Case Study

Improving the Applicability of Clinical Trial Simulations to Real People

  • Translation

Understanding a Complicated Problem

The results from a randomized controlled trial (RCT) should not be directly extrapolated to the general population, even within the region where the trial was performed. Patients enrolled in clinical trials for investigational new medicines are carefully selected to optimize the likelihood of good safety and favorable outcomes. The selection process, based on inclusion and exclusion criteria, creates a contrived study sample that mimics experimental controlled conditions but may not reflect real patient populations. The difference between experimental RCT and real patient populations means there is often a discrepancy in the efficacy of a drug provided in a trial when compared to clinical practice setting.

When the focus of an RCT is a drug intervention, one attempt to solve this translation problem is to use pharmacokinetic pharmacodynamic (PKPD) models to incorporate more information about the drug in the study design. However, PKPD models often suffer from a lack of large, real populations to inform optimal dosing regimens. These models rely on sufficient understanding of the relevant biology. Otherwise, the simulations may diverge from real patient experiences.

Integrating real-world evidence (RWE) with PKPD models can provide a better understanding of the drug interventions studied in an RCT. RWE is the evidence derived from the analysis of real-world data (RWD). In trials taking place in high income countries (HIC), it is common to look at both PKPD and RWE simultaneously.  The pairing of these two seemingly disparate types of data analyses buoys the interpretation of RCT results. Drug absorption models can be fuelled by more realistic inputs obtained from RWE. The combination of both PKPD and RWE provides more detail about real patients in the community.

When attempting to translate the results of clinical trials performed in HICs to patients in real situations, as well as in situations in low- and middle-income countries (LMIC), we need to consider similarities and differences in patient characteristics. While some characteristics are the same across regions, some are tempered by regional biologic and environmental differences. These intrinsic and extrinsic factors can have a marked effect on the safety and efficacy outcomes of medicines. For example, intrinsic factors could include age, gender, genetics, drug metabolism and elimination capabilities. Extrinsic factors could include environmental exposures, smoking, diet, medical access, and drug-drug interactions.

These and other factors can affect the context of a clinical trial in a new region, so it becomes important to understand the factors that create the context in the region where the trial was performed versus the new setting. As we learn more about the populations of interest and the relevant factors, we can understand the extent to which results of clinical trials are applicable for that population and ultimately make recommendations to adjust dosing or treatment protocols accordingly.  For example, based on other therapeutic research areas, especially those impacted by an individual’s genetic background, some factors will be shared and can inform both PKPD and RWE health outcomes in multiple regions.

Therefore, the translation of RCT results from HICs to LMICs can be beneficial, but requires understanding the context in HIC, identifying which trends that can be extrapolated into LMICs, and differentiating between the variables specific to LMIC as compared to HIC subjects. HICs already address the issue of translating the results from RCTs to the wider patient population using RWE and PKPD approaches. We believe LMICs should have a similar opportunity to use these approaches that may benefit LMICs by appropriately translating trial results from HICs to LMICs.

Research Questions

  • Can RCTs in new regions of interest be better informed by combining pharmacological analyses and RWE?
  • Can we improve identification of patient characteristics who are treated for a particular disease using a combination of pharmacologic and RWD?
  • Can we improve optimal dosing recommendations in LMICs by getting insight into real patient characteristics from HIC and LMICs?
  • What patient and health system characteristics from RWE are most predictive of treatment response for a particular disease or medical condition in clinical practice? Are these trends observed in other regions?

Experimental Design and Analysis Approach to Combine PBPK Models and RWE

Pharmacologic analyses, such as physiologically based pharmacokinetic models (PBPK) and RWE can be combined to better inform RCTs. The complexity of the biology and variability of the model can affect the results of population-based estimates of drug safety and efficacy outcomes. Therefore, one use of RWE is to help fill in missing information so that more context is available for fitting or interpreting or PBPK models.

RWE is the result of analysis based on RWD, which is an assembly and curation of data from global health studies and other sources, including retrospective and prospective observational studies, electronic medical records (EMRs), and individual-level patient experience data. RWD datasets are selected based on attributes of interest, such as diagnosis codes, outpatient or hospital visits, cost information, laboratory values, exposure (or lack of exposure) to selected medicines, pregnancy, age, gender, or clinical history.

In some settings, medical reporting systems are not well established, leading to a paucity of relevant data. Most of the RWD that is available is generated in HICs but there is a growing interest in developing more LMIC RWD and RWE. While LMIC-generated resources and information are being developed, we can still learn about interventions through RWD from HICs.

Initially, we can understand the clinically related and system-related characteristics of populations and subpopulations through RWE exploratory cohort-building. This lets us learn about different health contexts underlying administration of RCT drugs and identify potential sources of bias that can lead to erroneous results and conclusions. We adjust for preferential treatment administration bias using propensity scores for weighting results. The results of initial RWE exploratory analysis provide a clearer view of intrinsic and extrinsic factors. Those factors then become the inputs for different modelling platforms, such as PBPK models, depending on the nature of the data, questions of interest and outcomes.

To predict drug concentrations expected for alternative dosing regimens in a target population, we use a PBPK modelling platform. PBPK models describe the absorption, distribution, metabolism, and excretion of a drug. They can also be used to predict drug formulations and drug-drug interactions. The inputs of a PBPK model are the physicochemical properties of a drug and salient intrinsic and extrinsic factors of the target population. Intrinsic and extrinsic factors of the population can come from RWE.

The drug concentration derived from an RWE-adjusted PBPK model is, in turn, use as the input for pharmacodynamic models (PD) of drug effect in the body. These models can help with understanding potential impact of alternative dosing regimens through quantitative systems pharmacology models, which generate population-based estimates of drug safety and efficacy outcomes, as well help design future clinical trials. The results of clinical trial simulations can also help determine whether additional data is needed to fully assess the context and outcomes. PK and PD modelling can be a powerful tool to better understand efficacy and safety outcomes. These methods can bridge the results from prior clinical trials to new target populations of interest.

Example: RWE of route-dependent drug-drug interactions

One example of using RWE to provide new information for pharmacologic studies is a study of contraceptive failure due to drug-drug interactions.1 The hormonal contraceptive failure rate varies by administration route. This study wanted to determine if a specific type of drug-drug interaction was associated with higher risk of unintended pregnancy based on the route of administration.

The data for this study was drawn from case reports of adverse events from the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS) from 1971 through 2020. The study found that the route of administration of hormonal contraceptives had differential drug-drug interactions that resulted in unintended pregnancies. Those taking oral contraceptives and using implants had a higher risk of drug-drug interactions that led to unintended pregnancies, but not those using intrauterine or vaginal ring devices. In the future, the RWE generated from this study can be incorporated in pharmacologic models to better capture the risks of drug-drug interactions and unintended pregnancy based on the hormonal contraceptive route of administration.

Actionable Outcomes

When we consider the complexity of the factors that contribute to the context of the use of a drug in a particular population, we can more potentially bring new drugs to patients who can benefit. Integrated information from PK and RWE analyses provides insight into the safety and efficacy of medicines under actual clinical conditions of use, which may be different from the outcomes of controlled clinical trials.

RWD and RWE provide a rich source of information about the characteristics and outcomes of targeted patients receiving treatments for diseases or conditions of interest. Proper data analysis and interpretation, along with published results can lead to more complete clinical trial simulations that help in the design of future trials. The results of clinical trial simulations can also help determine whether additional data is needed to fully assess the context and outcomes. PK and PD modelling can be a powerful tool to better understand efficacy and safety outcomes. These methods can bridge the results from prior clinical trials to new target populations of interest.

The combination of RWE, published results, and pharmacokinetic and pharmacodynamic modelling might provide a scientific basis for regulatory reliance on HIC regulatory review by LMICs. One possibility is the ability to streamline clinical trials in LMIC based on information acquired from HIC. Another possibility is to check and adapt interventions to better meet local needs. Alternatively, this analysis could indicate that additional clinical trials are needed prior to market authorization in the target populations. Comparisons of drug therapy outcomes in HIC patients versus LMIC may also provide signals of critical differences and mechanistic insights in the causal relationships behind the differences in outcomes.


  1. Sunaga Tomiko, Brian Cicali, Stephen Schmidt, and Joshua Brown. Contraception. 2021;103(4),222-224. doi:10.1016/j.contraception.2020.12.002

Datasets Utilized

Real-world data (RWD) from U.S. Food and Drug Administration Adverse Event Reporting System (FAERS)


Randomized controlled trials (RCTs), clinical trials, pharmacokinetic pharmacodynamic (PKPD) models, physiological based pharmacokinetic (PBPK) models, real-world data (RWD), real-world evidence (RWE)