1) Controlled clinical trials are the most robust method to measure differences between a therapy under study and a control group. They consist of an intervention group directly compared to a control group.
2) Key aspects of controlled clinical trials include randomization of subjects to intervention or control groups, blinding of subjects and investigators, pre-specified inclusion/exclusion criteria, and institutional review board oversight of subject consent and safety.
3) Results of controlled clinical trials can be meaningfully applied to clinical practice if studies use appropriate methodologies like intention-to-treat analysis and measure adherence to assigned therapies.
1) Controlled clinical trials are the most robust method to measure differences between a therapy under study and a control group. They consist of an intervention group directly compared to a control group.
2) Key aspects of controlled clinical trials include randomization of subjects to intervention or control groups, blinding of subjects and investigators, pre-specified inclusion/exclusion criteria, and institutional review board oversight of subject consent and safety.
3) Results of controlled clinical trials can be meaningfully applied to clinical practice if studies use appropriate methodologies like intention-to-treat analysis and measure adherence to assigned therapies.
1) Controlled clinical trials are the most robust method to measure differences between a therapy under study and a control group. They consist of an intervention group directly compared to a control group.
2) Key aspects of controlled clinical trials include randomization of subjects to intervention or control groups, blinding of subjects and investigators, pre-specified inclusion/exclusion criteria, and institutional review board oversight of subject consent and safety.
3) Results of controlled clinical trials can be meaningfully applied to clinical practice if studies use appropriate methodologies like intention-to-treat analysis and measure adherence to assigned therapies.
CONTROLLED CLINICAL TRIAL EVALUATION Validity of Clinical Trials
Term Meaning Application Controlled Clinical Trial Internal Quality of the Strong design ● Premiere study design to measure and quantify Validity study design should translate differences in effect of the intervention and control into reliable ● Consists of an investigational (intervention) group results being directly compared to a control group (e.g. External Ability to apply Study results standard therapy, placebo) Validity results to meaningful to practice practitioners and ● Most robust method to measure and quantify can be used for differences in effects between a therapy under study patient care and the control group METHODS - Study Design TITLE ● Study questions dictate the study design ● Reflective of the work, unbiased, specific, and concise ● Controlled clinical trial - prospectively measures a (usually ≲ 10 words) but not too general or detailed difference in effect between two or more therapies ● Declarative sentences that tends to overemphasize ○ Groups are similar and treated identically with the conclusions are not preferred exception of the therapies under study ● RCT should be identified in the title ○ Parallel design - subjects in the study are assigned ● Should include key words that are both sensitive to one of the groups and monitored (easing the task of locating the appropriate articles) and ● Controlled clinical trial - most rigorous method of specific (excluding those not being searched for) establishing a cause-and-effect relationship between treatment and outcome ABSTRACT ○ Treatment is the cause and outcome is the effect ● Concise overview of the study or a synopsis of the ○ Effect is compared to the effect of other groups major principles of the article ○ Magnitude of the difference in effect can be ● Includes information addressing the article objective, estimated methods, results, conclusions ● Should be thorough, complete, and unbiased in METHODS - Patient Inclusion/Exclusion Criteria wording selection The results of a controlled clinical trial should be extrapolated to the patient type enrolled in the study INTRODUCTION and readers must be aware of the limitations of ● Purposes of introduction: surrogate endpoints and subgroup analysis results. 1. Discussing the study rationale 2. Discussing the study purpose/objective Surrogate endpoint - a study measurement (lab value ● After formulating the objective, formulate: or physical assessment) that serves as a substitute 1. Research hypothesis - difference in the therapy marker for an actual clinical outcome (LDL levels for under investigation and control cardiovascular events) 2. Null hypothesis - no difference between the two groups Inclusion criteria: lists subject demographics that must ● Not all clinical trials includes hypotheses be present in order for the subject to be enrolled in the trial METHODS ● Design of the study is important for the results to be Exclusion criteria: characteristics that prevent a subject valid from enrollment in the trial or necessitates withdrawal ● Includes: types of subjects enrolled, the comparative from the study therapy description, outcome measures, and statistics
PH-INFO REVIWER (TOPIC 3) | Ballada 4A-PH
● These are pertinent to the extrapolation of the study Any identified differences can be attributed to results (applying study results to practice [external intervention rather than other factors. validity]) Controlled clinical trial - the word “control” indicates ● Ex: CONDOR (Celecoxib versus Omeprazole and another therapy is serving as the measuring point for Diclofenac in patients with Osteoarthritis and the effect of the intervention to be assessed. rheumatoid arthritis) trial ○ Subjects are either: METHODS - Institutional Review Board/Subject (1) at least 60 y/o; or Consent (2) 18-59 y/o with a history of gastroduodenal Research projects that use human as study subjects ulceration or GI hemorrhage must be approved before investigators begin enrolling ○ Thus, results cannot be extrapolated to patients subjects in the trial. less than 60 years since only those less than 60 years with history of gastroduodenal ulcer were included IRB - committee charged with ensuring the subjects are protected and not exposed to unnecessary harm or ● Selection bias may be present unethical medical procedures ○ Occurs after subjects meet the inclusion and exclusion criteria, but are not enrolled in the study Informed consent form: subject is presented with the ○ Investigators prevent them from being enrolled form to notify study procedures, their rights and because they may alter the results either positively responsibilities of participating in the study,the risks, or negatively benefits, compensation, voluntary participation, and ○ Common form in the run-in phase (Lead-in phase) right to withdraw without penalty
● Run-in phase (Lead-in phase) METHODS - Blinding
○ 2-4 weeks before being officially enrolled ● A technique in which subjects and/or investigators are ○ Can identify subjects who may or may not adhere unaware of who is in the intervention or control group to the therapy regimen, experience side effects from ● To reduce the influence of bias on measuring a the therapy, or did not meet prespecified criteria difference in effect between the intervention and ○ They are excluded from participating even if they control met the original inclusion criteria ○ Produces a bias by selecting a group who do not Types of Definition completely represent the population Blinding No blinding Investigators and subjects are aware METHODS - Intervention and Control Groups (open-label) of the assignment to the intervention Intervention group: consists of the therapy under or control investigation (medication or procedure) Control group: Single Either investigators or subjects, but consist of no therapy (placebo), another therapy (active not both, are aware of the assignment control) or be compared to existing data (historical Double Both investigators and subjects are not aware of the assignment data) Triple In addition to both investigators and subjects not being aware, trial Both are to be as similar as possible in all respects personnel involved in data (average age, number of male/female, medication use, interpretation are not aware of disease states) other than the treatment received. subject assignment
PH-INFO REVIWER (TOPIC 3) | Ballada 4A-PH
METHODS - Randomization ● measuring the adherence of therapy ● An essential component of all controlled clinical trials ○ medication dosage unit counts and a significant differentiator from other study designs ○ serum drug levels ● All persons in a clinical trial who have an equal chance ○ regular follow-up communications (i.e., telephone to be in the intervention or control group conversations) ● Results are more dependable than nonrandomized ● insufficient and/or inappropriate data collection trials methods and nonadherence usually lead to biased results METHODS - Endpoints Primary endpoint - one effect caused by the METHODS - Sample Size intervention and control ● appropriate sample size is vital for the study results to Secondary endpoints - routine and useful measure; not have any significant meaning; conducting a power considered to be the primary purpose of the study analysis is important to determine a suitable sample size The primary endpoint should be appropriate for the study purpose and measured using valid techniques and Sample size (n) - refers to the number of subjects methods. randomized into a study and is of considerable importance to the validity of the study results Composite endpoints - combination of endpoint measures into one primary endpoint ● sample size should not be determined on the basis of ● Consists of clinical outcomes directly related to convenience, arbitrarily, or by the number of easily morbidity and mortality as opposed to a recruited subjects pharmacological action (reduction in any incidence of ● dependent on the expected magnitude of difference stroke/MI/CV-related death vs. lowering cholesterol in the endpoint effect between the intervention and levels control ● To measure an overall effect of therapy ○ large sample size is needed to detect a small difference in effect between the intervention and METHODS - Follow-Up Schedule/Data control outcome, while a smaller sample size is Collection/Adherence needed to detect large differences between the two Considerations: groups 1. study should be conducted for an appropriate duration ●regardless of the method selected to determine the 2. data need to be consistently collected throughout the appropriate sample size, it must be calculated prior to entire trial initiating the clinical trial
Monitoring of the trial results at predetermined
intervals is important throughout the duration of the trial.
● the protocol for discontinuing the clinical trial early is
established prior to enrolling study subjects ● prior to the start of the study, data collection methods are established ● investigators should ensure trial personnel are properly trained and have sufficient resources to complete data collection
PH-INFO REVIWER (TOPIC 3) | Ballada 4A-PH
METHODS - Statistical Analysis ○ continuous data are assessed via parametric statistics; ● means to analyze sample data and apply it to the common tests are Student’s t-test, analysis of variance population (ANOVA), and analysis of covariance (ANCOVA) ● a biostatistician is consulted as one of the trial ○ Nonparametric tests are used for nominal and ordinal investigators to perform the statistical analysis of the data; examples are chi-square (χ 2) and Mann-Whitney trial results U test ● to collect sufficient evidence to reject H0 in favor of accepting the research hypothesis (H1 ) (new Descriptive statistics terminology may refer to this as failure to accept H0 ) ○ describe the characteristics of the sample (e.g., ● appropriate tests are selected based on the type of average subject age, baseline endpoint values, number data that will be collected and analyzed of subjects with another disease present) and the results in some studies (e.g., X% had an adverse effect) Type of Definition Examples ○ presented as measures of central tendency (e.g., Data mean [average], median, mode) and/or measure of Nominal Categorical data; Yes/No; variability (e.g., range, standard deviation [SD], data placed in alive/dead; colors variance) one category, but of cars in a not more than parking lot into Non-Parametric Test one category, five categories of Chi-square test Comparison of nominal mutually either red, white, data for independent exclusive blue, black, or groups (2x2) other Fischer’s exact test Comparison of nominal Ordinal Ranking, ordered Likert scale; visual data for 2 groups when analog scale expected frequency is <5 Interval Data with Temperature in McNemar’s test Comparison of nominal measurable equal degrees data for 2 matched or distances Fahrenheit paired groups between points, Contingency table Comparison of nominal but no absolute analysis (R x C) data when there are >2 zero groups or >2 possible Ratio Data with Temperature in outcomes measurable equal degrees Kelvin, Cochran Mantel- Comparison of nominal distances blood pressure, Haenszel test data for multiple 2x2 between points cholesterol levels, tables and an absolute white blood Wilcoxon Rank Sum test Comparison of zero count Mann-Whitney U test continuous data taken from 2 independent Inferential statistics (e.g., Student’s t-test, chi-square groups test) Wilcoxon signed rank Comparison of ○ used to draw conclusions, based on the sample, for test continuous data taken the application of the trial results to the population from 2 paired groups ○ used to determine if a statistical difference is present Kruskal Wallis test Comparison of between the intervention and control groups continuous data taken ○ p-value is calculated based on trial results and from >3 independent statistical tests; afterward, the p-value is compared to groups the alpha (α)-value established prior to the beginning of Friedman’s test Comparison of the trial continuous data taken ○ selection of the statistical test depends on the data from >3 paired groups being parametric (i.e., normal distribution) versus nonparametric
PH-INFO REVIWER (TOPIC 3) | Ballada 4A-PH
Parametric Test RESULTS - Subject Demographics Student’s T-test Comparison of continuous ● describes the subjects actually enrolled and data taken from 2 randomized in the clinical trial independent groups ● usually presented in a table of demographic Paired T-test Matched/paired groups information 1 – Way ANOVA Comparison of continuous ● includes average age, gender ratio, disease states, data taken from >3 and/or drug therapy use among the study participants independent groups at the time of enrollment Repeated measures For paired / matched ● patient baseline demographic data need to be groups compared between treatment groups to ensure the 2 – Way ANOVA Similar to 1 – way ANOVA, groups are as similar as possible but can be made for >2 factors RESULTS - Subject Dropouts/Adherence Other Tests ● After the baseline subject information, data regarding Pearson Regression Determines if a linear the follow-up (i.e., subject dropout or attrition) and correlation between 2 adherence should be presented groups when data is ● DROPOUT/Loss to follow-up - subjects randomized in normally distributed a clinical trial did not complete the entire duration Spearman Regression Determines if a linear ○ lack of desire to continue, subject relocation correlation between 2 (e.g., moving to another city), difficulty finding groups when data is NOT transportation to clinic visits, subject protocol normally distributed violation, side effects, and death Multivariate Regression Determines relationship ● not all subjects will be compliant with the therapy of variable with single ● investigators need to report the number of subjects dependent continuous and major reasons for discontinuing the study, variable adherence rates, and the techniques of assessing the Logistic Regression Determines relationship data of multiple variable with single dependent ● attrition rates of 60% among both groups of a clinical dichotomous variable trial, although not ideal, is much less concerning than a study where attrition rates are 10% in one group and ● Type I error (alpha error) 50% in another - H0 is falsely rejected and H1 is falsely accepted ○ Due ○ indicator of significant medication safety to chance considerations ● intention-to-treat (ITT) principle - study results are ● Type II error (beta error) analyzed using data collected from all randomized - H0 is falsely accepted and H1 is falsely rejected ○ due subjects, regardless of whether they completed the to either by chance or small sample size entire study duration ○ mimics real-life application of an intervention RESULTS into practice because, similar to real life, all ● contains primary and secondary endpoint results and subjects in a clinical trial may not complete other useful information (patient demographics, therapy as prescribed dropout information, and safety information) ● Per protocol (PP) principle - analyzing data only from subjects completing the trial per the protocol ○ determining the effects of the intervention in subjects who followed the study protocol and completed the entire course of therapy
PH-INFO REVIWER (TOPIC 3) | Ballada 4A-PH
RESULTS - Endpoints/Safety CONFIDENCE INTERVAL ● Primary endpoint results - presented clearly and ● The use of 95% CI can assist the reader in assessing completely, using transparent and unbiased methods ○ the magnitude of difference in effect between the investigators need to explain the results and present intervention and control to apply to the population. probability values (i.e., p-values) ● Secondary endpoint results - presented in a fashion No Difference Does Not Indicate Equivalency similar to the primary endpoints ● H0 is not written to state the intervention and control ● Surrogate endpoints - a measure of the efficacy of a are the same, but stated as no difference in the effect treatment can be defined as laboratory values (e.g., (i.e., endpoint measurement) between the intervention HDL-C/LDL-C), symptoms (e.g., pain), or clinical and control parameters (e.g., blood pressure) that are employed as ● Nonstatistically significant results do not equate to a substitute for a clinical endpoint (e.g., morbidity, the intervention and control being the same or equal mortality) ○ convenience (easily and readily assessable) BIBLIOGRAPHY ○ well-established relationship between the ● provides documentation to support the information surrogate and clinical outcomes (e.g., provided in the manuscript or acknowledgment for the hemoglobin A1C and risk/severity of diabetes) work of other authors ○ determination of clinical benefit as a result of ● Readers should scan the references listed in the changes in the surrogate endpoint bibliography to determine if the authors used material from reputable sources RESULTS - Subgroup Analysis ● Investigators often analyze the results of subsets of ACKNOWLEDGEMENTS the study subjects, as divided into various groups that ● list of individuals/groups contributing to the clinical often include gender, age, and presence of diseases or trial, but who do not meet the requirements for other complicating factors (i.e., diabetes versus no authorship diabetes) FUNDING DISCUSSION/CONCLUSION Controlled clinical trial investigators and authors should ● to evaluate and/or interpret the results of the clinical disclose any funding sources and potential conflicts of trial interest. ● begin with a summary of the key findings of the study ● pharmaceutical companies, government agencies ● the trial may be discussed in comparison to other (e.g., National Institutes of Health [NIH]), national trials assessing the intervention or the disease state organizations (e.g., American Heart Association), under investigation university grants (e.g., faculty development grants), and ● it can contain biased wording private donation. ● most commonly cited criticisms of clinical trials by clinicians is the lack of external validity of the trials ● study strengths and limitations should be addressed
DISCUSSION/CONCLUSION - Result Interpretation
STATISTICAL SIGNIFICANCE VS. CLINICAL DIFFERENCE
● Interpreting the p-values correctly is crucial in evaluating a controlled clinical trial; not all statistically significant p-values are clinically important. The magnitude of difference in effect between the intervention and control cannot be determined solely with the p-value.