This was a qualitative study focusing on seven hospitals on the two Northern States, covering 2 tertiary hospitals and 5 secondary and district hospitals. Between January to April 2007, 74 semi-structured interviews were conducted with: 54 health providers (including nurses, medical officers, gynaecologists, medical social workers, counsellors and hospital managers) responsible for providing services to abused women; 8 policy makers at regional and ministerial levels; and 12 key informants (see Table ). Very few counsellors were interviewed as they are primarily available at specialised hospitals, but also because counselling to women who experienced abuse is offered primarily by medical social workers.
Details of in-depth interviews
Snowball sampling was used to identify health providers and key informants who were knowledgeable about the research topic and the researched model, with the assistance of the local partners. Respondents were selected according to their profession and their experience with violence issues (primarily violence against women) and their connection with OSCCs.
Semi-structured interview guides were developed for each respondent type as they offered core questions around their views about violence and their challenges when providing services to abused women. Respondents were asked about providers’ experiences in offering services, the challenges and opportunities they faced, and their views on the adaptation and scale-up of OSCCs. The structure of the interviews tried to encourage respondents to discuss matters they might have otherwise not revealed without prompting such as their beliefs towards IPV. It also ensured that issues relevant to the research were not completely overlooked and allowed comparison across sites and between States.
Prior to the fieldwork, the guides were tested before use, with the support of the local senior partners from the University Sains Malaysia (USM), and further refined and finalized. After piloting the instruments, the guides were modified accordingly and some questions simplified and refined also during the interview period. The majority of interviews were conducted in English. Malay interviews (primarily with nurses) were conducted by two local research assistants, who had been previously trained on the interviews-topic guides.
These interviews were complemented by a document analysis of hospital protocols, and site observations in each facility. In particular, a brief facility checklist was developed for facility observation and was used to collect information about the services offered at the selected facilities. For instance, it contained questions about hospital characteristics (e.g. number of beds, outpatients, services provided), about OSCC features (staff, services on site and on referral, observation of the room), about guidelines and protocols used, and about training offered. These data will be used for describing the visited settings in another article.
Digitally recorded interviews were saved onto a computer and subsequently transcribed. Malay transcripts were translated into English and their accuracy was checked by a local person and the field supervisor in Kelantan.
A framework analysis method was used to analyse the emerging themes. The transcripts of in-depth interviews were read repeatedly to familiarise with the text, to have a full picture of the data collected, and to begin to identify some main themes throughout such initial reading. Once finalised the cross-cutting thematic code framework, interviews were coded and managed by using NVIVO (N7), a qualitative software package (N7) used to help analyse and code narrative texts.
All through the analysis, the code framework was further revised to add new categories of codes, where new sub-codes and themes were identified. The coding system was further refined when we began to group together some linked themes and we came up with fewer broad overarching issues, linked to the main study objectives:
health providers’ attitudes and views (e.g. their views on IPV, on their role, on abused women, etc.). This contains a series of sub-themes that reflect providers’ characteristics as individuals;
challenges faced by the health providers when managing OSCC cases at health care delivery (e.g. lack of time, lack of training, etc.). This contains sub-themes about challenges met at individual level while delivering care, but also at organisational and health care delivery level (e.g. lack of human resources, lack of appropriate training, little collaboration, etc.);
challenges at policy-making level (as perceived by policy-makers).
Once coded, relevant information related to each main theme was combined and examined more closely.
Data were triangulated internally by interviewing different groups and through the use of a variety of data collection methods. The same issues were explored with different study groups (at health service and at policy level), in individual discussions and in different States. In addition, dissemination workshops helped verify that our own understanding of some of the issues were truthful to the providers’ views.
Ethical approval was granted by the Ethical Committees of the LSHTM, the World Health Organisation Scientific and Ethics Review Group (SERG), and the national Malaysian ethical review committee. Additional permissions to conduct interviews with health providers in the selected hospitals were also obtained by the State Health Departments in Penang and Kelantan.
Written informed consent was obtained from each participant in the study. Each informant was asked to sign an informed consent form and anonymity of all information was assured by asking them whether they agreed to be quoted in disseminating reports. A copy of the consent form was given to each respondent to keep. A Malay version was also available for respondents who would not be comfortable in English.
A coding system for all research sites within countries was put in place. In particular, confidentiality of records, tapes and transcripts was assured through numeric coding. Interviews were recorded, where consent was given to tape. Each interview was coded and no name appeared either on the informed consent or on the checklist. In addition, when the respondent agreed to be taped, a recording code was attached without the name of the interviewee, so that the respondent would not be identifiable by the transcribers. Transcripts had no signs of identification for participants. Care was taken in ensuring that any quotes used during the dissemination were anonymised to ensure that the individual respondents/participants could not be identified.