More comprehensive reporting of methods in studies using respondent driven sampling is required: a systematic review of the uptake of the STROBE-RDS guidelines

Viability of Web-Based Respondent-Driven Sampling of Belgian Men Who Have Sex With Men: Process Evaluation

BACKGROUND

Obtaining a representative sample is a substantial challenge when undertaking health research among hidden and hard-to-reach populations such as men who have sex with men Web-based respondent-driven sampling (WEB RDS) was developed to overcome such sampling challenges and to create population estimates based on network and sampling characteristics. Despite a decade of research, it remains unclear whether WEB RDS is suitable for sampling hidden populations such as men who have sex with men.

OBJECTIVE

This study aims to evaluate how viable the WEB RDS methodology is for obtaining a nationwide sample of men who have sex with men, suitable for population inference of sexual health characteristics, in Belgium.

METHODS

We adapted the Medical Resource Council process evaluation framework for interventions, to evaluate an empirical WEB RDS. Viewing "WEB RDS" as a complex intervention with respondent-driven recruitment as the aim, we evaluated indicators of context, implementation, mechanisms of impact, and performance. We analyzed the data using a mixed methods approach that integrated findings from quantitative analysis, such as RDS diagnostics, and qualitative thematic analysis.

RESULTS

Sampling did not reach a sufficient sample size (n=193) to compensate for an RDS design effect of 3 and the number of recruitment waves was low (waves=7). A visual examination of the convergence and bottleneck plots indicates that many more waves of recruitment would be needed for population estimates to become independent of the seeds. However, producing further waves was impeded by challenges inherent to the research context and process. Men who have sex with men and their community organization representatives indicated that, in Belgium, men who have sex with men are overresearched, with low motivation for the topic of sexual health and digital etiquette dictating not sharing survey links. A moderate reward of €10-€30 (US $11.2-$33.6) with a dual incentive structure was insufficient to overcome these barriers.

CONCLUSIONS

This study indicates that WEB RDS, even with a moderate incentive, is not a viable sampling strategy for obtaining valid population estimates of sexual health traits of men who have sex with men in Belgium. The study emphasizes the need to understand men who have sex with men research motivation and topic saliency. Additionally, the study highlights the importance of digital etiquette. Finally, the study showcases the use of the adapted Medical Research Council framework for evaluating WEB RDS methodology.

Characteristics of anesthesia program that graduating medical students find important: A cross-sectional survey

BACKGROUND

Applying to residency programs and constructing a rank order list is a challenging process and is a vital task that fourth year medical students must undergo each year. The aim of our study was to identify common themes in what residency applicants are considering when applying to an anesthesiology program and compare them between male and female applicants.

METHODS

A cross-sectional survey was distributed to all 1447 medical students who applied to the Brown University Anesthesiology residency program for the 2022 National Residency Matching Program (NRMP). The survey consisted of 53 questions designed to assess the factors that candidates deem important when ranking a residency program highly. Exploratory factor analysis was performed to determine the item-factor assignment. Confirmatory analysis was done to test the associations between the average scores for each of the identified factors. Associations and average scores among male and female responders were explored.

RESULTS

922 medical students completed the survey with a response rate of 63.7 %. An exploratory factor analysis (n = 457) performed on a random split sample (RSS) revealed a 45-item, seven-factor structure consisting of the following themes: exposure, teaching quality, gender and racial diversity, financial considerations, work life balance, prestige, and living in the same area. A confirmatory factor analysis was performed for the seven-factor model on the second RSS (n = 465). The model demonstrated adequate fit (Comparative Fit Index, CFI = 0.90, root mean square error of approximation [RIMSEA] = 0.085). Post hoc modification significantly improved the model fit (CFI = 0.98, RMSEA = 0.057).

CONCLUSION

Anesthesia applicants prioritize clinical exposure, teaching quality, diversity, and practical training factors like simulation and POCUS training. In addition, applicants also weigh financial considerations and work-life balance heavily in their ranking of anesthesia programs. Understanding these preferences can help anesthesia programs effectively market their programs to improve recruitment outcomes and to attract top applicants.

Serum Levels of Bupivacaine After Bilateral Ultrasound-Guided Deep Parasternal Intercostal Plane Block in Cardiac Surgery with Median Sternotomy

OBJECTIVE

To evaluate systemic levels of bupivacaine after bilateral ultrasound-guided deep parasternal intercostal plan (PIP) block in cardiac surgical patients undergoing median sternotomy.

DESIGN

Prospective, observational study SETTING: Single institution; academic university hospital PARTICIPANTS: Twenty-eight adult patients undergoing cardiac surgery with median sternotomy received a PIP block with 2.5 mg/kg bupivacaine with or without dexamethasone and dexmedetomidine.

MEASUREMENTS

Arterial blood samples were analyzed for total serum bupivacaine concentration at 5, 15, 30, 45, 60, 90, 120, and 150 minutes after placement of PIP. Local anesthetic volume, local anesthetic adjuncts, time to extubation, postoperative pain scores, and opioid consumption were recorded.

MAIN RESULTS

The mean peak bupivacaine concentration was 0.60 ± 0.62 µg/mL, and the mean time to maximum concentration (Tmax) was 16.92 ± 12.97 minutes. Two patients (7.1%) had a concentration >2.0 µg/mL within 15 minutes of block placement. The mean Tmax of bupivacaine was significantly greater in patients who did not receive additives compared to those patients who did (22.86 ± 14.77 minutes v 10.0 ± 5.22 minutes; p = .004). The times to extubation and postoperative pain were not improved with additives.

CONCLUSIONS

Bilateral PIP placed at the end of cardiac surgery resulted in low systemic bupivacaine levels. The inclusion of additives shortened Tmax without improving outcome.

The Development and the Assessment of Sampling Methods for Hard-to-Reach Populations in HIV Surveillance

Due to stigma or legal issues, populations with higher HIV risk are often hard to reach, which impedes accurate population estimation of HIV burden. To better sample hard-to-reach populations (HTRPs) for HIV surveillance, various sampling methods have been designed and/or used since HIV epidemic following the first reported AIDS cases in 1981. This paper describes the development and the assessment (i.e., validity and reproducibility) of approximately eight sampling methods (e.g., convenience sampling, snowball sampling, time location sampling, and respondent-driven sampling) for HTRPs in HIV surveillance, with a focus on respondent-driven sampling (RDS). Compared to other methods, RDS has been greatly assessed. However, current evidence is still inadequate for RDS to be considered the best option for sampling HTRPs. The field must continue to assess RDS and to develop new sampling approaches or modifications to existing approaches.

Evaluation of respondent-driven sampling in seven studies of people who use drugs from rural populations: findings from the Rural Opioid Initiative

BACKGROUND

Accurate prevalence estimates of drug use and its harms are important to characterize burden and develop interventions to reduce negative health outcomes and disparities. Lack of a sampling frame for marginalized/stigmatized populations, including persons who use drugs (PWUD) in rural settings, makes this challenging. Respondent-driven sampling (RDS) is frequently used to recruit PWUD. However, the validity of RDS-generated population-level prevalence estimates relies on assumptions that should be evaluated.

METHODS

RDS was used to recruit PWUD across seven Rural Opioid Initiative studies between 2018-2020. To evaluate RDS assumptions, we computed recruitment homophily and design effects, generated convergence and bottleneck plots, and tested for recruitment and degree differences. We compared sample proportions with three RDS-adjusted estimators (two variations of RDS-I and RDS-II) for five variables of interest (past 30-day use of heroin, fentanyl, and methamphetamine; past 6-month homelessness; and being positive for hepatitis C virus (HCV) antibody) using linear regression with robust confidence intervals. We compared regression estimates for the associations between HCV positive antibody status and (a) heroin use, (b) fentanyl use, and (c) age using RDS-1 and RDS-II probability weights and no weights using logistic and modified Poisson regression and random-effects meta-analyses.

RESULTS

Among 2,842 PWUD, median age was 34 years and 43% were female. Most participants (54%) reported opioids as their drug of choice, however regional differences were present (e.g., methamphetamine range: 4-52%). Many recruitment chains were not long enough to achieve sample equilibrium. Recruitment homophily was present for some variables. Differences with respect to recruitment and degree varied across studies. Prevalence estimates varied only slightly with different RDS weighting approaches, most confidence intervals overlapped. Variations in measures of association varied little based on weighting approach.

CONCLUSIONS

RDS was a useful recruitment tool for PWUD in rural settings. However, several violations of key RDS assumptions were observed which slightly impacts estimation of proportion although not associations.

Outpatient compared to inpatient thyroidectomy on 30-day postoperative outcomes: a national propensity matched analysis

BACKGROUND

To address the postoperative outcomes between outpatient and inpatient neck surgery involving thyroidectomy procedures.

METHODS

A cohort analysis of surgical patients undergoing primary, elective, total thyroidectomy from multiple United States medical institutions who were registered with the American College of Surgeons National Surgical Quality Improvement Program from 2015 to 2018. The primary outcome was a composite score that included any 30-day postoperative adverse event.

RESULTS

A total of 55,381 patients who underwent a total thyroidectomy were identified comprising of 14,055 inpatient and 41,326 outpatient procedures. A cohort of 13,496 patients who underwent outpatient surgery were propensity matched for covariates with corresponding number of patients who underwent inpatient thyroidectomies. In the propensity matched cohort, the occurrence of any 30-day after surgery complications were greater in the inpatient group, 424 out of 13,496 (3.1%) compared to the outpatient group, 150 out of 13,496 (1.1%), P < 0.001. Moreover, death rates were greater in the inpatient group, 22 out 13,496 (0.16%) compared to the outpatient group, 2 out of 13,496 (0.01%), P < 0.001. Similarly, hospital readmissions occurred with greater frequency in the inpatient group, 438 out of 13,496 (3.2%) compared to the outpatient group, 310 out of 13,496 (2.3%), P < 0.001.

CONCLUSION

Thyroidectomy procedures performed in the outpatient setting had less rates of adverse events, including serious postoperative complications (e.g., surgical site infection, pneumonia, progressive renal insufficiency). In addition, patients who had thyroidectomy in the outpatient setting had less 30-day readmissions and mortality. Surgeons should recognize the benefits of outpatient thyroidectomy when selecting disposition of patients undergoing neck surgery.

Hidden population size estimation and diagnostics using two respondent-driven samples with applications in Armenia

Estimating the size of hidden populations is essential to understand the magnitude of social and healthcare needs, risk behaviors, and disease burden. However, due to the hidden nature of these populations, they are difficult to survey, and there are no gold standard size estimation methods. Many different methods and variations exist, and diagnostic tools are needed to help researchers assess method-specific assumptions as well as compare between methods. Further, because many necessary mathematical assumptions are unrealistic for real survey implementation, assessment of how robust methods are to deviations from the stated assumptions is essential. We describe diagnostics and assess the performance of a new population size estimation method, capture-recapture with successive sampling population size estimation (CR-SS-PSE), which we apply to data from 3 years of studies from three cities and three hidden populations in Armenia. CR-SS-PSE relies on data from two sequential respondent-driven sampling surveys and extends the successive sampling population size estimation (SS-PSE) framework by using the number of individuals in the overlap between the two surveys and a model for the successive sampling process to estimate population size. We demonstrate that CR-SS-PSE is more robust to violations of successive sampling assumptions than SS-PSE. Further, we compare the CR-SS-PSE estimates to population size estimations using other common methods, including unique object and service multipliers, wisdom of the crowd, and two-source capture-recapture to illustrate volatility across estimation methods.

Emulation of the control cohort of a randomized controlled trial in pediatric kidney transplantation with Real-World Data from the CERTAIN Registry

BACKGROUND

Randomized controlled trials in pediatric kidney transplantation are hampered by low incidence and prevalence of kidney failure in children. Real-World Data from patient registries could facilitate the conduct of clinical trials by substituting a control cohort. However, the emulation of a control cohort by registry data in pediatric kidney transplantation has not been investigated so far.

METHODS

In this multicenter comparative analysis, we emulated the control cohort (n = 54) of an RCT in pediatric kidney transplant patients (CRADLE trial; ClinicalTrials.gov NCT01544491) with data derived from the Cooperative European Paediatric Renal Transplant Initiative (CERTAIN) registry, using the same inclusion and exclusion criteria (CERTAIN cohort, n = 554).

RESULTS

Most baseline patient and transplant characteristics were well comparable between both cohorts. At year 1 posttransplant, a composite efficacy failure end point comprising biopsy-proven acute rejection, graft loss or death (5.8% ± 3.3% vs. 7.5% ± 1.1%, P = 0.33), and kidney function (72.5 ± 24.9 vs. 77.3 ± 24.2 mL/min/1.73 m² P = 0.19) did not differ significantly between CRADLE and CERTAIN. Furthermore, the incidence and severity of BPAR (5.6% vs. 7.8%), the degree of proteinuria (20.2 ± 13.9 vs. 30.6 ± 58.4 g/mol, P = 0.15), and the key safety parameters such as occurrence of urinary tract infections (24.1% vs. 15.5%, P = 0.10) were well comparable.

CONCLUSIONS

In conclusion, usage of Real-World Data from patient registries such as CERTAIN to emulate the control cohort of an RCT is feasible and could facilitate the conduct of clinical trials in pediatric kidney transplantation. A higher resolution version of the Graphical abstract is available as Supplementary information.

Drivers of US health care spending for persons with seizures and/or epilepsies, 2010-2018

OBJECTIVE

This study was undertaken to characterize spending for persons classified with seizure or epilepsy and to determine whether spending has increased over time.

METHODS

In this cross-sectional study, we pooled data from the Medical Expenditure Panel Survey (MEPS) household component files for 2010-2018. We matched cases to controls on age and sex of a population-based sample of MEPS respondents (community-dwelling persons of all ages) with records associated with a medical event (e.g., outpatient visit, hospital inpatient) for seizure, epilepsy, or both. Outcomes were weighted to be representative of the civilian, noninstitutionalized population. We estimated the treated prevalence of epilepsy and seizure, health care spending overall and by site of care, and trends in spending growth.

RESULTS

We identified 1078 epilepsy cases and 2344 seizure cases. Treated prevalence was .38% (95% confidence interval [CI] = .34-.41) for epilepsy, .76% (95% CI = .71-.81) for seizure, and 1.14% (95% CI = 1.08-1.20) for epilepsy or seizure. The difference in annual spending for cases compared to controls was $4580 (95% CI = $3362-$5798) for epilepsy, $7935 (95% CI, $6237-$9634) for seizure, and $6853 (95% CI = $5623-$8084) for epilepsy or seizure, translating into aggregate costs of $5.4 billion, $19.0 billion, and $24.5 billion. From 2010 to 2018, the annual growth rate in total spending incurred for seizures and/or epilepsies was 7.6% compared to 3.6% among controls.

SIGNIFICANCE

US economic burden of seizures and/or epilepsies is substantial and warrants interventions focused on their unique and overlapping causes.

Respondent-Driven Sampling: a Sampling Method for Hard-to-Reach Populations and Beyond

Purpose of Review

We provided an overview of sampling methods for hard-to-reach populations and guidance on implementing one of the most popular approaches: respondent-driven sampling (RDS).

Recent Findings

Limitations related to generating a sampling frame for marginalized populations can make them “hard-to-reach” when conducting population health research. Data analyzed from non-probability-based or convenience samples may produce estimates that are biased or not generalizable to the target population. In RDS and time-location sampling (TLS), factors that influence inclusion can be estimated and accounted for in an effort to generate representative samples. RDS is particularly equipped to reach the most hidden members of hard-to-reach populations.

Summary

TLS, RDS, or a combination can provide a rigorous method to identify and recruit samples from hard-to-reach populations and more generalizable estimates of population characteristics. Researchers interested in sampling hard-to-reach populations should expand their toolkits to include these methods.