The need for observational study data has increased significantly over the past few years, due largely to the growing demand from regulatory authorities, payers, physicians and patients for real-world information on the performance of a marketed drug product. Phase III clinical trials are the gold standard for supporting the marketing approval of a product, but what happens after approval?

How is safety and effectiveness determined once the product is introduced to patients who would not have been eligible for the phase III trial? What everybody really wants to know is is how the product is going to perform in the real world.

Observational research addresses these questions, yet one question often raised is how an observational study is different from a traditional phase III controlled clinical trial. The answer to this question lies in the fundamental reason for conducting either type of study.

A phase III clinical trial is conducted to test specific hypotheses (generally regarding safety and efficacy) about a new drug or an existing treatment being tested in a new indication to obtain regulatory approval. These studies are highly regulated and have very clear guidelines on how they are to be implemented. Additionally, they are designed to limit the effects that external influences (i.e. co-morbid conditions, concomitant medication) may have on the product being tested.

Alternatively, an observational study, or registry, is designed to collect real-world data on the natural history, management and outcomes of patients who have a particular disease or have been prescribed a specific product. These programmes are designed not to infringe upon a physician's management of a patient in both frequency of follow-up and administration of medical tests, labs and treatment.

While these differences are fairly straightforward and intuitive, it is the similarities between the two that often create the greatest confusion. Development of a protocol, recruiting physicians, analysing data and reporting results are common features of both observational and traditional clinical studies. However, the execution of each of these and other critical elements clearly distinguishes the two methodologies.

Why observational?

The one thing that holds true for observational research studies is that physicians are not mandated to prescribe or treat their patients based on a set of procedures outlined in a protocol - they treat and manage their patients as they would in everyday practice. An observational study simply provides the physician with a framewo rk for collecting and reporting data. As such, these studies can be designed to address a number of pre- or post-approval objectives, often addressing many objectives in one study. For example:

Post-approval safety surveillance:Monitor the long-term safety outcomes of a product (drug or device) in a large, multi-faceted patient cohort, either in response to aregulatory agency's mandate or as part of a larger outcomes study.

Natural history of a disease: Collect data on disease incidence/prevalence, symptomatology and disease progression, as well as physician practice and treatment patterns.

Long-term outcomes: Gather multi-year clinical and economic data (from the physician) and quality-of-life data (from the patient) to support publications, economic modelling and treatment guidelines.

Best practices: Improve quality of care by assessing evidence-based best practices based on real-world patient data.

Product acceptance: Demonstrate a product is safe and effective in a large cohort of patients over many years, as well as foster positive business relationships with prescribers.

In many cases, an observational study is the only viable approach to achieving these objectives, as the costs, as well as the onus on sites and patients, would make conducting a clinical trial unfeasible.

Implementing an observational study

The strategic purpose for undertaking an observational research study is clearly different from that of a phase III clinical trial. However, even with this understanding, if those responsible for operationalising and managing the observational study apply trial-like processes and procedures, there is a risk of jeopardising both the real-world nature of the study and the study's objectives themselves. For example, a clinical trial is statistically powered to evaluate a specific treatment effect. In contrast, an observational study is not powered overall, as there is no specific hypothesis to test. Rather, the size of the study is determined based on the desired outreach to practitioners and ensuring sufficient patient data to allow for numerous robust analyses over time.

There are a number of other key factors that differentiate an observational study from a phase III clinical trial.

Study conduct

Randomising patients to an active, experimental compound or placebo is common in a clinical trial, as is blinding the physician and patient to the treatment that is being administered. In an observational study, randomising a patient to one treatment or another would interfere with the physician's treatment decision and therefore would not reflect the real world.

Physician participants

Recruiting and qualifying sites are critical activities in both clinical trials and observational studies. For a clinical trial, the best sites are those that have a very solid research infrastructure, as well as previous clinical trial experience. In most cases, these sites have a study co-ordinator who can help co-ordinate the study's start-up and ongoing processes. These sites are also generally familiar with Good Clinical Practice. Sites participating in an observational study, however, are often either research-naïve or have only limited experience with clinical research. Perhaps more importantly, they also generally lack the infrastructure to do so - most observational research sites operate without the benefit of a designated study co-ordinator or other dedicated research staff.

Another difference is that clinical trial sites will have an in-person investigator's meeting to discuss the study and train all physicians and site staff on the study protocol and other relevant procedures. Due to the fact that observational studies recruit many more sites over a longer period of time, a one-time investigator's meeting is usually not possible. Site recruitment and training activities are ongoing; sites can proceed through the recruitment process as quickly as they are able.

Institutional review board approval

In most cases, IRB approval is sought to ensure the scientific integrity of the study, regardless of which type of study is undertaken. One marked difference is that most observational study sites do not have a local IRB and will therefore be able to utilise a central IRB approval. Central IRBs are used in clinical trials, but often the centres have their own local IRB to which the study materials must be submitted. Due to the variability of local IRBs, the time to study approval and enrolling the first patient at these sites will most certainly be longer (and impact the overall study timeline).

Another key consideration is that many trials and observational studies are global in scope, recruiting sites from all around the world. While the data collected are extremely valuable, again the study approval process (via ethics committees) can be strikingly different. This is further complicated by the fact that multiple levels of ethics approval may be necessary, both at the national and local site level. In Europe, for example, each member state was required to enact the Clinical Trials Directive (2001/20/EC), but the interpretation and implementation of the directive is up to the individual member state. The process for ethics committee approval will differ based not only on whether it is a traditional clinical trial or observational study, but also on the rules established by each member state.

Patient population

The patient population in a clinical trial is generally small in proportion to the prevalence of the disease being studied. Observational studies, on the other hand, recruit a large number of patients to ensure adequate data for the broad, population-based statistics that will be applied.

Inclusion/exclusion criteria

In order to create a patient sample that is as homogeneous as possible, strict inclusion and exclusion criteria are established for patients participating in a clinical trial. This is absolutely necessary to demonstrate that the product alone is responsible for the outcomes noted in the study.

In an observational study, however, there are very few inclusion and exclusion criteria for entry into the study, to assure that the patient population reflects the true treated population in real-world practice. Patients in the real world are complex - they often have co-morbidities and take many different drugs. To limit their participation would undermine the fundamental objective of observational research - to generate information describing real patients, on real therapies, in real medical practice. Hence, the primary inclusion criteria in many studies will be only that a patient has the disease or is taking a particular product.

Study materials

Materials developed in clinical trials and observational studies are similar in purpose but very different in content. The clinical trial site will be more accustomed to dealing with complex contracts and protocols of 50-100 pages, as this level of detail is necessary to clearly describe the study requirements and processes. When working with sites that are research-naïve, documents of this magnitude not only would be excessive, but would likely intimidate sites and impact the ability to meet site-enrollment projections.

Accordingly, all of the materials created for an observational study - confidentiality agreement, physician agreement, informed consent, protocol and case report forms - must be simple and easy to understand, while still providing the necessary guidance and protections for which they are intended. This is indeed a challenge in many studies, and one with which many study sponsors struggle. Successful observational study design, including materials development, requires a level of flexibility, resourcefulness and even creativity that is difficult to achieve. Retrofitting an observational study into existing - albeit tried-and-true - clinical trial processes is not the answer, and will, in the end, limit the study's potential and outcomes.

Data requirements

Regardless of methodology, data are the very foundation of any study. But again, there are very important differences in data requirements based on the type of study. In a clinical trial, sites are required to complete an extensive, complex case book for each patient. This usually includes a medical history, concomitant medications, and/or laboratory and test results. These are completed based on a pre-determined visit schedule that requires the patient to return to the physician's office. Conversely, there is a nominal set of case report forms completed for each patient in an observational study. The forms are simple and contain targeted information related to the overall objectives of the programmes. This approach lessens the burden on the sites and their staff.

In a similar way, the query-generation and resolution process differs. In a clinical study, there is an exhaustive process to collect all missing data and to ensure they are as clean as possible. Additionally, on-site monitoring is required to verify that the data reported on case report forms are documented in a patient's chart. These visits can occur many times throughout the course of the trial and require considerable resources on the part of the site. In an observational study, queries are generally limited to key variables, edit checks are kept to a minimum, and there is limited, if any, on-site monitoring. Almost all management of the site and resolution of data issues happen remotely via telephone. If monitoring is deemed necessary for an observational study, in most cases it will be limited to reviewing a small percentage of sites and patients.

Pharmacovigilance

Clinical studies incorporate a mechanism for reporting all serious and non-serious adverse events. Full safety narratives are necessary for each serious adverse event, and a Data Safety Monitoring Board will review safety data throughout the study. Safety data are also monitored very explicitly through source-data verification, both to assure that all events recorded in the patients' medical records are reported through the study, and to identify the event's treatments and outcomes.

Safety reporting in an observational study, though, will in many cases mirror what is done in normal medical practice. This process can range in complexity from simply requiring sites to report events through the MedWatch system in the US, to much more stringent reporting requirements in some other programmes. The decision around how detailed a process to include is driven by the study's objectives; if the study is an FDA/EMEA-mandated safety-surveillance study, the requirements will be understandably more rigorous. In other - even most - observational studies, a process that assures both appropriate reporting, and ease of use, is employed for both serious and non-serious adverse events.

Why is this important?

The changing regulatory environment, pressure from payers, even questions from patients themselves, are all influencing the demand for observational data in today's market. The purpose of this article is to show that there are important variables to consider when designing and implementing an observational study. Although the application of these studies is indeed growing, with more and more registries, large-simple trials and other observational studies being implemented, there are still a lot of questions regarding what is and is not appropriate.

While a phase III trial is the gold standard for supporting a product up to and through regulatory approval, observational studies are necessary to examine what occurs in the real world. It is only with a better understanding of the operational challenges associated with observational studies that research will be equipped to successfully manage them.

Box: Common differences at a glance

Phase III clinical trial

* Tests specific hypotheses about a new drug or existing treatment being tested in a new indication to obtain regulatory approval.

* Design limits the effects of external influences on the product being tested.

* Physicians treat patients based on a set of procedures outlined in the protocol.

* Statistically powered to evaluate a specific treatment effect.

* Patients randomised to an active, experimental compound or placebo.

Observational study

* Collects data on the natural history, management and outcomes of patients who have a particular disease or have been prescribed a specific product.

* Design reflects real-world environment.

* Physicians treat and manage patients as they would in everyday practice.

* Patients not randomised.

* Sites are research-naïve and do not have study co-ordinator or other dedicated research staff.

Matthew Gordon is registry consulting director at ICON. For more information visit www.iconclinical.com