Real-world evidence (RWE) can help neuroscience sponsors account for disease heterogeneity, identify development risks earlier, and build more precise evidence strategies from Day 1.

For decades, Parkinson’s disease has often been approached as a single clinical entity. Sponsors identify a population with Parkinson’s, evaluate an intervention, and compare average outcomes across treatment groups.

But Parkinson’s does not have an average course.

Patients can differ markedly in their motor and non-motor symptoms, rates of progression, underlying biology, and responses to treatment. Alpha-synuclein pathology, genetic variation, mitochondrial dysfunction, neuroinflammation, and interactions among these pathways may also contribute differently across patients.

“Parkinson’s is a diverse disease with no ‘average’ course,” says Michael Markowitz, MD, P95 Julius Clinical Expert Medical Scientist. “Progression can vary over decades, and treatments may help a specific subgroup more or produce a greater response in certain symptoms.”

This heterogeneity is more than an academic concern. If Parkinson’s represents a collection of biologically and clinically distinct subtypes, trials designed around an average patient may dilute meaningful treatment effects, enroll populations unlikely to respond, or rely on endpoints that do not adequately reflect how patients experience the disease.

The answer is not to move away from rigorous, controlled clinical trials. It is to design those trials with a more accurate understanding of the patients, biological pathways, disease trajectories, and clinical settings they are intended to represent. That requires real-world data (RWD) before protocols are finalized, not only RWE after pivotal trials have ended.

Average Outcomes Can Conceal The Most Valuable Signal

A conventional trial endpoint may show a modest average treatment effect across a study population. That average, however, can mask several very different outcomes.

One subgroup may experience a clinically meaningful benefit. Another may remain stable. A third may receive little or no benefit. If those groups differ according to phenotype, biomarker status, genetic profile, baseline disease burden, or progression pattern, a population-level average can obscure both scientific insight and potential product differentiation.

The challenge is especially pronounced in Parkinson’s because progression is neither uniform nor necessarily linear. Tremor, rigidity, bradykinesia, postural instability, gait impairment, and freezing may follow different trajectories. Non-motor symptoms can be equally consequential, yet may not be fully represented by endpoints centered primarily on periodic, clinic-based motor assessments.

Emerging biomarker research further illustrates this heterogeneity. A 2023 Parkinson’s Progression Markers Initiative study evaluated an alpha-synuclein seed amplification assay using cerebrospinal fluid samples. The assay detected alpha-synuclein pathology in most participants with Parkinson’s, but results differed across clinical and genetic subgroups.

Approximately 12% of the 545 participants with a clinical Parkinson’s diagnosis were assay-negative. Negative results were more frequent among participants with normosmic Parkinson’s and LRRK2-associated disease. These findings demonstrate the limitations of defining Parkinson’s through clinical presentation alone and reinforce the possibility that different biological processes may underlie similar diagnoses.

Biomarkers such as alpha-synuclein seed amplification assays could become important components of biologically informed disease classification. But biomarkers should not be viewed in isolation.

“Biomarker data can help bridge the gap between tightly controlled clinical trials and the diversity of the real-world patient population,” Markowitz says. “EHR data and longitudinal registries can help map disease progression, while wearables may provide more continuous information about changes in motor symptoms.”

The objective is not to divide every study population into increasingly narrow groups. It is to identify which sources of variation are relevant to a therapy’s mechanism of action, expected benefit-risk profile, target population, and evidence needs.

Real-World Data Can Reveal The Patient Population Before The Protocol Defines It

Sponsors have traditionally used RWE to address post-approval questions, assess long-term safety, or demonstrate economic value. Those applications remain important, but they represent only part of RWE’s potential. The more strategic opportunity is to use RWD before a trial begins.

Electronic health records, medical and pharmacy claims, disease registries, laboratory results, imaging, genomics, patient-reported outcomes, and digital health technologies can collectively help sponsors understand:

  • How many potentially eligible patients may exist
  • Where those patients receive care
  • Which comorbidities and concomitant medications are common
  • How symptoms and functional outcomes evolve over time
  • Which eligibility criteria may unnecessarily exclude patients
  • Whether proposed visit and testing schedules reflect clinical practice
  • Which endpoints are measurable, meaningful, and sufficiently reliable
  • Which subgroups may have distinct progression or treatment-response patterns
  • Which outcomes may matter most to patients, clinicians, regulators, and payers

The U.S. Food and Drug Administration defines real-world data as routinely collected information relating to patient health status or the delivery of healthcare. Relevant sources include EHRs, claims, registries, and digital health technologies. The agency has emphasized the potential for fit-for-purpose RWD to support therapeutic product development and regulatory decision-making across the product lifecycle.

That lifecycle perspective matters. If sponsors begin RWD analysis only when they need an external control, post-authorization study, or payer-facing value dossier, the analysis cannot correct assumptions embedded years earlier in the development program.

“Using RWE early can help optimize trial design, refine patient selection, and align development with patterns observed in clinical practice,” Markowitz says. “It may also support better recruitment planning, reduce the risk of avoidable protocol changes, and generate evidence that is more relevant to regulators, patients, and payers.”

Five Development Risks Real-World Data Can Help Sponsors Address Before Phase III

1. Overestimating the available patient population

A prevalence estimate is not an enrollment forecast. After applying requirements related to disease stage, symptoms, biomarker status, prior treatment, concomitant medications, comorbidities, geography, and site access, the practically recruitable population may be considerably smaller than anticipated. Longitudinal EHR, claims, and registry data can help sponsors model the effect of proposed criteria and identify where the eligible population contracts.

This is particularly important in Parkinson’s. The number of people carrying a broad clinical diagnosis may be large, while the number who match a therapy’s biological hypothesis and operational requirements may be much smaller.

A structured feasibility assessment can help answer critical questions earlier:

  • What percentage of the observed population meets each proposed criterion?
  • Which criteria account for the greatest eligibility attrition?
  • Where are potentially eligible patients receiving care?
  • How frequently are the variables needed to determine eligibility captured?
  • Is the target subgroup sufficiently concentrated to support recruitment?

These insights do not guarantee successful enrollment. They can, however, replace assumptions with empirical estimates before the study enters its most costly stages.

2. Creating eligibility criteria that are scientifically clean but operationally unworkable

Restrictive eligibility criteria can reduce variability, but they can also slow recruitment and produce a study population that poorly reflects clinical practice. In Parkinson’s, criteria related to age, cognitive status, disease duration, medication use, comorbidities, symptom profile, and caregiver availability can substantially affect the eligible population.

The Clinical Trials Transformation Initiative recommends using fit-for-purpose RWD to assess planned eligibility criteria, understand patient and site needs, and support recruitment planning. Its 2025 recommendations describe how RWD-supported approaches may improve trial efficiency, shorten timelines, and expand patient access to research.

For Parkinson’s sponsors, an early eligibility attrition analysis can estimate how many otherwise appropriate candidates would be excluded by each requirement. Development teams can then distinguish between criteria that are essential for safety or interpretability and those that persist largely because they were used in previous trials. This can produce protocols that are both scientifically rigorous and more executable.

3. Selecting sites based on reputation rather than access to the right patients

Experienced investigators remain essential, but historical site performance does not guarantee access to the specific patients required for a new protocol. RWD can help identify geographic concentrations of patients, referral pathways, high-volume treatment centers, and institutions serving populations with relevant clinical characteristics. When biomarker testing is required, sponsors can also evaluate whether sites have the necessary diagnostic pathways, sample-collection capabilities, and specialist networks.

“Site selection can be improved by using information about where relevant patient populations are located and where high-volume centers can be found,” Markowitz says.

This does not replace qualitative site assessment. Investigator experience, infrastructure, staffing, competing studies, and local referral relationships still require careful evaluation. RWD can add a patient-centered layer to that process by showing whether a site’s addressable population aligns with the actual protocol, rather than with the disease category in general.

4. Choosing endpoints that do not adequately reflect disease behavior or patient priorities

Periodic clinical assessments provide essential standardized data, but they capture only a small portion of a patient’s day-to-day experience. Wearables, smartphones, gait sensors, and other digital health technologies may provide higher-frequency measures of tremor, gait, activity, and freezing episodes. Patient-reported outcomes and preference studies can help identify which changes patients consider meaningful. Longitudinal clinical data can clarify whether a proposed endpoint changes consistently enough during the planned study period to support reliable measurement.

These sources should not be adopted simply because they generate more data. Sponsors must determine whether a measure is valid, reliable, interpretable, operationally feasible, and appropriate for its intended context of use.

Endpoint planning should also reflect multiple stakeholder perspectives. Regulators may focus on clinical benefit and measurement validity. Patients may prioritize function, independence, cognition, mobility, or relief from burdensome non-motor symptoms. Payers and health technology assessment bodies may require evidence related to resource utilization, caregiver burden, quality of life, and longer-term value.

A 2024 multinational real-world study involving 5,299 people with Parkinson’s found that approximately 15% had advanced disease. People with advanced disease had more comorbidities and poorer quality of life, while mean annual healthcare resource utilization increased substantially with disease progression. The study found that estimated mean annual Parkinson’s-related healthcare resource utilization costs were approximately two to three times higher for people with advanced disease than for those with intermediate disease in the United States and Europe A separate analysis estimated the 2024 U.S. economic burden of Parkinson’s disease and atypical parkinsonisms at $82.2 billion.

For sponsors, these findings reinforce the need to build evidence around outcomes that matter beyond a clinical score, including function, independence, quality of life, healthcare utilization, and caregiver burden.

5. Discovering clinically important population differences after the pivotal study is complete

If a differential response is identified only through a post hoc Phase III analysis, the sponsor may not have the sample size, biomarker data, or longitudinal context needed to interpret it confidently. Early RWD analyses can identify candidate phenotypes and progression profiles that merit prospective evaluation, helping sponsors make more informed decisions about stratification and enrichment strategies, biomarker collection, sample-size assumptions, endpoint selection, prespecified subgroup and sensitivity analyses, and long-term follow-up. The FDA’s recent RWE guidance also makes clear that regulatory-grade evidence requires more than access to a large dataset. Data must be relevant and reliable for the research question, with appropriate attention to study design, standards, traceability, missingness, validation, and the quality of key variables.

More data does not automatically create better evidence. Sponsors need a clearly defined research question, a fit-for-purpose data assessment, and an evidence plan developed early enough to collect information that existing sources cannot provide.

From Real-World Insight to Clinical Execution

Identifying heterogeneity in RWD does not automatically produce a better clinical trial. The findings must be translated into a scientifically defensible protocol that sites can execute and the intended patient population can realistically join.

This is where a CRO with integrated clinical and RWE expertise can add particular value.

For example, an RWD analysis may identify a potentially meaningful Parkinson’s subgroup based on phenotype, progression pattern, genetic profile, or biomarker status. Before incorporating that subgroup into a clinical trial, development teams must determine whether it can be defined consistently, identified by participating sites, recruited at sufficient scale, and evaluated within an appropriate statistical framework.

These decisions require the combined perspectives of epidemiologists, medical experts, biostatisticians, data scientists, and clinical operations teams. They also require an understanding of how a change in one part of the protocol may affect the rest of the study.

A CRO that works across both clinical research and RWE can help sponsors:

  • Translate population analyses into eligibility and stratification decisions
  • Test scientific hypotheses against patient availability and site capabilities
  • Connect patient mapping with country and site selection
  • Evaluate the operational implications of biomarker and digital-measurement strategies
  • Align trial evidence with longer-term regulatory, medical, and payer needs
  • Reassess the evidence plan as new clinical and real-world findings emerge

The CRO does not replace sponsor decision-making. It helps ensure that development decisions are informed by evidence and tested against clinical, statistical, and operational realities.

Evidence Generation From Day 1

An evidence-generation-from-day-1 strategy begins by defining the intended patient population and the questions that different stakeholders will eventually ask about the therapy. It then connects those questions to a coordinated plan for clinical, biomarker, digital, real-world, patient-reported, and health-economic evidence.

That plan should evolve with the asset. Early RWD analyses can characterize the population and test protocol assumptions. Biomarker and data-collection roadmaps can preserve the ability to evaluate important subgroups later. Input from patients, investigators, regulators, payers, and health technology assessment bodies can help ensure that the program measures outcomes that are both scientifically credible and meaningful outside the trial.

“Starting early with RWE collection can provide long-term value to physicians, patients, and payers by supporting robust efficacy, safety, and economic evidence,” Markowitz says. “The key is aligning clinical, medical, and commercial objectives from the beginning.”

Precision Neuroscience Is Becoming An Integrated Data Challenge

Over the next five to ten years, precision neuroscience will likely depend less on any single data source and more on the ability to integrate different forms of evidence over time. EHRs can provide clinical context. Claims can reveal comorbidities, healthcare utilization, and treatment patterns. Registries can support long-term follow-up. Imaging, laboratory, genetic, and biomarker data can help characterize biological differences. Patient-reported outcomes can capture lived experience. Digital health technologies may provide more continuous measures of motor function outside the clinic.

“Wearables can add real-time information through more continuous motor measurements, including data about tremor, gait, and freezing,” Markowitz says. “When combined with clinical, imaging, genetic, and biomarker data, these measures may help us understand variability in Parkinson’s more precisely.”

With appropriate validation and governance, these technologies may eventually help clinicians and patients recognize meaningful changes between visits. In clinical development, they may also support more sensitive measurement, decentralized data collection, or a more complete view of treatment effects.

The value comes not from collecting every available variable. It comes from connecting the right data to a clearly defined scientific and clinical question.

The Commercial Value Follows The Scientific Value

Earlier insight into disease heterogeneity can help sponsors assess whether an asset has a plausible responder population, prioritize indications, evaluate feasibility, and sharpen product positioning. Outcomes that reflect patient and payer priorities can also strengthen the evidence package required for adoption and access.

Not every RWD analysis will change a protocol, and no dataset can remove the uncertainty inherent in drug development. The advantage comes from identifying consequential uncertainties early enough to act on them. For Parkinson’s therapies, where biological and clinical variability can affect both treatment response and trial execution, that timing matters.

Make RWE a Development Strategy, Not a Late-Stage Add-On

Parkinson’s does not follow one biological pathway, produce one clinical trajectory, or affect every patient in the same way. Trials designed around a single average population risk overlooking the differences that may determine whether a therapy succeeds, for whom, and on which outcomes.

RWE cannot eliminate that complexity or replace rigorous prospective trials. It can make heterogeneity visible early enough to inform patient selection, stratification, feasibility, site strategy, and endpoint design.

For neuroscience sponsors, the central question is no longer: “Where can we add RWE?”

It is: “What do we need to understand about real patients before we design the next study?”

If Parkinson’s is not one disease, its trials should not be designed around the assumption of one average patient. Evidence generation from Day 1 allows sponsors to design for heterogeneity, rather than discover its consequences after the trial is complete.

That is how precision neuroscience becomes a practical development strategy.

Designing Evidence Around the Real Parkinson’s Population

P95 Julius Clinical combines neuroscience, clinical development, and real-world evidence expertise to help sponsors translate patient-level complexity into practical development decisions.

Planning a Parkinson’s development program or refining an upcoming neuroscience study? Connect with our team to explore how earlier real-world insight can strengthen protocol design, feasibility, patient stratification, and the long-term evidence strategy.