Meta-Analysis of Studies on the Effects of Digital Therapeutics

Digital therapeutics (DTx), novel treatment methods that have the potential to surpass traditional approaches such as pills, have received considerable research attention. Various efforts have been made to explore effective treatment methods that actively integrate DTx. This review investigates DTx treatment outcomes comprehensively through a meta-analysis. The analysis—a manual search of studies on “digital therapeutics”—includes DTx studies from January 2017 to October 2022. Hedges’ g is used to quantify effect size for fifteen studies analyzed, encompassing eight control groups. Further, a quality assessment is performed using the Bias Risk Assessment Tool. The Hedges’ g analysis results provide weighted average effect sizes across the eight control groups, revealing a substantial value of 0.91 (95% CI: 0.62 to 1.20); this signifies a moderate to large effect size. Further refinement, which excludes one study, yields an increased weighted average effect size of 1.13 (95% CI: 0.91 to 1.36). The quality assessment results consistently indicate a low risk of bias across studies. The meta-analysis results indicate that DTx can provide significant pivotal therapeutic impacts and offer a means to personalize treatment approaches and streamline the management of patients’ treatment processes.


Introduction
Digital therapeutics (DTx) are software-driven therapeutic interventions that combine various types of treatment methods to prevent, control, and treat diseases, ultimately leading to enhanced treatment outcomes [1].DTx represent an emergent research area within digital healthcare that is distinct from the broader concepts of digital health or digital medicine; DTx extend beyond the realm of conventional small molecular compounds, such as pills and capsules, and biological agents, such as antibodies, proteins, and cells [2,3].In other words, they utilize "therapeutic intervention" in products.As stated by Sepah et al. [4], behavioral and self-care programs can be considered primary forms of DTx.
Although DTx products have not yet been approved in Republic of Korea, there has been remarkable growth in the global market, and this sector is projected to have an average annual growth rate of 31.4% by 2026 [5].Regarding DTx products for improving childhood attention deficit hyperactivity disorder (ADHD) symptoms, there have been instances of FDA approval [6] and reports indicating authentic therapeutic effects [7,8].However, no studies have investigated the average overall therapeutic effect of DTx.While studies on DTx interventions targeting hypertension, body mass index (BMI), and pain, such as those by Jasik [9] and Choi et al. [10], have concluded positive effects, no specific research has been dedicated to examining the overall therapeutic outcomes of DTx.
Therefore, our study is aimed at assessing the overall therapeutic outcomes of DTx use by conducting a meta-analysis, based on data from relevant research fields where DTx has been applied.To achieve this goal, a selection of academic papers was conducted by referencing the studies of Matthews [11], Wu et al. [12], Alipanah et al. [13], and Webb et al. [14].The overall therapeutic effects reported in each study were investigated based on the provided data.The significance of the therapeutic effects was determined by calculating Hedges' g, a method first proposed by Hedges in 1981 [15][16][17].Additionally, a subgroup analysis was performed to discern specific therapeutic effects within each subgroup.Last, a funnel plot was created to assess the homogeneity of the research data and provide an overall evaluation of effect size analysis.
The expected outcomes of our study can be broadly divided into three categories.First, by conducting a meta-analysis based on data from relevant existing studies, we can determine the specific and overall therapeutic effects of DTx.Second, we can establish a framework to assess the actual levels of their efficacy.Third, we can present data that specifically demonstrate the treatment effects on individual symptoms for future research in the field.

Study Selection and Data Extraction
Our study systematically analyzed research findings on the validation of the therapeutic effects attributed to DTx.The analysis focused on studies published between January 2017 and October 2022.
Adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol, our literature search involved querying both PubMed and manually identified articles, resulting in 14,211 and 91 articles, respectively (see Figure 1).Additionally, we conducted a manual search of the reference lists from related articles and journals.This search was based on the keywords "digital therapeutics" and "randomized controlled trial".To ensure methodological rigor, we employed a thorough cross-validation process between two reviewers.This involved the identification and exclusion of duplicates, papers that did not meet the predefined inclusion criteria, non-DTx research, studies that did not address our topic (substance abuse, musculoskeletal, anxiety, or mental health), any meta-analysis studies, studies with risk of bias, and studies with incomplete data.After considering the research method and data accessibility, we found 13 qualitative studies and a total of 15 articles that were deemed eligible for inclusion in the meta-analysis.
Table 1 presents the list of these 15 selected articles and offers a comprehensive overview of the individual study characteristics.Further, the descriptive statistics summarizing the key aspects of the included studies are provided in Table 2.
The meta-analysis was executed within the framework of the Participants, Intervention, Comparison, Outcome, and Study design (PICOS) model:

•
Participants: The focus was on patients necessitating treatment through DTx.

•
Intervention: The study encompassed DTx-based experimental interventions.• Comparison: Comparative data were extracted from experimental groups subjected to DTx.

•
Outcome: The primary focus was on assessing the therapeutic effects of DTx.

•
Study Design: Studies utilizing randomized controlled trials (RCTs) and pre-post designs were included.
This comprehensive approach allowed for a nuanced examination of the therapeutic landscape of DTx across diverse study designs and participant profiles, providing a robust foundation for the subsequent meta-analysis.

Study Quality Assessment
To determine the risks of bias in each study, the Cochrane Group's Risk of Bias 2 (ROB2) tool was employed.Each signaling question had the response options of Yes (Y), Probably Yes (PY), Probably No (PN), No (N), and No Information (NI).The assessment categorized bias into three levels: low risk, some concerns, and high risk.

Data Analysis
To analyze the effect sizes, we used Stata 17.0 software.As effect size estimate statistics for group comparisons, standardized mean differences were computed using Hedges' g with the confidence level set at 95%.Further, we generated a forest plot to evaluate statistical heterogeneity.

Quality Assessment of the Selected Papers
We conducted a methodological quality assessment on the 15 studies included in the meta-analysis.The evaluation consistently indicated a low risk of bias for 12 of the 15 studies and "some concerns" for 3 of the 15 studies.Consequently, all 15 studies were selected as reliable data sources for further analysis (Table 3).
Table 4 presents the results of investigations into the effects on various subgroups and diseases, and the combined results of selected studies categorized by subgroups, which provide insights into the pooled effects across different domains, are presented in Table 5.

Discussion and Conclusions
The current meta-analysis systematically evaluated the therapeutic effects of DTx by rigorously analyzing 15 independent studies.Employing stringent screening processes and carefully addressing issues related to heterogeneity, our findings unequivocally demonstrated the significant positive impacts of DTx on patients' overall health outcomes.Specifically, the heterogeneity between subgroups can be useful for comprehensively examining the effects of DTx, including the evaluation and comparison of DTx between subgroups.Moreover, the related results can be used to encourage DTx-related research activities in each subgroup and further expand related technologies.The analysis demonstrated that DTx exhibited a remarkable reduction in adverse effects, especially in alleviating pain (effect size = 0.255, p = 0.049, 95% CI = 0.000-0.389).This finding holds significant implications for improving patients' quality of life and reducing healthcare costs associated with pain management.
A further breakdown by disease group revealed intriguing insights.We observed substantial therapeutic effects in studies on substance use disorders [21] and anxiety associated with peripheral intravenous cannulation [22].The analysis of effect sizes indicated significant enhancements in perception measures (FPS-P, GAD-7, and HADS).These findings underscore the potential of DTx as valuable treatment methods for addressing substance abuse and anxiety-related disorders.
Notably, the five studies focusing on musculoskeletal disorders did not yield statistically significant effects; similarly, the four studies centered on mental health did not exhibit substantial impacts either.The lack of effectiveness of DTx in such disease categories may relate to the limitations of treatment methods through behavioral interventions versus those that act directly on the body.This may be particularly true for diseases and techniques that do not fully reflect individual and subjective physical and mental characteristics.Specifically, in mental health-related studies, the intricate interplay of individual psychological states with various contextual elements may contribute to the observed lack of significant effects.Mental well-being is inherently subjective and influenced by diverse personal experiences and environments.Furthermore, if DTx do not present comprehensive solutions for specific mental health conditions, alternative or complementary approaches may be warranted.A holistic approach that integrates DTx with traditional therapeutic methods could yield more promising outcomes for certain mental health conditions.Regarding musculoskeletal disorders, it is possible that DTx may not provide the requisite level of physical intervention necessary for this particular condition.These conditions often necessitate tailored physical therapeutics, exercises, and other hands-on treatments, which DTx may not adequately incorporate them.A combination of DTx and traditional physical therapeutic approaches could potentially address the limitations of either method.
However, DTx can be considered as effective approaches for the management of substance abuse and anxiety, particularly when implemented through well-structured behavioral programs and self-care in specific fields.Our findings underscore that DTx interventions have significant positive health outcomes, further highlighting their therapeutic feasibility.Additionally, the meticulous application of the PRISMA protocol here helped us identify future directions for presenting DTx-related research findings.Our study emphasized the importance of precise result presentation and the establishment of universally acceptable criteria to further validate and demonstrate the effects of DTx.Further, it should be noted that the meta-analysis conducted in this study provided valuable insights into the overall research trends and outcomes regarding the therapeutic effects of DTx.
However, our study does have some limitations.First, the diversity of the various research designs and measurement tools makes consistent analyses and interpretations difficult.Second, securing sufficient reliability was challenging, based on the limited number of studies in a specific disease category.That is, our analyses may be influenced by various study characteristics, designs, and factors, such as sample size, study duration, types of DTx employed, and measurement tools.It is therefore crucial to acknowledge that the results may vary depending on the target population and disease categories.To ensure the generalizability of the study findings, it is imperative to track ongoing research and validate the results over time.To further enhance our understanding of DTx, it is necessary to diversify research areas and explore their strengths and weaknesses in various fields.This comprehensive approach will enable the identification of effective treatment methods and strategies tailored to specific conditions and patient needs.Moreover, it is necessary to standardize study designs and methodologies to minimize the influence of these confounding factors and provide more definitive conclusions regarding the efficacy of DTx across different disease categories.
DTx technology and related markets are emerging rapidly with the concurrent development of science and technology.Their potential for improving the quality of life of numerous patients is anticipated to grow exponentially in the future, along with the efficient employment of advanced technologies.However, at the same time, there may be several barriers to adopting and applying DTx.These include high development costs, technical difficulties, traditional resistance from the medical community, and the attitudes of patients and medical personnel.To overcome these barriers, integrated cooperation is required in various aspects, such as the effective integration with the existing medical system, the improved understanding of medical personnel through education and training programs, and the introduction of cost-effective models.

Figure 1 .
Figure 1.Article selection using the PRISMA protocol.Figure 1. Article selection using the PRISMA protocol.

Figure 1 .
Figure 1.Article selection using the PRISMA protocol.Figure 1. Article selection using the PRISMA protocol.

Table 1 .
General information on selected studies.

Table 2 .
Descriptive statistics of selected studies.

Table 3 .
Quality assessment of the selected papers.

Table 4 .
Meta-analysis of the selected studies.
Notes: Explanation of abbreviations and symbols: Studies: Number of studies; Subgroup: Subgroup classification; ds: Effect size; 95% CI: 95% Confidence Interval; Weight (%): Weight percentage; p: Significance level; LL: Lower limit of the confidence interval; UL: Upper limit of the confidence interval.

Table 5 .
Meta-analysis of the selected studies by subgroup.