Interoperability Project Plan

Technological Forecasting & Social Change 78 (2011) 650–660

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

Applying the Technology Acceptance Model to the introduction of healthcare information systems

Fan-Yun Pai a,⁎, Kai-I Huang b

a Department of Business Administration, National Changhua University of Education, No. 2, Shi-Da Road, Changhua City, 500, Taiwan b Department of Business Administration, Tunghai University, No. 181 Section 3, Taichung Harbor Road, Taichung, Taiwan

a r t i c l e i n f o

⁎ Corresponding author. E-mail addresses: [email protected] (F.-Y. Pai)

0040-1625/$ – see front matter © 2010 Elsevier Inc. doi:10.1016/j.techfore.2010.11.007

a b s t r a c t

Article history: Received 27 May 2010 Received in revised form 23 October 2010 Accepted 18 November 2010 Available online 16 December 2010

With the rapid development of information systems and advancesinhealthcare technology paired with current concerns arise over patients' safety and how to cure them efficiently, the healthcare information systems are attracting the attention of more and more people. The purpose of this study is to propose a conceptual model, appropriate for the intention to use healthcare information systems, by adopting the system, service, and information qualities covered in the Information System Success Model proposed by DeLone and Mclean [1] as the external variables and integrating the three dimensions of perceived usefulness, perceived ease of use, and intention to use — referred to in Venkatesh and Davis' updated Technology Acceptance Model, TAM [2]. This study first analyzes relevant researches on the intention to use such systems as the basis for the questionnaire design, then conducts questionnaire survey among district hospital nurses, head directors, and other related personnel. After the questionnaires are collected, SEM is used to analyze the data. The analysis shows that the proposed factors positively influence users' intention to use a healthcare system. Information, service and system quality influence user's intention through the mediating constructs, perceived usefulness and perceived ease-of-use. Managerial implications are provided accordingly. Suggestions for introducing healthcare information system are then provided as well.

© 2010 Elsevier Inc. All rights reserved.

Keywords: Information system success model Technology Acceptance Model (TAM) Healthcare information system

1. Introduction

Currently, with the rapid development of information systems and the advancement of healthcare technologies, nurses are often required to learn how to operate relevant care assistance equipments while providing clinical care for patients. As the severity of patients' illnesses increase, nurses must spend more time taking care of them, therefore, lots of scholars assert that how to apply current information technology in assisting healthcare to effectively improve the quality of healthcare service and promote electronic case history has currently become an important subject in healthcare information management [3,4].

In recent years, regarding the factors which may impact the implementation of a healthcare information system, the questionnaire surveys conducted by Hsiao and Chang [5] among a total of 85 regional hospitals found that such factors include among others, the support from the senior management level, the skills of the special committee, and the coordination of organizational resources and user participation. However, Choe [6] claims in research that those factors are among others, user participation, support from the senior management level, training, background of the special committee, and task type. In addition, other scholars point out the following factors also have to be considered while implementing the system, including the nurses' preparation, the coordination among each department and the evaluation on continuous supervising, the acceptance of computers by nurses, and organization and management support. Therefore, how to use information technology to develop healthcare systems is an important subject that deserves lots of attention [7–9].

, [email protected] (K.-I. Huang).

All rights reserved.

651F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

In the past, most researches on healthcare information were about the planning and discussion of hospitals as an entire unit, for example, Tsai et al. [10] studied the factors that influence the information systems in hospitals; or about the brief introduction of healthcare information systems, for example, Chang et al. [3] introduced the system and how to use it, in the context of a specified hospital. None of them focused on the study of users' actual use of the system, while those that come close mainly included qualitative descriptions with a lack of quantity analysis. At the same time, systems concerning patients' safety also deserve more attention, such as the alerting system on patients' life safety, recording system on vital signs and accident notifying system. Therefore, by adopting the system, service and information qualities covered in the Information System Success Model proposed by DeLone and Mclean [1] as the external variables and integrating the three dimensions of perceived usefulness, perceived ease of use, and intention to use referred in Venkatesh and Davis' [2] updated Technology Acceptance Model, TAM, this study is expected to propose a evaluation model appropriate for healthcare information systems, in order to identify the cause and effect relationships between the relevant factors affecting the intention to use information systems and provide reference for hospitals equipped or unequipped with the system to evaluate, improve, and plan.

2. Literature review

2.1. Healthcare information system

This system is known as the healthcare planning system or hospital information system. Its development can be dated back to 1960 when its major functions were limited to administrative management only. After 1970, sizable hospitals gradually set up internal information sectors, and private information companies started to develop high commercial value computer information systems, which contributed to the prosperous development of the healthcare information system [10]. The creation of this system is mainly a set of standards based on healthcare diagnosis, symptoms, cause, healthcare target and measurements. Such computerized programs provide nurses with the necessary contents, healthcare plans, and additional functions including addition, revision, inquiry and printing [11]. In order to get a more efficient system, Simpson and Weaver [12] believes that by integrating the healthcare information system with the hospital system, clinical care and administrative management can be combined to enhance the efficiency of the system.

To appropriately evaluate the efficiency of such systems, many scholars adopt different methods. For instance, Hortman and Thompson [13] carried out open Q&A in both questionnaires and forms to identify users' satisfaction and opinion, while Lee et al. [14] used one-to-one or one-to-many quality interviews to analyze in depth the users' opinion on a system. Lising and Kennedy [15] mainly verified the quality of case history to figure out whether a healthcare process has been recorded completely as they also used the behavioral observation method to get a better idea of the time allocation during the healthcare process. In the recent 5 years, healthcare information system use has mainly been evaluated in the forms of questionnaire surveys, in-depth interviews, individual case studies, material collections. The questionnaire survey method is most widely used, generally targeted at system use satisfaction and attitudes relevance with its major components as the nurses' age, seniority, education, and user satisfaction [16,17]. These researches show that nurses feel positively on the system in these aspects: it reduces paper work, provides healthcare instruction, and is equipped with learning functions; in contrast, they feel negatively in terms of insufficient computers and evaluation contents, disconnecting with other information system, complicated operation procedure, etc. [14]. In recent years, the application of healthcare information systems and relevant research results are fruitful, which can be separated into four major categories: the factors which can impact the input of the system, the structure of the system, the components of the system, and the efficiency of the system.

2.2. Behavior theory and Technology Acceptance Model

In 1975, Ajzen and Fishbein [18] proposed the Theory of Reasoned Action, TRA, which mainly illustrates a person's behavioral tendency, for the purpose of predicting, changing and interpreting an individual's particular behavior. TRA posits that individual behavior is driven by behavioral intentions where behavioral intentions are a function of an individual's attitude toward the behavior and subjective norms surrounding the performance of the behavior. In this theory, attitude and subjective norms are independent of each other and they could each exert indirect influence on an individual's behavior through behavioral intention. Attitude toward the behavior is defined as the individual's positive or negative feelings about performing a behavior. Subjective norm reflects social pressures when an individual is performing a behavior and his perception of whether people important to the individual think the behavior should be performed.

In 1985, Ajzen [19] proposed the Theory of planned behavior, TPB. It is an extension of the Theory of Reasoned Action that strived for a more appropriate prediction and interpretation of behavioral theory. The difference between TPB and TRA is that the former predicts behavior under comparatively less controllable circumstances, while the latter predicts behavior based on the assumption that all behaviors and behavioral motivations are under control. TPB also adds the concept of perceived behavioral control as a third variable. It refers to an individual's perceived ease or difficulty of performing a particular behavior [20]. It is assumed that perceived behavioral control is determined by the total set of accessible control beliefs. In other words, if an individual feels that he obtains more resources and opportunities while the difficulty of performing a behavior is comparatively less, his perceived behavioral control would be stronger.

In order to explore the relationship between the perceived emotions factor and the use of science technology, Davis [21] developed the Technology Acceptance Model, TAM that shows how users come to accept and use a technology and is based on the

652 F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

Theory of Reasoned Action (TRA) and Theory of Planned Behavior (TPB) [22]. TAM assumes that there are two specified beliefs that determine computer usage: perceived usefulness and perceived ease of use, eliminating subjective norms and normative beliefs. The model suggests that perceived usefulness and perceived ease of use influence users' attitudes towards using a new technology. User shows positive feelings about the new technology if he or she believes it is good for his or her job performance, thus users' attitudes towards using a new technology will be more positive. Such attitudes will furthermore influence the user's behavioral intention and actual system use. In addition, external variables can also have some impact on users' internal attitudes, beliefs, and intentions, further influencing the Technology Acceptance Model [23]. Previous studies proved that different external variables actually influence perceived usefulness and perceived ease of use. Hong et al. [24] asserted that the following five external variables influence individual perception: the computer's self-efficacy, the knowledge of the search domain, the relevance, the terminology and the screen design. However, Lewis et al. [25] found the external variables to include the institutional factor, the social factor, and the individual factor.

After a long period of research, TAM has been successfully tested across a wide range of computing technologies, organizational settings, and user populations [26]. Although Hsu and Lu [27] mention that comparative results are mixed, TAM is still one of the most frequently tested models in IS literature. Many scholars have revised the Technology Acceptance Model to enhance its interpretation abilities. They not only revised the structure of TAM, but also added external variables and mediators. By studying the relationships between all the variables, they have created better predicting models [2,28–32]. This study uses the revised TAM proposed by Venkatesh and Davis [2] which includes Perceived Usefulness, Perceived Ease-of-Use and Intention of Use.

2.3. Information system success model

DeLone and McLean [33] created a multidimensional IS success model, which integrates the model of communication developed by Shannon and Weaver [34], and the information impact theory found by Mason [35]. The updated model consists of six interrelated dimensions of IS success: system and information quality, IS use, user satisfaction, individual impact and organization impact. Studies show that system and information quality can influence user satisfaction. The degree of IS use can influence the degree of user satisfaction directly, the individual's performance indirectly, and eventually affect the whole organization.

However, Pitt et al. [36] argue that DeLone and McLean's information system success model did not include a measure of IS service quality. They believe that it is necessary to include IS service quality, and assert that system, information, and service quality together have an impact on IS use and user satisfaction.

Referring to many scholars' arguments in the past, and agreeing with Pitt et al. [36] on the service quality perspective, DeLone and McLean [1] proposed an updated IS success model, by adding the dimension of service quality into the original version. Information, system and service quality may separately or simultaneously affect the two interrelated dimensions of IS use and user satisfaction while these two dimensions directly affect net benefits. This is also the first time subsequence use is introduced into the measuring of the IS success model [37].

3. Research design

3.1. Research model

Based on the purpose of the study, as well as the results of sorting relevant research articles, the factors that may affect the healthcare information system are illustrated in Fig. 1. The study is mainly based on the external variables covered in the IS success model proposed by DeLone and McLean [1], including system quality, information quality and service quality, together with

Fig. 1. Research model.

653F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

perceived usefulness, perceived ease of use, and intention to use as research dimensions demonstrated by Venkatesh and Davis [2] in the updated Technology Acceptance Model (TAM). In addition to extensive literature review, we also conducted in-depth interviews with experts in healthcare institutes and experts familiar with healthcare information systems or general information system to understand the underlying factors influencing users' usage intention of healthcare information system and to modify the proposed model.

3.2. Research hypothesis

Ahn et al. [38] used the Technology Acceptance Model to explore the online and offline features of Internet shopping malls and their relationships with the acceptance behaviors of customers. The results show that the external variables which affected the online features included information, service and system quality. Meanwhile, these variables directly influence perceived usefulness and perceived ease of use. Davis et al. [39] and Venkatesh and Davis [2] pointed out in relevant researches on Technology Acceptance Model (TAM) that information quality positively affects perceived usefulness, in other words, if the information quality of the knowledge management system is good, the output charts would be correct, the output knowledge would be fruitful and could be reused, thus, users believe the system is capable of providing correct information and knowledge. Consequently, the study puts forward hypothesis 1 (H1) based on above related researches.

H1. Information quality is positively related to IS user's perceived usefulness.

In exploring the factors for a successful website, Chou [40] mentioned that service quality includes on-time, professional and personalized service, and it influences perceived usefulness positively. The same was found in the studies of Gefen and Keil [41], as well as Zhang and Prybutok [42]. They updated TAM to be in line with the context of online shopping. The results of their studies show that service quality affects not only customer loyalty, but also the perceived ease of use of the online shopping system. Based on above relevant articles, hypothesis 2 (H2) is put forward as follows:

H2. Service quality is positively related to IS user's perceived usefulness.

Ahn et al. [38] used the Technology Acceptance Model to explore online and offline features of Internet shopping malls. The results show that the external variables which affected the online features include information, service and system quality. These variables positively influenced perceived usefulness and perceived ease of use. Thus, hypothesis 3 (H3) is represented follows:

H3. Service quality is positively related to IS user's perceived ease of use.

Chiou and Fang [43] explored internet users' behavior and found that system quality include design quality, response time, and accessibility. Design quality refers to the inquiry function of the system and file transfer speed. Online response time means how soon the response is and how long the response takes. Accessibility refers to whether the software and hardware of the website are accessible. These have significant impacts on an IS user's perceived ease of use. Thus, the following hypothesis 4 (H4) is put forth:

H4. System quality is positively related to IS user's perceived ease of use.

Hung et al. [44] studied previous research articles on TAM, and found among 39 articles, more than 30 researches claim that user's perceived ease of use affects perceived usefulness positively, in accordance to what Mathieson [45] found in the research of the use of word processing software. Lee and Kim [32] also found a positive relationship between perceived ease of use and users' perceived usefulness. Based on above results, hypothesis 5 (H5) is presented as follows:

H5. Perceived ease of use is positively related to IS user's perceived usefulness.

Tsai et al. [46] analyzed the association between individual motivation and user acceptance of a knowledge management system. The results demonstrate that whether the use of a knowledge management IS can improve users' work performance, productivity and efficiency will affect users' frequency in using the system. The results show a positive relationship between perceived usefulness and users' intention to use. Chiou and Fang [43] explored users' behavior in using internet, and concluded that frequently updating useful information on a website can affect users' willingness to use the website. The study proved the positive relation between perceived usefulness and users' intention to use. Hence, this relationship is hypothesized as follows:

H6. Perceived usefulness is positively related to IS user's intention to use.

Lee and Chao [47] explored researches on hospital employees' use of electronic case histories, and found the users' intention to use electronic case histories were affected by their feelings about whether they are easier to use than the conventional method. The study therefore concludes that perceived ease of use has a positive impact on user's intention to use. Chen et al. [48] studied electronic public service, and the results show that a simplified electronic public service system attracts user to reuse the system. Thus, this leads us to establish the following hypothesis:

H7. Perceived ease of use is positively related to IS user's intention to use.

654 F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

3.3. Sample and data collection

According to the list of healthcare nursing centers published by the Bureau of National Health Insurance (BNHI) in Taiwan in 2008, there are 23 medical centers, 70 regional hospitals, 359 district hospitals. In order to analyze in-depth how hospitals at different levels use the healthcare information system and considering that the number of district hospitals alone meet the sample quantity, the study adopted district hospitals as sample targets. Mainly based on the template proposed by Krejcie and Morgan [50], the sampling survey was conducted in 100 randomly selected district hospitals. Each hospital was given 10 questionnaires on the intention to use healthcare information systems and those were filled in by related nurses. It was not known beforehand whether the hospital already has a functional healthcare information system in place. In the end, a total of 420 questionnaires were returned. Eliminating the incompletely filled-in questionnaires, the remaining valid questionnaires came to a total of 366. The response rate, based on the number of distributed, is 36.6%, however, it is 87.15% based on the number of questionnaires returned. Tests for a non-response bias were carried out by comparing early respondents (responses received within the first 2 weeks) and later respondents (responses received within the third week or later). The analysis indicated the absence of a non-response bias.

3.4. Questionnaire design

We undertook an intensive study of literature of interest to identity existing measures for related constructs. The questionnaire was pilot tested with fifteen industry experts. And we conducted face-to-face discussions with these experts after they completed the questionnaire. We modified, added and deleted questions to refine the survey based on their feedback.

Likert Scale is used in this study as it is the most commonly used measure in scale design, with the 3-point and 7-point Likert scales generally enjoying the largest popularities. However, Berdie [49] addresses this questionnaire design in his study and defends the 5-point Likert scale for the following three reasons. First, in most cases, a 5-point Likert Scale is the most reliable measuring method. Once the questions are over five, it is hard for people to distinguish the right point. Secondly, a 3-point Likert Scale depresses people's strongest and mildest opinion, while a 5-point Likert Scale can express it ideally. Thirdly, a 7-point Likert Scale causes confusion for those people with poor distinguishing ability. Hence, the study adopts the 5-point Likert Scale, with the responses rated as follows: 1 as strongly disagree, 2 as disagree, 3 as somewhat agree, 4 as agree, and 5 as strongly agree.

4. Empirical study and discussion

4.1. Basic information

Among all the respondents, there are only 22 men (6.1%) compared to 344 women (93.9%). A majority of the respondents, a total of 155 people (44.2%), are between 30 and 39 years old. Most of them are nurses, a total of 194 (55.3%), with head nurses (23.4%). They are mainly bachelor holders. In terms of seniority, most of them have more than 5 years of experience, a total of 242 (68.9%). Regarding the source of the healthcare information system, most hospitals purchase the system from outside (66.7%). There are a total of 125 respondents (35.6%) that have been using a healthcare information system for 3 years. A total of 140 respondents (39.9%) once accepted training on the system while another total of 172 respondents (49%) have never received any training courses. Concerning familiarity with the operations of the system, a majority of the samples, a total of 180 (51.3%), are good with it.

4.2. Reliability and validity analysis

The reliability of the questionnaire is measured by Cronbach's coefficient alpha (α). The results from the study show a Cronbach's α score of each dimensional scale: information quality at 0.956, service quality at 0.940, system quality at 0.960, perceived usefulness at 0.963, perceived ease of use at 0.943, intention to use at 0.944, while the construct as a whole is at 0.923. This indicates that the questionnaire has the sufficient homogeneity (internal consistency) by exceeding the acceptable coefficient alpha of 0.90. Table 1 shows the reliability analysis of construct.

The related construct and measurement of the questionnaire are based on the theories covered in previous relevant research articles, meanwhile, the questionnaire is verified by a professor of the Information Management Department in National Chung

Table 1 Reliability and validity analysis.

Construct Reliability KMO

Information Quality 0.956 0.935 Service Quality 0.940 0.917 System Quality 0.960 0.938 Perceived Usefulness 0.963 0.953 Perceived Ease of Use 0.943 0.750 Intention to Use 0.944 0.760 Whole 0.983

655F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

Cheng University, a deputy professor of Medical Information Research Institution in Taipei Medical University, and experts, such as the supervisor of the nursing department in Kuang Tien General Hospital, in order to guarantee the validity of the questionnaire.

Apart from the above, the study adopts factor analysis to measure the construct validity of the questionnaire, applying KMO value in the factor analysis. More higher the KMO value is, more correlating factors the variables share, in turn, more appropriate it is for factor analysis. A KMO value above 0.5 justifies the use of factor analysis, it is not fit for factor analysis otherwise [51].Hence, as Table 1 shows, the KMO value of each variable is above 0.5, indicating each as appropriate for factor analysis as there are some correlating factors among the variables. It also suggests that the questionnaire have sufficient construct validity, as all the factor loadings exceed the acceptable 0.5.

4.3. Characteristics of construct

Table 2 demonstrates an average value of every construct variable. In the information system success model, variables include information quality (mean=3.53), service quality (mean=3.43), and system quality (mean=3.35).Under the Technology Acceptance Model (TAM), variables include perceived usefulness (mean=3.55), perceived ease of use (mean=3.46), and intention to use (mean=3.61). At the study's conclusion, healthcare information system users have responded positively in all the information, service, and system qualities, which indicate the models applied in the study are appropriate.

4.4. Correlation analysis

To analyze the relationship between variables, matrices of Pearson product moment correlation coefficients are used to measure the related index between variables. Samples to be tested are based on the questions of every construct. As Table 3 shows, factors of healthcare information system are correlated with each other positively, with each Pearson correlation coefficient ranging from 0.213 to 0.756.

4.5. Measurement model

Regarding the criteria for evaluating model fit, the study is based on the Bagozzi and Yi [52] proposed preliminary fit criteria, overall model fit, and fit of internal structure of the model.

The evaluation of model fit covered in the study is based upon the following scholars' suggestions on the ideal criteria, for example, Byrne [53] proposed a goodness-of-fit model (as measured by the GFI, Goodness-of-Fit Index), claiming that GFI index must exceed 0.80. According to Gefen et al. [41], it is a basic criterion that both indices of NFI and IFI exceed 0.90 for acceptable model fitness, while the recommended fit values for CFI should be more than 0.90 and AGFI more than 0.80. In general, if the value of χ2/df is smaller than 5, it is considered to be a good fit. Conversely, a RMSEA of less than 0.08 suggests a good fit.

This method is adopted to measure a series of model errors, input errors, or identification problems. We can know whether our loadings are more than 0.50 and in the acceptable range or not from the measurement errors. Table 4 indicates that the loading of each construct is more than 0.50 and there is no negative number, which indicates that all indices are within the acceptable range.

Based on what Hair et al. [54] suggested, the study examines the various goodness-of-fits of the overall model and information observations in three types: Absolute Fit Measures, Incremental Fit Measures and Parsimonious Fit Measures.

4.5.1. Absolute fit measures These measures determine the degree to which the overall model (structural and measurement models) predicts the observed

covariance or correlation matrix. The indices of measures commonly used include: the chi-square statistic, GFI, and RMSEA (Root Mean Square Error Approximation). As Table 5 illustrates, the absolute fit index of the study's overall theory models are: χ2=2895.655, df=695χ2/df=4.166, GFI=0.871, and RMSEA=0.080, indicating that all of them are within acceptable range.

4.5.2. Incremental fit measures The second class of measures compares the proposed model to some baseline model, most often referred to as the null model.

The null model should be some realistic model that all other models should be expected to exceed. The indices of these measures are: NFI (Normal Fit Index), CFI (Goodness-of-fit Index) and AGFI (Adjusted Goodness-of-fit Index). Table 5 yields AGFI=0.884, NFI=0.907,IFI=0.901, and CFI=0.901, indicating that all indices are within applicable range.

Table 2 Mean and variance analysis.

Variables Item Mean Variance

Information Quality 8 3.53 0.74 Service Quality 6 3.43 0.77 System Quality 10 3.35 0.75 Perceived Usefulness 9 3.55 0.76 Perceived Ease of Use 3 3.46 0.75 Intention to Use 3 3.61 0.75

Table 3 Correlation analysis.

Information Quality Service Quality System Quality Perceived Usefulness Perceived Ease of Use Intention to Use

Information Quality 1 Service Quality 0.596 ** 1 System Quality 0.756 ** 0.662 ** 1 Perceived Usefulness 0.244 ** 0.253 ** 0.262 ** 1 Perceived Ease of Use 0.267 ** 0.213 ** 0.306 ** 0.224 ** 1 Intention to Use 0.463 ** 0.378 ** 0.437 ** 0.234 ** 0.300 ** 1

*means pb 0.05; ** means pb 0.01;*** means pb 0.001.

Table 4 Measurement model.

Variables Factor Loadings(λ) CR AVE

Information Quality 0.861 0.854 1. The information covered in the healthcare information system meet my needs. 0.892 2. The healthcare information system can provide correct information. 0.753

Service Quality 0.850 0.855 1. When I am facing difficulty, service people from the information center can help me solve the problems. 0.888 2. Service people from the information center have good service attitudes. 0.771

System Quality 0.850 0.849 1. I can get related information while using the healthcare information system. 0.900 2. The healthcare information system can be linked to or integrated with information from other systems. 0.775

Perceived Usefulness 0.870 0.842 1. The healthcare information system can improve my professional skills. 0.873 2. The healthcare information system can reduce the paper work time. 0.777

Perceived Ease of Use 0.923 0.917 1. I think the healthcare information system is easy to use. 0.953 2. I think the interface of the system is clear. 0.876

Intention to Use 0.923 0.909 1. I am willing to use the healthcare information system. 0.943 2. I am glad to learn new healthcare information systems. 0.868

656 F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

4.5.3. Parsimonious fit measures These measures are sometimes called adjusted fit measures. They can be used to compare models with differing numbers of

parameters to determine the impact of adding additional parameters to the model. Common parsimonious fit measures are the parsimonious normal fit index (PNFI) and the parsimonious goodness of fit (PGFI). Table 5 yields PNFI=0.785, PGFI=0.817, and PGFI=0.640, indicating that all indices are within applicable range.

These measures are mainly used to examine significant differences among the estimated parameters to the model and the reliability of the potential variables of various indices. These can be judged by whether both individual item reliability and potential variables composite reliability (CR) are more than 0.70, and average variance extracted (AVE) more than 0.50, the acceptable range. As the Table 4 Measurement Model demonstrates, the CR of information, service, and system qualities, perceived usefulness, perceived ease of use and intention to use are 0.861, 0.850, 0.850, 0.870, 0.923, and 0.923 respectively, while the AVE for each item is 0.854, 0.855, 0.849, 0.842, 0.917, and 0.909 respectively. All of these values exceed the marginally acceptable range, suggesting a good fit of the internal structure of the model.

4.6. Structural model

Before taking the next step, the study first confirms that every construction has certain reliability and validity, in other words, the study establishes some hypotheses to analyze how variables of information, service and system qualities, perceived usefulness, perceived ease of use and intention to use affect the intention to use healthcare information system together. According to Fig. 2, the hypotheses are tested, and the tested results are sorted into Table 6.

Regarding the associations between information and service qualities and perceived usefulness, the standard coefficient of information quality and perceived usefulness is 0.407 with a p-value of 0.05 which supports H1. Meanwhile, it is also concluded that information quality positively affects the users' perceived ease of use of the healthcare information system. In addition, the standard coefficient of service quality and perceived usefulness is 0.172 with a p-value of 0.05, thus, H2 is supported, reflecting that service quality positively affects the users' perceived usefulness of a healthcare information system.

Concerning the relations between service quality, system quality, and perceived ease of use, the standard coefficient of service quality and perceived ease of use is 0.196 with a p-value of 0.05, thus, H3 is proved, showing that service quality has a positive impact on the users' perceived ease of use of the healthcare information system. The standard coefficient of system quality and

Table 5 Measurement model.

Measures Index Ideal Suggest Value Overall Model N=366

Absolute Fit Measures χ2 − 2895.655 df − 695 χ2/df b5 4.166 GFI N0.80 0.871 RMSEA N0.08 0.080

Incremental Fit Measures AGFI N0.80 0.884 NFI N0.90 0.907 IFI N0.90 0.901 CFI N0.90 0.901

Parsimonious Fit Measures PNFI N0.50 0.785 PCFI N0.50 0.817 PGFI N0.50 0.640

Fig. 2. Model path analysis.

657F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

perceived ease of use is 0.421 with a p-value of 0.05, hence, H4 is supported, displaying that system quality exert a positive influence on users' perceived ease of use of the system.

In terms of the relationship between perceived ease of use and perceived usefulness, the standard coefficient of β21 is 0.394, thus, H5 is supported, translating into that perceived ease of use positively influences the users' perceived usefulness of the system.

As for the associations among perceived usefulness, perceived ease of use and intention to use, the standard coefficient of β31 is 0.387 with a p-value of 0.05, consequently, H6 is supported, showing that perceived usefulness positively impacts users' intention to use the system. The standard coefficient of β32 is 0.498 with a p-value of 0.05, therefore, H7 is supported, representing that perceived ease of use positively affects users' intention to use the system.

4.7. Direct and indirect effects

The effects of variables are grouped into three categories: direct, indirect and overall effect, while the last one refers to the direct effects plus the indirect ones. According to hypotheses 1 and 6, information quality indirectly affects users' intention to use, through the path γ11β31, and the indirect influence is 0.158 (by multiplying the path coefficients — 0.407*0.387), without any direct influence. As a result, the overall influence is 0.158. The indirect influence exerted by the service industry on users' intention to use have 3 paths, γ12β31, γ22β21β31, and γ22β32, with values respectively at 0.067, 0.030, and 0.098 (by multiplying the path coefficients — 0.172*0.387, 0.196*0.394*0.387, and 0.196*0.498 respectively), without any direct influence, hence, the overall influence is 0.195. System quality affects users' intention to use indirectly though two paths γ23β21β31 and γ23β32, with values at 0.06 and 0.210 respectively (by multiplying the path coefficients — 0.421*0.394*0.387 and 0.421*0.498 respectively), without any direct influences, as a result, the overall influence is 0.274. Perceived usefulness directly affects users' intention to use, through path β31, value at 0.387 (path coefficient 0.387), without any indirect effects, therefore, the overall influence is 0.387. Perceived ease of use affects users' intention to use both directly and indirectly, through paths β21β31 and β32, with values at 0.156 and 0.498 (by multiplying 0.394*0.387, plus the path coefficient 0.387), hence, the overall influences is 0.654. The study shows that

Table 6 Result of whole model hypotheses test.

Hypothesis Path Coefficient Result

H1: Information quality is positively related to IS user's perceived usefulness 0.407 ⁎ Supported H2: Service quality is positively related to IS user's perceived usefulness 0.172 ⁎ Supported H3: Service quality is positively related to IS user's perceived ease-of-use 0.196 ⁎ Supported H4: System quality is positively related to IS user's perceived ease-of-use 0.421 ⁎ Supported H5: Perceived ease-of-use is positively related to IS user's perceived usefulness 0.394 ⁎ Supported H6: Perceived usefulness is positively related to IS user's intention to use 0.387 ⁎ Supported H7: Perceived ease-of-use is positively related to IS user's intention to use 0.498 ⁎ Supported

⁎ Means p b0.05.

658 F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

perceived ease of use has the largest impact on users' intention to use, with an overall influence of 0.654, following by perceived usefulness (overall influence of 0.387), next to system quality (overall influence of 0.274), as a result, it is clear that perceived ease of use plays the most important role in healthcare information system.

5. Discussion and conclusion

This study proposes that information quality positively affects perceived usefulness, with a path coefficient of 0.497 and p- value at 0.001, supporting H1. The result is in line with previous studies. DeLone and McLean [1] updated the Information System Success Model based on the Technology Acceptance Model, believing that users' behavioral intention to use will be impacted by each individual's perceived usefulness and attitudes towards the system. When the user's attitude towards the information quality is more positive, the perceived usefulness of information will be higher. As a result, this study asserts that while introducing healthcare information systems, we should emphasize the following aspects: making sufficient information available, having good interface design and ensuring on-time updating of information on the system.

In their research of online shopping, Zhang and Prybutok [42] updated TAM to be in line with the context of online shopping. The results of their studies show that service quality affects not only customer's loyalty, but also the perceived ease of use of the online shopping system. The study once again proves that service quality of the healthcare information system positively influences users' perceived usefulness, with a path coefficient of 0.172 and p-value at 0.001, supporting H2. In addition, service quality also has a positive influence on users' perceived ease of use, with a path coefficient of 0.196 and p-value at 0.001, supporting H3. The result is consistent with what Ahn et al. [38] concluded in their study on using the Technology Acceptance Model to explore online and offline features of Internet shopping malls and their relationships with the acceptance behaviors of customers. The above analysis suggests that when users feel more satisfied with the service quality of the healthcare information system, their perceived usefulness and perceived ease of use will be higher. Therefore, medical centers should not only focus on these influential forces during the system introduction period, but also continuously improve their service qualities. All of these affect users' feelings about the information system. By continuously enhancing its service qualities, the system would be able to reach its potential full performance.

The results of the study demonstrate that system quality positively influences users' perceived ease of use, with a path coefficient of 0.421 and p-value at 0.001, supporting H4. This is in accordance with what Chiou and Fang [43] found in exploring internet users' behavior. Their study concluded that system quality has a significant relationship with perceived ease of use. Their system quality included design quality, response time, and accessibility. Design quality refers to the inquiry function of the system and file transfer speed. Online response time means how soon a response is given and how long the response takes. Accessibility refers to whether the software and hardware of the website are accessible. According to the above analysis, the more the user agrees with the system quality, the more he or she perceive its ease of use, consequently, hospitals should pays more attention to the healthcare information system's stability, its information provided, its information integration ability and its flexibility, so as to improve users' perceived ease of use of the information system.

Perceived usefulness and perceived ease of use significantly affect users' intention to use, with a path coefficient of 0.498 and p- value at 0.001, supporting H6 and H7. This is in line with the results of the Tsai et al. [46] analysis on the associations between individual motivation and user acceptance of a knowledge management system. The results demonstrate that the use of a management information system can improve users' work performance, productivity, and efficiency. In turn, it will affect these users' frequency in using the system. These results show a positive relationship between perceived usefulness and users' intention to use. Lee and Chao [47] performed researches on hospital employees' use of electronic case histories, and found that users' intention to use electronic case histories were affected by their feelings about whether they are easier to use than the conventional method. The study therefore concludes that perceived ease of use has a positive impact on a user's intention to use. In addition, the results show that perceived ease of use has a significant positive impact on perceived usefulness, with path coefficient of 0.394 and p-value at 0.001, supporting H5. This is in line with the results of the Kwon and Wen [22] analysis on the factors affecting social network service use. In short, users' intention to use a system would be influenced positively and directly by their perceived usefulness and perceived ease of use, which in turn displays the absolute importance of perceived ease of use. Therefore, while introducing a healthcare information system into hospital, it is necessary to prompt users to use the system by making the operation ways and interface of the system simple and easy to learn.

659F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

This study integrates TAM and the Information Systems Success Model to justify and extend the Technology Acceptance Theory to healthcare information systems. Although this study makes significant contributions to both academia and practice, there are several limitations which open up venues for further research. There are several factors not discussed that may influence the constructs in TAM and the Information Systems Success Model. For example, speed of response may influence service quality. The effects of the antecedences of these two models, therefore, can be investigated in detail. In addition, only district hospitals were selected as samples to develop and test the proposed model. Future studies should further develop the proposed model and verify the proposed model with broader samples such as medical centers, regional hospitals, and clinics.

References

[1] W.H. DeLone, E.R. McLean, The DeLone and McLean model of information systems success: A ten-years update, J. Manage. Inf. Syst. 19 (4) (2003) 9–30. [2] V. Venkatesh, F.D. Davis, A theoretical extension of the technology acceptance model: four longitudinal field studies, Manage. Sci. 46 (2) (2000) 186–204. [3] M. Chang, J.S. Lin, Y.M. Chang, T.T. Lee, Applications of nursing information systems: Sharing the experience of implementation in a hospital, J. Nurs. 55 (3)

(2008) 75–80. [4] W. Raghupathi, J. Tan, Strategic uses of information technology in health care: A state-of-the-art survey, Top Health Manage 20 (1) (1999) 1–15. [5] J.L. Hsiao, I.C. Chang, An empirical study of establishing nursing care plan systems, J. Inf. Manage. 12 (2) (2005) 27–43. [6] J.M. Choe, The relationship among performance of accounting information systems: influence factors, and evolution level of information systems, J. Manage.

Inf. Syst. 12 (4) (1996) 215–239. [7] T.Z. Li, R.H. Wang, S.S. Hsu, L.F. Chen, H.Y. Chang, Information technology and nursing: Using PDAs in clinical nursing care, J. Nurs. 45 (1) (1998) 69–76. [8] H.L. Lai, L. Chen, Nurse satisfaction with the clinical use of personal digital assistant, TZU CHI Med. J. 15 (2) (2003) 97–103. [9] T.W. Dillon, D. Lending, T.R. Crews, R. Blankenship, Nursing self-efficacy of an integrated clinical and administrative information system, CIN: Computers,

Inform. Nurs. 21 (4) (2003) 198–205. [10] Y.J. Tsai, S. Wu, B.C. Chiang, Exploring factors affecting the performance of hospital information systems, J. Inf. Manage. 11 (2) (2004) 191–210. [11] P.A. Mehmert, C.A. Dickel, R.J. Mckeighen, Computerizing nursing diagnosis, Nurs. Manage. 20 (7) (1989) 24–30. [12] R.L., C.A. Simpson, Administrative application of information technology for nursing managers, in: V.K. Saba, K.A. McCormick (Eds.), Essentials of nursing

informatics, McGraw Hill, Boston, 2005. [13] P.A. Hortman, C.B. Thompson, Evaluation of user interface satisfaction of a clinical outcomes database, Comput. Inform. Nurs. 23 (6) (2005) 301–307. [14] T.T. Lee, C.H. Yeh, L.H. Ho, Application of a computerized nursing care plan system in one hospital: Experiences of ICU nurses in Taiwan, J. Adv. Nurs. 39 (1)

(2002) 61–67. [15] M. Lising, C. Kennedy, An multi method approach to evaluating critical care information systems, Comput. Inform. Nurs. 23 (1) (2005) 27–37. [16] T.T. Lee, T.Y. Lee, K.C. Lin, P.C. Chang, Factors affecting the use of nursing information systems in Taiwan, J. Clin. Nurs. 50 (2) (2005) 170–178. [17] H. Alquraini, A.M. Alhashem, M.A. Shah, R.I. Chowdhury, Factors influencing nurse's attitudes towards the use of computerized health information systemsin

Kuwaiti hospitals, J. Adv. Nurs. 57 (4) (2007) 375–378. [18] M. Fishbein, I. Ajzen, Belief, Attitude, Intentions and Behavior: An Introduction to Theory and Research, Addison-Wesley, Boston, MA, 1975. [19] I. Ajzen, From intention to actions: A theory of planned behavior, in: J. Huhl, J. Bechmann (Eds.), Action-Control: From Cognition to Behavior, 1985, Springer-

Verlag, Heidelberg. [20] I. Ajzen, Attitude structure & behavior, in: A.R. Pratkanis, S.J. Breckler, A.G. Greenwald (Eds.), Attitude structure and function, Lawrence Erlbaum Associates,

Hillsdale, NJ, 1989. [21] F.D. Davis, Perceived usefulness, perceived ease of use, and user acceptance of information technology, MIS Q. 13 (3) (1989) 319–340. [22] O. Kwon, Y. Wen, An empirical study of the factors affecting social network service use, Comput. Hum. Behav. 26 (2) (2010) 254–263. [23] F.D. Davis, R.P. Bagozzi, P.R. Warshaw, User acceptance of computer technology: A comparison of two, Manage. Sci. 35 (8) (1989) 982–1003. [24] W. Hong, J.Y.L. Thong, W.M. Wong, K.Y. Tam, Determinants of user acceptance of digital libraries: An empirical examination of individual differences and

systems characteristics, J. Manage. Inf. Syst. 18 (3) (2002) 97–124. [25] W. Lewis, R. Agarwal, V. Sambamurthy, Sources of influence on beliefs about information technology use: An empirical study of knowledge workers, MIS Q.

27 (4) (2003) 657–678. [26] B. Hasan, M.U. Ahmed, Effects of interface style on user perceptions and behavioral intention to use computer systems, Comput. Hum. Behav. 23 (6) (2007)

3025–3037. [27] C.L. Hsu, H.P. Lu, Consumer behavior in online game communities: A motivational factor perspective, Comput. Hum. Behav. 23 (3) (2007) 1642–1659. [28] M. Igbaria, N. Zinatelli, P. Cragg, A.L.M. Cavaye, Personal computing acceptance factors in small firms: A structural equation model, MIS Q. 2 (13) (1997) 279–305. [29] C. Jackson, S. Chow, R. Leitch, Toward and understanding of the behavioral intention to use an information system, Decis. Sci. 31 (2) (1997) 291–311. [30] H.C. Lucas, V.K. Spitler, Technology use and performance: A field study of broker workstations, Decis. Sci. 30 (2) (1999) 291–311. [31] P. Luarn, H.H. Lin, Toward an understanding of the behavioral intention to use mobile banking, Comput. Hum. Behav. 21 (6) (2005) 873–891. [32] S. Lee, B.G. Kim, Factors affecting the usage of intranet: A confirmatory study, Comput. Hum. Behav. 25 (1) (2009) 191–201. [33] W.H. DeLone, E.R. McLean, Information systems success: The quest for the dependent variable, Inf. Syst. Res. 3 (1) (1992) 60–95. [34] C.E. Shannon, W. Weaver, The Mathematical Theory of Communication, University of Illinois Pres. London, IL, 1949. [35] R.O. Mason, Measuring information output: A communication system approach, Inf. Manage. 1 (5) (1978) 219–234. [36] L.F. Pitt, R.T. Watson, C.B. Kavan, Service quality: A measure of information systems effectiveness, MIS Q. 19 (2) (1995) 173–188. [37] T. McGill, V. Hobbs, J. Klobas, User-developed applications and information systems success: A test of DeLone and McLean model, Inf. Resour. Manage. J. 16 (1)

(2003) 24–45. [38] T. Ahn, S. Ruy, I. Han, The impact of the online and offline features on the user acceptance of Internet shopping malls, Electron. Commer. Res. Appl. 3 (4)

(2004) 405–420. [39] F.D. Davis, R.P. Bagozzi, P.R. Warshaw, Extrinsic and intrinsic motivation to use computers in the workplace, J. Appl. Soc. Psychol. 22 (14) (1992) 1111–1132. [40] J.H. Chou, Using DeLone and McLean model to analyze determinants of portal sites success, J. Inf. Manage. Concepts Syst. Appl. 8 (1) (2006) 109–132. [41] D. Gefen, M. Keil, The impact of developer responsiveness on perceptions of usefulness and ease of use: An extension of the technology acceptance model,

DATA BASE Adv. Inf. Syst. 29 (2) (1998) 35–49. [42] X. Zhang, V.R. Prybutok, An empirical study of online shopping: A service perspective, Int. J. Serv. Technol. Manage. 5 (1) (2004) 1–13. [43] Y.W. Chiou, G.D. Fang, A study of web portal user behavior, Web J. Chin. Manage. Rev. 8 (1) (2005) 43–60. [44] S.Y. Hung, T.P. Liang, C.M. Chang, A meta-analysis of empirical research using TAM, J. Inf. Manage. 12 (4) (2005) 211–234. [45] K. Mathieson, Predicting user intentions: comparing the technology acceptance model with the theory of planned behavior, Inf. Syst. Res. 2 (3) (1991)

173–191. [46] C.H. Tsai, W.H. Chih, W.L. Hsu, Using tam to study the personal motivations and social context factors influence the acceptance of knowledge management

systems-based on high technology firms' samples, Electron. Commer. Stud. 5 (1) (2007) 81–108. [47] W.I. Lee, P.J. Chao, The research of information Technology Acceptance Model of electronic patient record system in health care industry's employees-An

example of central and southern Taiwan, Health Manage. 5 (2) (2004) 243–269. [48] Y.C. Chen, R.N. Shang, T.Y. Ho, S.C. Hesieh, The behavioral intention to use e systems for employees in public utility company: Analysis based on TAM and TTF,

J. E Bus. 10 (1) (2008) 305–327.

660 F.-Y. Pai, K.-I. Huang / Technological Forecasting & Social Change 78 (2011) 650–660

[49] D.R. Berdie, Reassessing the Value of High Response Rates to Mail Surveys, Mark. Res. 1 (3) (1994) 52–64. [50] R.V. Krejcie, D.W. Morgan, Determining sample size for research activities, Educ. Psychol. Meas. 30 (1) (1970) 607–610. [51] H.F. Kaiser, An Index of Factorial Simplicity, Psychometrika 39 (1) (1974) 31–36. [52] R.P. Bagozzi, Y. Yi, On the evaluation for structural equation models, J. Acad. Mark. Sci. 23 (4) (1988) 272–277. [53] B.M. Byrne, Structural equation Modeling with LISREL, PRELIS, and SMPLIS, Lawrence Erlbaum Associates, Mahwah, NJ, 1998. [54] J.F. Hairs, R.E. Anderson, R.L. Tatham, W.C. Black, Multivariate Data Analysis, 5th ed., Prentice-Hall, Upper Saddle River, NJ, 1998.