• No results found

In this section the nature and limitations of this research, and opportunities for further research are outlined.

1.13.1 Nature of the Research

This research is exploratory, and the findings should be interpreted with care when applied to different contexts. The primary purpose is to use existing theory to gain a deeper understanding of quality in healthcare, its enablers and barriers and the underpinning mechanisms that influence it. It integrates hitherto separate theoretical domains, developing a conceptual model to help guide stakeholders in health care when designing and implementing change.

1.13.2 Limitations of the Study

In this section, the limitations of this research are discussed. First, the limitations associated with the systematic review are addressed. Then, the limitations associated with the interview techniques (repertory grid technique and semi-structured interview technique) employed in this research are outlined, and finally, the limitations associated with the use of mixed-methods are covered.

1.13.2.1 Systematic Review

In this review, every effort has been made to follow the steps designed to reduce bias and increase rigor and transparency by systematically searching all available literature and extracting relevant evidence. However, as with any academic work, this review has limitations associated with it.

First, only one reviewer was employed in this systematic review, and the possibility for researcher bias is therefore present. By following a strict and predefined protocol (see Appendix A.1 from page 317) and carefully documenting each step, the researcher has attempted to reduce any untoward bias during the search, extraction and synthesis of the evidence presented in this thesis.

Second, interventions reviewed were often based on previous research, therefore lacking a strong theoretical basis. The research on interventions identified in this review typically focus on the clinical or administrative needs for interventions. As no consensus on what constitutes optimal pharmaceutical consumption, and outcomes data on administrative interventions are lacking, it appears that the choice of intervention is not well grounded in theory, and at the same time, may be inadequately described in the literature in terms of Theory of Planned Behaviour, Agency Theory or decision under uncertainty. However, each study has been evaluated from the theoretical perspectives identified in the scoping study, and relevant mechanisms of influence in health care quality identified within the same theoretical domains.

Finally, no restriction on study type for inclusion was made. The tendency for reviews published within the medical domain of the literature (approximately 80%) is focused on randomized controlled trials as the highest standard of evidence quality. This positivist approach may not be the best when dealing with complex issues such as physician prescribing behaviour, and elements of realist synthesis have therefore been employed.

1.13.2.2 Repertory Grid and Semi-Structured Interview Technique

Three main limitations to repertory grid technique (RGT) apply in this study. First, RGT is a method that is time consuming, which makes it difficult to generate large amounts of data and thus produce general knowledge. Second, RGT may be of limited use if the respondents find it difficult to understand the technique or if the constructs elicited are not handled in a sensitive manner (Cassell and Walsh, 2004). Finally, interpretation of the data may be problematic and due to the fact that interviews are subject to researcher bias (Goffin, 2002).

The study was designed to capture how physicians frame quality of health care delivery in a hospital setting. The interview subjects in this study provided real patient cases where the outcome was death, either indirectly or directly caused by the physician being interviewed. This indicate that the respondents were both able and willing to tell the truth, and is therefore a testimony to the first two criteria suggested by (Gottschalk et al., 1945); see Table 1-8 Checklist of criteria suggested by Gottschalk et al. (1945 p. 35). With respect to the third criteria, the researcher has been as diligent as possible to provide example quotes to substantiate every inference made from the data in this study. The use of low inference descriptors, such as quotes, is described as a strategy for improving the quality of qualitative research (Johnson, 1997 p. 283). Finally, Projects 1-3 and literature have been used to corroborate the evidence presented.

Table 1-8 Checklist of criteria suggested by Gottschalk et al. (1945 p. 35)

# Description Checked

1 Was the ultimate source of the detail (the primary witness) able to tell the truth? Yes

2 Was the primary witness willing to tell the truth? Yes

3 Is the primary witness accurately reported with regard to detail under examination? Yes 4 Is there any external corroboration of the detail under examination? Yes

Though qualitative research does not seek statistical generalizability, but generalizability to theory, there are elements that help establish the quality of the study. First, low inference descriptors (quotes) have been widely used. Second, theory triangulation (Agency Theory, Theory of Planned Behaviour and Decision Theory) and methods triangulation (repertory grid and semi-structured interviews) have been used. Finally, discussions with peers and implementation of findings in practice have yielded accepted changes in the way emergency medical care is organized on a national level. In Norway, quality in health care is a key topic in the national debate particularly towards the election in the fall 2013.

One weakness of the semi-structured interview technique is that there may be a limited scope for the respondent to answer questions in sufficient detail or depth.

In order to ensure sufficient detail and depth in the responses, a laddering approach was employed by asking follow-up questions.

Furthermore, during the interview, the researcher may influence the way a respondent answers various questions, thereby biasing the responses. The

researcher followed the interview guide closely, and made every effort possible not to influence the answers provided. Open ended questions with laddering were employed.

1.13.2.3 Triangulation (mixed-methods)

As described above, in this research, a multi-methodical approach suited to the individual projects in this research has been employed. Even though its use goes back more than 100 years, it is still debated by scholars (Blaikie, 2010).

Triangulation, also known as mixed-methods, is advocated by several scholars;

for example, (Denzin, 1988) and (Robson, 1993). However, (Blaikie, 2010) argues that it is inappropriate to combine methods based on different ontological positions. Blaikie (2010), p. 227, does however note that the use of mixed-methods is of particular benefit when more than one research question is being addressed, as is the case in this thesis.

1.13.3 Opportunities for Future Research

The results from this research have highlighted many opportunities for further research, which are addressed in this section.

The systematic literature review revealed a classification of factors influencing physician decision making. However, a deeper understanding of the link between contexts and interventions in determining health outcomes is an area for further research.

The systematic literature review employed factor analysis and extended the thematic review of extant literature. Further development of research based tools to synthesis knowledge is an area with much potential for further work.

The thesis provides an exemplar of the usefulness of repertory grid and semi-structured interview techniques in exploring physician perceptions of quality in health care delivery based on patient-physician encounters. Their application to the professional service industries in general may be an area for further research.

No grand theory of medical decision making has yet been formulated. Decision making in medicine is closely linked to quality; thus, medical decision making is an area for further theoretical research.

The research undertaken in this thesis employed real patient-physician interactions to gain insight into how physicians perceive and construct quality. In this research, the patient is not included, and the dyadic patient-physician relationships is an area to be further studied in the context of the findings presented in this thesis.

This study has identified how physicians frame quality in health care delivery, identified influencers in quality of health care service delivery, and may serve as a framework for further studies in this field. The model presented in this research will benefit from further mathematical development and prospective testing. Thus, further studies validating the framework presented in this thesis is warranted.