval (CI): 0.10-0.48).

DISCUSSION

High prevalence of STIs was observed among Vietnamese MSM, and perceived and enacted stigma was related to HIV and STI testing. Our findings reaffirm the importance of regular STI screening among this population as well as additional outreach to promote safe HIV and STI healthcare engagement.

Lipid testing trends before and after hospitalization for myocardial infarction among adults in the U.S., 2008–2019

Background

Routine monitoring of low-density lipoprotein cholesterol (LDL-C) identifies patients who are candidates for initiation of lipid-lowering therapies (LLT), assesses adherence and response, and measures clinical indications for modifications to treatment. This study investigated LDL-C testing trends among patients hospitalized for incident myocardial infarction (MI), a population at high risk of future atherosclerotic cardiovascular disease (ASCVD) events who may benefit from additional LLT.

Purpose

Our objective was to estimate rates of LDL-C testing before and after MI hospitalization, hypothesizing that testing rates would increase following MI.

Methods

Using claims data from a large population with Medicare supplemental or commercial insurance, we identified U.S. adults aged ≥20 years discharged from a MI hospitalization. Patients qualified for the study cohort if their first MI hospitalization occurred during 1/1/2008–3/31/2019, with one year of continuous enrolment in insurance leading up to the hospitalization. Patients were required to be discharged alive to the community (excluded if died or transferred to another facility). LDL-C testing was assessed in the year before admission (pre-MI) and for up to one year after discharge (post-MI), with censoring due to insurance plan disenrollment. To evaluate changes in LDL-C testing, we fit an overdispersed Poisson model to the time-series of pre-MI rates and extrapolated the model to the post-MI period, accounting for person-time and seasonality. We predicted the LDL-C testing rates if no MI occurred (i.e., based only on pre-MI testing trends) and estimated rate differences with 95% confidence intervals (contrasting observed vs. model-predicted rates).

Results

A total of 389,367 patients were hospitalized for their first MI during the study period with 60% aged <65 years, 64% were male, and 51% filled a statin prescription after discharge (half being high-intensity statins). In the year pre-MI, 25% had a history of diabetes mellitus, 35% used statins, and 4% used ezetimibe. In the year post-MI, only 52% had an LDL-C test. Mean observed rates (tests per 1,000 patients per month) in the pre- and post-MI periods were 51.9 (95% CI: 51.7, 52.1) and 84.4 (95% CI: 84.1, 84.6), respectively. Across the 12 months post-MI, observed rates were higher than model-predicted rates, with the magnitude of rate differences changing over time (Figure). The observed testing rate peaked two months post-MI (rate difference: 65.7, 95% CI: 64.6, 66.7).

Conclusions

Our findings indicate that LDL-C testing rates increased following a MI hospitalization and stayed elevated throughout the following year. Despite this increase, overall rates remained low, with only one in two patients receiving an LDL-C test in the year after MI. These results highlight a potential gap in care, particularly given the importance of LDL-C monitoring for this population to reduce risk of future ASCVD events.

Funding Acknowledgement

Type of funding sources: Private company. Main funding source(s): This study was funded by Amgen, Inc. Several authors are employees and own stock in Amgen.

Estimating the Effect of Depression on HIV Transmission Risk Behaviors Among People Who Inject Drugs in Vietnam: A Causal Approach

The burden of depression and HIV is high among people who inject drugs (PWID), yet the effect of depression on transmission risk behaviors is not well understood in this population. Using causal inference methods, we analyzed data from 455 PWID living with HIV in Vietnam 2009-2013. Study visits every 6 months over 2 years measured depressive symptoms in the past week and injecting and sexual behaviors in the prior 3 months. Severe depressive symptoms (vs. mild/no symptoms) increased injection equipment sharing (risk difference [RD] = 3.9 percentage points, 95% CI -1.7, 9.6) but not condomless sex (RD = -1.8, 95% CI -6.4, 2.8) as reported 6 months later. The cross-sectional association with injection equipment sharing at the same visit (RD = 6.2, 95% CI 1.4, 11.0) was stronger than the longitudinal effect. Interventions on depression among PWID may decrease sharing of injection equipment and the corresponding risk of HIV transmission.Clinical trial registration ClinicalTrials.gov NCT01689545.

Lipid Testing Trends in the US Before and After the Release of the 2013 Cholesterol Treatment Guidelines

Background

The 2013 ACC/AHA cholesterol treatment guidelines removed the recommendation to treat adults at risk of cardiovascular disease to goal levels of low-density lipoprotein cholesterol (LDL-C). We anticipated that the frequency of LDL-C testing in clinical practice would decline as a result. To test this hypothesis, we evaluated the frequency of LDL-C testing before and after the guideline release.

Methods

We used the MarketScan^® Commercial and Medicare Supplemental claims data (1/1/2007–12/31/2016) to identify four cohorts: 1) statin initiators (any intensity), 2) high-intensity statin initiators, 3) ezetimibe initiators, and 4) patients at very high cardiovascular risk (≥2 hospitalizations for myocardial infarction or ischemic stroke, with prevalent statin use). Rates of LDL-C testing by calendar year quarter were estimated for each cohort. To estimate rates in the absence of a guideline change, we fit a time-series model to the pre-guideline rates and extrapolated to the post-guideline period, adjusting for covariates, seasonality, and time trend.

Results

Pre- and post-guideline rates (LDL-C tests per 1,000 persons per quarter) were 248 and 235, respectively, for 3.9 million statin initiators; 263 and 246 for 1.3 million high-intensity statin initiators; 277 and 261 for 323,544 ezetimibe initiators; and 180 and 158 for 42,108 very high-risk patients. For all cohorts, observed post-guideline rates were similar to model-predicted rates. On average, the difference between observed and predicted rates was 8.5 for patients initiating any statin; 2.6 for patients initiating a high-intensity statin; 11.4 for patients initiating ezetimibe, and −0.5 for high-risk patients.

Conclusion

We observed no discernible impact of the release of the 2013 ACC/AHA guidelines on LDL-C testing rates. Rather, there was a gradual decline in testing rates starting prior to the guideline change and continuing throughout the study period. Our findings suggest that the guidelines had little to no impact on use of LDL-C testing.

Using negative control outcomes to assess the comparability of treatment groups among women with osteoporosis in the United States

PURPOSE

In contrast to randomized clinical trials, comparative safety and effectiveness assessments of osteoporosis medications in clinical practice may be subject to confounding by indication. We used negative control outcomes to detect residual confounding when comparing osteoporosis medications.

METHODS

Using MarketScan Commercial and Supplemental claims, we identified women aged ≥55 years who initiated an oral bisphosphonate (BP) (risedronate, alendronate, or ibandronate), denosumab (an injected biologic), or intravenous zoledronic acid (ZA) from October 1, 2010 to September 30, 2015. Women with Paget's disease or cancer were excluded. We compared individual oral BPs to each other, denosumab to ZA, denosumab to oral BPs, and ZA to oral BPs, with respect to 11 negative control outcomes identified by subject matter experts. We estimated the 12-month cumulative risk difference (RD) using inverse probability of treatment and censoring weights.

RESULTS

Among 148 587 women, most initiated alendronate (57%), followed by ibandronate (12%), ZA (11%), risedronate (10%), and denosumab (10%). Compared with denosumab, patients initiating ZA had similar risks of all negative control outcomes. Compared with oral BPs, patients initiating denosumab had a higher risk of a wellness visit (RD = 1.2%, 95% CI: 0.4, 1.9) and a lower risk of receiving herpes zoster vaccine (RD = -0.6%, 95% CI: -1.1, -0.2). Comparing ZA with oral BP initiators resulted in two outcomes with positive associations.

CONCLUSIONS

Caution is warranted when comparing injectable vs oral osteoporosis medications, given the potential for unmeasured confounding. Evaluating negative control outcomes could be a standard validity check prior to conducting comparative studies.

Longitudinal Analysis of Depressive Symptoms, Perceived Social Support, and Alcohol Use among HIV-Infected Men Who Inject Drugs in Northern Vietnam

Background: Limited research examines depressive symptoms, alcohol use, and social support among HIV-infected people who inject drugs. Objectives: Using longitudinal data, we investigated whether perceived social support moderates the relationship between depressive symptoms and alcohol use among HIV-infected men who inject drugs in Vietnam. Methods: Data were collected from participants (N = 455; mean age 35 years) in a four-arm randomized controlled trial in Thai Nguyen, Vietnam. Data were collected at baseline, 6, 12, 18, and 24 months with 94% retention excluding dead (N = 103) or incarcerated (N = 37) participants. Multilevel growth models were used to assess whether: (1) depressive symptoms predict when risk of alcohol use is elevated (within-person effects); (2) depressive symptoms predict who is at risk for alcohol use (between-person effects); and (3) within- and between-person perceived social support moderates the depressive symptoms-alcohol relationship. Results: Participants reported high but declining levels of depressive symptoms and alcohol use. Participants with higher depressive symptoms drank less on average (B = -0.0819, 95% CI -0.133, -0.0307), but within-person, a given individual was more likely to drink when they were feeling more depressed than usual (B = 0.136, 95% CI 0.0880, 0.185). The positive relationship between within-person depressive symptoms and alcohol use grew stronger at higher levels of within-person perceived social support. Conclusions: HIV-infected men who inject drugs have increased alcohol use when they are experiencing higher depressive symptoms than usual, while those with higher average depressive symptoms over time report less alcohol use. Social support strengthens the positive relationship between within-person depressive symptoms and alcohol use.

Prevalence and Transmission Dynamics of HIV-1 Transmitted Drug Resistance in a Southeastern Cohort

Background

Transmitted drug resistance (TDR) compromises clinical management and outcomes. Transmitted drug resistance surveillance and identification of growing transmission clusters are needed in the Southeast, the epicenter of the US HIV epidemic. Our study investigated prevalence and transmission dynamics in North Carolina.

Methods

We analyzed surveillance drug resistance mutations (SDRMs) using partial pol sequences from patients presenting to 2 large HIV outpatient clinics from 1997 to 2014. Transmitted drug resistance prevalence was defined as ≥1 SDRMs among antiretroviral therapy (ART)–naïve patients. Binomial regression was used to characterize prevalence by calendar year, drug class, and demographic and clinical factors. We assessed the transmission networks of patients with TDR with maximum likelihood trees and Bayesian methods including background pol sequences (n = 15 246).

Results

Among 1658 patients with pretherapy resistance testing, ≥1 SDRMs was identified in 199 patients, with an aggregate TDR prevalence of 12% (95% confidence interval, 10% to 14%) increasing over time (P = .02). Resistance to non-nucleoside reverse transcriptase inhibitors (NNRTIs; 8%) was common, followed by nucleoside reverse transcriptase inhibitors (4%) and protease inhibitors (2%). Factors associated with TDR were being a man reporting sex with men, white race, young age, higher CD4 cell count, and being a member of a transmission cluster. Transmitted drug resistance was identified in 106 clusters ranging from 2 to 26 members. Cluster resistance was primarily NNRTI and dominated by ART-naïve patients or those with unknown ART initiation.

Conclusions

Moderate TDR prevalence persists in North Carolina, predominantly driven by NNRTI resistance. Most TDR cases were identified in transmission clusters, signifying multiple local transmission networks and TDR circulation among ART-naïve persons. Transmitted drug resistance surveillance can detect transmission networks and identify patients for enhanced services to promote early treatment.

Prevalence and predictors of depressive symptoms among HIV-positive men who inject drugs in Vietnam

Background

HIV infection is common among people who inject drugs (PWID), and HIV-positive PWID may be particularly vulnerable to depression. This study measured the prevalence of depressive symptoms and the factors associated with severe symptoms among 455 HIV-positive PWID in Thai Nguyen, Vietnam.

Methods

We used cross-sectional data from PWID in a randomized controlled trial of an intervention to reduce high-risk injecting and sexual behaviors in Thai Nguyen from 2009–2013. Depressive symptoms were measured with the Center for Epidemiologic Studies Depression Scale (CES-D). We used logistic regression to assess demographic, clinical, and psychosocial predictors of severe depressive symptoms (CES-D≥23) with prevalence odds ratios (POR) and 95% confidence intervals (CI).

Results

The prevalence of severe depressive symptoms (CES-D≥23) was 44%. 25% of participants had mild to moderate depressive symptoms (16≤CES-D<23), and 31% experienced no depressive symptoms (CES-D<16). Not being married, self-rated poor health, greater frequency of injection drug use, history of overdose, no alcohol use, and daily cigarette smoking were positively associated with severe depressive symptoms in unadjusted models and remained predictive in a multivariable model. The strongest predictors of depressive symptoms were self-reported poor health (POR = 2.94, 95% CI: 1.82, 4.76), no current alcohol use (POR = 2.35, 95% CI: 1.47, 3.77), and not currently married or cohabitating (POR = 2.21, 95% CI = 1.40, 3.47).

Conclusion

Severe depressive symptoms were common among HIV-positive PWID in Thai Nguyen and were strongly associated with demographic, clinical, and psychosocial factors. Interventions that promote social support from family and reduce drug dependence may particularly benefit PWID experiencing severe depressive symptoms. Greater recognition and treatment of depressive symptoms has the potential to enhance quality of life and improve HIV clinical outcomes for PWID.