Role of perceived ease of use, usefulness, and financial strength on the adoption of health information systems: the moderating role of hospital size

Measurement model

The measurement model includes the relationship of the overall construct with its respective items. Four common measures used for the measurement model when using a structural equation model are item reliability, construct reliability, convergent validity, and discriminant validity.

Reliability and convergent validity

Reliability means the degree to which the result of a measurement or calculation is accurate. Two common types of reliability used in the measurement model are item reliability and construct reliability. The outer loading values are used to assess item reliability, and Cronbach’s alpha and composite reliability are used to assess construct reliability. The threshold value for both reliability measures is 0.7, and a value of 0.6 is also acceptable if the basic criteria of convergent validity are established. Table 3 shows that all the reliability values are within the threshold value limit, indicating that the items’ and constructs’ reliability is established. The next measure for the measurement model is convergent validity, which means how closely a test relates to other tests that measure the same (or similar) constructs. The measure used for convergent validity is AVE. The threshold value for the AVE is 0.5 or above. Table 3 shows that all the constructs have an AVE value greater than the threshold value, indicating that all the constructs are convergently valid.

Discriminant validity

Discriminant validity specifically measures whether constructs are related or not. The Fornell–Larcker criteria, HTMT ratios, and cross-loading are all common measures for discriminant validity. But most social scientists recommend the Fornell–Larcker criteria and HTMT as the most robust measures for discriminant validity. Table 4 shows the Fornell–Larcker criteria for the model. The threshold value for the Fornell–Larcker criteria is that the diagonal square AVE values must be greater than the values of their respective columns and rows. Table 4 of the Fornell–Larcker criteria shows that all the diagonal values are greater than the values of their respective columns and rows, indicating that the constructs’ discriminant validity is established based on these criteria.

The second measure used for discriminant validity is HTMT ratios. The threshold value for the HTMT ratio is that all the HTMT values must be less than 0.85. Table 5 of the HTMT ratios shows the HTMT values of the constructs. Table 5 shows that all the values are smaller than the threshold range of HTMT, which indicates that the constructs have achieved their discriminant validity based on the HTMT scale.

Common method bias

The term “common method bias” refers to a spurious variance that attributes the measurement method rather than the construct that the measures are supposed to represent. It is a significant issue for researchers working with primary data. VIF values reflect the multicollinearity issue of the model and address the common method bias problem. If a model has VIF values less than 3.0, it indicates that the model is free from the issue of common method bias. Table 3 of the reliability, multicollinearity, and convergent validity shows that all the individual construct items have a VIF value less than the threshold value, indicating that the model is free from the issue of common method bias.

Structural model

Below Fig. 2 represents the study’s structural model, which indicates the relationship among all the variable of the study.

Tested structural model the adoption of health information system.

Hypothesis testing and regression analysis

Regression analysis is a very old statistical technique used to estimate the significance level of a cause-and-effect relationship between the two variables. Regression was first coined by the social scientist Francis Galton in the nineteenth century. Different techniques are used in the regression analysis: ordinary least square, Partial least square, most likelihood error estimation, etc. The ordinary least square is used for the secondary data, while the PLS and MLE are used for the primary data. Partial least square is used for the hypothesis testing on a model based on predetermined solid theories, while MLE is used for the model being tested for the first time; as this study is based on pre-developed theories, a partial least square technique was adopted to estimate the hypothesis. Table 6 shows the list of the hypotheses based on the model of this study and their significance level. The table shows six hypotheses, of which three are based on direct relationships, and the rest are based on moderate relationships. The measures used for a relationship’s statistical significance are the T and p values. The threshold value for the t value is 1.96 and above, while the threshold value for the p value is 0.05 or less. From Table 6, it was identified that, among the six hypotheses, four are statistically significant. At the same time, the other two are statistically insignificant, not meeting the basic requirements of the threshold values. While the beta value for each relationship shows the strength of that relationship. Interpretation of the hypothesis’s analysis, as shown in Table 6 are given below.

H₁: FC has a positive relationship with HIS adoption.

The values of regression analysis show that there is positive and significant relationship between FC and the adoption of HIS with a β value of 0.16, a p value of 0.02, a t-statistic of 1.995. These values provide evidence that the assumption about the significant role of financial capability or strength in the adoption of HIS is positive and significant. In other words, FC positively influence the adoption of HIS.

H₂: PEU has a positive relationship with HIS adoption.

The values of regression analysis show that there is positive and significant relationship between PEU and the adoption of HIS with a β value of 0.076, a p value of 0.044 and a t-statistic of 1.998. These values provide evidence that the assumption about the significant role of PEU in the adoption of HIS is positive and significant. In other words, PEU positively influence the adoption of HIS.

H₃: PU has a positive relationship with HIS adoption.

The values of regression analysis show that there is positive and significant relationship between PU and the adoption of HIS with a β value of 0.254, a p value of 0.000 and a t-statistic of 4.16. These values provide evidence that the assumption about the significant role of PU in the adoption of HIS is positive and significant. In other words, PU positively influence the adoption of HIS.

H₄: The size of a hospital moderates the relationship between FC and HIS adoption.

The values of regression analysis show that there is no positive and significant moderating role of hospital size exist on the relationship between FC and the adoption of HIS with a β value of −0.075, a p value of 0.213 and a t-statistic of 1.426. These values provide evidence that the assumption about the significant moderating role of hospital size on FC and HIS adoption is not significant. In other words, hospital size does not moderate this relationship.

H₅: The size of a hospital moderates the relationship between PEU and HIS adoption.

The values of regression analysis show that there exist a positive and significant moderating role of hospital size on the relationship between PEU and the adoption of HIS with a β value of 0.101, a p value of 0.009 and a t-statistic of 2.618. These values provide evidence that the assumption about the significant moderating role of hospital size on PEU and HIS adoption is not significant. In other words, hospital size does not moderate this relationship.

H₆: The size of a hospital moderates the relationship between PU and HIS adoption.

The values of regression analysis show that there is no positive and significant moderating role of hospital size exist on the relationship between PU and the adoption of HIS with a β value of 0.08, a p value of 0.283 and a t-statistic of 1.07. These values provide evidence that the assumption about the significant moderating role of hospital size on PU and HIS adoption is not significant. In other words, hospital size does not moderate this relationship.

Model fitness

In statistical modeling, model fit is essential because it directly affects the validity of inferences obtained from the data and the reliability of findings. For the model fitness, several measures are available in the SmartPLS, like the SRMR, Chi-square, NFI, etc., but most of the researchers recommend the SRMR for the model fitness in the PLS-SEM. When applying a structural equation model based on PLS, a value less than 0.08 is generally considered a good fit. Table 7 shows that the SRMR value is 0.068, which is less than the threshold value, which indicates that the model fitness has been achieved.

R square

R square is a measure used for the computation of the coefficient of determination, which explains how the combined effect of the independent variables causes variations in the dependent variable. R square is the collective effect of the variables, which also denotes the explanation power of the model. The greater the R square value, the better the model explanation power. A value of R square for the primary data greater than 10% is considered a reasonable explanation power. However, a value greater than 40% is considered good for the secondary data. Table 8 shows that the R square value for the said model is 0.126, indicating that the 12.6% variation on the dependent variable is due to the independent variables in this research model.

Predictive relevance of the model

Predictive relevance is an advanced tool used in the SmartPLS to detect the prediction power of a model. According to social scientists, a model with more than zero prediction power based on primary data is considered good. The measure used for the prediction power is Q square. Table 8 shows that the model has a Q square value of 0.052, which indicates a moderate level of prediction power.

IPMA analysis

IPMA stands for importance and performance analysis, an advanced technique used in the SmartPLS. This technique explains the importance and performance of each variable for the variable of interest, known as the dependent variable. According to Table 9, perceived usefulness is the most important and performed variable for the aforementioned model to adapt the IS. The importance value for the PU is 36.5%, and the performance value is 68.29%. The IPMA analysis provides guidelines to the policymakers about the importance of variables and needs to be considered. For example, performance values indicate how well the system works in each category. PEU and Financial Capability have 58.02 and 62.5 values respectively, while PU received a highest score of 67.29, which indicates strong performance. These statistics helps to prioritize areas that require improvement. It emphasizes on importance of PU as the highest priority during the adoption of HIS.

MGA analysis

A multigroup analysis is an advanced technique used in the structural equation model to compare the difference between the two groups about the statistical significance of a relationship. Table 10 compares the statistical differences of the respondents in each relationship based on gender. The table indicates no significant impact of gender. It is a significant sign that there is no difference between the data based on the groups, which shows that the data is homogenous. There is no heterogeneity in the data because heterogeneity is one of the basic assumptions of a proper and efficient regression analysis considered free from the bias of the data.

Table 11 compares the respondents’ differences in each model’s relationship. The table shows no difference due to the designation of any relationship except perceived usefulness to adapting the IS. According to the table, doctors show much behavior about the perceived usefulness of the adaptation of ISs.

Discussion

The IS is an important aspect of technology that is overcoming all the manual management systems in most organizations; the health sector is one among them (Alotaibi and Subahi 2022). There are several factors, according to the literature, which is responsible for the adaptation of the IS (Chen et al. 2020). Six hypotheses were claimed based on this study’s model, among which three are based on direct relationships, and the other is based on moderating relationships. Among these six, only two hypotheses were not supported by the findings of this study, while the other four were supported. The first hypothesis claims that financial capability will lead to a better adaptation of the IS in the healthcare sector of Pakistan. However, the findings of this study support the claimed hypothesis with the β = 0.168; p = 0.023. Several studies from the past literature also have similar findings to this study that financial capability or strength is a significant factor that leads toward adapting ISs (Deepu and Ravi 2021; Shahbaz et al. 2019). The second hypothesis based on direct relationship claimed that perceived ease of use would lead toward adaptation of ISs in Pakistan’s healthcare sector. However, the findings of this study support the claimed hypothesis with β = 0.076; p = 0.044. Although several studies from the past literature have the same findings, perceived ease of use is a sufficient factor for adopting healthcare ISs in different sectors along with the healthcare sector. According to them, people adopt new technology products and services mostly based on how useful and easy to use the new one is (Nikou and Maslov 2021). The third and last direct relationship based supported hypothesis claims that perceived usefulness will lead to better adaptation of ISs in the healthcare sector; however, the findings of this study support the claim that perceived usefulness will encourage the adaptation of ISs with β = 0.254; p = 0.000. Although there are several studies from past researcher which has been conducted in different geographical regions on different IS adaptation other than the healthcare sector, they show their findings also in line with the said study (Chen and Aklikokou 2020; Lin et al. 2012).

The fourth hypothesis claimed that a change in the size of the hospital would moderate the relationship of financial capability with the adaptation of ISs in the healthcare sector. However, from the findings of this study, it was found that the hypothesis is not supported by a statistically insignificant relationship with β = 0.075; p = 0.213. Past literature also claimed that several studies have the same findings as the said study about the said hypothesis (Kuek and Hakkennes 2020). This may be because the hospital’s size does not mean the hospital is more capable of financing or having more profit margin to invest in these technological things (Shahbaz et al. 2019). The fifth hypothesis based on this study’s model claims that the hospital’s size will affect the perceived ease of use impact on the adaptation of ISs in the healthcare sector. However, the findings of this study also support the said argument that the increase in the hospital size perceived ease of use will have more effect on the adaptation of ISs in the healthcare sector with β = 0.101; p = 0.009. Several studies from the past also have the same type of findings; the reason for that how much the size of a hospital is increased it will lead a better communication between the employees and it will indirectly enhance the people’s psyche to understand the said technology to be easy to be used (Dhagarra et al. 2020). The sixth and last hypothesis argues that the size of the hospital will moderate the perceived usefulness impact on the adaptation of the IS in the healthcare sector of Pakistan. However, the study results don’t show consistency with the argument claimed with β = 0.080; p = 0.283. Several studies from the past literature also have the same types of findings, which are in line with the findings of this study (Zhao et al. 2019). The reason for that may be that size of the hospital may not be a significant factor affecting its usefulness. If a product or service is useful, people will use and adopt it regarding the size of that organization (Tsai et al. 2019).

PU and PEU are popular terms, emphasizing that people are more inclined to accept technology that they believe to be useful and easy to use. PU and PEU become significant considerations in the healthcare context because of the complicated tasks and sensitive nature of data, which need the adoption of useful and user-friendly technologies (Huarng et al. 2022). Similarly, financial capability (Endriyas et al. 2023) of a healthcare organization is also critical as the adoption of HIS requires large investments in technological facilities, training, and maintenance. Hospitals might face difficulties in the adoption and integration of HIS if they have financial difficulties. This could impede their capacity to fully utilize HIS for better services. Therefore, consideration of PEU, PU and financial capability are essential determinants for HIS adopting strategies. In addition, the research also investigated the moderating influence of hospital size (Kraus et al. 2021) on the impact of PU, PEU and financial capability on the adoption of HIS. It is clear from the findings that hospital size has a moderating role on the impact of PEU on HIS adoption. In other words, the role of PEU on HIS adoption is dependent on the hospital size. PEU may impact the adoption of HIS in large hospitals more than in smaller hospitals, as large hospitals have more organizational resources and complexities than the smaller ones. Due to the importance of this moderating effect, PEU-enhancing measures must to be adapted according to the specific needs associated with hospital sizes. It is essential to understand the relationship between hospital size and PEU in influencing the adoption of HIS in order to establish tailored strategies and measures according to the various needs and difficulties, faced by different size healthcare organizations. As shown by the findings, there exists no significant moderating role of the hospital size on the relationship between financial capability, PU and the adoption of HIS. The influence of financial capability and PU on HIS adoption is similar in hospitals of various sizes. In other words, PU and funding capacity have a considerable impact on the decision to adopt HIS, regardless of the hospital’s size. The absence of this moderating impact suggests that tactics to increase PU and provide funding for the adoption of HIS should be widely used in all hospital of all size. It emphasizes the significance of these elements as broadly applicable drivers for the HIS adoption across all hospitals. The research has the following implications.

Theoretical implications

The study investigated the role of PU, PEU, and financial strength or capability on the adoption of HIS while considering the hospital size a moderating variable. The research identifies these factors as necessary for the adoption of HIS and broadening the horizon of TAM and RBV. The study integrates TAM and RBV and develop its own theoretical model. Further, the research also makes an important contribution toward the understanding of successful adoption of HIS.

Policy implications

The study offers some important implications for the policy makers, who are involved in the decisions regarding the adoption of HIS. Policymakers need to recognize the critical role that funds play in the adoption of HIS in hospitals and provide them enough funding. It is also necessary for policy makers to provide training and awareness among the healthcare professionals regarding the PEU and PU of HIS. This will provide a clear picture and purpose of the HIS adoption to the employees. Policy makers should adapt tailored approach while selecting and adopting HIS according to the needs and uniqueness of hospitals, e.g., size and financial strength.

Practical implications

The study also provides some essential implications for hospitals. Hospital managers should asses and do careful financial planning about the viability of adopting and implementing a HIS. Hospitals should actively include healthcare workers in the HIS implementation process to improve PEU and PU, and provide training. In order to successfully adopt HIS, hospitals of all sizes should work together and with other healthcare organizations to share experiences, lessons learned, and best practices.