Randomised Clinical Trials - Essay Example

This report provides a researched account of clinical trials. Initially a history of the development of clinical trials from primitive resources and methods to modern day technology intensive trials is chronicled. This is followed by a detailed elaboration on the ethical principles guiding research involving human subjects. Ultimately various methods to determine sample size for a randomized control trial have been studied and illustrated with examples. Methodology Internet has been the primary source of information for this report. Several journal articles were accessed from the Liverpool Metalib, few others were open access and some required registration. The critical resource of D G Altman's "Practical Statistics for Medical Research" was also available at Google Books. Various search terminologies were used which include "Evolution/History/Review of Clinical Trials", "Ethical issues of clinical trials", "Nazi medical war crimes", "Nuremberg/Helsinki/Belmont" codes. Formulae for the sample size calculations were also accessed from journal publications and some power point presentation slides which have been cited and calculations were performed manually. History of clinical trial Owing to the practical significance of implications of therapeutic interventions for patients, historians have shown an active interest in the charting out of evolution of clinical trials. Histories of clinical trials have been recorded and they have analysed the development of quantification in therapeutic evaluation, the emergence of probabilistic thinking, the application of statistical methods and theory and the sociology, ethics and politics of clinical trials as succinctly summarized by Chalmers (1) in 2001. The basic concept behind the modern day clinical trial is not a new one. In fact, the earliest recorded reference to something resembling a clinical trial can be found in none other than the Bible. The extract, which is found in the Book of Daniel, describes the efforts of the Babylonian king, Nebuchadnezzar II (605-562 BCE), to compare his recommended diet, consisting of meat and wine, with one of legumes and water over a 10 day period. At the end of the trial the king noted that those on the legumes and water diet were fitter than those who had been fed meat and wine and switched the latter to legumes as well. Inclusion of blinding and placebos to reduce observer biases comprise an important factor in planning an effective RCT. Records of these point out to as early as eighteen century when Dr.Benjamin Franklin was appointed by King of France in the Royal Commission to judge the authenticity of "Animal Magnetism" which alleged that sicknesses were caused by some apparent "obstacle" in the flow of body fluids and could be removed by the touch of a physicians finger or by pointing an iron rod. In a retort to the small percentage of success, Franklin replied, " the Spirits given by the Hope of Success them to exert more Strength in moving their Limbs " Clearly, Franklin was aware of what we now call 'the placebo effect' as described by Huth (2) in 2006 Further, Booth (3) in his book in 2005 documented the role of Physician John Haygarth in 1810-1820 attacked the widespread quack of Perkinism which involved "redirecting the natural body electricity" by using bi-material rods and was the first physician to carry out a single blind clinical trial using a placebo. May it be the instance of the challenge put forth by Flemish physician Jean Baptist Von Helmont of ensuring that like is compared with like in the case of people suffering from fevers, pleurisies without bloodletting in the 1700's or Amberson flipping a coin for unbiased allocation for assigning treatment in the Tuberculosis trial in 1938; medical practitioners have considered the ethical aspects of fair allocation and avoidance of undue advantage since a long time now. Unbiased comparison groups according to Chalmers (1) in those times would result either from "schedules (coin tosses, selection of different coloured beads from an urn, reference to random number tables, and so on) or by using unbiased systematic processes such as strict alternation or rotation of patients in a consecutive series to one of two or more comparison groups." Besides excluding allocation bias, the real evolution in therapeutic "intervention" also occurred in the eighteenth century. A number of trials were conducted around that time utilising both historical and concurrent controls. Perhaps the most renowned out of these is James Lind's classical scurvy experiment whilst aboard the British naval ship the H.M.S. Salisbury. During this mini trial Lind aimed to test the efficacy of six potential antiscorbutic agents on twelve sailors afflicted with scurvy. After a period of 6 days those who had been given the citrus fruit treatment (oranges and lemons) were recovered and back on their feet. This was then followed by the contributions of P.C.A Louis in 1836. Louis observed that early bloodletting seems to reduce the duration of a pneumonitis disease in patients who survive from this disease but may also increase the overall short-term mortality. Modern Analysis of Louis's data also confirms with his original conclusions which was ascertained by Morabia (4) in 1997 Hrbjartsson (5) in 1998 shed light on a very good example of a controlled therapeutic experiment in the 19th century is the evaluation of anti diphtheria serum in 1898 by Danish Nobel Laureate physician Johannes Fibiger. Fibiger stressed on the four methodological features of the trial. Though the knowledge of Diphtheria epidemics was limited in those days, he realized that large numbers of subjects and long duration was required for the study. Then he also took account the seasonal variation in the mortality count of Diphtheria. His fourth contribution was a concise discussion of the importance of random allocation and bias. However, the most important milestone in the history of the clinical trial is probably the publication of the Medical Research Council's streptomycin trials in 1948. Designed by Sir Austin Bradford Hill, who is often credited for having conceived the modern day RCT, these trials set an example for future research. The distinguishing feature of these trials was randomisation of subjects. This concept was introduced by RA Fisher in 1926 and had already been applied to agricultural research. Thus one can observe a dynamic succession and evolution in Clinical Research with regards to random allocation, planning, conduct and reporting of trials. Methodology of Clinical Trials Before proceeding to the ethical questions raised in the execution of clinical trials, let us first, in brief understand the flowchart of steps occurring in a clinical trial. Thus we can identify with the exact steps at which these ethical questions are raised. The Broad Phases of Clinical Trials include: I. Dose Determination (Threshold) II. Efficacy at fixed dose (Safety and Efficacy) III. Comparing Treatments (Randomized Control Trials) IV. Late/Uncommon Effects (Expanded Safety) Schematic Representation of a Simple RCT Recruit and Register RCT Study Sample Rx A (n=100) Rx B(n=100) Treat, Monitor, Evaluate and Compare Outcomes There are various classifications of clinical trials. The best classification of frequently used terms was offered by Jadad (6). According to Jadad (6), randomized controlled trials can be classified as to "the aspects of intervention that investigators want to explore, the way in which the participants are exposed to the intervention, the number of participants included in the study, whether the investigators and participants know which intervention is being assessed, and whether the preference of nonrandomized individuals and participants has been taken into account in the design of the study". This has also been described well in Stolberg's (7) paper. The classification scheme is adopted from Jadad (6) and is completely paraphrased by me suiting to my understanding of the topic. Randomized Controlled Trials Classified According to the Different Aspects of Interventions Evaluated Explanatory or pragmatic trials: Explanatory trials are those which assess whether or not a trial is effective and if so how. Pragmatic trials determine not only test whether the trial works but also all the consequences of the intervention and its effects in daily use. Efficacy or effectiveness trials: Interventions conducted in ideal circumstances of temperature, pressure and period are termed efficacy interventions whereas those conducted during everyday conditions are effectiveness trials. Phase 1, 2, 3, and 4 trials.-These terms describe the different types of trials used for the introduction of a new intervention, traditionally a new drug, but could also encompass trials used for the evaluation of a new embolization material or type of prosthesis, for example. The Phase 1 trials are conducted after sufficient animal testing has been performed and it has been deduced that the drugs are safe for use in human bodies. Phase 1 studies are usually performed on healthy volunteers. Post the successful completion of phase 1 which shows no adverse reactions in healthy individuals, phase 2 begins. Phase 2 involves evaluating the efficacy of the concerned drug by administering through various routes and types(syrup,pills,injections). Phase 2 studies focus on efficacy while still providing information on safety. Phase 3 studies are essentially effectiveness trials which are conducted only after it has been deduced by phase 1 and 2 that the drug is safe and provides sufficient therapeutic advantage. Phase 4 studies are tantamount to postmarketing studies of the intervention; they are performed to identify and monitor possible adverse events not yet documented. Randomized Controlled Trials Classified According to Participants' Exposure and Response to the Intervention Parallel design.-When participants are exposed to a single intervention, it is termed as a parallel design. Crossover design: There are instances where a clinical study might involve multiple interventions. Thus a crossover design is one in which each patient receives each of the intervention in a random fashion. Factorial design: The use of factorial designs is considered when a number of therapies are being evaluated in combinations. Thus for example in a 3-by-3 factorial, three interventions are tested each alone and in combination, and compared to a control. This means that investigators can efficiently test two interventions with only marginal increases in sample size. In addition, the benefits of treatment combinations can be evaluated in a controlled manner. This design is most useful when interactions are either very strong or nonexistent as suggested by Hebert and colleagues (8) in 2002 Randomized Controlled Trials Classified According to the Number of Participants "Randomized controlled trials can be performed in one or many centers and can include from one to thousands of participants, and they can have fixed or variable (sequential) numbers of participants" according to Jadad (6) "N-of-one trials."-Randomized controlled trial with a single participant is termed as "n-of-one trials" or "individual patient trials." Randomized controlled trials with a simple design that involve thousands of patients and limited data collection are called "megatrials". Usually, megatrials require the participation of many investigators from multiple centers and from different countries. Sequential trials.-A sequential trial is one in which the investigator does not estimate the sample size of participants in advance. Rather, the trial is continued until either of the intervention shows a significant advantage over the other or there is clearly no significant difference between the interventions over a large breadth. Fixed trials: Fixed trials as the name suggests have pre decided sample size estimation. These sample sizes are calculated using various statistical formulae depending on the design and outcome of the study. This is explained further in detail. Randomized Controlled Trials Classified According to the Level of Blinding In addition to randomization, the investigators can incorporate other methodologic strategies to reduce the risk of other biases. These strategies are known as "blinding." The purpose of blinding which is sometimes also referred to as masking is to reduce the predictive value or subjective bias while planning an RCT." An open randomized controlled trial is one in which everybody involved in the trial knows which intervention is given to each participant. A single-blinded randomized controlled trial is one in which only one group of individuals involved in the trial (usually patients) does not know which intervention is given to each participant. A double-blinded randomized controlled trial, on the other hand, is one in which two groups of individuals involved in the trial (usually patients and treating physicians) do not know which intervention is given to each participant" as explained by Jadad (6) Controlled Trials Classified According to Nonrandomized Participant Preferences Jadad (6) continues to explain that "it may be the case that suitable individuals may refuse to participate in a randomized controlled trial. Then again there might be some which may participate provided they are allocated to a particular arm (treatment or placebo). At least three types of randomized controlled trials take have been designed taking into account the preferences of eligible individuals as to whether or not they take part in the trial. These are called preference trials because they include at least one group in which the participants are allowed to choose their preferred treatment from among several options offered. Such trials can have a Zelen design, comprehensive cohort design, or Wennberg's design." Ethics Development of RCT's took place in the twentieth century since the World War II. The ethical issues inherent in human experimentation were at the forefront after the Nazi "medical" war crimes. The German medical team employed some extremely brutal and inhuman experiments on the prisoners. The doctors involved in these mal practices were tried at the Nuremberg trial. The judgment by the war crimes tribunal at Nuremberg laid down 10 standards to which physicians must conform when carrying out experiments on human subjects in a new code that is now accepted worldwide as the Nuremberg Code. This judgment established a new standard of ethical medical behavior for the post World War II human rights era. The pivotal requirement laid down in the Nuremberg Code was informed voluntary participant consent. The principle of voluntary informed consent protects the right of the individual to control his own body. This code also recognizes that the risk must be weighed against the expected benefit, and that unnecessary pain and suffering must be avoided as pointed out by Schuman (9). The Nuremberg code further paved the path for the Helsinki Declaration. Adopted by the 18th World Medical Assembly, Helsinki, Finland,June 1964,the charter established a series of non binding principles of basic, Clinical Research and Non Clinical Research involving human subjects as appearing in World Medical Orhanisation (10) A significant modification in the Helsinki Declaration was the clause that informed voluntary consent was no longer essential in the case of non responsive patient like in the instance of a cardiac arrest and could be given proxy for by a guardian thus facilitating treatment. Clinical trials often pose ethical concerns. Advocates for these studies sometimes respond to challenges largely by reiterating important discoveries facilitated by clinical experiments. Has importance of science or knowledge gained a counterweight to concerns about ethics Is it justified that clinical trials sacrifice well being of current subject participants for the future benefits of other individuals The concept of "Equipoise", or the "uncertainty principle" has been central to the development of experiments involving humans. It holds that there be genuine uncertainty regarding the outcome of the intervention and only then can patients be enrolled for the trial. Thus mathematically "it is required that for the two treatments A & B under consideration to be in equipoise require that the probability of outcome of each experiment be 0.5" as put forth by Fries and Krishnan (11). However this definition appears to be too rigid as it is highly unlikely that one formulates an experiment with no predictive value or devises another control intervention with as much predictive value as the proposed treatment. Thus the definitions of equipoise were required to be relaxed. In a paper that reviewed Industry sponsored RCT's accepted for the 2001 American College of Rheumatology favoured all 45 of the 45 industry sponsored RCT's. The reason behind this hundred percent success was probed further by Fries and Kishwar (11) and they came to four plausible explanations. Firstly, a placebo control was used as an arm in 32 of the 45 studies and pre study evidence suggested that drug was superior to the placebo. One then questions whether equipoise is violated while employing placebo as one of the arms. However Stang (12) and colleagues assert that "a placebo-controlled trial may be ethically acceptable, even if proven therapy is available, for compelling and scientifically sound methodological reasons its use is necessary to determine the efficacy or safety of a prophylactic, diagnostic or therapeutic method." Stang (12) observed insightfully that "in the absence of a placebo group, it may be impossible to interpret a drug's potential for harm." Secondly, out of the 45 studies 22 studies were derived from clinical trials already conducted previously and had just referred to different end points viz.cost or quality of life. "Publication bias" according to Turner (13) results in the tendency for studies that are positive to be published and the tendency of negative and indeterminate studies to not be released. This also might attribute to the roseate picture of hundred percent success of industry sponsored RCT's. Lastly, another factor called "design bias" may have an implication on the results of an RCT and is discussed at length below. The concept of Equipoise, though at face value appealing, is self contradictory in nature. From an industry perspective the drug development process must involve designing for success. "A funding commitment by a for-profit entity to an RCT that may cost hundreds of millions of dollars simply will not be made unless a positive outcome may be predicted with considerable certainty. This is termed as "design bias" However is design bias necessarily a bad thing question is a provoking question by Fries and Krishnan (11) "Should study be conducted with no promise of success or study drugs which are superior than a placebo with no knowledge of their efficacy over existing remedies" Then there is the issue of social tradeoffs. Some would argue that, in a world with limited resources, the benefit of a therapy must also be related to the resources consumed "[e.g. is it good medicine to spend $500,000 per patient for an average gain in life of 3 weeks ]" as questioned by Fisher (14) in 1999. Should studies that fail to identify an appropriate, perhaps even a narrow therapeutic niche for the drug be conducted Some authors argue that increased number of trials would actually assist in limiting cost of medical care and would generate large databases for public review and debate. To enrol humans in large RCTs without preliminary studies might pose truly major risks to participants, but after preliminary studies have been conducted true uncertainty no longer exists. "The principle of equipoise becomes the paradox of equipoise" as aptly summarised by Fries and Krishnan (11). A step ahead of "Equipoise" was The Belmont Report which adopt a more inclusive approach. The Belmont Report attempts to summarize the basic ethical principles identified by the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research as appearing on NIH (15) The Belmont principle instead of giving priority to equipoise identifies the three principles of autonomy, Beneficence and Justice. Thus a patient unless in a condition of diminished mental capacity or under force has the right to decide to participate in trials provided all the aspects of research are communicated to him. Beneficence implies that study designs to the greatest degree possible maximize benefits and minimize risks to the patient. Justice refers to fairness in distributing the benefits and burdens of research. Finally the role of Data Monitoring Committees is also reviewed." It is argued that clinical trials with data monitoring committees that use statistical stopping guidelines should generally not be stopped early for large observed efficacy differences because efficacy estimates may be exaggerated and there is minimal information on treatment harms." It is not the obligation of the Data Monitoring Committee to merely consider large differences in efficacy but also to balance the significant difference with possible harm which may not be apparent before the trial. This view was put forth by Goodman (16) Finally, RCT's are considered as the gold-standards for proving advantage of novel clinical interventions because of their ability to reduce bias. However it has been observed that this tool is being increasingly misused and manipulated. Many reviews have documented deficiencies in reports of clinical trials. According to Moher's (17) research, "information on the method used in a trial to assign participants to comparison groups was reported in only 34% of 616 reports indexed in PubMed in 2006.Similarly, only 45% in 2006 reported a sample size calculation. Reporting is not only often incomplete but also sometimes inaccurate. Of 119 reports stating that all participants were included in the analysis in the groups to which they were originally assigned (intention-to-treat analysis), 15 (13%) excluded patients or did not analyse all patients as allocated." Thus a team of editors and scientists have charted what is known as the CONSORT (Consolidated Standards of Reporting Trials) First published in 1996, revised in 2001 and again in 2010 CONSORT aims at improving the reporting of clinical trial by providing a checklist of items which should be Strictly reported in the Trial results. The checklist could be viewed in Moher's (17) extensive work about the same topic. Endorsement of International Journals and Medical Groups lays a strong foothold for CONSORT. Thus a checklist of various items such as rationale behind study design, description of trial design including allocation ratio, changes to methods after trial commencement, eligibility criteria of patients and settings and locations of data collection et al has been drawn. Thus one can observe that consistent efforts have been undertaken to provide a morally sound yet practically feasible Clinical Trial solution. Sample Size Calculation: One of the most common requests made to statisticians by Medical Researchers is to determine the sample size required to conduct a trial conclusively. "The sample size is the number of patients or other experimental units included in a study, and determining the sample size required to answer the research question is one of the first steps in designing a study", as defined by Noordzij (18) in 2010 Yet mathematical formula for determining sample size have been developed only in recent years. A review of 172 RCT's performed by Ambros (19) in journals like NEJM and Lancet (1973-76) showing no estimation of numbers prior to the trial prove the fact. Estimation of sample size is crucial in medical research where the sample size is just large enough to provide a reliable answer to the research question. If the sample size is too small, it may lead to inaccurate or non conclusive results and if the sample size is too large extra effort might be involved, the concerned therapy might be risky and costs will be high. Sample size estimation provides a realistic idea of the cost of the project and also provides the investigator to determine the duration of their study. For example, "the calculated sample size may be 50 (a manageable number) but if the yearly accrual of subjects is 10 (assuming all subjects give consent to be in the study), it will take at least five years to complete the study! In that case a multicentre study is encouraged." As illustrated by Chan (20) in 2003 A scientific study begins with postulating a Null Hypothesis. A null hypothesis is always set to be rejected by an Alternate Hypothesis. Thus to prove that drug A is superior than drug B, the null hypothesis is drug A is NOT superior than drug B and the claim is rejected by proving the null hypothesis false. In doing so, we might sometimes commit a type I error wherein we reject the null hypothesis when it is true. When by statistical testing a significant difference is observed in two groups it can be attributed to two reasons. Firstly, that there is an actual difference between group A and group B and secondly this difference might arise out of chance. The p value gives us the amount of chance. Thus a p value of 0.05 implies that there is a 5% probability that the difference in the two groups has risen due to chance. Thus the researcher has to individually determine what value of p to be assigned. A type II error is performed when one is not able to reject the null hypothesis when it is actually false. This is determined by the Power of the study. "Conventionally, the power is set at 80% or more, the higher the power, the bigger the sample size required. To be conservative, a two-sided test (more sample size required) is usually carried out compared to a one-sided test which has the assumption that the test therapy will perform clinically better than the standard or control therapy" concludes Chan (20) 2003 If the primary outcome of interest is dichotomous (success/failure, yes/no, etc) then a different formula is employed. For example, 15% of the subjects on the standard therapy had a successful outcome and it is of clinical relevance only if we observe a 50% (effect size) absolute improvement for those on the study therapy (i.e. 35% of the subjects will have a successful outcome). How many subjects do we need to observe a significance difference For a two-sided test of 5%, a simple formula to calculate the sample size is given by m (size per group) = c X where c = 7.9 for 80% power and 10.5 for 90% power, 1 and 2 are the proportion estimates. Thus from the above example, 1 = 0.15 and 2 = 0.35. For a 80% power, we have m (size per group)=7.9 X [0.15 (1 - 0.15) + 0.35 (1 - 0.35)]/(0.15-0.35)2 = 7.01 Hence 7 X 2 = 14 subjects will be needed in each group. For example suppose we were to calculate the sample size required to assess whether a new drug for Multi drug resistant strains of Tuberculosis increases survival by 5%. Thus we have, New Tuberculosis Treatment with 95% survival rate, standard Standard Tuberculosis cure at 90% survival. Thus according to above formula, P1=0.9 (Std Treatment) P2=0.95 (New Treatment) m (size per group)=7.9 X [0.9 (1 - 0.9) + 0.95(1 - 0.95)]/(0.9-0.95)2 = 10.86 =11 subjects per group with 80% power of study Simplest formula for a continuous outcome and equal sample sizes in both groups, assuming alpha ( or the Type I error) = 0.05; beta ( or the Type II error) = 0.20 and Power = 0.8 or 80% is given by Noordzij (19) as follows: N = Where, N =the sample size in each of the groups 1 =population mean in treatment Group1 2 =population mean in treatment Group2 1 2 =the difference the investigator wishes to detect 2 =population variance(SD) a =conventional multiplier for alpha=0.05 which is 1.96 b =conventional multiplier for power=0.90 which is 1.28 Consider the example of a trial of two interventions to reduce blood pressure, where the primary outcome will be mean fall in systolic blood pressure. Lets stipulate a clinically important difference to be 15mm of Hg. Lets assume the standard deviation of the difference between successive systolic readings = 20 mm Hg. Assigning power to be P = 0.90 = 90%, = 0.05 = 5%. We have 1 2= 15; Thus (1 2)^2= 225 N= 2(1.96+1.28)2*400/225 =37.32 samples in each group i.e 74 samples in all The same sample size in the above example can be determined by Altmans Nomogram (21) which is a simple smart graphical method of determining sample size. It consists of three scales, two vertical and one slanting. The left vertical scale is marked for standard deviation, the right vertical scale is the scale of power and the center slanting scale gives the sample number. Thus for the above example mark standardized difference= 1 2 / = (15/20) = 0.75 Thus the line connecting the stipulated standardized difference and power cuts the slanting scale precisely at 74 which gives us the sample size in accordance with the previous method. Once the sample size is determined, one can then proceed onto conducting the actual trial. The difference between the arms can then be determined by various statistical tools like t test or chi square test or Wilcoxon test and Mann Whitney in case of non parametric samples to determine whether there is a statistically significant difference or not. Thus Randomized Clinical Trials have evolved over a long period of time to assume their current stature. May it be to assess the Prenatal use of drug AZT: will it prevent HIV infection in newborns of HIV positive mothers or Laser photocoagulation: will it reverse age related macular degeneration, Tight glycemic control: will it reduce vascular complications of Type I and Type II diabetes Routine PSA screening: will it reduce mortality from prostate cancer ; they have been applied in numerous instances, they have been used in numerous occasions to come to rational conclusions about crucial medical inventions thus benefitting the society at large. For RCT's to remain a gold standard stronger vanguards of Ethics will be required to make sure that this powerful tool does not fall in wrong applications. 1. Chalmers I. Comparing like with like: some historical milestones in the evolution of methods to create unbiased comparison groups in therapeutic experiments. Int J Epidemiol. 2001 October;30(5):1156-1164. Available from: http://dx.doi.org/10.1093/ije/30.5.1156. 2. Huth EJ. Benjamin Franklin's place in the history of medicine. The James Lind Library (www.jameslindlibrary.org). 2006 Accessed Wednesday 28 April 2010. 3. Booth C. John Haygarth, FRS (1740-1827): a physician of the Enlightenment. Philadelphia: American Philosophical Society, 2005 (ISBN: 0-87169-254-6) 4. A Morabia P. C. A. Louis and the birth of clinical epidemiology. J Clin Epidemiol. 1996 December; 49(12): 1327-1333. 5. Hrbjartsson. A, Gtzsche.P, and Gluud C The controlled clinical trial turns 100 years: Fibiger's trial of serum treatment of diphtheria BMJ. 1998 October 31; 317(7167): 1243-1245. 6. Jadad AR. Randomised controlled trials: a user's guide. London, England: BMJ Books, 1998 7. Stolberg HO, Norman G, Trop I. Randomized Controlled Trials. Am J Roentgenol. 2004 December;183(6):1539-1544. Available from: http://www.ajronline.org/cgi/content/abstract/183/6/1539. 8. Hebert, PC The design of randomized clinical trials in critically ill patients. CHEST April 2002 121(4) 1290-1300. Available from http://chestjournal.chestpubs.org/content/121/4/1290.full#cited-by 9. Schuman . The Nuremberg Code (1947) In: Mitscherlich A, Mielke F. Doctors of infamy: the story of the Nazi medical crimes. New York:, 1949: xxiii-xxv. Available from: http://www.cirp.org/library/ethics/nuremberg/ 10. World Medical Organization. Declaration of Helsinki. British Medical Journal (7 December) 1996;313(7070):1448-1449. Available from: http://www.cirp.org/library/ethics/helsinki/ 11. Fries JF, Krishnan E. Equipoise, design bias, and randomized controlled trials: the elusive ethics of new drug development. Arthritis Res Ther. 2004;6(3). Available from: http://dx.doi.org/10.1186/ar1170. 12. Stang A, Hense H, Jckel K, Turner E, and Tramr M. Is It Always Unethical to Use a Placebo in a Clinical Trial PLoS Med. 2005 March; 2(3): e72. Published online 2005 March 29. doi: 10.1371/journal.pmed.0020072. 13. Turner, E. Matthews A, Linardatos E, Tell R, and Rosenthal R Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med. 2008 January 17; 358(3): 252-260. doi: 10.1056/NEJMsa065779. Available from http://content.nejm.org/cgi/content/full/358/3/252 14. Fisher L Advances in Clinical Trials in the Twentieth Century. Annual Review of Public Health 1999 20, 109-124 Available from http://arjournals.annualreviews.org/doi/abs/10.1146%2Fannurev.publhealth.20.1.109 15. National Institute of Health. The Belmont Report Ethical Principles and Guidelines for the protection of human subjects of research April 18,1979. Available from: http://ohsr.od.nih.gov/guidelines/belmont.html 16. Goodman SN. Stopping at nothing Some dilemmas of data monitoring in clinical trials. Ann Intern Med. 2007 June;146(12):882-887. Available from: http://view.ncbi.nlm.nih.gov/pubmed/17577008. 17. Schulz K, Altman D, Moher D, Group C. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. Trials. 2010 March;11:32+. Available from: http://dx.doi.org/10.1186/1745-6215-11-32. 18. Noordzij M, Tripepi G, Dekker FW, Zoccali C, Tanck MW, Jager KJ. Sample size calculations: basic principles and common pitfalls. Nephrology, dialysis, transplantation. 2010 January;Available from: http://dx.doi.org/10.1093/ndt/gfp732. 19. Hahn P. Sample Size Estimation. Queens University. 2009. Available from: http://meds.queensu.ca/medicine/obgyn/apog/sample.size_p.pdf 20. Chan YH. Randomised controlled trials (RCTs)-sample size: the magic number Singapore medical journal. 2003 April;44(4):172-174. Available from: http://view.ncbi.nlm.nih.gov/pubmed/12952027. 21. Altman DG. Practical Statistics for Medical Research (Statistics texts). 1st ed. Chapman & Hall/CRC; 1990. Available from: http://books.google.co.in/booksid=v-walRnRxWQC&dq=Altman+DG.+Practical+Statistics+for+Medical+Research+%28Statistics+texts%29.+1st+ed.+Chapman+%26+Hall/CRC%3B+1990.+Availab&printsec=frontcover&source=bn&hl=en&ei=sYHYS6_EOMyyrAfpu4SvBw&sa=X&oi=book_result&ct=result&resnum=4&ved=0CBUQ6AEwAw#v=onepage&q&f=false Read More