Difference-in-differences (DiD) is a powerful, quasi-experimental research design widely used in longitudinal policy evaluations with health outcomes. However, DiD designs face several challenges to ensuring reliable causal inference, such as when policy settings are more complex. Recent economics literature has revealed that DiD estimators may exhibit bias when heterogeneous treatment effects, a common consequence of staggered policy implementation, are present. To deepen our understanding of these advancements in epidemiology, in this methodologic primer, we start by presenting an overview of DiD methods. We then summarize fundamental problems associated with DiD designs with heterogeneous treatment effects and provide guidance on recently proposed heterogeneity-robust DiD estimators, which are increasingly being implemented by epidemiologists. We also extend the discussion to violations of the parallel trends assumption, which has received less attention. Last, we present results from a simulation study that compares the performance of several DiD estimators under different scenarios to enhance understanding and application of these methods.

Objectives. To test whether paid family leave policies in California and New Jersey improved breastfeeding practices, overall and among key subgroups.Methods. We conducted difference-in-differences analyses, comparing pre–post policy changes in California and New Jersey with changes in states where no paid family leave policies were implemented. We examined a large, diverse sample of children born during 2001 to 2013 (n = 306 266), drawn from the 2003 to 2015 National Immunization Survey waves. Outcomes included ever breastfed, breastfed exclusively at 3 and 6 months, and still breastfed at 6 and 12 months, as well as duration of any breastfeeding and exclusive breastfeeding. We examined heterogeneity in policy response by maternal characteristics.Results. Paid family leave policies resulted in a modestly greater likelihood of exclusively breastfeeding at 6 months. Subgroup analyses were mixed, although several breastfeeding outcomes were consistently improved among married, White, higher-income, and older mothers.Conclusions. Exclusive breastfeeding improved after implementation of paid family leave policies in the overall sample, and additional benefits were noted for more advantaged mothers. This contributes critical evidence to an ongoing policy discussion, suggesting that subsequent paid family leave policies should be designed to target more vulnerable mothers.

Large increases in health care costs combined with an economic slowdown have created pressures for health plans and employers to reconsider cost containment strategies that were scaled back after the managed care backlash. In this paper we examine how plans' approaches to cost containment and care management have evolved since 2001. Plans reintroduced and refocused some utilization management techniques during 2002 and 2003 while continuing to invest in disease and case management. Some also began to experiment with new variants of managed care, including tiered provider networks and incentive-based provider payments. However, few respondents believed that these strategies alone would greatly reduce future costs.

Background: Zambia experienced declining aggregate fertility and increasing aggregate contraceptive use from 1990 to 2000. Yet, in rural Zambia, progress in family planning has lagged far behind the advances made in Zambia's urban areas. The contraceptive prevalence rate in Lusaka and other urban areas outstripped the rate in rural Zambia by nearly 25 percentage points (41.2 percent versus 16.6 percent) in 2001. The total fertility rate varied between urban and rural areas by 2.5 children (4.3 versus 6.9 children). This paper considers the urban-rural differentials in Zambia and assesses family planning outreach as a tool to narrow this divide. Methods: This study uses the Zambia Demographic and Health Survey (DHS) data, collected between 2001 and 2002. Logistic regression techniques were employed to examine factors associated with contraceptive use. The first analysis tested modern contraceptive use versus traditional method use and no use. In addition, separate models were run for samples stratified by type of residence (rural or urban) to determine if different factors were associated with use by residence. A simulation determined the effect of all women receiving at least one household visit from a health worker if all other variables were held constant. Results: Differences in modern contraceptive use between urban and rural areas persist (OR: 1.56, 95 percent CI: 1.24–1.96) even after adjusting for a number of demographic, socioeconomic, cognitive, and attitudinal factors. Household visits by a community health worker significantly increased the likelihood of modern contraceptive use among rural women (OR: 1.83; 95 percent CI: 1.29–2.58). If all rural women received at least one outreach visit per year, the prevalence rate for modern contraceptive methods would be expected to increase for this group by 5.9 percentage points, a marked increase but less than one-quarter of the total urban-rural differential. Conclusion: Outreach in the form of health worker visits can improve access to family planning services, but it does not eliminate barriers to access or address continued high-fertility desires in Zambia. Until policymakers consider strategies that address both family planning demand creation and supply of services, progress in Zambia and the rest of sub-Saharan Africa will continue to lag behind the rest of the world.

Introduction: Peer support has been recommended to promote smoking cessation, but results from prior meta-analyses have not established its efficacy. We conducted a systematic review and meta-analysis to assess current evidence and identify potential modifiers of efficacy. Methods: Randomized controlled trials of peer-support interventions with a smoking cessation outcome were identified in January 2022 from PubMed and references listed in identified studies. The meta-analysis outcome measure was mean risk ratio (RR, 95% confidence interval [CI]) for abstinence at the longest follow-up timepoint between 3 and 9 months from baseline. Potential modifiers tested were peer smoking status (former, current, or unknown), follow-up timepoint, abstinence measure, and cumulative engagement time between peers and smokers (“dose”). Studies were assessed for risk of bias and certainty of evidence. Results: We identified 16 trials, which varied in abstinence effect size (RR 0.61–3.07), sample size (23–2121), dose (41–207 minutes), and follow-up timepoint (<1–15 months). Across 15 trials with follow-up between 3 and 9 months (N = 8573 participants; 4565 intervention, 4008 control), the pooled Mantel-Haenszel RR was 1.34 (95% CI: 1.11–1.62). Effect sizes were greatest among interventions with formerly smoking peers (RR 1.43, 95% CI 1.17–1.74; five trials). We found positive effects for follow-up timepoints ≥3 months but no effect of intervention dose. The overall quality of evidence was deemed “very low.” Conclusions: Peer-support interventions increased smoking abstinence. There remains a lack of consensus about how to define a peer. Intervention features such as peer smoking status appear to have explanatory power. Additional high-quality and more comparable trials are needed. Implications: This study reviewed the latest evidence from randomized controlled trials and found that peer-support interventions enhance smoking cessation. Efficacy varies with key intervention features such as peer smoking status and follow-up timepoint, which may be used to facilitate development of more effective peer-support interventions. Future trials and reviews would benefit from careful consideration and clear reporting of peer smoking status, length of follow-up, abstinence measures, and intervention dose.