What is a Clinical Study?

A clinical study is scientific research involving people that is conducted to assess the efficacy and safety of a new drug or to expand the indications of an already known drug. Worldwide, clinical studies are an integral stage of drug development preceding registration and broad medical use. During a clinical study a new drug is studied for efficacy and safety data; on that basis the competent health authority decides on registration or refusal. A drug that has not undergone clinical studies cannot be registered or brought to market.

Extrapolating animal and biological-model results to humans is only possible in general terms — sometimes not possible at all. Human pharmacokinetics differ even from primates. But preclinical analysis is essential for assessing the likelihood and character of adverse effects and for calculating the starting human dose.

Clinical studies can begin only after encouraging preclinical results (biological models and lab animals), ethics-committee approval and a positive decision from the competent health authority. Initially the drug is studied in a small number of patients/volunteers. As safety and efficacy data accumulate, the subject pool grows and the drug is compared with drugs already used in practice.

Clinical studies as a drug-development stage

According to the US Pharmaceutical Research and Manufacturers of America (PhRMA), of 10,000 drug candidates US pharma companies take into development, only 250 reach preclinical studies. Only 5 make it to clinical studies. Only one becomes a drug and enters broad medical practice.

In the US only 11% of metabolic drugs, 14% of CNS drugs, 15% of cardiology drugs, 20% of respiratory drugs, 27% of oncology drugs and 40% of antibiotics that enter clinical trials receive marketing authorization. In 2009 the US FDA approved 34 innovative drugs; in 2008 — 31; in 2007 — 23 and so on.

From 1979 to 2005 the cost of developing a drug grew from $100M to $1.3B.

Today developing an innovative drug takes on average 10–12 years and $0.8–1.2B.

Over the last decade, trial procedures and designs have grown significantly more complex. From 1999 to 2005 the number of procedures — tests, examinations, etc. — per clinical study rose from 96 to 158 (+65%). Meanwhile the enrollment rate (patients meeting the tightening inclusion criteria) dropped from 75% to 59% (−21%), and completion rates from 69% to 48% (−30%). Average study length rose from 460 to 680 days (+70%).

In 2009 worldwide, 17,057 clinical studies were initiated.

Overview

Before a study the sponsor defines what will be studied. Typically the question is medical — e.g. “does drug A help patients with ischemic heart disease?” Results from a limited sample are extrapolated via statistics to the whole population. Biomedical statisticians are always involved in study design — they develop data-collection and analysis methods ensuring representative results. Before a full-scale study, exploratory studies are usually conducted. Decisions on comparator (one or several drugs or placebo) and patient groups are made with recommendations from subject-matter and clinical-trial experts.

Clinical studies can be single-center (one country) or multi-center (many countries).

Investigators enroll patients per inclusion criteria and collect health data during participation (lab results, drug-concentration data, health status changes, etc.). Data are sent to a data-processing center for analysis and statistical summary. Data analysis is the final step — answering the study questions, confirming or rejecting hypotheses and sometimes generating new ones.

Example study goals:

assess safety and efficacy of a new drug in patients with a specific disease (e.g. Alzheimer’s);
assess safety and efficacy of different doses of an already widely used drug (e.g. 10 mg vs 5 mg);
assess safety and efficacy of an already widely used drug in a new indication;
whether a new drug is more effective than the currently used drug (“gold standard” comparison);
comparing two already used drugs for a disease (therapy A vs therapy B).

Most studies compare two drugs; some compare three or more.

The goal, objectives, design, methodology, statistics and organization are described in the clinical-study protocol — a kind of instruction for investigators. They must strictly follow it — this guarantees that the study is conducted identically at every site. Multi-center studies always follow a single protocol. Non-compliance can lead to exclusion of the investigator/site.

Clinical studies are funded by the sponsor — a pharma company, research institute, government body, etc. Sponsor-funded studies are often outsourced to CROs.

History

For a long time clinical-trial conduct, including ethics, was not legally regulated. Even in the mid-20th century, many unethical studies — rightly called “inhumane experiments on humans” — took place, including experiments in Nazi death camps.

In 1932 the US Department of Health launched a study of untreated syphilis in 399 African-American men in Tuskegee, Alabama (Tuskegee syphilis study). Subjects were not told they had syphilis or that they were in a study — they were told they would be treated for “bad blood.” The study was planned in two phases: 6–9 months of natural-course observation, then treatment (Salvarsan, mercurial ointment — not very effective but the only options at the time). When the Rosenwald Foundation withdrew funding, the researchers decided to run a prospective untreated study. In 1947 penicillin’s efficacy against syphilis was established and doctors nationwide treated it successfully — yet the Tuskegee study continued unchanged. Letters pressured subjects into painful spinal taps (“last chance for special free treatment”); families were offered funeral cost coverage in exchange for autopsy consent. When national free-treatment programs appeared, researchers prevented subjects from joining. The study ran until 1972 — halted only after a media leak. Many subjects had died of syphilis or its complications.

Thus until the mid-20th century rights of trial subjects were not protected. The voluntary-participation norm was first enshrined in the Nuremberg Code (1947) and then the WMA Declaration of Helsinki (1964), which is the basis of all subsequent guidelines on human-subject research.

There was no pre-market regulatory barrier for drugs until 1938. The first Pure Food and Drug Act (1906) defined adulterated and misbranded “patent” drugs and prohibited their manufacture, sale and transport, but set no efficacy or safety requirements.

Sulfonamides — introduced in 1935 — were the first effective anti-infectives. In 1937 M.E. Massengill released a pediatric liquid form. Sulfonamide is poorly soluble in ordinary solvents, so diethylene glycol — a toxic antifreeze component — was used. No preclinical or clinical studies were run. In October 1937 the FDA learned of 8 children and 1 adult dying after taking the drug. The company sent 1,100 telegrams; warnings went out by radio and newspaper; all 239 FDA inspectors plus local police were mobilized. Buyers were tracked by prescription addresses. The product was withdrawn quickly, but product already sold killed 107 people — most children.

In 1938, under FDA pressure following the sulfonamide deaths, the US passed the Federal Food, Drug, and Cosmetic Act, giving the FDA authority over food, cosmetic and drug safety. Manufacturers now had to conduct safety studies and submit data for FDA review before marketing. The law also expanded FDA authority over labeling — broadly interpreted in court to cover brochures, scientific paper copies, annual reports and efficacy claims. The FDA began regulating efficacy claims and pre-marketing oversight even before these duties were statutorily codified.

The thalidomide tragedy of 1959–1961 triggered comprehensive clinical-trial rules. Without adequate preclinical or clinical study, thalidomide was marketed in Europe as a sedative and anti-nausea drug for pregnant women. In the US it was not registered, but the manufacturer sent free samples to doctors. First in Germany, then in 40 other countries, cases of phocomelia (“seal-like limbs”) were registered in children whose mothers took the drug during pregnancy.

Between 1956 and 1962, more than 10,000 children worldwide were born with thalidomide-related malformations. Spontaneous phocomelia is extremely rare; most clinicians had not seen such cases before 1959. Retrospective studies linked thalidomide to malformations, but the evidence was initially unconvincing. Prospective observations in prenatal clinics confirmed that thalidomide caused the greatest tragedy in modern pharmacotherapy.

Thalidomide has been used since 1998 for certain hematological cancers (including multiple myeloma) and severe leprosy. A complex safety and training system controls physicians and patients — patients must use the most effective contraception and are barred from donating blood or sperm.

After thalidomide it became clear that drug circulation must be state-controlled and that registration should be based only on full, objective trials — not on individual experts’ subjective opinions.

In 1962 the US adopted the Kefauver-Harris Amendment to the Federal Food, Drug, and Cosmetic Act — also known as the Drug Efficacy Amendment. It required manufacturers to provide the FDA with proof of drug efficacy and safety — and accurate side-effect information — before approval. It also first introduced the requirement of informed patient consent for trial participation. Since 1962 no manufacturer can register a drug without a full dossier of preclinical and clinical data.

Because many drugs had been approved before the law only on safety grounds, the FDA launched the Drug Efficacy Study Implementation (DESI) — a retrospective efficacy review. By 1984 the FDA had evaluated 3,443 drugs; 2,225 were ruled effective, 1,051 not effective, 167 deferred.

In the 1960s many European countries adopted state-control laws for drug sales.

Emergence of clinical-trial rules. ICH GCP

Based on existing national quality requirements and the WMA Helsinki Declaration principles, clinical-trial rules developed and were codified in Europe. Eventually the need to harmonize requirements was recognized. In April 1990 a meeting of US, Japanese and European Economic Community representatives in Brussels — later called the 1st International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH-1) — tried to build common rules. Further meetings followed in Orlando (1993, ICH-2) and Yokohama (1995, ICH-3). On May 1, 1996 the international expert group finished a single document — the ICH Harmonized Tripartite Guideline for Good Clinical Practice (ICH GCP) — and called on participants to enshrine it in law.

ICH GCP took effect in the US, EU and Japan in 1997.

In 1998 Russia adopted OST 42-511-99 “Rules for conducting high-quality clinical trials in the Russian Federation,” based on ICH GCP. In 2005 a text identical to ICH GCP was adopted in Russia as national standard GOST R 52379-2005 “Good Clinical Practice.”

Compliance ensures reliable results, avoids exposing patients to unjustified risk, and protects rights and confidentiality. “Compliance with this standard provides public assurance that the rights, safety and well-being of study subjects are protected, consistent with the principles established by the WMA Helsinki Declaration, and that clinical trial data are reliable.”

Types of clinical studies

First classification — by whether the study intervenes in ordinary patient management (standard exam and treatment).

Observational study — a clinical study in which the researcher collects data by simple observation without actively intervening.

Non-interventional study (“no-intervention study”) — a study in which the drug is prescribed in the usual way under the terms of the marketing authorization. Assignment of a patient to a specific treatment strategy is not decided in advance by the protocol; it is made per existing practice and is clearly separated from the decision to enroll the patient. No diagnostic or monitoring procedures beyond standard practice are applied; epidemiological methods are used to analyze the data.

Interventional study — a study of new, unregistered drugs, immunobiologicals, medical devices — or a study in which drugs/immunobiologicals/devices are prescribed or used in a manner different from the registered label (a new indication, a new dose, new route, new administration, new patient category).

Another classification is by goal (US NIH):

Prevention trials — to find the best way to prevent diseases in people who have never had them, or prevent recurrence. Studied: drugs, vaccines, vitamins, minerals, lifestyle changes.
Screening trials — to find the best way to detect certain diseases/conditions.
Diagnostic trials — to find the best diagnostic method for a given disease/condition.
Treatment trials — to study the efficacy and safety of investigational drugs, new drug combinations, or new surgical/radiation methods.
Quality-of-life trials — to study ways to improve quality of life for patients with chronic illness.
Expanded access / compassionate use — use of an investigational drug in seriously/life-threateningly ill patients who cannot be enrolled in the clinical study because they do not meet the inclusion criteria. Typically for patients with no effective treatment available.

Clinical-trial designs

The design is the overall study plan — how the study will be conducted. Main observational designs — cohort and case-control, among others.

In a cohort study a group (cohort) is followed over time. Subgroups — those who received or did not receive (or received to different degrees) the investigational treatment — are compared. Prospective cohort studies are planned first, then data collected. Retrospective cohort studies use archives.
In a case-control study, people with a given disease are compared with people in the same population without the disease — to identify links between clinical outcomes and earlier exposure to risk factors.

Other observational designs include cross-sectional observational studies.

The reference design is the randomized controlled double-blind trial.

Randomization means random assignment to treatment groups — every patient has the same chance of getting the investigational or control drug. Treatment has effects regardless of whether it is active. The placebo effect must be considered. Today the two main control technologies are placebo control and active control. Placebo control — the control group gets placebo identical in form, color, taste and smell to the investigational drug. Active control — the investigational drug is compared with the established “gold standard” therapy.

Placebo control raises ethical issues — it may limit the patient’s right to the best available care. Placebo use is limited. The WMA Helsinki Declaration allows placebo only:

when no proven effective treatment exists;
when there are compelling, scientifically-justified methodological reasons to use placebo, and placebo/untreated patients are not exposed to serious or irreversible harm.

Psychological (“subjective”) factors matter greatly. Knowing which treatment a patient is receiving may bias safety/efficacy assessments. An investigator favoring one arm may unconsciously interpret outcomes favorably. To minimize subjective factors, blinding is used.

Single-blind — the patient does not know, the investigator does. Double-blind — neither the patient nor the investigator knows.

Blinding minimizes intentional distortion and spreads unintentional distortion equally between groups.

Clinical-study protocol

The protocol describes the study’s goal, objectives, design, methodology, statistics and organization. Every study starts with protocol development — the most important document. Regulatory agencies and ethics committees review the protocol to judge scientific adequacy and subject protection. It guides investigators, unifies work across sites, and is the basis for statistical analysis and later audits/inspections. A large study’s protocol may take years, with sponsor staff and outside consultants. Format and content for sponsor-funded studies are standardized (ICH GCP); regulators in Canada and Australia also follow ICH.

Informed consent is a process by which a patient or healthy volunteer freely confirms their wish to take part in a study. The signed document is also called informed consent. The investigator informs the patient of every aspect that could affect the decision — benefits, risks, time investment, possible side effects and so on. Hence “informed.” After explanation the patient receives a written document describing the study (duration, procedures, risks, potential benefits). The participant decides whether to sign.

A participant may withdraw at any time without giving reasons.

Study power

During planning, biomedical statisticians determine how many patients must be enrolled to obtain a statistically significant result showing the difference in efficacy. Patient number is set before the study — it drives cost. The required sample depends on the disease, the end-points, the design and so on. Showing efficacy of a new drug in otherwise untreatable metastatic renal cancer in a placebo-controlled trial needs far fewer patients than a placebo-controlled trial in well-treatable ovarian cancer. Where patients can improve without treatment, spontaneous improvement masks the effect. To isolate those who benefited from the drug, a large sample is needed.

Power — the study’s ability to detect clinically meaningful differences between the investigational product and the comparator if they really exist. The larger the sample, the higher the power.

To reliably show a small difference a larger sample is needed. But with a large sample you may statistically show differences that are no longer clinically meaningful — hence the distinction between statistical and clinical significance.

Phases of clinical studies

Preclinical studies include in vitro (test tube) and in vivo (animal) studies with different test-substance doses to gather preliminary data on pharmacology, toxicity, pharmacokinetics and metabolism. They help pharma companies decide whether to investigate a substance further. Human studies may begin if preclinical data show the drug may be useful, is sufficiently safe and does not put people at unnecessary risk.

Drug development is often described as four sequential clinical-study phases. Each phase is a separate clinical study; multiple studies within a phase may be required for registration. If the drug passes the first three phases, it receives a marketing authorization. Phase IV studies are post-marketing.

Note: this article summarizes the main points from the full Russian-language entry; for exhaustive references and citation notes see the original source.