Accurately measuring and monitoring patient medication adherence is critical in clinical practice and research settings but continues to be a challenging task globally [1]. Various methods are used to measure medication adherence, such as patient self-reports, pill counts, pharmacy refill records, drug metabolites or biomarker testing, and directly observed therapy (DOT) [1]. However, none of these methods have been accepted as a standard measure of medication adherence across a variety of settings [2]. More recently, sensor technologies have been increasingly used to track the medication-taking behaviors of patients [1]. For example, the Medication Event Monitoring System (MEMS) can record every time the patient opens the pill bottle via a sensor embedded in the pill cap [3,4]. Such technologies provide a unique opportunity to measure and monitor patient medication adherence over time [1]. The notion that medication adherence monitoring technology represents the gold standard of measurement of patient medication adherence has been voiced by some researchers [3-9] but continues to be disregarded by others [10-15]. There is limited consensus on how to determine or select the appropriate medication adherence monitoring technology for use, which may be due to the lack of appropriate technology assessment criteria in this field.

The advantages and limitations of the commonly used methods for measuring medication adherence have been described in the literature. For example, DOT allows for direct observation of patient medication-taking actions [16-18], but it is expensive to sustain and produces a constrictive time strain on both health care providers (HCPs) and patients’ daily schedules [1,12,18,19]. As a common way to measure medication adherence, patient self-reporting respects patient autonomy but carries the potential risk of patient overestimation or underestimation of their adherence abilities [20-22]. Medication adherence monitoring technologies with various types and features are being continuously developed and upgraded [1]. Some newly developed technologies may possess unique features that are unfamiliar to users [23]. Despite this literature, there is no summary or synthesis that reflects a clear understanding of the characteristics and values of a variety of medication adherence monitoring technologies. There is a growing need for technology assessment criteria to guide the development and selection of appropriate technologies for monitoring medication adherence to improve patient outcomes [24].

Stakeholders’ expectations regarding the use of health information technology for monitoring medication adherence also vary. From a clinical practice perspective, a user-friendly interface and the accurate monitoring of adherence are considered when selecting appropriate monitoring technologies [1]. From the technological development perspective, although system accuracy and data fidelity remain high priorities, developers also need to consider the feasibility of technical engineering of the system, such as energy consumption and battery lifetime [25]. Advanced medication adherence monitoring technologies may not be limited to a single method to gather patient medication adherence information [1]. In addition, human interactions with these technologies can be complicated owing to the comprehensive medical and pharmacological contexts, as well as multidimensional patient medication adherence behaviors [22]. A compiled summary and assessment of currently available applications of medication adherence monitoring technologies is important for a better understanding of their capacities and performance when making decisions for their adoption and use.

The purpose of this narrative review is to summarize literature reports on the current applications of medication adherence monitoring technologies and identify potential assessment criteria to support decisions related to technology development and adoption.

PubMed, Scopus, CINAHL, and ProQuest Technology Collection databases were searched because of their broad collection of literature focusing on health, health care, and technological domains. A combination of search terms was included as follows: (medication adherence) AND (measurement technology OR monitoring technology). A full list of search strategies used for each database is included in Multimedia Appendix 1. To gather the most recent collection of medication adherence monitoring technologies, searches were focused on scholarly articles published between January 2010 and June 2021 and written in English.

Studies were included in this narrative review if they met the following criteria: (1) described the development of medication adherence monitoring technologies, (2) assessed the characteristics of medication adherence monitoring technologies, or (3) tested the application of technologies for monitoring medication adherence. All study methods were included. Only articles published in English with their full text available were included. Considering the ease of dispensing medication and self-administration of pill form of medications, we focused on medication adherence technologies suited for pills. Medication adherence monitoring technologies that suited nonpill forms of medications, such as inhalers, eye drops, and injectable medications, were excluded. Studies that did not provide adequate descriptions of technology characteristics or used technologies that did not monitor patient medication adherence were also excluded. Study selection was performed manually using this set of eligibility criteria.

Data concerning medication adherence monitoring technologies were manually extracted from the reviewed articles by the first author (MM) and discussed with the research team. These elements included the following: (1) type of technology, (2) name of technology, (3) technical features, (4) data capture and applications, (5) perceived advantages, and (6) limitations of the identified technologies. Data directly related to publications, such as the country and publication year, were also gathered. Information regarding adherence monitoring technologies was extracted and organized into a table for further synthesis (Multimedia Appendix 2 [1-6,9-13,15-18,20,26-99]).

A descriptive analysis of the characteristics of the selected studies was conducted. Key characteristics, including the technical features, data capture methods, advantages, and limitations of each technology type, were assessed and summarized. Common and recurring elements were coded and categorized as potential assessment criteria. All identified potential criteria were discussed and evaluated among the team members until a consensus was reached. The final criteria were organized into categories and subcategories and presented as a matrix.

A total of 3865 records were retrieved from the database search. Of these 3865, the removal of duplicates left 3774 (97.65%) articles for title screening. After reviewing the titles and abstracts for relevance, 7.63% (288/3774) of the articles were identified for retrieval, of which 97.2% (280/288) were successfully gathered. Of the 288 articles, 8 (2.8%) articles were not retrievable because their full text was not available on the web. Following the assessment for eligibility via full-text review, 35% (98/280) of the articles were included in the final analysis. Figure 1 shows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart describing the overall search and selection process.

Among the reviewed articles, the vast majority (72/98, 73%) were published between January 2015 and June 2021. Over half (50/98, 51%) of the identified studies were published in the United States, followed by Canada (8/98, 8%), and Japan (4/98, 4%). The study types and designs varied greatly among the 98 reviewed articles. Most studies (41/98, 42%) were pilot tests of feasibility, acceptability, usability, or proof of concept. Only a few studies were randomized controlled trials, including pilot randomized controlled trials (5/98, 5%), retrospective cohort studies or secondary data analyses (6/98, 6%), or qualitative studies (8/98, 8%). Literature review articles (8/98, 8%), study protocols (4/98, 4%), and commentary and editorial comments (2/98, 2%) were included in the review analysis. The most common medications studied were tuberculosis treatment regimens (19/98, 19%) and antiretroviral therapy for HIV (16/98, 16%).

As the adoption and development of medication adherence monitoring technologies continue to increase, understanding their key characteristics is vital. This narrative review provides an overview of the technical features, data capture methods, and advantages and limitations of current medication adherence monitoring technologies reported in the literature and synthesizes 28 technology assessment criteria that can be used to guide the development and selection of relevant technologies. Overall, there were 8 types of medication adherence monitoring technologies, dominated by electronic pill bottles, electronic pill boxes or bags, and ingestible sensors. Although technical features varied by technology type, there were common expectations regarding the advantages of using these technologies for accurately monitoring medication adherence and increasing the adoption of these technologies in patients’ daily lives.

All current medication adherence monitoring technologies have varying degrees of technological restriction. The most commonly reported technology types, electronic pill boxes or bags and electronic pill bottles that use opening events as a proxy for medication ingestion, face undesired patient behaviors such as pocket dosing and curiosity openings, which are obstacles to the device’s accuracy for patient adherence estimates [2,9,11,14,16,27,30,33,37,41,48-50]. Despite this limitation, the popularity of developing and using pill monitoring devices remains, which may be due to their unobtrusiveness and convenience of use in patients’ everyday routines, suggesting an increasing adoption of objective measurement and monitoring of medication adherence through technological approaches.

Although electronic pill bottles possess extensive histories of being used in both clinical and research settings, the presence of many other medication adherence monitoring technology studies in the pilot and feasibility phase implies that the integration of newer technologies, such as motion sensor–based technologies and ingestible sensors, is still relatively new and ongoing [1,4,15,18,60,64,76]. Overall, technologies capable of monitoring patient medication adherence provide significant advantages, including real-time medication adherence data reporting, yet questions concerning the accuracy of these devices prohibit them from becoming a gold standard in clinical and research standings. Thus, until further developments in medication adherence monitoring technologies occur, multiple methods for patient medication adherence assessment must be used to evaluate patient medication adherence rates and behaviors [1].

Many medication adherence monitoring technologies possess software to organize patient medication adherence data to an extent; however, most of these devices require separate analysis and quantification of the data by HCPs or researchers [1], creating a significant burden of time consumption and the concern of further data integration with other technology applications. The development of a patient medication adherence data management software that can construct automatic visualizations of patient medication adherence estimates should be considered to provide an easy interpretation of patient medication adherence patterns. Moreover, the use of advanced software for adherence data processing and presentation may improve the adoption and integration of medication adherence monitoring technologies in clinical settings.

In addition to variances in technical features, current medication monitoring technologies differ in their data capture methods and the subsequent use of such data in relation to patients’ medication adherence assessment. The ability of most medication adherence monitoring technologies to provide real-time observations of patient medication adherence behaviors is beneficial to HCPs and researchers to prevent nonadherence and facilitate appropriate interventions [1,15,36,37,79]. However, most of these technologies rely on proxy measures of medication adherence, such as device opening events, thereby limiting their data accuracy [2,9,11,14,16,27,30,33,37,41,48-50]. Furthermore, successful implementation of these technologies in clinical settings or the integration of patient medication adherence monitoring data into clinical practice has rarely been reported. One of the major barriers is the interoperability of these monitoring technology systems with established clinical information systems and workflow. To facilitate the adoption of medication adherence monitoring technologies in clinical systems to improve patient care, the method of adherence data capture must be feasible for targeted patients and the acquired data must be easily integrated into standard electronic health record systems. The medication adherence data capture methods and data use presented in this review can help guide HCPs and researchers toward the appropriate selection of medication adherence monitoring technology. Developers must also consider the implications of medication adherence data capture within clinical and research settings to ensure greater ease of use for both patients and providers.

To the best of our knowledge, this is the first collection of assessment criteria focused on technologies to monitor patient medication adherence. The proposed assessment criteria include five major categories as follows: development information, technology features, adherence to data collection and management, feasibility and implementation, and acceptability and usability. The identified criteria highlight significant aspects of medication adherence monitoring technologies that must be considered during technology development and adoption. For example, an important component of medication adherence monitoring technology implementation is cost; however, a common limitation of these technologies is their expensive price tags [1,2,5,7,9,12,18,27,38,54,78,85,102]. The proposed criteria emphasize the cost efficiency of medication adherence monitoring technologies within the feasibility and implementation category. The high cost of devices restricts their adoption in clinical and research settings because other methods, such as patient self-reports, are significantly cheaper [1,78].

As this compilation of assessment criteria was formed by reviewing the current literature, other existing challenges within technology acceptance or technology design features were also addressed, such as risks to patient privacy or the effect of large device size on user adoption owing to daily life inconveniences [25]. Given the multifaceted nature of the proposed assessment criteria, they can be used to guide the improvement of these technologies for better medication adherence measures and monitoring.

In addition, our set of proposed assessment criteria possessed a structure similar to that of other validated mobile health assessment frameworks. For example, a pyramid for app evaluation framework, proposed by Henson et al [104] and adapted by the American Psychiatric Association as the App Evaluation Model, introducing a similar 5-level structure of evaluation categories, including access and background, privacy and security, clinical foundation, usability, and data integration toward therapeutic goals [104,105]. Similarly, each category covers a few specific evaluation criteria; for example, ease of use is assessed under the usability category [105]. Certain general technology assessment criteria can be applied to both mobile health apps and medication adherence monitoring technology, such as usability, privacy and security, and data integration. However, adherence monitoring technology possesses technical features to support medication storage and management, which results in its unique assessment criteria, such as the medication storage capacity of the device or date-and-time stamps indicating medication-taking actions. The collection of medication adherence monitoring technology assessment criteria was generated from an extensive literature review and information synthesis, which demonstrates its solid evidence foundation but also suggests that further empirical tests and validation are needed in the future.

This narrative review has some limitations. First, our database selection and search strategies might not have been sufficiently extensive to capture all published literature. Moreover, we limited the studies to those published in English, potentially excluding other existing medication adherence monitoring technologies from non-English sources. The proposed medication adherence monitoring technology assessment criteria are representative of the elements identified in our literature review and synthesis, which are subject to further validation and evaluation. We did not review detailed information published by specific manufacturers. Finally, given that the scope of this review was focused on medication adherence technologies used for the monitoring of pill form medications, the assessment criteria and the rest of our findings may not be generalizable to all types of medication. It is noteworthy that a large proportion of the identified articles were pilot or feasibility studies. Consequently, our assessment domain of the criteria may also be limited to the early stages of technology development.

Overall, this narrative review presents a summary of the current technological features and data capture methods, reports the advantages and limitations of medication adherence monitoring technologies for pill form medications, and proposes a potential technology assessment criteria. Our constructed assessment criteria are crucial for the development and adoption of these technologies. Specifically, further technological development is required to expand the interoperability of medication adherence monitoring technology systems in clinical settings. The increased implementation of technologies that monitor patient medication adherence has demonstrated the potential to improve patient medication adherence behaviors. Although this technological method of patient medication adherence monitoring cannot be defined as the gold standard method for medication adherence monitoring, the functionalities that they possess may improve patient medication adherence and support greater patient health outcomes over time.