Introduction | Part 1: Hierarchy of Evidence | Part 2: Case Reports | Part 3: Cross-Sectional Studies | Part 4: Case Control Studies | Part 5: Cohort Studies | Part 6: Randomized Controlled Studies | Part 7: Meta-Analysis | Part 8: Systematic Reviews
Status: Draft (this section includes both RCT's and Cross-Over studies)
A Randomized Controlled Study (also: experimental study, controlled clinical trial, random controlled trial or RCT) is an investigation where the researcher assigns subjects to both study and control groups through the process of randomization. The study may be further refined by blinding (alt: masking) the subjects and/or the investigator (single vs. double blinding) to the type of intervention. An additional and distinct process of concealment involves a separate investigator who allocates subjects and who is unaware of which group each study participant is assigned.
The RCT is evidence based medicine's most powerful strategy to establish evidence regarding the treatment of patients. As in the cohort study, controlled clinical trials are prospective; first stating a hypothesis and then following patients forward in time.
As mentioned previously in cohort and case-control studies, the inclusion of a control group into the study is fundamental to reducing or eliminating the effect of bias on the interpretation of outcome. In clinical trials, once the hypothesis is explicitly stated, a group of treatment and control or comparison cohorts are selected. And, to further improve the reliability of the study, subjects are randomly (by chance alone) allocated to study and control groups. The overall probabilities about allocation to a group are known, but no one subject is assigned to a group through an overt or occult bias (or at least that is the goal). The random allocation (by chance) also doesn't necessarily mean that subjects are allocated equally. Randomization is performed by using a pseudo-random number algorithm generated by computer (more common) or historically by random number tables. The fact that distribution may not be equal can be illustrated by the examples at the end of this page.
The process of assigning patients to a group, even when using a random number scheme, must be carefully planned. Subjects must first be evaluated to determine their eligibility in the trial. Past and concurrent co-morbid illness, medications and potential allergies to the class of medication being studies are some examples of selection criteria which should be considered prior to assignment. For some patients, allocation to an observational only group may simply be unethical (e.g. epileptics). Also, if the investigator handling the enrollment knows ahead of time which group a patient belongs to there may additional bias. The process of concealment involves additional masking of the enrollment decision investigator from who is allocated to which group.
Since clinical trials may be undertaken over a considerable time, patients may drop-out of the study, never fully complete the intervention, or deviate from the protocol within their group (e.g. 10% of control patients who may be doing poorly could switch to an alternative treatment outside of their group assignment). Because of the nature of random allocation, there is an inherent bias to the loss of these subjects - their loss disrupts the basis of comparison and the study is no longer considered random. Intention-to-treat analysis is a strategy designed to analyze all original subjects, including those who did not complete the intervention or those who were non-compliant in the group to which they were originally assigned. Noncompliance is a real-world phenomena and inclusion of an intention-to-treat analysis is more realistic when compared to the somewhat artificial completed treatment analysis. However, there is a difference between those who did not complete the intervention and drop-outs. The latter group may no longer even be present to be evaluated and these subjects may make analysis more difficult if not impossible.
The control or comparison group should receive, in an identical fashion to the treatment group, either:
- Observation alone (see Hawthorne Effect): Patients are aware they are being observed.
- Nothing at all: Subjects may not be aware they are being observed.
- Placebo: See placebo effect
- Gold Standard Therapy: This is unquestionably the current best measure used to define a disease. This does not mean however that the gold standard is 100% accurate in diagnosing a disease (few tests reach sensitivity or specificity of 1.0) nor does it mean that there may be potentially different gold standards to define a disease within different clinical scenarios or within individual species.
Once a hypothesis is stated treatment cohorts subjects are randomized and allocated, the test may proceed in one of three ways:
- Open-label: Subjects and investigators are aware of who has been assigned to a group
- Single-masked (blinded): Patients blinded to their group assignment
- Double-masked (blinded): Patients and investigators both blinded to group assignments
Trials are focused on the effectiveness of treatment as well as other outcomes such as side-effects, adverse events, etc. Although a majority of clinical trials make a comparison between two treatment interventions, they may also, additional treatments are evaluated concurrently with an expected increase in the complexity and interpretation of the results.
While adverse event reporting has gained momentum in human medicine, structured and reliable capture of adverse event information is rather negligible in veterinary medicine. Controlled standard terminologies or ontologies are used for pharmacovigilence such as MedDRA (Medical Dictionary for Regulatory Activities) and CTCAE v.4 (Common Terminology Criteria for Adverse Events used on oncology clinical trials). CTCAE 4.0 uses MedDRA terms with the addition of a qualitative grading scale to record severity. The Veterinary Cooperative Oncology Group has adopted CTCAE for creation of a terminology unique to the veterinary domain for reporting adverse events following chemotherapy or biological neoplastic therapy.
The term efficacy when used to describe an outcome from a trial, essentially describes the reduced probability of experiencing an adverse outcome or the extent it provides a benefit within the environment of a clinical trial. Effectiveness however is distinctly different; it describes the ability of treatment to produce a benefit under normal clinical conditions and not within the artificial environment of a controlled clinical trial. This may seem like an uneccessary semantic exercise, however it is is important because your target patient may not fall under the same conditions as the research subjects given the presence of co-morbid diseases, other concurrent medications, patient type, etc.
|From 2000-2010, meta-analyses were reported most often in Poultry Science (280), Journal of Animal Science (278), JAVMA (270), American Journal of Veterinary Research (215), Veterinary Anaesthesia and Analgesia (197), Veterinary Record (185), Journal of Dairy Science (173), Veterinary Parasitology (145), Journal of Veterinary Pharmacology & Therapeutics (142), Journal of Veterinary Internal Medicine (125).|
|Top concept representation (MeSH Terms), excluding randomized-controlled trials, were dogs (physiology), cattle (physiology), animal feed (analysis), dog diseases (drug therapy), diet, treatment outcome, animal nutritional physiological phenomena, horses (physiology), chickens (growth and development), and time factors.|
A randomized double-blind comparison of ondansetron and metoclopramide in the prophylaxis of emesis induced by cyclophosphamide, fluorouracil, and doxorubicin orepirubicin chemotherapy. Bonneterre J, et al. J Clin Oncol 1990 Jun;8(6):1063-9
In this study, 75 breast cancer patients participated in a double-blind crossover study to compare the efficacy and safety of ondansetron and metoclopramide. The authors noted a"period" interaction in the "analysis of emetic response in the first 24 hours". This required a parallel analysis of cohorts for the first treatments alone. It was also noted that there was a significant patient preference for ondansetron over metoclopramide (63% v 26%).
This is a paired design method where the same study subject is exposed to both the target (or sequence of treatments) as well as the control and individual outcomes are assessed for each. The advantage is patients already enrolled in a study can be used to evaluate outcomes on both sets of cohorts and thereby eliminate some of the administrative and financial costs of testing a larger number of individuals. And, by having patients exposed to a series of treatments, patient preferences, such as the preference for ondansetron over metoclopramidein the study mentioned above, can be assessed as well. However, inherent to patients being subjected to multiple treatments, they are in the trial longer which increases the chance patients will drop-out before the completion of the trial . And because subjects are the target for more than one intervention, cross-over trials are generally not appropriate for studies involving surgery, critical-care illnesses (e.g. ARDS), cancer, infectious diseases, etc. Cross-over trials are better designed for diseases that tend to be chronic such as osteoarthritis, allergic bronchitis, hypertension, etc.
Cross-over trails can exhibit the phenomenon of carry-over where the effects of one treatment on a patient persists into the next treatment period. This relationship of carry-over effects is often expressed as a period interaction. Investigators may not even be aware of the influence on outcomes on the same subject between each intervention. When the period effect is considered, a simple analysis can still be considered if it is feasible to assume a negligible effect of the carry-over and the patients are randomly allocated to each sequence as well. One method to eliminate carry-over effects into the next intervention is to implement a 'washout-out' period.
|From 2000-2010, cross-over studies were reported most often in the American Journal of Veterinary Research (137), Journal of Veterinary Pharmacology and Therapeutics (128), Veterinary Anaesthesia and Analgesia (55), Equine Veterinary Journal (49), Journal of Animal Science (45), Journal of Dairy Science (38), Equine Veterinary Journal Suppl (37), Journal of Veterinary Internal Medicine (37) and Veterinary Record (30).|
|Top concept representation (MeSH Terms), excluding cross-over studies, were horses, dogs, injections (intravenous), random allocation, area under curve, administration (oral), dose-reponse relationship (drug), time factors and heart rate|
|Example (prospective, randomized, cross-over trial)|
Warming with a forced air warming blanket minimizes anesthetic-induced hypothermia in cats.
In this study, 8 cats were anesthetized and were subjected to 4 separate trials. Each trial consisted of a 90-minute anesthetic period separated into two consecutive 45 minute periods. A forced-air warming blanket was either activated or inactivated to create the four trials (off/off, on/off, on/on, off/on).
Random controlled trials: A users guide | Alejandro R Jadad | BMJ Publishing
Importance of clinical trials in evaluating therapy of renal diseases. Polzin DJ, Vet Clin North Am Small Anim Pract. 1996 Nov;26(6):1519-25. Review.
Veterinary randomized clinical trial reporting: a review of the small animal literature. Lund EM, J Vet Intern Med 1998 Mar-Apr;12(2):57-60
This series has been loosely organized from a set of lectures given by the author within graduate courses in Biomedical Informatics beginning in 2002. Content is being edited to improve organization, depth, correct inaccuracies as well as updates with new information during Winter/Spring 2012. Feedback is greatly appreciated. © 2011, Stuart Turner.