Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sadia Farzana has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Supplementary Material for "Pfizer versus Moderna COVID-19 Vaccine Effectiveness against Mortality: Evidence from Linked Mortality and Vaccination Records in Indiana"

    by Sadia Farzana, Francesco Maria Rossi, Qian Luo, Jeff Whittle, Kevin McGUrk, Benjamin W. Weston, Ye Yuan, Ali Moghtaderi, Vladimir A. Atanasov, Bernard S. Black

    This Supplementary Material provides additional results and methods details for Farzana et al. (2026), Pfizer versus Moderna COVID-19 Vaccine Effectiveness against Mortality: Evidence from Linked Mortality and Vaccination Records in Indiana.

  • Open Access

    Pfizer Versus Moderna COVID-19 Vaccine Effectiveness Against Mortality: Evidence from Linked Mortality and Vaccination Records in Indiana

    by Sadia Farzana, Francesco Maria Rossi, Qian Eric Luo, Jeff Whittle, Kevin McGurk, Benjamin W. Weston, Andy Ye Yuan, Ali Moghtaderi, Vladimir Atanasov, Bernard S. Black

    Objective: It is commonly believed in the public health community that the mRNA vaccines from Pfizer and Moderna were broadly similar in preventing COVID-19 mortality, with perhaps a small advantage for Moderna. We revisit this question, using a research design that controls for healthy-vaccinee bias. Methods: We studied deaths among persons aged 15 years or greater from January 2021 through December 2022 in Indiana, linked to vaccination records, using an outcome measure, namely COVID-19 Excess Mortality Percentage (CEMP), which uses non-COVID natural mortality to control for HVB. Results: We find both large healthy-vaccinee bias for both vaccines, and larger bias for Pfizervaccinees: Pfizer recipients were healthier and thus would have faced lower COVID-19 mortality without vaccination. For ages 60+, we find almost 50% higher COVID-19 remaining mortality risk (RMR) for Pfizer vaccinees receiving primary vaccination over a “core period” from 2Q2021 through 1Q2022: the average Pfizer/Moderna RMR ratio was 1.49. For ages 15–59, this ratio was 1.25, but was not statistically significant. For booster dose for ages 60+, the average Pfizer/Moderna RMR ratio was 1.34. For 2Q2022 through 4Q2022, after the major Omicron wave, we find much lower vaccine effectiveness and no significant Pfizer–Moderna difference. Conclusions: During the period with both good vaccine availability and high COVID-19 mortality (2Q-2021–1Q2022), the Moderna vaccine was associated with lower COVID-19 mortality, relative to the Pfizer vaccine. A possible explanation is the higher Moderna dose (100 μg vs. 30 for Pfizer). After that period, a high percentage of the population had natural immunity from prior COVID-19 infection; we find only small mortality gains from vaccination, with no significant Pfizer–Moderna difference. Policy Implications: Controlling for healthy-vaccinee bias, Moderna was associated with lower COVID-19 mortality, relative to Pfizer. This difference was not known in real-time due to the failure of the available studies to address healthy-vaccinee bias. Controlling for such bias is vital when making vaccination policy.

  • The Effect of Medicaid Expansion on Diagnosis, Care, Treatment, and Health for Persons with Diabetes: Evidence from Medicaid Expansion in Wisconsin

    by Sadia Farzana, Andrew Owen, John Meurer, Ronald Ackermann, Bernard Black

    We report time-series evidence on the effects of expansion of Medicaid coverage under the 2014 Affordable Care Act (ACA), on healthcare and outcomes for persons with diabetes, including medium- and longer-term effects. We focus on diabetes, as a common, often-underdiagnosed condition for which effective treatments exist, and on childless adults, who were a particular target for ACA expansion. We use difference-in-differences (DiD) analysis applied to a longitudinal, uniquely detailed (relative to prior work) dataset that includes visit-level electronic healthcare records over 2011–2022 from the Medical College of Wisconsin (a major Milwaukee-centered health system) linked to Medicaid enrollment records. We compare a treatment group of 1,679 newly enrolled childless adults with diabetes to a propensity-score-balanced control group of 1,600 already-insured adults with diabetes (mostly parents with children at home). Gaining insurance leads to sharply higher utilization (outpatient, emergency department, and hospitalization); and large increases in new diagnoses of diabetes and other chronic conditions; related prescriptions; and testing rates. We find improvement in intermediate health outcomes (blood sugar, blood pressure, and cholesterol levels), and suggestive evidence of improved longer-term outcomes (macrovascular events; advanced kidney disease; amputations). We also compare persons with diabetes to other persons; DiD coefficients are generally larger for persons with diabetes, but percentage changes are often higher for other persons.

  • Open Access

    Online Appendix for: COVID-19 Vaccination for Young Adults: Estimated Lives Saved for Ages 15-24

    by Francesco Maria Rossi, Vladimir A. Atanasov, Sadia Farzana, Qian Luo, Katie Allen, Jeff Whittle, Kevin McGUrk, Ali Moghtaderi, Bernard S. Black

    This Appendix provides dataset and methods details, as well as additional results and robustness checks, for Rossi et. al, COVID-19 Vaccination for Young Adults: Estimated Lives Saved for ages 15-24 (working paper 2026). The paper can be downloaded from SSRN at: http://ssrn.com/abstract=6846360

  • Patient Race/Ethnicity, Socioeconomic Status, and Acute Myocardial Infarction Mortality

    by Sadia Farzana, Ali Moghtaderi, Andy Ye PhD Yuan, Qian (Eric) Luo, Vinay Md Kini, Bernard Jd Black

    Background: Prior work often finds racial/ethnic differences in mortality following acute myocardial infarction (AMI), but is subject to selection bias. Similar selection issues arise for differences associated with socioeconomic status (SES). Objectives: Re-examine the association between patient race/ethnicity, area-level socioeconomic status (area-SES), and post-AMI mortality using a data source that limits patient selection of hospital or cardiologist, or vice-versa, and which lets us distinguish between-hospital from within-hospital sources of differences. Research Design: We compare mortality disparities with no controls, with controls typical of the prior literature (“typical controls”), and more extensive controls, including hospital, cardiologist, and calendar quarter fixed effects (FEs). Subjects: The study uses a 100% sample of 681,000 Medicare Fee-for-Service patients aged 68+ hospitalized for incident (first) AMI over 2008–2019. Measures: Post-AMI mortality in-hospital, within 30 days after discharge and over periods up to 3 years post-discharge. We study patients with ST-segment elevation MI (STEMI) and non-ST-segment elevation MI (nSTEMI) separately. Results: With no or typical controls, Blacks, Hispanics, and Asians have higher in-hospital mortality than Whites; lower SES also predicts higher mortality. However, higher mortality is substantially explained by between-hospital differences in mortality rates. Longer-term mortality is higher for Black patients, consistent with the importance of post-discharge pathways. Conclusions: Post-AMI disparities in outcomes can be strongly affected by selection effects, especially the tendency for poor and minority persons to be treated at lower-quality hospitals. Disparities measurement and policy discussions should distinguish access-related hospital sorting from within-hospital processes and pay greater attention to post-discharge pathways.

  • Open Access

    Appendix for: COVID-19 Vaccination: Actual and Potential Lives and Life-Years Saved

    by Qian Luo, Francesco Maria Rossi, Ali Moghtaderi, Sadia Farzana, Jeff Whittle, Kevin McGUrk, Benjamin W. Weston, Vladimir A. Atanasov, Bernard S. Black

    This Appendix contains additional details on the data sources, data cleaning, simulation procedure and additional results for Luo et al. (2026), COVID-19 Vaccination: Actual and Potential Lives and Life-Years Saved, http://ssrn.com/abstract=6841624.

  • Open Access

    COVID-19 Vaccination for Young Adults: Estimated Lives Saved for Ages 15-24

    by Francesco Maria Rossi, Vladimir A. Atanasov, Sadia Farzana, Qian Luo, Katie Allen, Jeff Whittle, Kevin McGUrk, Ali Moghtaderi, Bernard S. Black

    COVID-19 mortality risk was small for young people and rose rapidly with age. Dr. Vinay Prasad, the Dirctor of the FDA's Center for Biologics Evaluation and Research (CBER) has questioned the net benefit of COVID-19 vaccination for children and young adults, citing evidence of 10 and perhaps more cases of probable young-adult deaths from vaccine-induced myocarditis. These deaths must be balanced against lives saved in this age group. We estimate how many lives of people aged 15-24 were saved by COVID-19 vaccination over July 2021-March 2022. We use a combination of US national data and linked mortality and vaccination records from Indiana, and several estimation strategies which account for healthy vaccinee bias. There were 1,387 COVID-19 deaths of people aged 15-24 during this period in the US. We estimate that COVID-19 vaccination prevented 1,086-1,532 additional deaths (mean=1,303) in this age group. Encouraging vaccination was a reasonable strategy at the time and was highly effective in hindsight. We caution that these results do not extend to children under 15, who had much lower COVID-19 mortality, or to booster doses (which had limited incremental value for this age group).

  • Open Access

    COVID-19 Vaccination: Actual and Potential Lives and Life-Years Saved

    by Qian Luo, Francesco Maria Rossi, Ali Moghtaderi, Sadia Farzana, Jeff Whittle, Kevin McGUrk, Benjamin W. Weston, Vladimir A. Atanasov, Bernard S. Black

    COVID-19 killed millions of people worldwide, but rapid vaccine development saved millions of lives. We model lives saved in the U.S., by age range, under three alternative scenarios: (i) no vaccination; (ii) full vaccination (95% vaccination of all persons); and (iii) a faster booster scenario, with 95% booster uptake among those with primary vaccination plus booster update speed matching the actual speed achieved for primary vaccination. There were 983,000 COVID-19 deaths in the U.S. over the pandemic period (March 2020-April 2023); this would have been 1,787,000 with no vaccination (804,000 lives saved). Full vaccination would have saved 62,000 additional lives and faster booster uptake would have saved another 47,000 lives. Mean life-years lost per decedent were 14.70; total life-years saved by vaccination were 14.5 million; full vaccination and faster booster uptake would have saved an additional 1.7 million and 649,000 life years, respectively. Most lives saved from vaccination, or deaths not saved due to lack of full vaccination, were for the middle-aged and elderly, but the percentage of preventable COVID-19 deaths due to partial vaccine uptake was largest for younger persons, due to lower vaccination rates. The effect of partial uptake on life-years lost was largest for the middle aged.

  • Open Access

    COVID-19 Vaccine Effectiveness Against Mortality in the Omicron Period: Evidence from Linked Mortality and Vaccination Records

    by Sadia Farzana, Francesco Maria Rossi, Katie S. Allen, Qian Luo, Jeff Whittle, Kevin McGurk, Benjamin W. Weston, Andy Ye Yuan, Ali Moghtaderi, Vladimir Atanasov, et al.

    Objective: This study aimed to assess COVID-19 vaccine effectiveness against death (VE), controlling for healthy vaccinee bias. Methods: We link all adult deaths through year-end 2022 in the State of Indiana, U.S.A., to vaccination records and identify which deceased received primary vaccination (measured as either one or two initial doses) and which received one or two booster doses. We measure COVID-19 mortality with the COVID Excess Mortality Percentage (CEMP). CEMP is calculated, for a group defined by various characteristics (age, sex, time period), as COVID-19 deaths divided by non-COVID natural deaths. The CEMP outcome measure accounts for healthy vaccinee bias by using non-COVID natural mortality to control for differences in population health. Results: We find a large healthy vaccinee bias. Controlling for this bias, we find substantial VE for primary vaccination and the first booster dose during the first five vaccine-available calendar quarters, from 1Q2021 through 1Q2022 (end of Omicron infection wave). However, over 2Q–4Q2022, we find no evidence for primary-vaccination VE, and find moderate but statistically insignificant booster VE, which largely wanes by 4Q2022. Conclusions: It is known that by 2Q2022, most people had natural immunity from prior COVID-19 infection. Thus, our results for 2Q–4Q2022 largely reflect comparing hybrid (infection plus vaccination) immunity to infection-only immunity. In this period, we find negligible mortality benefit from primary vaccination, and moderate but waning benefit from a booster dose. Policy Implications: Controlling for healthy vaccinee bias is crucial when estimating VE. We found limited VE against COVID-19 mortality over 2Q–4Q2022, but lacked data for more recent periods.


  • Open Access

    The Impact of Health Insurance on Healthcare Utilization and Health: Evidence from Medicaid Expansion in Wisconsin

    by Sadia Farzana, Andrew Owen, John Meurer, Ronald Ackermann, Bernard S. Black

    We extend the Oregon Health Insurance Experiment (OHIE) study of the effects of gaining health insurance by reporting time-series evidence, from difference-indifferences (DiD) analysis of visit-level healthcare records, before and after Medicaid expansion in Wisconsin in 2014 under the Affordable Care Act (ACA). Wisconsin expanded Medicaid to cover childless adults with incomes up to 100% of the federal poverty limit (FPL). We follow these persons over 2011-2022 using electronic health records (EHR) records for a major, Milwaukee-based health system. We compare healthcare use for newly-Medicaid-enrolled childless adults aged 36-64 (treatment group) with a propensity-score-balanced control group of previously enrolled persons. Gaining health insurance leads to a sharp, immediate, rise in health care utilization. Outpatient visit rates for new enrollees roughly double, to levels similar to the control group. ED visit and hospitalization rates jump from well below to well above the control group, but by 2016 fall back to control-group levels. ED visits rise for both visits leading to discharge and visits leading to hospital admission. Medicaid expansion predicts an increase in a range of new diagnoses (e.g. diabetes, microvascular disease, peripheral vascular disease), new medication prescriptions (e.g., diabetes, cholesterol, blood pressure), testing rates (e.g., glucose, HbA1c for diabetics, cholesterol), and lower levels for chronic disease markers (glucose, cholesterol, blood pressure, but no change in macrovascular event rates. Treatment effects are broadly similar across gender, race/ethnicity, and age.