Data from clinical trials involving human participants are essential in establishing an evidence base about the safety and effectiveness of our treatments. This first article describes the steps involved in designing and setting up a clinical trial, from establishing the research question(s) to searching the literature. Acquiring some knowledge about how to set up a clinical trial will allow the conscientious clinician to use the most relevant information to provide the highest possible standards of clinical care for his/her patients.
CPD/Clinical Relevance: Even if a clinician is not, has never been, nor is ever planning to be involved in research, he/she should understand and be able to interpret the data from clinical trials.
Article
Clinical trials involve observations or interventions undertaken with human participants (usually patients) to provide information concerning specific questions about the safety or effectiveness of treatment. Laboratory and animal studies might provide some initial indications in these areas, however, they almost always lack clinical validity and can rarely replace clinical data obtained in a scientific manner.1 The evidence derived from clinical trials may be used together with clinical judgement and patients' values as part of an evidence-based approach to care (Figure 1).
Figure 1. What is evidenced-based medicine?
There are various study designs (which can be broadly classified into quantitative and qualitative approaches), but the most appropriate design depends upon the research question to be answered. Current conventional wisdom is that a quantitative randomized controlled trial (RCT), which attempts to reduce the potential influences of both the patient and researcher, is the best design for interventional studies in both medicine and dentistry. However, there may be practical, ethical or cost considerations that prevent the use of this design. The study of human motivations and behaviour may require a qualitative approach but, for the purposes of this article, we will mainly focus on the design and setting up of a clinical trial using a quantitative approach.
A number of organizations can provide guidance on designing, conducting, analysing and publishing clinical trials. The first organization to contact is the local Clinical Research Network (CRN), which works closely with both the National Health Service (NHS) and Higher Education Institutional Research and Development offices (R&D). The CRNs are funded by the National Institute for Health Research (NIHR), a UK organization that supports research in the NHS. Applications for a clinical trial can be developed with the help of a local Clinical Trial Unit (CTU), which has expertise in all areas of trial design and management.
The NIHR has devised an online clinical trial toolkit routemap designed to help researchers set up and manage a Clinical Trial of an Investigational Medicinal Product (CTIMP).2 At first glance it may seem quite complicated, as the regulations involving CTIMPs are more stringent than those regulating most orthodontic trials. The routemap, however, does provide good guidance about the many requirements that must be satisfied when setting up and running a clinical trial.
This article will describe the steps involved in designing and setting up a clinical trial, from establishing the research question(s) to searching the literature and writing a formal protocol.
Designing the research question for the clinical trial
Step 1: Identify a knowledge gap
The first step for any research project is to decide what question the study is attempting to answer. When the question(s) has or have been identified, it is useful to determine if anyone has attempted to answer the question(s) before. If the answer is ‘yes’, then ask ‘how good was this attempt’?
The NIHR recommends that the development of a clinical trial starts with a systematic review of the existing literature.3 A systematic review will provide reliable information to justify your research and should help develop your study design.3 Do not worry if the research has been carried out before and therefore is not considered innovative. Almost every area of clinical interest, certainly in dentistry, requires confirmation that the original results are reproducible, applicable in a variety of settings and with a large number of the target population (ie generalizable).
Step 2: Formulate the research question
The FINER criteria can be used to help formulate a good research question (Table 1).4
FINER
CRITERIA
EXPLANATION
F
Feasible
Evidence of previous work (eg pilot, feasibility studies, scientific literature)
Manageable in scope (eg subjects, clinical expertise, time and money)
I
Interesting
Getting the answer intrigues investigators, peers and community
N
Novel
Investigators are in equipoise regarding the question
Research findings are anticipated to resolve clinical uncertainty
E
Ethical
A study that institutional review board would approve
R
Relevant
To scientific knowledge
To clinical and health policy
To future research
Step 3: Focus the research question
Focusing the question for the clinical trial can be undertaken using the acronym PICO:
P = Patient, population or problem;
I = Intervention being investigated (eg medical, surgical, preventive);
C = Comparator or control (best proven intervention(s), no intervention, placebo);
O = Outcome(s) attributable to a specific disease, condition or injury.
Step 4: Decide on the study design
Clinical trials can have several designs (Figure 2):
Observational studies (eg cross sectional, cohort, case control);
Experimental or interventional studies (eg randomized controlled trials (RCTs)).
Figure 2. Flow chart of clinical trial study designs. Strength of evidence is colour coded: high = green, moderate = orange, low = red.
Observational studies are generally non-interventional, because the researcher is not able to influence the treatment or environment in which the study takes place, usually for practical or ethical reasons. If an observational study is deemed the most appropriate or the only feasible approach, the researcher then needs to decide if data will be collected for a group of individuals:
At one point in time (eg cross-sectional);
Over the course of time (eg longitudinal);
Without an outcome of interest, being divided into subgroups based upon their exposure to a potential cause to determine the development of an outcome of interest (eg cohort study);
With an outcome of interest being compared to a suitable control group to determine the occurrence/timing of exposure to a potential cause (eg case-control).5
Once this has been decided, the researcher then needs to decide if data will be collected retrospectively or prospectively. Prospective data collection generally provides stronger evidence, because changes within individuals can be assessed and any loss of data or participants from the study can be accounted for.
Experimental studies involve comparing the outcomes in a group of individuals who have received a treatment (usually novel or new) with a group who have not received the treatment or received an alternative treatment (usually the conventional or placebo). The decision about who receives the new treatment is either allocated randomly (ie by chance) or by a quasi-random technique, such as alternates (not ideal). This aims to ensure that the groups are comparable at the start of the trial, with any differences that might influence the outcome (eg confounders) evenly distributed between the groups. This will depend on the correct use of an unpredictable, random allocation order and hiding that sequence until assignment, to remove any possible biases of the treating clinician or the patient on assignment (ie blinding).5
The theoretical strengths of evidence provided by each type of study design are shown in Figures 2 and 3. Ideally, once selected, the study design should be included in the research question, so that it can be easily indexed and identified from electronic databases.6
Figure 3. The hierarchy of the evidence-based medicine pyramid. Strength of evidence is colour coded: high = green, moderate = yellow/orange, low = red.
Step 5: Eliminate bias
A bias is a systematic error, or deviation, which could affect the interpretation of the results and lead to an over or underestimation of the intervention effect.7 There are several sources of potential bias within a clinical trial (Table 2) and these should be addressed during the design of the trial.7
Type of bias
Description
Factors to consider in clinical trial design to avoid bias
Selection bias
Significant differences in the baseline characteristics of the groups due to bias in the selection of participants
Large sample sizes of the target population and clear inclusion and exclusion criteria
Allocation bias
Significant differences in the baseline characteristics of the groups due to bias in the allocation of participants
Random sequence generationAllocation concealment
Performance bias
Bias due to participants and investigators knowing which intervention was received
Blinding of participants and investigators to intervention
Detection bias
Bias due to assessors knowing which intervention was received
Blinding of outcome assessment
Attrition bias
Bias due to significant differences between the groups in terms of withdrawals
Per protocol vs intention to treat analysis
Incomplete outcome data can be accounted for by either worst case scenario, last observation carried forward, imputation and sensitivity analysis
Reporting bias
Bias due to selective reporting of outcomes
Compare outcomes in protocol/research methodology to those reported
Publication bias
Selective publication of reports, usually in favour of statistically significant results
Trial registration
Step 6: Refine the specific objectives and hypotheses of the trial
Objectives: Questions the trial is designed to answer;
Hypothesis: Specific question being tested to help meet the objectives of the trial and amenable to statistical testing.
Step 7: Patient and Public Involvement
Patient and Public Involvement (PPI) is an important consideration to confirm that the research question is important and relevant to the people it directly affects. The trial should also be practical and feasible.
Step 8: Write a protocol for the clinical trial
Writing a formal protocol is helpful for a number of reasons:
Provides a step-by-step guide which can account for any problems and concerns (eg bias and confounding);
Essential for obtaining funding, sponsorship and the necessary regulatory approvals (eg ethical and NHS research and development (R&D) approvals);
Should limit the possibility of undeclared changes once the trial has begun and/or selective outcome reporting.
Summary
This first article has described the steps involved in designing and setting up a clinical trial from establishing the research question(s) to searching the literature. A forthcoming second article will describe how to write a protocol and gain the necessary approvals for a clinical trial.
