Immunization acceptation and refusal
among healthcare workers
“An ounce of prevention is worth a pound of cure.” – Benjamin Franklin
Bachelor thesis
Global Health
Lisa Lammers - 10000313
Algemene Sociale Wetenschappen (General Social Sciences)
UvA Faculty of Social and Behavioral Sciences
Year: 2013-2014
First reader: Barbara Da Roit
Second reader: Debby Gerritsen
Preface
Writing a thesis is like the proverbial cherry on top of your bachelor education. And truly it is. In my study time I always liked the research assignments, so I was excited to start my very own research project, although writing a thesis is no easy job.
While I have written this thesis myself, I have had the help of some wonderful people. First of all is of course my supervisor and guide, Barbara Da Roit, who was the best I could have wished for. Right from the start I noticed that she was very willing to help her students and has many times read my writings, giving great feedback. After our meetings I always felt very optimistic about this project and sometimes I really needed that boost. Secondly, my second reader and guide, Debby Gerritsen has been a great help in the department of quantitative research. Apparently, I had created a difficult research question and model, but with her suggestions (with some help from Pepijn Olders as well) I think an interesting methodology was put together. I would like to thank her for taking the time to sit down with me to discuss possible methods and even introducing me to new options. Before writing my thesis, I knew a bit about doing research, but now with all the feedback both my supervisors have given, that knowledge most definitely has grown. For me, the hardest part was gathering all the data and so I want to thank those who have helped me with this. I want to thank the Zonnehuisgroep Amstelland for their cooperation, and especially Emmy Doornbos, who was enthusiastic about spreading the questionnaire around and has tremendously helped me in reaching my 200 respondents mark. My father also has helped with quite a bit chunk of the respondents, as he had contact around 150 colleagues on his own. Others have helped to and I am very thankful for your help. All healthcare workers that wanted to be interviewed I also am grateful of. Lastly, I would like to thank my family, boyfriend and friends, for sticking with me in those oh so busy months. Discussing my findings, bringing me food when I was locked up in my room behind my laptop and even helping with gathering respondents, it all is greatly appreciated. Thank you all for your feedback, support and trust.
Now that my thesis has been written, I have finished my bachelor and while it feels a little like a book closes, I am glad for this experience. In a few months I have went from an idea about researching immunization, to a finished product that I would like to present to you.
Index
Abstract ... 5
1. Introduction: immunization acceptation and refusal among healthcare workers ... 6
2. Theoretical framework ... 9
2.1. Key concepts ... 9
2.2. Explaining vaccine uptake ... 10
2.2.1. Trust ... 10 2.2.2. Expert system ... 13 2.2.3. Professionalism ... 14 2.2.4. Expert power ... 15 3. Research question ... 15 4. Methodology ... 18 4.1. Research strategy ... 18 4.2. Research design ... 18 4.3. Research methods ... 19 4.4. Population ... 19
4.5. Sampling and data collection ... 19
4.6. Operationalization ... 20
4.7. Ethical issues ... 22
5. Quantitative analysis ... 23
5.1. Sample ... 23
5.2. Predictor and mediator variables ... 24
5.4. Influence predictors and mediator on decision vaccine ... 26
5.5. Influence predictors on perception trust/risk vaccine... 28
5.6. Influence mediation model on decision vaccine ... 29
5.6.1. Influence trust/risk on decision vaccinated ... 30
5.6.2. Influence mediation model: professionalism ... 30
5.6.3. Influence mediation model: expert power ... 31
5.6.4. Influence mediation model: expert system ... 32
6. Qualitative analysis ... 39
6.1. Sample ... 40
6.2. Reasons for decision vaccination ... 41
6.2.1. Types of reasons ... 41
6.2.2. The role of professionalism, expert power and position expert system ... 45
6.3. Trust and risk perception vaccination ... 48
6.3.1. Types of trust ... 48
6.3.2. The role of professionalism, expert power and position expert system ... 49
7. Conclusion ... 52
8.1. Policy advice ... 54
9. Appendix ... 61
9.1. Research instrument: questionnaire ... 61
9.2. Research instrument: semi-structured interview ... 65
Abstract
Immunization is still contested and to understand this, it is important to understand the perspective of healthcare workers themselves. Concepts such as the perception of trust/risk, professionalism, expert power and position expert system are researched in order to see if they had an influence on whether the position of healthcare workers does influence their decision regarding the uptake of the influenza vaccine. To answer this, a mixed methods design is used with questionnaires and semi-structured interviews, in order to find relationships and also the mechanisms behind those relations between the researched concepts. The level of
professionalism and amount of expert power were found to be no significant predictors, while various groups of healthcare workers in the expert system of healthcare did seem to have an effect, most healthcare workers were less likely to be trusting and getting vaccinated when compared to medical doctors and researchers. Trusting the influenza vaccine had a positive influence on getting vaccinated, as was predicted. Another interesting result was that
education had a positive effect on trust. Finally, receiving clear information that every level of healthcare worker could understand was noticed as key in making the decision of getting vaccinated. The position of healthcare workers does seem to influence their decision, however other factors need to be found and researched.
1. Introduction: immunization acceptation and refusal among healthcare workers
Immunization always has met difficulties and the presence of anti-vaccine groups testifies that immunization is still contested. Greenough (1995) mentions for example that the struggles surrounding the compulsory vaccination laws in the 19th century were hardly unique: “Yet the potential for resistance is always present, because encounters with government vaccinators are never about immunization alone. Public health measures derive their authority from the police power of the state, and people do not lightly offer themselves (or their immune systems) to government, even when its authority is legitimate.” (Greenough, 1995: p. 633).
Since the general population sometimes seems to struggle with accepting vaccination programs it is important to look at the healthcare workers. If people who work in the field of healthcare themselves have doubts about vaccines, they can give (un)intentionally a message to the rest of society. To understand immunization refusal, understanding the perceptions of healthcare workers is key. This is why the focus of this thesis is on healthcare workers in the Netherlands and whether they choose to get the yearly influenza vaccination. This vaccination is not mandatory, although it is recommended. While many healthcare workers do get vaccinated, there still are those who do not get the influenza vaccination. This thesis researches different groups of healthcare workers, such as for example medical doctors and nurses, but also homecare workers. The objective of this research is finding out what the reasons are for different groups of healthcare workers for (not) getting the influenza vaccine. Such results could clarify the decision to get vaccinated and help health organizations and employers with getting higher vaccination rates.
Societal relevance can be found, because influenza vaccine receipt by healthcare workers is important since they are at risk for occupational exposure to influenza and may act as vectors in the transmission of influenza in nosocomial (hospital) context (Martinello, Jones & Topal, 2003) and in other care contexts, such as homecare. Also regarding pandemics, maintaining the health and availability of healthcare workers is an essential component of pandemic preparedness (Prematunge, 2012). There is discussion whether influenza vaccination should be mandatory. However, as Babcock et al. (2010) notes, mandatory vaccination is a controversial strategy that pits healthcare worker autonomy against patient safety. Also, while there has been done some research about immunization refusal and hesitancy, scientists still find that large numbers of various groups of people are not vaccinated. There are very different reasons for this, however, understanding the reasons of healthcare workers and professionals could contribute to a better understanding of why people
in general do not wish to be vaccinated. Reasons may vary from a moral perspective to simple logistics reasons, such as no time to get the vaccine. Other vaccination research may benefit from such understanding. Understanding the reasoning of healthcare workers regarding their decision whether to get vaccinated could perhaps help contributing to the understanding of immunization refusal of citizens and even vaccination programs in global healthcare. Since healthcare workers themselves work in the healthcare system and thus are more embedded, they might have interesting perspectives regarding immunization and also can be viewed as an example, thus their behavior regarding immunization plays an important role.
In their literature review about this topic, Hofmann, Ferracin, Marsh & Dumas (2006) note that in many countries the vaccination coverage among healthcare workers is low. They found two main barriers to vaccine uptake: firstly, the misperception of influenza, its risks, and the role of healthcare workers in its transmission to patients, and the importance and risks of vaccination; secondly, the lack of (or perceived lack of) conveniently available vaccine. Another literature overview (Riphagen-Dalhuisen, Gefenaite & Hak, 2012) focused on predictors for vaccine uptake. Knowing that the vaccine is effective, being willing to prevent influenza transmission, believing that influenza is highly contagious, believing that influenza prevention is important and having a family that is usually vaccinated were statistically significantly associated with a twofold higher vaccine uptake.
In another study, a questionnaire was sent to Dutch hospital-based healthcare workers (Hopman, 2011). This resulted in several predictors for influenza vaccine uptake: age >40 years, the presence of a chronic illness, awareness of personal risk and awareness of risk of infecting patients, trust in the effectiveness of the vaccine to reduce the risk of infecting patients, the healthcare workers’ duty to do no harm and their duty to ensure continuity of care, finding vaccination useful despite the constant flow of visitors and having knowledge of the Health Council’s advice, social influence and finally, convenient time for vaccination.
Research about healthcare workers in Dutch nursing homes found that the average vaccination rate in nursing homes healthcare workers was 18.8% and 91.6% for residents (Looijmans-van den Akker, Marsaoui, Hak & Van Delden, 2009). In all, 68.1% nursing homes had a written policy, 87% actively requested that their employees be immunized, and 87% offered information to healthcare workers in any way. Despite the fact that the majority of nursing homes administrators (>69%) agreed with all arguments in favor of implementation of mandatory influenza vaccination, only a minority (24.3%) agreed that mandatory vaccination should be implemented if voluntary vaccination fails to reach sufficient vaccination rates. So while nursing homes actively try to get their employees vaccinated, they
are hesitant to implement a mandatory vaccine.
A final study from Opstelten et al. (2010) notes that since 2008 yearly influenza vaccination has been actively recommended in the Netherlands and also that in 2009 the Dutch government urged healthcare workers to receive additional vaccination against the pandemic influenza (A/H1N1). However, the effects of these recommendations are unknown (Opstelten et al., 2010). General practitioners and GP-trainees were mailed a questionnaire. In total, 63% of the GPs were vaccinated against seasonal influenza and 85% against pandemic influenza. For GP-trainees, these percentages were 47% and 77%, respectively. With regard to the medical staff working in the respondents’ practices, 60% received the seasonal and 76% the pandemic influenza vaccine. Reducing the risk of transmitting the virus to vulnerable patients and the individual's personal protection were the most frequently reported motives for vaccination, while having no medical indication for influenza vaccination and the conviction of being protected against influenza because of frequent professional exposure to the virus were the most frequently mentioned reasons for not being vaccinated (Opstelten et al., 2010).
Most previous research on this topic has been done from a social medicine perspective, adopting a quantitative research strategy. Social medicine does incorporate medical knowledge on the social, cultural and economic impact of medical phenomena (Stedman, 2005), however other social scientific perspectives, such as sociology, psychology and anthropology with each having different focus points, could possible find different results, which in turn can help in understanding how the perception of healthcare workers is formed and how it can influence their decision to get vaccinated. An interdisciplinary approach with several social scientific perspectives might shed better light onto decision making processes related to immunization. Theories from a sociological, psychological and anthropological perspective are integrated, helping understand the reasons and decisions of healthcare workers regarding vaccine uptake in a way that mono-disciplinary research cannot. Each of these scientific perspectives has its own focus point, such as human behavior in different groups, individuals and cultures.
Likewise, most of the previous research was conducted among hospital-based healthcare workers using quantitative questionnaires. A qualitative approach, combined with quantitative research, can contribute to a better understanding, since qualitative research incorporates the importance of understanding micro social mechanisms that can influence the behavior of people, while also adding a more interpretative perspective that can help understanding those mechanisms in a way quantitative research cannot. Researching different groups of healthcare workers, including ones that work outside of a hospital environment, can
also contribute to a better understanding. Different kinds of professionals might have different views and ideas about topics such as the influenza vaccine, resulting in various reasoning for deciding whether to get vaccinated or not. This thesis incorporated a comparative, cross-sectional study about several groups of healthcare workers and a mixed methods design (questionnaires and semi-structured interviews) to contribute to the gap in the literature that was mentioned before regarding healthcare workers and immunization.
The thesis is organized as follows. In the following chapter the key concepts and important theories are discussed, as they can help understanding how the perception of healthcare workers regarding influenza vaccination is formed and how it could influence their decision to get vaccinated. Then the central research question of this thesis is discussed, as well as two sub questions which each focus on an important step in the research process. Followed by the explanation of the chosen strategy, design, population, sampling, ethical issues and methods of both the quantitative and qualitative research. Then the operationalization and data collection are discussed, followed by the results of both methods. Finally, the found results will be discussed as a conclusion is given.
2. Theoretical framework
This paragraph explains the used concepts and theories in this thesis. Firstly, key concepts are defined, followed by the introduction of theories from sociological, psychological and
anthropological perspectives that are thought to be useful in researching immunization and might explain vaccine uptake. The primary concept of trust and risk perception is introduced first, then the three concepts of expert system, professionalism and (expert) power are discussed.
2.1. Key concepts
A few key concepts need to be explained first to better understand the chosen topic of this thesis and its relevance. Firstly, the World Health Organization (WHO, 2014) describes
immunization as “the process whereby a person is made immune or resistant to an infectious
disease, typically by the administration of a vaccine”. Vaccines stimulate the body’s own immune system to protect the person against subsequent infection or disease. Secondly, regarding specifically the influenza vaccination, the U.S. Centers for Disease Control and Prevention (CDC, 2014) note that it is an annual vaccination using a vaccine specific for a given year to protect against the highly variable influenza virus. Finally, a definition of
healthcare workers is given, WHO describes them as “all people engaged in the promotion,
protection or improvement of the health of the population” (Dal Poz et al., 2006).
2.2. Explaining vaccine uptake
Several theories from different social scientific perspectives are relevant to the study of immunization acceptance and refusal among healthcare workers. Theories from sociology referring to ‘trust’, ‘expert systems’, ‘expert power’ and ‘professionalism’ help understanding the problems regarding immunization from a perspective that includes the society as a whole. Concepts from psychology, such as ‘trust’ and ‘risk issues’ integrate a perspective that adds the more individual view of the healthcare workers on the topic of immunization. Finally, theories from anthropology provide a perspective on the importance of cultural influences. Those three disciplines interact when researching immunization refusal, as the decision to get vaccinated can be influenced by individual and psychological processes, but also by processes in social systems and society, and concepts such as trust and risk are socially constructed, so context is key.
2.2.1. Trust
The primary concept in this study is that of trust. The legitimacy and safety of immunization have been challenged in different contexts. Trust in public health institutions, in individual healthcare practitioners and in society, plays a key role in the decision-making process (Casiday, 2005). Risk is also an important concept that is closely linked with trust and thus is also discussed in this paragraph.
From a sociological perspective, the role of trust in social systems is important. Trust is one of several social constructs and an element of social reality (Searle, 1995). Giddens (1994) notes that trust in the multiplicity of “abstract systems” is a necessary part of everyday life, modern societies are becoming more complex and people cannot have knowledge about all areas of human life to handle individually all the time. Healthcare can be seen as an abstract social system. While an individual may have some knowledge about his/her own health, they cannot for example diagnose or treat themselves. Organizations such as hospitals or day-and home care organize certain areas of human life in the department of healthcare. This connects with the theory of expert systems (Giddens, 1991), which are systems of technical accomplishment or professional expertise that organize large areas of the material
and social environments. This theory is discussed later on in more detail, it however is also relevant for the concept of trust. Regarding this thesis, trust in healthcare and vaccines is very important and central to the research question, because healthcare workers have to trust healthcare as a whole, their own organization or government, and the influenza vaccine of course. This trust can influence the decision of getting vaccinated. A healthcare worker who is trusting and perceives the vaccine as positive, is more likely to get it then a healthcare worker who perceives the vaccine as risky. Beck (1992) describes the current era as the ‘risk society’ of late modernity, where risk is defined as “a systematic way of dealing with the hazards and insecurities induced and introduced by modernization itself”. Because the new risks are largely invisible, knowledge plays an increasingly important role in the risk society, as people work to establish causal links between seemingly unrelated phenomena. So whether the perceived risk are real or not, they can have social and political consequences of great importance. As Beck (1992) puts it by quoting Thomas: “If people experience risks as real, they are real as a consequence.”
From a more psychological perspective, trust is believing that the trusted person will do what is expected of him/her. This also rings true for organizations and is important, because it means that healthcare workers have to trust the healthcare system and their organization. Healthcare workers who are trusting of the organization they work at, for example, might trust its judgment for distributing the influenza vaccine. Distrusting healthcare workers might not be so easily persuaded to get vaccinated. In this sense, trust is integral to the idea of social influence (Kelman, 1958), as it is easier to influence someone who is trusting. Kelman (1958) also identifies three types of social influence; compliance,
identification and internalization. With compliance, people seem to agree with others,
however they keep their own opinions private. Identification means that people are influenced by someone they like and who is respected. Finally, internalization happens when people accept a certain belief or behavior and agree with it both publicly and privately. Healthcare workers might discuss the vaccine among themselves or outside of work with friends and family. Concerning this thesis it might be interesting to see if the several groups of healthcare workers are influenced differently and how easily they are influenced. Also, there are three different forms of trust important to mention: trust, trustworthiness and trust propensity (Colquitt, Scott & LePine, 2007). Trust is defined as the intention to accept vulnerability to a trustee based on positive expectations of his/her actions, trustworthiness as the ability, benevolence and integrity of a trustee, and trust propensity as a dispositional willingness to rely on others (Colquitt, Scott & LePine, 2007). While trust is a key concept in this thesis, not
all forms of trust might be equally important.
Thirdly, from a more anthropological perspective, Casiday (2005) notes that trust is the strongest within social groups. This also implies that distrust occurs between different social groups, a relevant example would be the group of worried citizens and the government regulators. An interesting idea is that of Earle and Cvetkovitch (1995), who argue that humanity needs to move to a ‘state of cosmopolitan social trust’. This kind of trust would rely on flexibility, communication across social boundaries, and imagination to find common values across different social groups and develop solutions to problems that were previously beset by inter-group divisions (Earle & Cvetkovitch, 1995).
Risk perceptions can lead to the acceptance of vaccination, but also to the refusal (Streefland, Chowdhury & Ramos-Jimenez, 1999). People’s decisions are informed by different concepts of risks (Casiday, 2005). These concepts depend largely on (dis)trusting the government or the information sources about the vaccine’s safety. Mistrust in the competency and efficacy of vaccination programs often results in a negative risk perception. Although decisions about immunization are taken by individuals, as mentioned before, social context plays an important role in shaping the public debates and policies, as well as in forming individuals' notions and decisions about risk (Casiday, 2005).
Anthropologist Mary Douglas and political scientist Aaron Wildavsky (1986) note that all risks are socially constructed. This is because identifying a risk requires a certain configuration of ideas about what outcomes would be undesirable, and what conditions put us in danger of experiencing those outcomes. There also seems to be an important difference between ‘expert’ and ‘lay’ conceptions of risk (Casiday, 2005). Gifford (1986) describes that 'epidemiological risk' is framed in terms of `relationships which are objective, depersonalized, quantitative, and scientifically measured' at the level of the population. On the other hand, 'clinical' and 'lay risk' are lived and experienced at the level of the individual. This could mean there are differences of risk perception among various types of healthcare workers, but also that there is a difference between people who work in healthcare themselves and people who do not, as different criteria are used to frame their perception of the influenza vaccine.
Another assumption is that public concern about vaccination is due to a miscalculation of risk (Hobson-West, 2003). Risk communication research points out that factors such as media triggers and emotions are of influence of public risk perception. For example, the media attention surrounding the A/H1N1 influenza virus (Mexican influenza) and urging of the government to get the vaccine seems to have had an impact on the decision of healthcare workers to get vaccinated, as an increasing number of healthcare workers got the vaccine for
the pandemic A/H1N1 influenza virus in comparison to the seasonal influenza vaccine (Opstelten et al., 2010). Vaccination generally is perceived as risky since it often concerns little children and vulnerable people. People also do not have much control over the outcome, benefits are hard to see and if there is damage, the vaccination can be potentially long-term or even fatal (Hobson-West, 2003). Ironically, the success of vaccination programs such as the smallpox program seems to undermine itself. People underestimate the risk of such diseases, as they cannot remember a time when those were a real problem.
Trust in the vaccine, as well as the risk perception of vaccines and influenza as a disease are important concepts. The concept of trust can be linked with the next concept of expert system (Giddens, 1994), because people are reliant on others, whom they must trust to protect them from risks.
2.2.2. Expert system
Within sociological theory expert systems are systems of technical accomplishment or professional expertise that organize large areas of the material and social environments in which we live today (Giddens, 1991). As mentioned before, healthcare as a whole can be seen as such a system. Furthermore, expert knowledge is integrated into the society and it is ever present. People do not think about this all the time, but do have faith in it. This links back to the discussed concepts of trust and risk. People respect a concept of “authority” or superior construct that makes sure their world turns each day (Giddens, 1991). Sztompka (1999) adds: “We usually take them for granted, do not even notice their pervasive presence. And we have learned to rely on them, to the extent that their failure looks like a catastrophe." This ties in with the concept of trust, and because trust in health institutions plays a big role it is important to understand healthcare as an expert system in order to understand the decisions of healthcare workers. Healthcare workers can have different positions within this expert system, which could possibly influence their decisions regarding immunization, because of different levels of expertise and knowledge. Citizens perhaps accept the authority of healthcare as an expert system, while people who work in healthcare themselves might contest this authority based on their own position in the system. A medical doctor has a different position than for example a nurse or a homecare worker in this expert system and they could have a different perspectives about vaccines, perhaps due their educational background or work experience. The various positions can lead to different levels of professionalism, which are discussed in the next paragraph, which in turn can influence the perception of healthcare workers regarding their
decision to get vaccinated or not.
2.2.3. Professionalism
Sociologist Freidson (1970) notes: “A profession can be seen as an occupation which has assumed a dominant position in a division of labor, so that it gains control over the determination of the substance of its own work.” (p. 17). Professionalism can be described as the occupational control of work (Freidson, 1999). Furthermore, there is a distinction between institutional constants and institutional variables. The constants are an officially recognized body of knowledge and skill, an occupationally negotiated division of labor, an occupationally controlled labor market and an occupationally controlled training program (Freidson, 1999). The institutional variables include the organization and policy of state agencies, the organization of the occupation itself, the dominant ideologies of the time and place, and the substance of particular bodies of knowledge and skill (Freidson, 1999). This can be linked to the concept of expert power, which is discussed in the following paragraph.
Professionalism can also be linked to the concept of trust, as Freidson (1970) mentions that professions keep their dominant position because of the trustworthiness of its members. This trustworthiness includes ethicality and knowledgeable skill. As mentioned before, healthcare workers can have different positions within the healthcare system, based on their expertise and knowledge. A healthcare worker’s trust and trustworthiness is influenced by their profession, in turn influencing their views on ethicality and their knowledge. Again, this could influence the decisions of healthcare workers of getting vaccinated. A healthcare worker that is viewed as professional and is trusted might influence others in their perspective of the influenza vaccine. Also, professionalism could influence the decision of immunization because people might feel it is their duty and part of their profession.
Another important concept regarding professionalism is discretionary power: “having or using the ability to act or decide according to your own discretion or judgment” (Freidson, 1986; Paden, 1986). A professional may make autonomous decisions, based on their expertise insight, although they have learned the standard protocols. This is of great importance when researching the perspective of healthcare workers regarding the influenza vaccine as different levels of professionalism and the ability to make autonomous decisions can vary among healthcare workers, thus influencing their perspective and ultimately their decision. The different levels of professionals and power could influence not only a healthcare worker’s own decision, but could also influence those around them. Depending on their level of
professionalism and expertise, healthcare workers can influence the ‘lower’ levels of healthcare workers, but also the normal citizens, and their perception of vaccines. Professionals and healthcare workers play a key role in the decision-making process (Casiday, 2005). This also entails the idea that different groups (or levels) of healthcare workers can have different reasoning, such as more moral reasons or practical reasons, for their decision on getting the influenza vaccine, based on their position and level of professionalism at work.
2.2.4. Expert power
A final important concept and sociological theory is that of power. French & Raven (1959) came up with five bases of power: coercive, reward, legitimate, referent and expert. However, for this thesis only the last one, expert power, is of importance. It is described as “the ability to administer to another information, knowledge or expertise” (French & Raven, 1959). Regarding this thesis, medical doctors can be a great example. As mentioned in the explanation of expert systems, medical doctors have a different and perhaps higher position in this system than, for example, a nurse or someone who works in homecare. Because of this position, and their educational background and expertise, a medical doctor might have more expert power. But nurses could possibly rank higher than a homecare worker, etcetera. Furthermore, a leader or expert is able to convince his ‘subordinates’ to trust him because of his expert power or knowledge. It is important to note that this expertise does not even have to be genuine or real, the mere perception provides the power. Healthcare workers with expert power indeed have the power to influence how people perceive an interference such as immunization. As there are different levels of expertise and knowledge, healthcare workers themselves can be influenced by experts in their own field and influence others in return. There could be a hierarchy of expertise, which in turn could influence different types of healthcare workers in their perception and decision regarding vaccination. So there can be influence within and outside the expert system that is healthcare.
3. Research question
The central research question in this thesis is: ‘How does the position of healthcare workers
influence their influenza immunization decision?’ To answer this research question and
consistently with the above-discussed theoretical framework it is important to answer two sub questions: (1) what is the influence of professionalism, expert power and the position in the healthcare expert system on healthcare workers’ perception of the influenza vaccination? (2)
How does their perception about the influenza vaccination influence the decision of healthcare workers to get vaccinated?
To be able to grasp healthcare workers’ perception regarding influenza vaccination and how it can influence their decision to get vaccinated, it is of importance to first look at the position of healthcare workers and how this influences their perception of trust and risk regarding the influenza vaccine. To answer this, the position of healthcare workers is conceptualized along three dimensions: their level of professionalism, the amount of (expert) power they have or possibly are influenced by and their position within healthcare as an expert system. These concepts could influence each other, but ultimately they could influence the way a healthcare worker thinks about immunization, and the influenza vaccination in particular. Secondly, the trust and risk perception of the influenza vaccine could influence the decision of healthcare workers to get the vaccine. The perspective of trust/risk regarding the influenza vaccine is expected to be a mediation effect, thus effecting the decision of getting vaccinated. These two sub questions and research steps will ultimately make it possible to answer the main research question.
The first hypothesis regarding the first sub question is concerned with the three predictors and their relation with the trust/risk perception. Namely, the higher healthcare workers score on their position within the expert system, their level of professionalism and their amount of (expert) power, the more likely they are to have trust in the influenza vaccine. Although decisions about immunization are taken by individuals, Casiday (2005) notes that social context does play an important role, which is why the position of the healthcare worker is analyzed. The position of a healthcare worker in the expert system of healthcare could influence their perspective of the influenza vaccine (Giddens, 1991). A professional can make autonomous decision based on their expertise insight (Freidson, 1986; Paden, 1986) and a higher level of professionalism possibly could have a positive influence on trust and ultimately the decision of getting the influenza vaccine. Professionals might feel that it is their duty to get the vaccine, for example. This correlates with a high position in the expert system of healthcare. Although a high level of professionalism could also allow such healthcare workers to deviate from what is normal as they can make their own judgment, moral reasoning could override this issue and let healthcare workers be more trusting. Expert power could influence a healthcare worker in his/her decision (French & Raven, 1959), having the power or being influenced by someone else. A hierarchy of expertise, again referring to the concept of an expert system, might positively influence the decision of getting vaccinated, when people with a high amount of expert power are trusting of the influenza vaccine.
A second hypothesis is that the perspective of trust/risk regarding the influenza vaccine will influence the decision of healthcare workers to get vaccinated, meaning that the higher healthcare workers score on trust, the more likely they are to get vaccinated. This also means that if a healthcare worker scores low on trust, meaning they perceive the vaccine as risky, they are less likely to get vaccinated. This hypothesis is concerned with the second sub question. As mentioned before, trust in public health institutions, in individual healthcare practitioners and in society, plays a key role in the decision-making process (Casiday, 2005), thus a positive effect is expected. A more ‘expert’ perception of risk is based on objective, scientifically measured terms and experienced at the level of the population, while ‘lay risk’ is lived and experienced at the level of the individual (Gifford, 1986). As mentioned before, this perception might be influenced by factors such as the level of professionalism, the amount of expert power and the position of a healthcare worker within the expert system.
A final hypothesis is about the direct effect between the predictors and the decision of getting vaccinated, it is expected that the higher healthcare workers score on their position within the expert system, their level of professionalism and their amount of (expert) power, the more likely they are to get the influenza vaccine. Different levels of expertise and knowledge could influence the decision of getting the influenza vaccine. As was with the possible relationship between these predictors and the perception of trust/risk regarding the influenza vaccine, they are expected to have a positive influence. Professionals might make the decision to get the vaccine based on their expertise insight (Freidson, 1986; Paden, 1986) or based on moral issues, such as feeling it is their duty or that they need to be a role figure. Such reasoning may also be influenced by expert power (French & Raven, 1959), whether it is their own or someone else’s, as their position in the expert system (Giddens, 1991). To summarize the three hypotheses:
H1a: The higher healthcare workers score on position within expert system, the more likely
they are to have trust in the influenza vaccine.
H1b: The higher healthcare workers score on level of professionalism, the more likely they are
to have trust in the influenza vaccine.
H1c: The higher healthcare workers score on amount of (expert) power, the more likely they
are to have trust in the influenza vaccine.
H2: The higher healthcare workers score on trust, the more likely they are to get vaccinated.
H3a: The higher healthcare workers score on position within expert system, the more likely
they are to get vaccinated.
H3b: The higher healthcare workers score on level of professionalism, the more likely they are
to get vaccinated.
H3c: The higher healthcare workers scores on amount of (expert) power, the more likely they
are to get vaccinated.
With these questions and hypotheses this thesis looks at what types of relations can be found between various concepts such as perceptions of (mis)trust, professionalism, power or authority and healthcare as an expert system, but also looks at the mechanisms that lie underneath.
4. Methodology
4.1. Research strategy
In order to answer the research question and contribute to the research gap, mixed methods research is needed. Bryman (2008: 608-609) mentions several rationales that can explain the choice of combining different methodological strategies, such as offset (combination offsets weaknesses), completeness (more comprehensive account of the studied area), explanation (one method is used to explain findings from the other method), credibility (enhancing the integrity of findings), context (qualitative research providing contextual understanding of results quantitative research) and illustration (use of qualitative data to illustrate quantitative findings). These are the factors that in one way or another prove to be relevant for this study in choosing mixed methods, as it helps with understanding the perspectives of healthcare workers regarding immunization and their decision whether to get the influenza vaccination. A quantitative strategy is applied first, followed up by a qualitative strategy. Expected is that the quantitative research generates some results that can be studied in depth conducting qualitative research, resulting in a better understanding of found results. The two strategies complement each other, with the quantitative data providing concrete results, and the qualitative data helps with finding more in-depth interpretations and mechanisms behind the data from the questionnaire.
4.2. Research design
A cross-sectional design (Bryman, 2008: 44) is used, as the variation between several groups of healthcare workers are researched. The data of more than one case are collected at a single point in time, becoming a body of quantitative or quantifiable data in connection with at least
two variables. Those variables then are examined to detect patterns of association (Bryman, 2008: 44). Furthermore, it can be said that a comparative design (Bryman, 2008: 58) is used, as several different groups of healthcare workers are compared in what their reasons are for their decision on getting vaccinated or not and what their views on and experiences are of various concepts such as trust or professionalism.
4.3. Research methods
As mentioned before, both a quantitative method and a qualitative method are needed. For the quantitative method, self-completion questionnaires (Bryman, 2008: 216) are used. The questionnaires are distributed to the groups of healthcare workers through an online survey. This method provides for concrete, quantifiable results and shows what sort of relations that are between the researched concepts. For the qualitative part, semi-structured interviews (Bryman, 2008: 438) are conducted. Respondents from the questionnaire pool are asked to participate in an interview. This method helps with understanding the findings from the questionnaires and perhaps explains why certain relations are found, as mentioned above.
4.4. Population
The population of the quantitative research are several groups of healthcare workers. Six different groups were identified: medical doctors/researchers, nurses/assistants, behavior experts (orthopedagogues and psychologists), homecare workers, helpers of disabled and finally, healthcare workers that a have more coordination function. The first two groups are more hospital-based, while the other groups are most likely working outside of a hospital. The population of the qualitative research originate from the respondent pool of the questionnaire and consist of healthcare workers who expressed an interest in wanting to get interviewed. They have different functions, such as a medical doctor, an orthopedagogue, several nurses and homecare workers and originate from several different organizations.
4.5. Sampling and data collection
Respondents have been sampled through convenience and snowball sampling (Bryman, 2008: 183-184), because of the limited time and access, ruling out any form of probability sampling. This could possibly in some form influence the found results and also means results probably will not be able to be generalized to other healthcare workers, meaning the data might not be
as representative as it would have been when respondents are selected randomly. While access is available in a few healthcare locations, respondents have to be willing to participate. Aside from personal connections, healthcare locations were contacted through e-mail, and individual healthcare workers were reached through social media. An appeal to fill in the questionnaire was also placed on social media, such as Facebook, Twitter and LinkedIn. Some respondents and companies in turn may refer to new respondents, meaning snowball sampling also plays an important role in collecting enough respondents. However, something else that needs to be taking in consideration is that the data most likely is nested, as most respondents are contacted through their organization and are nested within this very organization. This also could influence the generated data and have an effect on the results that are found, as being nested in an organization itself perhaps is an explanatory variable.
Primary research locations are the Academic Medical Centre (AMC) Amsterdam, University Medical Centre (UMC) Utrecht, the organization Zonnehuisgroep Amstelland and the Hartekamp organization. Most respondents originate from one of these locations. Other organizations and several individual healthcare workers were contacted as well. Most of the organizations that were contacted did not respond or did not want to participate because they did not have the time. Respondents at AMC, UMC and Hartekamp were acquired through personal connections, however the organization Zonnehuisgroep Amstelland was contacted as well and was willing to distribute the questionnaire among its employees. Qualitative data collection was done through the questionnaire as well, respondents that were interested and willing to be interviewed could leave their e-mail address behind for contact purposes.
4.6. Operationalization
In the questionnaire the concepts are operationalized through several questions per concept. Questions originate from several pre-existing questionnaires and scales, which have served as inspiration for the created questionnaire of this thesis. As they all are existing questionnaires and scales, the questions that are used are thus validated in measuring the wanted concepts. For the concept of trust the Healthcare System Distrust Scale (Rose, Peters, Shae &
Armstrong, 2004) and the Medical Mistrust Index (LaVeist, Nickerson, Boulware & Powe, 2001) are used as inspiration, resulting in four questions. For the concept of (expert) power the Semantic Differential Power Perception Survey (Bartos et al., 2008) was used, resulting in ten questions. Finally, for the concept of professionalism, three questionnaires were used as inspiration and resulted in four questions: the Medical Professionalism in the New
Millennium Physician Charter (ABIM Foundation, 2002), the Penn State College of Medicine Professionalism Questionnaire (Blackall et al., 2007) and a Profession Survey on Medical Professionalism (Campbell, 2007).
Concept Operationalization Measurement
Healthcare workers Work occupation Q: What is your occupation?
Q: How many years have you worked in the
healthcare sector? Influenza vaccination Take up of influenza
vaccine
Q: Did you get the influenza vaccine this year?
Q: Did you ever get the influenza vaccine in your career?
Trust, risk issues (Perception of) trust and risk about influenza vaccine
Disagree – Agree
Q: The healthcare sector has the health of the patient as the first priority. Q: Healthcare
organizations know what they are doing, they have enough capacity.
Q: People need to be wary concerning vaccinations. Q: I expect that the healthcare sector keeps up-to-date with new information regarding the influenza vaccine. Expert system Position of the healthcare
workers in their work place
Q: What is your highest education?
Q: What is your occupation?
Power Power of the healthcare
workers in their work place Q: At my work, I have… knowledge – ignorance. Q: At my work, I have… authority – no authority. Q: At my work, I have… influence – no influence. Q: At my work, I have… respect – no respect. Q: At my work, I have… power – no power. Q: At my work, I feel… complimented – critized. Q: At my work, I feel… 21
autonomous – not autonomous. Q: At my work, I feel… experienced – not experienced. Q: At my work, I feel… above my colleagues – beneath my colleagues Q: At my work, I feel professional – not professional. Professionalism Professionalism of the
healthcare workers in their job
Disagree – Agree Q: I demonstrate
adaptability in responding to changing needs and priorities of the patient/client. Q: I assume leadership in client or patient management. Q: I practice my own knowledge and capacities. Q: I assume personal responsibility for
decisions regarding client or patient care.
For the qualitative method semi-constructed interviews were conducted. A topic list was developed, with most questions being based on some of the questions from the questionnaire, while other questions were added as well to find out what kind of information healthcare workers receive regarding the influenza vaccine and how they want to be approached through their organization about the vaccine. The topic list can be found in the Appendix. After conducting the interviews, they were transcribed so the data could be analyzed through open coding, labeling the concepts and finding categories. Through the open coding, mechanisms behind the results of the questionnaires can be investigated and might offer more in-depth explanations for relationships between the discussed concepts of the theoretical framework.
4.7. Ethical issues
Anonymity is an important ethical issue. Healthcare workers might not want to admit that they do not get the influenza vaccination or what their reasons are for their decision.
Questions about their work, professionalism and authority could also be badly received. The anonymity of the questionnaire hopefully helps with gaining the trust of the respondents and getting them to answer truthfully, some respondents might also be more willing to answer truthfully through personal connections. This anonymity most likely also helps with answering the more personal questions about professionalism and authority (expert power).
The interviews however are not entirely anonymous, however personal information (e.g. names) are removed in the final version and through objectivity as a researcher, the given answered will not be judged. All given information is confidentially. Finally, through these steps any harm to respondents, groups or organizations will be prevented.
5. Quantitative analysis
In this paragraph the results of the quantitative analysis are presented. First the sample and the variables that were used are introduced, followed by a description of the mediation model that is constructed. Assumptions have been tested, but no notable problems were found. To be able to answer the research question several steps need to be taken. First a logistic regression is carried out to determine the direct effect of the independent variables on the dependent variable, without the mediator present. Then, the mediator variable ‘trust/risk vaccine’ is added. Because a mediation effect is expected, a linear regression was also conducted to measure the direct effect of the predictor variables on the mediator variable of ‘trust/risk’
5.1. Sample
A total of 241 respondents filled in the questionnaire completely, however 306 respondents had started to fill in the questionnaire, resulting in some variables with a higher total. In table 1 the descriptives are presented of the outcome variable, being vaccinated or not, and the control variables ‘gender’, ‘age’, ‘education’, ‘years working in healthcare’ and ‘percentage contact with patients/clients’.
Table 1.
N Mean SE Min Max
Vaccinated No (0) Yes (1) 306 155 151 - - 0 1 Gender Female (0) Male (1) 305 247 58 - - 0 1 Age 294 42.75 12.533 19 70 Education High school (1) MBO (2) HBO (3) WO (4) 305 25 83 112 75 - - 1 4 Years working in healthcare 243 17.13 12.011 0 45 Percentage contact with patients 0-25% (1) 25-50% (2) 50-75% (3) 75-100% (4) 303 99 48 63 93 2.50 1.234 1 4
5.2. Predictor and mediator variables
To create the predictor variables ‘professionalism’, ‘expert power’ and the mediator variable ‘perspective trust/risk vaccine’, the subscales for these concepts from the questionnaire were added together to create the new variables. Which questions were used for which concept can be found in the discussed operationalization in paragraph 4.6.
The predictor variable ‘position expert system’ was conducted from the healthcare workers’ occupation. Healthcare workers were split up in several groups, based on their occupation they mentioned in the questionnaire. The first group consists of medical doctors and researchers, The second group consists of nurses, assistants and paramedics, the third group consists of behavior experts (psychologists and orthopedagogues), the fourth group consists of homecare workers, the fifth group consists of helpers of the disabled and finally
the sixth group consists of the more coordination occupation such as management, teachers, HR and coordinators. Each group was giving a value from 1 to 6 in the dataset. Both linear and logistic regressions were used in the analysis of the questionnaire, meaning that a categorical variable like ‘position expert system’ also needed to be dummy coded (Field, 2009). The group of medical doctors and researchers was chosen as the baseline, or the reference group.
Table 2.
Dummy 1 Dummy 2 Dummy 3 Dummy 4 Dummy 5
Medical doctors/researchers 0 0 0 0 0
Nurses/assistants/paramedic 1 0 0 0 0
Behavior experts 0 1 0 0 0
Home care 0 0 1 0 0
Helpers disabled, elderly 0 0 0 1 0
Other/Coordinators 0 0 0 0 1
Finally, the two variables regarding vaccine uptake (‘last year vaccinated’ and ‘ever vaccinated in career’) were combined to create a variable called ‘vaccinated yes/no’ that distinguished healthcare workers being either vaccinated or not.
Table 3.
N Mean SE Min Max
Trust/risk vaccine 284 13.26 2.004 7 20 Professionalism 269 15.41 1.876 10 20 Expert power 246 38.03 4.839 22 47 Expert system Medical doctor/researcher Nurse/assistant/paramedic Behavior expert Homecare/IG
Helper disabled, elderly Other: management/ education/HR/coordinator 301 24 61 24 78 34 80 - - 1 6
Note: Cronbach’s α: ‘trust/risk vaccine’ .54, ‘professionalism’ .69, ‘expert power’ .83
5.3. Mediation model
As a mediator variable is used, an additional macro called PROCESS (Hayes, 2008) was needed to create a mediate model. A mediated relationship with the concepts in this thesis looks as follows:
Figure 1.
a b
c’
While a simple relationship looks as follows:
Figure 2.
c
The model in figure 1 as a total shows the indirect effect of the predictor on the outcome variable, with a mediator variable present. Apart from the indirect effect, several other relationships can be analyzed. Path a presents the relationship between the predictor and the mediator, path b the relationship between the mediator and the outcome variable and path c’ shows the relationship between the predictor and the outcome, while the mediator is present in the model and is called the direct effect. The model in figure 2 shows the total effect between the predictor variable and the outcome variable, with no mediator present and is called path c.
5.4. Influence predictors and mediator on decision vaccine
To analyze the data from the questionnaire and in order to answer the research question, first a logistic regression is carried out to determine the direct effect of the independent variables on the dependent variable, without the mediator present. For this analysis the outcome variable is ‘vaccinated yes/no’, with several predictors such as ‘professionalism’, ‘expert power’ and
Predictor (Expert System, Professionalism, Expert Power)
Mediator (Perspective trust/risk vaccine)
Outcome (Decision get vaccinated yes/no)
Predictor (Expert System, Professionalism, Expert Power)
Outcome (Decision get vaccinated yes/no)
‘expert system’, together with several control variables; ‘age’, ‘gender’, ‘education’, ‘years working in healthcare’ and ‘percentage contact with patients’. In the second step, the mediator ‘trust/risk vaccine’ is added to the model.
Table 4.
Model 1 Model 2
95% CI for Odds Ratio 95% CI for Odds Ratio
B (SE) Lower Odds
Ratio
Upper B (SE) Lower Odds
Ratio Upper Included Constant -1.66 (1.94) - .19 - -3.72 (2.20) - 0.02 - Professionalism 0.15 (0.09) 0.96 1.16 1.40 0.13 (0.10) 0.93 1.13 1.37 Expert Power 0.06 (0.04) 0.98 1.06 1.14 0.05 (0.04) 0.97 1.05 1.14 Expert System * Nurses/Assistants/Paramedic Behavior experts Daycare workers
Helpers disabled, elderly
Others/Coordinators -2.88*** (0.81) -2.21** (0.81) -1.67 (0.89) -2.49** (0.88) -2.09** (0.79) 0.01 0.02 0.03 0.02 0.03 0.06 0.11 0.19 0.08 0.12 0.28 0.53 1.08 0.46 0.59 -2.71*** (0.82) -2.05* (0.82) -1.61 (0.90) -2.25* (0.89) -2.02* (0.81) 0.01 0.03 0.3 0.02 0.03 0.07 0.13 0.19 0.11 0.13 0.33 0.64 1.17 0.60 0.66 Trust/Risk Vaccine - - - - 0.19* (0.09) 1.02 1.21 1.44 Control variables Gender 0.98* (0.44) 1.12 2.66 6.32 0.94* (0.45) 1.06 2.55 6.12 Age 0.02 (0.02) 0.98 1.015 1.06 0.02 (0.02) 0.98 1.02 1.06 Education -0.29 (0.26) 0.45 0.74 1.22 -0.39 (0.27) 0.39 0.67 1.13
Years working in healthcare -0.02
(0.02) 0.94 0.98 1.02 -0.02 (0.02) 0.94 0.98 1.02 Percentage contact patients/clients -0.11 (0.18) 0.63 0.89 1.28 -0.09 (0.18) 0.64 0.92 1.32 27
Note: Model l R2 (Nagelkerke) = .196. χ2 = 8.79, p = .361. Model 2 R2 (Nagelkerke) = .225. χ2 = 3.31, p = .914. * p < .05, ** p < .01, *** p < .001
The Hosmer-Lemeshow test is a statistical test for goodness of fit for the logistic regression model and shows for the first model χ2 = 8.79, p = .361, where small Chi-squared values (with larger p-value closer to 1) indicate a good logistic regression model fit. For the second model the test shows χ2 = 3.31, p = .914.
In the first model, various occupation categories of the ‘expert system’ variable were associated with a decreased likelihood of being vaccinated when compared to the reference group ‘medical doctor/researcher’: the ‘nurse/assistant/paramedic’ group (b = -2.88, Wald = 12.65, p < .001), the ‘behavior expert’ group (b = -2.22, Wald = 7.49, p < .01), the ‘helper disabled/elderly’ group (b = -2.49, Wald = 8.05, p < .01) and the ‘other: management/education/HR/coordination’ group (b = -2.09, Wald = 6.86, p < .01). Also, the variable ‘gender’ was significant, being male was associated with an increased likelihood of being vaccinated (b = 0.98, Wald = 4.94, p < .05). Finally, the first model explained 19.6% (Nagelkerke R2) of the variance in the decision to get vaccinated.
When adding the variable ‘trust/risk vaccine’ to check if it has any influence, the previous significant groups of the ‘expert system’ and ‘gender’ still were found to be significant, although the p-values sometimes slightly decreased. Trust/risk perception regarding the influenza vaccine was associated with an increased likelihood of being
vaccinated (b = 0.19, Wald = 4.85, p < .05). The second model explained 22.5% (Nagelkerke
R2) of the variance in the decision to get vaccinated, when ‘trust/risk vaccine’ was added. 5.5. Influence predictors on perception trust/risk vaccine
When the variable ‘trust/risk vaccine’ was added, p-values slightly decreased and the variable itself was found to be positively associated with being vaccinated. Because of these results a mediation effect of (mis)trusting the influenza vaccine is expected, meaning the relationship between the predictors ‘professionalism’, ‘expert power’ and ‘position expert system’ and the outcome variable ‘vaccinated yes/no’ can be explained by a healthcare worker’s perspective of viewing the vaccine as trustworthy or risky.
In order to understand this relationship, a linear regression is carried out to measure the direct effect of the predictor variables on the mediator variable of ‘trust/risk’. For this regression analysis the outcome variable is ‘perception trust/risk vaccine’ and the predictor variables are ‘professionalism’, ‘expert power’ and ‘expert system’ (which is dummy coded),
and the same control variables as before: ‘age’, ‘gender’, ‘education’, ‘years working in healthcare’ and ‘percentage contact with patients’.
The dummy variable for the group ‘nurse/assistant/paramedic’ predicted the amount of trust in the influenza vaccine (b = -1.48, t = -2.38, p < .05), as did the dummy variables for the ‘behavior expert’ group (b = -1.30, t = -2.03, p < .05) and the ‘helper disabled/elderly’ group (b = -1.80, t = -2.57, p < .05). Since all three b-values are negative, this means that healthcare workers who are nurses, assistants, paramedics or helpers for the disabled and elderly are likely to have less trust in the influenza vaccine when compared to medical doctors and researchers. Also, the ‘education’ variable predicted the amount of trust in the influenza vaccine (b = 0.45, t = 1.97, p = .050). With a positive b-value, this means that healthcare workers with a higher education are more likely to have trust in the influenza vaccine.
Finally a R2 value of .174 was found, meaning this model explains 17.4% of the variance in the perspective trust/risk regarding the influenza vaccine.
5.6. Influence mediation model on decision vaccine
Finally, mediation models can be built with the help of the SPSS macro PROCESS (Hayes, 2008), which analyses the indirect effect of the predictor variables and creates several models. The first model in PROCESS (Hayes, 2008) consists of a linear regression with the outcome variable being ‘perspective trust/risk vaccine’ and the predictor variables ‘professionalism’, ‘expert power’ and ‘position expert system’, this however was already analyzed in paragraph 5.4 and output from this model are omitted in this paragraph to prevent repetition.
The second model in PROCESS (Hayes, 2008) consists of a logistic regression with the outcome variable ‘vaccinated yes/no’, the predictor variables ‘professionalism’, ‘expert power’ and ‘position expert system’ and the mediator variable ‘trust/risk vaccine’ added as another predictor as well. However, this too was already analyzed in the second step of paragraph 5.3 and thus the output from this model is omitted as well in this paragraph.
The third model in PROCESS (Hayes, 2008) consists of the same logistic regression as mentioned before, with the outcome variable being ‘vaccinated yes/no’ and the predictor variables ‘professionalism’, ‘expert power’ and ‘position expert system’. The variable
‘perspective trust/risk vaccine’ is not in this model, so a direct effect is analyzed here. This too has been discussed in paragraph 5.3 as the first step, so output regarding this again is omitted. However, this third model also analyzes the influence of the mediator variable on the outcome variable. The results from this, the influence of ‘perspective trust/risk vaccine’ on ‘vaccinated
yes/no’, is shown below.
The fourth and final model in PROCESS (Hayes, 2008) is the indirect effect model has not been discussed in previous paragraphs, and is key in answering the research question, so the results from this model are presented below.
5.6.1. Influence trust/risk on decision vaccinated
Perspective trust/risk vaccine → Decision vaccine yes/no
Part of the output of the second model in PROCESS (Hayes, 2008) shows that trust in the vaccine does significantly predicts the decision of getting vaccinated (b = 0.19, z = 2.20, p < .05. Since b is positive, this means that as trust in the influenza vaccine increases, the
likelihood of being vaccinated also increases.
5.6.2. Influence mediation model: professionalism
Professionalism → (Perspective trust/risk vaccine) → Decision vaccine yes/no
Results of both the total and direct effect can be found again in this model, however the results of the indirect effect are important here. The output shows b = .026, a bootstrapped standard error of .028 and a confidence interval of -0.0088 and 0.976, meaning that the true b-value for the indirect effect falls between these two b-values. Since the confidence interval does contain a slightly negative value, a genuine indirect effect could be questioned, but the perception of trust/risk regarding the influenza vaccine might be a mediator between the relationship of professionalism and the decision of getting vaccinated. A Sobel test is also run, showing the b-value again, as well as a z-score of 1.21 and a p-value of .226, meaning that there is no significant indirect effect. However, tests such as Sobel can be misleading and are prone to problems when working with smaller sample sizes, and with the confidence interval containing an only slightly negative value, there still might be a small mediation effect. Other effect sizes sadly are not available for an indirect effect model with a dichotomous outcome,
Finally, it can be concluded that there might be a small indirect effect of professionalism on the decision of getting vaccinated through the perspective of trust/risk regarding the influenza vaccine, b = 0.03, BCa CI [-0.009, 0.976], it is however unsure and might not be likely.
Figure 3
b = 0.14, p = .11 b = 0.19, p = .028
direct effect, b = -0.13, p = .21
indirect effect, b = .03, 95% CI [-0.01, 0.98]
5.6.3. Influence mediation model: expert power
Expert power → (Perspective trust/risk vaccine) → Decision vaccine yes/no
Results of both the total and direct effect can be found again in this model, however the results of the indirect effect are important here. The output shows b = .0065, a bootstrapped standard error of .0095 and a confidence interval of -0.0071 and 0.0323, meaning that the true
b-value for the indirect effect falls between these two values. Since the confidence interval
does contain a slightly negative value, a genuine indirect effect could be questioned, but the perception of trust/risk regarding the influenza vaccine might be a mediator between the relationship of expert power and the decision of getting vaccinated. A Sobel test is also run, showing the b-value again, as well as a z-score of 0.8213 and a p-value of .4115, meaning that there is no significant indirect effect. However, tests such as Sobel can be misleading and are prone to problems when working with smaller sample sizes, and with the confidence interval containing an only slightly negative value, there still might be a small mediation effect. Other effect sizes sadly are not available for an indirect effect model with a dichotomous outcome.
Finally, it can be concluded that there might be a small indirect effect of professionalism on the decision of getting vaccinated through the perspective of trust/risk regarding the influenza vaccine, b = 0.01, BCa CI [-0.007, 0.032], it is however unsure and might not be likely.
Predictor (Professionalism)
Mediator (Perspective trust/risk vaccine)
Outcome (Decision get vaccinated yes/no)
Figure 4
b = 0.03, p = .33 b = 0.19, p = .028
direct effect, b = 0.06 p = .17
indirect effect, b = .01, 95% CI [-0.007, 0.032]
5.6.4. Influence mediation model: expert system
In PROCESS (Hayes, 2008) each dummy separately has to be picked as the independent variable, with the other dummy variables (plus the other independent variables and control variables) as covariates, meaning the test needs to be executed five times for ‘position expert system’. For this variable all PROCESS models are discussed, because of the use of dummy variables which are needed since PROCESS will not allow to specify a categorical predictor. In this regression analysis the outcome variable is ‘vaccinated yes/no’, the predictor variables are per step the expert system groups dummy variables and the mediator is ‘perspective trust/risk vaccine’. The remaining predictors (as well as the other dummy variables) again are added together with the control variables as covariates.
In the first step, a simple linear regression is created regarding the variable ‘perspective trust/risk vaccine’ predicted from a healthcare worker’s position in healthcare as an expert system, using the five dummy coded variables.
Expert system positions→ Perspective trust/risk vaccine.
First is the ‘nurse/assistant/paramedic’ group. Output shows that being part of this expert system group does significantly predict the healthcare worker’s perspective of trust/risk regarding the influenza vaccine (b = -1.48, t = -2.38, p < .05). Since b is negative, this means that if a healthcare worker is part of this group, trust/risk regarding the vaccine declines, when compared to the reference group of medical doctors and researchers.
Being part of the second, the ‘behavior expert’ group that consists of orthopedagogues and psychologists, does significantly predict the healthcare worker’s perspective of trust/risk regarding the influenza vaccine (b = -1.29, t = -2.03, p < .05). Since b is negative, this means
Predictor (Expert power)
Mediator (Perspective trust/risk vaccine)
Outcome (Decision get vaccinated yes/no)
