After the analysis of the main themes emerging from the document analysis and the semi- structured interviews, and exploring how these inform our understanding of the policy process and the analysis of frames of the problem of poor-quality medicines, I proceeded to undertake a comparative analysis of the policy processes and frames used by respondents across the three case countries. This was conducted in two phases.
First, I reflected on the similarities and differences in the policy development processes and how comparable the policy solutions to address the issue of poor-quality medicines were. I noted whether, in the chronological progression of policy efforts between these three countries, there was evidence of policy coherence, overlaps of initiatives, or evidence of policy transfer between one context to another. I also reflected on potential gaps in policy processes from one country to the next, and the role of policy actors and how different categories of actors, or actor configurations influence the policy process in the three countries. The comparative analysis process was made easier with the visualisation of the policy processes in a timeline format. I then compared the dominant framings of the problem of poor-quality antimalarial medicines in Cambodia, Laos and Thailand. Through the mapping & interpretation stage of the thematic content analysis approach, I charted the similarities and differences between the five dominant frames and the sub-themes emerging from the data across the three countries. I noted the similarities and differences in official definitions of ‘poor-quality medicines’ across the three countries in the policy and legal documents. I compared how different policy actors, across institutional and national settings, defined the policy issue and if relevant, I depicted variations in how the same frame was articulated across the three countries. This was done by juxtaposing the list of themes and subthemes that emerged from the data, and investigating how these reflect different articulations of each frame.
Comparative analysis offered the opportunity to improve the reliability of my observations from the data. I compared these observations across national contexts to uncover any discrepancies in my empirical analysis. In this process, I also referred to the data from global expert interviews to inform the arguments in the final discussion of findings. Besides these triangulation methods, I sought to improve the reliability of my findings by cross-referencing them across the three methods. Respondents’ contributions cannot be considered to be flawless reflections of reality, and their accounts had to be interpreted carefully and in light of other information obtained through document analysis, or from the literature.
I also made sure that all of the data was coded comprehensively, ensuring that disconfirming evidence and deviant cases are addressed and included in the analysis (Green & Thorogood 2014; Silverman 2011). Additionally, I consulted my supervisors and the advisory committee on certain batches of data and their coding structures. This ensured the reliability of my analytical framework and demonstrated that the codes and thematic structure that emerged from the data were not solely reliant on my subjective analysis (Green & Thorogood 2014). To ensure the validity of my findings, I also sought to obtain validation of my findings from respondents whenever possible, by going back to participants and asking them if findings seemed relevant and accurate (Green & Thorogood 2014). This was a useful approach, especially as it provided an opportunity to verify elements of confidentiality with the participants who could approve or reject quotations extracted from their accounts, when applicable. This method also offered the possibility to share some key findings and recommendations with the decision makers that were interviewed in the data collection phase of this study, as an organic reporting mechanism.
Introduction to results
Respondents
A total of 51 respondents30 were interviewed during my fieldwork in Cambodia, Laos and Thailand. Before leaving London, 23 key informants were first identified through the stakeholder map exercise and some initial contacts on the ground were made with the help of my research advisors based in Southeast Asia. A further 30 were recruited while in the field, identified through snowballing technique following recommendations from interviewees. This respondent-driven approach was iterative and largely driven by data refinement (Ritchie et al. 2014). It offered the possibility to address information gaps when required and was well-suited to the exploratory nature of this study which delves into a politically-sensitive and complex, multi-faceted phenomenon.
Table 6 on page 96 summarises the number of respondents per country and for each category of respondent. Under ‘government-level respondents’, I include representatives from the Ministry of Health, National Regulatory Authorities (NRAs), national malaria control programs and law enforcement institutions such as customs or police. ‘Non-government respondents’ include local civil society, local representatives from academia or from the industry, in-country representatives from donor organisation, inter-governmental organisations or international non-government organisations based locally, some of which were key implementing partners in the policy process. Under non-government respondents, I was also able to interview three industry representatives. They were included in the sample of respondents so far as they were involved in the agenda-setting process, and because the opportunity presented itself by referral. The sample of respondents did not include patients. However, respondents in each country shared interpretations of consumer behaviour as a contributing factor to the illicit trade in poor-quality antimalarial medicines.
Country Respondents No. Total
Cambodia Government-level 2 17
Non-government 15
Laos Government-level 9 17
Non-government 8
Thailand Government-level 7 17
Non-government 10
Table 6 Number of local respondents per sector, per country.
Simultaneously, in the course of the year 2015 and 2016, I conducted semi-structured interviews with 23 international experts on access to quality essential medicines, in order to gauge their opinion on the complexity of the issue and on the factors they believe contribute to the prevalence of falsified essential medicines globally. Most of interviews with experts were conducted by tele-conference, although five of those were conducted in person. As summarised in Table 7 (page 96), the sample of expert respondents included senior members of international non-government, inter-governmental and donor organisations, representatives from the pharmaceutical industry, as well as academics and one respondent from the media. The results from these interviews served to inform the research design and triangulate findings in the discussion chapter.
Category of Expert Respondents No. of Respondents International Non-Governmental Organisations 6 Inter-Governmental Organisations 2 Donor Organisations 2 Pharmaceutical Industry 5 Academia 7 Other (media) 1 Total 23
Table 7 Number of expert respondents per sector
Interviews were conducted in English and a few in French. All local policy actors were proficient in either language. Medical or pharmaceutical studies in Laos for example are offered mainly in French. In their role, most local policy actors re are required to work collaboratively with counterparts from other countries, in English. For these reasons, no interpreter was required for the interviews. As a French educated in English speaking countries, I was comfortable conducting interviews in either language.
Data management and anonymity
Various respondents clearly expressed their fear at associating their name or that of their organization to this sensitive topic, as this might affect their professional relations or personal safety. For this reason, effective data management and preservation of anonymity were key concerns through the data collection and analysis process. there were four categories of anonymity as described in the consent forms shared with respondents ahead of the scheduled interview. A total of only 25 respondents provided full consent, and a total of 12 respondents refused for the information to be cited directly. The reasons provided for refusing to be cited as part of this study were to preserve close working ties with government officials in country, because of non-disclosure agreements under donor polices, or because of fears of disapproval from senior management. As most respondents requested that their contributions anonymized and only few agreed to be quoted directly, I chose to anonymise all quotations in the results chapters.
Color Code Consent Type
GREEN 1
BLUE 2
ORANGE 3
RED 4
Consent type 1. Respondent agrees to take part in this interview and for quotes and other material arising from his/her participation to be used and attributed by name. (Note some
individual statements were still requested to remain anonymous during the interview.)
Consent type 2. Respondent agrees that material from my interview may be quoted, but would like his/her name to be anonymised.
Consent type 3. Respondent agrees that material from my interview may be quoted, but would like his/her name to be anonymised as well as any other information that might be used to identify him/her, including the organisation that employs him/her.
Consent type 4. Respondent does not agree for any material from the interview to be quoted but the researcher may use the information from the interview to inform her analysis.
To guarantee the anonymity of respondents and respect confidentiality levels, each respondent is attributed a unique code, used to reference quotations and statements in the results chapters. The master sheet with the respondent codes and corresponding respondent information is kept in a secure password-protected folder on my computer. The information about respondents has been omitted from all published documents when anonymity was requested, ensuring that no additional information could accidentally reveal the identity of the respondents (i.e. institution, name of department etc.). Additionally, each interview recording was re-named with the numerical code assigned to the respondent. To keep track of anonymity requirement, each respondent and matching transcript was given an anonymity colour code. In the numerical code for each respondent, STK stands for locally based policy actors across all categories. EXP stands for expert respondents interviewed outside the three countries. Each stakeholder and expert respondent was then given a number from 01 onwards, in chronological order. References for each respondent’s contribution includes the category of the respondent and the date of the interview. For example, the seventh stakeholder interviewed (07) was a respondent from a national regulatory authority, interviewed on the 1st of April 2015. The code attributed was therefore STK07G_NRA20150401. I use this numerical code structure to reference inputs from respondents in the results chapters.
All recordings were stored in a password-protected folder and removed from the recording device. The data collected is stored in a secure encrypted folder, using the software 7-zip. Data which was stored on my personal laptop during fieldwork was immediately transferred to the secure LSHTM server upon my return to London. To ensure that no data was accidentally lost, it was also stored on a password protected folder on Filr and on an external hard-drive securely stowed in a locked drawer at LSHTM. This data will be stored on a secure platform for a minimum of ten years following study completion, in accordance with the Retention and Disposal schedule of LSHTM.
The categories of respondents are attributed as follows:
G – Government NG – Non-Government
NRA - National Regulatory Agency NMCP - National Malaria Control Program MoH - Ministry of Health
LEA - Law Enforcement Agency IGO - Inter-Governmental Organisation
INGO - International Non-Governmental Organisation LNGO - Local Non-Governmental Organisation
Donor - Donor Organisation (or Country) Academia – Academia
Consult - Consultant
Pharma - Pharmaceutical Manufacturer (or Association)
Country Category Organisation Type
Organisation Names
Cambodia G NRA Department of Drugs and Food (1)
NMCP Cambodia National Malaria Center (1)
NG IGO World Health Organisation Country Office (2), United Nations United Nations Office for Project Services (1)
INGO Malaria Consortium (1), Clinton Health Access Initiative (1), Population Services International (1), United States Pharmacopeia (1), ACT Watch (1)
LNGO Cap Malaria (1). University Research Cambodia (1)
Donor Australian Department of Foreign Affairs and Trade (1), United States Agency for International Development (2), France Fond Solidaire
Prioritaire Mékong (1)
Consult Independent health consultant (1) Laos G NRA Food and Drug Department of Lao (6)
NMCP National Center of Malariology, Parasitology and Entomology (1)
MoH (1)
LEA Customs (1)
NG IGO World Health Organisation Country Office (2) INGO United States Pharmacopeia (1)
Academia Lao-Oxford Mahosoty Medical Research Unit (2), Worldwide Antimalarial Resistance Network (1)
Donor Anonymous (1)
Pharma Domestic Manufacturer (1)
Thailand G NRA Thai Food and Drug Administration (2) NMCP Bureau of Vector-Borne Diseases (1) MoH Bureau of Drugs and Narcotics (1) LEA Customs (2), Police (1)
NG IGO United Nations Office for Drugs and Crime (1) INGO International Organisation for Migration (2)
LNGO PhraRed (1)
Academia Chulalongkorn University (3) Pharma PreMa (1), Local Manufacturer (2)
Presentation of results
The next three chapters present the results for Cambodia, Laos and Thailand in three separate chapters (6-8). Each empirical chapter starts with an introduction to the historical and political context of the country. This background section provides a brief overview of the health system in each country of recent reforms and of the main policy documents regulating pharmaceuticals. It also provides a short introduction to the malaria burden situation in each country. The second section of the empirical chapters offers a chronological overview of the policy processes to improve access to quality medicines and to reduce the circulation of poor- quality AMLs in Cambodia. This is represented in the form of a timeline. The narrative around this timeline is informed by the results of the document analysis and supplemented, where necessary, with data from interviews and from the literature. For each country, I include more information about the regulatory system for each country – by regulatory function. This information is available in the Appendices (Appendices 9, 12, 15). The stakeholder mapping provides an overview of a dynamic network of actors. I present information on these policy actors in a table format which is available in the Appendix section (Appendices 10, 13, 16). Gathering information on all the partners was useful for both the data collection and analysis processes of the thesis. The relationships between the policy actors were too complex to represent visually. However, where relevant, I do include a diagram of their interactions over
a specific program (as in the case of Cambodia). These visual representations only represent a snap shot in time however and the network of policy actors is as dynamic as the issue of poor- quality medicines is complex. The third section of the empirical chapters presents the results from the semi-structured interviews conducted with policy actors on site. I present the main themes alluded to by government and non-government level respondents under each of the six frames that reflect policy actors’ perceptions of the problem of poor-quality medicines - its drivers and challenges. For each country, this section provides an overview of which frames were most prevalent among government and non-government respondents. The last section of each empirical chapter summarises respondents’ interpretations of the challenges and main contributing factors to the problem of poor-quality AMLs. In this last section, I reflect on the implication of findings for policy development in each country.