The Epi Department hosts a weekly seminar series, occuring every Wednesday from 12:30pm to 1:30pm in Kresge 502. Below, please read about some of the latest seminars, as well as those still to come. Also, read student's summaries of the monthly special seminars.
Upcoming Seminars
November 11
No seminar (Veterans Day)
November 18
Immaculata DeVivo, PhD, MPH
Associate Professor, Department of Medicine
Harvard Medical School
"Telomeres and Chronic Disease"
November 25
No seminar (Thanksgiving)
December 2 (Featured Seminar)
Alfredo Morabia, MD, PhD
Professor of Epidemiology
City University of New York and Mailman School of Public Health
"When triumphant epidemics finally met epidemiology"
December 9
JoAnn Manson, MD, DrPH
Chief, Division of Preventive Medicine, Brigham and Women's Hospital
Professor of Medicine and the Elizabeth F. Brigham Professor of Women's Health, Harvard Medical School
"Vitamin D: Is it as good as it seems?"
December 16
James Meigs, MD, MPH
Assistant Professor of Medicine, Harvard Medical School
Director, MGH Clinical Research Program Disease Management Research Unit
Topic: TBA
Seminar (Fall 2009)
September 9
Marc Lipsitch, D.Phil.
Professor, Epidemiology Department, HSPH
"2009 H1N1 Influenza"
September 23
David Christiani, MD, MPH, MS
Professor, HSPH. Professor, HMS. Physician, Massachusetts General Hospital
"Genetic Predictors of Survival in Lung Cancer"
September 30
John Seeger
Topic: TBA
October 7
Kari Stefansson, MD, PhD (deCODE Genetics)
"Genetics of Common/Complex Traits"
October 14
Francine Laden, ScD
Mark & Catherine Winkler Associate Professor of Environmental Epidemiology, HSPH
Assistant Professor of Medicine, Channing Laboratory, Brigham & Women's Hospital
"Health Effects of Air Pollution in the Nurses' Health Study"
November 4 (Featured Seminar)
Henrik Grönberg, M.D., Ph.D.
Professor of Cancer Epidemiology
Chairman, Department of Medical Epidemiology and Biostatistics
Karolinska Institutet
"The clinical utility of genetic markers in prostate cancer today and tomorrow"
Click above for VIDEO of lecture.
Summaries from 2009/2010
September featured seminar:2009 H1N1 Influenza
Marc Lipsitch, D.Phil. Professor, Department of Epidemiology
With the return of the academic year comes renewed concern about the spread of infectious diseases as September drags with it long hours of groups huddled in classrooms, offices, and laboratories. On the bright side, September also heralds the return of the Department of Epidemiology’s weekly seminar series!
This year, the H1N1 influenza is at the forefront of infectious disease concerns among many of us at the Harvard School of Public Health (HSPH). As epidemiology’s origins are rooted strongly in the mitigation of infectious disease outbreaks, what better way to introduce the seminar series than by learning about how epidemiology is applied to inform today’s crucial public health decisions? This year, HSPH’s own Dr. Marc Lipsitch, Professor of Epidemiology, kicked off the series with his discussion of recent developments in “the swine flu affair.”
Dr. Lipsitch’s talk, informally titled the “Missing Denominator,” emphasized that most of the numbers reported by the media and other outlets are wrong, and public health officials know they are wrong. Nevertheless, critical and time-sensitive public health decisions need to be made, and so the rapid data collection and efforts to correct the data for bias are of utmost importance [1,2]. Dr. Lipsitch began by discussing the anticipated severity of the pandemic, which can be categorized using the Pandemic Severity Index – a system that considers the case fatality ratio and the projected number of deaths to assign the pandemic to one of five discrete categories with increasing severity (Category 1 to Category 5). The best data available at the time of the seminar suggested that the current pandemic would be classified as a category 1, the lowest category. Recommendations regarding interventions then depend on the severity category and can be found at www.flu.gov.
Dr. Lipsitch went on to describe several recently completed and ongoing studies conducted to understand the dynamics of H1N1 and to estimate the potential public health impact of the infection. The examples represented creative solutions to what may be best described as on-the-ground, real-time information generation. In one example, Dr. Lipsitch described a study in which he and his co-investigators estimated the number of cases of H1N1 in Mexican residents by leveraging measures of incident influenza A/H1N1 among a relatively small and well observed group of US, UK, Spanish, and Canadian travelers who had visited Mexico [3]. Dr. Lipsitch and colleagues estimated that a minimum of 113,000 to 375,000 cases of novel influenza A/H1N1 occurred in Mexicans during the month of April 2009; a figure that is about 100 times more than had been virologically confirmed. Additional discussion focused on efforts to estimate the basic reproductive number [4] and case-fatality ratio [5] for the infection in the United States.
A major theme of the discussion was that the data and information environment surrounding the pandemic is continually updated and that the emerging picture is constantly evolving. Nevertheless, public health decision makers must use the best evidence available at the time of each decision, which they rely on epidemiology to generate.
By Joshua J. Gagne
- Lipsitch M, Hayden FG, Cowling BJ, Leung GM. How to maintain surveillance for novel influenza A H1N1 when there are too many cases to count. Lancet. 2009 Oct 3;374(9696):1209-11. Epub 2009 Aug 11. Review. PubMed PMID: 19679345. http://www.sciencedirect.com.ezp-prod1.hul.harvard.edu/science?_ob=ArticleURL&_udi=B6T1B-4X03JF9-2&_user=209690&_rdoc=1&_fmt=&_orig=search&_sort=d&_docanchor=&view=c&_acct=C000014438&_version=1&_urlVersion=0&_userid=209690&md5=fa1182557ffa2bbfbf13d7feb2460c29
- Lipsitch M, Riley S, Cauchemez S, Ghani AC, Ferguson NM. Managing and reducing uncertainty in an emerging influenza pandemic. N Engl J Med. 2009 Jul 9;361(2):112-5. Epub 2009 May 27. PubMed PMID: 19474417. http://content.nejm.org/cgi/content/full/361/2/112
- Lipsitch M, Lajous M, O’Hagan JJ, et al. Use of cumulative incidence of novel influenza A/H1N1 in foreign travelers to estimate lower bounds on cumulative incidence in Mexico. PLoS ONE 2009;4(9):e6895. http://www.plosone.org/article/info:doi%2F10.1371%2Fjournal.pone.0006895
- White L, Wallinga J, Finelli L, Reed C, Riley S, Lipsitch M, Pagano M. Estimation of the Reproductive Number and the Serial Interval in Early Phase of the 2009 Influenza the Current Influenza A/H1N1 Pandemic in the USA. Influenza and Other Respiratory Viruses, 22 September 2009 DOI:10.1111/j.1750 2659.2009.00106.x http://www3.interscience.wiley.com/cgi-bin/fulltext/122610761/HTMLSTART
- Presanis A, Lipsitch M De Angelis D, New York City Department of Health and Mental Hygiene Swine Flu Investigation Team, Hagy A, Reed C, Riley S, Cooper B, Piedrzycki P, Finelli L. The severity of pandemic H1N1 influenza in the United States, April – July 2009. PLoS Currents Influenza (non peer-reviewed moderated discussion board – currently under peer review) http://knol.google.com/k/anne-m-presanis/the-severity-of-pandemic-h1n1-influenza/agr0htar1u6r/16#
October featured seminar: Genetics of Common/Complex Traits
Kari Stefansson, MD, PhD (deCODE Genetics)
“Genetics is just a study of information. For reasons I don’t understand, genetics strikes more fear than any other area of science. This comes from a culture that sells cigarettes, alcohol, and fast-moving cars… telling us it’s bad to learn more about ourselves. We should be encouraged to learn as much about ourselves as we can.” -- Kari Stefansson
The role of genetics in disease etiology is far more complex than previously thought, involving the interplay between common variants, rare variants, and the built environment. Current epidemiological evidence supports the association between common variants and disease initiation. Although an individual polymorphism may confer minimal disease risk, they can exert substantial effects when combined with others. Rare variants may have larger effects in individuals, but are expected to account for low population attributable risk due to their low frequency. As a consequence of current limitations in technology and sample size, population-based research has focused on identifying common variants that influence risk of disease.
The detection of risk variants has advanced our understanding of the biology and pathophysiology of many diseases, including kidney stones, thyroid cancer, atrial fibrillation, diabetes, and prostate cancer. For example, variants in TCF2 confer prostate cancer risk, but are also protective against type 2 diabetes (T2D). Additionally, the frequency and effects of variants have been shown to differ by population and geography. These results are promising and could potentially aid in clinical decision-making and the identification of high-risk individuals. However, the utility of investigating genetic susceptibilities is limited when attempting to elucidate complex disease etiology that involves both genetic and environmental components. For example, a recent genome-wide association study identified a common variant in the nicotinic acetylcholine receptor that may confer risk of lung and cardiovascular disease through an effect on behavior (nicotine dependence), highlighting the importance of gene-environment interactions in disease pathogenesis. Thus, the interactions between multiple factors may play an important role in the etiology of certain diseases, but its very presence complicates their study.
“The most important choice of our lives is the choice of our parents.” Parent-of-origin effects are currently being investigated by utilizing a novel approach based on the concept of surrogate parenthood. Recent work demonstrates that sequence variants confer different risk depending on parental origin. In fact, the transmission of paternal and maternal alleles may not only differ in frequency, but also confer risk when transmitted from one parent and protection when transmitted from the other. For example, the single nucleotide polymorphism (SNP) rs2334499 is an important genetic variant that confers risk of type 2 diabetes. When not considering parent-of-origin, this SNP was not associated with disease risk. However, when evaluated by parent-of-origin, the variant was associated with an increased risk of T2D (OR=1.3) if maternally transmitted, but exhibited a protective effect (OR=0.87) if paternally transmitted. Thus, by conducting conventional case-control studies without considering parent-of-origin, investigators may underestimate or miss an important effect.
In sum, the parent-of-origin effect is likely important and may explain some, but not a large part, of the “dark matter”. There remains much to discover in the genetic determinants of disease. Future studies need to investigate the influence of parent-specific variants and gene-environment interactions, while keeping the characteristics of different populations in mind. Not only will this potentially have clinically meaningful implications for disease risk prediction – but more importantly – it will enable us to learn more about ourselves.
by Shanshan Li, Mengmeng Du, Jennifer Nguyen
Summaries from 2008/2009
April Special Seminar: "Strategically Significant Findings: The False, the Inflated, and the Useless"
Almost all published studies are statistically significant. Even the few published studies lacking statistical significance claim some sort of non-significant trend or offer an excuse why the findings fail to be statistically significant. Statistical significance alone, however, should not be the primary metric used in evaluating a study. The true value of the study is its credibility.
Credibility is how likely a research finding is to be true. Credibility has little to do with statistical significance, but rather is related to the amount of bias and random error present. Issues affecting credibility include sample size, effect size, the number of biological factors involved, and replication. It is further complicated by multiple comparisons.
While the field of Epidemiology has come a long way with the weight it places on statistical significance, there remains much room for improvement. Examples of non-replicated diminishing effects abound in the literature. In fact, a study by Dr. Ioannidis found 49 highly cited medical research studies to be contradicted over time or to have exaggerated effects. To make matters worse, resistance to refutation of original findings and failure to acknowledge unsuccessful replication are plentiful.
Nevertheless, there is an evolving emphasis on credibility. Fields such as genetic epidemiology, where replication was previously almost unheard of, now stresses the importance of timely replication. Replication alone is not enough. Efforts to grade the credibility of published findings by assessing factors including pre-evidence odds, data at hand, consistency, bias, and the field of study are needed. A push toward the use of positive predictive value rather than just statistical significance is also necessary. Failing to acknowledge limitations in studies with significant results remains a major problem to address.
Biomedical journals have tremendous potential to increase credibility. Their continued effort to avoid publication bias, selective reporting, and fabrication of results are essential. Journals must also expand their focus beyond novelty, importance, and significance, to require explanations of limitations, uncertainty, and replication. Potential remedies for focusing on credibility rather than statistical significance abound. By promoting multidisciplinary communications, fostering systematic approaches to research, and encouraging replication over discovery, a more credible system of research reporting that rewards creative thinking and robust methodology can be created.
by Mitch Machiela
November Special Seminar: "Unplanned Triumphs and Elusive Enigmas: On the Remarkable Epidemiology of Gastric Esophageal Cancer"
The rate of stomach cancer has steadily declined in both the United States and Sweden over the course of the late 19th and 20th centuries. Changes in smoking behavior and diet were initially proposed as possible explanations, but were later ruled out when studies revealed a weak association between diet and stomach cancer, and an attributable risk for smoking that was far less than the observed decrease in risk. Over time, evidence began to support the hypothesis that Helicobacter Pylori (H. Pylori) is an important risk factor and may even be a necessary cause. H. Pylori is a class I carcinogen and studies pooled in six different meta-analyses showed an odds ratio between 1.92 and 2.56. Furthermore, a Swedish population based case-control study that addressed misclassification due to false negative seroprevalence found much higher seroprevalence in cases compared to controls (odds ratio = 21, etiologic fraction=71%). In a prospective study in Japan, participants testing negative for H. Pylori did not develop stomach cancer, while 70% of participants in H. Pylori positive strata developed stomach cancer. Further support for this hypothesis is a marked decrease in seroprevalence by birth cohort over 1920 through 1980, possibly explaining some of the observed decrease in stomach cancer prevalence during this time period.
Esophageal cancer was rare until the 1970s, when the incidence rate increased nearly six-fold over the following 20 years. This was remarkably the most rapid increase of all cancers during this time period and the scientific community agreed that it could not be attributed to misclassification. Protective factors considered were antioxidants and H. Pylori seropositivity. Meta-analysis yielded an odds ratio of 0.53 for antioxidants, and an odds ratio .51 for CagA seropositivity (a protein marker for particularly virulent H. Pylori strains). In the Swedish Cancer Registry, the sharpest increase was observed around 1990, and it is conjectured that this increase may be due to decreasing seroprevalence of H. Pylori in the overall population.
While H. Pylori's relationship with esophageal cancer is not as well characterized as its relationship with stomach cancer, the findings provide the opportunity to consider how to balance eliminating a risk factor for one disease that may be protective for another.
By John Jackson
October Special Seminar: "Harvard Catalyst: How can we help improve your research life?"
On October 1, Steven Freedman, MD, PhD, Associate Professor of Medicine at Harvard Medical School, addressed students and faculty as the first monthly "featured speaker" in the weekly Department of Epidemiology Seminar Series. Dr. Freedman presented on the Harvard Catalyst, a resource created last May to unite basic scientists, clinicians, and public health investigators from all Harvard schools and affiliated healthcare centers for the common goal of promoting human health. As suggested by its name, the Catalyst aims to speed up efforts to understand, treat, and prevent disease through eliminating barriers to interdisciplinary collaboration.
Dr. Freedman gave an overview of the recently launched Harvard Catalyst website, http://catalyst.harvard.edu, which facilitates such collaboration through its comprehensive investigator and core facilities databases. The investigator database allows one to easily search for potential research partners by name, institution, and keyword. Each investigator profile contains a full publication list with direct links to PubMed and displays links to researchers with similar interests. The core facilities database provides information to investigators on the special services available to them in research and medical facilities throughout the university. Of particular benefit to those in the Department of Epidemiology is the Biostatistics Consultation Program, which provides free assistance with grant and IRB submissions, study design, and data analysis. Some of those in attendance at the seminar expressed concern that there is currently no epidemiology core that can similarly offer consultations, but Dr. Freedman appeared receptive to the suggestion that the Department of Epidemiology play a larger role in the Catalyst.
Funding for the Harvard Catalyst comes from a $117.5 million Clinical and Translational Science Award from the National Institutes of Health and a $75 million grant from the University and Academic Health Centers. The Catalyst will draw on these funds to provide $2.5 million in pilot grants this year, with financial support expected to be expanded in subsequent years. Dr. Freedman stressed that these pilot grants, which are available to all researchers regardless of degree status, are designed to promote high-impact, "out-of-the-box" research. By involving researchers at all levels and encouraging innovative, broad-based research, the Harvard Catalyst promises to be an invaluable tool for collectively finding solutions to current and future health challenges.
By Sarah Aroner and Leena Merdad