Regression discontinuity designs in epidemiology: causal inference without randomized trials.

When patients receive an intervention based on whether they score below or above some threshold value on a continuously measured random variable, the intervention will be randomly assigned for patients close to the threshold. The regression discontinuity design exploits this fact to estimate causal treatment effects. In spite of its recent proliferation in economics, the regression discontinuity design has not been widely adopted in epidemiology. We describe regression discontinuity, its implementation, and the assumptions required for causal inference. We show that regression discontinuity is generalizable to the survival and nonlinear models that are mainstays of epidemiologic analysis. We then present an application of regression discontinuity to the much-debated epidemiologic question of when to start HIV patients on antiretroviral therapy. Using data from a large South African cohort (2007-2011), we estimate the causal effect of early versus deferred treatment eligibility on mortality. Patients whose first CD4 count was just below the 200 cells/µL CD4 count threshold had a 35% lower hazard of death (hazard ratio = 0.65 [95% confidence interval = 0.45-0.94]) than patients presenting with CD4 counts just above the threshold. We close by discussing the strengths and limitations of regression discontinuity designs for epidemiology.

Correction: Addressing vaccine hesitancy in developing countries: Survey and experimental evidence.

[This corrects the article DOI: 10.1371/journal.pone.0277493.].

Addressing vaccine hesitancy in developing countries: Survey and experimental evidence.

Vaccine hesitancy is proving to be a significant impediment to COVID-19 vaccination campaigns in some developing countries. This study focuses on vaccine hesitancy and means of reducing it. Data come from a large, representative phone survey and online randomized survey experiment, both run in Papua New Guinea, a developing country with low vaccination rates. Less than 20% of relevant respondents to the phone survey were willing to be vaccinated, primarily because of fear of side effects and low trust in the vaccine. Although vaccine hesitancy was high in the online experiment, participants who received a message emphasizing that the vaccine was safe and COVID-19 dangerous were 68% more likely to state they planned to be vaccinated than those in the control group. A message appealing to social norms was also effective in reducing vaccine hesitancy, although its efficacy was limited to certain types of people.

Effect of a Prize-Linked Savings Intervention on Savings and Healthy Behaviors Among Men in Kenya: A Randomized Clinical Trial.

Importance: Interventions to reduce men's alcohol use and risky sexual behaviors are essential for reducing new HIV infections in high-prevalence settings in sub-Saharan Africa. Prize-linked savings accounts can motivate savings and may decrease expenditures on risky behaviors, but few studies have examined the HIV prevention potential of such savings interventions among men. Objective: To evaluate the effect of prize-linked savings accounts on savings behavior and expenditures on alcohol, gambling, and transactional sex among men in Kenya. Design, Setting, and Participants: Randomized clinical trial among communities in Siaya County, Kenya. Participants were men 21 years or older who owned a mobile phone, were engaged in fishing or transportation sector work, and were willing to open an account with a local bank; they were screened for eligibility between September 3 and October 5, 2018. Interventions: Eligible participants were offered savings accounts endowed with 1000 Kenya shillings (US $10) and randomized (1:1) to receive weekly lottery-based rewards contingent on growth in savings balance or to a control group that received standard interest. Main Outcomes and Measures: The primary outcome was an indicator of whether a participant saved any money in the bank account (intent-to-treat analysis) during the study period. Secondary outcomes included total amount saved in the bank account, total amount saved in all sources, and expenditures on alcohol, gambling, and transactional sex. Results: A total of 425 men were screened, 329 (77.4%) met eligibility criteria, 300 (70.6%) were enrolled (with 152 randomized to the intervention group and 148 to the control group), and 270 of 300 (90.0%) opened bank accounts. Participants' mean age was 33.7 years (interquartile range, 13.5 years), 84.3% (253 of 300) were married, and the mean weekly earnings were US $30 (interquartile range, US $23). During a mean (SD) follow-up of 9 (2) weeks, 37.3% (50 of 134) in the intervention group saved money in a bank account vs 27.2% (37 of 136) in the control group, although the difference was not statistically significant (odds ratio, 1.62; 95% CI, 0.96-2.74). The intervention group had higher growth in bank savings balances (US $10.26; 95% CI, US $5.00-US $58.20 vs US $4.87; 95% CI, US $0.67-US $9.00) and higher total savings from all sources (US $201; 95% CI, US $133-US $269 vs US $145; 95% CI, US $88-US $202), but neither difference was statistically significant. The intervention did not have a significant effect on alcohol, gambling, and transactional sex expenditures. Conclusions and Relevance: Prize-linked savings accounts modestly increased savings among high-risk men in Kenya over a 9-week period, but the difference compared with standard-interest savings accounts was not significant. Testing of more powerful savings products is needed to assess whether such savings-led interventions can reduce men's expenditures on alcohol, gambling, and transactional sex. Trial Registration: Social Science Registry identifier: AEARCTR-0003224, and ClinicalTrials.gov identifier: NCT04013295.

Protecting the poor? Impact of the national health equity fund on utilization of government health services in Cambodia, 2006-2013.

INTRODUCTION: Cambodia's health equity fund (HEF) is the country's most significant social security scheme, covering the poorest one-fifth of the national population. During the last two decades, the HEF system was scaled up from an initial two health districts to national coverage of public health facilities. This is the first national study to examine the impact of the HEF on the utilisation of public health facilities. METHODS: We first investigated the level of national HEF population coverage and health service use made by HEF eligible members using an administrative HEF operational dataset. Second, through multilevel interrupted time series analysis of routine monthly utilisation statistics during 2006-2013, we evaluated the impact of the HEF on hospital and health centre utilisation. RESULTS: The proportion of HEF beneficiaries using hospital services in a given year (4.6%) appeared to exceed rates in the general population (3.3%). The introduction of the HEF was associated with: a significant level change in the monthly number of consultations at HCs followed by a gradual slope increase in time trend and a significant level change in the monthly number of deliveries. Overall, this was equivalent to a 15.6% net increase in number of consultations and 5.3% in deliveries in the first year. At RHs: a significant level change in the number of RH inpatient cases, followed by a sustained slope increase; a significant slope increase in the number of outpatient consultations and in the overall number of newborn deliveries. Overall, this was equivalent to a 47.9% net increase in inpatient cases, 24.1% in outpatient cases and 31.4% in deliveries in the first year. CONCLUSION: The implementation of the HEF scheme was associated with increased utilisation of primary and secondary care services by the poor.

Association of State Laws Permitting Denial of Services to Same-Sex Couples With Mental Distress in Sexual Minority Adults: A Difference-in-Difference-in-Differences Analysis.

Importance: Recent evidence suggests that state policies affecting sexual minorities are associated with health disparities. Twelve states have laws permitting the denial of services to same-sex couples, and the US Supreme Court is considering whether states can prohibit the denial of services to same-sex couples. Objective: We investigated whether state laws permitting individuals to refuse services to sexual minorities were associated with changes in the proportion of sexual minority adults reporting mental distress. Design, Setting, and Participants: This difference-in-difference-in-differences linear regression analysis with state fixed effects used Behavioral Risk Factor Surveillance System (BRFSS) data from 2014 through 2016 from adults aged 18 to 64 years in 3 states that implemented laws permitting the denial of services to same-sex couples (Utah, Michigan, and North Carolina) and 6 nearby control states (Idaho and Nevada, Ohio and Indiana, and Virginia and Delaware, respectively). Sexual minority adults were defined as those who identified as gay, lesbian, bisexual, or not sure of their sexual orientation under a module on sexual orientation that BRFSS implemented in 2014 and each state could opt to include. Analysis controlled for year and individual-level sex, race, ethnicity, age group, educational attainment, income, employment, and marital status. A permutation test was conducted to precisely estimate statistical significance. Exposures: An interaction term indicating whether individuals identified as a sexual minority and lived in a state with a law permitting denial of services to same-sex couples in 2015. Main Outcomes and Measures: Mental distress, defined as poor mental health on 14 or more of the past 30 days. Results: Of 109 089 participants, 4656 (4.8%; all percentages incorporate survey weights) identified as sexual minorities, 86141 (72.1%) were non-Hispanic white, and ages were uniformly distributed between 18 and 64 years. In 2014, 2038 of 16637 heterosexual adults (12.6%) and 156 of 815 sexual minority adults (21.9%) in the 3 same-sex denial states reported mental distress. The proportion of sexual minority adults reporting mental distress increased by 10.1 percentage points (95% CI, 1.8 to 18.5 percentage points, permutation-adjusted P value = .046) between 2014 and 2016 in states that passed laws permitting denial of services to same-sex couples compared with control states, a 46% relative increase in sexual minority adults experiencing mental distress. Laws permitting denial of services to same-sex couples were not associated with significant changes in heterosexual adults experiencing mental distress (-0.36 percentage points, 95% CI, -1.73 to 1.01 percentage points). Conclusions and Relevance: Laws permitting denial of services to same-sex couples, which exist in 12 states and are under consideration by the US Supreme Court, are associated with a 46% increase in sexual minority adults experiencing mental distress.

Causal language and strength of inference in academic and media articles shared in social media (CLAIMS): A systematic review.

BACKGROUND: The pathway from evidence generation to consumption contains many steps which can lead to overstatement or misinformation. The proliferation of internet-based health news may encourage selection of media and academic research articles that overstate strength of causal inference. We investigated the state of causal inference in health research as it appears at the end of the pathway, at the point of social media consumption. METHODS: We screened the NewsWhip Insights database for the most shared media articles on Facebook and Twitter reporting about peer-reviewed academic studies associating an exposure with a health outcome in 2015, extracting the 50 most-shared academic articles and media articles covering them. We designed and utilized a review tool to systematically assess and summarize studies' strength of causal inference, including generalizability, potential confounders, and methods used. These were then compared with the strength of causal language used to describe results in both academic and media articles. Two randomly assigned independent reviewers and one arbitrating reviewer from a pool of 21 reviewers assessed each article. RESULTS: We accepted the most shared 64 media articles pertaining to 50 academic articles for review, representing 68% of Facebook and 45% of Twitter shares in 2015. Thirty-four percent of academic studies and 48% of media articles used language that reviewers considered too strong for their strength of causal inference. Seventy percent of academic studies were considered low or very low strength of inference, with only 6% considered high or very high strength of causal inference. The most severe issues with academic studies' causal inference were reported to be omitted confounding variables and generalizability. Fifty-eight percent of media articles were found to have inaccurately reported the question, results, intervention, or population of the academic study. CONCLUSIONS: We find a large disparity between the strength of language as presented to the research consumer and the underlying strength of causal inference among the studies most widely shared on social media. However, because this sample was designed to be representative of the articles selected and shared on social media, it is unlikely to be representative of all academic and media work. More research is needed to determine how academic institutions, media organizations, and social network sharing patterns impact causal inference and language as received by the research consumer.

Difference-in-Differences Analysis of the Association Between State Same-Sex Marriage Policies and Adolescent Suicide Attempts.

Importance: Suicide is the second leading cause of death among adolescents between the ages of 15 and 24 years. Adolescents who are sexual minorities experience elevated rates of suicide attempts. Objective: To evaluate the association between state same-sex marriage policies and adolescent suicide attempts. Design, Setting, and Participants: This study used state-level Youth Risk Behavior Surveillance System (YRBSS) data from January 1, 1999, to December 31, 2015, which are weighted to be representative of each state that has participation in the survey greater than 60%. A difference-in-differences analysis compared changes in suicide attempts among all public high school students before and after implementation of state policies in 32 states permitting same-sex marriage with year-to-year changes in suicide attempts among high school students in 15 states without policies permitting same-sex marriage. Linear regression was used to control for state, age, sex, race/ethnicity, and year, with Taylor series linearized standard errors clustered by state and classroom. In a secondary analysis among students who are sexual minorities, we included an interaction between sexual minority identity and living in a state that had implemented same-sex marriage policies. Interventions: Implementation of state policies permitting same-sex marriage during the full period of YRBSS data collection. Main Outcomes and Measures: Self-report of 1 or more suicide attempts within the past 12 months. Results: Among the 762 678 students (mean [SD] age, 16.0 [1.2] years; 366 063 males and 396 615 females) who participated in the YRBSS between 1999 and 2015, a weighted 8.6% of all high school students and 28.5% of students who identified as sexual minorities reported suicide attempts before implementation of same-sex marriage policies. Same-sex marriage policies were associated with a 0.6-percentage point (95% CI, -1.2 to -0.01 percentage points) reduction in suicide attempts, representing a 7% relative reduction in the proportion of high school students attempting suicide owing to same-sex marriage implementation. The association was concentrated among students who were sexual minorities. Conclusions and Relevance: State same-sex marriage policies were associated with a reduction in the proportion of high school students reporting suicide attempts, providing empirical evidence for an association between same-sex marriage policies and mental health outcomes.

Regression Discontinuity for Causal Effect Estimation in Epidemiology.

Regression discontinuity analyses can generate estimates of the causal effects of an exposure when a continuously measured variable is used to assign the exposure to individuals based on a threshold rule. Individuals just above the threshold are expected to be similar in their distribution of measured and unmeasured baseline covariates to individuals just below the threshold, resulting in exchangeability. At the threshold exchangeability is guaranteed if there is random variation in the continuous assignment variable, e.g., due to random measurement error. Under exchangeability, causal effects can be identified at the threshold. The regression discontinuity intention-to-treat (RD-ITT) effect on an outcome can be estimated as the difference in the outcome between individuals just above (or below) versus just below (or above) the threshold. This effect is analogous to the ITT effect in a randomized controlled trial. Instrumental variable methods can be used to estimate the effect of exposure itself utilizing the threshold as the instrument. We review the recent epidemiologic literature reporting regression discontinuity studies and find that while regression discontinuity designs are beginning to be utilized in a variety of applications in epidemiology, they are still relatively rare, and analytic and reporting practices vary. Regression discontinuity has the potential to greatly contribute to the evidence base in epidemiology, in particular on the real-life and long-term effects and side-effects of medical treatments that are provided based on threshold rules - such as treatments for low birth weight, hypertension or diabetes.

Regression discontinuity designs are underutilized in medicine, epidemiology, and public health: a review of current and best practice.

OBJECTIVES: Regression discontinuity (RD) designs allow for rigorous causal inference when patients receive a treatment based on scoring above or below a cutoff point on a continuously measured variable. We provide an introduction to the theory of RD and a systematic review and assessment of the RD literature in medicine, epidemiology, and public health. STUDY DESIGN AND SETTING: We review the necessary conditions for valid RD results, provide a practical guide to RD implementation, compare RD to other methodologies, and conduct a systematic review of the RD literature in PubMed. RESULTS: We describe five key elements of analysis all RD studies should report, including tests of validity conditions and robustness checks. Thirty two empirical RD studies in PubMed met our selection criteria. Most of the 32 RD articles analyzed the effectiveness of social policies or mental health interventions, with only two evaluating clinical interventions to improve physical health. Seven out of the 32 studies reported on all the five key elements. CONCLUSION: Increased use of RD provides an exciting opportunity for obtaining unbiased causal effect estimates when experiments are not feasible or when we want to evaluate programs under "real-life" conditions. Although treatment eligibility in medicine, epidemiology, and public health is commonly determined by threshold rules, use of RD in these fields has been very limited until now.