Jesus N. Sarol Jr., PhD, recently joined IHSI as Senior Research Biostatistician, expanding the expertise of the Biostatistics Core to now also include Epidemiology and Research Design. Dr. Sarol is a biostatistician and epidemiologist with more than 25 years of experience in biostatistics, epidemiology, research methods and data management. After earning a bachelor’s in statistics from the University of the Philippines, he went on to earn his master’s and doctorate in epidemiology from UCLA and completed post-doctoral training in biostatistics at the Free University Berlin. His experience in public health and medical research came from involvement in various capacities as principal investigator, co-investigator, epidemiologist, biostatistician, data manager, computer programmer, and technical writer in several large-scale health surveys, epidemiologic and clinical studies, and program evaluations in the Philippines.
Tell us more about your background. How did you develop such broad research experience?
I am both a biostatistician and an epidemiologist with training obtained from universities in the Philippines, USA, and Germany. I taught graduate-level biostatistics and epidemiology courses and served as advisor and panel member of graduate students’ theses/dissertations at the University of the Philippines Manila (UPM) for more than 25 years. UPM was the only university with a school of public health in the Philippines, so we trained many of the individuals who became leaders in the Philippine Department of Health and public hospitals. UPM was also a regional training center for the Association of Southeast Asian Nations (ASEAN) for public health and occupational health, so we provided consultants to guide programs, policies, interventions, and studies for the Department of Health, health research institutions, and international organizations. Outside of academia, I have worked for international organizations such as the World Health Organization, International Organization for Migration, FHI 360 and German Technical Cooperation. This involvement provided experiences where I actually applied epidemiologic and biostatistical expertise in describing the magnitude of health problems that are global in scale, identifying factors associated with them, and evaluating interventions for their control and prevention. Since there are Illinois researchers who deal with similar problems, I believe I can contribute valuable inputs into the design of their studies and interventions.
How did you get involved with health research?
At first, my research involvement was as a computer programmer for faculty research studies at the University of the Philippines. A senior biostatistician would provide the list of required outputs, and I would use SAS programming to produce them. I was later tasked to perform statistical analyses, planning the required outputs based on study objectives. When I returned from my doctoral studies at UCLA, I was invited to become co-investigator in studies where I provided inputs in proposals, took charge of data management and analysis, and contributed to reports and papers for publication. In the last ten years, I have been commissioned by the Department of Health and international organizations for several health studies and evaluations. My research involvement covered wide variety of areas such as primary health care, tuberculosis, mental illness, blindness, social re-insurance, air pollution, oral health, urban health, nutrition, hospital performance, essential drug availability and migration health. My UCLA training and the mentorship I received under renowned biostatistician, Prof. Dankmar Böhning contributed significantly to my skills in biostatistics and epidemiology. I studied at a crucial time when prominent UCLA professors, Prof. Sander Greenland and Hal Morgenstern, were very actively involved in the clarification of many epidemiologic concepts such as measures of association, study designs, confounding, biases, and causal inference. I worked with Prof. Böhning’s team in three separate occasions studying novel statistical methodologies on heterogeneity in populations.
Can you clarify the difference between what a biostatistician versus an epidemiologist provides in a research study?
Biostatisticians perform many tasks in a research study. The common ones include preparing a sampling or randomization plan, computing sample size or power, preparing and executing the statistical analysis plan, and finding best ways to present results (tables or graphs). These are very valuable inputs from biostatisticians such that often, the expectation is that a biostatistician performed them in a study. As an epidemiologist, I like to infuse an epidemiologic perspective into studies of human populations. Epidemiologists often look for associations between people’s characteristics and health (e.g. diseases, disabilities, disease sequelae) and health-related outcomes (e.g. health services utilization, social determinants of health). In particular, they try to find something that would help in understanding causal mechanisms in the development of these outcomes in the research studies. Three associations are often studied in human populations: risk factor association, prognostic factor association, and markers for disease. Epidemiologists would like to see which of these questions is/are addressed to assess the study’s significance. Ideally, we’d like studies that lead to clearer answers to at least one of these associations, rather than studies that cannot distinguish between them in the results. Both groups of professionals ensure that the study objectives are going to be met. Biostatisticians will make data talk in presenting results and epidemiologists will ensure that these results make sense. Both fields of expertise look into the same features of a research study—design, data, and data analysis. Biostatisticians contribute to more efficient designs (e.g. control of variability), choice of appropriate distribution to represent data, and functional relationship of variables, epidemiologists would be concerned about possibility of selection biases in study design, validity of operational definitions of variables, and choice of meaningful statistical models.
Can you provide some specific examples of how these differences would play out in a research study?
I will provide two examples: 1. In studying exposure-disease associations, a biostatistician might decide to use the Poisson distribution for rate data and an exponential equation for the relationship of the exposure with disease because he sees that this combination fits the data. An epidemiologist, on the other hand, might look into minimizing selection and information biases, such as the use of new cases of disease for the dependent variable (rate) for a risk-factor hypothesis. The epidemiologist would also agree to the choice of the exponential relationship of the biostatistician because it leads to a suitable interpretation like the multiplicative effects of risks factors. 2. In national prevalence studies of mental health in a developing country, a biostatistician would be needed to come up with a cost-efficient sampling design, say a stratified two-stage cluster sampling design with probabilities proportionate to population size. An epidemiologist would be concerned about how to obtain valid and reliable data in feasible ways. For example, to determine the presence of mental illnesses, a large survey cannot employ trained psychiatrists because of the inadequacy of numbers of these professionals. So other individuals with health backgrounds, that are abundantly available, would have to be engaged as interviewers. Furthermore, an epidemiologist would also advise on the choice of a diagnostic tool such as the World Health Organization World Mental Health Composite International Diagnostic Interview (WHO WMH-CIDI), which he might have found to have good psychometric properties (e.g. sensitivity and specificity) when used by trained lay interviewers (non-psychiatrists). I should emphasize that these are not separate contributions from the two fields of expertise. Biostatisticians are usually trained in epidemiologic methods as an area of application, while epidemiologists rely on biostatistical methods for data analysis. Many schools of public health have these professionals in one department.