Difference-in-differences Estimate Research Articles

Costs of an environmental regulation in livestock farming: Evidence from pig production in rural China

AbstractWe investigate the cost and benefit of one of the most stringent Chinese environmental regulations that led to a shutdown of a large number of livestock farms. The temporal and spatial variation in programme implementation allows us to employ a staggered difference‐in‐difference (DID) to identify the causal effects of the regulation. Our DID estimates show that while the regulation significantly reduced NH3‐N, it has no significant effect on the other three important livestock related pollutants (pH, DO and COD). In contrast, the regulation consistently reduced the number of pigs slaughtered, inventory of live pigs and pork production by 8.3%, 10.3% and 11.2%, which alone is equivalent to a 2.9 percentage point loss of China's entire agricultural output value in 2016. Further analyses reveal evidence of partial substitution between pig and other livestock animals and the possibility of relocation to other regions. The policy‐induced reduction in export and increase in pork price is consistent with the reduction of pig and pork production. Overall, we find that the regulation achieved rather limited environmental benefit at a large economic cost.

The Effect of Pretrial Detention on Labor Market Outcomes

ObjectiveTo study the effect of pretrial detention on post-verdict labor market outcomes.MethodWe combine Chilean individual administrative data for criminal cases and labor market outcomes to estimate the effect of pretrial detention for first-time defendants on labor outcomes using the Difference-in-differences (DiD) method, controlling by individual fixed effects, and an instrumental variables (IV) approach. The IV approach takes advantage of the quasi-random assignment of judges.ResultsThe IV results show that pretrial detention reduces the probability of having formal employment and the average monthly wage by 39% and 56% during the six months following the final trial verdict. DiD estimation delivers estimates that are between one-third and one-half smaller. The magnitudes of the effects shown continue to be relevant as much as 24 months after the final trial verdict. The negative effect of pretrial detention also exists for accused individuals whose trial does not end with a custodial sentence. The results of our analysis suggest that the negative effect of pretrial detention is (at least) driven by the lasting effect of being excluded from the labor market during the trial, the accompanying social stigma, and the impact of pretrial detention on the probability of post-verdict incarceration.ConclusionDespite the relevant and lasting effects of pretrial detention on labor market performance, the evidence suggests that individuals in both the control and treatment groups keep the same labor dynamics pretrial and posttrial if they return to work immediately after its conclusion. Therefore, conditional on the decision to keep pretrial detention as an ongoing policy, a possible avenue to attenuate the negative effect on labor outcomes is to design public policies that support access to the labor market and finding an employment position immediately following pretrial detentions and trial proceedings.

Investigating Applications of Prognostic Scores to Balancing Weights: Evidence from an Evaluation of Accountable Care Organizations

Research ObjectivePropensity score methods, based on the predicted probability of treatment, are ubiquitous in health services research. Over the past decade, an increasing amount of research has investigated applications of prognostic scores: the predicted probability of the study outcome of interest (e.g., illness, mortality). However, this research has focused on the use of prognostic scores in matching, either in conjunction with, or in place of, more commonly‐implemented propensity score matching approaches. There has been little, if any, research focused on the use of prognostic scores with other analytic strategies such as weighting or subclassification typically associated with propensity scores. Prior research has also focused on prognostic scores for binary outcomes. This study presents several analytic strategies based on a continuous pseudo‐prognostic score (P‐PS) and assesses their practical value relative to a public evaluation of a Medicare value‐based payment model.Study DesignWe used Medicare claims and enrollment data covering all 41 Medicare accountable care organizations (ACOs) that participated in the ACO Investment Model (AIM): the same data used to evaluate the first AIM performance year. The evaluation's primary outcome of interest was total Medicare spending, per‐beneficiary‐per‐month. We estimated a regression model for total spending using each ACO's comparison group and predicted the P‐PS (i.e., predicted spending) for intervention and comparison beneficiaries using the model's parameters.While the original evaluation used entropy‐balancing weights to balance all covariates between intervention and comparison beneficiaries within each ACO, we calculated three types of enhanced entropy‐balance weights based on the P‐PS. We balanced intervention and comparison beneficiaries on all covariates and P‐PS, and then incorporated a quadratic P‐PS as well. Lastly, we classified beneficiaries into five quintiles of P‐PS, and separately balanced covariates within each quintile.We assessed the effect of the enhanced weights on the magnitude and precision of corresponding impact estimates, as well as the validity of the parallel trend assumption underlying the difference‐in‐differences (DID) evaluation design, relative to the original weights.Population StudiedBeneficiaries with fee‐for‐service Medicare attributed to ACOs in AIM and corresponding comparison beneficiaries eligible for attribution in 2013–2016.Principal FindingsIncorporating P‐PS directly into the weights did not meaningfully affect estimated differences in baseline parallel trends or impact estimates. However, across the 41 ACOs, weighting by covariates within PPS quintiles reduced average absolute baseline differences by 14.7% at the mean and 21.2% at the median. These differences shifted the mean impact estimate by 6.7%, while increasing the mean standard error of the impact estimates by 2.3%.ConclusionsBalancing covariates within P‐PS quintiles improved the validity of DID estimates by substantially reducing differences in baseline trends, and meaningfully reduced bias in the subsequent impact estimates. Bias reduction was greater than the loss in precision. Incorporating P‐PS directly into the weights did not meaningfully affect baseline or impact estimates.Implications for Policy or PracticeBalancing covariates within prognostic score quintiles shows promise for reducing bias in observational study designs. Unlike matching‐based prognostic score approaches, this method can be implemented when outcomes of interest are continuous, and does not require discarding unmatched data.Primary Funding SourceCenters for Medicare and Medicaid Services.

Effects of the Affordable Care Act Medicaid Expansion on in‐Hospital Mortality, Access to Rehabilitation, and Readmission Rates Among Young Adult Trauma Patients

Research ObjectiveTrauma is the leading cause of death and disability among young adults. Prior to the implementation of the Affordable Care Act (ACA), more than 30% of young adult trauma patients were uninsured, and uninsured patients had higher in‐hospital mortality rates and worse access to rehabilitation. We aimed to assess the impact of the ACA Medicaid expansion on insurance coverage and rates of in‐hospital mortality, access to rehabilitation, and unplanned readmissions and return emergency department (ED) visits in young adult trauma patients. Expansion States Non‐expansion States Difference‐in‐Differences Estimate (95% CI) P 2011–2013(N = 56,196) 2014–2017 (N = 62,893) 2011–2013 (N = 104,344) 2014–2017 (N = 143,595) Uninsured 27.1 9.7 32.7 31.8 −16.50 (−17.08 to −15.93) <0.001 In‐hospital mortality 2.22 2.03 2.04 1.78 0.07 (−0.34 to 0.47) 0.74 Discharged to inpatient rehabilitation 6.18 8.07 5.91 6.47 1.33 (0.68 to 1.97) <0.001 Discharged to a skilled nursing facility 1.97 2.24 1.76 1.61 0.42 (0.11 to 0.73) 0.008 Discharged to home healthcare 8.25 7.85 7.77 7.59 −0.22 (−1.51 to 1.07) 0.74 30‐day unplanned readmission 7.47 7.88 8.09 8.20 0.30 (−0.70 to 1.31) 0.55 30‐day return ED visit 18.31 18.96 21.58 22.67 −0.44 (−1.66 to 0.78) 0.48 Risk‐adjusted marginal percentages are shown. Study DesignWe used the state inpatient databases of 5 states that expanded Medicaid through the ACA (Arkansas, Iowa, Illinois, Kentucky, and Maryland) and 5 states that did not (Florida, Georgia, Kansas, Missouri, North Carolina) to perform a difference‐in‐difference (DD) analysis comparing changes in insurance coverage and risk‐adjusted outcomes in young adults hospitalized for traumatic injuries before (2011–2013) and after (2014–2017) Medicaid expansion and open enrollment.Population StudiedWe included adults aged 19–44 years who were hospitalized for traumatic injury in their state of residence. Patients admitted electively or for burn injuries were excluded.Principal FindingsOf the 367,028 trauma patients included, 26.7% were women and the mean age was 30.8 years. As shown in the table, Medicaid expansion was associated with a more than 16 percentage point decrease in the uninsured rate. This drop increased over time, with the annual DD estimates for 2014–2017 being −11.9, −17.4, −17.8, and − 19.3 percentage points respectively. Medicaid expansion was not associated with decreases in in‐hospital mortality or 30‐day readmission or return ED visit rates. It was associated with increases in access to inpatient rehabilitation, and this access grew over time (annual change in the DD: 0.8 percentage points, p < 0.001).ConclusionsAmong young adult trauma patients, Medicaid expansion increased insurance coverage and access to inpatient rehabilitation, and these increases have grown over time. It has not reduced in‐hospital mortality or readmission rates.Implications for Policy or PracticeMedicaid expansion has enabled more young adults with serious injuries to access inpatient rehabilitative care, likely resulting in improved functional outcomes. However, given the unexpected nature of traumatic injury and the long recovery period that is sometimes required, these benefits may be eliminated if Medicaid waivers adding work requirements or eliminating retroactive Medicaid eligibility go into effect.Primary Funding SourceNational Institutes of Health.

Socioeconomic inequalities in low birth weight risk before and during the COVID-19 pandemic in Argentina: A cross-sectional study

BackgroundThe coronavirus disease 2019 (COVID-19) pandemic may have exacerbated existing socioeconomic inequalities in health. In Argentina, public hospitals serve the poorest uninsured segment of the population, while private hospitals serve patients with health insurance. This study aimed to assess whether socioeconomic inequalities in low birth weight (LBW) risk changed during the first wave of the COVID-19 pandemic.MethodsThis multicenter cross-sectional study included 15929 infants. A difference-in-difference (DID) analysis of socioeconomic inequalities between public and private hospitals in LBW risk in a pandemic cohort (March 20 to July 19, 2020) was compared with a prepandemic cohort (March 20 to July 19, 2019) by using medical records obtained from ten hospitals. Infants were categorized by weight as LBW < 2500 g, very low birth weight (VLBW) < 1500 g and extremely low birth weight (ELBW) < 1000 g. Log binomial regression was performed to estimate risk differences with an interaction term representing the DID estimator. Covariate-adjusted models included potential perinatal confounders.FindingsOf the 8437 infants in the prepandemic cohort, 4887 (57•9%) were born in public hospitals. The pandemic cohort comprised 7492 infants, 4402 (58•7%) of whom were born in public hospitals. The DID estimators indicated no differences between public versus private hospitals for LBW risk (−1•8% [95% CI −3•6, 0•0]) and for ELBW risk (−0•1% [95% CI −0•6, 0•3]). Significant differences were found between public versus private hospitals in the DID estimators (−1•2% [95% CI, −2•1, −0•3]) for VLBW risk. The results were comparable in covariate-adjusted models.InterpretationIn this study, we found evidence of decreased disparities between public and private hospitals in VLBW risk. Our findings suggest that measures that prioritize social spending to protect the most vulnerable pregnant women during the pandemic contributed to better birth outcomes.FundingNo funding was secured for this study.

Difference-in-differences with variation in treatment timing

The canonical difference-in-differences (DD) estimator contains two time periods, ”pre” and ”post”, and two groups, ”treatment” and ”control”. Most DD applications, however, exploit variation across groups of units that receive treatment at different times. This paper shows that the two-way fixed effects estimator equals a weighted average of all possible two-group/two-period DD estimators in the data. A causal interpretation of two-way fixed effects DD estimates requires both a parallel trends assumption and treatment effects that are constant over time. I show how to decompose the difference between two specifications, and provide a new analysis of models that include time-varying controls.

A Women’s Group-Based Nutrition Behavior Change Intervention in India Has Limited Impacts Amidst Implementation Barriers and a Concurrent National Behavior Change Campaign

Abstract Objectives Women's self-help groups (SHGs) have become one of the world's largest institutional platforms of the poor, reaching over 70 million Indian women in 2020. Limited evidence exists on effects of nutrition interventions through SHGs on maternal and child nutrition outcomes. Methods The Women Improving Nutrition through Group-based Strategies (WINGS) study was a quasi-experimental impact evaluation, comparing 16 matched blocks where communities were provided support to form SHGs and improve women's livelihoods; 8 blocks (1 in each matched pair) received a 3-year nutrition intervention (NI) with nutrition education, agriculture- and rights-based information, facilitated by a trained female volunteer; the other 8 blocks received standard activities (STD) to support savings &amp; livelihoods. We conducted repeated cross-sectional surveys of mother-child pairs in 2017–18 (n = 1609 pairs) and 2019–20 (n = 1841 pairs). We matched treatment groups over time and applied difference-in-difference (DID) regression models to estimate NI impacts. Outcomes were knowledge domains (nutrition for pregnant women, breastfeeding, complementary feeding, child health), child feeding (e.g., early breastfeeding initiation, dietary diversity, animal source food (ASF) consumption), woman's diets, woman's BMI and child anthropometry. Matching covariates included woman, child, and community characteristics. Results About 40% of women were SHG members. Nutrition intervention exposure was low; only ∼10% of NI women had heard of intervention content at endline. There were large improvements in women's knowledge in both groups. DID estimation revealed a positive NI impact on knowledge of timely introduction of ASFs to children (P &amp;lt; 0.05), but knowledge of nutrition for pregnant women unexpectedly improved more in the STD group (P &amp;lt; 0.05). No impacts were observed for any anthropometry or diet indicators except child ASF consumption (P &amp;lt; 0.01). Conclusions Limited impacts on nutrition outcomes may be due to limited exposure, low motivation or skills of volunteers, and a concurrent national nutrition behavior change program targeting mothers and children in all study areas. Our findings add to a growing literature on SHG-based behavior change interventions and the conditions necessary for their success. Funding Sources Bill &amp; Melinda Gates Foundation.

A Nutrition Behaviour Change Intervention Delivered Through Women’s Self-Help Groups in India Is Protective Against Depression and Reduces Time Spent in Market Work

Abstract Objectives Women's self-help groups (SHGs), which operate at large scale in India, are an important platform for delivering behaviour change communication (BCC) and social support interventions to rural women. Little is known about how such group-based interventions affect women's mental health and time use. Methods The Women Improving Nutrition through Group-based Strategies (WINGS) study was a quasi-experimental impact evaluation, comparing 16 blocks (8 matched pairs) with SHG formation support; 8 blocks received a 3-year nutrition intervention (NI) with BCC topics such as nutrition, home-gardens and women's well-being, facilitated by a trained female volunteer; the other 8 received standard activities (STD) to support savings &amp; livelihoods. We conducted repeated cross-sectional surveys of mother-child pairs in 2017–18 (n = 1609) and 2019–20 (n = 1841). We matched treatment groups over time and applied difference-in-difference (DID) regression models to estimate NI impacts. Outcomes assessed: (1) common mental disorder symptoms (CMD) (Self Reporting Questionnaire (SRQ) score, 8 or higher) and (2) time use, constructed using 24-hour recall data. Time indicators were the proportion of time spent on productive work (employed, agricultural work), reproductive work (cooking, caring for children etc.), and time spent on social-leisure activities (hobbies, socializing). Results Overall, women were 25 years old with 5 years of education and worked 10.7 hours/day. CMD were reported by 17% of women. DID estimates showed that CMD prevalence doubled over time among women in STD areas but did not change in NI areas (P &amp;lt; 0.01). Compared to STD areas, women in NI areas reported a larger decrease in time spent on productive work (DID: −5 percentage points (pp); P &amp;lt; 0.01) and larger increases in time spent on reproductive work (DID: +5 pp; P &amp;lt; 0.01) and on social-leisure activities (DID: +22 minutes, P &amp;lt; 0.01). Conclusions A BCC intervention delivered through SHGs in rural India protected against a secular trend in declining mental health and shifted women's time from market work to domestic and social-leisure activities. These findings add to a growing evidence base on the effectiveness of group-based interventions to improve women's wellbeing in developing countries. Funding Sources Bill &amp; Melinda Gates Foundation.

The effects of an air quality alert program on premature mortality: A difference-in-differences evaluation in the region of Paris.

Daily exposure to air pollution has been shown to increase cardiovascular and respiratory mortality. While increases in short-term exposure to air pollutants at any daily concentrations has been shown to be associated to adverse health outcomes, days with extreme levels, also known as air pollution peaks based on specific thresholds, have been used to implement air quality alerts in various cities across the globe. We aimed at evaluating the potential effects of the Air Quality Alerts (AQA) system on different causes of premature mortality in Paris, France. Air quality alerts (AQA) based on particulate matter (PM10) levels and related interventions were implemented in the region of Paris in 2008 and were revised to be more stringent in 2011. In this study, we applied a difference-in-differences (DID) approach coupled with propensity-score matching (PSM) to daily mortality data for the period 2000 to 2015 to evaluate the effects of the Paris AQA program on different causes of premature mortality for the entire population and for adults>75years old. Overall, results did not show evidence of a reduction in mortality of the PM10 AQA program when first implemented in 2008 with initial thresholds (80µg/m3); DID estimates were slightly above 1 for cardiovascular and respiratory mortality. However, when evaluating the drastic reduction in revised thresholds in 2011 (50µg/m3) to trigger interventions, we identified a reduction in cardiovascular (DID=0.84, 95% CI: 0.755 to 0.930) mortality, but no change in respiratory mortality was detected (DID=0.97, 95% CI: 0.796, 1.191). Our study suggests that AQA may not have health benefits for the population when thresholds are set at high daily PM10 levels. Given that such policies are implemented in many other metropolitan areas across the globe, evaluating the effectiveness of AQA is important to provide public authorities and researchers a rationale for defining specific thresholds and extending the scope of these policies to lower air pollution levels.

Impact of Turkish policy reform on labor force status of disabled males: A difference-in-difference analysis

This paper aims to study the effect of a Turkish policy reform enacted in 2008 that requires firms to hire disabled applicants. Our attention is only on males to avoid complications arising from gender differences in disability and labor force participation. The data is from the Turkey Health Survey (THS) of the Turkish Statistical Institute (TurkStat) for the years 2008 and 2012. We define ?disability? as an impairment of long-term health conditions that lasts more than six months and that restricts the individual in daily activities. We use difference-in-difference (DD) estimation, in which the DD estimator is the difference between disabled and non-disabled individuals in the difference in labor force participation before and after the new policy. The results suggest an insignificant effect of the treatment on the treated, implying that the policy reform does not create any incentive for disabled males to participate in the labor force.

Is There An Association Between Bundled Payments and "Cherry Picking" and "Lemon Dropping" in Orthopaedic Surgery? A Systematic Review.

Is There An Association Between Bundled Payments and "Cherry Picking" and "Lemon Dropping" in Orthopaedic Surgery? A Systematic Review.

Impact of family law reform on adolescent reproductive health in Ethiopia: A quasi-experimental study

• Ethiopia passed the Revised Family Code in 2000, which set the minimum age of marriage for girls at 18 and strengthened women’s rights within marriage. • Analysis using synthetic control method and difference-in-differences helps to overcome empirical challenges of identifying impact of the law. • Implementation of the law was associated with large reductions in rates of adolescent birth, child marriage, and early sexual initiation. • Reductions were similar in rural and urban areas. • Findings indicate that strong legal frameworks for gender equality may be effective catalysts in facilitating social change on child marriage. In sub-Saharan Africa, nearly 2 in 5 girls are married before the age of 18. Child marriage has adverse consequences for women’s physical, emotional, and social wellbeing and development. Emerging evidence has identified the importance of gender equality as a social context for shaping adolescent sexual and reproductive health norms. In 2000, Ethiopia’s government passed the Revised Family Code, increasing the minimum age of marriage from 15 to 18 for girls without exceptions, and strengthening women’s rights within marriage regarding marital property, divorce, and employment. I evaluated the impact of the law on adolescent reproductive health indicators and newborn mortality rates. I used difference-in-differences (DID) and synthetic control methods (SCM) to compare cohort trends in Ethiopia to those in comparison countries. I show that implementation of the law was associated with a 9-percentage-point reduction in risk of adolescent birth for exposed cohorts, an 8-percentage-point reduction in child marriage, and a 10-percentage-point reduction in sexual initiation before age 18 in SCM models. There was no association of the law with changes in risk of termination of pregnancy, unmet need for contraception, infant mortality rates, or neonatal mortality rates. Results were consistent across SCM and DID models, although DID estimates were slightly attenuated. I discuss mechanisms and policy implications. The results of this study provide evidence that strong legal frameworks for gender equality may be effective catalysts in facilitating social change around child marriage.

Inclusive Governance and Biodiversity Conservation: Evidence from Sub-Saharan Africa

We examine the conservation effects attributable to changes in the size of community-governed protected areas (PAs) by adopting a generalized difference-in-difference (DID) design with a two-way fixed effect regression model and synthetic control methods. Panel data from the extraordinary datasets of the World Database on Protected Areas (WDPAs) and the Red List of International Union for the Conservation of Nature (IUCN) are used for 32 Sub-Saharan African countries in this study. Our generalized DID estimates show that countries with community-governed PAs have reduced the IUCN Red List threat level by 17% for mammals. We also find stronger evidence of the effect of community-governed PAs on the IUCN threat level using synthetic control method that allows us to match the “intervention countries” with those countries that exhibit similar pre-intervention threat level. Our results are robust on alternate specifications in which we exploit variations in the cumulative size of the designated PAs differentiated by the IUCN governance types. We also compare the effect of strictly state-governed PAs with community-governed PAs. Our findings provide evidence in support of recent qualitative studies that find positive responses of community participation towards common goods that carry potential economic incentives. This paper contributes to the idea that inclusive environmental policies and legislations yield environmental gains not at the cost of social exclusion.

Racial/Ethnic Disparities in Very Preterm Birth and Preterm Birth Before and During the COVID-19 Pandemic

The coronavirus disease 2019 (COVID-19) pandemic may exacerbate existing racial/ethnic inequities in preterm birth. To assess whether racial/ethnic disparities in very preterm birth (VPTB) and preterm birth (PTB) increased during the first wave of the COVID-19 pandemic in New York City. This cross-sectional study included 8026 Black, Latina, and White women who gave birth during the study period. A difference-in-differences (DID) analysis of Black vs White disparities in VPTB or PTB in a pandemic cohort was compared with a prepandemic cohort by using electronic medical records obtained from 2 hospitals in New York City. Women who delivered from March 28 to July 31, 2020, were considered the pandemic cohort, and women who delivered from March 28 to July 31, 2019, were considered the prepandemic cohort. Reverse transcription-polymerase chain reaction tests for the presence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) were performed using samples obtained via nasopharyngeal swab at the time of admission. Clinical estimates of gestational age were used to calculate VPTB (<32 weeks) and PTB (<37 weeks). Log binomial regression was performed to estimate Black vs White risk differences, pandemic cohort vs prepandemic cohort risk difference, and an interaction term representing the DID estimator. Covariate-adjusted models included age, insurance, prepregnancy body mass index, and parity. Of 3834 women in the pandemic cohort, 492 (12.8%) self-identified as Black, 678 (17.7%) as Latina, 2012 (52.5%) as White, 408 (10.6%) as Asian, and 244 (6.4%) as other or unspecified race/ethnicity, with approximately half the women 25 to 34 years of age. The prepandemic cohort comprised 4192 women with similar sociodemographic characteristics. In the prepandemic cohort, VPTB risk was 4.4% (20 of 451) and PTB risk was 14.4% (65 of 451) among Black infants compared with 0.8% (17 of 2188) VPTB risk and 7.1% (156 of 2188) PTB risk among White infants. In the pandemic cohort, VPTB risk was 4.3% (21 of 491) and PTB risk was 13.2% (65 of 491) among Black infants compared with 0.5% (10 of 1994) VPTB risk and 7.0% (240 of 1994) PTB risk among White infants. The DID estimators indicated that no increase in Black vs White disparities were found (DID estimator for VPTB, 0.1 additional cases per 100 [95% CI, -2.5 to 2.8]; DID estimator for PTB, 1.1 fewer case per 100 [95% CI, -5.8 to 3.6]). The results were comparable in covariate-adjusted models when limiting the population to women who tested negative for SARS-CoV-2. No change was detected in Latina vs White PTB disparities during the pandemic. In this cross-sectional study of women who gave birth in New York City during the COVID-19 pandemic, no evidence was found for increased racial/ethnic disparities in PTB, among women who tested positive or tested negative for SARS-CoV-2.

Introduction of digital reporting platform to integrate community-level data into health information systems is feasible and acceptable among various community health stakeholders: A mixed-methods pilot study in Mopti, Mali.

BackgroundIntegration of community-level health data within Mali’s web-based District Health Information System (DHIS2) is underexplored. This study conducted in Mopti, Mali examined challenges and enablers affecting integration and investigated how digital technology optimizes data quality, availability, and use.MethodsThis pre-post mixed-methods study compared community health workers’ (CHWs’) experiences reporting routine community-level data using the DHIS2 digital application on tablets and paper forms. 141 CHWs participated in quantitative surveys and focus group discussions at baseline and endline. In-depth interviews were conducted with 18 and eight CHW supervisors and 12 and 11 other stakeholders at baseline and endline, respectively. We calculated changes in CHW performance, and job satisfaction among intervention and comparison groups using the difference-in-difference (DID) estimator controlling for baseline characteristics. Routine longitudinal DHIS2 data described timeliness and completeness of CHW reporting. Thematic analysis of qualitative data explained implementation challenges and enablers, and considerations for data use.ResultsThe median number of health talks and household visits among intervention group CHWs increased from baseline to endline compared to the comparison group (DID estimator; P < 0.05 for both), as did aspects of job satisfaction (satisfaction with opportunities to contribute ideas to improve services and coordination with supervisors and stakeholders, P < 0.1). CHWs reported high levels of comfort and confidence navigating the tablet for data collection and on-time reporting. While CHWs experienced challenges –tablet quality, limited network connection and increased workload, they preferred the digital to paper system. Stakeholder, supervisor and CHW roles in data review and decision-making appear unchanged from baseline to endline, though some supervisors found the tablets improved data timeliness and completeness. Routine longitudinal DHIS2 data confirm high rates of data timeliness and completeness before and after the intervention, with little or no change over time.ConclusionsCHW tablet use for data collection and reporting is feasible and desirable, however, program and policy changes are needed for this to be a fully-functional system. Future efforts need to consider how to ensure site-level network connectivity; quality, compatibility and functionality of digital technology; and routine supportive systems for CHWs and community health actors on data use.

Market-oriented extension and farming efficiency in small-scale maize farmers: evidence from Northern Vietnam

PurposeNorthern Vietnam is a mountainous area where most farmers' livelihoods depend on maize production. Beside training in farming techniques, since 2013 extension courses on market access have been gradually delivered to maize farmers to help them save production costs and increase earnings.Design/methodology/approachBy applying robust regression and Tobit models to estimate technical efficiency (TE), the difference-in-difference (DID) technique is used in this study to determine the change in the TE of 1,035 farmers in the period 2013–2015 in Northern Vietnam.FindingsThe DID estimator indicated the initial success of extension courses by showing a faster increase in the TE of the receivers compared to that of non-receivers. However, co-operation in production and marketing by maize farmers and access to micro-credit resulted in an insignificant impact on the TE of maize farmers in 2015. These low impact results highlighted the inefficiency of the extension and micro-credit programmes and that raised the concern of policymakers. The distance to local market signifies the determination of farmers on their crops affected by how close their location to their input/output marketplaces.Research limitations/implicationsThe Land Policy of the government should take into careful consideration measures to exploit all potential production and to mitigate poverty as well as foster the development of the region.Originality/valueThe expansion of extension courses on access to market has been recognised as an effective method to increase farmer's efficiency.

Socioeconomic and physical health status changes after visual impairment in Korea using difference-in-difference estimations

This study analyzed the impact of visual impairment on socioeconomic and physical health status and its heterogeneity by severity of visual impairment. We used nationally representative cohort data based on Korean national health insurance claims (2002–2013), which were extracted for 11,030 persons (2206 visually impaired, 8824 control). This was restructured as monthly data for each person (person-month). Multivariate and ordered logistic regressions were conducted, and the pre-impairment status between the visually impaired and non-visually impaired people was adjusted by difference-in-difference (DiD) estimation. Focusing on medical aid (a public healthcare service assistance program for people who cannot afford health insurance premiums), the DiD estimate showed that the likelihood of receiving aid was higher among visually impaired compared with non-impaired people. Mildly and severely visually impaired people were more likely to be medical aid recipients than their counterparts. The severely visually impaired group was more likely to be unemployed. The visually impaired group were less likely to have no comorbidity. Our findings show that the socioeconomic and physical health status of visually impaired people is more likely to deteriorate than that of their non-visually impaired counterparts following onset of impairment.

Changes in Health Care Use Among Undocumented Patients, 2014-2018

The 2016 presidential campaign was marked by intensified rhetoric around the deportation of undocumented immigrants. The association of such rhetoric with primary, emergency, and inpatient care among undocumented immigrants is unclear. To examine the association of increased anti-immigrant rhetoric during the 2016 presidential campaign with health care use among a group of Medicaid-ineligible patients largely composed of undocumented immigrants. Using a difference-in-differences (DID) approach, this cohort study analyzed health care use between January 1, 2014, and May 31, 2018, in a retrospective cohort of Medicaid and Medicaid-ineligible (>90% undocumented) adult and pediatric patients. The inflection point of interest was June 16, 2015, the date of Donald Trump's announcement of candidacy, which represented a documented increase in anti-immigration rhetoric during the presidential campaign. Analyses were controlled for age, self-reported sex, and baseline comorbidities. Data analysis was conducted from August 28, 2018, to September 1, 2020. The DID of the number of completed primary care encounters before and after June 16, 2015, in Medicaid compared with Medicaid-ineligible patients. Secondary outcomes included the DID of emergency department (ED) visits and inpatient discharges over the same period. There were 20 211 patients included in the analysis: 1501 (7.4%) in the sample of predominantly undocumented Medicaid-ineligible patients (861 [57.4%] female) and 18 710 (92.6%) in the Medicaid control group (10 443 [55.8%] female). The mean (SD) age as of 2018 in the Medicaid-ineligible group was 38.2 (15.4) years compared with 22.2 (16.5) years in the control group. There was a differential decrease in completed visits among Medicaid-ineligible children compared with Medicaid children (DID estimate, 0.8; 95% CI, 0.7-0.9) and Medicaid-ineligible adults (DID estimate, 0.8; 95% CI, 0.8-0.9). There was also a significant differential increase in ED visits among Medicaid-ineligible children (DID estimate, 2.3; 95% CI, 1.1-5.0). In addition, there was a differential decrease in inpatient discharges among Medicaid-ineligible adults (DID estimate, 0.5; 95% CI, 0.4-0.7), with no significant change in ED visits or ED admission rates in this group. In this cohort study, there was a significant decrease in primary care use among undocumented patients during a period of increased anti-immigrant rhetoric associated with the 2016 presidential campaign, coincident with an increase in ED visits among children and a decrease in inpatient discharges among adults, with the latter possibly attributed to a decrease in elective admissions during this period.

Abstract SS1-10: Impact of Medicaid expansion on racial disparities in time to adjuvant chemotherapy administration among breast cancer (BC) patients

Abstract BACKGROUND: Delays in adjuvant chemotherapy initiation have a detrimental effect in outcomes and are more frequently seen among racial/ethnic minorities, uninsured patients and those with low socioeconomic status. A provision in the Affordable Care Act called for expansion of Medicaid eligibility in order to cover more low-income Americans. In this study we evaluate the impact of Medicaid expansion on racial disparities in time to adjuvant chemotherapy administration in a large cohort of BC patients.METHODS: Women (aged 40-64) diagnosed with primary invasive BC (stage I-III) between 01/01/2007 and 12/31/2016 were identified in the National Cancer Database. All patients underwent surgery as initial treatment modality and received adjuvant chemotherapy within 6 months of surgery. Chemotherapy delay was defined as &amp;gt;60 days from surgery to the first dose of chemotherapy. The cohort was limited to those residing in states that underwent Medicaid expansion in 2014 (KY, NV, CO, OR, NM, WV, AR, RI, AZ, MD, MA, ND, OH, IA, IL, VT, HI, NY, DE). For comparison purposes, 2007-2013 was considered the pre-expansion period and 2014-2016 the post-expansion period. We calculated difference-in-difference (DID) estimates using multivariable linear regression models. Results are presented as adjusted differences (in % points) between race/ethnicity groups in the pre and post-expansion periods, and as adjusted DID with 95% CI. A negative DID estimate indicates that Medicaid expansion reduces the racial/ethnic disparity in chemotherapy administration delay. The parallel trend assumption was tested. Variables in the final model included age, comorbidities, BC subtype, insurance status, distance to treatment facility, region, education, household income, facility type and facility case volume. RESULTS: 105,385 patients were included (median age 53), of them 75,663 (71.8%) were diagnosed in the pre- and 29,722 (28.2%) in the post-expansion period. 77.5% of the patients were White, 11.7% Black and 4.9% Hispanic. The proportion of patients experiencing chemotherapy initiation delays was greater among Blacks and Hispanics in both study periods compared to Whites. The proportion of chemotherapy delays decreased for all races between the pre and the post-expansion period. We observed a statistically significant decrease in the chemotherapy initiation racial disparity. The adjusted DID was -2% (95%CI -3.8 to -0.3, P=0.02) between Whites and Blacks was and -3% (95%CI -5.5 to -0.5, P=0.02) between Whites and Hispanics. In a subgroup analysis among 10,777 Medicaid patients the parallel trend assumption was not valid, therefore we estimated the DID between Whites and Blacks using a propensity-score based weight sample, identifying a reduction in the racial disparity of even greater magnitude (DID -9.1% [-14.4 to -3.8, P&amp;lt;0.001]). CONCLUSIONS: We demonstrate that Medicaid expansion reduced racial disparities by decreasing the proportion of Blacks and Hispanics experiencing delays in adjuvant chemotherapy initiation and decreasing the gap that exists when compared to Whites. These important results highlight the positive impact of policies aimed at improving equity and increasing access to health care. Pre-ExpansionPost-ExpansionRaceNAdjusted Rate of Delayed Chemo &amp;gt; 60 daysAdjusted Difference Between White Race (% points)NAdjusted Rate of Delayed Chemo &amp;gt; 60 daysAdjusted Difference Between White Race, (%points)Adjusted DID (95% CI), Percentage PointsPWhite5930527.202232524.20RefRefBlack856036.18.9372031.16.9-2.0 (-3.8 to -0.3)0.02Hispanic343232.35.1170426.32.1-3.0 (-5.5 to -0.5)0.02 Citation Format: Mariana Chavez-MacGregor, Xiudong Lei, Hui Zhao, Tina Shih, Sharon H Giordano. Impact of Medicaid expansion on racial disparities in time to adjuvant chemotherapy administration among breast cancer (BC) patients [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr SS1-10.

Impact of a telemonitoring intervention in patients with chronic heart failure in Germany: A difference-in-difference matching approach using real-world data

Introduction The aim of this study was to evaluate the effects of a non-invasive telemonitoring intervention on mortality, healthcare costs, and hospital and pharmaceutical utilisation in patients with chronic heart failure (CHF) of a large statutory health insurer in Germany. Methods In a retrospective observational cohort study using real-world data, we assessed differences between 635 patients who received a telemonitoring intervention versus 635 receiving usual care covering 36 months after intervention. We used propensity score matching on a set of 102 parameters collected in the 24-month pre-intervention period to correct for observed differences, as well as difference-in-difference (DiD) estimators to account for unobserved differences. We analysed the effect of the intervention for up to three years on (i) all-cause mortality; (ii) costs (i.e. inpatient stays, ambulatory care, pharmaceuticals, and medical aids and appliances); and (iii) healthcare utilisation (i.e. length and number of hospital stays, number of prescriptions). Results DiD estimates suggest lower inpatient costs of the telemonitoring group of up to €1160 (95% confidence interval (CI): –2253 to –69) in year three. Ambulatory care costs increased significantly in all three years up to €316 (95% CI: 1267 to 505) per year. Telemonitoring had a positive effect on survival (hazard ratio = 0.71; 95% CI: 0.51 to 0.99) and increased the number of prescriptions for diuretics. Effects were more prominent for patients with severe CHF. Discussion The study suggests that the telemonitoring intervention led to a significant decrease in mortality and a shift in costs from the inpatient to the ambulatory care sector 36 months after intervention.