Clinical research is conducted to generate unbiased data, to improve existing knowledge, and to generate results valid for the majority of patients. To achieve these goals, clinical researchers must adhere to rigorous scientific standards.
At the start of clinical research, a detailed plan, or study protocol, should be written. This step is essential because it helps the researcher design the study. The purpose of the study protocol is to define the research question, to look for the significance of the study, and to search for the best design. These three steps are explained in this tutorial.
Answering the research question is the primary goal of the study. The question is often raised by the problems a surgeon has encountered and for which a solution is needed. However, before defining the exact question, one should first conduct an extended literature search to be aware of all the knowledge available in this domain.1
A good research question is one that is Feasible, Interesting, Novel, Ethical, and Relevant, which forms the mnemonic FINER.2 Feasible means that one should know the effort needed in terms of time and money, patients, technical expertise, and the amount of data required. Interesting means that the investigators will have incentives (e.g., scientific, career, funding) to carry out the study. Novel means that the study will provide new information to the scientific community. The literature search is mandatory to appreciate this point. Ethical means that the study is acceptable, physically and psychologically, for all patients who are included. In almost every country, the research question should be submitted to an institutional review board for an official ethical approval. Relevant means that the results are likely to be found. The study of these results will improve the knowledge in the field, thereby increasing the quality of care for patients or generating new hypotheses for further research.
Example of a research question:
What are the effects of treatment with nonsteroidal anti-inflammatory drugs (inhibitor of cyclooxygenase-2) compared with a placebo on the radiographic development of heterotopic ossification after total hip arthroplasty for patients with primary hip osteoarthritis?
When designing clinical research, the first task is to define if one would like to describe the events, observe the events, or introduce an intervention and examine the events afterward (Slide 1).
Descriptive studies include cross-sectional studies, correlational studies, case reports, and case series.
- Cross-sectional studies show the prevalence of an event in a specified population. An example is the study of the prevalence of residual pain after shoulder arthroplasty.
- Correlational studies examine the possible link between the prevalence of a risk factor and a possible consequence of the risk factor in a specific population. An example is the study of osteoporosis prevalence and femoral neck fracture in men older than 70 years of age.
- Case reports (a detailed presentation, <10 cases) or case series (a detailed presentation, >10 cases) are a good way to present rare diseases, secondary effects of treatments, or associations of diseases to generate new scientific hypotheses in terms of diagnostics, potential risks, or treatments. However, there is no way to estimate if the cases are rare or frequent. The group of cases is often highly selected and subject to bias (even rare events may appear simultaneously without being necessarily connected). To remediate these limitations, analytic studies have been designed.
Analytic studies include case-control studies, cohort studies, and randomized clinical trials. Case-control studies and cohort studies are observational studies.
- Case-control studies are case series to which a control group is added. The group of patients is increased, which diminishes the chance of observing random results. For example, a group of patients who had one episode of dislocation of their total hip arthroplasty can be compared to a group of patients who did not dislocate their hip prothesis and were operated on during the same period. To improve this design, one can match the patient with a control patient according to supposed confounding factors, and the study becomes a matched-paired case-control study. Confounding factors have the potential of influencing the outcome if they do not occur in the same proportions in both groups.3 For example, these factors may be age, sex, comorbidities, etc. The best control for the patient is himself or herself. When the patient can be studied in two different situations, the study is then called a crossover case-control study. This design is only possible if the baseline status of the patient can be determined before observing him/her in a second situation. However, even if some bias can be avoided in this design, it remains a retrospective design and, as such, may give rise to imperfect memory bias and interpretation bias.
- Cohort studies are prospective studies where a group of patients is followed over time while the baseline parameters and their evolution are recorded. The measurement tools are chosen before the patients are included in the study. For example, researchers can study the quality of life and the range of motion of the knee in patients before and after total knee arthroplasty using the EuroQol 5-dimension questionnaire and Physilog (BioAGM, CH) device, respectively. Although the "time" bias has now been cancelled with this prospective design, there is still a potential bias when the patients are selected for the study; surgeons can choose "good" cases to appreciate the results of a new procedure.
- The randomized clinical trial is the gold standard design for clinical research. The randomization process allows the control of all potential confondant factors, even the unknown ones by distributing them equally between the two groups of patients under study. In addition, if the patient, the surgeon, and/or the observer are blinded as to which group of intervention the patient belongs, then the best level of confidence in the data recorded is achieved. These are the single-blind (patient or surgeon or observer) or double-blind (patient and surgeon/observer) randomized clinical trials. These studies provide the highest level of reliable evidence.4,5
Apart from these studies, systematic reviews, meta-analysis studies, and decision analysis studies are also used. These are less used analytic studies based on previous publications on a specific topic and allow a global vision. These study designs rely on the quality and availability of the data.
As in all fields of medicine, various study designs are available for clinical research in orthopedic surgery. Although randomized clinical trials, in principle, are preferred, sometimes it is difficult to blind surgeons and patients to the type of surgery performed, or it might be unethical to compare a surgical treatment to the corresponding conservative or placebo treatment. When determining a research design, a surgeon must consider the advantages and disadvantages of the types of designs so that optimal evidence can be obtained for the specific context in which the study is performed.
- Creswell JW. Research Design. Qualitative, Quantitative, and Mixed Methods Approaches. Thousand Oaks, Calif: Sage Publications, Inc.; 2003.
- Hulley SB, Cummings SR, Browner WS, Grady D, Hearst N, Newman TB. Designing Clinical Research. Philadelphia, Pa: Lippincott Williams & Wilkins; 2001.
- Fletcher RH, Fletcher SW, Wagner EH. Clinical Epidemiology - The Essentials. Baltimore, Md: Lippincott Williams & Wilkins; 1996.
- Jolles BM. Why evidence-based medicine in orthopaedic surgery? [Orthopedics Hyperguide Web site]. Available at: http://www.ortho.hyperguides.com.
- Levels of Evidence and Grades of Recommendation. Available at: http://www.cebm.net/levels_of_evidence.asp.