Table 1.

Characteristics of included studies.

Authors	Study design	Population	Main findings
Au-Yeung et al. 22	Pilot demonstration project	n  =  111 healthy volunteers and patients with tuberculosis, heart failure, and hypertension.	The positive and negative detection accuracy of the system when compared to directly observed ingestion was 97.1% and 97.7%, respectively.
Belknap et al. 24	Pilot demonstration project	n  =  30 tuberculosis patients who are enrolled for DOTS co-ingested the sensor with their ATT medication.	The positive detection accuracy of the ingestible sensor system was 98.1% when the three percent data missed because of prototype malfunction and when a repeat measure model was used it increased to 99.6%. The identification accuracy was 100%.
Bonacini et al. 10	Pilot demonstration project	n  =  28 patients with chronic HCV infection treated on fixed dose combination of ledipasvir/sofosbuvir.	89% of the patients had medication adherence >/ = 95% during the study period. 26 (93%) of the study participants attained sustained virological response when assessed after 12 weeks or more of posttreatment. Of the two unresponsive cases one had suboptimal adherence (<90%) and other was due to drug resistance.
Chai et al. 25	Pilot demonstration project	n  =  15 opioid-naïve individuals who were discharged from the emergency department with acute fracture pain.	The study was able to detect the actual patterns of opioid ingestion and found that most of the opioid naive patients with acute fracture only use opioids during initial days depending on their perception of their pain.
Chai et al. 26	Pilot demonstration project	n  =  10 patients with acute extremity fractures.	The DPS was able to detect the ingestion events with 87% accuracy. The decrease in accuracy is because of two participants who failed to charge the batteries of the system due to acute pain. The satisfaction survey among patients following the DPS use revealed that the DPS is palatable, easy to use, and valuable.
Daar et al. 27	Pilot demonstration project	n  =  15 PLWH on stable antiretroviral therapy (ARV) with suboptimal adherence (<90%) over last 28 days self-reported or found by clinician on 6 months of clinical evaluation)	More than 70% of the participants stated that they tolerated the system well and it is comfortable to use. Over 80% agreed that the system is helpful.
DiCarlo et al. 28	Pilot demonstration project	n  =  37 hypertensive patients.	The wearable sensor detection of 510 valsartan doses was 98%. Mean taking and timing adherence rates between clinical visits were 90% and 83%, respectively.
Eisenberger et al. 29	Pilot demonstration project	n  =  20 stable adult kidney transplants.	Out of 4136 ingestible event markers, enteric coated mycophenolate sodium capsules (IEM-ECMPS) prescribed 2824 (68%) ingestions have been detected without being directly observed. 100% of 34 directly observed ingestion events were detected accurately. The detection rate of daily medication ingestion was 99.3% in patients who took two IEM-ECMPS capsules BID and 100% in those who took one IEM-ECMPS capsule twice daily.
Flores et al. 30	Pilot demonstration project	n  =  20 healthy volunteers.	100% of directly observed ingestion events are detected. Of 384 self- administered ingestion events 371 got detected accurately (96.6%). Of the 391 recorded ingestion events 385 (98.47%) were remotely transmitted from the reader to a secure cloud storage. The overall adherence to the prescribed study capsules as measured by the DPS was 97.75%.
Frias et al. 31	Cluster-randomized pilot clinical trial	n  =  120 participants with uncontrolled hypertension and diabetes mellitus.	The participants who used the DPS had a significant reduction of −21.8 mmHg (SE 2.5) in systolic blood pressure (BP) when compared to −12.7% (SE 2.8) for the usual care group at 4 weeks. 81% from the digital medicine group achieved their BP goal of 140/90 mmHg, whereas only 33.3% among the usual care group achieved the BP goal. The patients who used the DPS had greater reductions in HbA1c, DBP, and LDL-C.
Kamal et al. 32	Pilot demonstration project	n  =  15 people living with HIV (PLWH) and n  =  6 health care providers (HCPs) who are directly involved in their health care.	Eight of the 15 PLWH used the digital medicine were supportive, and the rest found it inconvenient. HCPs commented that the text message reminder in the study is the most important aspect, and they would recommend it to people with adherence difficulties for short-term use, up to 6 months.
Moorhead et al. 33	Post-hoc analysis	Data of n  =  57 patients participated in a cluster-randomized clinical trial among patients with uncontrolled hypertension and type 2 diabetes gathered in this efficacy study and n  =  74 patients in the safety study.	This analysis suggested that the digital health medication dose reminder in patients with low medical adherence was the most important function of DPS and there was no incidence of overdosing due to medication dose time reminder during the study period.
Noble et al. 34	Pilot demonstration project	n  =  39 hypertensive patients and n  =  15 practicing pharmacists.	The mean change in SBP in patients who used the DPS over the 2-week evaluation period was − 7.9  ±  22.1 mmHg; mean change in DBP was − 2.8  ±  12.9 mmHg. 32% of the study participants have uncontrolled hypertension due to poor medication adherence. 15 patients participated in the post intervention survey and all of them found the system experience was positive. 91% of the pharmacists surveyed agreed that the DPS helped them in assessing the actual medication ingestion trends and giving their patients specific recommendations.
Peters-Strickland et al. 35	Pilot demonstration project	n  =  67 adult patients with schizophrenia and their health care providers.	78% of patients rated were somewhat satisfied, satisfied, or extremely satisfied with the DPS and 70% of them found it somewhat helpful, helpful, or extremely helpful. 65% of them rated it as somewhat easy, easy, or very easy to use the system.
Profit et al. 36	Pilot demonstration project	n  =  30 and n  =  29 healthy volunteers for two sub studies.	The overall accuracy of ingestion detection was 78.3% during sub study A. The overall accuracy of ingestion detection in sub study B is 96.6%. The mean latency of time from ingestion signal detection by wearable sensor to detection by cloud-based server is 6.2–10.3 min. 50% of the ingestions detected before 2 min and 90% of the detection occurred within 30 min of ingestion.
Sulkowski et al. 37	Pilot demonstration study	n  =  288 adults with chronic HCV infection.	Sustained virological response was achieved in 99.1% of 218 participants who had HCV RNA assessed at ≥ 10 weeks post-treatment.
Thompson et al. 38	Pilot demonstration project	n  =  28 patients with high CVD risk attending cardiac prevention and rehabilitation.	90% of the study participants self-reported that they have good adherence (never missed a medication) but when adherence was directly measured using the DPS only 57% found to have good adherence (> 80%).
Triplett et al. 39	Pilot demonstration project	n  =  33 pediatric solid organ transplant patients. 21 of them participated in the satisfaction survey. Five health care providers involved directly in the transplant completed the satisfaction survey.	27 participants discontinued the study prior to 6 months. The main reason behind withdrawal from the study was the inconvenience with the patch used in the system. Even though, majority of the study participants felt that DPS increases motivation and sense of control.
Chai et al. 40	Qualitative study	n  =  30 participants of age >18 years, self-reported HIV negative, cis-gender MSM, currently on pre-exposure prophylaxis (PrEP) or eligible for PrEP, and self-reported substance abuse excluding alcohol with in last six months.	Participants described that the DPS could be helpful during events of substance abuse and times of spontaneous sexual behavior, when there is a high chance of failure to adhere to medication. n  =  19 participants showed willingness to use the DPS for monitoring real-time preexposure prophylaxis adherence. While few had concerns of safety, size of the reader, stigma and some due to lack of knowledge about the system.
Holender et al. 41	Focus group discussion	n  =  12 cardiovascular patients aged >65 years participated in two focus groups.	The participants suggested that the DPS is welcoming once it's familiar and they feel that this could improve medication monitoring by the health care providers, prevent adverse events, reduce confusion with polypharmacy and prolong independence. The barriers discussed were lack of cognitive abilities and memory problems in elderly and difficulty in familiarizing new technology.
de Mendoza et al. 42	Qualitative study	n  =  10 healthcare providers treating breast cancer survivors participated in qualitative semi structured interviews. n  =  19 participated in an online survey.	Providers consider the side effects, access, forgetfulness, knowledge and beliefs, and communication are the barriers for the patients to adhere to hormonal therapy and many of their participants have adherence issues. They were confident that DPS can provide them timely and accurate adherence information and also help their patients to adhere to the medication. They think that privacy, safety, and access would be the patient's concerns with DPS.