Quality evaluation and functional classification of Arabic health apps: A systematic review

Abstract

Objective

The study aimed to systematically review mHealth apps available for Arabic speakers regarding quality and functional classification.

Methods

A systematic search was conducted on the Apple App Store. Eligible apps were downloaded and tested thoroughly and rated independently by two reviewers using the Mobile App Rating Scale (MARS). The National Institute for Health and Care Excellence (NICE) evidence standards framework for evaluating Digital Health Technologies (DHTs) was used to perform the functional classification for the included apps.

Results

Sixty-one health-related apps met the eligibility criteria and were included in the study. The average overall MARS score across all apps was 4.03 (SD = 0.38), indicating a generally high quality of the apps. The functions of the reviewed apps fit into 7 out of 10 NICE classifications. We found that 50% of the applied functions were equally split between ‘communicate’ and ‘inform’, 39% fell within the ‘self-manage’ and ‘preventive behaviour change’ classes, and only 11% of the apps were classified as having ‘treat’, ‘diagnose’ and ‘simple monitoring’ functions. Furthermore, our review indicated that 42% of the apps focused on supporting general health, and 28% covered diet and physical activity.

Conclusion

This work summarises the current state of available free-of-charge mHealth apps for Arabic speakers on the Apple App Store by highlighting their quality and coverage of functionalities and areas of health. In general, most of the apps were considered to be of decent quality but with partially limited functions and coverage of health conditions. Developers of future Arabic health apps should focus on engagement and aesthetic features, employing more advanced functions and supporting a wide range of health interventions.

Introduction

Digital health technologies represent various tools and platforms leveraging digital systems to enhance health outcomes and healthcare delivery. Digital health encompasses mobile health (mHealth), wearable devices, telemedicine, and other digital innovations designed to improve health services. ¹ Evidence from the literature suggests that digital health technologies have the potential to facilitate and enhance the quality of communication between patients with chronic diseases and healthcare providers (HCPs), as well as to increase patient satisfaction with treatment services and promote quality of life.^2,3 In particular, the increase in the use of mobile devices and smartphones has made mHealth applications, the focus of this study, an innovative and timely approach to delivering health interventions. ⁴ Several systematic reviews revealed that mobile applications (apps) have been increasingly reported as delivering health behaviour change interventions and showing promising results.^5,6 This could explain the observed growth of health-related app numbers in app stores. In 2024, an annual report on digital and mobile usage revealed that 257 billion apps were downloaded globally, an increase of 0.9% compared to 2023. ⁷ This growth reflects a sustained global demand for mobile applications, including health apps, which are among the fastest-growing app categories. ⁷

A search of the app market for health conditions would retrieve an endless list of commercial apps. Many health apps are available for download, whether designed for specific health conditions or for unspecified ones, but most have been strongly criticised. While the quality of mHealth apps has been widely reviewed and critiqued in other languages, there remains a significant gap in the literature regarding Arabic-specific mHealth apps. Several studies reviewed health apps and revealed that the majority of them (1) lacked reliable content, (2) lacked scientific evaluation and clarity about effectiveness regarding health outcomes, (3) lacked the involvement of users (patients) and input from HCPs in their development, and (4) lacked usability and engaging features.^8,9,10,11,12 These features are required for a high-quality app with understood potential in relation to its effectiveness and, in parallel, to its long-term engagement.^10,13 Health apps must adhere to international standards such as ISO/IEC 25010 to ensure high-quality interims of reliability, usability, and safety. ¹⁴ There is also a substantial concern that commercial low-quality apps might mislead individuals desperate for a solution to distressing conditions. ¹⁰ The reviews were on language-specific apps; none were in Arabic. A few studies focused on reviewing mHealth apps oriented to Arabic speakers, finding them either outdated and lacking a systematic approach or health condition-specificity.^15,16,17,18 The lack of systematic reviews evaluating the quality of these apps has left their functionality, usability, engagement, and information accuracy largely unexplored. This study addresses this critical gap by systematically reviewing and evaluating the quality of Arabic health apps and providing functional classification.

Methods

Where applicable, this review followed the guidelines for reporting systematic reviews according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) ¹⁹ to ensure methodological rigour and transparency (see Supplemental material for the PRISMA checklist). However, it was not prospectively registered.

Search strategy

A systematic approach was adopted to conduct the App Store search. The Apple App Store in Saudi Arabia was searched between 25 and 26 March 2022 to identify Arabic health apps. Eight carefully selected keywords, including Arabic and English, were employed for the search (see Supplemental material). These keywords were derived from the literature, and several pilot searches were conducted to identify the terms that would yield the most relevant results. For each keyword, the top 30 retrieved results were included for review. This decision was informed by the Apple App Store's ranking system, which is based on app popularity and considers factors such as the number of downloads and user engagement. ²⁰ By focusing on the top 30 results, we ensured the inclusion of the most visible and widely used apps for users. It also ensured feasibility, as the list of results in app stores can be almost endless and often becomes increasingly irrelevant as we move further down the list. This approach allowed us to balance comprehensiveness with practicality within the scope of the study.

App eligibility criteria

Apps were included if they were: (1) available in the Arabic language; (2) intended for public and patient use; (3) related to health or placed in the category of ‘Health and Fitness’ or ‘Medical’ in the app store; (4) free of charge.

Apps were excluded if they provided no direct health benefits or support to patients or public health. Specifically, apps were excluded if their purpose did not align with supporting patient health, promoting public health, or offering actionable health-related benefits, such as monitoring, education, or management of health conditions. Apps designed for professional use, those unrelated to health, or those that were dysfunctional, unresponsive, and constantly crashing during testing were also excluded.

App selection

After removing the duplicates, the process of app selection consisted of two stages. First, apps were screened based on their title, App Store description, listed functionality, and screenshots. Apps that did not meet the eligibility criteria were excluded at this stage. Second, apps identified as potentially relevant during the first stage were installed on two iPhones for a more detailed examination. At this stage, the apps were further assessed to confirm their eligibility based on their actual functionality and purpose. Apps failing to meet the inclusion criteria after this assessment were excluded from the study. Two reviewers (A.A.A.A and B.M.A.A) selected the apps independently through the two stages, and any disagreements between them were resolved through discussion involving a third reviewer (A.A.M.A).

Quality assessment and data extraction

Eligible apps were tested thoroughly for 10–15 min and rated independently by two reviewers (A.A.A.A and B.M.A.A) using the Mobile App Rating Scale (MARS), a reliable tool for assessing the quality of mobile health apps. ²¹ The scale contains 19 items grouped into four categories: engagement (five items), functionality (four items), aesthetics (three items), and information quality (seven items). All MARS items are scored on a 5-point scale (1—inadequate, 2—poor, 3—acceptable, 4—good, 5—excellent). MARS final scores are calculated using the four categories’ arithmetic mean (M). The means of the MARS scores from the reviewers were calculated. In addition, the mean and standard deviation (SD) were calculated for all the apps’ MARS scores and the subscale quality scores to reach a conclusion as to the apps’ quality. The intraclass correlation coefficient (ICC) with 95% CI was calculated to evaluate the interrater reliability for the raters. Following the scale developer's recommendations, the raters completed online training [4] in using the tool efficiently before rating the included apps. A pilot quality rating was also conducted to ensure consistency and reliability among the two raters. The pilot involved a randomly selected sample of apps that did not meet the inclusion criteria and were not part of the final review. Both raters independently assessed the selected apps using MARS. After that, the raters compared their scores and discussed any disagreements in interpretation or scoring criteria. Through this process, they reached a consensus by aligning their understanding of the MARS criteria and refining their evaluation approach as necessary.

A data extraction sheet was designed to extract the following data from each app: app name, developer, language, health area and Apple Store rating. The last was compared to the calculated average MARS scores for the apps using the Pearson correlation coefficient.

Functional classification

The National Institute for Health and Care Excellence (NICE) evidence standards framework for evaluating Digital Health Technologies (DHTs) ²² was used to perform the functional classification for the included health apps. The NICE framework includes ten functional categories for DHTs grouped into four evidence tiers: Tier 1 (System service), Tier 2 (Inform, Simple monitoring, and Communicate), Tier 3a (Preventive behaviour change and Self-manage), and Tier 3b (Treat, Active monitoring, Calculate, and Diagnose). These categories provide a structured approach for evaluating the functionality of DHTs. ²² According to the framework, each app was assigned to a single category that best described its main and obvious function. When an app presented multiple and equal evident functionalities, we followed the NICE framework's guideline to classify the app under the function belonging to the highest applicable evidence tier. This ensured that the app's most significant and impactful use case was prioritised in the classification. The apps’ classification was conducted individually by two reviewers (A.A.A.A and B.M.A.A). The results were then compared, and in case of any disagreement, agreement was reached by consensus. A third reviewer (A.A.M.A) was to be consulted if required. Descriptive statistics were then used to summarise the classification of the included apps.

Results

App selection

The PRISMA diagram in Figure 1 shows the Apple App Store search results and the selection steps. Sixty-one health-related apps met the eligibility criteria and were included in the study.

Figure 1.

PRISMA diagram.

App characteristics

The characteristics of the eligible apps are detailed in the Supplemental material. The apps were coded by number, as shown in the Supplemental material, for reference purposes. The Apple App Store rating average was 4.03 out of 5 for the included apps (Table 1). The lowest rating was 1.9 for only one app (53), and the highest rating was 5 for three apps (9, 60, 61). A total of 5 apps (8, 12, 52, 56, 57) had no rating. Almost all the apps (92%, n = 56) were available in at least two languages, mainly Arabic and English. Only five apps (8%) were available in Arabic alone. Almost 42% (n = 26) of the apps were intended to support general health, by means such as acquiring doctor consultations in different specialities (see Supplemental material). Apart from that, the remaining apps fell into eight areas of health: 28% (n = 17) were oriented to diet and physical activity, 11% (n = 7) were for mental health, 7% (n = 4) were ford diabetes disease, 3% (n = 2) for female health, 3% (n = 2) for pregnant health, 2% (n = 1) for rheumatic disease, 2% (n = 1) for COVID-19, 2% (n = 1) for Dermatology complications.

Table 1.

The MARS scores for the assessed apps.

App #	Engagement score	Functionality score	Aesthetics score	Information quality score	Overall MARS score	App store rating score
57	4.20	4.88	3.83	4.70	4.50	No rating
59	4.00	4.88	3.83	4.50	4.43	4
1	4.10	5.00	4.33	4.20	4.41	4.5
37	4.00	4.88	4.33	4.43	4.41	3.9
42	4.20	4.88	4.17	4.30	4.39	4.4
41	4.20	4.50	4.50	4.33	4.38	4.7
36	4.10	4.63	4.33	4.40	4.36	4.5
61	3.40	4.88	3.83	4.50	4.35	5
29	4.00	4.88	4.17	4.33	4.34	4.7
10	3.90	4.75	4.33	4.30	4.32	4.5
25	4.00	4.75	4.17	4.33	4.31	4.7
19	4.10	4.50	4.33	4.30	4.31	4.4
26	4.30	4.63	4.00	4.30	4.31	4
27	4.30	4.63	4.00	4.30	4.31	4.9
47	4.10	4.75	4.17	4.20	4.30	4.7
13	4.00	4.75	4.33	4.13	4.30	4
51	3.40	4.88	4.00	4.13	4.30	4.6
60	3.80	4.88	3.67	4.25	4.30	5
32	4.20	4.63	4.00	4.30	4.28	4.9
14	3.90	4.75	4.17	4.30	4.28	3.9
50	3.40	4.88	3.83	4.13	4.26	4.5
34	4.20	4.63	4.00	4.20	4.26	4.9
20	4.00	4.75	4.00	4.20	4.24	4.4
22	3.60	4.75	4.33	4.20	4.22	4.7
48	3.80	4.75	4.33	4.00	4.22	4.8
28	4.00	4.63	4.00	4.23	4.21	4.5
24	4.00	4.50	4.00	4.20	4.18	3.6
40	3.90	4.25	4.17	4.38	4.17	4.7
35	4.30	4.25	4.00	4.10	4.16	4.4
9	3.80	4.63	4.00	4.20	4.16	5
58	4.20	4.75	3.67	4.00	4.15	4
12	3.50	4.88	4.00	4.13	4.13	No rating
11	4.00	4.63	4.00	3.83	4.11	4.5
5	3.50	4.50	4.50	3.90	4.10	3.3
30	3.10	4.88	4.17	4.23	4.09	4.7
54	2.80	4.88	3.83	3.75	4.09	3
49	3.00	4.75	3.67	3.88	4.07	3.7
15	3.90	4.88	3.50	4.00	4.07	3.8
52	3.20	4.75	3.83	3.50	4.05	No rating
33	3.60	4.50	4.00	4.08	4.04	4.3
38	4.10	4.50	3.67	3.90	4.04	4.1
17	3.70	4.00	4.00	4.45	4.04	4.4
18	3.60	4.50	3.67	4.10	3.97	4.6
8	4.00	3.88	4.00	3.90	3.94	No rating
21	4.00	4.25	3.33	4.13	3.93	3.1
55	3.40	4.50	3.83	4.10	3.91	3.4
39	3.90	4.25	3.00	4.13	3.82	3.1
16	3.40	4.13	3.50	4.20	3.81	4.5
31	3.50	3.88	3.33	4.00	3.68	4
44	3.30	4.50	3.50	3.33	3.66	3.3
2	4.10	3.63	3.00	3.60	3.58	3.3
3	3.50	3.13	3.50	4.17	3.57	3.3
43	3.00	3.25	3.83	4.08	3.54	3.8
56	3.80	3.75	3.00	3.50	3.54	No rating
7	3.40	4.00	3.67	3.00	3.52	2
4	3.70	3.63	3.00	3.70	3.51	2.8
6	3.10	3.38	3.50	3.96	3.48	3.9
46	2.70	4.50	2.67	3.63	3.37	2
53	2.40	3.13	2.83	2.88	3.01	1.9
45	2.50	3.75	2.17	3.53	2.99	3
23	2.30	4.25	2.17	2.50	2.80	3.2
M (SD)	3.70 (0.49)	4.45 (0.48)	3.80 (0.51)	4.04 (0.40)	4.03 (0.38)	4.03 (0.78)

App quality assessment

Table 1 outlines the two raters’ average MARS domain and total MARS scores, along with the ratings extracted from the App Store. The average overall MARS score across all apps was 4.03 (SD = 0.38), indicating a generally high quality of the apps reviewed, determined by a MARS threshold score of 3.0 = acceptable, out of 5.0 = excellent. Of the 61 apps, only two (23 and 45) scored slightly less than the threshold of 3.0, with an overall MARS score of 2.80 and 2.99, respectively. The highest score was 4.50, given to only one app (57). Among the four scale domains, engagement and aesthetics had the lowest ratings in most apps, with the means score being 3.70 (SD = 0.49) and 3.80 (SD = 0.51), respectively, representing acceptable to good quality. The interrater reliability between the reviewers, calculated from the MARS overall scores for all apps, was fair, using a two-way mixed intraclass correlation coefficient (ICC) of 0.619 (95% CI 0.365–0.772). Upon further analysis, ICCs were calculated for the individual MARS subdomains to better understand variability. The greatest agreement between raters was observed in the functionality domain, with an ICC of 0.779 (95% CI 0.632–0.867), reflecting substantial agreement. Conversely, the aesthetics domain showed the greatest variability, with an ICC of 0.346 (95% CI 0.046–0.598), indicating fair to poor agreement. These findings suggest that variability in subjective domains such as aesthetics may have influenced the overall ICC score, reflecting the challenges of evaluating subjective criteria. The total MARS scores showed a strong positive correlation (r = .73, P < .01) with the app store ratings.

Functional classification

The functions of the reviewed apps (n = 61) fit into 7 out of 10 NICE classifications, as shown in Figure 2 (see Supplemental material for complete classification). The dominant functions of the apps were ‘communicate’ and ‘inform’, representing 25% (n = 15) for each, followed by ‘self-manage’ and ‘preventive behaviour change’ with 21% (n = 13) and 18% (n = 11), respectively. Only 11% (n = 7) of the apps were classified with ‘treat’, ‘diagnose’, and ‘simple monitoring’ functions.

Figure 2.

The NICE functional classification of the reviewed apps.

Discussion

This work summarises the current state of available free-of-charge mHealth apps for Arabic speakers on the Apple App Store by highlighting their quality and coverage of functionalities and areas of health. We reviewed 61 health apps intended for public and patient use. The review revealed that the overall quality of the apps was considered good, as the average overall MARS score across all apps was 4.03 (SD = 0.38) (Table 1). This finding is consistent with the public impression of the apps, confirmed by the strong positive correlation (r = .73, P < .01) with the app store ratings. Although this could be seen as a positive aspect of the reviewed Arabic health apps, there is still room for improvement.

Across all four subscales of MARS, the app engagement and aesthetics are slightly less than the good quality score compared to the other two subscales, scoring an average of 3.70 (SD = 0.49) and 3.80 (SD = 0.51), respectively. These two domains were found to have the lowest scores in a recent study that assessed nine breast cancer apps for Arabic speakers using the MARS framework. ¹⁶ The study reported a mean score of 2.32 for the engagement and 3.00 for the aesthetics. Another study reviewed eighteen diabetes apps in the Arabic language using a different assessment approach and highlighted the need to focus on engagement and appealing features in future apps. ¹⁷ The engagement dimension of MARS showed the lowest average score in apps for multi-health conditions and other language speakers.^23,24 This could indicate that achieving high user engagement in mHealth apps, in general, needs further work.

It is vitally important to ensure that mHealth apps are engaging and interesting to users. This is because the extent of a user's exposure to the app content is entirely mediated by engagement, which is necessary for the app to be effective.^25,26 Evidence identifies user engagement as an essential factor in the success of digital health interventions, including mHealth apps. ²⁷ Additionally, difficulty in using mHealth apps may limit user retention rates, so greater emphasis should be placed on usability. Evidence suggests that actively involving end-users in the early stages of design and throughout the development process, as well as understanding and considering their requirements and preferences, increases the likelihood that the resultant products will be engaging and used by them.^28,29

Regarding the functional classification of the reviewed apps, we found that 50% (n = 30) of the applied functions were equally split between ‘communicate’ and ‘inform’ (Figure 2). In other words, most reviewed health apps serve as two-way communication channels between the public/patient and the healthcare providers or as information and resource channels for specific health conditions or general health and lifestyle. Likewise, another study found that information provision is the largest class, representing two-thirds of the categorised medication apps (n = 601). ³⁰

The second largest cluster (39%) of reviewed apps in the current study falls within the ‘self-manage’ and ‘preventive behaviour change’ classes. The role of these apps is to address public health issues such as healthy eating and exercise, and to promote self-management of a specified condition such as diabetes. Such functions require the employment of behaviour change techniques to ensure a high level of effectiveness.^22,31

On the other hand, we found that all other defined NICE functions have either a low number of apps, such as ‘treat’ (n = 1), ‘diagnose’ (n = 2), and ‘simple monitoring’ (n = 4) or none, such as ‘system service’, ‘active monitoring’, and ‘calculate’. The scarcity and absence of these classes may be attributed to the need for complex and advanced functionalities compared to providing health information or basic services. Conversely, apart from ‘system service,’ these functions appear to offer significant user-directed benefits and health outcomes, thereby broadening the range of health app functionalities available for Arabic speakers. The lack of apps classified under the ‘system service’ function can be attributed to the scope of this review. According to the NICE framework, ‘system service’ apps primarily deliver system-level benefits, such as operational or administrative tools, rather than direct advantages to users. ²² As this review focused exclusively on apps designed to provide direct health-related benefits to patients or the public, those primarily serving a ‘system service’ function were not included in the study.

Furthermore, our review indicated that the apps focused on supporting general health (42%), by means such as acquiring doctor consultations in different specialities, followed by supporting diet and physical activity (28%). The few remaining apps (30%) covered a combination of health conditions, such as mental health, Diabetes disease, and Dermatology complications (Supplemental material). This highlights the gap in the current apps available for Arabic speakers and the need to cover a more extensive range of different health conditions in future apps.

Strengths and limitations

Our study uniquely contributes by applying the NICE functional classification framework to Arabic-language health apps, offering a structured analysis of their functionalities. This approach highlights significant gaps in advanced features, such as active monitoring, diagnosis, and treatment. These findings underscore the need for future Arabic health app development to focus on incorporating these advanced functionalities to better meet user needs and align with global app standards.

The study followed a systematic approach to searching and identifying available apps for Arabic speakers exclusively in the Apple App Store, which may not capture apps available on other platforms. The search could have been widened by including the Google Play Store, a leading mobile operating system besides iOS. ³² However, this was not undertaken mainly because of resource limitations. No Android devices were accessible for the researcher to use to test apps, and more time would have been needed. Consequently, the findings may not fully represent the breadth of Arabic mHealth apps accessible to users on Android devices. Furthermore, the search was conducted in the Saudi Arabia app store and considered the publicly available apps. We might have missed Arabic apps that are only available in country-specific app stores or for consumers of specific healthcare providers. Additionally, the inclusion of only the top 30 results for each keyword search may have excluded less popular or newly released apps that could still be of high quality or relevance. While this approach was chosen to ensure feasibility and focus on widely used apps, it may have limited the comprehensiveness of the app review. Future studies could expand the scope by including additional app stores and broader search parameters.

Conclusion

The rapid growth of health-related apps in app stores has raised critical questions about their quality and utility for the public and patients. This study systematically reviewed the current landscape of mHealth apps available for Arabic speakers, offering valuable insights into their quality and functional scope. While many apps demonstrated decent overall quality, they often lacked advanced functionalities and comprehensive coverage of diverse health conditions. To enhance the impact of Arabic health apps, future development should prioritize improving user engagement and aesthetic features, incorporating advanced functionalities, and supporting a broader range of health interventions. These efforts can help bridge existing gaps and ensure that mHealth apps effectively meet the needs of Arabic-speaking populations.