Self-medication in Vietnam: Why do consumers purchase medicines without prescriptions?

Abstract

Abstract: The research aims to establish the aspects that motivate individuals to buy medicines without prescription to self-medicate in Vietnam. A Health Belief Model diagram is used for the research, and a total of 426 valid answers are used for the data analysis. The outcomes prove that Perceived Threat, Perceived Benefits, Perceived Barriers, Perceived Self-efficacy and Modifying factors towards self-medication are the key reasons motivating individuals to buy medicine without prescription (self-medication) in Vietnam. The research shows several issues related to self-medication in Vietnam: Perceived Threat, Perceived Benefits, Perceived Barriers, and Perceived Self-efficacy.

Keywords:

• self-medication
• drug stores
• consumer behaviour
• health belief model
• Vietnam

1. Introduction

People choose to self-medicate for many reasons, such as affordable means of medicines, guidance from relatives and friends, and fear of seeing the doctor (Partha et al., 2002). Self-medication in Vietnam is a broad concept entailing motivating factors such as individual behaviours, attitudes, socioeconomic factors, gender, and age (Hoai & Dang, 2017). According to Awad et al. (2005), self-medication is drug usage by persons to treat self-recognized sickness or signs. In other words, the use of unprescribed medicines by individuals on the grounds of their creativities (Gutema et al., 2011) also referred to self-administration as accessing and using medicine deprived of professional regulation concerning suggestions dosage and treatment period. Nevertheless, self-medication does not essentially mean consuming modern medicines alone and consuming herbs (Partha et al., 2002). Glanz et al. (2008) state that an individual’s beliefs and behaviours contribute to self-medication. As much as individuals like self-medicating, many lack the basic knowledge of using medicines (Kumari et al., 2012). This attracts possible side effects such as vomiting and adverse symptoms and illness. Ha et al. (2019) report that perceived susceptibility among individuals in Vietnam is shallow about self-medication. This low sensitivity often brings about complications due to ignorance and insufficient knowledge about medical complications. Individuals in Vietnam typically choose different forms of self-medication because they want to treat particular illnesses. People opt for self-administration to treat various diseases and are not limited to flu, skin problems, heartburn, cold, headache, and insect bites. Self-treatment is yet common in Vietnam, amounting to 40% to 60% of entire treatments.

However, the health ministry has reported congestion in hospitals, which have been functional at near 150% of their capacity (Hoai & Dang, 2017). The research connects the respondent’s socioeconomic aspects to their decisions on self-administration. The outcomes discovered that residential location, marital status, and gender motivate self-administration in Vietnam (Ha et al., 2019). Predominantly, females, not married, and city inhabitants incline to have more self-medication than other people. Therefore, we intended to systematically review the Health Belief Model for self-medication to gain inclusive information concerning incidences, population samples, most used medicines, targeted diseases, and self-medication reasons in Vietnam. This research is essential for scientists and related parties who want to understand why people are ever motivated to buy medicines over the counter without doctors’ prescriptions. The primary research target is to study the aims of self-medication for Vietnamese people. Besides, the primary research target is to provide an overview of current self-medication behaviour in a developing country such as Vietnam. This research contributes to consumer behaviour analysis and will be a valuable resource to improve and develop health policies for the community.

2. Literature Review

2.1. Health Belief Model in Self-Medication

The Health Belief Model (HBM) was advanced in 1950 to describe changes and maintain health-related behaviours. Since its inception, HBM has remained instrumental as a conceptual framework for studying health-related behaviour among individuals. The model has also been enhanced to offer an intervention outline for health-related behaviour (Champion & Skinner, 2008). In general, HBM can support adopted intervention measures towards a behaviour change. In the psychological realm, HBM intends to explain observed behaviour concerning psychosocial theories. Primarily, HBM is used to explain and predict health-related challenges regarding the uptake of healthcare services. Initially, HBM was created by welfare psychologists in the U.S. in the Community Health Sector. The initial intention for the formation of HBM was to examine and explain people’s widespread failure to engage in activities that negated the prevention and detection of disease (Champion & Skinner, 2008). In the subsequent years, the model was extrapolated to examine individuals’ reactions to symptoms and their related behaviour concerning the identified sickness; the model integrates the social learning theory to explain and predict factors that influence behaviour (Rosenstock et al., 1988). At this stage, the simulation of HBM focused on individual response and adherence to the medical regimen and their behaviour toward self-medication practices (Strecher & Rosenstock, 1997). Over time, the HBM has become practical in understanding public health concerns and developing intervention measures to support quality living.

Besides, HBM has become integral to psychological theories and concepts to understand and diagnose psychological challenges. The HBM is relevant to this research as it will help formulate a conceptual framework for explaining why people in Vietnam prefer self-medication. The researcher will use this model to predict individual self-medication decisions’ probable causes based on several factors. Some of the principal factors in the HBM for self-medication include perceived severity, perceived susceptibility, and expected potential benefits of action (Hoai & Dang, 2017). To further enhance the topic’s comprehension, the study uses quantitative methods to analyze chapter three of the Health Belief Model (Champion & Skinner, 2008) to examine Vietnam’s self-examination behaviour trends. A thorough analysis of the HBM proves that the model is sufficient in analyzing the research phenomenon. Having been used in various academic realms, HBM is more practical in giving an exact purpose and explanation of a study topic. Ochieng (2009) indicates that quantitative methods emphasize the need for studies to have objective measurements and statistical and mathematical numerical analysis of collected data. In perspective, data can be collected through research questionnaires, polls, surveys, or manipulation of existing data using specific computational research adopted for the present study. Furthermore, Leung (2015) states that quantitative research emphasizes collecting arithmetical data and generalizing it to the entire population under study. As such, quantitative data makes research studies valid and reliable. According to Noble and Smith (2015), research validity is significant since it gives credibility to the research findings, conclusion, and recommendations. Hence, the present study will analyze primary data collected from the sample population of the designed survey and secondary data from existing literature related to the study topic (Hussein, 2009). However, most secondary data will be used as references in developing this study. Therefore, this study will analyse data, logic, and mathematical testing models. The diagram below is a summary of the significant concepts of HBM.

2.2. Self- medication

The examination of self-medication determinants is progressively being carried out in developed and developing nations, especially Vietnam. A systematic approach to these factors gives comparatively probable aspects that enhance Vietnam’s self-medication behaviours, with a prevalence in urban centres (Nguyen & Nguyen, 2015). The practice of self-administration is a health problem globally that continues to be addressed and discussed in many states (Haukkala et al., 2000). Adverse types of self-medication can be initiated in previous literature reviews, including usage of unprescribed medicines, medication fill-ups, use of medicines that have been prescribed to relatives and friends, leftovers of medicines from the previous prescription, and exchange of dosage of prescribed medicines (Figueiras et al., 2000). Self-medication can similarly be in-home therapies, traditional remedies, pharmaceutical medicines, and supplements (Haukkala et al., 2000). Research has revealed that countless health challenges are associated with self-medication. Practice is an essential feature motivating people to buy medicine without a prescription. Figueiras et al. (2000) report that people find it easy to buy medicine from the nearest shops instead of travelling long distances searching for the same medicine.

Similarly, people take self-medication as the first form of treatment to cool down their pain. Typically, also, individuals opt to self-medicate because of the availability of home herbal remedies, ignorance of the sickness, the doctor and high fees for doctors, cheap medicine, high cost of original, insufficient information about health, patient confusion and success of other patients who took the same medicine (Ruiz, 2010). Overall, various factors can cause self-medication practice, and such aspects can differ from individual behaviours, ethnicity, and socioeconomic factors to culture, age, and gender (Yousef et al., 2008). The current research examines the robustness of self-medication and explains the motivating features of individuals who do this.

2.3. Factors of the health belief model in self-medication

3. Modifying factors

Glanz et al. (2008) confirm that socio-psychological and demographic might indirectly impact perceptions and affect health-connected behaviour. For instance, sociodemographic features, principally education achievement, are supposed to impact behaviour indirectly by prompting the perception of barriers, benefits, and susceptibility (Thuan et al., 2008; Yousef et al., 2008). Ramalhinho et al. (2014) provide an essential conclusion by linking central factors to self-administration with antibiotics in Vietnam’s highland part. Sociodemographic, ethnicity, gender, and age were the variables used. Ramalhinho et al. (2014) demonstrated that the non-prescription purchase of medicine is powerfully linked to gender, age, and socioeconomic factors such as employment, income, and education.

4. Perceived threats (PTH)

These are dangers that are associated with self-medication. The PTH has two elements: Perceived severity and perceived susceptibility:

4.0.1. Perceived susceptibility (PSU)

According to Champion and Skinner (2008), perceived susceptibility is the certainty about the likelihood of experiencing risks or getting sickness or infection. Despite the wide range of self-medication risks, a familiar theme in the literature is low perceived susceptibility to this practice. Many individuals tend not to see themselves as susceptible to the medicines they take without a prescription (Green & Murphy, 2014). According to Shaghaghi et al. (2014), people will continue to use unprescribed medicine if they do not pose any adverse risks and complications. Individuals do not understand that the risks might show later in stages with many complications.

H1a: PSU is positively related to Self-medication.

4.0.2. Perceived severity (PES)

According to Glanz et al. (2008), perceived severity is the belief of how severe a health disorder can be. There are specific consequences that come with self-medication. Still, many individuals tend to ignore it since there are no signs of risks and severity of the condition, such as antibiotic resistance, skin problem, hypersensitivity, and allergy (Bennadi, 2013). The emotional weightiness of getting a disease or even leaving it untreated comprises clinical evaluation and medical repercussions such as pain, death, and disability (Montastruc et al., 2016). There are also other perceived social severities, such as social relations and effects on job, life, and family. The mixture of seriousness and susceptibility has remained a perceived threat, which every individual should be cautious about before opting for self-medication (Ruiz, 2010). The outcomes are long-lasting and detrimental to one’s health.

H1b: PES is positively related to Self-medication.

4.0.3. Perceived benefits (PBE)

When an individual notices an individual susceptibility to a severe health state, the perceived threats, even if this perception causes a change of behaviour, the individual will be influenced by the perceived benefits of several available benefits for minimizing the threat of the disease (Shamsi et al., 2009) which means that people will opt for self-medication to relieve the pain and reduce its risks. It might be a contagious illness; thus, individuals will self-medicate to curb its spread. Mostly, over-the-counter medicines are misused since the individual can readily get them. According to Jaberee et al. (2009), there are several non-health-related perceptions, such as financial savings. Over-the-counter medicines are much cheaper than prescribed medicines from the doctor. Bennadi (2013) states that self-medication saves on costs compared to professional care. Also, the benefits of not walking for so long searching for suitable drugs and doctors are significant factors in people opting for self-medication. According to Ruiz (2010), individuals perceive self-medication easiest method to prevent and treat illness and symptoms, which do not require doctors, and the patients get instant relief. According to Ruiz (2010), this reduces the pressure on medical services healthcare services, which probably might be unavailable and insufficient. Sometimes medicines can be addictive, and people can opt to use drugs for their pleasure. Therefore, individuals giving maximum susceptibility beliefs are not anticipated to welcome any slightly suggested health actions lest they similarly see the actions as possibly advantageous by minimizing the threat.

H2: PBE is positively related to Self-medication.

4.0.4. Perceived barriers (PBA)

These are negative potentials associated with particular health actions (Glanz et al., 2008). They might perform as obstructions to taking suggested behaviours. A nonconscious cost-benefit analysis happens when people consider the benefits of the anticipated actions with perceived barriers. An individual doing self-medication might undergo a self-conflict. The rationale is that taking medicine could have side effects and be unpleasant and inconvenient, but they will again help relieve the pain (Rahbar et al., 2017). Even if a person perceives a health state as hostile and believes a specific action would efficiently lessen the threat, barriers might stop assignation in health-promoting behaviour. Thus, the combined levels of severity and susceptibility provide force or energy to act and benefit perception minus the barriers that give an ideal path of action (Rosenstock, 1974).

H3: PBA is positively related to Self-medication.

4.0.5. Perceived self-efficacy (PSE)

In a review done by Bandura (1982), Self-efficacy is a condition-specific assurance that individuals can survive situations of high risks deprived of falling back to their earlier behaviours. It can also be said to be beliefs about the individual capacity to perform behaviours that bring out desired (Haukkala et al., 2000). Bandura (1982) did not differentiate self-efficacy from the expected outcomes. Expectations of the outcome are similar, although different from the HBM aspect of perceived benefits. As discussed systematically in chapter eight of Social Cognitive Theory by Bandura (1997), the literature supports the significance of self-efficacy in instigating and preserving behaviour change. To succeed in the change of behaviour, individuals should (as per initial HBM hypotheses) feel susceptible to their existing behavioural outlines (perceived severity and susceptibility) and be sure that a change of a definite type might produce a valued result at a modest price (perceived benefit). Similarly, they should consider themselves skilled (self-efficacious) to overpower alleged barriers to execute an action. Individuals take medicines that match their sickness; fortunately, they opt to do the same when they get sick when they feel better. Pride is said to be the primary cause of self-efficacy (Shaghaghi et al., 2014); patients feel they can do more than enough for themselves without a doctor’s help. The review is done by Mehta and Sharma (2015); extraction of information from the internet, magazines, and periodicals makes individuals courageous enough to treat their sickness. Self-efficacy can lead to habituation, adverse allergic reactions, and injury of the organs (Bandura, 1997).

H4: PSE is positively related to Self-medication.

4.0.6. Cues to action (CUE)

According to Glanz et al. (2008), cues to action are tactics for operational readiness. According to Hochbaum (1958), several early articulating of the Health Belief Model comprised the aspects of cues that can trigger action. Several initial constructions of the HBM involved the aspect of a cue that could initiate actions. For instance, assuming that willingness to explore an act (perceived susceptibility and benefits) might be pushed by other aspects, chiefly through cues to initiate an action, like environmental actions, media publicity or bodily events. It means that individuals get triggered to self-medication by the influence of social media and body reactions. However, they could not empirically research the cue’s role or systematically study cues to act. Undeniably, though the idea of cues as prompting devices is pleasing, cues to actions are hard to review in descriptive studies; a cue could be as brief as sneezing or the hardly aware of posters perception.

H5: CUE is positively related to Self-medication.

4.1. Individual behaviors in self-medication

According to Beck and Ajzen (1991), Individual behaviours are subjective to the likelihood that they will produce a particular outcome that might be favourable or unfavourable. Individuals are motivated by certain factors to carry out recommended behaviour. People will be motivated by cheap medicine or its easy accessibility to opt for self-medication. The study carried out by Trafimow and Fishbein (1995) illustrates three vital factors that determine a person’s behavioural intention, which results in performance. First, if an individual has solid behavioural intentions to do something, he requires skills and knowledge to bring out the behaviour. Secondly, there ought to be very little or no environmental constraints that can somewhat hinder or make it impossible and difficult for someone to carry out or make behavioural performance. Third, the behaviour should be relevant to the individual (Becker, 1974a, 1974b). Behaviour also is connected to self-efficiency. Barbara (2008) stated that perceived self-efficiency by the people of Vietnam contributes to self-medication. According to Ruiz (2010, p. 318), people are believed to self-medicate because they do not take their sickness seriously; they do not get any simple justification to take their illness for treatment in the hospital. Carrasco-Garrido et al. (2008) indicated that some age-related issues are common among peers, and many symptoms are triggered by ageing. Thus, it is useless to see a doctor. As Panda et al. (2016), during their interview with participants, one elderly respondent reported that he was diagnosed with high blood pressure ten years ago and prescribed some pills to take every day. He said it is tiring to see a doctor each after drugs are over for others, yet he can get them from the pharmacy. There are logical concerns about self–medication (Haukkala et al., 2000), such as long waits to see a doctor and lengthy procedures when taking tests and treatments. This context has employed various experimental studies, the quantitative method following the research to reveal the significant self-medication factors in Vietnam.

5. Research methodology

The methods apply descriptive statistical analysis, Cronbach’s Alpha scale, exploratory factor analysis (EFA) and linear regression model. The data were analyzed by SPSS 20.0 software. The sample size is diversified with various individuals like students, employed people, and pharmacists. English and Vietnamese languages were used in the questionnaires. The information obtained was to be used only for research purposes to ensure that participants’ privacy was prioritized. Before answering the questionnaire, participants can read the disclaimer information, and the academic council has approved the topic of the university. The sample size was 426 participants in the survey. Hair et al. (1998) argued that they would take at least five times the number of samples compared to the total number of variables. Gagne, and Hancock (2006) suggests that the above number of samples is appropriate and accurate for factor analysis. More specifically, the formula for taking sampling is n = 5 * m, in that m means the total number of questions in the research. There are 31 questions in the questionnaire. Therefore, the representative samples for this thesis are n = 5 * 31 = 155, so 426 samples are collected.

Consequently, the offline survey was conducted in Da Nang, Ha Noi, and Ho Chi Minh cities at the drug stores. This offline process ended once we got adequate evidence of a wide range of substantial drug stores. The online survey was also presented via Google Form, where the participants answered the questions posed to them using the link we provided. Residents of developing countries opt to buy medicines on their own instead of visiting the doctor. The questionnaire survey employs the Likert-5 scale to measure the agreeableness of respondents from 1 (total disagree) to 5 (totally agree).

6. Results

Demographic data is the general information that can describe the characteristics of the respondents who participate in answering this research questionnaire (Table 1). In this study, five questions were used to determine. Online and offline surveys conducted a total of 481 questionnaires. Out of them, 426 questionnaires were valid (hold 88,6%). The final data was analyzed by SPSS software.

Table 1. Demographic information

		Count	Column N %
Age	From 18 to 25	174	40.9%
	From 26 to 45	161	37.8%
	Above 45	91	21.4%
	Total	426	100.0%
Gender	Male	195	45.8%
	Female	231	54.2%
	Total	426	100.0%
Income	Below 5 million VND	55	12.9%
	From 5 to 10 million VND	99	23.2%
	From 11 to 15 million VND	139	32.6%
	Above 15 million VND	133	31.2%
	Total	426	100.0%
Personality	Rational	195	45.8%
	Emotional	231	54.2%
	Total	426	100.0%
Educational level	Under college	102	23.9%
	College and University	227	53.3%
	Postgraduate	97	22.8%
	Total	426	100.0%

According to Tavakol et al. (2011), the authors tested the scales with Cronbach’s Alpha coefficient (higher than 0.6) to conclude whether the scales are reliable and the extent to which the items are related. Table 2 below shows values tested using Cronbach’s Alpha.\

Table 2. Reliability test

Items/Scales	Scale Mean if Item Deleted	Scale Variance if Item Deleted	Corrected Item-Total Correlation	Cronbach’s Alpha if Item Deleted
Reliability test of “PSU” with α = 0.872
PSU1	7.7911	1.766	.723	.775
PSU2	7.8028	1.839	.715	.783
PSU3	7.8146	1.846	.700	.797
Reliability test of “PES” with α = 0.873
PES1	11.7418	3.580	.631	.873
PES2	11.9061	2.895	.792	.810
PES3	11.8756	3.074	.764	.822
PES4	11.8709	3.115	.730	.836
Reliability test of “PBE” with α = 0.864
PBE1	14.7019	4.789	.682	.836
PBE2	14.8286	4.848	.642	.845
PBE3	14.9507	4.442	.677	.838
PBE4	14.9178	4.358	.756	.816
PBE5	14.8075	4.730	.666	.839
Reliability test of “PBA” with α = 0.877
PBA1	14.6831	5.459	.718	.850
PBA2	14.9155	4.600	.764	.840
PBA3	14.8286	4.985	.781	.833
PBA4	14.6714	5.746	.685	.859
PBA5	14.8920	5.320	.631	.870
Reliability test of “PSE” with α = 0.890
PSE1	14.7465	7.352	.781	.857
PSE2	14.7019	7.678	.661	.882
PSE3	14.7160	7.973	.684	.879
PSE4	14.9155	6.553	.832	.842
PSE5	14.9765	6.574	.741	.869
Reliability test of “CUE” with α = 0.819
CUE1	11.6667	2.670	.577	.800
CUE2	11.7089	2.494	.625	.779
CUE3	11.7629	2.539	.626	.778
CUE4	11.7629	2.318	.736	.725
Reliability test of “IBE” with α = 0.792
IBE1	15.8826	3.614	.607	.740
IBE2	15.9061	4.076	.459	.788
IBE3	15.9108	3.705	.633	.732
IBE4	15.9742	4.025	.511	.771
IBE5	15.9601	3.888	.661	.727

Based on the above table, we can see that all scales have Cronbach’s Alpha coefficients over 0.7 and can be considered appropriate. Besides, the items of each scale have corrected Item-total correlation over 0.3. Furthermore, Cronbach’s Alpha If Items deleted of each item is higher than Cronbach’s Alpha. Therefore, these scales are reliable. Analysis of factors EFA will help reconfigure groups of scales, check convergence and differentiation of groups of variables, and help eliminate complications to improve research results. So, the primary purpose of EFA is to examine the relationship between observed variables and group them into groups of explanatory variables for factors (Rahn, 2014).

In Table 3, the results achieved KMO coefficient = 0.773 > 0.5 (Velicer, W. F., & Jackson, D. N. 1990) and Barlett’s test value is 6932.771 with significance level Sig = 0.000 < 0.05, showing that observed variables belonging to the same factor are strongly correlated with together. At the same time, the total variance extracted is 66.623%> 50%, showing that five factors explain 66.623% of the variation of the data set, and the Eigenvalues are 4.699; 4.189; 3.527; 3.243 and 1.664, respectively, and they are all higher than one qualifies for factor analysis. As it is shown in the table that all factors loading are higher than 0.5, which means the observed variables achieve convergence value. Besides, no observed variables with factor loading values load more than two factors, so the observed variables reach differentiated values. In conclusion, the statistical coefficients are satisfactory to prove that the observed variable has practical significance and can be used to build a model to test the initial hypothesis (Table 3).

Table 3. Results of EFA for independent variables

Factors	Component
	1	2	3	4	5
PES2	.867
PES3	.835
PSU1	.828
PES4	.823
PSU2	.778
PSU3	.769
PES1	.667
PSE4		.888
PSE1		.868
PSE5		.834
PSE3		.803
PSE2		.788
PBA3			.868
PBA2			.855
PBA1			.827
PBA4			.807
PBA5			.748
PBE4				.838
PBE1				.782
PBE3				.778
PBE2				.776
PBE5				.772
CUE4					.856
CUE3					.774
CUE2					.756
CUE1					.735
Initial Eigenvalues	4.699	4.189	3.527	3.243	1.664
Rotation sums of squared loadings	17.266	30.844	44.042	56.617	66.623
KMO					0.773
Sig of Bartlett’s test					0.000

The results in Table 4 show that KMO is 0.817 > 0.5. So, it is suitable for this study. Barlett’s test value is 598.924, with a significance level Sig = 0.000 < 0.05; therefore, EFA analysis for the dependent variable was statistically significant. The eigenvalue is 2.761 > 1, so it certainly meets the condition. The cumulative percentage equals 55.212%>50%, which means the extracted factor (IBE) could explain 50% of the variation in the data set (Widaman, 1990)

Table 4. Results of EFA for the dependent variable

	Component
	1
IBE5	.813
IBE3	.790
IBE1	.772
IBE4	.694
IBE2	.630
Initial Eigenvalues	2.761
Rotation sums of squared loadings	55.212
KMO	0.817
Sig of Bartlett’s test	0.000

Table 5 shows the results of testing the correlation between 5 independent variables and one dependent variable. There are two critical indicators to analyze. They are r coefficient: Firstly, the r coefficient was positive with the correlation of 5 independent factors (PTH, PBE, PBA, PSE and CUE) and one dependent factor (IBE). That means concluding that as each independent factor increases, so does the dependent factor (IBE). Secondly, Pearson’s r level between predictors (independent factors) and dependent factors (IBE) ranged from 0.3 < r < 0.5 (Cohen J., 1988). It is proved that they correlate.

Table 5. Also shows that all factor loadings are higher than 0.5, which means the observed variables achieve convergence value. Also, no observed variables with factor loading values load more than two factors, so the observed variables reach differentiated values. After analyzing EFA, we find that the scales reach convergent and discriminant values. Therefore, they can be used for further research, as shown below

Table V: Correlations
		PTH	PBE	PBA	PSE	CUE	IBE
PTH	Pearson Correlation	1	−.025	−.108*	.062	.043	.305**
	Sig. (2-tailed)		.611	.026	.204	.373	.000
	N	426	426	426	426	426	426
PBE	Pearson Correlation	−.025	1	−.032	.022	.415**	.417**
	Sig. (2-tailed)	.611		.516	.654	.000	.000
	N	426	426	426	426	426	426
PBA	Pearson Correlation	−.108*	−.032	1	.006	−.025	.339**
	Sig. (2-tailed)	.026	.516		.897	.600	.000
	N	426	426	426	426	426	426
PSE	Pearson Correlation	.062	.022	.006	1	−.035	.392**
	Sig. (2-tailed)	.204	.654	.897		.477	.000
	N	426	426	426	426	426	426
CUE	Pearson Correlation	.043	.415**	−.025	−.035	1	.450**
	Sig. (2-tailed)	.373	.000	.600	.477		.000
	N	426	426	426	426	426	426
IBE	Pearson Correlation	.305**	.417**	.339**	.392**	.450**	1
	Sig. (2-tailed)	.000	.000	.000	.000	.000
	N	426	426	426	426	426	426

Sig. 2- tailed level: It can be seen that the Sig. 2-tailed between each independent and dependent factor is 0.000 (less than 0.05), which means that the correlation results are statistically significant. In addition, it is enough conditions to make a regression analysis.

From Table 6, the adjusted R² (Adjusted R-square) is 0.645 > 0.5. This statistic means that five independent, conditional-responsive variables can explain 64.3% of the change in the dependent variable (IBE). 1 < Durbin-Watson <3 is in the acceptance zone. In this test, the Durbin-Watson coefficient is 1.601. Therefore, there is no relationship between the residuals.

Table 6. Regression analysis results

Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig.	Collinearity Statistics
		B	Std. Error	Beta			Tolerance	VIF
1	(Constant)	−1.503	.200		−7.497	.000
	PTH	.268	.025	.316	10.808	.000	.980	1.020
	PBE	.260	.029	.288	9.016	.000	.824	1.214
	PBA	.330	.025	.388	13.308	.000	.987	1.013
	PSE	.271	.021	.375	12.902	.000	.993	1.007
	CUE	.318	.030	.340	10.642	.000	.823	1.215
Adjusted R^2: 0.643
Durbin–Watson: 1.601
ANOVA^a (sig.): 0.000
Dependent Variable: IBE

ANOVA sig (F) = 0.000, which is less than 0.05; each independent variable has a linear relationship with the dependent variable. The independent variables can explain the variation of the dependent variable IBE. The VIF magnification coefficients of PTH, PBE, PBA, PSE and CUE are less than 10, so the multicollinearity phenomenon does not occur (Hair et al., 1998) Therefore, the relationship between the independent variables does not affect the interpretation of the multiple linear regression model. So, with all results, we see that the regression model is consistent and statistically significant. We have a regression model with unstandardized beta coefficients (Dawn L. and Gilbert A., 2010):

$\mathbf{I}\mathbf{B}\mathbf{E}\mathbf{=}\mathbf{-}\mathbf{1.503}\mathbf{+}\mathbf{0.268}\hspace{0.17em}\mathbf{P}\mathbf{T}\mathbf{H}\mathbf{+}\mathbf{0.260}\hspace{0.17em}\mathbf{P}\mathbf{B}\mathbf{E}\mathbf{+}\mathbf{0.330}\hspace{0.17em}\mathbf{P}\mathbf{B}\mathbf{A}\mathbf{+}\mathbf{0.271}\hspace{0.17em}\mathbf{P}\mathbf{S}\mathbf{E}\mathbf{+}\mathbf{0.318}\hspace{0.17em}\mathbf{C}\mathbf{U}\mathbf{E}$

T-test or one-way ANOVA is a method that can be used to compare the mean of a qualitative factor (Modifying factors) with one or more quantitative factors (IBE factor).

The Independent Samples Test with the Sig is 0.000 < 0.05. As a result, we conclude that there is a difference between males and females in Individual behaviours. According to Table 7, we can see that males’ behaviours are higher than females’ (4.1887 > 3.8069). Besides, there is also a difference between Rational and Emotional Personality in Individual behaviours. Table 7 shows that the mean value of Rational Personality is lower than Emotional (3.6400 < 4.2701). The one-way ANOVA test shows a difference among Income groups and Education levels in Individual behaviours (sig = 0.000 < 0.05). However, we have not found a difference among groups of Age in Individual behaviours (sig = 0.592 > 0.05).

Table 7. The results of the T-test and one–way ANOVA

Modifying factors		Mean	Test value	Sig.
Age	Under 18	3,9477	F = 0.637	0.592
	From 18 to 25	4,0312
	From 26 to 45	3,9578
	Above 45	3,9890
Income	Below 5 million VND	4.1527	F = 13.959	0.000
	From 5 to 10 million VND	4.1111
	From 11 to 15 million VND	4.0144
	15 million VND	3.7805
Educational level	Under college	4.1765	F = 92.535	0.000
	College and University	4.1004
	Postgraduate	3.4990
Personality	Rational	3.6400	t = −17.908	0.000
	Emotional	4.2701
Gender	Male	4.1887	t = 8.923	0.000
	Female	3.8069

7. Conclusion and Discussion

The study’s main objective is to use the Health Belief Model to study factors influencing the self-medication behaviour of people in developing countries, particularly Vietnam. Based on the research results collected by the authors, the findings from this study can enhance further self-medication research projects in Vietnam and other countries in the world. According to the structure of the Health Belief Model, four main factors influence consumer self-buying behaviour, namely Perceived Threats (PTH), Perceived Barriers (PBA), Perceived Self-Efficacy (PSE) and Cues to Action (CUE). Research results show that all four hypothetical factors are accepted. Based on the analysis results, the factor “PBA” has the most substantial impact with b = 0.388. It can be seen that the barriers to medical examination are the reason why consumers decide to buy drugs themselves for treatment. Hospitals should optimize the registration process and make treatment more convenient. This finding is consistent with the results from Nasir et al. (2020), which prove that the scarcity of drugs and healthcare support are the common causes of self-medication during the COVID-19 outbreak in Dhaka city. The second place is “PSE” with b = 0.375, which shows consumers feel the effectiveness when buying drugs themselves. Self-medication is a common practice in Asia as it provides a low-cost alternative for people (Hoai & Dang, 2017; Nepal & Bhatta, 2018; Okumura et al., 2002; Yeika et al., 2021). The third factor is “CUE” with b = 0.340; consumers greatly influence society on self-buying behaviour. The two factors with the weakest impact are “PTH” and “PBE”, with b = 0.316 and b = 0.288. Consumers perceive the risk factor as relatively low, a severe problem because they do not fully understand self-treatment harms. It shows that consumers do not appreciate the benefits of self-medication; they still have confidence in professional medical services. Consumers perceive self-medication involves inappropriate and injudicious use of medicines to treat self-recognized symptoms by the people (Nasir et al., 2020)

This study has some advantages, which show that men have higher self-buying behaviour than women (4.1887 > 3.8069). Consumers in Vietnam also buy drugs based on Emotional Personality rather than Rational Personality (3,6400 < 4,271). This finding is consistent with many previous findings (Stewart et al., 2010; Wang et al., 2019). The one-way ANOVA test shows a difference among Income groups and Educational levels in self-medication (sig = 0.000 < 0.05). This finding is supported by Al-Azzam et al. (2007) and Schmid et al. (2010).

However, we have not found a difference among groups of Age in Individual behaviours (sig = 0.592 > 0.05), and it is consistent with previous studies (Tamseel et al., 2022; Yousef et al., 2008). However, Awad et al. (2008) report that self-medication use increased with age and differed between males and females in Kuwait. These facts call for further research on the age and gender differences in self-medication practices worldwide.

This study has shown some significant disadvantages related to the access and use of drugs in Vietnam. The author provides an overview of self-medication in Vietnam based on the research results. The limitations of access to prescription drugs and self-medication behaviour have been found. Vietnam has an incomplete control system related to drug trading; drug lists should be thoroughly researched before being licensed to be widely sold in drugstores. Hence, consumers use drugs safely, and the government can control purchases to ensure people’s health. Currently, policies and regulations on drug purchase and sale have not been implemented ineffectively. Consumers quickly buy many particular medicines, even antibiotics, at drug stores. This happens very commonly in society in Vietnam, especially in rural areas. Besides the price barrier when buying prescription drugs, the health insurance law should optimize people’s benefits. Many people believe they cannot use good medicines if they use health insurance when examined in hospitals. This fact leads to the behaviour of buying them from pharmacies.

Another problem is that patients often use too many drugs during medical examinations. These unnecessary drugs cause the examination cost or lead to consumers’ hesitation when going to patients’ hospitals. For this reason, the prescription of unnecessary drugs from healthcare providers should also be carefully considered. Hospitals and other stakeholders should also consider barriers related to healthcare policy for patients. Nowadays, many patients are afraid to go to the hospital for other reasons, such as long medical examination hours or poor service quality. It can be seen that the self-buying behaviour of Vietnamese people is also formed by other issues, such as the lack of knowledge of consumers, and they will self-treat because of the influence of the common perception of society. Besides, they feel the effectiveness when using the drug, making them ignore the side effects of using the wrong dose or drugs. Therefore, government intervention in disseminating knowledge about drug use is critical and necessary to improve the health and well-being of the people. Finally, support policies on medical costs should be optimized and addressed when patients use hospital medical services.

Disclosure statement

Authors would like to thank Industrial University of Ho Chi Minh City for funding the research project.

Additional information

Funding

Authors would like to thank Industrial University of Ho Chi Minh City for funding the research project.

Appendix

Table

QUESTIONAIRE
(PSU) Perceived Susceptibility
PSU1: I buy medicines myself because my relatives usually choose this way.
PSU2: I know that trying to buy medicines without visiting the doctor can lead to taking the wrong prescription.
PSU3: I buy medicine at the pharmacy instead of going to the doctor because I am afraid of spreading other diseases at the hospital.
(PES) Perceived Severity
PES1: I buy medicines myself instead of visiting the doctor because my symptoms are mild.
PES2: I suppose my symptoms are not too severe to go to the hospital.
PES3: I am aware that buying medicines without meeting the doctor at the hospital can cause resistance or side effects.
PES4: I am aware that taking drugs of unknown dosage can lead to side effects.
(PBE) Perceived benefits
PBE1: I buy medicines myself to reduce the cost of treatment.
PBE2: Distance when buying medicines is the priority factor when I choose for treatment.
PBE3: I want to choose medical stores instead of going to hospital because of the procedural burden at the hospital.
PBE4: I suppose that the number of medicines I buy by myself will be less than use the prescription given by the doctor.
PBE5: I buy medicines at drugstores because the assistants there understand my medication-taking habits.
(PBA) Perceived barriers
PBA1: It is more expensive of buying medicine at the hospital than at the medical stores.
PBA2: I do the self-meditation instead of visiting doctor because of saving time or finding hard to schedule my timetable to visit hospital.
PBA3: I buy medicines myself instead of visiting the doctor because the hospitals/clinics are located so far from place where I live.
PBA4: I buy the medicines myself instead of visiting doctor because I am afraid to follow many hospital procedures.
PBA5: My concern about the attitude of doctors or staffs, it pretends me to go to the hospital.
(PSE) Perceived self-efficacy
PSE1: I think that I only need to go to the hospital when I get serious symptoms.
PSE2: I buy the medicines myself instead of visiting the doctor because I have used this medicine many times before.
PSE3: I buy medicines myself for the prompt treatment of acute pains.
PSE4: I buy the type of medicine that my friends or relatives recommend its efficacy.
PSE5: I buy medicine based on seeking suggestion from the internet.
(CUE) Cues to action
CUE1: I obtain medicine when I feel that there are some problems with my health.
CUE2: Changing the weather leads my health getting sick.
CUE3: I obtain medicine when a new pandemic appears.
CUE4: I buy medicine because there are advertisements say it is good for my health.
(IBE) Individual behaviors
IBE1: I do the self-meditation because this behavior is very popular to my friends and relatives.
IBE2: I will seek the information of medicines before buying them myself.
IBE3: I normally choose buying medicine myself because it allows me to save the cost.
IBE4: I will choose to buy medicine at pharmacies near my area.
IBE5: I feel comfortable when buying medicines myself.

Note: All constructs are measured on a five-point Likert scale, which ranges from 1: strongly disagree with the statements to 5: strongly agree with the statements.