Determinants of patient behavioural loyalty on primary health centres: Evidence from a cross-sectional study in Indonesia

Service provider and service quality: Service providers affect the service quality of PHCs (Hypothesis 1)

The service quality of the health care provider is a multi-dimensional concept (Ghahramanian et al., 2017) and providing high-quality service in the health care sector is the key to success in achieving better public health outcomes. From the perspective of patients, service providers can be the most influential factor of perceived service quality. Service providers are doctors, nurses, and other staff in hospitals or other health care organizations, and they interact directly with patients. Their interpersonal skill builds patient perception in the quality of health care and shapes the behavioural loyalty of the patients (Mosadeghrad, 2014; Fiaz et al., 2018).

The service provider plays a significant role in both technical and functional quality dimensions. The former refers to the technical precision of the medical diagnoses and treatments according to medical professional specifications and standards, while the latter deals with the way patients receive a health care service (Padma et al., 2010). Physicians are the main actors in diagnosing and deciding on the medical treatment to be delivered to patients and at the same time, they are the dominant person interacting with the patient. As a result, the service provider is among one of the most important determinants of service quality, and both are measured by patients and other stakeholders.

Previous studies have revealed that service provider character and personality affect service quality. For instance, Mosadeghrad (2014) investigated factors that contributed to health care quality in Iranian health care providers. They found that service quality was shaped by the character and personality of patients, doctors, and environmental factors. Moreover, (Atinga & Baku, 2013) studied determinants of service quality of antenatal care in Ghana. The study uncovered how physician attentiveness contributes to a positive impact on service quality. Based on the above discussion, the first hypothesis proposed is as follow:

Service process and service quality: Service process affects service quality of PHCs (Hypothesis 2)

The indicators of a good service process can be how fast patients receive the medical treatment, since they register any interactions between patients and staff in the health care facility (Elleuch, 2008). The service process can also be measured by consistency in terms of quality and standards in health care provision.

A good service process, for example, a shorter waiting list, will improve service quality (Atella et al., 2019). The service process affects service users, but it depends on several factors such as the nature, character of service users, and waiting times (Wood et al., 2009). Previous investigators found a relationship between service process and service quality. For instance, Elleuch (2008) studied Japanese patient satisfaction and found that it is determined by process character (patient-doctor interaction) and physical attributes (settings and appearance).

Service environment and service quality: Service environment affects service quality of PHCs (Hypothesis 3)

The service environment refers to the atmosphere of the health care provider that is created by both the physical environment and the social environment, that shape the patient impression. A good service environment is reflected by a comfortable condition, both outside and inside the building, with a friendly and informative social environment supported by various facilities, and infrastructure, based on policies to improve the quality of PHCs services (Meier & Krug, 2009).

Moreover, the service environment also relates to social experiences gained by patients during their visits to the health care provider. This includes patient-friendly environment, responsiveness, and interaction between patient and health care providers. In their study in Pakistan, Fatima et al. (2018) revealed a positive association between health care service quality aspects, for instance, physical environment and customer-friendly environment, with patient loyalty through patient satisfaction.

Service quality and behaviour loyalty: Service quality of PHCs affects behavioural loyalty of PHCs users (Hypothesis 4)

In the health care sector, increasing access to hospitals through social insurance may lead to increasing patient interest and concern regarding the quality of health care services (Fatima et al., 2018). Notwithstanding, the rapid growth of private hospitals and private clinics provides alternatives for patients in selecting which hospital they want to visit. Thus, patient behaviour loyalty has become an important issue, both in a relatively strict referral system or a relatively less strict referral system. In this issue, service quality seems to be a key factor of the loyalty of patients.

Ngoma and Ntale (2019) believe that loyalty encourages users to repeat their use and provide positive feedback to the other members of the public, that would encourage actual and prospective users. Fatima et al. (2018) classified two types of loyalty, namely, attitudinal and behavioural loyalty. Attitudinal loyalty refers to loyalty created by distinctive sentiments of customers towards a product or service, while behavioural loyalty can be defined as repeating use of certain products or services from the same providers, expanding the use volume and promoting the product or service among other people (Fatima et al., 2018). The present study has a focus on behavioural loyalty.

Patients normally use their prior service experience of using health care services to decide whether they would wish to, or not wish to, visit the health care service again. A body of literature has confirmed that the relationship between service quality and behavioural loyalty is emerging. Fatima et al. (2018) examines the influence of service quality on patients’ satisfaction and patients’ behavioural loyalty in six private hospitals in Pakistan. Those authors unveiled the fact that better quality of health care services may increase patient satisfaction and loyalty. Ambrose et al. (2016) undertook their study in the southern US, and examined the association between service quality and patients’ willingness to recommend a hospital to their relatives and friends. The study revealed that service quality has a positive influence on recommendation behaviour among patients.

The above literature review shows that the relationship between service quality and behavioural loyalty in health care industry is still relatively scanty. Most studies have been conducted in other sectors, for instance in banking industry. Besides, prior studies have been undertaken mostly in developed countries and in a hospital setting. Service quality and behavioural loyalty in the context of PHCs are still uncovered. Unlike hospitals, PHCs have different setting of organisation and limited offered services. For instance, PHCs only provides basic medication for their patients. However, it has an unreplaced role as a gatekeeper to hospital.

Referring to the theoretical framework of the relationship between variables as described previously, the model to be tested in this study is depicted in Figure 1. This research aims to test the aforementioned hypotheses as well as to test the mediating role of service quality on the relationship between internal service factors (service provider, service process and service environment) and behavioural loyalty.

Figure 1. Proposed model of the relationship of service provider, service process, service environment, service quality and behavioural loyalty of patients on primary health centres.

H1–H4 indicate the hypotheses that are tested in this study.

Research design

A cross-sectional study, adhering to the STROBE guidelines (Mardaleta, 2022b), was conducted in 14 out of 23 districts in Aceh Province, Indonesia to assess the determinants of patient behavioural loyalty on PHCs. The PHC users, patients who received the services from PHCs, were interviewed directly. The questionnaire-assisted interviews were conducted by the authors after receiving the consent from the patients. The data collection was conducted between September and December 2020. Due to the coronavirus disease 2019 (COVID-19) pandemic, we were unable to interview all the patients directly and therefore the questionnaires were provided to the patients and were then collected. The patients were provided with the phone number of the investigator and therefore were able to ask the questions to the investigators during completing the questionnaire.

The protocol of this study was approved by Komisi Program Studi Pendidikan Doktor Ilmu Manajemen, Universitas Syiah Kuala, Banda Aceh, Indonesia (No. B/359/UN11.1.1/DIM/TD.06/2020). The patients provided written consent prior to being included in this study. Due to the COVID-19 pandemic, some patients provided verbal consent only, and the interview was conducted in a social distancing manner. In such conditions, the investigators wrote the notes stating that the patients agreed to participate and signed the consent sheet. Such an approach was approved by the Indonesian ethical committee due to force majeure to avoid the transmission of the virus between patients and investigators.

Sample and sampling method

Using purposive sampling method, 14 out of 23 districts in Aceh Province were selected based on regionalisation namely, the Central, South, West, and East region. In total, there are 137 PHCs (Puskesmas) that have Inpatient Unit in Aceh Province. The number of PHCs included in this study was determined using the Slovin minimum sample formula: ($n=\frac{N}{1+N{\left(5\%\right)}^2}),$ where $n=\frac{137}{1+137{\left(5\%\right)}^2}$= 102 PHCs. Therefore, 102 PHCs were randomly selected using Randomizer, an online randomization tool. Although Slovin formula is a back-of-the-envelope calculation to calculate the number people to be included a study, in this present study the formula was used to calculate the number of PHC which could be represented the sample of the study. The patients who received the services from the selected 102 PHCs, were selected proportionally based on the annual number of inpatient patients of each PHC (ranged between 2–10 patients each PHC).

Study variables

To assess the factors associated with behavioural loyalty of PHC users, this study assessed two main variables: exogeneous variable (internal service factors) and mediating variable. In this study, exogeneous variables are also called internal service factors since all the variables are part of the PHCs. The exogeneous variable has three main components: service provider, service process, and service environment. Detailed operational definition of each variable are provided in Table 1. In this study, we also examined the role of service quality as an intervening variable of the relationship between internal service factor and behavioural loyalty. Previous studies have shown a relationship between service quality and loyalty. For instance, Wu (2011) found a direct impact of service quality on re-visit intention of patients in Taiwanese hospitals. Krishna et al. (2016) demonstrated the indirect effect of one health care service quality, namely, empathy, to behavioural intention. Therefore, the role of service quality to mediate the impact of internal service factors on behavioural loyalty is supported by the previous literature.

Study questionnaire and its assessment

The questionnaire was developed based on previous literature (Ahmed et al., 2017; Atella et al., 2019; Fairchild et al., 1997; Fatima et al., 2018; Ferreira & Marques, 2019; Fornara et al., 2006; Lovelock & Wirtz, 2016; Meier & Krug, 2009; Mosadeghrad, 2014; Okoye & Obeta, 2015; Parasuraman et al., 2005; Paulus et al., 2016; Zeithaml et al., 2002). The questionnaire consisted of four parts. The first part collected the demographic data while the last three parts assessed the exogeneous, mediating and endogenous variable domain (see the questionnaire in the Extended data (Mardaleta, 2022a)). For exogeneous, mediating and endogenous variables, the possible answers were provided in the Linkert-scale ranged from “Strongly disagree (score 1)” to “Strongly agree (score 5)”. The raw scores were used in the final analysis.

The convergent validity and discriminant validity tests were conducted to ensure the validity of the questionnaire. Based on previous literature, the validity of the questionnaires is confirmed if the loading factor at convergent validity is higher than 0.5 (Ghozali, 2014). Discriminant validity is verified if the cross loading at the construct is higher than another construct and average variance extracted (AVE) is higher than 0.5 (Ghozali, 2014). The reliability of the questionnaires is considering good if composite reliability (CR) value is higher than 0.5 and the value of Cronbach’s Alpha higher than 0.6 (Malhotra & Morris, 2009). Our validity tests showed that all questionnaires’ items were valid as p-value of convergent validity, discriminant validity and AVE were all > 0.50. All of the items were also reliable because the CR value was > 0.70 and Cronbach’s Alpha was > 0.60. Outer loading of each indicator of all variables ranged between 0.703 and 0.913. These indicated that the items within questionnaire are valid and reliable.

Data analysis and hypothesis testing

The multicollinearity possibility was assessed by using Tolerance dan Variance Inflation Factor (VIF). We found that the VIF value was smaller than 10 indicating no multicollinearity between the domains. To examine the data normality, critical ratio of skewness and kurtosis ± 2.58 were used. Hypothesis testing was carried out by using the structural equation model (SEM). This model was used because this model has several advantages. Firstly, SEM analysis is able to carry out complicated tests of decision-making processes in various public sector management and accounting sciences, and others. Secondly, SEM can be used to address both regressive and dimensional research questions, and it can measure the influence of theoretically existing relationships, including mediating relationships (Ferdinand, 2014, Holbert & Stephenson, 2003). Thirdly, SEM can analyse down to the level of indicators, or find the root of the problem, because it is not limited to observed variables. Therefore, SEM is the most appropriate method to solve complex and difficult problems, as SEM can distinguish empirical data and latent data that can be defined from the error value and loading factor. Lastly, SEM can perform regression analysis, factor analysis, and path analysis. The confirmatory factor analysis was used to determine the value of the loading factor. The initial measurement model was carried out to examine the goodness of fit value. All analyses were conducted using IBM SPSS Amos version 23.

Respondents’ characteristics

During the study, 402 questionnaires were distributed and all of them were returned. However only 389 respondents answered all the questions completely and were included in the final analysis (Mardaleta, 2022a). The characteristics of the respondents are provided in Table 2. More than half (59%) of the respondents were married and there was an equal gender proportion between male and female (49% vs 51%). Most on the patients were aged between 18 and 54 years old and almost half of them (48%) completed the senior high school. Based on the type of occupation, only 5% of respondents were working as civil servant while the rest were working in non-civil servant sector (Table 2).

SEM results

Our assessment of criteria measurement of SEM indicated that the data were normal and there were no outliers indicating the data could be used to test our SEM structural model and the hypotheses. An evaluation of the goodness of fit criteria of a model with several index suitability criteria and a cut off value was conducted in order to ensure whether a model can be accepted or rejected. Our SEM model had a p-value > 0.05, chi-square fit statistics/degree of freedom was less than 2 goodness-of-fit index (GFI), root mean square error of approximation < 0.08, Tucker Lewis index and normed fit index were both > 0.90, parsimonious goodness-of-fit index <1.0 and GFI > 0.90. These suggested that our SEM model was acceptable.

The SEM test was assessed the influence between variables; if the variable has a probability value p < 0.05, the hypothesis is accepted. The results of the SEM analysis are provided in Table 3. It can be clearly seen that not all of the hypotheses are supported by the study. Two components of internal service factors, service provider and service environment, had a positive and significant influence on service quality of the PHCs with p < 0.001 and p = 0.021, respectively. The service process has a positive, but insignificant, impact on service quality. Our data also suggested that service quality, indeed, was a determinant of behavioural loyalty of patients to the PHCs (p = 0.003) (Table 3).

We also assessed the intervening role of service quality. Here, we assessed its role as mediator variable that influences the relationship between the independent variable (internal service factors) and the dependent variable (behavioural loyalty). Our data suggested that service quality did not serve as an intervening variable between internal service factors (service provider, service process, and service environment) and behavioural loyalty of the patients, with p = 0.091, p = 0.230 and 0.260, respectively (Table 4).

Study limitation

This study is not free from limitations. Firstly, only PHCs in Aceh province were included in this study. Therefore, the generalisation of results is limited. Secondly, this study did not take into account patient characteristics as independent variables that may contribute to the different perception of service quality. Therefore, this study suggests a wider research scope for further research. Qualitative-based research is also suggested to explore patient perception in service quality, particularly in the current pandemic situation. Thirdly, it is suggested for further study to compare determinants of service quality between PHCs and private clinics, to gain a more comprehensive understanding of service quality from the perspective of health care users. Lastly, since the COVID-19 pandemic, we were unable to interview all the patients directly and some of the data were collected through the self-administered questionnaire. Therefore this might could cause bias.

Underlying data

Figshare: ‘Determinants of patient behavioural loyalty on primary health centres: evidence from Indonesia’, https://doi.org/10.6084/m9.figshare.19326731 (Mardaleta, 2022a).

This project contains the following underlying data:

Extended data

Figshare: ‘Determinants of patient behavioural loyalty on primary health centres: evidence from Indonesia’, https://doi.org/10.6084/m9.figshare.19326731 (Mardaleta, 2022a).

This project contains the following extended data: