Stephen B. Soumerai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Stephen B. Soumerai has a total of 152 co-authors. Here are the top co-authors by number of shared publications:
Dennis Ross-Degnan
Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, USA
10 shared publications- Dennis Ross-Degnan
Department of Population Medicine, Harvard Medical School, Boston, MA
9 shared publications Fang Zhang
Department of Population Medicine Harvard Medical School and Harvard Pilgrim Health Care Institute Boston MA USA
8 shared publicationsJerry H. Gurwitz
Division of Geriatric Medicine University of Massachusetts Medical School Worcester MA USA
6 shared publicationsChristine Y. Lu
Department of Population Medicine Harvard Medical School and Harvard Pilgrim Health Care Institute Boston Massachusetts
4 shared publications- Helene Levens Lipton
Department of Clinical Pharmacy, School of Pharmacy, Institute for Health Policy Studies, School of Medicine, University of California at San Francisco, 3333 California Street Suite 265, St Laurel Heights, San Francisco, CA 94143-0936, USA
4 shared publications
Latest Publications
- Open Access
- 11 March 2021
- Published by American Psychiatric Association Publishing in Psychiatric Research and Clinical Practice
- Vol. 3 (2) , 98-101
Response to “Black Box Warning Did Not Cause Increased Suicides”
- 11 March 2021
- Published by American Psychiatric Association Publishing in Psychiatric Research and Clinical Practice
- Vol. 3 (2) , 98-101
Open AccessResponse to “Black Box Warning Did Not Cause Increased Suicides”
- 1 February 2021
- Published by Elsevier in Journal of Affective Disorders
- Vol. 281, 41-50
Experiences of health care costs among people with employer-sponsored insurance and bipolar disorder
- 1 February 2021
- Published by Elsevier in Journal of Affective Disorders
- Vol. 281, 41-50
Experiences of health care costs among people with employer-sponsored insurance and bipolar disorder
- Open Access
- 7 October 2020
- Published by American Psychiatric Association Publishing in Psychiatric Research and Clinical Practice
- Vol. 2 (2) , 43-52
Increases in Suicide Deaths Among Adolescents and Young Adults Following US Food and Drug Administration Antidepressant Boxed Warnings and Declines in Depression Care
ObjectiveStudies show decreased depression diagnosis, psychotherapy, and medications and increased suicide attempts following US Food and Drug Administration antidepressant warnings regarding suici...
- 7 October 2020
- Published by American Psychiatric Association Publishing in Psychiatric Research and Clinical Practice
- Vol. 2 (2) , 43-52
Open AccessIncreases in Suicide Deaths Among Adolescents and Young Adults Following US Food and Drug Administration Antidepressant Boxed Warnings and Declines in Depression Care
ObjectiveStudies show decreased depression diagnosis, psychotherapy, and medications and increased suicide attempts following US Food and Drug Administration antidepressant warnings regarding suici...
- Open Access
- 1 January 2020
- Published by American Medical Association (AMA) in JAMA Internal Medicine
- Vol. 180 (1) , 144-147
Risk Factors Associated With Food Insecurity in the Medicare Population
Food insecurity is defined as the inability to obtain adequate food for explicitly financial reasons. Associations between food insecurity and worse health outcomes have been attributed to insufficient nutritional intake, trade-offs between food and medical treatment, and functional impairments.1 Medicare enrollees, who are 65 years or older or who have long-term disabilities, are especially vulnerable to food insecurity. We present detailed national estimates of food insecurity prevalence within the Medicare population.
- 1 January 2020
- Published by American Medical Association (AMA) in JAMA Internal Medicine
- Vol. 180 (1) , 144-147
Open AccessRisk Factors Associated With Food Insecurity in the Medicare Population
Food insecurity is defined as the inability to obtain adequate food for explicitly financial reasons. Associations between food insecurity and worse health outcomes have been attributed to insufficient nutritional intake, trade-offs between food and medical treatment, and functional impairments.1 Medicare enrollees, who are 65 years or older or who have long-term disabilities, are especially vulnerable to food insecurity. We present detailed national estimates of food insecurity prevalence within the Medicare population.
- 12 March 2019
- Published by Wiley in Health Services Research
- Vol. 54 (3) , 537-542
Instrumental variables: The power of wishful thinking vs the confounded reality of comparative effectiveness research
- 12 March 2019
- Published by Wiley in Health Services Research
- Vol. 54 (3) , 537-542
Instrumental variables: The power of wishful thinking vs the confounded reality of comparative effectiveness research
- Open Access
- 25 November 2017
- Published by Springer Nature in Implementation Science
- Vol. 12 (1) , 1-8
The impact of a physician detailing and sampling program for generic atorvastatin: an interrupted time series analysis
In 2011, Manitoba implemented a province-wide program of physician detailing and free sampling for generic atorvastatin to increase use of this generic statin. We examined the impact of this unique combined program of detailing and sampling for generic atorvastatin on the use and cost of statin medicines, market share of generic atorvastatin, the choice of starting statin for new users, and switching from a branded statin to generic atorvastatin. We conducted a retrospective study of Manitoba insurance claims data for all continuously enrolled patients who filled one or more prescriptions for a statin between 2008 and 2013. Data were linked to physician-level data on the number of detailing visits and sample provision. We used interrupted time series analyses to assess policy-related changes in the use and cost of statin medicines, market share of generic atorvastatin, the choice of starting statin for new users, and switching from a branded statin to generic atorvastatin. The detailing program reached 31% (651/2103) of physicians who prescribed a statin during the study period. Collectively, these physicians prescribed 61% of statins dispensed in the province. Free sample cards were provided to 61% (394/651) of the detailed physicians. The program did not change the level or trend in the overall statin use rate and the total cost of statins or increase the number of patients switching from another branded statin to generic atorvastatin. We found the program had a small impact on atorvastatin’s market share of new prescriptions, with a level increase of 2.6%. Though physician detailers were skilled at targeting high-prescribing physicians, a combined program of detailing visits and sample provision for generic atorvastatin did not lower overall statin costs or lead to switching from branded statins to the generic. The preceding introduction of generic atorvastatin appeared sufficient to modify prescribing patterns and decrease costs.
- 25 November 2017
- Published by Springer Nature in Implementation Science
- Vol. 12 (1) , 1-8
Open AccessThe impact of a physician detailing and sampling program for generic atorvastatin: an interrupted time series analysis
In 2011, Manitoba implemented a province-wide program of physician detailing and free sampling for generic atorvastatin to increase use of this generic statin. We examined the impact of this unique combined program of detailing and sampling for generic atorvastatin on the use and cost of statin medicines, market share of generic atorvastatin, the choice of starting statin for new users, and switching from a branded statin to generic atorvastatin. We conducted a retrospective study of Manitoba insurance claims data for all continuously enrolled patients who filled one or more prescriptions for a statin between 2008 and 2013. Data were linked to physician-level data on the number of detailing visits and sample provision. We used interrupted time series analyses to assess policy-related changes in the use and cost of statin medicines, market share of generic atorvastatin, the choice of starting statin for new users, and switching from a branded statin to generic atorvastatin. The detailing program reached 31% (651/2103) of physicians who prescribed a statin during the study period. Collectively, these physicians prescribed 61% of statins dispensed in the province. Free sample cards were provided to 61% (394/651) of the detailed physicians. The program did not change the level or trend in the overall statin use rate and the total cost of statins or increase the number of patients switching from another branded statin to generic atorvastatin. We found the program had a small impact on atorvastatin’s market share of new prescriptions, with a level increase of 2.6%. Though physician detailers were skilled at targeting high-prescribing physicians, a combined program of detailing visits and sample provision for generic atorvastatin did not lower overall statin costs or lead to switching from branded statins to the generic. The preceding introduction of generic atorvastatin appeared sufficient to modify prescribing patterns and decrease costs.
- Open Access
- 18 October 2016
- Published by Springer Nature in Journal of General Internal Medicine
- Vol. 32 (2) , 204-209
False Dichotomies and Health Policy Research Designs: Randomized Trials Are Not Always the Answer
Some medical scientists argue that only data from randomized controlled trials (RCTs) are trustworthy. They claim data from natural experiments and administrative data sets are always spurious and cannot be used to evaluate health policies and other population-wide phenomena in the real world. While many acknowledge biases caused by poor study designs, in this article we argue that several valid designs using administrative data can produce strong findings, particularly the interrupted time series (ITS) design. Many policy studies neither permit nor require an RCT for cause-and-effect inference. Framing our arguments using Campbell and Stanley’s classic research design monograph, we show that several “quasi-experimental” designs, especially interrupted time series (ITS), can estimate valid effects (or non-effects) of health interventions and policies as diverse as public insurance coverage, speed limits, hospital safety programs, drug abuse regulation and withdrawal of drugs from the market. We further note the recent rapid uptake of ITS and argue for expanded training in quasi-experimental designs in medical and graduate schools and in post-doctoral curricula.
- 18 October 2016
- Published by Springer Nature in Journal of General Internal Medicine
- Vol. 32 (2) , 204-209
Open AccessFalse Dichotomies and Health Policy Research Designs: Randomized Trials Are Not Always the Answer
Some medical scientists argue that only data from randomized controlled trials (RCTs) are trustworthy. They claim data from natural experiments and administrative data sets are always spurious and cannot be used to evaluate health policies and other population-wide phenomena in the real world. While many acknowledge biases caused by poor study designs, in this article we argue that several valid designs using administrative data can produce strong findings, particularly the interrupted time series (ITS) design. Many policy studies neither permit nor require an RCT for cause-and-effect inference. Framing our arguments using Campbell and Stanley’s classic research design monograph, we show that several “quasi-experimental” designs, especially interrupted time series (ITS), can estimate valid effects (or non-effects) of health interventions and policies as diverse as public insurance coverage, speed limits, hospital safety programs, drug abuse regulation and withdrawal of drugs from the market. We further note the recent rapid uptake of ITS and argue for expanded training in quasi-experimental designs in medical and graduate schools and in post-doctoral curricula.
- 21 July 2016
- Published by Wiley in Health Services Research
- Vol. 52 (1) , 9-15
The Reliability of Instrumental Variables in Health Care Effectiveness Research: Less Is More
- 21 July 2016
- Published by Wiley in Health Services Research
- Vol. 52 (1) , 9-15
The Reliability of Instrumental Variables in Health Care Effectiveness Research: Less Is More
- Open Access
- 23 June 2016
- Published by Centers for Disease Control and Prevention (CDC) in Preventing Chronic Disease
- Vol. 13, E82
History Bias, Study Design, and the Unfulfilled Promise of Pay-for-Performance Policies in Health Care
Recently, PCD published a longer, related article, “How Do You Know Which Health Care Effectiveness Research You Can Trust? A Guide for the Perplexed,” that was translational in nature. The article used simple graphs and easy-to-understand text—in five case studies—to illustrate how powerful biases, combined with weak study designs that cannot control for them, can yield untrustworthy research on several widespread interventions: influenza vaccination policy, health information technology, drug safety, prevention of childhood obesity and hospital safety (“mortality reduction”)\ud programs. Since this was not a formal methods paper, the target audiences were policy makers, journalists, trainees, and the public. The primary goal was to understand how weak or strong study designs are likely to fail or succeed in controlling for these pervasive biases. At the start of that article, we promised to add to these case examples of common biases and research designs to show why “caution is needed in understanding and accepting the results of research that may have profound and long-lasting effects on health policy and clinical practice.” In this sixth case study, the authors revisit one of the most common and virulent biases, threats of\ud history. Studies can mislead policymakers and clinicians because they fail to control for history, which represents pre-existing or co-occurring changes in study outcomes that were happening with or without the intervention. The policy in this case, pay-for-performance (PfP, see below), is extremely sensitive to this powerful bias because medical practice is always changing as a result of factors unrelated to a\ud policy, such as widespread media or national guidelines supporting a life-saving treatment, e.g., beta blockers for acute MI (1). Without investigating and visualizing outcomes over time before and after a policy or intervention, it is likely that the investigators will attribute such ongoing changes to “effects” of the quality improvement policy, resulting in millions of dollars of waste implementing ineffective PfP policies worldwide
- 23 June 2016
- Published by Centers for Disease Control and Prevention (CDC) in Preventing Chronic Disease
- Vol. 13, E82
Open AccessHistory Bias, Study Design, and the Unfulfilled Promise of Pay-for-Performance Policies in Health Care
Recently, PCD published a longer, related article, “How Do You Know Which Health Care Effectiveness Research You Can Trust? A Guide for the Perplexed,” that was translational in nature. The article used simple graphs and easy-to-understand text—in five case studies—to illustrate how powerful biases, combined with weak study designs that cannot control for them, can yield untrustworthy research on several widespread interventions: influenza vaccination policy, health information technology, drug safety, prevention of childhood obesity and hospital safety (“mortality reduction”)\ud programs. Since this was not a formal methods paper, the target audiences were policy makers, journalists, trainees, and the public. The primary goal was to understand how weak or strong study designs are likely to fail or succeed in controlling for these pervasive biases. At the start of that article, we promised to add to these case examples of common biases and research designs to show why “caution is needed in understanding and accepting the results of research that may have profound and long-lasting effects on health policy and clinical practice.” In this sixth case study, the authors revisit one of the most common and virulent biases, threats of\ud history. Studies can mislead policymakers and clinicians because they fail to control for history, which represents pre-existing or co-occurring changes in study outcomes that were happening with or without the intervention. The policy in this case, pay-for-performance (PfP, see below), is extremely sensitive to this powerful bias because medical practice is always changing as a result of factors unrelated to a\ud policy, such as widespread media or national guidelines supporting a life-saving treatment, e.g., beta blockers for acute MI (1). Without investigating and visualizing outcomes over time before and after a policy or intervention, it is likely that the investigators will attribute such ongoing changes to “effects” of the quality improvement policy, resulting in millions of dollars of waste implementing ineffective PfP policies worldwide
- Open Access
- 14 April 2016
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
- Vol. 23 (6) , 1143-1149
Missing clinical and behavioral health data in a large electronic health record (EHR) system
Objective Recent massive investment in electronic health records (EHRs) was predicated on the assumption of improved patient safety, research capacity, and cost savings. However, most US health systems and health records are fragmented and do not share patient information. Our study compared information available in a typical EHR with more complete data from insurance claims, focusing on diagnoses, visits, and hospital care for depression and bipolar disorder.Methods We included insurance plan members aged 12 and over, assigned throughout 2009 to a large multispecialty medical practice in Massachusetts, with diagnoses of depression ( N = 5140) or bipolar disorder ( N = 462). We extracted insurance claims and EHR data from the primary care site and compared diagnoses of interest, outpatient visits, and acute hospital events (overall and behavioral) between the 2 sources.Results Patients with depression and bipolar disorder, respectively, averaged 8.4 and 14.0 days of outpatient behavioral care per year; 60% and 54% of these, respectively, were missing from the EHR because they occurred offsite. Total outpatient care days were 20.5 for those with depression and 25.0 for those with bipolar disorder, with 45% and 46% missing, respectively, from the EHR. The EHR missed 89% of acute psychiatric services. Study diagnoses were missing from the EHR’s structured event data for 27.3% and 27.7% of patients.Conclusion EHRs inadequately capture mental health diagnoses, visits, specialty care, hospitalizations, and medications. Missing clinical information raises concerns about medical errors and research integrity. Given the fragmentation of health care and poor EHR interoperability, information exchange, and usability, priorities for further investment in health IT will need thoughtful reconsideration.
- 14 April 2016
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
- Vol. 23 (6) , 1143-1149
Open AccessMissing clinical and behavioral health data in a large electronic health record (EHR) system
Objective Recent massive investment in electronic health records (EHRs) was predicated on the assumption of improved patient safety, research capacity, and cost savings. However, most US health systems and health records are fragmented and do not share patient information. Our study compared information available in a typical EHR with more complete data from insurance claims, focusing on diagnoses, visits, and hospital care for depression and bipolar disorder.Methods We included insurance plan members aged 12 and over, assigned throughout 2009 to a large multispecialty medical practice in Massachusetts, with diagnoses of depression ( N = 5140) or bipolar disorder ( N = 462). We extracted insurance claims and EHR data from the primary care site and compared diagnoses of interest, outpatient visits, and acute hospital events (overall and behavioral) between the 2 sources.Results Patients with depression and bipolar disorder, respectively, averaged 8.4 and 14.0 days of outpatient behavioral care per year; 60% and 54% of these, respectively, were missing from the EHR because they occurred offsite. Total outpatient care days were 20.5 for those with depression and 25.0 for those with bipolar disorder, with 45% and 46% missing, respectively, from the EHR. The EHR missed 89% of acute psychiatric services. Study diagnoses were missing from the EHR’s structured event data for 27.3% and 27.7% of patients.Conclusion EHRs inadequately capture mental health diagnoses, visits, specialty care, hospitalizations, and medications. Missing clinical information raises concerns about medical errors and research integrity. Given the fragmentation of health care and poor EHR interoperability, information exchange, and usability, priorities for further investment in health IT will need thoughtful reconsideration.