Randomized Controlled Trials (RCT) - Park's PSM

What is an RCT?

• Oral hypoglycaemic agents
• Varicose vein stripping
• Tonsillectomy
• Hospitalization of all MI patients
• Multiphasic screening

Design of an RCT (Fig. 9 from Park)

1. Drawing up a Protocol
2. Selecting Reference and Experimental populations
3. Randomization
4. Manipulation (Intervention)
5. Follow-up
6. Assessment of outcome

Step 1: The Protocol

• Specifies aims, objectives, questions to be answered
• Criteria for selection of study and control groups
• Sample size, procedures for allocation, treatments to be applied
• Standardization of procedures and schedules
• Responsibilities of all parties involved
• Essential especially in multi-centre trials
• Must be strictly adhered to throughout the study
• Purpose: to prevent bias and reduce sources of error

Pilot studies (preliminary test runs) may be done to test feasibility, operational efficiency, and to detect flaws before the final protocol is agreed upon.

Step 2: Selecting Reference and Experimental Populations

Criteria for participants (volunteers):

1. Must give informed consent - agree after being fully informed of purpose, procedures, and possible dangers
2. Should be representative of the reference population
3. Must be eligible/qualified - e.g., must be anaemic if testing an anti-anaemia drug; must be susceptible (non-immune) if testing a vaccine

A stable population whose cooperation is assured should be chosen to minimize losses to follow-up.

Step 3: Randomization

• Eliminates selection bias
• Allows comparability - "like compared with like"
• Distributes known and unknown confounding factors equally between groups
• The investigator has no control over who goes into which group

"Randomization is the heart of a controlled trial." - Park

• Randomization ensures every individual gets an equal chance of being placed in either group
• Done only after the participant has entered the study (after qualifying and giving informed consent)
• Best done using a table of random numbers
• If variables like age/sex could affect outcome, the population is stratified into subgroups, then randomized within each subgroup

Step 4: Manipulation (Intervention)

• The experimental group receives the intervention (drug, vaccine, dietary component, habit modification, new procedure)
• This creates an independent variable (the intervention)
• The outcome (e.g., incidence of disease, survival time, recovery period) is the dependent variable

Step 5: Follow-up

• Both groups examined at defined intervals in a standard manner with equal intensity under the same circumstances
• Duration based on when a significant difference (e.g., mortality) is expected to be demonstrable
• Some losses to follow-up (attrition) are inevitable - due to death, migration, loss of interest
• Substantial attrition makes it difficult to generalise results

Step 6: Assessment of Outcome

• Positive results: Benefits - reduced incidence/severity of disease, reduced cost
• Negative results: Side-effects, complications, deaths

Sources of Bias in Assessment:

Randomization and sample size cannot guard against these biases. The solution is blinding.

Blinding

When outcome is death, blinding is not essential.

Study Designs in RCTs

1. Concurrent Parallel Study Design

• Two randomly assigned groups: one gets treatment, other does not
• Patients remain in their assigned group for the entire duration
• Standard RCT design

2. Cross-over Study Design

• Each patient serves as his own control
• Group A gets treatment, Group B gets placebo - then they switch after a washout period
• Washout period allows elimination of drug ("carry-over effects")

• All patients receive the new therapy at some point
• Requires fewer patients

• The drug cures the disease (nothing left to cross over with)
• Drug effective only during a certain stage of disease
• Disease changes radically during the study period

Types of Randomized Controlled Trials

1. Clinical Trials

• Evaluate therapeutic agents, mainly drugs
• Examples:
  • Beta-blockers in reducing cardiovascular mortality post-MI
  • Folate supplementation to prevent recurrence of neural tube defects
  • Aspirin on cardiovascular mortality
  • Beta-carotene on cancer incidence
  • Efficacy of tonsillectomy for recurrent throat infection
  • Coronary bypass surgery for prevention of MI
• Not all clinical trials can be blinded (e.g., tonsillectomy - surgical vs. non-surgical is obvious)

2. Preventive Trials

• Test primary preventive measures - vaccines and chemoprophylactic drugs
• Classic example: 1946 MRC UK trial of whooping cough vaccine
• May require application to groups rather than individuals

3. Risk Factor Trials

• Test whether elimination of a suspected risk factor reduces disease incidence
• E.g., dietary fat reduction and coronary artery disease
• More difficult - risk factors are deeply ingrained habits

4. Cessation Experiments

• Withdrawal of a harmful agent already in use
• Classic example: Retrolental fibroplasia (RLF) - RLF was caused by high oxygen concentrations given to premature infants. A controlled trial confirmed oxygen was the cause; withdrawing it led to a dramatic fall in RLF incidence (see the graph showing peak in 1952-53, then sharp drop after 1953)

5. Caesarian (Natural) Experiments / Unplanned Experiments

• Evaluating effects of naturally occurring "experiments"
• Since most diseases are fatal/disabling, deliberate human experiments to confirm aetiology are rarely possible

6. Evaluation of Health Services

• Assessing effectiveness and efficiency of health care delivery systems
• Classic Indian example: Controlled trials in chemotherapy of tuberculosis in India - demonstrated that domiciliary (home) treatment of pulmonary TB was as effective as costlier hospital/sanatorium treatment - ushered in the era of domiciliary treatment
• South-East London multiphasic screening trial - led to withholding vast resources for national multiphasic screening in UK
• Studies showing nurses and paramedical workers can perform tasks traditionally done by physicians = health services research

Non-Randomized (Non-Experimental) Trials

• Ethical reasons (e.g., smoking and lung cancer - cannot deliberately expose people)
• Some preventive measures applicable only on community-wide basis (e.g., water fluoridation)
• Disease frequency is low and natural history is long (e.g., cancer cervix) - requires thousands of people for a decade or more - cost prohibitive

Types of Non-Randomized Trials:

1. Uncontrolled trials - no comparison group (e.g., Pap smear/cervical cancer screening introduced in 1920s without RCT; evidence built from over a dozen uncontrolled studies)
2. Historical controls - compare with past data
3. Community trials - intervention applied to entire community

In non-randomized trials: no randomization → low comparability → higher chance of spurious result. Validity of causal inference is largely a matter of extra-statistical judgement.

Advantages of RCTs

1. Randomization ensures comparability between groups (controls for both known and unknown confounders)
2. Blinding reduces observer and subject bias
3. Prospective design - allows direct measurement of incidence and causal inference
4. Results can be analysed for statistical significance
5. Gold standard for evaluating new therapies and preventive measures

Limitations/Disadvantages of RCTs

1. Ethical issues - cannot withhold proven treatment or expose to harm
2. Expensive and time-consuming
3. Difficult to maintain blinding for surgical or behavioural interventions
4. Attrition (drop-outs) can compromise results
5. Results may not be generalizable if study population differs from reference population
6. Volunteer bias - those who consent may differ from those who do not
7. Cannot be done for rare diseases or conditions with long natural history

High-Yield Exam Points Summary

Incidence vs Prevalence - Park's PSM

Definitions

Incidence

"The number of new cases occurring in a defined population during a specified period of time."

Important: Incidence rate MUST include the unit of time (e.g., "per 1,000 per year"). Writing "16.7 per 1,000" alone is inadequate.

Prevalence

"All current cases (old AND new) existing at a given point in time, or over a period of time, in a given population."

Types of Prevalence

(a) Point Prevalence

• All current cases (old + new) at one point in time
• The "point" may practically span a few days or weeks (time needed to examine the population)
• When "prevalence rate" is used without qualification, it means point prevalence

(b) Period Prevalence

• All current cases (old + new) existing during a defined period (e.g., annual prevalence)
• Includes cases arising before the period but extending into it, PLUS cases arising during the period

The Classic Diagram (Fig. 2 from Park)

Comparison Table

Relationship Between Incidence and Prevalence

• P = Prevalence
• I = Incidence
• D = Mean duration of illness

• Incidence = P/D
• Duration = P/I

The Bathtub/Tap Analogy (Fig. 3 from Park)

• Incidence = the tap (rate at which new cases pour in)
• Prevalence = the water level in the tub (all existing cases)
• Recovery/Death = the drain (cases leaving the pool)

Coffee House Analogy (Park)

• Counting people inside the coffee house at 10 AM = Prevalence
• Rate at which people enter per hour = Incidence

What Affects Prevalence?

• Incidence increases
• Duration of disease increases (e.g., effective treatment that prevents death but does not cure)
• In-migration of cases
• Out-migration of healthy people

• Incidence decreases
• More rapid recovery (shorter duration)
• More rapid death (shorter duration)
• New effective cure introduced

Paradox from Park: A treatment that prevents death but does not cure may paradoxically increase prevalence even without any change in incidence (e.g., early antiretrovirals in HIV).

Conversely, if duration decreases sufficiently, prevalence can fall despite a rise in incidence.

Uses of Each Measure

Uses of Incidence Rate

1. To control disease - identifies rate of new case occurrence
2. Aetiological research - studies causation, distribution, and pathogenesis
3. Testing efficacy of preventive and therapeutic measures
4. Monitoring and evaluation of disease control activities
5. Optimal measure for formulating and testing aetiological hypotheses

Uses of Prevalence Rate

1. Estimates the magnitude of health/disease burden in the community
2. Identifies high-risk populations
3. Administrative and planning purposes - estimating hospital beds, manpower needs, rehabilitation facilities
4. Used when incidence data is unavailable (with caution - duration element must be assessed)

Limitations

High-Yield Exam Points