Randomized controlled study

from Wikipedia, the free encyclopedia
Classification of clinical studies
 
 
 Intervention study
 
 
 
 
 
 Observational study
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 comparative
groups
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Randomized
controlled study
 
 non-randomized
controlled
study
 
 Descriptive
study
 
 Analytical
study
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Cohort
study
 
 Case-control
study
 
 Cross-sectional
study

The randomized controlled trial ( RCT English: r andomized c ontrolled t rial) is in the medical research that demonstrated best study design in order to obtain a clear statement in a clear question and to prove the causality. This is why the “gold standard” of study planning and research design is also spoken of. In addition, RCTs are also used in psychological and economic research.

The evidence-based medicine emphasizes the importance of RCTs as the best basis for empirical evidence of the effectiveness of treatments to meet patient-centered decisions.

RCTs are a type of experiment (“a methodical study for the empirical acquisition of information ( data )”).

Definitions

Randomization

Randomization means that the assignment to a treatment group (such as drug A or B) is made at random. The purpose of randomization is

  1. to exclude the influence of the examiner (bias) on the assignment of a treatment and thus on the study results and
  2. ensure the even distribution of known and unknown influencing factors across all groups. For this, the number of people to be examined must be sufficiently large.

The form and implementation of the randomization must be specified in the study.

In some cases, for ethical reasons (e.g. unacceptability of placebo treatment in the case of severe illness), randomization must be dispensed with.

Controlled study

The study is called controlled because the results in the study group are compared with those of the control group (the group with a reference value ) without intervention or a control intervention. The control intervention is either the most effective measure so far or a sham intervention (for drugs: placebo ).

If the data from previous cases are used as the control group, e.g. B. the control group of a previous study, one speaks of a historically controlled study . However, they are not assigned a high level of evidence.

The study group is also called the test , intervention or verum group (Latin verum , "the real thing"; in contrast to placebo, the real form of treatment, e.g. the drug containing active ingredients). The control group is also known as the comparison or placebo group (if placebos are used).

The so-called cross-over test plan represents an extension . Here, the intervention and control group are switched to mid-study.

example

The Indian state set up a vaccination system in a poor Indian region, which had hardly been vaccinated before. There was a vaccination nurse in every village. Nevertheless, only a few percent of the parents had their child vaccinated five times (the vaccination protection was only complete after the fifth vaccination). Using an RCT, researchers examined the effectiveness of measures against this vaccine fatigue. At the same time, they tested the hypotheses why the poor did not have their babies vaccinated:

  • Maybe they didn't understand the benefit. Children vaccinated against measles can continue to develop malaria or diarrhea .
  • Perhaps parents thought the vaccination was ineffective.
  • Maybe a superstition was to blame: children could catch the evil eye (that's why babies are rarely taken outside until their first birthday).
  • Many helpers believed that any vaccination program was powerless against such beliefs.

Using a random generator, the researchers divided 134 villages into three groups:

  • In 74 villages everything stayed the same.
  • In 30 villages, an infirmary was set up once a month, in which a nurse vaccinated the children free of charge.
  • The 30 remaining villages not only had the monthly vaccination camp, but parents also received two pounds of lenses when they brought their child to be vaccinated. At the end of the fifth vaccination, the parents were also given two plates.

After a year and a half, the researchers took stock: In the villages with lenses and plates as attractants, 38 percent of all children had received five vaccinations. In the villages with regular vaccination appointments, 17 percent of all children were vaccinated, in the 74 other villages it was only 6 percent.

This proved that the superstition (if it existed among parents) was obviously not strong enough to withstand the lure of a few pounds of lentils.

So RCTs can

  • examine how efficient measures are or
  • Compare the efficiency of several alternative measures.

Example: an institution is faced with the choice of either subsidizing basic foodstuffs with money (a measure from which all buyers benefit) or providing food and vitamins free of charge to pregnant women, young mothers and children. RCTs show that the latter is the much more efficient measure: the benefit per euro spent is significantly higher.

development

For centuries, peer group clinical trials have been conducted. As early as 1753 , James Lind published the results of his attempt to treat scurvy with oranges and lemons in Great Britain . In the German-speaking area, the Hungarian doctor Ignaz Semmelweis (1818–1865), who worked in Vienna, was the first author of the introduction of “systematic clinical observation” into medical research (1848). But it was not until the end of the 19th century that people became aware of the distortions (bias) that a non-random assignment to the therapy groups could bring with it. Therefore , at the beginning of the 20th century, alternate assignment trials became popular, in which the patients were alternately assigned to the therapy arms .

The randomized clinical trial was then established and standardized as a term in the 1940s by Austin Bradford Hill , a pioneer in medical statistics and epidemiology , with the support of the British Medical Research Council . He was also the statistician on the Medical Research Council Streptomycin in Tuberculosis Trials Committee . Its study of the treatment of tuberculosis with streptomycin is widely regarded as the first randomized clinical trial.

Initially, the ethical concerns about such studies were substantial, as the control group was denied the new, promising therapy. It was not until the thalidomide scandal that there was a rethink and the Americans were the first to require in the Kefauver-Harris Drug Amendment 1962 for the approval of new drugs that their efficacy should be verified in "adequate and well-controlled studies" ( adequate and well- controlled investigations ) had to be proven. In 1970 it was explicitly made clear that this meant randomized clinical studies. All other western countries followed the Food and Drug Administration and issued similar ordinances. This is how the randomized clinical study began its triumphant advance.

Subsequently, the pharmaceutical industry became the main sponsor of randomized clinical trials, ahead of government funding programs and universities, and by the 1980s they were considered the gold standard of medical science. In the hierarchical structuring of the study types in the context of evidence-based medicine , randomized clinical studies are at the top of the value order.

Also in the surgery randomized clinical trials were done early, and in 1959 the first study appeared with a control arm in which a sham operation ( sham surgery ) was used to ensure a do double. A significantly higher mortality was found in patients with angina pectoris , in whom the then estimated ligation of the internal mammary artery was performed, than in the control group. In this way, this study saved many heart patients from a nonsensical and dangerous operation, which shortly afterwards disappeared completely from the agenda.

However, there have been repeated warnings against "slavish faith in studies" and in many cases results only prevail if this is physiologically understandable. Nevertheless, there are also many randomized clinical studies, the results of which have had no or only very late influence on applied medicine. Few studies have also been massively falsified, and sometimes results only come when the study therapy is already outdated and replaced by other more innovative methods.

In addition, the demand for multicenter randomized clinical trials is now limiting scientific development. Studies require massive bureaucratic and institutional effort, and a Phase III registration study costs an average of $ 30 million at the beginning of the 21st century. Accordingly, randomized clinical studies disproportionately present Western medicine and pharmaceutical research, while other methods are often not examined in randomized clinical studies due to a lack of sponsors.

With companies that carry out clinical studies ( contract research organizations, CRO ), an industry with an annual turnover of 25 billion dollars has emerged. Randomized clinical trials, for example, have little public health effects, but a high marketing potential for new drugs in high-income countries.

disadvantage

While the randomized controlled trial (RCT) is the gold standard in the development and introduction of new drugs, this is not entirely the case in all other medical fields, and there are some disadvantages and limitations of RCTs that can be offset with other study designs :

  • High costs and long study duration : Due to the advancing regulation, the implementation of an RCT has become a time-consuming and expensive company. Many years pass from planning and approval to implementation and analysis, which can inhibit innovation. New developments may mean that the results of an RCT are already outdated by other innovations at the time of the study analysis.
  • Not an emergency instrument : when an epidemic occurs , rapid public health measures must be taken for sometimes little-known pathogens, so that an RCT cannot be carried out in advance. For example, the annual composition of the flu vaccine cannot be checked in advance in the RCT. Their effect can only be determined in prospectively collected data and then the composition can be adjusted for the following season if necessary.
  • No individual effects measurable : If therapy results depend to a large extent on individual effects, RCTs can be disrupted and their results are prone to error. This applies e.g. B. towards studies in surgery in which surgical technique and performance are a major influencing factor.
  • Rare diseases : In rare diseases , the number of patients that would be necessary for an RCT is often missing.
  • Problem of external validity : Due to the exclusion and inclusion criteria, the RCT often examines highly selected patients, and a transfer to other populations is sometimes not possible. In order to obtain sufficient "endpoints" and to limit the size of the study, it is advisable in RCTs to limit them to high-risk patients, but then the transferability to patients with medium or low risk is questionable.
  • Surrogate parameters are sometimes used to reduce the cost and duration of the study, but do not always correlate with the actual study goal. In the therapy of osteoporosis until the introduction of aminobisphosphonates and the "FIT-Trial" leading to the approval of alendronate, changes in bone density were often used as a surrogate, which led to a significant increase in some drugs, especially fluorine , but did not reduce the risk of bone fractures and today can no longer be used.
  • Problem of the short study duration : The duration of an RCT is limited, but with the drugs, which are often taken chronically, it remains to be seen whether the effect wears off later, after years, or whether other undesirable side effects accumulate over time. Vaccines in particular often lose their effectiveness ( waning ) after years , which is why refreshments are necessary for many vaccines, but this cannot be recorded in the RCT.
  • Population-wide analyzes often cannot be controlled in the RCT. Large observational studies have shown that an increased sodium intake is associated with an increased risk of high blood pressure and secondary diseases; there are also RCTs that show that sodium reduction lowers blood pressure. But population-wide public health -measures can hardly study in RCTs because they would have to run very long, very many patients would have to include, and there are numerous possibilities of distortion ( bias exist).

Comparison to observational studies

A study without randomization is considered an "observational study" because it does not interfere with the choice of treatment. In addition to simple clinical case series, this also includes case-control studies and cohort studies . The lack of randomization can lead to numerous unconscious falsifications ( English biases ) that can distort the results. However, the Cochrane Collaboration , which deals with the methodology of clinical studies and which helped develop evidence-based medicine , was able to show in a large-scale review study in 2014 that well-designed observational studies usually perform well compared to randomized controlled studies and do not show any significant difference . Fourteen methodological systematic reviews were included that compared the results of randomized controlled trials with those of observational studies with the same research question: in eleven cases there was no significant statistical difference, and the pooled and weighted odds ratio was 1.04 across all studies a 95% confidence interval of 0.89–1.21. However, this already indicates the high degree of heterogeneity and inconsistency between the studies compared, which is also evident in the heterogeneity estimate, which is very high at I 2 = 68%. Above all, however, the results differ significantly in three reviews, which thus precisely highlighted the problems in the evaluation of observational studies. Nonetheless, when evaluating studies, the special circumstances of each individual study should be carefully considered, instead of being based solely on blanket assessments of study types.

literature

Individual evidence

  1. ^ A b Deutscher Ärzteverlag GmbH, editorial office of Deutsches Ärzteblatt: Rare diseases: Randomized controlled studies are the gold standard here too. February 20, 2015, accessed February 4, 2019 .
  2. ^ Evidence-based Public Health. Retrieved February 4, 2019 .
  3. ↑ clinical trial designs. November 23, 2015, accessed on February 4, 2019 (German).
  4. ↑ Abhijit Vinayak Banerjee, Esther Duflo, Rachel Glennerster, Dhruva Kothari: Improving immunization coverage in rural India: clustered randomized controlled evaluation of immunization campaigns with and without incentives . In: British Medical Journal . tape 340 , 2010, p. c2220 ( bmj.com ).
  5. ↑ Max Rauner: What really helps. - Two economists stir up development policy. You compare different strategies in an experiment - and find out something amazing. In: time online. September 1, 2011, from: Die Zeit. No. 36/2011.
  6. ↑ Medical Research Council Streptomycin in Tuberculosis Trials Committee: Streptomycin treatment of pulmonary tuberculosis. In: British Medical Journal . Volume 2, 1948, pp. 769-783.
  7. ^ LA Cobb, GI Thomas, DH Dillard, KA Merendino, RA Bruxe: An Evaluation of internal-mammary-artery ligation by a double-blind technic. In: New England Journal of Medicine . Volume 260, 1959, pp. 1115-1118.
  8. ↑ Laura E. Rothwell, Jeremy A. Greene, Scott H. Podolsky, David S. Jones: Assessing the Gold Standard - Lessons from the History of RCTs. In: New England Journal of Medicine . Volume 374, Issue 22, June 2, 2016, pp. 2175–2181. doi: 10.1056 / NEJMms1604593 .
  9. ↑ Thomas R. Frieden: Evidence for Health Decision Making - Beyond Randomized, Controlled Trials . New England Journal of Medicine 2017, Volume 377, Issue 5, August 3, 2017, pages 465-475, [DOI: 10.1056 / NEJMra1614394].
  10. ^ A. Anglemyer, HT Horvath, L. Bero: Healthcare outcomes assessed with observational study designs compared with those assessed in randomized trials. In: The Cochrane database of systematic reviews. Number 4, April 2014, S. MR000034, doi : 10.1002 / 14651858.MR000034.pub2 , PMID 24782322 (free full text) (Review)

See also