Developing a Research Question

• The research question is the uncertainty an investigator aims to resolve through a study.
• Clinical trials have shown that treatments blocking estradiol synthesis reduce breast cancer risk in women with early-stage cancer
• New questions have emerged, such as optimal treatment duration, prevention in BRCA mutation carriers, managing osteoporosis side effects, and the effectiveness/safety of these treatments for primary prevention in healthy women.
• Studies should start with a research question that addresses what the investigator wants to know.
• The goal is to develop an important research question into a feasible and valid study plan.
• Questions usually come from your "laboratory," but sometimes you are answering someone else's question.
• The study plan includes developing a research question, designing the study, implementing the study, analyzing results, and inferring conclusions
• A study regarding the efficacy of a 9-valent pneumococcal conjugate vaccine in Soweto, South Africa.
• In a population of South African infants, immunization with a 9-valent pneumococcal conjugate vaccine is associated with reduced incidence of invasive pneumococcal disease.

Origins of a Research Question

• Mastering the literature is key.
• Be alert to new ideas and techniques.
• Low bone density is a risk factor for fractures.
• Women with low bone density may experience cognitive decline because of low estrogen levels.
• Researchers should be imaginative by observation and teaching.
• Choosing a mentor is important.

Characteristics of the Research Question: FINE

• F: Feasible
• I: Important
• N: Novel
• E: Ethical

Problems and Solutions if the Question is not Feasible

• If too broad, narrow the question using fewer variables.
• If not enough subjects are available, expand inclusion criteria, modify exclusion criteria, add subject sources, or lengthen the study timeframe.
• If methods are beyond the investigator's skills, collaborate, consult experts, review literature, or learn the skills needed.
• If too expensive, consider a less costly study design, fewer subjects/measurements, less extensive measurement, or fewer follow-up visits.

Problems and Solutions - FINE

• If a research question is not important, consult with mentors, communicate with funders, or modify the question.
• If a research question is not novel, thoroughly review the literature and consult experts familiar with unpublished ongoing research.
• It can be worthwhile to replicate previous observations, apply findings to other populations, or use new measurement methods to clarify risk factors and diseases.
• A confirmatory study can be done if it avoids previous weaknesses of existing studies or if the result to be confirmed was unexpected
• To address uncertain ethical suitability, consult with the institutional review board or modify the research question.
• If a question is not relevant, NIH reviewers emphasize the significance of the study, its impact on scientific knowledge, and potential changes to concepts, methods, or clinical services.

Problems and Solutions

• To solve a vague research question, create a plan outline early on.
• In the study protocol, include how subjects will be sampled and how variables will be measured.

Approaches

• Getting advice from at least one senior scientist or specialists is important.
• The study plan should emerge from an iterative process, which involves incremental changes in design, sample size estimation, reviews with colleagues, pretesting key features, and revisions.

Primary vs. Secondary Questions

• A study to determine if reducing dietary fat reduces breast cancer risk can also determine the effect on coronary events.
• Efficiency is an advantage with several answers from a single study, disadvantages include increased complexity of designing, implementing, and drawing statistical inferences
• Focus the study and sample size on a single primary research question.
• Adding secondary research questions about other predictors or outcomes may produce valuable conclusions without affecting the primary question settings.