Studi Eksperimental

01Core Concepts and Definitions

Experimental studies are the gold standard for establishing causality between an intervention and an outcome. The defining feature that separates experimental from observational studies is that the investigator assigns the exposure. In an observational study (cohort, case-control, cross-sectional), the researcher merely observes what happens without intervening. In an experimental study, the researcher actively intervenes and then observes.

The concept of clinical equipoise is foundational. An RCT is only ethically justified when genuine uncertainty exists within the medical community about whether the new intervention is better than the current standard of care. If strong evidence already shows one treatment is superior, it would be unethical to randomize patients into an inferior arm.

Experimental studies can be categorized by the unit of assignment. When individual patients are assigned, it is a clinical trial. When healthy individuals in the community are assigned, it is a field trial. When entire communities or clusters are assigned, it is a community trial. Exam questions frequently test your ability to identify the study design from a vignette and to recognize whether the unit of randomization is the individual or the group.

02Types of Experimental Studies

Randomized Controlled Trials are the most commonly tested type. The investigator enrolls patients who already have a condition, randomizes them to an intervention group or a control group, and follows them forward in time. The direction of inquiry is always prospective: intervention first, outcome later.

Field Trials differ in that they enroll healthy individuals who are at risk for developing a disease. The goal is prevention rather than treatment. Because the event rate in healthy populations is low, field trials typically require very large sample sizes, making them expensive and logistically demanding. Vaccine efficacy trials are the prototype.

Community Trials assign interventions at the level of entire communities. These are used when individual-level randomization is not feasible. The classic example is comparing towns that receive fluoridated water to towns that do not. A critical limitation is that individual-level confounders are harder to control, and the number of "units" (communities) is usually small.

03Elements of a Randomized Controlled Trial

This section covers the building blocks that exam questions test most frequently.

Randomization

Randomization is the process by which participants are assigned to groups by chance. Its primary purpose is to distribute both known and unknown confounders equally between groups. This is the single most important method for controlling confounding in experimental studies. Without randomization, the study becomes a quasi-experimental or observational study.

Allocation Concealment

Allocation concealment ensures that the person enrolling a participant does not know which group the next participant will be assigned to before enrollment. This is distinct from blinding. Even in an open-label trial, allocation concealment can be maintained. Failure of allocation concealment introduces selection bias because investigators may consciously or unconsciously direct sicker patients toward or away from the treatment arm. Methods include sequentially numbered sealed opaque envelopes, central telephone randomization, and pharmacy-controlled dispensing.

Blinding (Masking)

Blinding prevents information bias (measurement bias). If the investigator knows which group the patient belongs to, they may unconsciously assess outcomes differently. If the patient knows, they may report symptoms differently or behave differently (the Hawthorne effect).

When blinding is impossible (e.g., comparing surgery to medication), using hard endpoints (death, laboratory values) rather than subjective endpoints helps reduce the bias introduced by lack of blinding.

Control Groups

The Declaration of Helsinki states that a placebo-controlled trial is only justified when no proven intervention exists, or when compelling methodological reasons necessitate it and patients will not suffer irreversible harm.

04Clinical Trial Phases

Phase I trials are the first-in-human studies. They are primarily designed to evaluate safety, tolerability, pharmacokinetics, and pharmacodynamics. Most Phase I studies use dose-escalation designs. A key exam point: Phase I trials typically enroll healthy volunteers, not patients. The exception is oncology, where Phase I trials often enroll patients with advanced cancer because the drugs are too toxic for healthy individuals.

Phase II trials are the first time the drug is tested in patients who actually have the disease. They establish whether the drug has a therapeutic effect (proof of concept) and determine the optimal dose. Phase II is often divided into Phase IIa (dose-finding) and Phase IIb (efficacy exploration).

Phase III trials are large, multicenter, randomized controlled trials that compare the new drug to the current standard of care (or placebo). These provide the data needed for regulatory approval. The result of a Phase III trial is what a drug regulatory agency (such as BPOM in Indonesia or the FDA in the United States) uses to decide whether to grant a marketing authorization.

Phase IV trials occur after the drug has been approved and is on the market. Their purpose is to detect rare adverse effects that were too uncommon to appear in Phase III trials (which enroll thousands but not millions of patients). Phase IV is also called post-marketing surveillance. A classic exam question asks about a drug that has been on the market for years and is suddenly linked to a rare but serious side effect; this is Phase IV data.

Indonesian Regulatory Context

In Indonesia, the conduct of clinical trials is governed by Law No. 17 of 2023 on Health (the "Health Law") and BPOM Regulation No. 8 of 2024 on Procedures for Approval of Clinical Trial Implementation. These regulations mandate that all clinical trials must obtain prior approval from BPOM before they begin.

In practice, sponsors or research organizations must undergo several critical stages: obtaining ethical clearance from an accredited Health Research Ethics Committee, submitting comprehensive clinical trial documentation, and undergoing scientific and administrative evaluation by BPOM.

Indonesia has adopted Good Clinical Practice (GCP) standards and requires all clinical researchers to hold GCP certification as regulated by the Ministry of Health and BPOM. Ethics committees have been established across national and teaching hospitals. Additionally, all clinical trials conducted in Indonesia must be registered in the national clinical trial registry, and the Ministry of Health oversees review and approval of Material Transfer Agreements for the transfer of biological specimens.

Drug registration in Indonesia requires applicants to establish the efficacy and safety of non-clinical studies and clinical studies for Phases I, II, and III before a drug may be granted marketing authorization.

05RCT Design Variants

Crossover trials are heavily tested. The concept to understand is the washout period: the time between treatments during which the effect of the first drug must completely dissipate before the second drug is administered. If the washout period is too short, there is a carryover effect that contaminates the results. Crossover designs are only appropriate for chronic, stable conditions (not for conditions that resolve or progress rapidly), and the treatment must provide temporary relief rather than a cure.

Factorial designs allow the testing of two interventions in a single trial. In a $2\times 22\; \backslash times\; 2$ factorial design, subjects are randomized into four groups: Drug A + Drug B, Drug A + Placebo B, Placebo A + Drug B, and Placebo A + Placebo B. The key assumption is that there is no interaction between the two drugs. If an interaction exists, the factorial design becomes much harder to interpret.

Non-inferiority trials are increasingly common on exams. The hypothesis is reversed: instead of trying to prove the new drug is better, the investigator tries to prove it is not worse by more than a pre-defined margin ($\mathrm{\Delta }\backslash Delta$). A common trap is confusing non-inferiority with equivalence. Non-inferiority only tests one direction (not worse), while equivalence tests both directions (not worse AND not better).

06Analysis Approaches

Intention-to-treat analysis is the single most tested concept in this category. It preserves the benefits of randomization. Even if a patient in the treatment group stops taking the drug, that patient is still counted in the treatment group for analysis. This prevents bias because patients who drop out or cross over may be systematically different from those who stay.

The exam will often present a scenario where a patient randomized to treatment switches to control (or vice versa), and ask which group they should be analyzed in. The answer for ITT is always: analyze them in the group to which they were originally randomized.

For non-inferiority and equivalence trials, the per-protocol analysis is preferred because ITT analysis makes the two groups appear more similar (by diluting the treatment effect), which would artificially favor a conclusion of non-inferiority or equivalence. Using ITT in a non-inferiority trial is anti-conservative and increases the risk of a Type I error (falsely concluding non-inferiority).

07Sources of Bias and Validity

Internal vs. External Validity

Internal validity refers to the degree to which the results of the trial are attributable to the intervention and not to bias or confounding. A well-designed, well-conducted RCT with proper randomization, blinding, and allocation concealment has high internal validity.

External validity (generalizability) refers to whether the results can be applied to patients outside the study population. Strict inclusion/exclusion criteria improve internal validity but reduce external validity. Exam questions often present a well-conducted RCT and ask whether the results apply to a different patient population (e.g., elderly, pregnant, patients with comorbidities who were excluded from the trial). The answer is usually that external validity is limited.

08Ethical and Regulatory Framework

Informed Consent must include the purpose of the study, procedures involved, potential risks and benefits, alternatives to participation, and the right to withdraw at any time. In Indonesia, obtaining ethical clearance from an accredited Health Research Ethics Committee is a mandatory stage before clinical trial conduct.

The Data Safety Monitoring Board (DSMB), also called an independent Data Monitoring Committee, is an independent group that monitors patient safety and treatment efficacy during the trial. The DSMB has the authority to recommend early termination of a trial if the intervention is shown to be clearly superior (making it unethical to continue the control arm), clearly harmful, or futile (no chance of showing benefit even if the trial continues).

Stopping rules are pre-specified criteria for early termination. They are directly linked to interim analyses performed by the DSMB. A common exam question involves a scenario where a DSMB reviews interim data and finds overwhelming benefit in the treatment group; the next best step is to stop the trial early for ethical reasons.

Key Indonesian Regulations

BPOM Regulation No. 34 of 2025 was issued on December 31, 2025, to adjust provisions on clinical trial procedures and establish technical provisions for clinical trials based on each commodity, including cosmetics, natural medicines, health supplements, and processed food, as well as licensing provisions for clinical trial activities in the pharmaceutical sector.

In July 2023, the Ministry of Health signed a decree establishing the Indonesia Clinical Research Center (INA-CRC), whose function is to coordinate and facilitate clinical trials across all hospitals and serve as a catalyst for accelerating health research implementation in Indonesia. Under this framework, the government mandates that each hospital establish a Clinical Research Unit and engage in capacity-building initiatives.

09Traps and High-Yield Pearls

The most common way students lose points on experimental study questions is by confusing concepts that sound similar but have different meanings. Here are the traps that appear repeatedly:

Randomization vs. Allocation Concealment vs. Blinding. These three concepts are distinct. Randomization determines group assignment. Allocation concealment prevents the enrolling investigator from knowing the next assignment before enrollment (prevents selection bias). Blinding prevents patients, investigators, or analysts from knowing group assignment after enrollment (prevents information/measurement bias). A study can have randomization without blinding (open-label RCT), and blinding without proper allocation concealment is still vulnerable to selection bias. Exam questions will describe a flaw in one of these three and ask what type of bias results.

ITT in non-inferiority trials. Students who memorize "always use ITT" will get trapped. In a non-inferiority trial, ITT analysis is anti-conservative because it biases toward finding no difference between treatments, which is exactly what a non-inferiority trial is trying to show. The preferred analysis for non-inferiority is per-protocol.

Crossover trials and carryover effects. A vignette may describe a crossover trial where the washout period was too short. The trap is recognizing that the results are contaminated by carryover and cannot be interpreted at face value. If asked about the flaw, the answer is inadequate washout period, not "lack of randomization."

Phase I vs. Phase IV. Students often confuse the two because both focus on safety. The discriminator: Phase I tests safety in a small group of healthy volunteers before the drug is approved, while Phase IV monitors safety in the general population after the drug is on the market. If the vignette describes a newly approved drug and a newly discovered rare adverse effect, that is Phase IV.

Hawthorne effect vs. Placebo effect. The Hawthorne effect is a change in behavior because participants know they are being observed; it affects both groups. The placebo effect is an improvement in the control group because they believe they are receiving treatment. If both groups improve, the Hawthorne effect is the likely explanation. If only the control group improves, the placebo effect is more relevant.

Equipoise and when an RCT is unethical. If a question states that strong evidence already favors one treatment, conducting an RCT comparing it to an inferior treatment (or placebo) violates equipoise and is unethical. The DSMB stopping rule is the in-trial safeguard for this.

The core competency being tested is whether you can identify the study design from a clinical vignette, recognize the sources of bias that threaten its validity, understand the ethical constraints governing human experimentation, and correctly match the analytical approach (ITT vs. per-protocol) to the study design (superiority vs. non-inferiority).