Clinical Epidemiology Yellow Book
1.2 Selecting a Suitable Thesis Topic
In the past, determining what constitutes a suitable MSc thesis topic has been contentious in the Clinical Epidemiology & Health Care Research Program (Clin Epi), with considerable variability found across the final thesis topics of MSc graduates, from a meta-analysis to a randomized controlled trial. It was felt that some of the Clin Epi students’ MSc theses were comparable to a PhD thesis in other academic units. Now that a PhD, as well as an MSc, is offered in Clin Epi, guidance is even more necessary to lead students towards a suitable type and level of topic for their thesis research. This document has been written to assist students and their supervisors in their selection of a thesis topic
Purpose of Master’s Thesis
The master’s thesis in Clin Epi is designed to provide the student with research training above that acquired in courses. Courses cannot teach the student the nuts and bolts of carrying out actual studies, all of which have particular quirks and barriers to successful completion.
All students should learn about study design, writing a proposal, literature review and critical appraisal, and statistical analysis. Students are strongly encouraged to choose a thesis topic that is original, and publishable. The study should involve methods that are appropriate for the type of study undertaken (epidemiologic, clinical trial, economic evaluation, etc). Ideally the student should gain experience in data collection and data analysis so that when they are independent researchers they will know the trials and pitfalls of undertaking research. The student should have the experience of designing and finishing a research project and in many cases write a publishable paper.
Some students and their supervisors have elected to carry out studies that do not involve primary data collection (e.g. secondary analyses of administrative data). This represents a trade-off, in that the student finishes in a timely fashion but does not learn directly about study subject recruitment or retention, or how to collect and manage data. This trade-off needs to be taken under consideration when choosing a topic.
Purpose of a PhD Thesis
Although the taught courses are an important part of the PhD programme, it is the research and thesis writing that provide the main opportunities for developing both a broad perspective and in-depth knowledge of a particular area of research, either clinical or methodological (or both).
The PhD research work must be of sufficient breadth and depth to permit a thorough examination of the research topic. In the past, the need for breadth and depth was embodied in the ‘three papers rule’: that is, that a PhD should consist of sufficient work for publication of three peer reviewed publications. This guidance was not intended to suggest that students should undertake work on three unrelated topics, nor were the three publications intended to substitute for a properly written thesis. Rather, the rule was intended to give a rough measure of the quantity and quality of work needed for a PhD.
Earning a PhD degree is more than simply completing a few courses, carrying out a piece of research and writing a thesis; it is the process of becoming a professional clinical researcher in a given area. The thesis demonstrates a wide, critical knowledge of the field; an ability to ask appropriate questions and set the research in its proper context; that the appropriate research techniques have been mastered; and that the ideas and results have been communicated effectively to a group of peers. It is about making an original contribution to a particular research field within a framework of research training. ‘Original contribution’ means finding a gap in knowledge and answering a question posed by that gap.
Original contributions in the area of clinical epidemiology may include:
- Scholarly enquiry into the theoretical foundations of clinical epidemiology, in areas such as study design, measurement of health or health care, and approaches to analysis and interpretation of the data derived from such studies.
- Development of new methods or tools pertinent to study design, measurement of health or health care.
- Significant advancement of existing methods or tools in study design, measurement or analysis.
- Application of existing methods or tools in such a way as to advance knowledge of health or health care.
Guidelines for Choice of Thesis Topic
The following are proposed guidelines for choice of thesis topic in the Clin Epi program based on methodological approaches to data collection, not necessarily research design. Students who wish to pursue other topics should consult with their supervisor and the Program Director. As a rough guide, students should expect to spend 5 months FTE undertaking and writing their MSc thesis, while PhD students should expect to spend 1.5 years FTE.
NOTE: Regardless of the thesis topic or design, ethics approval must be obtained for all thesis work prior to undertaking the research.
A. Primary Data Collection
- Randomized Controlled Trials
Generally, it is difficult for a student to complete a RCT in a suitable time frame, even as a PhD. As a guideline, for an MSc thesis, it may be possible for a student to undertake the design of a RCT and complete a pilot study. Alternatively the student could start the RCT and provide an ‘interim’ analysis of the trial for the thesis.There are some circumstances where a RCT could be completed, and these may provide a suitable area for a PhD. These include questions where there are very large numbers of potential patients and where the required sample size is relatively small, around 50-60 patients (eg new epidural drug for obstetrical patients; paper & pencil versus videotape for patient education). Students should bear in mind that a very careful examination of feasibility should be undertaken, because suitable patients for a trial somehow disappear when trial recruitment begins.
- Observational Studies
Surveys (particularly mailed or self-administered in clinic or office setting) are usually appropriate for a master’s thesis. The circumstances should be that the sampling frame is large (eg physicians in Ontario), easily obtainable (sampling frame is available and accessible to students) and the sample size is manageable (eg one-time survey rather than longitudinal). Some examples include a survey of physicians’ attitudes towards prescribing a new drug; survey of patients’ satisfaction with a new intervention, etc. A more complex survey may be suitable for a PhD.
- Assessing patients’ preferences
Interviewing patients or using a self-administered questionnaire about their preferences for treatment choices, waiting times, etc, can usually be achieved with a manageable sample size. Again the issue should be one where access to patients will be easily obtained and large numbers of participants can be anticipated (eg breast cancer, heart disease). A patient preference study may be included as part of a PhD thesis, or a methodological development to do with measuring patient preferences may be suitable.
- Other types of observational studies
Generally a case-control study would be more suitable for a master’s thesis than a cohort study, as a case-control study can usually be performed in a more timely fashion. Methods include mailed questionnaires to cases/controls and/or chart reviews or interviews with patients. The question should be one in which there are large numbers of cases and controls (eg cancer, heart disease, diabetes) who can be easily accessed (eg cases/controls identified by patient registry, hospital patients, etc). A more complex and involved observational study may be suitable for a PhD.
- Instrument development and testing
The development of a new instrument from scratch would generally be too complex and take too long for a timely completion of a master’s thesis. However, portions of instrument development would be suitable, eg item generation, or reliability or validity studies. A PhD thesis might include the complete development of an instrument, along with validation work.
- Reliability/validity study of existing database
An MSc student project may involve the ascertainment of the reliability/validity of an existing administrative or clinical database. This might involve primary data collection such as a comparison of computerized records and hospital or physician hospital charts. Such a study might form a component of a PhD, but would be unlikely to provide a complete thesis topic.
- Qualitative study
This is an area of increasing interest among Clin Epi students and faculty; the Clin Epi Program has growing experience with qualitative research. Students should take the appropriate courses prior to undertaking this type of study and must have an experienced qualitative researcher as a member of his/her thesis committee. Qualitative studies can range from focus groups, key informant surveys or in-depth interviews with content analysis of interviews of patients or other groups. Qualitative research should not be undertaken lightly – it requires considerable expertise. Qualitative research can form the basis of either master’s or doctoral research.
B. No Primary Data Collection
- Administrative data or clinical databases
These have grown in popularity in recent years and are generally suitable for use for a Master’s thesis project. Several research questions can be addressed and the student does not have to spend the time gathering his/her own data. This is a trade-off since the student would not have the experience of his/her own data collection.If administrative data or clinical databases are to be used, then all the data analyses should be performed by the student (help by a programmer can be given, but in principle, all the computer analyses should be carried out by the student). Ideally database linkages (more than one database) or longitudinal data analysis (more than one year) should form part of the analysis. The scope and depth of the thesis topic will determine if it is suitable for an MSc or PhD.
- Decision analysis
Several thesis projects in recent years have involved decision analysis. Ideally at least a portion of the project should involve some primary data collection (eg obtaining utilities from providers or patients). The scope and depth of the thesis topic will determine if it is suitable for an MSc or PhD.
Meta analysis may be suitable for a master’s thesis in some circumstances. There needs to be a very large number of trials (approximately 100 studies) available to be considered in the project and the final number of quality studies should be more than 10. The quality of the studies should be assessed by at least 2 reviewers. Students should justify that the meta analysis topic is of sufficient scope to warrant a master’s thesis.
- Statistical models
Some students have requested to undertake their thesis research in the area of statistical modelling. This is usually suitable in conjunction with administrative databases (see above), eg in developing a clinical decision rule, or comparing risk-adjusted models. As a rule, the use of statistical models as a thesis project should have a direct clinical or health policy application and not just be a statistical exercise. (Students very interested in statistical theory could consider transfer to the MSc in biostatistics). If the topic is sufficiently complex, it may be suitable for PhD research.
1. Monson R R. Occupational epidemiology. Second edition. Boca Raton, FL: CRC Press, 1990
2. Hernberg S. Introduction to occupational epidemiology. Chelsea, MI: Lewis Publishers, 1992
3. Checkoway H, Pearce N, Kriebel D. Research methods in occupational epidemiology. Second edition. New York: Oxford University Press, 2004
4. Rothman K J, Greenland S. Modern epidemiology. Second edition. Philadelphia: Lippincott‐Raven Press, 1998
5. Koepsell T D, Weiss N S. Epidemiologic methods: studying the occurrence of illness. New York: Oxford University Press, 2003
6. Miettinen O S. Theoretical epidemiology. NY: Wiley, 1985
7. Eisen E A, Holcroft C R, Greaves I A. et al A strategy to reduce the healthy worker effect in a cross‐sectional study of asthma and metalworking fluids. Am J Ind Med 199731671–677. [PubMed]
8. Eisen E A. The healthy worker effect in morbidity studies. Med Lav 199586125–138. [PubMed]
9. Christiani D C, Ye T ‐ T, Wegman D H. et al Cotton dust exposure, across‐shift drop in FEV1, and five‐year change in lung function. Am J Respir Crit Care Med 19941501250–1255. [PubMed]
10. Christiani D C, Wang X ‐ R, Pan L ‐ D. et al Longitudinal changes in pulmonary function and respiratory symptoms in cotton textile workers. Am J Respir Crit Care Med 2001163847–853. [PubMed]
11. Wang X R, Zhang H X, Sun B X. et al A 20‐year follow‐up study of chronic respiratory effects of exposure to cotton dust. Eur Respir J 200526881–886. [PubMed]
12. Newman S C. Commonalities in the classical, collapsibility, and counterfactual concepts of confounding. J Clin Epidemiol 200457325–329. [PubMed]
13. Lubin J H, Gail M H. Biased selection of controls for case‐control analyses of cohort studies. Biometrics 19844063–75. [PubMed]
14. Prentice R L. A case‐cohort design for epidemiologic cohort and disease prevention trials. Biometrika 1986731–11.
15. Siemiatycki J. Risk factors for cancer in the workplace. Boca Raton, FL: CRC Press, 1991
16. Thomas D B, Gao D L, Self S G. et al Randomized trial for breast self‐exam in Shanghai: methodology and preliminary results. J Natl Cancer Inst 199789355–365. [PubMed]
17. Wernli K J, Astrakianakis G, Camp J. et al Development of a job exposure matrix (JEM) for the textile industry in Shanghai, China. J Occup Environ Hyg 20063521–529. [PubMed]
18. Astrakianakis G, Seixas N S, Camp J E. et al Modeling, estimation and validation of cotton dust and endotoxin exposures in Chinese textile operations. Ann Occup Hyg 200650573–582. [PubMed]
19. Wernli K J, Fitzgibbons E D, Ray R M. et al Occupational risk factors for esophageal and stomach cancers among women textile workers in Shanghai, China. Am J Epidemiol 2006163717–725. [PubMed]
20. De Roos A J, Ray R, Gao D L. et al Colorectal cancer incidence among female textile workers in Shanghai, China: a case‐cohort analysis of occupational exposures. Cancer Causes Control 2005161177–1188. [PubMed]
21. Li W, Ray R, Gao D L. et al Occupational risk factors for pancreatic cancer among female textile workers in Shanghai, China. Occup Environ Med 200663788–793. [PMC free article][PubMed]
22. Chang C ‐ K, Astrakianakis G, Thomas D B. et al Occupational exposures and risks of liver cancer among Shanghai female textile workers: a case‐cohort study. Int J Epidemiol 200635361–369. [PubMed]
23. Maclure M. The case‐crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 1991133144–153. [PubMed]
24. Sorock G S, Lombardi D A, Hauser R. et al A case‐crossover study of transient risk factors for occupational hand injury. Occup Environ Med 200461305–311. [PMC free article][PubMed]
25. Janes H, Sheppard L, Lumley T. Case‐crossover analyses of air pollution exposure data: referent selection strategies and their implications for bias. Epidemiology 200516717–726. [PubMed]
26. Navidi W. Bidirectional case‐crossover designs for exposures with time trends. Biometrics 199854596–605. [PubMed]