We performed a narrative literature review of definitions of scratching in patients with skin inflammation and a targeted literature review of sleep in the context of the period during which such scratching occurred. Then we synthesized these findings to propose an evidence-based definition of the novel measure, nocturnal scratch. This process is illustrated in Figure 1. A formal protocol was not registered prospectively.

The extracted data were synthesized into themes based on study characteristics: scratch as a behavior, other characterization of the scratching movement, and measurement parameters for both scratch and sleep. Because of the heterogeneities in study designs, outcome measures, and reporting of outcomes, meta-analytic techniques could not be used to further summarize the studies.

The synthesis of the results was performed with 2 goals: (1) to identify the most fitting and contextually correct definitions for “scratch” and “nocturnal” (in the context of sleep) and (2) to identify the parameters of the measurement of “scratch” and “nocturnal” to inform the development of ontologies for the novel measurement.

To achieve the first goal, we reviewed all the findings from the literature search with members of a multistakeholder project team composed of experts in AD, sleep, technology development, drug development, clinical care, and the patient and care partner experience (see the Acknowledgments), and ranked the metrics based on clinical and patient relevance.

For the development of ontologies for the novel measurement of nocturnal scratch, we adopted the contextual definitions from the first step to serve as a main class under which the related concepts and properties were nested. We evaluated the properties of our selected measures of scratch and sleep and then organized them under the core concepts connected to the definitions. These core concepts represent both the specific subparts of the measure of interest and the technical features of each measurement to be reported when assessing nocturnal scratch. To increase understanding and promote adoption, we added example values (or nested properties, where applicable) that illustrate the information the specific properties add to the overall measurement.

Finally, we reviewed these findings with regulatory colleagues from the US Food and Drug Administration during a Critical Path Innovation Meeting on July 22, 2022 [53].

We decided to apply a 25-year time constraint on our literature search (ie, between 1996 and 2021) to capture the most recent and up-to-date definitions or descriptions of “scratch.” We purposefully started the search in the 1990s, the era of the proliferation and maturing of new accelerometry-based devices measuring movement and activity [54]. From 1996 to 2021, a total of 1253 unique publications with at least 1 of the search terms in their titles appeared in PubMed. Additional publications were found in the searches of IEEE (n=136), Scopus (n=71), and Web of Science (n=463). After removing duplicates and applying the inclusion/exclusion criteria to all found articles, 29 studies contained a definition of scratch, itch, or pruritus. The terms “itch” and “pruritus” were added due to the frequently interchanged use of the terms itch and scratch. A summary of these studies is presented in Table 1 [11,25-52], and their full details are included in Table S3 in Multimedia Appendix 1.

The definitions for scratching differed widely. The earliest study [26] had the following description for 3 categories of scratching: “light rubbing, touching or scratching” with any part of the body that lasted <5 seconds (light scratching), 5-30 seconds (short bouts of scratching), or >30 seconds (long bouts of scratching). The most detailed definition of scratching [25] found among all of the articles was “any repetitive rubbing or scratching of the body—including through fabric...performed with any part of the body.” In this publication, the definition of time spent scratching (scratching time) was considered to start with the first video frame showing movement consistent with this action and to end with the last contact with the scratched area or if no movement had occurred in that area for at least 3 seconds. Another common method of defining scratch for the purposes of measurement took the form of instructing study participants to perform a willful scratching movement and learning from that recorded action. Among the 29 publications, we found 7 variables commonly specified, even though most publications only specified 1 or 2 of these variables. The variables were (1) properties of the scratching movement (eg, speed, frequency, pressure, intensity), (2) the scratching tool (hand, fingers, other body part), (3) the body part or area scratched, (4) the duration of scratching bout, (5) minimum separation between bouts, (6) scratch outcome to be measured, and (7) the digital sensing product or method used to record or observe scratching (Table 2).

Compared with the term “scratch,” the concept of “nocturnal” or, more contextually accurately, “sleep” is more comprehensively researched and well defined. Leveraging the large corpus of existing literature on sleep and existing definitions, we performed a targeted review of the sleep-related search terms describing a period of time one spends with the intention to sleep or in actual sleep, either on its own or together with a specific symptom or condition happening during sleep. The search for secondary symptoms during sleep returned 154 papers (title search) in PubMed. The search for “sleep opportunity” returned 23 papers (title search) in Google Scholar and 127 (title and abstract search) papers in PubMed. Searching for the term “attempted sleep” resulted in 354 papers in Google Scholar and 0 papers in PubMed. We used these papers to find standardized definitions of the period during which sleep occurs. We used the information from the found papers to inform the ontology of sleep.

We also searched for definitions or clarifications of what the researchers meant by “nocturnal” in the papers that have detailed descriptions of “scratch.” Among the final 29 publications from the scratch search, only 2 had any description of sleep variables and their definitions for nocturnal scratching (Table S1 in Multimedia Appendix 1) [27,29]. Other publications did not specifically define the sleep period during which scratching was measured, despite using methods measuring sleep such as polysomnography.

These 2 studies used identical descriptions for sleep. They defined the sleep start time and sleep end (wake) time. These were defined as when the participants attached the scratch monitor before going to sleep until the time when the participants removed the scratch monitor when they woke up. However, a definition of the measurement period overlapping with participants’ sleep was not provided in more detail or by using standard sleep outcome measure terms, such as “sleep opportunity.” No control method for observing or measuring sleep was used in these studies. Hence, based on the findings in our targeted literature search, we developed ontologies to clarify the terms defining the sleep period for the context of nocturnal scratching, which could be easily confused otherwise.

Our analysis of both scratch and sleep publications brought a conclusion that nocturnal scratch should not be measured during the period of “night” but rather during the period of “sleep.” This is an important recontextualization of the term that has not occurred before. The measurement of scratching within the context of attempted and actual sleep will provide a common understanding of the term, better define the measurement period, and accommodate patients’ needs and behaviors.

To begin the effort toward standardization of definitions and selection of fit-for-purpose outcome measures, we developed ontologies for the digital measurement of sleep scratching (Figures 2 and 3). For “nocturnal” (Figure 2), this resulted in recontextualizing this term as the period of intended and actual sleep. From this, main concepts were defined (sleep episode, wakefulness episode, metadata), with specific properties and values appearing for each concept. As an example, the properties that were measured for the concept “sleep episode” included duration, total sleep time, sleep stage, the main sleep event, and start and end.

For “scratching” (Figure 3), we created an ontology that was structurally identical to that created for sleep. For this term, the concepts were scratching episode, scratch area, and metadata. These concepts were described by the properties of duration, tool used, intensity, number of events, and movement type, with specific values attached to each property.

By deconstructing the terms “sleep/rest/nocturnal” and “scratch” and building ontologies to define constituent concepts and their relationships, we reached our goal of developing a standardized definition of scratch during rest or sleep, otherwise known as nocturnal scratch. This evidence-based and patient-centric definition we propose for nocturnal scratching, or scratching during sleep, is that it is an action of rhythmic and repetitive skin contact movement performed during a delimited time period of intended and actual sleep, not restricted to any specific time of the day or night (Table 3). For further clarity, the concepts underlying this definition can be dissected according to the ontologies in Figures 2 and 3.

It is important that a main outcome measure for nocturnal scratching is brought forward in this work. Based on the literature review and developed ontologies, we can propose the main outcome measure of nocturnal scratch: the total scratch time (sum of all recorded scratching bouts) measured during a delimited period of intended and actual sleep within the total sleep opportunity (Table 3). Total scratch time can be also represented as a percentage of the measured sleep period. These definitions of measurement properties and main outcome measures of nocturnal scratching are technology agnostic and can be universally adopted across various sensor technologies or algorithms developed to measure and assess these phenomena.

In this review of the literature, definitions for scratching-related variables differed widely. These differences—due to varying definitions appropriated for devices, analytical techniques, and study goals—hamper the ability to develop standardized outcome measures to assess and compare digital technologies and to monitor various scratching-related disease states in diverse patient cohorts. In addition, only 2 studies provided any description of the variables used to define sleep, both of which used the same device-specific characteristics. A variety of techniques with inadequate a priori definitions were used to determine the sleep/wake times in the other studies, also hampering cross-study comparisons.

For nocturnal scratching, the term “nocturnal” is itself an exclusionary misnomer. Conventionally, the term describes actions or events that happen at night (eg, nocturnal animals). However, in medicine, this term is often used for the actions or activities occurring during the time of sleep (eg, nocturnal incontinence). Although the term “nocturnal” in the measurement of nocturnal scratching has been historically used and well established, it has not been precisely defined in the context of patients’ sleep. In a world where researchers and drug developers are becoming more focused on patient centricity, the term “nocturnal” has become insufficient, as patients’ sleep can occur during different parts of the day due to various reasons or situations. Examples include shift workers, cross time zone travelers, infants, or people with sleep disorders. As this term is rooted in a period of sleep, using the closest sleep period definition and outcome measure—sleep opportunity—provides an elegant solution to establish the term for the purposes of measurement.

While the term “scratch” is widely understood, a unified definition has never been proposed. A definition discerning it from the sensation of itch is often not provided in research studies or during interactions with patients. Moreover, the literature review showed that different research studies often have different definitions of scratching, its measurement, and its properties.

To aid in a unified understanding, we created ontologies for the digital measurement of scratching during sleep based on our literature review. In the ontologies, the properties and definitions for the main concepts of sleep and scratch, along with their nested properties, present a road map for the development of detailed standardized measurements for use in clinical research and practice. A rich list of properties available to add to assessments will ensure improved specificity of measurement and thus add valuable information about the patient populations.

Per our ontologies, the inclusion of more advanced properties remains to be decided by specific research groups or tool developers. To use sleep as an example, this would include whether rapid eye movement (REM) and non-REM sleep periods are included in the definition; whether wake periods are included in the overall sleep period measured; and whether resting on the bed before falling asleep, resting in bed before waking up, and waking after sleep onset “count” toward the sleep period. These aspects will need to be specified by the researchers prior to every measurement of scratching during sleep, and we recommend incorporating them based on the measurement ability of the tools used. While some of these variables are optional, we believe that several of the properties should be required at all times, including the definition of the sleep period considered and the device or method used to measure sleep—whether polysomnography, actigraphy, wearable smart watches, or observer recordings.

In this work, we propose a unified definition of sleep time scratch, or “nocturnal scratch,” together with a list of concepts and properties allowing for the efficient measurement of this phenomenon. Our goal is to enable better communication and sharing of results between various stakeholders taking part in research in AD and other skin inflammatory conditions. The evidence-based definition of the measurement itself, its properties, and main outcome measures provide everyone with a powerful “dictionary” that can ultimately drive rapid acceptance of novel DHT-based measurements in clinical research by the patients themselves, clinicians, industry, technology developers, regulators, and payers. The definition also provides clear specifications for developing digital products that will no longer need to differentiate themselves based on the varied measurement properties but that can now focus on usability, utility, and other features. As a result, researchers, clinicians, and patients will be able to make better choices across high-quality measurement tools.

Digital biomarkers and digitally measured clinical outcome assessments must be subject to reproducible and meaningful verification and algorithmic and clinical validation [55]. Without meaningful and comprehensive definitions, however, there can be no reliable verification or validation procedures. In addition, researchers can tune the properties (variables mentioned above) to achieve better performance or specificity of information, but that does not necessarily mean better performance and usefulness in the actual application of a digital biomarker. The context of the disease and specifics of the patient population for the measurement will always be important aspects to consider when deliberating the parameters of measurements, as will the value of the measure to the patients themselves [56].

We are hopeful that these definitions and ontologies will become standardized and common through their wide adoption. Even the origination of the idea for this research supports these outlooks, as it was sparked by a precompetitive collaboration [57] between industry and invited experts from academia, clinical space, patient organizations, and regulatory bodies. We encourage all these stakeholders to use the proposed ontologies and terminology in their research and evaluation of evidence on the digital measurement of nocturnal scratch to drive unified advancement in the adoption of DHT-based measurements. We are aware that agreeing on the standard values across all research requires a consensus between technology developers, clinicians, patients, and regulators, which is not easily achievable. However, we believe by the standardization of definitions and properties based on data and evidence, as well as by expert discussions and future collaborations, the goal of introducing a unified measurement will eventually be achieved.

This study has limitations. First, the considerable variations in study design and populations for the studies included in this review prevented the assessment of study quality through established evaluation tools [58]. We thus attempted to quantify the quality of evidence in simpler terms by describing the cohort size and particulars of the measurement instruments and methods. In addition, some relevant studies might have been overlooked in our searches, although we included as many applicable terms as possible in the search strings. Finally, studies with negative results might not have been published at all (due to positive publication bias), possibly skewing our findings.

Wearable digital technologies and machine learning algorithms show potential in the assessment and monitoring of scratching during sleep. Standardized evidenced-based definitions for variables related to this symptom will aid in evaluations of these tools and their wide adoption in clinical research.