Online health management continuance and the moderating effect of service type and age difference: A meta-analysis

Abstract

Numerous empirical studies have been carried out to explore factors of online health management continuance. However, results were not unified. We thus conducted a meta-analysis to identify influential factors and potential moderators. A systematic literature search was performed in nine databases (PubMed, Web of Science, the Cochrane Library, Ovid of JBI, CINAHL, Embase, CNKI, VIP, and CBM) published up to December 2020 in the English or Chinese language. Meta-analysis of combined effect size, heterogeneity, moderator analysis, publication bias assessment, and inter-rater reliability was conducted. Totally 41 studies and 12 pairwise relationships were identified. Confirmation, perceived usefulness, satisfaction, information quality, service quality, perceived ease of use, and trust were all critical predictors. Service type and age difference showed their moderating effects respectively. The perceived usefulness was more noteworthy in medical service than health and fitness service. The trust was more noteworthy in young adults. The results confirmed the validity and robustness of the Expectation Confirmation Model, Information Systems Success Model, and trust theory in online health management continuance. Moderators included but are not limited to age difference and service type. The elderly research in the healthcare context and other analytical methods such as qualitative comparative analysis should be applied in the future.

Keywords

Age difference continuous use online health management service type

Highlights

The model integrating the Expectation Confirmation model (ECM), Information Systems Success Model (ISS), and trust theory are robust to understand online health management continuance.

Moderators of online health management continuance included but are not limited to age difference and service type.

Scholars should put more effort on studies targeting a specific type of application and age-fit users to analyze the influential factors.

Elderly research in the healthcare context and other analytical methods such as qualitative comparative analysis should be applied in the future.

Introduction

E-Health is a new paradigm to facilitate healthcare service delivery.¹ According to statistical data of iiMedia Research² and China Internet Network Information Center (CNNIC),³ the emergence of COVID-19 pandemic has caused a sharp rise in demand for delivering online health management service. Online health management, regarding internet information technology as a carrier, mainly including remote monitoring, telemedicine, and health-related mobile applications or websites.⁴ It provides health information and services to assist users in organizing their health records, monitoring their health trends, communicating with medical professionals, and sharing their health information with healthcare providers, caregivers, or families.⁵ It has been proved as an effective strategy to meet the individuals’ health management needs and improve healthcare service delivery.⁶

The premise of the health promotion effect is that users are willing to accept and use the online health management tools. Bhattacherjee⁷ pointed out that though the initial adoption is an important first step to realize the information systems success, the determinant factor for the final success is continuous use. Different from the adoption, continuous intention⁷ is conceptualized as users’ intention to continue using information systems after they reevaluating their initial acceptance decision. However, the average 30-day retention rate for online health management was less than 50%.^8–10 The low retention rate revealed that there is a great potential for improvement in users’ continuance. Consequently, post-adoption behavior is a factor worthy of exploration and attention.

There are multiple factors, positively or negatively, influence the continuous use of online health management. Factors such as perceived usefulness, ease of use, and trust were concluded with mixed statistical significance, direction, and magnitude.^11–14 It may result from inconsistent sample sizes, research populations, contextual perspectives, and study qualities. The complex evidence and interactive healthcare system makes healthcare providers difficult to inform evidence-based interventions. Hence, using a quantitative synthesis technique to integrate complex results in the health field is necessary. This paper only focuses on users who employ online health management carrier to manage their health, but not include the healthcare providers and technology managers.

Combing influential factors could provide a general understanding of continuous use. However, it cannot explain why some online health management is more sustained than others within the same user population and why the same technology service engenders different usage intentions or behavior across different ages.¹⁵ Based on prior research,^16–18 service type and age difference have the potential to be moderators. Previous research had explored several classifications such as the e-health system versus tool,¹⁸ online hedonic versus utilitarian versus communication service.¹⁵ Reference to the IMS Institute⁸ and detailed classifications in prior research,¹⁹ this paper divided online health management type into two primary categories: the health and fitness service type are wellness-focused such as fitness, diet, and stress, and the medical service type are disease-focused such as diabetes app, Haodaifu consultation platforms and mental health applications. For the age difference, researchers had already speculated and verified age difference to be a meaningful moderator on mobile health adoption.^20–23 But did the age difference still moderate online health service continuance remains unclear. Identifying the moderating effect of service type and age difference can provide additional insight for the subject.

Meta-analysis is a statistical method that can synthesize different homogeneity research results to provide rigorous quantitative empirical evidence.²⁴ This article aims to: (1) elaborate influential factors of online health management continuance; (2) depict a consistent model to understand the users’ continuous use of online health management, and provide some inspirations for future research; (3) specify whether the age difference (young adults versus all population) and service type (medical service versus health and fitness service) act as moderators on online health management continuance.

Research model

Expectation confirmation model

Expectation confirmation theory (ECT), proposed by Oliver²⁵ in 1980, exerts wide-reaching implications in marketing studies to elucidate post-purchase satisfaction. The main constructs of this theory are expectation, perceived performance, confirmation, and satisfaction. ECT emphasizes that customer satisfaction with services or products is based on the comparison result of expectation and performed performance, and the satisfaction will influence the intention of continued use.²⁶ Bhattacherjee extended the original model and pointed out the potential predictive power of perceived usefulness.⁷

Information systems success model

ISS model centers on the belief that system quality and information quality determine users’ satisfaction to information systems, especially in e-commerce systems.²⁷ This model illustrates the connotation of information quality contained accuracy, relevance, understand-ability, completeness, currency, dynamism, personalization, and variety. In 2003, Delone perfected the ISS model and added service quality.²⁸ The revised model regarded satisfaction as an important index to measure the success of an information system. Plenty of studies that combined this model with other models or theories in an information technology context can be found,²⁹ and its efficiency has been widely confirmed.

Trust theory

Trust theory was originally widely applied in the industrial marketing and purchasing group.³⁰ McKnight introduced it into information system research and developed the trust model in 2002.³¹ Critical constructs in the trust model are disposition to trust, institution-based trust, trusting beliefs, trusting intentions, and trust-related behaviors. It deems that trust beliefs are positively related to trusting intention and finally lead to trust-related behaviors.³¹ Trust theory emphasizes, far from being unitary, trust is multidimensional. Yet, the complex, multidimensional nature of trust has not been considered in the current empirical evidence of E-Health.

Our review of the extant literature revealed that researchers often integrate information system theories to enable theory development.^14,32,33 Specific, they used a traditional theory in information systems as a theoretical backbone, and then distilled critical constructs proposed by other complementary theories into the traditional theory.³² The expectation confirmation model (ECM) and Information systems success model (ISS) has been two extensively used theories in information systems continuance area, and their applicability has been well recognized.^34,35 Integrating ECM and ISS is an appropriate choice to explore users’ continuance in using an online health system. However, most existing studies integrating ECM and ISS employed only a subset of the constructs. The predicted power of the ECM and ISS hybrid model hasn’t been widely confirmed in an online health management context. Therefore, there is still the need for systematic integration of the constructs of ECM and ISS that would apply to online health management continuance.

In addition to integrating traditional theories from information system domination, theories that focus on a specific context and identify relevant factors are important in enriching theories and comprehensively understanding the focal phenomenon.³² Different from offline health services, online health services are multi-source, non-face-to-face, and untouchable. We thus assume that trust is a determinant for online health management continuance due to the mixed information and contactless interaction.

Moreover, our review of this body of work revealed that perceived ease of use was a factor with great divergence in current studies.^11,12,36 To figure out its effect, we took perceived ease of use into our theoretical model. Altogether, this study tried to integrate ECM, ISS, trust models and verify the model through meta-analysis. Figure 1 and Table 1 provided a clear presentation about our conceptual model and the definition of its vital constructs.

Figure 1.

Theoretical model.

Table 1.

Critical constructs of the proposed conceptual model.

Construct	Definition	Source
Perceived ease of use (PEOU)	The degree to which a person believed that using a particular system would be free of effort	Davis³⁸
Perceived usefulness (PU)	The degree to which a person believed that using a particular system would enhance their job performance	Davis³⁸
Information quality (IQ)	The degree to which a person believes that the information provided is complete, relevant, easy to understand, accurate, timely and comprehensive	Delone³⁶
Service quality (SeQ)	The degree to which a person evaluates the services and support provided by the services provider	Delone³⁶
System quality (SyQ)	The degree to which a person perceived the functionality, reliability, and flexibility of the system	Delone³⁶
Satisfaction (SAT)	An affect, captured as a positive (satisfied), indifferent, or negative (dissatisfied) feeling	Bhattacherjee⁷
Confirmation (CON)	The extent to which users' expectation of IS use is realized during actual use	Bhattacherjee⁷
Continuous use (CU)	Users' intention or behavior to continue using information systems	Bhattacherjee⁷
Trust (TRU)	The willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trust, irrespective of the ability to monitor or control that other party	Davis³⁹

Methods

This paper applied meta-analysis to analyze the combined effect size of each path through R 4.0.3, using Pearson correlation coefficient r as a common metric.³⁹

Article selection

We conducted a comprehensive search of the PubMed, Web of Science, Cochrane, Ovid of JBI, CIANHL, Embase, CNKI, VIP, CBM databases to identify eligible studies published up to December 2020 in English or Chinese. The search string was developed by combining two categories: online health management and continuance. The detailed search terms were displayed in Table 2. The search was performed in title, keywords, and abstract based on the combination of interventions and outcomes. Reference lists of relevant studies were also manually searched. Two independent authors used reference management software (Endnote X9) to delete duplicate articles, scan the titles, abstracts, and keywords for preliminary screening, and read the full article to identify eligible studies according to the inclusion and exclusion criteria.

Table 2.

Detailed search terms ( two combined with Boolean operator AND).

Online health management	Continuance
“Online health”, “online medical”, “mhealth”, “e-health”,“telehealth”, “telemedicine”, “mobile health”, “internet hospital”, “internet-based”, “technology-based”, “web-based”, “on-line”, “online healthcare service”, “smartphone”, “app*". (Combined with boolean operator OR)	“Continu*”, “intention to continue”, “substance use”, “user stickiness”, “post adoption”, “adherence”, “engagement”, “compliance”. (Combined with boolean operator OR)

Inclusion criteria: we included (1) P (population): study populations were users who employ online health management to manage their health; (2) I (interest): influential factors on online health management continuance. Online health management, mainly included remote monitoring, telemedicine, health-related mobile applications and websites; (3) O (outcome) was continuous use intention or behavior; (4) studies quantitatively examined at least one relationship from our theoretical model; (5) studies in English or Chinese; (6) studies must report the sample size and correlation coefficient r, otherwise, they must report standard regression coefficient or t-value; (7) If the same author conducted several studies based on the same sample, we only included the one with the largest sample size.

Exclusion criteria: we excluded: (1) editorials, interviews, comments; (2) the full text were not available.

Eventually, 41 articles were selected for meta-analysis. The entire process of article screening was shown in Figure 2. The distribution of articles by publication year was shown in Figure 3. Detailed information of the final included articles was elucidated in Table 3.

Figure 2.

The entire process of article screening.

Figure 3.

The distribution of included articles by publication year.

Table 3.

Overview of articles included in the meta-analysis.

Article (author)	Year of publication	Sample size	Theoretical model	Type	Country or area	Age	Journal
Shahriar Akter³⁰	2011	216	-	Medical	Bangladesh, Asia	All population	Journal of the American society for information science and technology
Shahriar Akter¹¹	2013	473	ISS	Medical	Bangladesh, Asia	All population	Behaviour and information technology
Sekyoung Youm⁴¹	2015	97	TAM	Health and fitness	Korea, Asia	All population	Telemedicine and e-Health
Shupei Yuan⁴²	2015	317	UTAUT2	Health and fitness	USA, Europe	Young adults	Telemedicine and e-Health
Jaehee Cho¹²	2016	343	PAM、TAM	Medical	Korea, Asia	All population	International Journal of medical informatics
He Minggui⁴³	2016	221	ECM-ISC	Medical	China, Asia	All population	New Media Research
Jian Mou⁴⁴	2016	436	ISS	Medical	South Africa	Young adults	Information development
Xi Zhang⁴⁵	2016	273	ECM	Medical	China, Asia	Young adults	Information technology for development
Ardion D³⁷	2017	476	TAM	Health and fitness	German, Europe	Young adults	International Journal of Human–Computer interaction
Meng Yin⁴⁶	2017	260	ECM-ISC	Medical	China, Asia	All population	Journal of Dalian University of technology (social sciences)
Min Zhang⁴⁷	2017	288	ECM	Medical	China, Asia	All population	Data analysis and Knowledge Discovery
Yuan Chen⁴⁸	2018	284	ELM	Medical	China, Asia	All population	International Journal of medical informatics
Dongxiao Gu⁴⁹	2018	494	ECM	Medical	China, Asia	All population	International Journal of Environmental research and public Health
Bing Wu¹⁴	2018	469	ECM-ISC	Medical	China, Asia	All population	Journal of medical internet research
Jia Yang⁵⁰	2018	224	TAM	Medical	China, Asia	All population	Journal of traditional Chinese medicine management
Dong Qingxing⁵¹	2019	105	ECM	Medical	China, Asia	All population	Journal of Modern information
Huo Haoshuang⁵²	2019	440	SCT	Medical	China, Asia	All population	Chinese Journal of medical library and information science
YuWei Chang⁵³	2019	241	S-O-R	Medical	China, Asia	All population	Patient Education and Counseling
Yue Chen⁵⁴	2019	307	CVT	Medical	China, Asia	All population	Dissertation
Hui Li⁵⁵	2019	302	ECM-ISC	Health and fitness	China, Asia	Young adults	Dissertation
Jia Li⁵⁶	2019	211	ECM	Health and fitness	China, Asia	All population	Informatics for health and social care
Yaoyao lin⁵⁷	2019	251	TAM、ECM、 ECM-ISC	Health and fitness	China, Asia	All population	Mathematics practice and theory
Changrui sun⁵⁸	2019	308	ECM	Health and fitness	China, Asia	Young adults	Dissertation
Linglu Yin⁵⁹	2019	456	UGT	Health and fitness	China, Asia	All population	Dissertation
Yiyu Zhang⁶⁰	2019	746	UTAUT	Medical	China, Asia	All population	Journal of medical internet research
Yimei Zhu⁶¹	2019	338	ISS	Health and fitness	China, Asia	Young adults	Dissertation
Ziying Hong⁶²	2019	283	TPB	Medical	China, Asia	All population	Health information Journal
Nanyan Cao⁶³	2020	473	TAM	Medical	China, Asia	All population	Statistics and management
Dongxiao Gu⁶⁴	2020	311	TPB	Medical	China, Asia	All population	Information science
Zizhen Qin⁶⁵	2020	505	ECM	Medical	China, Asia	All population	Journal of Modern information
Antoine Grenier Ouimet⁶⁶	2020	178	ISS	Medical	Canadian , Europe	All population	Journal of medical internet research
Jaehee Cho⁶⁷	2020	108	ECM、 TAM	Health and fitness	Korea, Asia	All population	Disability and health Journal
Hongcheng Cui⁶⁸	2020	338	TAM	Health and fitness	China, Asia	All population	Journal of Capital University of physical Education and sports
Guanxiong Huang⁶⁴	2020	449	TAM	Health and fitness	China, Asia	Young adults	Computers in Human behavior
Md Abdul Kaium¹³	2020	400	UTAUT、 ECM	Medical	Bangladesh, Asia	Elderly	International Journal of pharmaceutical and healthcare marketing
K Anil Kumar⁶⁹	2020	453	ECM	Medical	India, Asia	All population	Innovative marketing
Tomer shemesh⁷⁰	2020	168	TAM	Medical	Israel, Asia	Young adults	Telemedicine and e-Health
Minsun shim⁷¹	2020	506	ISS	Health and fitness	Korea, Asia	All population	International Journal of medical informatics
Reginald A silver⁷²	2020	1142	TAM、UTAUT2	Medical	USA, Europe	Young adults	Journal of medical internet research
Chunyan Xie⁷³	2020	373	TAM、ECM	Medical	China, Asia	Elderly	Risk management and healthcare policy
Xiaoxiao Zhang⁷⁴	2020	379	ISS	Health and fitness	China, Asia	Young adults	International Journal of nursing sciences

Note: TAM refers to Technology Acceptance Model; ECM refers to Expectation Confirmation Model; ECM-ISC refers to Expectation-Confirmation Model of IS Continuance; ISS refers to the Delone and McLean (D&M) Information Systems Success Model; ELM refers to Elaboration likelihood Model; UTAUT refers to Unified Theory of Acceptance and Use of Technology Model; UTAUT2 refers to Unified Theory of Acceptance and Use of Technology2 Model; PAM refers to Post-acceptance Model; S-O-R refers to stimulus-organism-response theory; UGT refers to Use and Gratifications Theory; SCT refers to Social cognition theory; ”-” refers to did not have a theoretical model.

Study quality assessment

Two independent authors used an amended version of the Joanna Briggs Institute’s checklist for prevalence studies⁷⁴ to assess the methodological quality of eligible studies. All disagreements were resolved through the discussion with a third author. There are nine items included in JBI Critical Appraisal tools. If all items in the study are judged as “yes,” the quality is evaluated as A level; if only some of the items are judged as “yes,” it is evaluated as B level; if none of the items is judged as “yes,” it is evaluated as C level, and studies with C level should be excluded.^75,76

Inter-rater reliability

According to McHugh,⁷⁷ we used percent agreement statistic and Cohen’s kappa statistic to measure inter-rater reliability for study quality assessment.

Study coding

Foremost, we extracted the data into a pre-created excel spreadsheet according to an information list: author, year of publication, path relationship, effect size, sample size, research model, subgroup, the significance of the path relationship, and critical constructs. All extracted data was cross-checked by two authors. For some studies that reported the standard regression coefficient or the t-value of the path, we treat the standard regression coefficient as an effect size as the correlation coefficient r²⁴. For the t-value, we used a common formula proposed by Wolf⁷⁸ to convert it to an effect size: $r = \sqrt{t^{2} / (t^{2} + d f)}$ , where t is the t-value of the path, df is the degree of freedom.

Data analysis

The following is the basic statistical analysis by R 4.0.3, using the metacor (Meta-analysis of Correlation coefficients) command to achieve inverse variance based on sample weighting.⁷⁹ The correlation coefficients and sample sizes were collected to complete the integration of effect sizes with a random-effects model using the method of DerSimonian and Laird.⁸⁰ The statistics for the accuracy of this article are 95% confidence interval, heterogeneity statistic Q-value, and heterogeneity index I². Q-values of group differences were conducted to examine moderation.⁸¹ The index of publication bias was fail-safe N.⁸² The corresponding formula for calculating the fail-safe N⁸³ is as follows: p = 0.05, Nf_s=(ΣZ/1.645) ²- K; p = 0.01, Nf_s=(ΣZ/2.33) ²- K; Where K is the number of included studies, and Z is the Z-value of each independent study.

Results

Descriptive statistics

There were 41 empirical studies included in the meta-analysis to summarize research findings of 12 pairwise relationships on online health management. Overall, the qualities of 10 studies were assessed as A level, 31 studies were B level, and generally should minimize the risk of sampling methods and insufficient coverage of the sample. Appendix A presented the detailed outcome of the study quality assessment. Of the 41 studies, 27 were associated with medical service, 14 were health and fitness service. For the significance of relationships, Perceived usefulness-Continuous use (PU-CU), Perceived usefulness-Satisfaction (PU-SAT), Perceived ease of use-Perceived usefulness (PEOU-PU), Satisfaction-Continuous use (SAT-CU), Service quality-Perceived usefulness (SeQ-PU), Service Quality-Satisfaction (SeQ-SAT), Information quality-Perceived usefulness (IQ-PU), Information quality-Satisfaction (IQ-SAT), Confirmation-Perceived usefulness (CON-PU), and Confirmation-Satisfaction (CON-SAT) paths were studied with the significant rate of 100%, while the significant rate of Trust-Continuous use (TRU-CU), Perceived ease of use-Continuous use (PEOU-CU), and System Quality-Satisfaction (SyQ-SAT) were 83.3%, 66.7% and 50%, respectively. In terms of 14 pairwise relationships, Satisfaction-Continuous use (SAT-CU) path was the most widely investigated with 26 studies, following is the Perceived usefulness-Continuous use (PU-CU) path with 23 studies. Notably, System Quality-Satisfaction (SyQ-SAT) path was only reported by two studies, and no study reported the System Quality-Perceived usefulness (SyQ-PU) path. Due to insufficient data, SyQ-PU and SyQ-SAT paths were not available to conduct a meta-analysis. Therefore, a total of 12 pairwise relationships were conducted for meta-analysis. Table 4 provided a detailed quantitative description in statistics.

Table 4.

Descriptive statistics.

Path relationship	Path relationship (abbreviation)	Number of studies	Correlation coefficients		Significance			Range of sample sizes		Total sample size	Average Sample size
Path relationship	Path relationship (abbreviation)	Number of studies	Lower	upper	Significant	Non- significant	Significant (%)	Lower	upper	Total sample size	Average Sample size
Confirmation-perceived usefulness	CON-PU	11	0.16	0.72	11	0	100%	108	494	3277	297.90
Confirmation-satisfaction	CON-SAT	12	0.40	0.82	12	0	100%	108	494	3788	315.67
Perceived ease of use-Perceived usefulness	PEOU-PU	9	0.18	0.68	10	0	100%	224	746	3583	398.10
Service quality-Perceived usefulness	SeQ-PU	7	0.11	0.57	7	0	100%	178	469	2125	303.75
Service quality-satisfaction	SeQ-SAT	5	0.22	0.98	5	0	100%	308	506	2209	441.80
Information quality-perceived usefulness	IQ-PU	5	0.16	0.68	5	0	100%	260	469	1639	327.80
Information quality-satisfaction	IQ-SAT	5	0.19	0.58	5	0	100%	241	506	1990	398.00
Trust-continuous use	TRU-CU	12	-0.06	0.48	11	1	83.3%	97	453	3331	277.58
Perceived ease of use-Continuous use	PEOU-CU	12	0.02	0.47	8	4	66.7%	108	746	4452	371.00
Perceived usefulness-Continuous use	PU-CU	23	0.14	0.96	23	0	100%	108	746	7835	340.70
Perceived usefulness-Satisfaction	PU-SAT	14	0.11	0.68	14	0	100%	105	1142	5094	364.00
Satisfaction-continuous use	SAT-CU	26	0.12	0.94	26	0	100%	97	1142	9570	368.00
System quality-perceived usefulness	SyQ-PU	-	-		-		-			-	-
System quality-satisfaction	SyQ-SAT	2	0.07	0.15	1	1	50%	436	506	942	471.00

Note: “-” means that there was no available data.

Inter-rater reliability results

Results of the inter-rater reliability for each item of JBI Critical Appraisal tools were shown in Table 5. For the percent agreement statistic, all parameters of percent agreement were over 92%. For Cohen’s kappa statistic, all parameters of Cohen’s kappa were over 0.7. According to McHugh,⁷⁷ the level of inter-rater reliability for each criterion was moderate or higher.

Table 5.

Inter-rater reliability results for study quality assessment.

	Percent agreement, %	Value of kappa	Level of inter-rater reliability
Q.1	92.68	0.77	Moderate
Q.2	95.12	0.73	Moderate
Q.3	100.00	1.00	Almost perfect
Q.4	90.24	0.86	Strong
Q.5	92.68	0.85	Strong
Q.6	100.00	1.00	Almost perfect
Q.7	100.00	1.00	Almost perfect
Q.8	100.00	1.00	Almost perfect
Q.9	100.00	1.00	Almost perfect

Reliability statistics

Reliability refers to the consistency and stability of the constructs.²⁴ This paper posed Cronbach’s alpha as an index to describe the reliability of each construct. If articles only displayed the composite reliability (CR) but not Cronbach’s alpha, we recorded the composite reliability to replace the Cronbach’s alpha.²⁴ Reliability statistics have been listed in Table 6. The average Cronbach’s alpha of each construct is higher than 0.8, which means all constructs were deemed highly reliable across the study.

Table 6.

Reliability statistics.

Constructs	Abbreviations of constructs	Minimum	Maximum	Average	Variance	Number of studies
Perceived usefulness	PU	0.61	0.95	0.87	0.006	19
Perceived ease of use	PEOU	0.63	0.97	0.87	0.009	13
Information quality	IQ	0.78	0.93	0.89	0.002	10
Service quality	SeQ	0.76	0.93	0.96	0.003	11
Satisfaction	SAT	0.75	0.95	0.89	0.003	21
Confirmation	CON	0.73	0.96	0.88	0.004	11
Continuous use	CU	0.68	0.98	0.87	0.007	29
Trust	TRU	0.82	0.96	0.91	0.004	13
System quality	SyQ	0.69	0.94	0.84	0.100	7

Heterogeneity test

This paper is according to the benchmarks of I² and Q-value to test the heterogeneity. High Q (p < .001) and I² showed significant heterogeneity of all 12 pair-wise relationships. Table 6 exhibited the results of combined effect size. According to the standard raised by Cohen, which divided the magnitude of the effect degree into three levels (around 0.1=weak, around 0.3=moderate, reach 0.5=strong),⁸⁴ we observed that SQ-SAT (combined effect size=0.84), CON-SAT (combined effect size=0.76), SAT-CU (combined effect size=0.61) and PU-CU (combined effect size=0.52), had a strong impact between the dependent variable and the independent variable. The level of TRU-CU was weakest, with a combined effect size of 0.23. The combined effect size of CON-PU, IQ-PU, IQ-SAT, SQ-PU, PEOU-CU, and PU-SAT were all from 0.3 to 0.5, reporting the moderate relationship strength. By scanning the 95% confidence interval (CI), we can easily identify the statistically significant paths. If the 95%CI crosses the null, the path is non-significant in the statistic.²⁴ As shown in Table 7, none of the 95%CI crossed the null value, and it reflected that the 12 paths were all statistically significant.

Table 7.

The results of the heterogeneity test.

Relationship strength	Path relationship	Q-value	I ²	P ₁	Combined effect size	95%CI	Z-score	P ₂	Nfs_0.05
Strong	SeQ→SAT	1404.50	99.7%	.00^***	0.84	0.05–1.62	2.09	.04^*	2899
Strong	CON→SAT	200.65	94.5%	.00^***	0.76	0.62–0.90	10.78	.00^***	9337
Strong	SAT→CU	1172.14	98.1%	.00^***	0.57	0.41–0.73	7.04	.00^***	45,006
Strong	PU→CU	1223.92	98.2%	.00^***	0.52	0.35–0.68	6.06	.00^***	17,623
Moderate	CON→PU	154.98	93.5%	.00^***	0.48	0.35–0.62	6.90	.00^***	2689
Moderate	PEOU→PU	161.54	95.0%	.00^***	0.44	0.29–0.59	5.74	.00^***	2093
Moderate	IQ→PU	106.78	96.8%	.00^***	0.43	0.28–0.78	4.10	.00^***	894
Moderate	IQ→SAT	51.10	92.9%	.00^***	0.40	0.24–0.56	4.93	.00^***	585
Moderate	PU→SAT	408.97	96.3%	.00^***	0.39	0.26–0.52	5.76	.00^***	5779
Moderate	SeQ→PU	85.24	93.0%	.00^***	0.38	0.22–0.55	4.62	.00^***	834
Weak	PEOU→CU	87.65	87.5%	.00^***	0.24	0.15–0.32	5.48	.00^***	1009
Weak	TRU→CU	51.58	78.7%	.00^**	0.23	0.16–0.31	6.05	.00^***	759
-	SyQ-PU	—	—	—	—	—	—	—	—
-	SyQ-SAT	—	—	—	—	—	—	—	—

Note: “-” means data not available; P₁ is the significant level of Q-Test; P₂ is the significant level of Z-Test; 95%CI is 95% confidence interval; Nfs_0.05 is the results of publication bias; ^*** p < 0.001; ^** p < .01; ^* p < .05.

Moderator analysis

The heterogeneity test results indicated a moderating effect on each pairwise relationship. Drawing upon prior review, we identified service type and age difference as two possible moderators. We divided the dataset into separate groups for each moderator and conducted a meta-analysis separately. According to study contexts and methods reported in each research, we coded each paper into medical service or health and fitness service. With reference to previous research,⁸⁵ we classified age difference into three categories: young adults, elderly, and all population. According to the World Health Organization,⁸⁶ we coded the study whose population aged over 60 as “elderly,” aged 18–44 years as “young adults,” and the study whose population engaged all ages as “all population.” Some relationships were not performed owing to inadequate data such as SeQ-SAT, IQ-SAT, and CON-PU. The detailed information is shown in Table 8. The following results are obtained: (1) Except PU-SAT and TRU-CU two relationships, the effect sizes of each relationship across different age and service type was minor difference. (2) For the moderating effect of service type, PU had a significantly stronger positive effect on SAT in medical services (combined effect size=0.46, CI= 0.32–0.61) than health and fitness service (combined effect size= 0.22, CI= 0.16–0.28). (3) For the moderating effect of age difference, the impact of trust towards continuous use was more noteworthy in the young adults (n=3, combined effect size=0.37, CI = 0.23–0.52) than all population (n=9, combined effect size=0.18, CI = 0.12–0.25). Figure 4 and Figure 5 gave a clear presentation of the variations.

Table 8.

The results of the subgroup analysis.

Path relationship	Service type (effect size)			Age difference (effect size)
Path relationship	Medical	Health and fitness	Q-value	Young	Elderly	All population	Q-value
CON-PU	0.43	0.63	1.70	—	—	—	—
CON-SAT	0.72	0.82	0.27	—	0.79	0.72	0.07
PEOU-PU	0.38	0.51	0.54	0.50	—	0.41	0.08
SeQ-PU	0.45	0.38	0.25	0.48	—	0.31	1.26
SeQ-SAT	—	—	—	—	—	—	—
IQ-PU	—	—	—	0.68	—	0.43	1.11
IQ-SAT	—	—	—	0.46	—	0.36	0.22
TRU-CU	0.26	0.14	1.27	0.37	—	0.18	5.87 ^*
PEOU-CU	0.22	0.26	0.16	0.13	—	0.24	1.04
PU-CU	0.22	0.26	0.16	0.48	—	0.55	0.23
PU-SAT	0.44	0.26	4.83 ^*	0.40	—	0.24	3.47
SAT-CU	0.65	0.56	0.32	0.54	—	0.60	0.21
SyQ-PU	—	—	—	—	—	—	—
SyQ-SAT	—	—	—	—	—	—	—

Q-value refers to the results of subgroup differences; “-” means data not available ; ^* p < .05.

Figure 4.

The moderating effect on PU-SAT.

Figure 5.

The moderating effect on TRU-CU.

Publication bias

An important step in this article was to assess publication bias through the Rosenthal’s fail-safe N. All path relationships have passed the fail-safe N test (see Table 7), proving that each relationship’s overall effect sizes were robust. A consistent model supported by solid results was depicted in Figure 6.

Figure 6.

Revised model. Note: ^*** p < .001; ^** p < .01; ^* p < .05; The dashed relationships represented that there wasn’t available data in the meta-analysis.

Discussions

Principal findings The development of continuous use is not influenced by a single factor but rather by the combination of multiple factors

This paper examined seven influential factors and two potential moderators of online health management continuance. Meta-analysis results proved that confirmation, perceived usefulness, satisfaction, information quality, service quality, perceived ease of use, and trust were all critical predictors of online health management continuance. Information quality, service quality, and perceived ease of use significantly impact perceived usefulness. Service type played a moderating impact on PU-SAT, and age difference played a moderating effect on TRU-CU. Our findings suggested that online health management continuance factors are diverse, complex, and interactive. The development of continuous use is not influenced by a single factor but rather by the combination of multiple factors.

Online health management continuance grounded in ECM, ISS, and trust theory

The relationship strengths of CON-SAT, CON-PU, PU-SAT, PU-CU, SAT-CU, SeQ-SAT, SeQ-PU, IQ-SAT, IQ-PU, and TRU-CU paths confirm the excellent predictive power of our model from an empirical perspective. Almost half of the original studies applied at least one of the three theories, but only five original studies guided the practice through the combination of theories. Studies used a single theory may less effective and comprehensive in informing interventions. The theory integration of ECM, ISS, and trust provide a comprehensive understanding of online health management continuance and its determinant factors.

Perceived ease of use of online health management technology is significant, but can’t go beyond perceived usefulness

One of significant findings of this paper was the positive result of PEOU-PU and PEOU-CU. Our result failed to confirm some previous studies which demonstrated PEOU had no significant positive effect on CU.^59,65 Online health management technology being not mature can tend to undergo updates frequently.⁸⁷ The recurring modifications to the interface or functions will deter users from becoming familiar with the technology.³⁶ Thus, maintaining online health management systems ease to use is significant for users’ continuous use. Nevertheless, the relationships strengthen of PEOU-CU were substantially weaker than PU-CU. In line with some previous studies,^37,88 the impact of perceived usefulness on technology continuance is stronger than that of perceived ease of use. The findings indicated that, users are driven to sustain online health management primarily for its functions and secondarily for its ease-to-use system. If online health management cannot perform critically needed healthcare functionality, no amount of ease of use can compensate for it.

Compared with the general population, trust is more notable to young adults’ continuance

With regard to moderating analysis of the age difference, trust had a more notable impact on young people than general population. This finding was confirmed by prior studies.^89–92 Young adults often struggle with mental health issues, sexual behavior, drug and alcohol abuse, smoking, and overeating habits.⁹³ Fear of stigma and negative experiences to face-to-face health consultation will influence their online health engagement.^94,95 For online health management, they pay more attention to service confidentiality and personal privacy than general population. Moreover, there was an announcement that trust is a multi-dimensional, dynamic process rather than a static construct.⁴³ Due to the insufficient data, this paper performed an analysis of trust from an overall perspective. Future research can adopt more detailed dimensions and dynamic stage of trust to pursue a comprehensive understanding.

Compared with health and fitness services, perceived usefulness had a stronger impact to satisfaction on medical services

The moderating results of service type illustrated that perceived usefulness had a stronger positive effect on users’ satisfaction in medical service than in health and fitness service. This finding suggested that medical services of online health management tend to be more useful-focused. Medical services emphasize the usefulness of its functions such as medical information delivery, online booking access, and channels of dialogue with doctors.^46,96 In most cases, solving the disease-related problem is the primary demand for medical services users. When patients encountered disease issues, medical services provide valuable features that can satisfy users' needs and significantly contribute to users' satisfaction.

Moderators included but are not limited to the age difference and service type

Among other path relationships, neither service type nor age difference became a moderator. It illustrates that detailed interventions are not majorly determined by service type, which was supported by a previous study.⁹⁷ Age has been confirmed in previous studies,⁸⁵ whereas, in this paper, only a minor difference was noticed. The finding was unexpected but reasonable. First, the inconsistency with previous studies can be explained by publication bias. Journals have a propensity to publish statistically significant results.⁹⁸ There may exist many insignificant results that resemble ours, which were unreported. For this reason, we included the master’s thesis and the gray literature in our meta-analysis to pursue a comprehensive result. Second, this finding may result from insufficient elderly data on online health management continuance. Among 41 included studies, only two were elderly, and others were young adults or the general population. As “digital refugee,” the elderly have differed from young people on online health usage.⁸⁵ Future researchers should carry out such studies focused on the elderly to support a more nuanced understanding of age differences.

Implication for practice

Integration of theoretical insights enables technology developers and healthcare providers to combine influential factors into practice appropriate to different specific users and technologies. Implementation should be tailored to service types and ages.

Technology developers should not overlook the usefulness of online health management, especially for medical services. Technology developers should ensure the best implementation of online health management systems for users, with an easy-to-use system, high-quality information and service to achieve users’ perception of usefulness and satisfaction. To achieve an easy-to-use system, failure of online access caused by system incompatibility or other technical issues should be avoided. And the interface must not be cumbersome. For the information, technology developers should provide authoritative, relevant, and readable information to increase information quality.⁹⁹ For high-quality service, they should tried to give a timeliness response service for users.¹¹ Studies^100,101 suggested healthcare providers respond to patients within 24 h, if possible within 8 h. Considering that it is difficult for caregivers to be always online, more advanced and sophisticated technologies such as artificial intelligence¹⁰² and data-driven healthcare decision-making systems^103–105 are encouraged in the online healthcare system.

As mentioned above, young people often struggle with mental health and need more privacy. For healthcare providers, they should build a long-term trusting relationship through their abilities, empathy, and high-quality service to preserve young adults’ trust.¹⁰⁶ For technology designers, they should implement privacy-protective with more advanced technology such as the dynamic searchable symmetric encryption (DSSE) technique.¹⁰⁷ Privacy features such as anonymous consultation are also needed.

Implication for future research

The meta-analysis held significant value for advancing theory and implicating future research directions.

Concentrate on continuance theories and import theories from other areas

Firstly, this study proposed and verified a consistent model to mitigate the confusion caused by the inconsistent results of online health management continuance, which contributed to the advancing theory of information technology continuance in a large sense. Secondly, the results of this study presented that some acceptance theories such as TAM and UTAUT were frequently used to explain online health management continuance. However, there was an argument that it is misapplication to employ acceptance theories for information system continuance because there have differences between acceptance and continuance behaviors.¹⁰⁸ The model in our study combined two continuance theories, ECM and ISS, which could be an illustration of how online health management continuance research can be incrementally advanced in a meaningful manner. We believed that future research should concentrate on the continuance theory to explain online health management continuance. In addition to employing classical theories in information continuance domination, Bhattacherjee¹⁰⁸ advocated importing theories from other areas to comprehensively describe information continuance. The results of this study revealed that some theories from other areas, such as use and gratifications theory (UGT), social cognition theory (SCT), and stimulus-organism-response theory (S-O-R), have been employed to try new relationships of online health management continuance. We thus suggest future research widely import theories from other areas such as psychology, sociology, and medicine to present new variables and insights.¹⁰⁹ Thirdly, extracting constructs merely from quantitative results might not enough to understand online health management continuance. Future research could use the influential factors inspired by qualitative research to further enrich the theoretical model.

Implication for future research of study population, content, design, and methods

For the study population, future researchers should carry out studies focused on the elderly to support a more nuanced understanding of age differences. For research direction, trust, service quality, and information quality are multidimensional, hierarchical, and context-specific. This paper only confirmed them from a general perspective, and future research should be carried out to adequately capture multiple dimensions of these factors in online health management continuance. Moreover, the existing data on system quality was not enough to perform the meta-analysis. Future research could carry out more quantitative exploration of system quality to provide available data for future meta-analyses. For study design, existing studies are mostly cross-sectional; more longitudinal studies were expected in the future. For the research methods, online health management continuance is impacted by multiple factors. Different combinations of factors may work differently. Future research could apply other analysis methods such as Qualitative Comparative Analysis (QCA) methods¹¹⁰ in online health management to determine which combinations of factors can explain variations in outcomes of interest.

Limitations

This paper as a reference for future research still had some limitations that need to be clarified. Firstly, some constructs such as perceived risk, habit, and facilitating conditions were not included in our conceptual model, which can be further explored. Similarly, our paper did not analyze other potential moderators such as gender and culture due to the lack of relevant information. These were general defects that were hard to avoid in meta-analysis. Future research can perform on analyzing those potential moderators. Secondly, this paper did not explore the relationship between continuous use intention and actual use. It is a direction left to be further explored. Thirdly, our findings should be interpreted with caution considering the Asian studies and cross-sectional study design of included studies. To generalize our results, more longitudinal studies in diverse countries are needed in the future.

Conclusion

This paper aimed to identify salient determinants of online health management continuous use and understand the moderating effect of age difference and service type. Here are some inspirations learned from the article: (1) There existed complex, dynamic interrelationships among factors of continuous use on online health management. Increasing users’ positive experience in practical applications is the key to its success. (2) A consistent model integrating the Expectation Confirmation model (ECM), Information Systems Success Model (ISS), and trust theory are robust for scholars to understand online health management continuance. (3) Moderators included the age difference and service type but are not limited. Follow-up research should put more effort into the elderly to support a more nuanced understanding of age difference and apply more methods such as Qualitative Comparative Analysis (QCA) to explore the combined effects of different combinations further.

Footnotes

Author’s contributions

Yucong Shen: Data curation, Statistical analysis, Writing-Original draft, Writing-Reviewing and editing.Wenxian Xu: Data curation, Statistical analysis. Andong Liang: Writing-Reviewing. Xinlu Wang: Data curation. Xueqin Lu: Writing-Reviewing. Zhongqiu Lu: Conceptualization, Supervision. Chenchen Gao:Conceptualization, Supervision, Writing-Reviewing and editing.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Zhejiang natural science foundation of China (LQ21G030017), the scientific research projects of Zhejiang Provincial Department of Education (Y201840389) and the Wenzhou science and Technology Bureau (Y20190026).

ORCID iD

Yucong Shen

Appendix

Appendix A

Study quality assessment.

Authors, Year of publication/Questions	Q.1	Q.2	Q.3	Q.4	Q.5	Q.6	Q.7	Q.8	Q.9	Quality
Shahriar Akter,2011³⁰	Y	Y	Y	Y	Y	-	-	Y	Y	A
Shahriar Akter,201311	Y	Y	Y	Y	Y	-	-	Y	N	B
Sekyoung Youm,2015⁴¹	Y	Y	Y	Y	Y	Y	Y	Y	Y	A
Shupei Yuan,2015¹¹¹	N	Y	Y	Y	N	-	-	Y	Y	B
Minggui He,2015⁴³	Y	Y	Y	Y	N	-	-	Y	Y	B
Jian Mou,2016⁴⁴	Y	Y	Y	Y	N	-	-	Y	Y	B
Xi Zhang,2016⁴⁵	N	N	Y	Y	Y	-	-	Y	Y	B
Jaehee Cho,2016¹²	Y	Y	Y	Y	N	-	-	Y	Y	B
Meng Yin,2017⁴⁶	N	Y	Y	N	N	-	-	Y	Y	B
Min Zhang,2017⁴⁷	Y	Y	Y	Y	N	-	-	Y	Y	B
Ardion D,2017³⁷	Y	Y	Y	Y	N	-	-	Y	Y	B
Dongxiao Gu ,2018⁴⁹	Y	Y	Y	Y	Y	Y	Y	Y	Y	A
Jia Yang,2018⁵⁰	Y	Y	Y	U	N	-	-	Y	U	B
Yuan Chen,2018⁴⁸	Y	Y	Y	U	N	-	-	Y	Y	B
Qingxing Dong,2018⁵¹	Y	N	Y	Y	Y	-	-	Y	Y	B
Haoshuang Huo,2018⁵²	Y	Y	Y	N	Y	-	-	Y	Y	B
Yuwei Chang,2019⁵³	Y	Y	Y	Y	Y	-	-	Y	Y	A
Bing Wu,2018¹⁴	Y	Y	Y	Y	N	-	-	Y	Y	B
Yue Chen,2018⁵⁴	Y	Y	Y	Y	N	-	-	Y	Y	B
Hui Li,2019⁵⁵	Y	Y	Y	Y	N	-	-	Y	Y	B
Yaoyao Lin,2019⁵⁷	Y	Y	Y	U	N	-	-	Y	Y	B
Changrui Sun,2019⁵⁸	Y	Y	Y	Y	N	-	-	Y	Y	B
Linglu Yin,2019⁵⁹	Y	Y	Y	U	N	-	-	Y	Y	B
Yimei Zhu,2019⁶¹	Y	Y	Y	Y	N	-	-	Y	Y	B
Jia Li,2019⁵⁶	U	Y	Y	Y	N	-	-	Y	N	B
Yiyu Zhang,2019⁶⁰	Y	Y	Y	Y	N	-	-	Y	U	B
Ziying Hong,2019⁶²	Y	Y	Y	Y	Y	Y	Y	Y	Y	A
Nanyan Cao,2020⁶³	Y	Y	Y	N	Y	-	-	Y	U	B
Dongxiao Gu,2020⁶⁴	Y	Y	Y	N	Y	-	-	Y	N	B
Zizhen Qin,2020⁶⁵	Y	Y	Y	N	Y	-	-	Y	N	B
Hongcheng Cui,2020⁶⁸	Y	Y	Y	Y	N	-	-	Y	N	B
Jaehee Cho,2020⁶⁷	Y	Y	Y	Y	Y	Y	Y	Y	Y	A
Antoine Grenier ,2020⁶⁶	Y	Y	Y	Y	N	-	-	Y	Y	B
Guanxiong Huang,2020¹¹²	Y	Y	Y	Y	N	-	-	Y	Y	B
Md Abdul Kaium,2020¹³	Y	Y	Y	Y	Y	-	-	Y	Y	A
K Anil Kumar,2020⁶⁹	Y	Y	Y	Y	Y	-	-	Y	Y	A
Tomer Shemesh,2020⁷⁰	U	Y	Y	Y	Y	-	-	Y	Y	B
Minsun Shim,2020⁷¹	Y	Y	Y	Y	Y	-	-	Y	Y	A
Reginald A Silver,2020⁷²	Y	Y	Y	Y	Y	-	-	Y	Y	A
Chunyan Xie,2020⁷³	Y	Y	Y	Y	N	-	-	Y	Y	B
Xiaoxiao Zhang,2020⁷⁴	Y	Y	Y	Y	N	-	-	Y	Y	B

Note: Q.1Was the sample frame appropriate to address the target population? Q.2Were study participants sampled in an appropriate way? Q.3Was the sample size adequate? Q.4Were the study subjects and the setting described in detail? Q.5Was the data analysis conducted with sufficient coverage of the identified sample? Q.6Were valid methods used for the identification of the condition? Q.7Was the condition measured in a standard, reliable way for all participants? Q.8Was there appropriate statistical analysis? Q.9Was the response rate adequate, and if not, was the low response rate managed appropriately? “Y” means “yes”; “N” means “no”; “U” means “unclear”; “-” means “not applicable”.

References

Akter

Ray

. mhealth - an ultimate platform to serve the unserved, 2020, https://pubmed.ncbi.nlm.nih.gov/20938579

Data analysis of medical industry: 74.4% of users participated in internet medical services during the epidemic period in 2020, https://www.iimedia.cn/c1061/76038.html (2020).

The 47th China statistical report on internet development, http://www.gov.cn/xinwen/2021-02/03/content_5584518.htm (2021).

Fields

Gandhi

. Tools for population health management. Prim Care: Clin Off Pract 2019; 46: 529–538.

Zettel-Watson

Tsukerman

. Adoption of online health management tools among healthy older adults: an exploratory study. Health Informatics Journal 2016; 22: 171–183. DOI: 10.1177/1460458214544047.

Hernandez Silva

Lawler

Langbecker

. The effectiveness of mHealth for self-management in improving pain, psychological distress, fatigue, and sleep in cancer survivors: a systematic review. J Cancer Survivorship 2019; 13: 97–107. DOI: 10.1007/s11764-018-0730-8.

Bhattacherjee

. Understanding information systems continuance: an expectation-confirmation model. Mis Q 2001; 25: 351–370. DOI: 10.2307/3250921.

IMS Institute for Healthcare Informatics . Patient adoption of mHealth. USA: IMS Institute for Healthcare Informatics, 2015.

Krebs

Duncan

. Health app use among US mobile phone owners: a national survey. JMIR Mhealth Uhealth 2015; 3: e101. DOI: 10.2196/mhealth.4924.

10.

Aljuraiban

. Use of weight-management mobile phone apps in Saudi Arabia: a web-based survey. J Med Internet Res 2019; 7: e12692.

11.

Akter

D'Ambra

Ray

, et al. Modelling the impact of mHealth service quality on satisfaction, continuance and quality of life. Behav Inf Technology 2013; 32: 1225–1241. DOI: 10.1080/0144929x.2012.745606.

12.

Cho

. The impact of post-adoption beliefs on the continued use of health apps. Int J Med Inform 2016; 87: 75–83. DOI: 10.1016/j.ijmedinf.2015.12.016.

13.

Kaium

Bao

Alam

, et al. Understanding continuance usage intention of mHealth in a developing country: an empirical investigation. Int J Pharm Healthc Marketing 2020; 14: 251–272. ahead-of-print.

14.

. Patient continued use of online health care communities: web mining of patient-doctor communication. J Med Internet Res 2018; 20: e126. DOI: 10.2196/jmir.9127.

15.

Sun

Bhattacherjee

. Looking inside the “it Black Box”: technological effects on it usage. Data Processor Better Business Education 2015; 54: 1–15.

16.

Yousafzai

Foxall

Pallister

. Technology acceptance: a meta-analysis of the TAM: part 2. J Model Management 2007; 2: 251–280.

17.

Sumak

Hericko

Pusnik

. A meta-analysis of e-learning technology acceptance: the role of user types and e-learning technology types. Comput Hum Behav 2011; 27: 2067–2077.

18.

Chauhan

Jaiswal

. A meta-analysis of e-health applications acceptance moderating impact of user types and e-health application types. J Enterprise Inf Management 2017; 30: 295–319. DOI: 10.1108/jeim-08-2015-0078.

19.

Liu

. mHealthApps: a repository and database of mobile health apps. JMIR mHealth and uHealth 2015; 3: e28. DOI: 10.2196/mhealth.4026.

20.

Xue

Yen

Chang

, et al. An exploratory study of ageing women’s perception on access to health informatics via a mobile phone-based intervention. Int J Med Inform 2012; 81: 637–648.

21.

Guo

Zhang

Sun

. The privacy–personalization paradox in mHealth services acceptance of different age groups. Electron Commerce Res Appl 2016; 16: 55–65.

22.

Guo

Han

Zhang

, et al. Investigating m-health acceptance from a protection motivation theory perspective: gender and age differences. Telemed J E Health 2015; 21: 661–669.

23.

Lai

. Technology acceptance model for internet banking: an invariance analysis. Inf Management 2005; 42: 373–386.

24.

Borenstein

Hedges

Higgins

JPT

, et al. An introduction to meta-analysis. New York: Wiley, 2009.

25.

Oliver

. A cognitive model of the antecedents and consequences of satisfaction decisions. J Marketing Res 1980; 17: 460–469.

26.

Oliver

. Effect of expectation and discontinuation on postexposure product evaluations: an alternative interpretation. J Appl Psychol 1977; 62: 480–486.

27.

Delone

Mclean

. Information systems success: the quest for the dependent variable. Inf Syst Res 1992; 3: 60–95.

28.

DeLone

McLean

. Information systems success revisited. In: Proceedings of the 35th annual hawaii international conference on system sciences, 2002.

29.

Daghan

Akkoyunlu

. Modeling the continuance usage intention of online learning environments. Comput Hum Behav 2016; 60: 198–211. DOI: 10.1016/j.chb.2016.02.066.

30.

Akter

D’Ambra

Ray

. Trustworthiness in mHealth information services: an assessment of a hierarchical model with mediating and moderating effects using partial least squares (PLS). J Am Soc Inf Sci Technology 2011; 62: 100–116. DOI: 10.1002/asi.21442.

31.

Mcknight

Chervany

Kacmar

. Developing and validating trust measures for e-commerce. Inf Syst Res 2002; 13: 344–359.

32.

Venkatesh

Thong

JYL

. Cousumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. Mis Q 2012; 36: 157–178.

33.

Franque

Oliveira

Tam

, et al. A meta-analysis of the quantitative studies in continuance intention to use an information system. Internet Research, 2020.

34.

Anol

. Understanding information systems continuance: an expectation-confirmation model. Mis Q 2001; 25: 351–370. DOI: 10.2307/3250921.

35.

Delone

McLean

. The DeLone and McLean model of information systems success: a ten-year update. J Management Inf Syst 2003; 19: 9–30.

36.

Beldad

Hegner

. Expanding the technology acceptance model with the inclusion of trust, social influence, and health valuation to determine the predictors of German users’ willingness to continue using a fitness app: a structural equation modeling approach. Int J Human–Computer Interaction 2017; 34: 882–893. DOI: 10.1080/10447318.2017.1403220.

37.

Davis

. Perceived usefulness, perceived ease of use, and user acceptance of information technology. Mis Q 1989; 13: 319–340.

38.

Mayer

Davis

Schoorman

. An integrative model of organizational trust. Acad Management Rev 1995; 20: 709–734.

39.

Borenstein

Higgins

JPT

. Meta-analysis and subgroups. Prev Sci 2013; 14: 134–143. DOI: 10.1007/s11121-013-0377-7.

40.

Youm

. The development and user satisfaction evaluation of internet-based N-screen healthcare walking content to increase continuous usage motivation. Telemed J E Health 2015; 21: 677–685. DOI: 10.1089/tmj.2014.0135.

41.

Yuan

Kanthawala

, et al. Keep using my health apps: discover users’ perception of health and fitness apps with the UTAUT2 model. Telemed E-Health 2015; 21: 735–741. DOI: 10.1089/tmj.2014.0148.

42.

Long

. The study on the effect of perceived risk and trust on continuous usage of mobile medical APP—based on the ECM-ISC. Journalism and Communication Review, 2016, pp. 152–165.

43.

Mou

Cohen

. Trust and online consumer health service success: a longitudinal study. Inf Development 2016; 33: 169–189.

44.

Zhang

Yan

Cao

, et al. The role of perceived e-health literacy in users’ continuance intention to use mobile healthcare applications: an exploratory empirical study in China. Information Technology for Development, 2016, pp. 1–26.

45.

Yin

. Research on the intention of continuous use of mobile app based on the theory of ECT and is success—take the health app as an example. J Dalian Universty Technology (Social Sciences) 2017; 38: 81–87.

46.

Zhang

Luo

Nie

, et al. Analysis of users’ willingness to continue using mobile medical app from the perspective of information ecology. Data Anal Knowledge Discov 2017; 1: 46–56.

47.

Chen

Ynag

Zhang

, et al. Central or peripheral? Cognition elaboration cues’ effect on users’ continuance intention of mobile health applications in the developing markets. Int J Med Inform 2018; 116: 33–45. DOI: 10.1016/j.ijmedinf.2018.04.008.

48.

Yang

, et al. Understanding the role of mobile internet-based health services on patient satisfaction and word-of-mouth. Int Journal Environmental Research Public Health 2018; 15: 1972. DOI: 10.3390/ijerph15091972.

49.

Yang

Xie

, et al. Analysis of the continuous use behavior of mobile medical app patients— from the perspective of trust. J Traditional Chin Med Management 2018; 26: 52–55.

50.

Dong

Zhou

Mao

, et al. An investigation on the users’ continuance intention in online health community———based on perceived value theory. J Mod Inf 2019; 39: 156.

51.

Huo

Zhang

, et al. Social cognition theorybased factors influencing continuing use behaviors of online health community users. Chin J Med Libr Inf Sci 2019; 28: 55–62.

52.

Chang

Hsu

Wang

, et al. Integration of online and offline health services: the role of doctor-patient online interaction. Patient Educ Couns 2019; 102: 1905–1910. DOI: 10.1016/j.pec.2019.04.018.

53.

Chen

. Research on continuance intention of mobile medical applications. In: Master Dissertation. Huazhong University of Technology, 2019.

54.

. Influencing factors assocated with the continuance intention of health management APP for youth group—based on the empirical investigation and analysis of Xi’an City. In: Master Dissertation. Chongqing University, 2019.

55.

Liu

, et al. Users’ intention to continue using social fitness-tracking apps: expectation confirmation theory and social comparison theory perspective. Inform Health Soc Care 2019; 44: 298–312. DOI: 10.1080/17538157.2018.1434179.

56.

Lin

Wei

. Research on influencing factors of sustainable use intention of sports fitness app users. Mathematics Pract Theor 2019; 49: 61–65.

57.

Sun

. Study on influencing factors of college students continuous use intention of sports app in shandong province. Master dissertation. Qufu Normal University, 2019.

58.

Yin

. An empirical study on the continuous intention of sports app users. In: Master dissertation. Yanshan University, 2019.

59.

Zhang

Liu

Luo

, et al. Factors influencing patients’ intentions to use diabetes management apps based on an extended unified theory of acceptance and use of technology model: web-based survey. J Med Internet Res 2019; 21: e15023. DOI: 10.2196/15023.

60.

Zhu

. Research on the factors influencing colloge students’ continuing using intention of health APPs master dissertation. Zhengzhou University, 2019.

61.

Hong

Deng

Zhang

. Examining factors affecting patients trust in online healthcare services in China: The moderating role of the purpose of use. Health Informatics J 2019; 25: 1647–1660. DOI: 10.1177/1460458218796660.

62.

Cao

Zeng

ZHang

, et al. Analysis on the influencing factors of psychological app users’ continuous use intention——Take “one psychology” as an example. Stat Management 2020; 35: 83–88.

63.

Shaanova

Yang

, et al. A study on user’s continuous searching of online health community. Inf Sci 2020; 38: 92–97.

64.

Qin

Huo

. Esearch on the continuous intention of health information seeking via social media———bade on the view of evenue expenditure. J Mod Inf 2020; 40: 66–77.

65.

Antoine

Gerit

Louis

, et al. Investigating patients’ intention to continue using teleconsultation to anticipate postcrisis momentum: survey study. J Medical Internet Research 2020; 22: e22081. DOI: 10.2196/22081.

66.

Cho

Lee

. Post-adoption beliefs and continuance intention of smart device use among people with physical disabilities. Disabil Health J 2020; 13: 100878. DOI: 10.1016/j.dhjo.2019.100878.

67.

Cui

Chen

. Study on the willingness of continuous mobile fitness app users. J Capital Univ Phys Education Sports 2020; 32: 75–96.

68.

Kumar

Natarajan

. An extension of the expectation confirmation model (ECM) to study continuance behavior in using e-Health services. Innovative Marketing 2020; 16: 15–28.

69.

Shemesh

Barnoy

. Assessment of the intention to use mobile health applications using a technology acceptance model in an Israeli adult population. Telemed J E Health 2020; 26: 1141–1149. DOI: 10.1089/tmj.2019.0144.

70.

Shim

. What quality factors matter in enhancing the perceived benefits of online health information sites? Application of the updated DeLone and McLean information systems success model. Int J Med Inform 2020; 137: 104093. DOI: 10.1016/j.ijmedinf.2020.104093.

71.

Silver

Subramaniam

Stylianou

. The impact of portal satisfaction on portal use and health-seeking behavior: structural equation analysis. J Medical Internet Research 2020; 22: e16260. DOI: 10.2196/16260.

72.

Xie

Jia

. An empirical study on the factors affecting elderly users’ continuance intention of shared nurses. Risk Manag Healthc Policy 2020; 13: 1849–1860. DOI: 10.2147/RMHP.S261827.

73.

Zhang

. Continuous use of fitness apps and shaping factors among college students: a mixed-method investigation. Int J Nurs Sci 2020; 7: S80–S87. DOI: 10.1016/j.ijnss.2020.07.009.

74.

Dong

Chen

, et al. Prevalence of suicide attempts among Chinese adolescents: a meta-analysis of cross-sectional studies. Compr Psychiatry 2015; 61: 78–89. DOI: 10.1016/j.comppsych.2015.05.001.

75.

. Evidence based nursing. Second ed. Beijing, China: People’s Medical Publishing House(PMPH), 2012.

76.

Munn

Moola

Lisy

, et al. Methodological guidance for systematic reviews of observational epidemiological studies reporting prevalence and cumulative incidence data. Int J Evidence-Based Healthc 2015; 13: 147–153. DOI: 10.1097/xeb.0000000000000054.

77.

McHugh

. Interrater reliability: the kappa statistic. Biochemia Med 2012; 22: 276–282. DOI: 10.11613/bm.2012.031.

78.

Fredric

Wolf

. Meta-analysis: quantitative methods for research synthesis, 23. Beverly Hills California Sage Publications, 1986, pp. 77–92.

79.

Zeng

Zhang

. Analysis of R and meta. China: Military Medical Science Press, 2015.

80.

Dankmar

Uwe

Ekkehart

, et al. Some general points in estimating heterogeneity variance with the derSimonian-laird estimator. Biostatistics 2002; 3: 445–457.

81.

Hong

Wang

, et al. Understanding the determinants of online review helpfulness: a meta-analytic investigation. Decis Support Syst 2017; 102: 1–11. DOI: 10.1016/j.dss.2017.06.007.

82.

Brand

. File drawer problem. APS Observer 1997; 10, http://works.bepress.com/jay_brand/53/

83.

Rosenberg

. The file-drawer problem revisited: a general weighted method for calculating fail-safe numbers in meta-analysis. Evolution 2005; 59: 464–468.

84.

Jacob

. Statistical power analysis for the behavioral sciences. Taylor and Francis, 1988, p. 2013.

85.

Zhao

Zhou

. What factors influence the mobile health service adoption? A meta-analysis and the moderating role of age. Int J Inf Management 2017; 43: 342–350.

86.

World Health Organization . UN task force on NCDs: scaling-up its efforts to reduce the catastrophic number of premature deaths, 2022, https://www.who.int/news/item/11-05-2017-un-task-force-on-ncds-scaling-up-its-efforts-to-reduce-the-catastrophic-number-of-premature-deaths

87.

iiMedia Research . China mobile healthcare market research report, 2016, https://www.iimedia.cn/c400/49397.html

88.

Davis

Wiedenbeck

. The mediating effects of intrinsic motivation, ease of use and usefulness perceptions on performance in first-time and subsequent computer users. Interacting Comput 2001; 13: 549–580. DOI: 10.1016/s0953-5438(01)00034-0.

89.

Moreno

Molina

. The impact of the transparency policy on university students’ trust and intention of continued use. In: 47th Hawaii international conference on system sciences, 2014.

90.

Zahedi

Song

. Dynamics of trust revision: using health infomediaries. J Management Information Systems 2008; 24: 225–248.

91.

Tao

. Understanding users’ initial trust in mobile banking: an elaboration likelihood perspective, 28. Elsevier Science Publishers B, 2012, pp. 1518–1525. V

92.

Kim

. Trust in health information websites: a systematic literature review on the antecedents of trust. Health Informatics J 2016; 22: 355–369. DOI: 10.1177/1460458214559432.

93.

Patton

Sawyer

Santelli

, et al. Our future: a Lancet commission on adolescent health and wellbeing. Lancet 2016; 387: 2423–2478. DOI: 10.1016/s0140-6736(16)00579-1.

94.

Dorstyn

Denson

. Predictors of mental health service use by young adults: a systematic review. Psychiatr Serv 2016; 67: 946–956. DOI: 10.1176/appi.ps.201500280.

95.

Dunne

Bishop

Avery

, et al. A review of effective youth engagement strategies for mental health and substance use interventions. J Adolesc Health 2017; 60: 487–512. DOI: 10.1016/j.jadohealth.2016.11.019.

96.

Akter

D’Ambra

Ray

. Development and validation of an instrument to measure user perceived service quality of mHealth. Inf Management 2013; 50: 181–195.

97.

Chauhan

Jaiswal

. A meta-analysis of e-health applications acceptance: moderating impact of user types and e-health application types. J Enterprise Inf Management 2004; 30: 295–319.

98.

Rosenthal Robert . Writing meta-analytic reviews. Psychol Bull 1995; 118: 183–192.

99.

Househ

Hossain

Jamal

, et al. A cross-sectional content analysis of Android applications for asthma. Health Informatics J 2017; 23: 83–95. DOI: 10.1177/1460458215627289.

100.

Couchman

Forjuoh

Rascoe

, et al.

E-mail communications in primary care: what are patients’ expectations for specific test results?

Int J Med Inform 2005; 74: 21–30.

101.

Mold

Hendy

Lai

Y-L

, et al. Electronic consultation in primary care between providers and patients: systematic review. JMIR Medical Informatics 2019; 7: e13042. DOI: 10.2196/13042.

102.

Uddin

Stranieri

Gondal

, et al. Rapid health data repository allocation using predictive machine learning. Health Informatics J 2020; 26: 3009–3036.

103.

Santala

Halonen

Martikainen

, et al. Automatic mobile health arrhythmia monitoring for the detection of atrial fibrillation: prospective feasibility, accuracy, and user experience study. Jmir Mhealth and Uhealth 2021; 9: e29933. DOI: 10.2196/29933.

104.

Woodcock

Mittelstadt

Busbridge

, et al. The impact of explanations on layperson trust in artificial intelligence-driven symptom checker apps: experimental study. J Med Internet Res 2021; 23(11): e29386. DOI: 10.2196/29386.

105.

Daley

Ni'Man

Neves

, et al. mHealth apps for gestational diabetes mellitus that provide clinical decision support or artificial intelligence: a scoping review. Diabetic Med 2022; 39: e14735. DOI: 10.1111/dme.14735.

106.

Stunden

Zasada

VanHeerwaarden

, et al. Help-seeking behaviors of transition-aged youth for mental health concerns: qualitative study. J Medical Internet Research 2020; 22: e18514. DOI: 10.2196/18514.

107.

Liu

Fan

, et al. Achieving privacy-preserving DSSE for intelligent IoT healthcare system. Ieee Trans Ind Inform 2022; 18: 2010–2020. DOI: 10.1109/tii.2021.3100873.

108.

Bhattacherjee

Barfar

. Information technology continuance research: current state and future directions. Asia Pac J Inf Syst 2011; 21: 1–18.

109.

Nabavi

Taghavi-Fard

Hanafizadeh

, et al. Information technology continuance intention: a systematic literature review. Int J E-Business Res (Ijebr) 2016; 12: 58–95.

110.

Jordan

Gross

Javernick-Will

, et al. Use and misuse of qualitative comparative analysis. Construction Management Econ 2011; 29: 1159–1173.

111.

Yuan

Kanthawala

, et al. Keep using my health apps: discover users’ perception of health and fitness apps with the UTAUT2 model. Telemed J E Health 2015; 21: 735–741. DOI: 10.1089/tmj.2014.0148.

112.

Huang

Ren

. Linking technological functions of fitness mobile apps with continuance usage among Chinese users: moderating role of exercise self-efficacy. Comput Hum Behav 2020; 103: 151–160. DOI: 10.1016/j.chb.2019.09.013.