Content uploaded by Klaus Fuchs
Author content
All content in this area was uploaded by Klaus Fuchs on Jun 18, 2020
Content may be subject to copyright.
Association for Information Systems Association for Information Systems
AIS Electronic Library (AISeL) AIS Electronic Library (AISeL)
Research Papers ECIS 2020 Proceedings
6-15-2020
Icon-based Digital Food Allergen Labels for Complementation of Icon-based Digital Food Allergen Labels for Complementation of
Text-based Declaration Text-based Declaration
Klaus L. Fuchs
ETH Zurich
, fuchsk@ethz.ch
Mirella Haldimann
ETH Zurich
, mirella.haldimann@autoidlabs.ch
Michael Zeltner
University of St. Gallen (HSG)
, michael.zeltner@student.unisg.ch
Alexander Ilic
ETH Zurich
, ailic@ethz.ch
Follow this and additional works at: https://aisel.aisnet.org/ecis2020_rp
Recommended Citation Recommended Citation
Fuchs, Klaus L.; Haldimann, Mirella; Zeltner, Michael; and Ilic, Alexander, "Icon-based Digital Food Allergen
Labels for Complementation of Text-based Declaration" (2020).
Research Papers
. 111.
https://aisel.aisnet.org/ecis2020_rp/111
This material is brought to you by the ECIS 2020 Proceedings at AIS Electronic Library (AISeL). It has been
accepted for inclusion in Research Papers by an authorized administrator of AIS Electronic Library (AISeL). For
more information, please contact elibrary@aisnet.org.
Fuchs et al. / Towards Digital Food Allergen Labels
. 1
ICON-BASED DIGITAL FOOD ALLERGEN LABELS FOR
COMPLEMENTATION OF TEXT-BASED DECLARATIONS
Research paper
Fuchs, Klaus, ETH Zurich, Zurich, Switzerland, klaus.fuchs@autoidlabs.ch
Haldimann, Mirella, ETH Zurich, Zurich, Switzerland, mirella.haldimann@autoidlabs.ch
Zeltner, Michael, University of St. Gallen, St. Gallen, Switzerland, m.zeltner@autoidlabs.ch
Ilic, Alexander, ETH Zurich, Zurich, Switzerland, ailic@ethz.ch
Abstract
Food allergen declarations on packaged products lack coherent regulation, standardisation and
therefore manifest in a variety of terminology. This unfortunately leads to a magnitude of preventable
allergy-related outbreaks and emergencies, as consumers rely on allergen declaration for self-
management. Together with dieticians, we therefore developed a purchase-related barcode scanning
mHealth application that utilises a standardised taxonomy of food allergens for the display of user-
friendly digital food allergen labels. The application design allowed for an in-the-wild randomised
controlled trial, randomly attributing users to two treatment groups that either received icon-based or
text-based digital food allergen labels. 74 users met the eligibility criteria for our study. The findings
suggest that users in both groups use the application similarly frequently, but users perceive icon-
based allergen labels more recommendable and more useful than text-based labels. Especially, users
with low fluency in the local language (German), and frequent travellers, both segments prone to food
allergy related emergencies, prefer the icon-based label, indicating that standardised, icon-based
digital food allergen declarations can complement current text-based declarations. Especially users
who are travelling internationally or are non-fluent in their local language can benefit from language-
agnostic icon-based digital food allergen labels.
Keywords: Food allergy, mHealth, visualisation, digital food allergy label, nutrition.
1 Motivation
The rate of population affected by food allergies is rising globally, constituting an increasing burden
on health-care systems and patients alike. Today, the prevalence of food allergies affects an estimated
10% of population (Sicherer and Sampson, 2018, 2014). While the exact drivers of this development
are still not fully understood, suspected factors include consumption of processed foods, surge in diet-
related diseases, increased hygiene, as well as environmental and (epi-)genetic interactions (Sicherer
and Sampson, 2018, 2014). Affected patients experience adversarial health responses mediated by
their immune system, resulting in hypersensitive reactions to certain kinds of foods. Such post-
exposure onsets depend very much on individual level of sensitivity and include rashes, hives, itching,
swelling, diarrhoea, nausea, running or congested noses, vomiting, as well as more serious and poten-
tially life-threatening consequences such as wheezing, cyanosis, fainting, anaphylactic shocks
(Sicherer and Sampson, 2018, 2014; Umasunthar et al., 2015). Within the United States (US) alone, it
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 2
is estimated that food allergies account for annual healthcare costs of over 25 billion USD (Gupta et
al., 2013), and cause roughly 30’000) emergency room visits and 150 deaths per year (FDA, 2017).
In absence of a cure, effective management of food allergies usually requires strict avoidance of rele-
vant allergens (Sicherer and Sampson, 2018), often hindered by non-standardized food labels (Versluis
et al., 2015). Although future therapies might be able to mitigate or even reserve food allergies, current
best practice today is patient compliance with strict avoidance of food items containing allergens
(Sicherer and Sampson, 2018). Unfortunately, when making food choices, consumers are confronted
with non-regulated allergen declaration on food packages (Sicherer and Sampson, 2018). While con-
sumers believe that allergen declaration is regulated, it is in fact subject to far less regulation than ex-
pected (Sicherer and Sampson, 2018). Especially the precautionary allergen labelling (PAL)
(Marchisotto et al., 2017) is voluntary and has lots of varying definitions, terminologies and no thresh-
old amounts (Marchisotto et al., 2017; Sicherer and Sampson, 2018; Versluis et al., 2015), leaving
customers confused and researchers calling for a global definition of food allergens (Allen et al.,
2014). Nevertheless, geographic regions still differ in their mandates to declare allergen presence in
food: For example, while the European Union (EU) (EU-1169/2011, 2014) requires the declaration of
14 different allergens, the US only requires declaration of eight allergens (FDA, 2017). The varying,
non-standardized implementations of allergen declarations on packaged foods represent one of the
largest drivers of food allergy related emergencies which are especially difficult to manage when hap-
pening during consumption of packaged foods at home (Versluis et al., 2015).
2 Related Work
2.1 Purchase-related mHealth
Mobile Health (mHealth) offers the opportunity for pervasive yet context adaptive health behavior
interventions (Goldstein et al., 2017). Research on food allergy related mHealth primarily focusses on
patients under ongoing medical supervision by a physician, e.g. in de-sensitization therapies. Such
mHealth focus on the initial diagnosis and diagnostic refinement of food allergies (Waibel et al.,
2019). However, with only few exception assessing products for a single food allergy (e.g. gluten)
(Dunford and Neal, 2017), or limited technical prototypes (Frey et al., 2016), such purchase-related
mHealth apps supporting allergy affected patients in food purchasing are not yet discussed in detail.
This is counter-intuitive, as similar purchase-related mHealth aimed at improving purchase decisions
based on the assessment of macro-nutrients have been already been discussed and adopted by consum-
ers (Dunford et al., 2014; Mhurchu et al., 2019; Volkova et al., 2016). Front-of-package labels (FoPL),
food labels in general, and in turn also standardized allergen labels are hard to regulate and implement
for food producers, brands and retails (Dunford et al., 2014; Julia et al., 2017; Volkova et al., 2016).
Food producers fear the costs, administrative burden, potential loss of revenue upon introduction of
novel labelling standards. In addition, production processes, product lifetime cycles and potentially
competing standards hinder the implementation of globally standardized allergen labels (Allen et al.,
2014). Nonetheless, since most packaged products within a retail environment carry a product-specific
barcode identifier (Brock, 2001) and most consumers carry a smartphone (Dunford et al., 2014), pur-
chase-related mHealth applications are a promising channel for informing users on present allergens
after scanning a barcode of a product. Such purchase-related mHealth for pre-consumption detection
of present allergens offers advantages for consumers. First, mHealth is inclusive, since most packaged
food products carry a product-specific barcode (Brock, 2001), and most consumers nowadays carry a
smartphone (Dunford et al., 2014). Second, in contrast to existing allergy-related mHealth that usually
focus on interpreting or treating symptoms of food allergies, such mHealth can provide relevant warn-
ings prior to consumption, especially relevant to preventing anaphylactic shocks at home (Versluis et
al., 2015). Third, mHealth can customize the language to the user, thereby overcoming current lan-
guage barriers of travellers, remote workers or illiterate citizens towards interpretation. This is espe-
cially relevant, since food allergies also affect socio-demographics segments prone to illiteracy, who
usually do not have access to diagnostics or treatment (Sicherer and Sampson, 2018). Fourth, mHealth
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 3
can display food allergens in standardized forms, potentially even in accordance to a future, global
standard framework (Allen et al., 2014). Fifth, mHealth can extend the scope and include allergens
currently not mandated by regulators. This is especially relevant, given that at least 170 different food
allergies and triggers exist (Boyce et al., 2010). Last but not least, mHealth can leverage tailoring
(Brug et al., 2003) and produce user-specific warnings to reduce the impact of false or vague warnings
of possible allergen contamination that are often unhelpful and can contribute to unintentional expo-
sure (Allen et al., 2014; Turner et al., 2011).
Since current allergen declarations use varying terminology, non-standardized items, ill-defined pre-
cautionary statements and require above-average literacy to be understood, researchers and practition-
ers are calling for an easier-to-understand labels (Declaratio, 2019), and a global definition framework
(Allen et al., 2014). Therefore, we designed and implemented an application by the name “Allergy-
Scan” and conducted a randomized controlled trial (RCT) to assess design choices for such a system.
Specifically, we demonstrate feasibility of mHealth to visualize food product allergen declarations in a
novel, i.e. standardized nomenclature of icon-based digital food allergen labels. Further, this study
aims to assess whether such language-agnostic icon-based digital food labels can lead to improve us-
age behavior and intention in comparison to current text-based declarations. We thus address the fol-
lowing research question in this study: do digital food allergen labels improve usage behaviour and
intention to use over contemporary text-based allergen labels in mHealth applications?
3 System Design
Similar to many mHealth apps, we relied on a design science approach (Gregor and Hevner, 2013;
Hevner et al., 2004; Peffers et al., 2008), which were also applied in similar mHealth research (Fuchs
et al., 2016). We hence focus on describing the application design in detail.
3.1 Application Design
We built on previous purchase-related mHealth studies (Dunford et al., 2014; Mhurchu et al., 2019;
Volkova et al., 2016) and hence designed and implemented a barcode-scanning mobile application to
provide users with digital food allergen labels to collect data on usage behavior within the automated
RCT. The application was developed specifically for release within Switzerland. We argue, that the
focus on Switzerland does not compromise generalizability to other regions for multiple reasons. First,
similar to other developed countries, also Switzerland shows increasing prevalence of diet-related dis-
eases and food allergies which current empirical studies indicate to affect a population rate of 6%
(Aha!, 2019). Second, the study team had access to the largest curated product database in Switzerland
(GS1-CH, 2018) and extended the product coverage through crowdsourcing, automatic and manual
data quality checks, thereby ensuring high data quality product data on nutrients, allergens and ingre-
dients. Similarly, also other regions offer food composition databases as mandated the European regu-
lation on public provision of food composition data (EU-1169/2011, 2014). In total, over 52,750 of the
most-frequently purchased food products in Switzerland were available in the study’s food product
composition database. Third, the study was supported by the Swiss Society for Nutrition (SGE-SSN)
with their dietary expertise, with similar national associations also existing in other countries, enabling
the adoption of the mHealth intervention towards local customs and cultural aspects. Throughout sev-
eral iterations with dietary experts from SGE-SSN and multiple rounds of user testing, the final design
of the “AllergyScan” was achieved. The app was then published in the Swiss Apple and Android app
stores. In the following, its most important functions as identified by users or experts are introduced.
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 4
Figure 1. AllergyScan user flow (1.1 home screen, 1.2 user profile, 1.3 scanning, 1.4 scan log)
3.1.1 Barcode Scanning
The app allows barcode scanning by activating the smartphone-integrated camera to capture a prod-
uct’s barcode-encoded global trade item number (GTIN) (Figure 1.3). After having set up a personal
profile (Figure 1.2), users can easily access the barcode scanning functionality from the application’s
home screen (Figure 1.1). After scanning and given that the product exists in the study database, users
receive a digital food allergen label, according to their allocation in the RCT.
3.1.2 Display of Product Images
The app displays product details including a product image after scanning a product (Figure 2). This is
important for visual confirmation, a key function in the search-phase of purchasing decisions, espe-
cially in supermarkets (Clement et al., 2013).
3.1.3 Inclusion of Weighted Products
Products that are weighed before purchasing (e.g. fruit, vegetables, meat, etc.) can carry price-specific
barcodes. Conventional barcode-scanning mHealth apps do not correctly identify these products, due
to their often uniquely generated barcode. We therefore programmed a reverse-mapping to identify the
corresponding base product (2110085000005 = ‘Le Gruyère Cheese’) of any weighted product (e.g.
barcode 2110085004959 identifies 4.95 Swiss Francs of ‘Le Gruyère Cheese’). This feature appears
important, since weighted products should not be neglected, e.g. cheeses, meat, fish, etc.
3.1.4 Tailoring
Through tailoring, mHealth can adapt interventions to the individual user (Nahum-shani et al., 2014).
Especially given the plethora of potential food allergy triggers not captured by current labelling prac-
tice (Boyce et al., 2010), a filtering mechanism to not differentiate relevant from irrelevant warnings is
needed. Therefore, we allowed users to select if they want to receive warnings for all allergens present
in a food item or just relevant ones for their individual profile. Accordingly, users either see i) tailored,
i.e. when a user’s allergy profile and a food item’s allergens conflict (Figure 2.1 and Figure 2.2) or ii)
non-tailored warnings of all present allergens (Figure 2.3 and Figure 2.4). Respectively, the label’s
disclaimer statement differed: Tailored allergen labels stated ‘This result is based on your allergy pro-
file. People with a different allergy profile cannot rely on it.’ The static labels stated, ‘This information
is provided by the manufacturer's food package and complies with the legal requirements’.
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 5
3.1.5 Usability & Learning
To support users in familiarizing themselves with the app, an introductory tutorial was included. Fur-
ther, to foster nutritional literacy, the app included educative tips regarding a scanned product’s cate-
gory or nutrients. In total, 130 of such texts and icons were produced by the Swiss Society for Nutri-
tion and shown while loading the list of product substitutes including images from the server.
Figure 2. Digital food allergen labels were icon-based or text-based. Users could select tailored
or static warnings. 2.1 icon-tailored, 2.2 text-tailored, 2.3 icon-static, 2.4. text-static.
3.1.6 Convenience
To support users in identifying healthier choices during shopping, the application allows storing dis-
covered products within an easy-to-access scan log list. From there, a user can quickly access the
product details of each stored product in order to check the digital food allergen label (Figure 1.4).
3.1.7 Error Reporting and Crowdsourcing
To continuously improve data quality, users of the ‘AllergyScan’ app were able to report erroneous
data entries. Further, the application would also automatically collect errors, if a product data entry
appears to contain errors by applying pre-configured checks (e.g. if an image is present, texts are well
defined, ingredients and allergens seem compliant). Also, to collect information on products that were
previously not yet in the study database, users were invited to add pictures of the front and back of
each product, an established approach in purchase-related mHealth applications (Codecheck, 2016;
Dunford et al., 2014; Gigandet, 2019). Research team members would then manually review the col-
lected errors and crowdsourced pictures to implement the necessary changes in the database.
3.1.8 Blocking of Product Categories
To reduce processing of irrelevant items or because of lacking data quality, we blocked certain food
categories (medicine, beauty and healthcare products, tobacco, alcohol) or special food categories that
ideally require advice from health-care professionals (e.g. baby food, supplements).
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 6
4 Methodology
4.1 Icon versus Text-based Allergen Labels
In this study, we want demonstrate feasibility of mHealth to visualize food product allergen declara-
tions in a novel, i.e. standardized nomenclature of icon-based digital food allergen labels and assess
whether such language-agnostic icon-based labels can lead to improve usage behavior and intention in
comparison to text-based declarations. mHealth can complement or even substitute the current printed,
text-based allergen declarations on packaged products through display of alternative icons that follow
intuitive, easy-to-understand, purpose-driven design. First, consumer confusion caused by varying
terminology and precautionary allergen statements could be mitigated. Further, such icons are promis-
ing to be understood by users unable to understand a local language e.g. travellers, migrants, or less
educated citizens. This is relevant, as food allergy related emergencies often affect socio-
demographics segments prone to illiteracy with less access to treatment (Sicherer and Sampson, 2018).
4.2 Participant Recruitment
The application was made available to smartphone users aged 16 years or older in Switzerland via the
iOS and Android app stores. Between November 2018 and March 2020, in total 1641 users installed
the AllergyScan application. There was no monetary incentive for users to participate, nor were any
promotions used for the study. To observe natural behavior among consumers, users did not receive a
study protocol and were free and able to use the app as they wanted.
4.3 Randomized Controlled Trial
In intervention delivery mode, food allergen labels were displayed for products, if matched successful-
ly with an existing product in the database (the rate of successful scans was approximately 72%). To
explore the impact of icon-based labels on user behavior, performance expectation and usage inten-
tion, we therefore randomly assigned users with either i) language-agnostic, intuitive icon-based labels
(‘icons’) (treatment group (TG)) or ii) text-based allergen labels in German (‘text’) (control group
(CG)) (as required by local regulation in Switzerland, i.e. one of the official languages). Users were
randomly assigned to receive either one of the two visualizations with a probability of 50%. Besides
overcoming potential language barriers, the icons also replace the currently ill-defined PAL statements
with a simple two stage color scheme. Via text mining, the current text fragments were transposed into
either ‘orange’ labels (e.g. trays of allergens may be expected) and ‘red’ labels (explicitly declared
presence). The icon-based labels follow a standardized nomenclature to mitigate varying terminology
for the allergens (e.g. milk=lactose or wheat=gluten). After randomly allocating users to the two inter-
vention types, the study protocol assessed the impact of icon-based labels on mHealth usage rates, per-
formance expectancy and usage intention. All users’ assignment to their allocation group were static
and could not be changed on a device. Generating the digital food allergen labels within the app relied
on each product’s corresponding allergen composition in the food composition database, interpreted
by the systematic generation of the food allergen labels (see System design).
4.4 Eligibility Checks and Data Collection
The system’s functionalities included automation of the RCT with randomized user segmentation, eli-
gibility screening, intervention delivery, data collection facilitation. The app also included duplicate
checks, automatic in-app logic checks, collection of informed consent, questionnaire administration,
outcome data collection from usage behavior and survey completion. To gather meaningful results,
only eligible users were invited to participate in the study. Therefore, eligible users were required to be
affected by at least one food allergy (self-reported within the user profile in the application). Further,
users were only invited to participate in the study after a minimum usage of eight product scans on at
least four separate days. In addition, the minimum number of successful scans that were found on the
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 7
study server was set to at least four products. Last but not least, the users that complied with the eligi-
bility criteria were invited via notification in-app to opt into the non-mandatory survey to participate in
the study. In total, 74 eligible users completed the survey and opted-in to share their usage statistics.
The participation rate of 4% indicates an acceptable response rate for public, non-requested invites to
surveys, such as e-mail marketing which often have response rates between 1-5%. In data collection
mode, eligible users were invited to share their profile, usage statistics and survey data. The following
ethical and security requirements were adhered to: (1) privacy-by-design: all collected data were
anonymized on device. No locations or addresses were stored to avoid processing of personalizable
data. (2) participant information statement was made available via app store and link.
5 Results
First, we included a descriptive analysis to describe the study sample. Second, we examined usage
behavior across the two treatment groups (number of scans and therefore usage behavior of the digital
allergen label). Third, we assessed the self-reported survey data. Specifically, self-reported intention-
to-use (four items), as well as performance expectation (three items) and effort expectancy between
the treatment groups via a one-way ANOVA. We performed a Tukey's honestly significant difference
(HSD) post hoc test for items found significant. To prevent confounding of the comparisons, we split
the data set into cross-sectional subsegments and compared socio-demographics prone to food allergy
emergencies accordingly (i.e. users who are non-fluent in German, or who frequently travelling abroad
to countries where they do not speak the language, or who suffer under multiple food allergies).
5.1 Sample Description
Variable
Type
N (%)
Variable
Type
N (%)
Gender
Male
16 (22%)
Age
16-29
32 (43%)
Female
48 (65%)
30-39
14 (19%)
Not stated
10 (14%)
40-49
10 (14%)
International
travel
Frequent
30 (41%)
50+
8 (11%)
Occasionally
33 (45%)
Not stated
10 (14%)
Rarely
11 (15%)
Checking prod-
uct ingredients
Frequently
61 (82%)
Number of food
allergies
One
47 (64%)
Occasionally
13 (18%)
Two
16 (22%)
Local language
skills
Fluent
63 (85%)
Three or more
11 (15%)
Non fluent
11 (15%)
Table 1. Baseline characteristics of study participants (N=74).
Between November 2018 and March 2020, 1641 users installed the ‘AllergyScan’ app, of which an
acceptable 4.5% met the eligibility criteria and also responded to the non-mandatory study survey (N =
74). Participants in the study had a mean age of 29.97 (SD = 12.66) years. The relatively young age
distribution is typical for mHealth studies and does not prevent generalizability. Also, this conven-
ience sample included more females than males, which is a frequent self-selection phenomenon in di-
et-related studies (Mhurchu et al., 2019). Among the study participants, most frequent food allergies
included gluten (N=56, 76%), followed by eggs (N=12, 16%) celery (N=8, 11%), crustaceans, milk
and peanuts (each N=7, 9%). This distribution is also in line with publicly available reports on self-
declared allergens and intolerances (Aha!, 2019). Most eligible users suffer under one food allergy
(N=47, 64%), while roughly one third (N=27, 36%) suffer under at least two food allergies in combi-
nation. The majority (82%) states that they have to manually read food items’ allergen declarations at
least once a week during their purchasing decisions. Regarding international travel, most (N=33, 45%)
spend between one and two weeks a year in a country whose language they do not speak. Contrary,
similarly many users spend more time abroad internationally (N=30, 41%), and less users stay in
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 8
Switzerland for at least 51 weeks a year (N=11, 15%). Among the sample, most users (N=63, 85%)
are either native or fluent in German, while a minority (N=11, 15%) only speaks basic or no German
at all.
5.2 Usage Statistics
App usage
All users
(NTG=32, NCG=42)
Frequent travellers
(NTG=12, NCG=18)
Not fluent in local language
(NTG=5, NCG=6)
All users
25.95 (21.66)
26.40 (20.02)
23.27 (20.09)
TG: Icon-based
27.44 (18.84)
26.83 (11.70)
35.80 (25.23)
CG: Text-based
24.81 (23.75)
26.11 (21.92)
12.83 (3.13)
DTG-CG
2.63
0.72
22.97
P
.61
.92
0.052
Table 2. Usage behavior: Mean (Standard deviation) of usage events, i.e. barcode scans
Between installing the application, and over the course of the at least eight-day long study period dur-
ing which users had to use the application on at least four different days, and until the time point com-
pleting the survey, participants scanned on average 26 products (Table 2). In total, 1920 scans were
conducted. One user even scanned 108 products, indicating that such a system can have a high value
during the food making process for an individual. This user was in the text-based group and has a high
impact in the assessment of the overall, average usage statistics, but was not removed as he passed the
survey including manipulation check. A paired t-test across the two treatment groups (CG: text-based,
TG: icon-based) showed no significant differences in effect on scanning usage. However, it can be
seen that on average, the users with the icon-based digital food label show the tendency to scan 11%
more products than users with the text-based declaration. Similarly, there was no significant difference
between frequent travellers in usage of icon-based or text-based labels. Still, again there was the ten-
dency observable that icon-based digital food labels were used more than text-based declarations. Fi-
nally, albeit at a small sample size, there was a big difference in label usage between users with no or
only basic levels of German. Concretely, users with a low proficiency in German scanned over 53%
more products in the icon-based group, compared to low proficient users in the text-based control
group. Given the small sample size of the hard-to-recruit citizens with a low proficiency who have at
least one food allergy, the difference in usage can be considered significant (albeit only at a p-value of
5.2%).
5.3 Usage Antecedents
In the following table (Table 3), we compared the self-reported survey items across three usage ante-
cedents between treatment group (icon-based label) and control group (text-based label), namely in-
tention to use, effort expectancy and performance expectancy. We highlighted the respective group
that was perceived superior on the five-item Likert scale in bold.
Overall, the icon-based treatment group was evaluated with relatively higher ratings for all nine out of
nine dimensions, with all ratings being higher than neutral (i.e. three on the five-level Likert scale). A
two-sided t-test reveals only few significances between the two groups. Significant differences were
observable regarding the intention to use and some aspects of performance expectancies (albeit only at
a p-value of below 10%). Specifically, the intention to use the app in the future is perceived signifi-
cantly higher within the icon-based treatment group (p=0.025). Overall, the results indicate that icons
labels seem to have a lot of non-significant and some significant positive impacts on intention to use
and performance expectancy when compared to text-based labels. Similarly, albeit weaker, icon-based
labels also seem to have a positive impact on at least some aspects of effort expectancy.
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 9
Usage antecedents
All users
Frequent travellers
Low fluency
TG
(Icon)
N=32
CG
(Text)
N=42
P
TG
(Icon)
N=12
CG
(Text)
N=18
P
TG
(Icon)
N=5
CG
(Text)
N=6
P
Intention to use A
4.08
(0.91)
3.56
(1.03)
0.025
4.28
(0.75)
3.70
(1.14)
.14
4.33
(0.97)
3.33
(0.99)
.13
I would recommend
the application to
friends.
4.16
(1.05)
3.62
(1.15)
0.042
4.25
(0.97)
3.61
(1.20)
.13
4.40
(1.34)
3.50
(1.22)
.27
I will keep using the
app in future for reg-
ular purchases.
4.25
(0.91)
3.79
(1.09)
0.057
4.25
(0.62)
3.94
(1.11)
.39
4.20
(0.83)
3.50
(0.83)
.20
I will keep using the
app in future during
travelling abroad.
3.84
(1.22)
3.26
(1.34)
0.059
4.33
(0.98)
3.56
(1.42)
.11
4.40
(0.89)
3.00
(1.26)
0.068
Performance expec-
tancy A
3.93
(0.82)
3.57
(1.01)
0.09
4.17
(0.63)
3.67
(1.10)
.17
4.53
(0.51)
3.83
(0.72)
.10
The app supports me
in identifying prod-
ucts that I am allergic
to.
4.00
(0.92)
3.62
(1.01)
0.099
4.25
(0.62)
3.67
(1.12)
.13
4.60
(0.55)
3.67
(0.82)
.058
The app saves time,
as I do not need to
check the list of in-
gredients for poten-
tial allergens myself.
3.91
(1.06)
3.45
(1.19)
0.093
4.25
(0.87)
3.50
(1.34)
0.098
4.60
(0.55)
3.50
(1.05)
.065
I like the application
and rate it highly.
3.91
(0.77)
3.64
(1.28)
.31
4.00
(0.74)
3.83
(1.29)
.69
4.40
(0.55)
4.33
(1.03)
.90
Effort expectancy A
3.85
(0.88)
3.68
(0.89)
.41
3.92
(0.78)
3.68
(1.04)
.42
3.73
(1.09)
3.67
(0.91)
.92
The app is conven-
ient to use for detec-
tion of allergens in a
product.
3.75
(0.62)
3.45
(1.06)
.26
3.83
(1.03)
3.44
(1.24)
.38
3.80
(1.30)
3.67
(0.82)
.84
The app’s visualiza-
tion format makes it
easy to notice aller-
gens in a product.
3.91
(1.02)
3.79
(1.05)
.62
4.00
(0.95)
3.72
(1.18)
.50
4.20
(1.30)
4.17
(0.98)
.96
Using the app is
worth my time.
3.91
(0.93)
3.81
(0.94)
.66
4.08
(1.08)
3.89
(1.02)
.62
4.20
(1.30)
3.67
(0.82)
.43
Table 2. Likert scale in survey: 1 (strongly disagree) to 5 (strongly agree)
A: equally weighted sums of all three sub-items below
Next, we compare the socio-demographic segments prone to food allergy related emergencies, i.e.
travellers that frequently spend time in countries where they do not speak the local language, as well
as users who are not fluent in the local language. Therefore, both groups are exposed to the risk of ac-
cidently purchasing food items that contain relevant allergens, as text-based declarations in a local
language might not be helpful to them. To prevent confounding, only the relevant cross-sections
among within the icon-based treatment group and text-based control group are selected, leading to
small sample sizes. Concretely, frequent travellers (i.e. spending at least two weeks a year in a country
of which the user does not speak the language of) do not show significant differences between icons
and text-based labels. Still, the tendency to prefer icons is even stronger than in the overall group, as
the absolute ratings as well as the differences between the groups are reported more favourable to-
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 10
wards the icons. Similarly, users who only have no or only basic fluency in German rate, give very
high intention to use ratings in comparison to the German text-based version (albeit not significant,
due to the low sample size).
6 Discussion
In this paper we developed, implemented and assessed a novel icon-based digital food allergen labels
to overcome the existing limitations of contemporary food allergy declaration printed on packaged
products. First, the introduction of icon-based labels that follow a standardized nomenclature instead
of text-based declarations with their varying terminology. Also, the application allowed for a novel
tailoring approach of digital food allergen labels towards the individual consumer instead of declaring
all allergens to all users at all times. This exploratory study was conducted as a fully automated RCT
across 74 users who fulfilled the eligibility criteria, agreed to anonymously sharing their usage behav-
ior and successfully completed the non-mandatory study survey.
Benefits of study protocol include the observable potential of icon-based digital food allergy labels on
usage behavior and the intention to use, effort expectancy and performance expectancy items assessed
in this study, all important usage antecedents towards adopting mHealth applications. Strengths of this
study include its rigorous RCT design with strict user assignment to one of the two treatment groups
and device-based, privacy-preserving duplicate checks, automatic in-app logic checks, collection of
informed consent, questionnaire administration, and outcome data collection from usage behavior and
survey completion. To our knowledge, this study was the first food allergy labelling RCT in a real-
world setting on digital food allergen labels for packaged products in any store across an entire coun-
try. Moreover, the study was able to attract users from different socio-demographic segments, includ-
ing users with multiple food allergies, low proficiency in the local language (i.e. German) as well as
frequent travellers who are often in regions where they do not understand the local language. These
segments are important to cater to, as they are especially prone towards suffering post-consumption
emergencies such as anaphylactic shocks.
Although icon-based labels did not correlate with significantly, but only a tendency towards, higher
application usage when compared to text-based labels, there were some significant in the usage ante-
cedents between icon-based and text-based labels. Concretely, compared to the text-based control
group, intention to use was perceived significantly among the overall group. Especially the self-
reported future usage intention and recommendablility were significantly higher among the icon-based
label group. Further, icon-based food allergy labels scored higher on every single dimension of inten-
tion to use when compared to their text-based counterparts. Also, the values for all dimension of inten-
tion to use, performance expectancy and effort expectancy were significantly higher than neutral (Lik-
ert scale for all items ranked from 1 to 5). Also, the icon-based label was always appreciated more
than the text-based alternative, as the ratings were constantly higher than for the text-based declara-
tion. This indicates that icon-based digital food allergen labels indeed have potential to positively im-
pact intention-to-use. Regarding performance expectancy, the positive impact of icon-based labels is
weaker, but still positive. More specifically, this study shows that especially in terms of supporting the
purchase decision process, icon-based labels perform superior. Regarding effort expectancy, icon-
based labels again perform better than text-based labels, especially in the aspect of convenience. Simi-
larly, the mHealth system scores better for intention to use, effort expectancy and performance expec-
tancy among frequent travellers (i.e. users that spend more than two weeks a year in a country of
which they do not understand the language), as well as users with low proficiency in German (i.e. the
local official language used to declare allergens in text form on packaged products). Albeit not signifi-
cant due to the low sample sizes, as non-fluent users with at least one food allergy are very hard to
recruit, these are promising tendencies that icon-based labels can be helpful for such segments that are
prone to food allergy related emergencies.
In summary, the study demonstrated feasibility of mHealth systems to cater to users with food aller-
gies through a standardized nomenclature, thereby overcoming existing barriers of current allergen
declaration. In addition, there was a significant increase in the intention to recommend the application
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 11
when receiving icon-based labels. In conclusion, even despite the fact that there was no observable
significant difference in actual label usage, the findings suggest the hypothesized potential of improv-
ing at least some important aspects food allergy usage antecedents, especially for users prone to aller-
gy related outbreaks.
The findings of this study ought to be considered with certain limitations. A first key limitation deals
with the fact that we selected individual established concepts which we saw as especially relevant in
the initial stage of a behavioral change intervention. Future study designs could assess users’ compli-
ance and intervention efficacy on actual food selection. The sufficient, yet low sample size (N = 74)
may represent another limitation for external validity and may have played a role in limiting the signif-
icance of some of the statistical comparisons conducted. Also, the self-selection bias towards female
users is limiting the generalization of the gathered insights towards the entire population. The study
protocol was rather strict and excluded users who did not have at least one self-reported food allergy
(given that only 6% of Swiss citizens suffer under food allergies, recruiting affected users can be ra-
ther cost- and effort-intense). It is also worth to mention the low retention rate in this context. As de-
scribed earlier, only 4.5% of all users decided to participate in the optional, non-supervised study. To
increase such figures future studies may thus opt for a more invasive study design, potentially requir-
ing users to complete the survey or to recruit patients supervised by health-care professionals to ensure
compliance with the study protocol. In any case, we recommend optimizing and testing the user con-
version rate for future studies on purchase-related mHealth. Another important limitation deals with
barcode identification. Approximately 72% of barcode scans were successfully identified in the server
backend. Even though, data of 52,750 products were available for the study, there are still thousands
of products missing in order to cover the product universe of the entire country of Switzerland. This
problem is compounded when users frequently buy groceries abroad, as can be the case for users who
live or work in border regions. Practitioners and researchers should be aware of the effort involved in
preparing and maintaining such a country-wide product database. A statistical power analysis can then
be performed to better ensure external validity.
In future works on this field, researchers should assess the concrete needs for users within socio-
demographic segments that are more prone to food allergy related emergencies under the current food
allergy declaration, e.g. illiterate, low-educated users or travellers. Naturally, research should try to
increase the sample size of a future study in order to enable more robust cross-sectional or stratified
sampling for the detection of further significant findings through increased statistical power. In addi-
tion, we suggest including assessment of the degree of nutritional quality and allergen composition of
the scanned products and categories within a future study protocol, to see if icon-based and tailored
labels lead to an increased uptake of healthy allergen-free food products. An interesting opportunity
for future research on purchase-related mHealth also arises from head-mounted devices such as Mi-
crosoft HoloLens or Magic Leap One. Such mixed-reality headsets can enable automatic, passive trig-
gering of icon-based, just-in-time delivered novel digital food allergen labels, as these devices have
always-active front-facing cameras and do not require the active user command to scan for barcodes or
recognize an item via computer vision, without the need of scanning a barcode (Krizhevsky et al.,
2012; Pandey et al., 2017). Especially, since operating a mobile application can be cumbersome while
carrying groceries within a fast-pace supermarket environment, such wearable camera devices repre-
sent a novel opportunity for passive, adaptive just-in-time interventions (Nahum-shani et al., 2015).
The present study adds to the body of research in meaningful ways. First, the icon-based food allergen
labels show positive impact on intention to use and some areas of effort and performance expectancy
when compared to text-based labels, suggesting that visual, easy-to-understand food allergen labels
indeed have potential to improve constructs relevant for mitigation of food allergy related onsets.
Counterintuitively, text-based labels are correlated with higher ratings for trust. This aspect could also
be caused by the introduction of the optional tailoring framework for personalization of digital food
allergen labels which had not been suggested before. But since tailoring is one of the key advantages
of mHealth, researchers, regulators and practitioners should assess the potential negative impacts of
tailoring digital food allergen labels on trust. Therefore, consequentially, future applications should
include both, i.e. a language-agnostic icon as well as a text-based declaration in the language that the
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 12
user can understand. Thereby, the text-based presentation that a user is used to, can be complemented
by the intuitive icon in order to create awareness for relevant allergens present in a food product.
Third, this study shows, that even in absence a global standardization of allergen declaration, mHealth
can produce standardized digital food allergen labels after scanning a barcode. Thereby, as this study
indicates, consumers, especially travellers, illiterate citizens, elderly with reduced eyesight, low-
educated consumers can potentially benefit from improved visualization of mHealth-mediated allergen
declaration in the future. Fourth, also the reverse-mapping of weight-encoded barcodes of this study
represent a valuable extension to current purchase-related and diet-related mHealth. This is especially
relevant, as the reverse-mapping enables barcode-scanning of weighted products, which previously
was not possible due to the uniqueness of their price encoding. Especially since weighted products
include processed products, their impact on diets cannot be ignored. Fifth, this study contributes to the
development of standards on food composition databases such as EuroFIR or LanguaL, which aim to
harmonize terminology on ingredients, nutrients and allergens. Besides the important work on stand-
ardizing databases, this study suggests that the visualization towards consumers can improve usage of
food allergen labels. This study suggests that similar to Nutri-Score (Julia and Hercberg, 2017), intui-
tive icon-based food allergen labels, in digital or potentially also in printed forms, seem to improve
intention to use (especially recommendablility and while travelling) and at least some aspects of per-
formance expectancy. Last but not least, this study does not only yield results for prevention of food
allergy related onsets and emergencies, but its insights are also relevant for researchers and practition-
ers working on mitigation of food intolerances which affect even more consumers (Acker et al., 2017).
As a final note, the introduction and increased usage of novel digital food allergen labels may also rep-
resent an approach to further increasing transparency between products’ ingredients and consumers.
There exists consistent evidence, that the introduction of transparency measures such as FoPL can lead
to healthier product reformulation by manufacturers (Volkova et al., 2016). Therefore, novel icon-
based digital food allergen labelling could have an additional, indirect health-beneficial effect besides
individual improvement of food choices via food reformulation (Volkova et al., 2016), which could
not be tested in the current study.
7 Conclusion
In conclusion, this study identified a promising context for mHealth research to expand to. Since mo-
bile medical apps are changing the way the world and health consumers handle their personal health
care. These applications allow health consumers to track their own health and make their adjustments
according to their lifestyles. In the assessed context of food allergies, this study contributed to identi-
fying a scalable method for the display of digital food allergen labels on users’ mobile devices after
scanning a barcode identifier. The usage statistics and usage antecedents indicate that especially users
who suffer under a language barrier in the local language can benefit from icon-based digital food al-
lergen labels. Further, there exists the tendency that also frequent travellers and all users in general
prefer icon-based digital food labels over contemporary allergen declaration. It remains vital to con-
tinue researching the ways in which such applications must work to prevent food allergy related onsets
and emergencies. Our study contributes to this effort, by examining the promising effect of digital al-
lergen declarations that could complement or even substitute non-standardised text-based declarations.
References
Acker, W.W., Plasek, J.M., Blumenthal, K.G., Lai, K.H., Topaz, M., Seger, D.L., Goss, F.R., Slight,
S.P., Bates, D.W., Zhou, L., 2017. Prevalence of food allergies and intolerances documented in
electronic health records. J. Allergy Clin. Immunol. https://doi.org/10.1016/j.jaci.2017.04.006
Aha!, 2019. Aha! Swiss Allergy Centre: Food allergy report [WWW Document]. Website. URL
https://www.aha.ch/swiss-allergy-centre/info-on-allergies/-allergies-intolerances/food-
allergy/food-allergies/?oid=1582&lang=en (accessed 4.22.19).
Allen, K.J., Turner, P.J., Pawankar, R., Taylor, S., Sicherer, S., Lack, G., Rosario, N., Ebisawa, M.,
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 13
Wong, G., Mills, E.N.C., Beyer, K., Fiocchi, A., Sampson, H.A., 2014. Precautionary labelling
of foods for allergen content: Are we ready for a global framework? World Allergy Organ. J.
https://doi.org/10.1186/1939-4551-7-10
Boyce, J.A., Assa’ad, A., Burks, A.W., Jones, S.M., Sampson, H.A., Wood, R.A., Plaut, M., Cooper,
S.F., Fenton, M.J., Arshad, S.H., Bahna, S.L., Beck, L.A., Byrd-Bredbenner, C., Camargo, C.A.,
Eichenfield, L., Furuta, G.T., Hanifin, J.M., Jones, C., Kraft, M., Levy, B.D., Lieberman, P.,
Luccioli, S., McCall, K.M., Schneider, L.C., Simon, R.A., Simons, F.E.R., Teach, S.J., Yawn,
B.P., Schwaninger, J.M., 2010. Guidelines for the diagnosis and management of food allergy in
the United States: report of the NIAID-sponsored expert panel. J. Allergy Clin. Immunol.
https://doi.org/10.1016/j.jaci.2010.10.007
Brock, D.L., 2001. White Paper: Integrating the Electronic Product Code (EPC) and the Global Trade
Item Number (GTIN). MIT Auto-ID Cent. 1–25.
Brug, J., Oenema, A., Campbell, M., 2003. Past, present, and future of computer-tailored nutrition
education. Am. J. Clin. Nutr. 77, 1028S-1034S.
Clement, J., Kristensen, T., Grønhaug, K., 2013. Understanding consumers’ in-store visual perception:
The influence of package design features on visual attention. J. Retail. Consum. Serv. 20, 234–
239. https://doi.org/10.1016/j.jretconser.2013.01.003
Codecheck, 2016. Codecheck: Produkte checken und gesund einkaufen [WWW Document]. URL
http://www.codecheck.info/ (accessed 4.4.16).
Declaratio, 2019. Food Allergen Symbols for Labeling of Food Items [WWW Document]. Website.
URL https://www.declaratio.net/hilfsmittel/allergensymbolik/materialen/ (accessed 4.22.19).
Dunford, E., Trevena, H., Goodsell, C., Ng, K.H., Webster, J., Millis, A., Goldstein, S., Hugueniot, O.,
Neal, B., 2014. FoodSwitch: A Mobile Phone App to Enable Consumers to Make Healthier Food
Choices and Crowdsourcing of National Food Composition Data. JMIR mHealth uHealth 2, e37.
https://doi.org/10.2196/mhealth.3230
Dunford, E.K., Neal, B., 2017. FoodSwitch and use of crowdsourcing to inform nutrient databases. J.
Food Compos. Anal. https://doi.org/10.1016/j.jfca.2017.07.022
EU-1169/2011, 2014. Regulation (EU) 1169/2011: Regulation on Food Information to Consumers
[WWW Document]. Regul. 1169/2011. URL
http://ec.europa.eu/food/safety/labelling_nutrition/labelling_legislation/index_en.htm (accessed
1.23.16).
FDA, 2017. Food Allergies: What You Need to Know, Consumer Reports.
Frey, R.M., Ryder, B., Fuchs, K., Ilic, A., 2016. Universal Food Allergy Number, in: Proceedings of
the 6th International Conference on the Internet of Things - IoT’16.
https://doi.org/10.1145/2991561.2998462
Fuchs, K., Huonder, V., Vuckovac, D., Ilic, A., 2016. Swiss Foodquiz: Inducing Nutritional
Knowledge via a Visual Learning based Serious Game, in: 24th European Conference on
Information Systems, ECIS 2016. Istanbul, Turkey.
Gigandet, S., 2019. Openfoodfacts Mobile Applications [WWW Document]. Openfoodfacts.com.
URL https://world.openfoodfacts.org (accessed 3.30.19).
Goldstein, S.P., Evans, B.C., Flack, D., Juarascio, A., Manasse, S., Zhang, F., Forman, E.M., 2017.
Return of the JITAI: Applying a Just-in-Time Adaptive Intervention Framework to the
Development of m-Health Solutions for Addictive Behaviors. https://doi.org/10.1007/s12529-
016-9627-y
Gregor, S., Hevner, A.R., 2013. Positioning and Presenting Design Science Research for Maximum
Impact. MIS Q. 37, 337–355. https://doi.org/10.2753/MIS0742-1222240302
GS1-CH, 2018. GS1 Switzerland: Trustbox Database [WWW Document]. URL
https://www.trustbox.swiss/ (accessed 8.28.18).
Gupta, R., Holdford, D., Bilaver, L., Dyer, A., Holl, J.L., Meltzer, D., 2013. The economic impact of
childhood food allergy in the United States. JAMA Pediatr.
https://doi.org/10.1001/jamapediatrics.2013.2376
Hevner, A.R., March, S.T., Park, J., Ram, S., 2004. Design Science in Information Systems Research.
MIS Q. 28, 75–105. https://doi.org/10.2307/25148625
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
Fuchs et al. / Towards Digital Food Allergen Labels
. 14
Julia, C., Hercberg, S., 2017. Development of a new front-of-pack nutrition label in France: the five-
colour Nutri-Score. Public Heal. Panor. 3, 712–725.
Julia, C., Péneau, S., Buscail, C., Gonzalez, R., Touvier, M., Hercberg, S., Kesse-Guyot, E., 2017.
Perception of different formats of front-of-pack nutrition labels according to sociodemographic,
lifestyle and dietary factors in a French population: Cross-sectional study among the NutriNet-
Santé cohort participants. BMJ Open. https://doi.org/10.1136/bmjopen-2017-016108
Krizhevsky, A., Sutskever, I., Hinton, G.E., 2012. Alexnet. Adv. Neural Inf. Process. Syst.
https://doi.org/http://dx.doi.org/10.1016/j.protcy.2014.09.007
Marchisotto, M.J., Harada, L., Kamdar, O., Smith, B.M., Waserman, S., Sicherer, S., Allen, K.,
Muraro, A., Taylor, S., Gupta, R.S., 2017. Food Allergen Labeling and Purchasing Habits in the
United States and Canada. J. Allergy Clin. Immunol. Pract.
https://doi.org/10.1016/j.jaip.2016.09.020
Mhurchu, C.N., Volkova, E., Jiang, Y., 2019. Effects of interpretive front-of-pack nutrition labels on
consumer food purchases: A randomized controlled trial. Obes. Res. Clin. Pract.
Nahum-shani, I., Hekler, E.B., Spruijt-metz, D., 2015. Building health behavior models to guide the
developmnet of just-in-time adaptive interventions: A pragmatic framework. Heal. Psychol. 34,
1209–1219. https://doi.org/10.1037/hea0000306
Nahum-shani, I., Smith, S.N., Witkiewitz, K., Collins, L.M., Spring, B., Murphy, S.A., 2014. Just-in-
time adaptive interventions (JITAIs): An organizing framework for ongoing health behavior
support., The Methodology Center Technical Report.
Pandey, P., Deepthi, A., Mandal, B., Puhan, N.B., 2017. FoodNet: Recognizing Foods Using
Ensemble of Deep Networks. IEEE Signal Process. Lett.
https://doi.org/10.1109/LSP.2017.2758862
Peffers, K., Tuunanen, T., Rothenberger, M. a., Chatterjee, S., 2008. A Design Science Research
Methodology for Information Systems Research. J. Manag. Inf. Syst. 24, 45–77.
https://doi.org/10.2753/MIS0742-1222240302
Sicherer, S.H., Sampson, H.A., 2018. Food allergy: A review and update on epidemiology,
pathogenesis, diagnosis, prevention, and management. J. Allergy Clin. Immunol. 141, 41–58.
https://doi.org/10.1016/j.jaci.2017.11.003
Sicherer, S.H., Sampson, H.A., 2014. Food allergy: Epidemiology, pathogenesis, diagnosis, and
treatment. J. Allergy Clin. Immunol. 133, 291-307.e5. https://doi.org/10.1016/j.jaci.2013.11.020
Turner, P.J., Kemp, A.S., Campbell, D.E., 2011. Advisory food labels: Consumers with allergies need
more than “traces” of information. BMJ. https://doi.org/10.1136/bmj.d6180
Umasunthar, T., Leonardi-Bee, J., Turner, P.J., Hodes, M., Gore, C., Warner, J.O., Boyle, R.J., 2015.
Incidence of food anaphylaxis in people with food allergy: A systematic review and meta-
analysis. Clin. Exp. Allergy. https://doi.org/10.1111/cea.12477
Versluis, A., Knulst, A.C., Kruizinga, A.G., Michelsen, A., Houben, G.F., Baumert, J.L., van Os-
Medendorp, H., 2015. Frequency, severity and causes of unexpected allergic reactions to food: A
systematic literature review. Clin. Exp. Allergy. https://doi.org/10.1111/cea.12328
Volkova, E., Li, N., Dunford, E., Eyles, H., Crino, M., Michie, J., Ni Mhurchu, C., 2016. “Smart”
RCTs: Development of a Smartphone App for Fully Automated Nutrition-Labeling Intervention
Trials. JMIR mHealth uHealth 4, e23. https://doi.org/10.2196/mhealth.5219
Waibel, K.H., Bickel, R.A., Brown, T., 2019. Outcomes From a Regional Synchronous Tele-Allergy
Service. J. Allergy Clin. Immunol. Pract. 7, 1017–1021.
https://doi.org/10.1016/J.JAIP.2018.10.026
Twenty-Eighth European Conference on Information Systems (ECIS2020) – A Virtual AIS Conference.
