Acta Clinica Belgica

International Journal of Clinical and Laboratory Medicine

ISSN: 1784-3286 (Print) 2295-3337 (Online) Journal homepage: https://www.tandfonline.com/loi/yacb20

Remote digital monitoring of medication intake:

methodological, medical, ethical and legal
reflections

Wim Van Biesen, Johan Decruyenaere, Katerina Sideri, Julian Cockbain &
Sigrid Sterckx

To cite this article: Wim Van Biesen, Johan Decruyenaere, Katerina Sideri, Julian Cockbain &
Sigrid Sterckx (2019): Remote digital monitoring of medication intake: methodological, medical,
ethical and legal reflections, Acta Clinica Belgica, DOI: 10.1080/17843286.2019.1708152

To link to this article: https://doi.org/10.1080/17843286.2019.1708152

Published online: 23 Dec 2019.

Submit your article to this journal

Article views: 10

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=yacb20
ACTA CLINICA BELGICA
https://doi.org/10.1080/17843286.2019.1708152

Remote digital monitoring of medication intake: methodological, medical,

ethical and legal reflections
Wim Van Biesena,b, Johan Decruyenaereb,c, Katerina Siderid, Julian Cockbaind and Sigrid Sterckxb,d
a
Department of Nephrology, Ghent University Hospital, Ghent, Belgium; bResearch consortium on Justifiable Digital Healthcare, Ghent
University Hospital, Belgium; cDepartment of Intensive Care Medicine, Ghent University Hospital, Ghent, Belgium; dBioethics Institute
Ghent, Department of Philosophy & Moral Sciences, Ghent University, Ghent, Belgium

ABSTRACT KEYWORDS
In 2017, regulatory approval was given in the US for a ‘digital pill’, a pill for which actual Digital pill; artificial
ingestion could be remotely monitored. The pill, Abilify Mycite is marketed by Otsuka but the intelligence; adherence; big
monitoring system derives from Proteus Digital Health. In this paper, we focus on this digital data; ethics; autonomy;
pill and another equivalent system from AiCure which relies on facial recognition. Both systems healthcare organisation
not only remind the patient to take a pill but also verify the actual intake. In this process,
patient-related data beyond the fact that the pill has been taken are also collected and sent to
a remote computer system of the system-providing company and possibly to third parties.
Although marketed as ‘innovative’, the introduction of such systems raises questions as to
the limitation of patient autonomy, secondary uses of patient data, impact on the physician’s
liability, and artificial inflation of drug prices. Whereas incorrect medication taking can be
problematic, it can be questioned whether remote intake-monitoring systems are, from an
ethical, legal and social perspective, the ideal way to address this. In this paper, we will reflect
on this question from the position of the different potential stakeholders involved.

Introduction
this kind, Abilify MyCite, contains the antipsychotic drug
Incorrect intake of medication is an important health- aripiprazole, but by modulating the intensity, duration
care problem that generates avoidable costs as well as and frequency of the induced electromagnetic signal,
additional suffering through poor outcomes [1]. different IEMs could be produced to discriminate intake
Recently, two different ways to try to address this of every possible medication. The IEM’s signal is
problem using digital technology have been launched detected by a patch on the patient’s chest and further
[2–4] (Figure 1). transmitted via the patient’s mobile phone to the cen-
Whereas these technologies were welcomed in the tral Proteus database. The patch/phone also detect
press as innovative and promising [5,6], it remains to meta-data, such as location, heart rate, and stress levels
be determined whether, to what extent, and at what [8], which can subsequently be merged with other
expense these devices will help improve the problem external datasets. Whereas this digital pill is officially
of incorrect medication intake. While at first sight promoted as a personal health management tool [7],
attractive, just as for many technologies, the conse- promoting patient empowerment, with a safety net of
quences of their implementation might have serious caregivers and physicians, this strategy would allow
impact on the way we organize care. Proteus to generate and sell algorithms predictive of
In this paper, we will consider the different stake- incorrect medicine intake to interested third parties.
holders involved and explore the potential conse- Proteus has announced that it has formulated 15 digital
quences of the introduction of devices that can track medicines for cardiovascular and metabolic conditions,
medication intake in an automated manner. 7 for infectious disease (including TB, hepatitis C and
HIV) and is working on digital cancer drugs [9]. They
have also paired up with nine health systems in the US
Examples of systems for remote monitoring of which have begun using the Proteus system with med-
medication intake ications for hypertension and hepatitis C [10].
The Proteus system uses a pill containing a drug The AiCure system in contrast relies on facial recog-
together with an ‘ingestible event marker’ (IEM) to nition image analysis [3]. An app on the patient’s
track its actual ingestion (Figure 1a).7 After ingestion, mobile phone reminds him/her to take a pill and pro-
the IEM generates a signal when it contacts the gastro- duces a video to verify the actual swallowing of the
intestinal fluid. The first commercially available pill of correct pill. The data are tracked and stored by the

CONTACT Wim Van Biesen Wim.vanbiesen@ugent.be Dept of Nephrology 0K12IA, Ghent University Hospital, Ghent, Belgium
It ain’t easy to swallow, it sticks in the throat (Bob Dylan)
© Acta Clinica Belgica 2019
2 W. VAN BIESEN ET AL.

Figure 1. The AiCure and Proteus systems to track medication intake.

AiCure computing system. To avoid patients cheating Adherence to a medication regimen can be defined
by pretending to take the pill or by taking a different as the extent to which patients take medications as
pill of similar shape and colour, AiCure has patented prescribed by their healthcare providers [1,13].
cheating identifying system, including one to recog- Although at first sight irrational, not following prescrip-
nise only pills surface-marked with a complex pattern. tions often represents a rational decision on the part of
The pattern could itself be registered as a trade mark patients. Influencing factors may be their views on
and licensed exclusively to a pharma company or, medication-taking, their life circumstances, available
alternatively, AiCure could become a producer of its resources, or competing priorities. The concept adher-
own, complex-marked, pills containing a generic active ence is frequently preferred over compliance: the for-
component. AiCure acknowledges collecting bio- mer assumes a therapeutic agreement between the
metrics, images, voice recordings, and even internet patient and the physician after ample explanation,
traffic of the patient, and holds itself out as ‘a beha- whereas the latter suggests that the patient is passively
vioral data analytics company’ [11]. following the doctor’s orders [14–16]. Compliance thus
indicates a paternalistic conceptualization of medica-
tion-taking behavior, disregarding patients’ expecta-
What problem would digital tracking of
tions and experiences. With the concept of
medication intake address?
adherence, the patient is not to be held solely respon-
Failure by patients to correctly take their medications is sible for the failure of a treatment plan because of
widespread, and rates have not changed substantially incorrect medicine intake, as his/her medication intake
over time [12]. Incorrect medication intake can lead to is the result of sound and convincing argumentation
suboptimal health for the patient and, counter-intuitively, by the prescriber [14]. Thus, for patients, adherence
may increase health-related costs. Society may suffer by means making reasonable efforts to follow the pre-
spread of infection, emergence of drug resistance, or by scribed dosage regime, whereas compliance means
increased physical risk if the drugs concerned are beha- following it irrespective of any undesired or intolerable
vior modifiers. In itself, improvement of correct medica- effects. Accordingly, a system which merely assists the
tion intake is thus a desirable goal. patient by reminding that a pill is due to be taken may
 ACTA CLINICA BELGICA 3

improve adherence, but a system which involves any gave Otsuka approval for the Abilify MyCite digital pill as
form of coercion or undue inducement is instead one a combination product (a device/drug combination)
for increasing compliance. and thus as a pharmaceutical as a result of its ‘primary
In competent patients (Table 1), patient autonomy mode of action’, so that it can be prescribed like any
is always violated by paternalistic decision making or other drug. Importantly, the FDA acknowledged that
lack of information, irrespective of the actual intake there was no evidence that the approved product actu-
being monitored or not. If after shared decision mak- ally improved adherence, and certainly not outcomes for
ing, the patient is willing to take the medication, patients [20].
a simple alerting system, e.g. an app on the patient’s In the EU, the EMA only approved the IEM for tracking
mobile phone, is sufficient to prevent the problem medication intake by the patient in clinical trials [19].
arising from forgetfulness [12]. Tracking the actual Proteus likely intends to compare the actual pill
drug intake in this setting is thus unnecessary, dispro- intake rates for aripiprazole recorded over years for
portionate and intrusive, and therefore in principle patients taking conventional pills to those recorded by
inappropriate. If a competent, informed patient refuses Proteus for Abilify MyCite in order to prove that the IEM
to take the medication, forcing or unduly inducing improves correct medication intake, i.e. to produce the
him/her to track medication intake would violate data for the second stage of regulatory approval men-
patient autonomy. tioned above. Only with that second stage of approval
When the patient cannot be involved in the decision would it be legal to advertise digital pills to interested
making because of incompetence or legal constraints, parties, e.g. physicians, insurers and national health ser-
monitoring intake cannot per se be considered to vices, for improving medication intake.
violate patient autonomy. Even then, however, propor- A recent systematic review found non-randomized
tionality requires that the indication for taking the controlled trials to evaluate Abilify Mycite, and, astonish-
drug should be sufficiently important to justify breach- ingly, the FDA approval was based on three non-
ing patient autonomy. Whereas some medications do randomized non-comparative cohort studies. There
have life-saving properties, for many this is not the was no evidence of any benefit of the digital over the
case. In these latter cases, tracking actual intake can non-digital version of aripiprazole [4]. Several clinical
be considered disproportionate, as there is no over- studies attempted to demonstrate the clinical and eco-
riding benefit to breaching patient autonomy. Thus, nomic benefits of using digital medicines to improve
whereas there are circumstances where the limitation outcomes in these patients. In a small open-label study
of patient autonomy by drug intake monitoring sys- (N = 109) an improved blood pressure control and
tems can be justified, they risk creating a slippery slope Hb1Ac in patients with uncontrolled hypertension and
towards unacceptable violation of patient autonomy. type 2 diabetes after the introduction of an IEM was
claimed [21]. However, these observations were made
after a short period (4 and 12 weeks), so the sustain-
How is the marketing approval of these
ability of the small effect would require further clarifica-
devices regulated?
tion. Of note, already by week 4, 9 out of 82 patients in
Regulatory approval (FDA or EMA) is concerned with the intervention group had dropped out of the study.
safety and efficacy. For the Proteus system, demonstrat- Remarkably, included patients were paid 150–525 dol-
ing efficacy has been turned into a two-stage exercise, lars, depending upon the study arm. Furthermore, revi-
involving first showing that the system provides accu- sion of in- and exclusion criteria, and expanding
rate verification of the ingestion of a pill, and only later recruitment to non-English native speakers was needed
demonstrating that it improves correct medication to ensure recruitment, possibly pointing to a certain
intake of pills containing active ingredients. Of note, reluctance of patients to take part in a study with such
there is no need for Proteus to demonstrate improved a device. The results of a study of the combination of the
clinical outcome for patients. In the first stage, the FDA Proteus system with Gilead’s HIV medication Truvada
approved the Proteus’ component of the pill, the IEM, as are currently still awaited [22]. Interestingly, this study
a new generic category of device, ‘ingestible event mar- was supported by Gilead, Proteus and the US National
ker’, with a Class II risk classification [17,18]. The exter- Institute of Mental Health.
nally worn patch was described as a miniaturized, As to AiCure’s system, a small (N = 28), randomized,
wearable, data-logger for ambulatory recording of phy- parallel-group [23] 12-week study concluded that the
siological and behavioral metrics including heart rate, system accurately monitored the ingestion of anticoagu-
activity, body angle, and time-stamped patient-logged lants in patients with recent ischemic stroke. However,
events, including events signaled by swallowing the the claim of the authors that the system also changed
IEM. FDA noted that the device enabled unattended patient behavior is unjustified, as there was no cross-over
data collection for clinical and research applications, design. Of note, only 28 of the 117 screened patients
and that it wirelessly sends and stores data to an exter- were actually included in the study, thus hampering
nal company computer [19]. In a separate stage, the FDA generalizability of results and suggesting selection bias.
 4

Table 1.
Potential concerns Consequences Potential solutions
Patient types
Competent patients Breaching autonomy Loss of freedom to make own treatment choice Informed consent and freedom to not use tracking system without fear of punishment
W. VAN BIESEN ET AL.

Incorrect medication intake because of Unintended health risks In principle, tracking systems are always disproportionate for these patients
negligence Strict regulation of access to and fees for health insurance and coverage, irrespective of non-compliance risk
Cherry-picking based on profiling Loss of solidarity/healthcare coverage on profiling
Undermining patient-physician relationship Loss of trust potentially resulting in future Strict regulation of freedom of healthcare professionals not to use or prescribe tracking systems
avoidance of seeking healthcare when needed
Non-competent patients Tracking systems become routinely used Extent of intrusion is disproportionate to the Strict regulation of which drugs are proven to be of sufficient benefit and for which conditions to justify the
benefits of the drug use of such intrusive systems
Definition of ‘non-competent’ ‘non-compliant’ may come to be interpreted as Strict regulation of whom should be considered as non-competent to justify breach of autonomy
‘non-competent’ Regular check-ups of the patient
Socially dangerous Incorrect medication intake Damage to society Digital tracking of medication intake only after proper investigation and approval by a competent court
persons Definition of ‘socially dangerous’ Socially dangerous may come to be defined on Strict regulation of which behavior is considered to be socially dangerous
the basis of prejudice
Types of disease conditions/medications
Contagious Incorrect medication intake Spread of infection Digital tracking of medication intake for selected infections
Emergence of resistance to anti-microbials
Medications with Patients do not take medication and sell it on Illegal market for certain medications (opiods, Limitation of tracking to suspect repeat or lengthy prescriptions
a criminal market the black market pain killers, neurotropics)
Data collection and usage
Privacy Identification of individuals as patients, Obligation to disclose which data is collected and how, for how long and where and to which other users it
potentially resulting in loss of prestige, is disclosed in which format (anonymized, aggregated, pseudonymised). This should be done in
credibility, employment, etc. a language understandable to persons with a minimum level of education.
Require clarification and justification of why each data item is collected.
Creation of publicly accessible registers of which data are collected by which company and to which other
users they sell their data
Bias and fairness Profiling based on tracker devices might not be Ascertain fairness of algorithms
equally accurate or reliable in all groups,
leading to inequity
Grouping/profiling Reduced or non-existent access to healthcare for Ascertain fairness of algorithms
certain groups
Financing/regulation/market exclusivity
Direct costs Cost of medication with tracking system Higher cost due to complexity (i.e. the Identify appropriate indications for use of tracking systems based on ethical, legal and cost-effectiveness
requirement for more components than criteria
a conventional pill)
Indirect costs Cost of medication intake with tracking system Significantly higher cost due to evergreening of Modify patent and other regulations to avoid evergreening
patents or other forms of exclusivity
 ACTA CLINICA BELGICA 5

In another RCT [24] involving subjects with schizophrenia, could potentially be limited by the use of a dedicated
the AiCure device-flagged participants having inap- mobile phone providing no location signal to the
propriate dosing parameters. However, one out of three insurer or the healthcare service, and by sufficiently
patients stopped using the device during the 12-week anonymizing data. However even then, with the facial
follow-up period. Based on these limited and low-quality recognition, patients could easily be de-anonymized,
data, AiCure nevertheless claims that its system has been e.g. via Facebook entries by cross-linking datasets.
validated to minimize risk and improve efficiencies across Lastly, patients might be harmed in different ways
the drug development pipeline in future trials and also by system failures. Patients might skip medications if
enables a profiling of patients at higher risk of not follow- their phone incorrectly suggests they were already
ing a drug regimen [25]. In conclusion, for the time being taken, or overdose if the phone did not correctly reg-
the evidence that these digital tracking devices actually ister medication already taken. Also, the patient could
do improve correct medication intake in the long term is be punished for actions incorrectly registered by the
at best uncertain, with data being suggestive that volun- system. It can be seriously questioned whether any
tary acceptance by patients might be rather low. potential benefits justify the infringements of patient
An unresolved issue might arise in cases where autonomy [10,26–29].
a specific remote monitoring system would be linked to
a specific medicinal product and would not be usable for
What is the potential impact on the healthcare
other medicinal products. In these circumstances, issues
professional?
regarding antitrust and competition law might also arise.
However, this is not relevant to the kinds of systems The fact that there is a tool to record the medication
discussed in this paper, as both most likely have the intake of a patient potentially raises an obligation on the
intention to have an as broad application as possible. physician to check up and react if the patient is not
Last, some privacy issues might potentially arise correctly following the prescription. It is unclear whether
when using these devices, and further evaluation is and under which circumstances the physician would be
needed to ensure the use of these devices is in line liable if he/she failed to timely react to such an alert. This
with the recent General Data Protection Regulations raises the problem of data overload, and the question of
(GDPR, EU (2016). Regulation (EU) 2016/679 of the when and how physicians should act [26]. A too low
European Parliament and the Council of 27 April 2016.) alert threshold would result in alert fatigue, whereas
a too high threshold would put into question the very
concept of tracking intake. What the appropriate reac-
Why would a patient accept such a remote
tions by healthcare professionals and the consequences
tracking system?
for the care of other patients not using such tracking
The most essential primary benefit to a competent systems would be, would need to be clearly defined.
patient willing to take the medication could be the Concerns must also arise regarding the confidentiality
reminder to take the pill. However, this could also be and the possibility of the data being used for purposes
achieved by a simple alarm, or a ‘nudge’, and the not agreed to or not known to the physician. Indirectly,
confirmation that the last pill was taken. In the case the system would also allow to monitor the performance
of competent patients, the digital tracking of medica- of the physician. This makes him/her vulnerable to sanc-
tion intake is disproportionately complex and expen- tion, e.g. in a liability action or by the fee-payer (e.g. an
sive. In settings where the patient has to pay himself insurer, a healthcare system, a hospital, or joint practice).
the full or partial cost of the medication, this creates Obligations to use the tracking system may seriously
additional (unnecessary) expenses and potential hamper the quality of the physician–patient relationship
inequity. More importantly, the existence of such sys- and physician autonomy, thus reducing job satisfaction
tems deprives the patient of the ability to not take the and increasing moral distress. Furthermore, since physi-
pill without others knowing, thus harming patient cians’ prescribing practices can be monitored by the
autonomy. Most fundamentally, the mere fact that pharmaceutical companies, this might result in undesir-
the actual intake of medication can be tracked and able pressures being put on the physician to prescribe
shared with third parties, potentially deprives the the more profitable digital pill.
patient from free choice. Insurers might withdraw
healthcare cover unless the patient uses the system.
What is the role of health services and
Furthermore, other patient-related data are harvested
insurance providers?
to no benefit to the patient, which raises various priv-
acy concerns. The problem is further enhanced by The direct financial consequences for healthcare
possible breaches of the security of the data storage payers and society can be immense. For example, the
system, and the risk that the data may be used for estimated US prices for a 30-day supply of aripiprazole
purposes the patient would not agree to, for instance, are: Abilify MyCite $1640; conventional Abilify $940;
marketing or ‘profiling’. In theory, violation of privacy and generic aripiprazole $30.
6 W. VAN BIESEN ET AL.

However, loss of solidarity is another, more insidious outcome, as more compliant patients often have bet-
aspect to be taken into account. Within the framework ter outcomes. However, excluding non-compliant
of value-based healthcare, quality standards for adher- patients from trials seriously hampers the generaliz-
ence to different medications, such as e.g. oral glycemia ability of findings to the real world and results in an
lowering, antihypertensive medications, lipid-lowering inflated expectation of benefits. Furthermore, it ser-
agents, antipsychotic medications, mood stabilizers, and iously reduces the occurrence of ‘informative drop
antiplatelet therapy, have been introduced. To achieve out’. There is potential information in the fact that
the standard, more than 75% of patients of a healthcare patients stop taking a medication, perhaps unpleasant
provider need to obtain at least 80% of the medication side effects, or perhaps perceived lack of effect. As
prescribed to them [13]. The calculations are currently a consequence, the results of the trial will again not
done on the basis of prescription refill records, but be representative of the effects in the real world [32].
digital tracking could be a more efficient system. For
healthcare services or insurers, the benefit is assurance
Benefits for the providers of these monitoring
that pills are taken, and that, if not, the healthcare
systems
service could change its way of ensuring uptake or
simply withdraw insurance cover, thus cherry-picking The producers can make a significant financial profit by
compliant patients. Worse, using patient profiling granting drug-specific exclusive licences to pharma-
based on collected data, such systems could already ceutical manufacturers, by granting institution-wide
ban patients before they enter an insurance program, licences to healthcare systems, or by themselves exclu-
or increase insurance fees, based on their presumed risk sively marketing the combination of a generic drug
of incorrect medication intake. This raises profound con- and the tracking system. In addition, the system provi-
cerns with regard to equity as especially the less literate, ders can financially profit from the secondary uses of
more socially deprived and poorer patients might have the patient data they amass and from any behavioral
such a risk profile. Given that healthcare providers them- predictive analytics they develop [33]. The more data
selves will also be monitored indirectly through their they have on people, the more closely they can target
patients, there is a high risk that these patients’ access to advertisements and the higher their earnings. Sensory
healthcare may be severely reduced or even eliminated. input such as keyboard strokes and voice can be used
Moreover, measures of medication intake surveillance to infer cognition, emotion and mood [34]. The data
might become obligatory with the results included in collected by the tracking system, consisting of beha-
the patient’s electronic health record for sharing among vioral, physiological and therapeutic data, can be
healthcare professionals and insurers. linked to data about the same individuals from many
different sources, including administrative records
from multiple government agencies and commercial
Use of medication intake trackers in clinical
data from other sources, making true anonymization
trials
nearly impossible in reality. Group profiling based on
In the framework of a clinical trial, the benefit of such algorithms can also negatively affect those who
a digital tracker to the trial investigator is reassurance have not given consent to the use of their data in the
that the drug has actually been taken by the patient in first place, as unrepresentativeness can induce serious
the trial, which may be perceived as making the trial bias in the performance of the algorithms [35].
results more trustworthy. In most trials, the higher cost
of the digital pills would be negligible in the context of
Impact on society at large
overall trial costs. However, using a tracking device for
the medication in a clinical trial challenges the inter- For society at large, medication intake-monitoring
pretation of the study results and requires sufficient systems may reduce the risk of the spread of infec-
consideration of transportability and generalizability tion, of increased drug resistance by microbes, and of
issues, as well as the question of how attrition due to drug abuse (through hindering patients from selling
side effects is handled. These serious concerns regard- on drugs prone to criminal abuse). Better health pro-
ing the validity and representativeness of randomized spects can be expected, provided of course that the
controlled trials [30,31], will only be exacerbated by the drugs in question actually improve health. However,
use of pill intake tracking systems. A digital pill offers major societal risks associated with these systems
the trial operator the possibility to select ‘compliant’ include their direct and indirect costs, loss of free-
participants based on characteristics identified by the dom, and loss of solidarity. Direct costs arise from
system providers, to limit the results used for regula- the enormous price differences between the conven-
tory clearance to those from subjects with a high like- tional and the digital pills [26]. Indirect costs however
lihood of actually taking the drug. Companies might also accrue from such monitoring systems as they
do this to speed up completion of trials, as less drop allow commercial players to use legal strategies to
out occurs, and to improve the chances of a positive obtain and extend monopolies covering medical
 ACTA CLINICA BELGICA 7

treatments. This seriously violates the ethical justifica- companies, system-provider companies, insurers, and
tion of the patent system in relation to drugs, namely healthcare systems to benefit, either by increased med-
that patents should encourage the arrival of new ication consumption or exclusion from cover of non-
drugs and new uses of them rather than to extend compliers. Risk profiling will result in cherry-picking of
the high-profit making period granted in relation to patients and loss of healthcare cover for those who
old drugs (evergreening). A correct balance between cannot afford the increased prices or who, for various
allowing innovators to recoup expenses through reasons of vulnerability, are reluctant to surrender their
exclusivity and achieving cost-containment by market right to refuse particular forms of medical treatment.
introduction of generics after this exclusivity period Thus, it should be clear that constraints, beyond the
results in the largest benefit for society. Medication normal ones of safety and efficacy applied by regulatory
intake tracking systems threaten this balance if they agencies, are required. The constraints required seem to
can have legal exclusivity. Therefore, non-exclusive fall into four major categories: type of patient; type of
licences should be available as a matter of law. disease and medication; type of data collection and use;
Otherwise, these ‘innovative’ technologies may jeo- and financial regulations/market exclusivity (see table).
pardize true innovation and only increase costs. To conclude, ‘smart’ pills are only a smart option if their
The introduction of tracking devices may also impact approval is smart with smart goals and smart constraints.
society in other ways. The mere presence of tracking
devices may at some future time curtail our autonomy
and serve to both discipline individual bodies and to Disclosure statement
regulate behaviors deemed to be a threat to society for No potential conflict of interest was reported by the authors.
whatever reason. It is probably no coincidence that the
first drug available in this category is a psychotropic
agent. Furthermore, tracking devices might be used to Funding
achieve optimal disease management through surveil- This work was supported by the Fonds Wetenschappelijk
lance and discipline, with the final aim to reduce the Onderzoek [FWO.OPR.2019.0045.01]; H2020 Marie Skłodowska-
financial burden of non-compliance on healthcare sys- Curie Actions [DLV-794604].
tems, and serve to further marginalize and differentiate
‘at-risk groups’ because of their unwillingness or inabil-
References
ity to conform [36]. Furthermore, there is a risk that
patient profiling based on medication tracking devices [1] Osterberg L, Blaschke T. Adherence to medication.
will lead to cherry-picking by healthcare providers and N Engl J Med. 2005;353:487–497.
insurers, with only low-risk patients having access to [2] Klugman CM, Dunn LB, Schwartz J, et al. The ethics of
smart pills and self-acting devices: autonomy,
health insurance and healthcare. There is thus a need truth-telling, and trust at the dawn of digital
for a strict and close regulation and/or oversight by an medicine. Am J Bioeth. 2018;18:38–47.
independent body, in order to protect patient and [3] [cited 2019 Sept 12]. Available from https://aicure.
public safety and best interests. Such a regulation com/
should ensure that the technology can be used opti- [4] Cosgrove L, Cristea IA, Shaughnessy AF, et al. Digital
aripiprazole or digital evergreening? A systematic
mally in situations where it could be of benefit, and at
review of the evidence and its dissemination in the
the same time safeguard against its use in conditions scientific literature and in the media. BMJ Evid Based
where it would be negative or even detrimental. Med. 2019.
Exploration on how this can be most optimally orga- [5] Sim I. Mobile devices and health. N Engl J Med.
nized should be the subject of further research. 2019;381:956–968.
[6] Chip in the pill. 2011. [cited 2019 Sept 12]. Available
from: https://nonadherence.blogspot.com/2011/01/
Concluding remarks [7] Proteus Announces Issuance of U.S. Patent for
Ingestible Digital Devices. [cited 2011 Sept 10].
Systems for remotely tracking medication intake may Available from: https://www.proteus.com/press-
well function safely from a technical point of view. releases/proteus-announces-issuance-of-u-s-patent-
However, their added value for patients and society at for-ingestible-digital-devices-2
[8] Speciality product launch insights. 2018. [cited 2019
large is unclear at best. There will be a range of circum- Sept 12]. Available from http://www.pharmexec.com/
stances at one extreme where the additional expense is specialty-product-launch-insights
warranted and the decrease in autonomy/agency can [9] Proteus Digital Health Announces Digital Medicines
be tolerated provided that appropriate controls are in Pipeline Development and Expansion into Oncology.
place. These controls should ensure the balance 2018. [cited 2019 Sept 10]. Available from https://
www.proteus.com/press-releases/proteus-digital-
between beneficial effects and the concomitant nega-
health-announces-digital-medicines-pipeline-
tive effects. Trust in the patient-physician relation will development-and-expansion-into-oncology
suffer. Introduction of these systems risks increasing [10] Belluck P First digital pill approved about biomedical
costs for society as a whole, while enabling drug “big brother”. New York Times 2017.
8 W. VAN BIESEN ET AL.

[11] Applying AICure’s artificial intelligence to healthcare [24] Bain EE, Shafner L, Walling DP, et al. Use of a novel
has the potential to impact millions of lives on a global artificial intelligence platform on mobile devices to
scale. 2019. [cited 2019 Sept 12]. Available from: assess dosing compliance in a phase 2 clinical trial in
https://aicure.com/neurobo-selects-aicure-advance- subjects with schizophrenia. JMIR Mhealth Uhealth.
pivotal-neuropathic-pain-clinical-trial/ 2017;5:e18.
[12] Kronish IM, Moise N. In search of a “Magic Pill” for [25] AICure in Final Stages of Phase II Clinical Trial for
medication nonadherence. JAMA Intern Med. Schizophrenia. 2014. [cited 2019 Sept 10]. Available
2017;177:631–632. from: https://aicure.com/aicure-in-final-stages-of-
[13] Steiner JF. Rethinking adherence. Ann Intern Med. phase-ii-clinical-trial-for-schizophrenia
2012;157:580–585. [26] Rowland C This $1650 pill will tell your doctors
[14] Chakrabarti S. What’s in a name? Compliance, adher- whether you’ve taken it. Is it the future of medicine?
ence and concordance in chronic psychiatric disorders. 2019. Available from https://www.washingtonpost.
World J Psychiatry. 2014;4:30–36. com/business/economy/this-1650-pill-will-tell-your-
[15] Marcum ZA, Sevick MA, Handler SM. Medication non- doctors-whether-youve-taken-it-is-it-the-future-of-
adherence: a diagnosable and treatable medical medicine/2019/04/28/393281b2-4c10-11e9-b79a
condition. JAMA. 2013;309:2105–2106. -961983b7e0cd_story.html?noredirect=on&utm_
[16] Nair KM, Levine MA, Lohfeld LH, et al. “I take what term=.d6adcd445cdc
I think works for me”: a qualitative study to explore [27] Tomlinson T. Getting off the leash. Am J Bioeth.
patient perception of diabetes treatment benefits and 2018;18:48–49.
risks. Can J Clin Pharmacol. 2007;14:e251–9. [28] Terrasse M, Sisti DA. Policing compliance: digital med-
[17] Letter from FDA to Proteus Biomedical. 2012. [cited icine and criminal justice-involved persons. Am
2019 Sept 12]. Available from: https://www.access J Bioeth. 2018;18:57–58.
data.fda.gov/cdrh_docs/pdf11/K113070.pdf [29] Carter J, Bababekov YJ, Majmudar MD. Training for our
[18] Evaluation of automatic class III designation (de novo) digital future: a human-centered design approach to
for Proteus personal monitor including ingestion graduate medical education for aspiring
event marker. 2012. [cited 2019 Sept 12]. Available clinician-innovators. NPJ Digit Med. 2018;1:26.
from: https://www.accessdata.fda.gov/cdrh_docs/ [30] Nagler E ERBP Guideline development methodology:
reviews/K113070.pdf towards the best possible guidelines. Nephrology, dia-
[19] European Medical Agency: Qualification opinion on lysis, transplantation: official publication of the
ingestible sensor system for medication adherence as European Dialysis and Transplant Association -
biomarker for measuring patient adherence to medi- European Renal Association 2013.
cation in clinical trials. [cited 2016 Sept 12]. Available [31] Rothwell PM. Factors that can affect the external valid-
from: https://www.ema.europa.eu/en/documents/reg ity of randomised controlled trials. PLoS Clin Trials.
ulatory-procedural-guideline/qualification-opinion- 2006;1:e9.
ingestible-sensor-system-medication-adherence- [32] Schepkens H, Vanholder R, Billiouw JM, et al. Life-
biomarker-measuring-patient_en.pdf threatening hyperkalemia during combined therapy
[20] FDA. 2017. [cited 2019 Sept 10]. Available from: with angiotensin-converting enzyme inhibitors and
https://www.accessdata.fda.gov/drugsatfda_docs/ spironolactone: an analysis of 25 cases. Am J Med.
nda/2017/207202Orig1s000Lbl.pdf 2001;110:438–441.
[21] Frias J, Virdi N, Raja P, et al. Effectiveness of digital [33] Howard H, Sterckx S, Cockbain J, et al. The conver-
medicines to improve clinical outcomes in patients gence of direct-to-consumer genetic testing compa-
with uncontrolled hypertension and type 2 diabetes: nies and biobanking activities: the example of
prospective, open-label, cluster-randomized pilot clin- 23andme. In: Wienroth M, Rodrigues E, editors.
ical trial. J Med Internet Res. 2017;19:e246. Knowing new biotechnologies. social aspects of tech-
[22] A Novel Wireless Ingestible Sensor System for nological convergence. London, Routledge; 2015. p.
Measurement of Medication Adherence in HIV 59–74.
Treatment and Prevention. 2015. [cited 2019 Sept [34] Insel TR. Digital phenotyping: technology for a new
16]. Available from: http://grantome.com/grant/NIH/ science of behavior. JAMA. 2017;318:1215–1216.
R01-MH110057-04 [35] Courtland R. Bias detectives: the researchers striving to
[23] Labovitz DL, Shafner L, Reyes Gil M, et al. Using artifi- make algorithms fair. Nature. 2018;558:357–360.
cial intelligence to reduce the risk of nonadherence in [36] Gagnon M, Jacob JD, Guta A. Treatment adherence
patients on anticoagulation therapy. Stroke. redefined: a critical analysis of technotherapeutics.
2017;48:1416–1419. Nurs Inq. 2013;20:60–70.