We identified available apps with an extensive search in the German Apple App Store and Google Play Store in April 2020. The search included the following keywords: “Morbus Crohn,” “Colitis ulcerosa,” “CED,” “Chronisch entzündliche Darmerkankungen,” “IBD,” “Crohn’s disease,” “ulcerative colitis,” “UC,” “inflammatory bowel disease,” “Crohn,” and “colitis.” The search was carried out semiautomatically, initially using a web crawler to retrieve available apps. The app store descriptions for the available apps were read by 2 raters (JK and MG), who then manually screened them for the inclusion and exclusion criteria. The screened apps did not have to be IBD specific, but had to at least be health specific. For example, they had to include functions such as medication reminders, toilet finders, or symptom diaries. Disease-specific apps that targeted other diseases were not considered to fit the inclusion criteria. Apps were included if they were (1) in the German language, (2) available in both app stores, (3) targeted patients or physicians, and (4) were clearly designed for IBD treatment, were relevant to IBD, or were at least relevant to health in general. Apps were excluded if they were (1) only usable with an additional device, (2) congress apps, (3) journal apps, (4) apps only available to study participants or physicians, or (5) inactive apps.

All 6 raters were physicians completing their internal medicine fellowships. Half the raters (n=3) used iPhones and the other half (n=3) used Android smartphones. As recommended by the developers of MARS, all participating raters viewed the training video by Stoyanov et al [24] before rating the apps, and the raters tested each app for at least 10 minutes. The different MARS rating aspects were discussed by the team in advance. The selected apps were downloaded and rated between July and October 2020. All the raters (n=6) tested 4 of the final 14 apps (with most downloaded from the app stores), and the remaining 10 apps were randomly allocated, so that each remaining app was rated by 1 iPhone and 1 Android user.

Statistical analysis was implemented following the same design as Knitza et al [25], who recently performed an analysis of German mobile apps for rheumatology. MARS section scores were calculated by taking the arithmetic mean of the score for each item in a section, with the overall score being the arithmetic mean of the section scores (excluding the subjective quality score). Overall scores and section scores were summarized as the median and range for each app, and apps were ranked based on the median overall MARS score. We analyzed item score deviations by section and rater using a random intercept–only mixed-effects linear regression model including the individual item scores as the dependent variable, a random effects term for the rater, and nested random effects terms for the MARS section and app. Using random intercepts from this model, we estimated how the item scores in each section for each app deviated from the overall mean item score to rank and plot the importance of the sections within each app. Similarly, we plotted the random effect intercepts and respective 95% CIs for the raters to rank the raters by their deviation from the overall mean item score as a measure of rater bias. Random intercept and fixed effect term CIs spanning both sides of 0 were considered insignificant. Finally, we analyzed interrater agreement at the item, section, and overall score levels for raters from a rater sample as the ICC2k (2-way random, average measures, absolute agreement). All data analysis was performed using the open-source R software package (v 3.5.3; R Foundation). Mixed-effect analysis was carried out using the lme4 R package.

Additional app functions and information are shown in Table 1. The result section was generated by manually screening the final apps, checking the home page of the developers of the apps, and reading the descriptions of the apps in the 2 different app stores. This search was performed by 1 of the 6 raters (MG). The following information was systematically assessed: target group, target disease, developer of the app, app category and technical aspects, studies available, medical product, and privacy policy. For the screening of available studies, we additionally searched PubMed and Google Scholar for the app names. MG manually collected the app store ratings and the number of ratings from both app stores on Aug 24, 2021.

We initially retrieved 1764 apps using the web crawler. We removed 1386 apps because they were not related to IBD, 317 apps because they were not available in German, 5 apps because they were used for congresses, and 6 clinical study apps. We also excluded 7 apps because they required an additional device, most frequently a fecal calprotectin test device. Several of these device-specific apps required an invitation for user registration or required a specific calprotectin test kit (eg, the partner apps from Abbvie, CalApp/IBDoc by Bühlmann Laboratories, and the CalproSmart by Calpro AS). The QuantOn Cal app was only usable with the specific QuantOn Cal test kit. Of the remaining apps, 20 were only available in 1 of the 2 app stores and were also excluded, as were 4 journal apps, 3 duplicates, and 1 inactive app. We included a final total of 14 apps in the MARS analysis (Figure 1).

Only 3 of the 14 rated apps (21%) addressed IBD in general, and none of them were specific to Crohn disease or ulcerative colitis. The other 11/14 apps (77%) were not IBD-specific but were relevant to IBD in other ways, such as by including pain and medication diary functions, stool protocols, or a toilet finder (see Table 1).

Importantly, we found that all of the final 14 analyzed apps addressed patients; none of them directly targeted physicians. A diary function was included in 9 of the 14 apps (64%); depending on the app, patients could track pain, frequency of defecation, or eating habits. A public toilet finder was included in 3 of the 14 apps (21%). Most of the apps had a reminder function, whether for appointments or medication. CED Forum was the only 1 of the 14 final apps that provided IBD patients an IBD-related social media platform with features similar to conventional social media platforms. It provided chatrooms on topics such as medication, symptoms, and other personal experiences, as well as diet.

Most of the apps were developed directly or indirectly by subcompanies of pharmaceutical companies. The app Carenity enabled patients to complete surveys for scientific studies; several of these can be found on PubMed or Google Scholar [34,35]. Patients did not receive compensation for completion. The app Deutsches Gesundheitsportal was the only app to directly quote research and to include articles and chapters [36]. There were also some studies based on the app Manage My Pain [30-32]. A past study analyzed patient adherence and acceptance for this app [37]. We identified 1 German study of the app Cara Care, which is for irritable bowel syndrome [33]. We classified 3 of the apps (3/14, 21%) as medical products [38-40].

Overall, app quality was varied. Median MARS scores ranged between 2.38/5 and 4.11/5. Figure 2 shows the individual MARS scores assigned by individual raters.

There was no significant difference between the iPhone and Android raters (P=.64, V=111.5). Rater agreement on the overall MARS score was good at the app level (ICC2k 0.84, 95% CI 0.68-0.93), for section score (ICC2k 0.82, 95% CI 0.76-0.88), and for individual item score (ICC2k 0.84, 95% CI 0.81-0.86). Random intercepts for observers from the mixed-effects model are presented in Figure 3.

The median total MARS and section scores are displayed in Table 2, as are the respective app store ratings and number of ratings of the respective apps. The MARS sections with the highest scores were functionality and aesthetics, with median scores of 4.12 and 4.00, respectively. Subjective quality had the lowest section score with a median of 2.88.

Multimedia Appendix 1 shows that the subjective quality section was rated systematically lower than the random intercept for each app. Otherwise, no systematic item score deviations were observed.

To our knowledge, no high-quality analysis of German IBD apps has yet been carried out. Our work was intended to inform patients and physicians alike about IBD apps based on the results of structured and objective testing criteria in order to guide and facilitate the selection and inclusion of appropriate IBD apps in the clinical routine.

In contrast to a previous analysis in rheumatology that used a similar search strategy [25], we did not find a single app that targeted physicians as users. This reflects an untapped potential and an opportunity, as physicians are increasingly using medical apps [41,42].

Only a few (3 of 14) of the rated apps were IBD specific: CED Dokumentation und Tipps, CED Forum, and Cara Care. None were specific to Crohn disease. Considering the relatively high incidence and prevalence of IBD [43], it was surprising that we could not identify a single disease-specific IBD app, either for Crohn disease or ulcerative colitis. In contrast, a previous analysis discovered multiple disease-specific German apps for rheumatic disease [25], including such comparatively rare types as systematic lupus erythematosus [44]. Overall, the retrieved apps addressed various patient-relevant functions and topics, but a single disease-specific app with combined app features would likely be more frequently and regularly used by IBD patients. Furthermore, such an app could include more specific topics, such as the fistulas associated with Crohn disease and IBD-associated arthritis or uveitis [45,46].

In general, information quality was rated rather poorly compared to aesthetics and functionality, representing another unmet need. As information concerning medication and disease are the top 2 features requested by patients suffering from chronic inflammatory rheumatic diseases [47], we infer that IBD patients likely also need this information. Accordingly, we are currently carrying out a patient survey to validate this assumption.

CED Dokumentation und Tipps was the app with the highest MARS score. This app was IBD specific and had several extra functions, such as a medication and appointment reminder, toilet finder, and diary function, that could be used to document pain level, stool frequency, weight, and eating habits. This app also provided nutrition recommendations, password protection, and had an especially intuitive interface and design.

Most of the apps were designed by pharmaceutical companies and did not explicitly report involving patients in their design. Notably, and in line with previous findings, very few supporting studies could be identified [25]. The app Deutsches Gesundheitsportal was the only app providing evidence (by quoting studies), while the app Carenity enabled patients to complete surveys for clinical studies. There was only one app, Manage My Pain, for which studies were available on function and patient adherence [30,32,37]. Another shortcoming of the examined apps was that none offered IBD-specific scores, such as the partial Mayo score for ulcerative colitis [48] or the Harvey-Bradshaw Index for Crohn disease [49].

In order to be eligible for prescription in Germany, developers need to provide evidence for the usefulness of their app. Only 1 of the 14 included apps, Cara Care, is expected to be among the first eligible apps in Germany related to irritable bowel syndrome and IBD. The developer homepage states that Cara Care is already eligible for prescription for patients with irritable bowel syndrome and that the developer has applied for eligibility for patients with IBD [50].

This study excluded apps for which a device was necessary for their use, such as the PartnerApp by Abbvie and QuantOn Cal, which both require a fecal calprotectin test device [51,52]. In our web search, we observed that these were the only devices with accompanying apps that could monitor inflammation activity in the bowel. In the future, these devices might be effective complementary apps that could provide objective data about actual disease status and other objective parameters, enabling improved remote monitoring.

Fecal calprotectin can predict relapses in IBD and indicate the response to medical treatment [53]. Furthermore, normalization of fecal calprotectin has recently been recommended as a treatment target in both ulcerative colitis and Crohn disease in the STRIDE (selecting therapeutic targets in inflammatory bowel disease) statements [54].

In several countries, including the United Kingdom and United States, IBD centers have already developed models for the use of telemedicine and have reported positive outcomes, such as decreased costs, decreased travel time, and reduced overall time for medical visits for patients [55-57]. Video consultations are used in most of these models. To the best of our knowledge, mobile apps have not been included in any of these models.

Some of the apps rated in our current study had a function allowing the creation of summary reports, for example of the last 3 months. Such reports can provide the treating physician a much more detailed and regular overview of the patient’s status, including treatment response and disease progression. The use of an additional app developed by a patient organization could be a useful supplement for improving telemedicine in IBD. Based on our study results, we suggest including the following disease-specific information in such apps: current therapy options (including evidence from major relevant clinical studies) and disease-specific scores, such as the Harvey-Bradshaw Index for Crohn disease [49] and the partial Mayo score for ulcerative colitis [48]. All major stakeholders, including patients, gastroenterologists, and scientists should be part of the app development process. In addition, studies of individual apps should be conducted to investigate their clinical and economic benefits and safety. Furthermore, the apps should be available in both app stores so that a recommendation can be made that is independent of the operating system. The developers of the apps should also be clearly identifiable. Finally, easy and secure data transmission to health care professionals should be ensured.

In the future, it may make sense to integrate the MARS score into the respective app stores in order to provide a standardized evaluation unit as an orientation aid for users. The implementation of MARS scores has been useful for the re-evaluation, optimization, and development of apps by revealing possible weaknesses of the apps and ways to improve them in a targeted manner. Evaluation with MARS should be carried out by patients and physicians as well as researchers, since future apps should ideally include 2-way communication and data exchange.

This study has several limitations. Importantly, the apps were all rated by physicians; no patients were included. There was a clear discrepancy between the physician MARS ratings and user ratings in the app stores, suggesting that there would also be significant differences in MARS scores if the same app was rated by a doctor and by a patient. To address this in a follow-up study, the results of this study will be discussed in a patient focus group and a reduced number of apps will be evaluated by patients. The IBD-specific apps had a significantly lower number of ratings compared to the non–disease-specific apps, which we consider was most likely due to the smaller target group. Using the web crawler, we performed an objective and automatic initial app search, similar to previous studies [58,59]. Nevertheless, some IBD apps may not have been recognized by our search strategy and might have been overlooked. Similarly, only apps available in both app stores were included. No detailed data safety analysis was performed, and we only assessed the availability of privacy policy information. Some apps also offered password protection. We excluded apps from our study that required additional devices, such as calprotectin test devices, because most of them were only accessible within specific clinical study programs and no funding was available to buy the devices. As some studies have already shown, the use of additional devices to provide objective and predictive laboratory data, such as from fecal calprotectin tests, is very useful for disease management and improves treatment [15-17,22,53,54]. The immense speed of mHealth development is also a general limitation on research in this area.

Our current study shows that at the moment, only a limited number of IBD-related apps are available to patients, and none are available to physicians. We found that app quality was varied, and we observed a general absence of clinical evidence and patient involvement.