How to Minimize Subjectivity in Dermatology Data

Dec 18, 2017

A dermatology clinical trial’s success is determined after the patient trials are over and the data is analyzed. But minimizing subjectivity in dermatology data is difficult. Physicians must assess visual symptoms, and variations in measurements threaten consistency. Even if a sponsor follows with strict definition the best practices of randomized, double-blinded trials, variations between physicians can be significant.

Since a trial’s success relies on the data, the statistics must be unbiased. Measuring treatment response as a score calculated on an established scale, such as the Psoriasis Area and Severity Index (PASI) or the Eczema Area and Severity Index (EASI) can provide some standardization to results, but neither index is a truly objective measure. For example, the PASI method includes a parameter of erythema, or redness, making it susceptible to bias.

Until more sophisticated measures are created, a combination of investigator training and new imaging methods is the best way to create accurate, reproducible data across sites.

Importance of Training

Thorough and proper training of investigators on the measures used for a study is a necessity. In-person meetings with a medical monitor give investigators opportunities to discuss the indication and receive feedback on how to assess them consistently. The medical monitor may share photos of patients and facilitate a discussion among all the meeting attendees to generate consensus. Training that incorporates this pattern of role playing, testing, and interactive feedback helps facilitate a baseline of repeatable data.

Web-based training—both webinars and dedicated online training sites—further allows investigators to become proficient in the examination and labeling of the study criteria, with additional support provided to the investigator on a one-on-one basis. Besides redness, other common subjective parameters include scaling and lesion thickness. Monitors can use this web information to compare how other investigators score the same images for such parameters.

Monitoring oversight and training are also important to creating objectivity. A CRA should watch for divergence between an Investigators Global Assessment (IGA) and condition-specific evaluations (i.e., PASI, EASI, or Scoring Atopic Dermatitis index [SCORAD]). For example, if an IGA score indicates significant clearance or overall improvement, but the PASI score does not, an investigator might not be measuring or recording data correctly, or there could be a difference in perception between the investigator and a staff member measuring the PASI score. It is crucial for a CRA to raise any such concerns immediately with the investigator to help ensure consistent data throughout the remainder of the trial.

Current State of Assessment

Subjectivity in clinical trials can blur patient outcomes. If investigators are not generating reproducible data, they could, in theory, lose evidence for a drug that works. Understandably, studying a drug in a reproducible manner is consistently one of pharma’s biggest hurdles.

In a recent webinar on biologics in dermatology, Dr. James Milbauer—dermatologist, medical monitor, and dermatology consultant for Novella Clinical—described how, in the not-too-distant future, imaging technology may potentially enable trial physicians to scan a subject for an objective, verifiable body surface area measurement, another parameter of PASI.

Currently, investigators examine subjects, and assuming the palm of the hand equals one percent of the body surface area, then estimate the affected body surface area, and try to pick representative lesions. There is tremendous inter-observer variation in these estimates; to a certain degree, there are even intra-observer variations. If, for example, an investigator sees the same subject in the morning and again in the afternoon, the estimation may change somewhat.

For these reasons, Dr. Milbauer suspects adoption of imaging in dermatological indications as a means of measurement is likely to rapidly increase. Photography is a common tool to assess dermatological conditions, but very few studies use computer assistance to objectively understand the degree of improvement over time. Such an advance in imaging could potentially be used for drug marketing as well, demonstrating unquestionable improvement before and after treatment.

Toward More Objective Assessments

There have already been advancements in turning previously subjective parameters objective. These parameters are starting to be incorporated into clinical trials. Addressing the aforementioned erythema parameter, for example, an article in the Journal of Medical Engineering & Technology stated that by using a chromameter to measure lightness, hue, and intensity, the “PASI erythema score for different skin types, i.e. low (fair skin) to highly pigmented (dark skin) skin types, can be determined objectively and consistent with dermatology scoring.”

Furthermore, in a study published in JAMA Dermatology, researchers introduced what they called the Objective Severity Assessment of Atopic Dermatitis (OSAAD), a score derived from electronically recorded measurements of skin hydration and trans-epidermal water loss, and computer-assisted body surface area estimation. The results of this study showed greater objectivity than the conventional SCORAD method.

These evolving methods hold promise for a truly objective future for dermatology clinical trials. If PASI becomes based on an area function accurately measured by imaging software instead of estimated, and if the redness, thickness, and scale were objectively measured as well, the PASI score will be a far more credible tool than it is now. Physicians will have more confidence in the meaningful and reproducible nature of the measure, and data differences from PASI scores, or other similar measures, from one study to another will be persuasive and informative for regulators and investigators alike.


Darcee Duke Strube, Senior Vice President, Dermatology Division, Novella Clinical

lorem ipsum