A quasi-experimental design will be followed in which health facility assessments will be undertaken in six health facilities and their catchment areas in three districts and in one large maternity hospital in Nairobi. A before and after design will be undertaken in the intervention area and among an equivalent comparison population living in areas and utilizing health facilities not served by a D&A program in order to control for potential time dependent confounding at least among observed characteristics. The study population will comprise health facilities, health providers and managers in selected facilities, national level managers and policy makers, women in labour and postpartum women and community members. The sites will be divided into those providing standards services, voucher program only sites, voucher and D&A sites and D&A sites only.
Proposed data collection procedures
All data collection instruments and procedures will be thoroughly pilot-tested prior to being used in settings similar to those in which they will be administered. Guidance will be sought from women who have recently given birth when finalizing the instruments to ensure that questions and the data collection procedures are conducted with appropriate sensitivity and are not perceived as stigmatizing. The study tools will be pre-tested among a small group of women with similar characteristics as the study population to identify potentially negative consequences and modified accordingly. Study tools for clients will be translated and pretested in the relevant local language. Overall the study design will comprise the following methods:
Focus groups discussions (FGDs)
Three to five FGD will be conducted with community members in each district. These will include women who had facility-based birth and those who delivered at home, family members, community health committee members, and local women’s and civil rights groups. Potential participants will be recruited from the communities around the health facilities. Standard FGD procedures will be followed; a group moderator will conduct the session, accompanied by a note-taker. The FGDs will explore broad themes around (i) motivations for giving birth in health facility and selection/use of the particular facility (ii) attitudes towards health facility deliveries (iii) quality and satisfaction with communication/interaction with the provider and (iv) perceptions of health system stigma and discrimination towards those accessing maternal health services. In addition the FGDs will explore and clarify views, opinions and perceptions and enhance the findings from client exit surveys and address unforeseen questions arising from other components of the study. The information generated from the FGDs will be used to; (i) focus and refine the nature, manifestations and indicators of D&A generated through the literature / document review and policy analysis; and (ii) to inform the development of baseline indicators that will measure the effect of interventions at end line.
In-depth interviews and informal discussions with senior health managers
We will conduct in-depth interviews with 25 senior reproductive health program managers at national and regional level; district health management teams (DHMTs), facility managers and civil society leaders. The objectives of the interviews will be to collect information similar to that collected through FGDs, with additional information on system level and governance factors that may contribute to abuse and disrespect.
Health facility assessments
We will conduct health facility assessments which comprise; (i) interviews with providers (ii) review of inpatient records (iii) structured facility inventory (iv) service statistics (v) observations of client-provider interactions during labour and childbirth and (iv) client exit interviews. We will conduct provider interviews to measure their perceptions of working conditions, respect, empathy with and prejudice against clients, and awareness of policy and service delivery guidelines concerning respect, dignity and client rights. Staff turnover, absenteeism, vacancy rates, workload, motivation and the challenges of managing and retaining maternity staff will also be measured through adapting self-administered instruments developed and validated in South Africa [16
]. All providers working in the MCH or Maternity units will be interviewed on their knowledge, practice and attitudes. Lickert scales will be used and providers given the opportunity to self administer part of the interview. The first part of the questionnaire will be administered by the researcher, a separate section with a range of statements covering the various issues will be given to the provider and asked to tick responses ranging from strongly agree to strongly disagree.
(i) Reviews of inpatient records will be used to assess the response to complications. We will conduct record reviews with maternity inpatient notes/partographs specifically to record for example, time of diagnosis of obstructed labor and time of cesarean section to measure aspects of untimely care and neglect. We will randomly review the delivery register and extract every fifth inpatient notes for review.
(ii) Structured facility inventory will be used to measure information on facility infrastructure, staffing levels, skills and training, and availability of equipment, commodities, test kits, data collection tools, as well as protocols for service delivery (guidelines, policies and standards).
(iii) Service statistics
, and reports of critical care behaviors where available, will be extracted to measure trends in the number of facility-based births and analyzed using time series charts [17
(iv) Observation of client provider interaction
during labour and delivery will measure both the process (how clients are treated and whether they actively participate) and the content (what they are told, technical competence, accuracy of information, provision of essential information during consultations). A structured non-participant observation will be undertaken of client-provider interactions for those clients recruited following the client’s consent. A structured checklist will be used in the observation of the client provider interaction to record their observations of the actions taken by the provider. The checklist comprises a set of items that together comprise pre-determined, agreed-upon and validated indicators of quality of care during labour and delivery. We will then observe the labor and normal delivery of all consenting clients per facility attending for childbirth over a two-week period. There is a precedent for doing this in Kenya, as the recently-completed Service Provision Assessment (SPA) observed normal deliveries using tools we adapted for this study [11
(v) Exit interviews of postpartum women
leaving the maternity unit: We will conduct exit interviews with all women aged 15–45
years who have recently delivered, at the selected facilities as they leave the maternity units after giving birth. Exit surveys are aimed to measure the prevalence of D&A during childbirth. All women satisfying these inclusion criteria will be recruited until the required sample sizes had been reached.
To estimate the sample size for evaluating the ultimate outcome indicator of the intervention package, “the reduction in the prevalence of D&A in facilities” we will use an estimated prevalence of 0.2 of women not using facilities due to provider related reasons, based on literature and regional experience where an additional study on D&A is also taking place. This is based on results from a voucher evaluation baseline population survey conducted in 2010 across the six districts, indicated that 22% of women said they did not seek medical care because of unfriendly providers. Therefore calculation of the sample size is based on the estimation that 20% of pregnant women who do not use a facility for delivery is due to provider based reasons. To measure a 10% decrease of D&A, with 80% estimated power for one-sample comparison of proportion with two sided alpha of 0.005 we will require a sample size of 381 for the exit survey for each district.
The distribution of the types of abuse will be assessed in terms of key socio-demographic variables such as age, parity, marital status, education, economic status, ethnicity, and other characteristics identified through the qualitative research that may be associated with discrimination or abuse.
Case narratives at community level
We will document examples from women who have experienced D&A and others who have not experienced D&A during childbirth. Women will first be interviewed exiting the postnatal ward and recruited for a follow-up interview at their home or a mutually convenient location within four weeks of childbirth to ask questions regarding their experience (using note taking and tape-recording).
Validation of data collection tools
We will develop a standard client exit questionnaire to assess abuse and disrespect in facilities that will be validated. This will aim to permit the instrument to be used in similar settings to assess prevalence and types of abuse and disrespect. To validate this tool a series of activities will be conducted. The process will constitute the following steps:
1. Finalizing a Construct Map (matrix) aimed at identifying the measureable elements of disrespect and abuse (Table
2. Conducting a series of interviews with clients to identify potential gaps in the Construct Map. This will be established by inviting 20 women from the focus groups and other women in the community identified by focus group participants who had experienced abuse during facility childbirth in the past year and 20 women with facility delivery in the past year who had not experienced abuse to complete a structured questionnaire. These women will then participate in a semi-structured in-depth interview by a different interviewer, who will be blinded to the responses of the questionnaire. The experiences of specific types of abuse established in the semi-structured interview will be used as the gold standard. We will then compare the results of the in-depth interviews with the results of the exit questionnaire to calculate the sensitivity, specificity, and the positive predictive value of the questionnaire for identifying any abuse and specific categories of abuse.
3. Finalizing the prototype exit interview questionnaire based on the pretested tool and the focus group discussions. This will help in clarifying any gaps in the client exit questionnaire.
4. Assessing the reliability of the exit interview. The questionnaire for “disrespect and abuse” is required to be consistent over the intended survey population. Reliability testing will be conducted to reduce the measurement error. These methods outlined will be used to reduce error caused by influences associated within respondents and the structure of the instrument.
5. Assessing the validity of the questionnaire. We will gather evidence for validity in three ways through internal structure, items analysis, and external measures. We will collect validity evidence based on internal structure evaluating whether the observed data (responses) are consistent with expectations both qualitatively and empirically using rank order correlation. We hypothesize that the items will arrange themselves from low to high along the D&A scale. We will then conduct item analysis (item by item) to determine whether the locations of the respondents for one item performed as expected on other items, comparing mean locations. This is a test of items design. We hypothesize that the mean location of each group will tend to increase as the scores increase.
The consortium for “Measuring D&A in Childbirth in Kenya” will be comprised of Population Council (PC), Federation of Women Lawyers – Kenya (FIDA) and the National Nurses Association of Kenya/Midwifery Chapter (NNAK/MC). This team was strategically created to bring complementary expertise in promoting women’s rights through local and national advocacy (FIDA), empowering health service providers to provide quality care (NNAK/MC), and implementation research to document and learn from the process (PC). Other critical stakeholders include the Ministry of Medical Services, Ministry of Public Health and Sanitation, the White Ribbon Alliance–Kenya and the Health Rights Advocacy Forum (HERAF).
Our strategy for maximizing the likelihood that successful interventions can be institutionalized is to address the five categories of contributing factors identified by Bowser and Hill through engaging with stakeholders at three levels: policy and governance, health system, and community. The landscape analysis that was conducted by Bowser and Hill of D&A in facility-based childbirth details how these levels are interrelated and why they need to be addressed simultaneously [5
Findings from the baseline study will inform development of the intervention package so that the most prevalent categories of D&A in Kenya are addressed directly. The design will be guided by a human rights-based approach, in particular, by analyzing and addressing the inequalities, discriminatory practices and unjust power relations between providers and clients as defined by international human rights treaties and corresponding governmental obligations and laws. In addition, the existing Kenya quality model and the national maternal care standards will be used to guide the interventions relevant at each level care of service delivery. We will conduct community dialogue meetings to share and validate the findings. In addition we will invite key stakeholders from all levels to a meeting at which key findings will be shared and discussed, contributing factors identified and prioritized. Broad approaches for the intervention package at national, facility and community levels will be discussed and agreed upon by the stakeholder group. To support this process, issues emerging from the data will be organized into the five broad categories of contributing factors at the three levels: Policy and Governance, Health System and Community, wherever possible linking these factors directly with the types of D&A experienced in the Kenyan context (Table
Proposed three-level intervention
Data management and analysis
Paper questionnaires and PDAs will be used to capture quantitative data. Checklists for the facility inventory and observations will use paper questionnaires whereas the client exit interview will be carried out using PDAs. Data from paper questionnaires will be keyed into Epidata 3.1 and exported into Stata 10 for analysis. Data from PDAs will be downloaded into an MS Access database before being exported into Stata 10 for analysis. Tests of proportions and relationships (between control and experimental or pre-intervention and post-intervention periods) will be made at 1% and 5% level of significance. We will use statistical analyses such as logistic regression models (with any form of D&A as dependent variable) and Z-tests examining D&A rates before and after to test the hypothesis described above, as well as to evaluate the impact of the intervention package on indicators measuring the key contributing factors. Additionally, multi-level regression models will evaluate which individual and group level factors are associated with a woman’s experience of D&A in childbirth, with the primary explanatory variable being which catchment area woman lived in.
Qualitative data will be captured on paper and audio tapes and later transcribed, translated and typed into MS Word, before being exported into QSR NVivo 10 software management and analysis. A thematic framework will be used in qualitative analysis, allowing for iterative use of both deductive and inductive approaches. We will compare analysis charts within and across sites to look for similarities and differences to support identification of key issues around abuse and disrespect. Final qualitative analysis will be organized around a description of the nature, manifestations and experiences at baseline and factors contributing to the abuse.
All researchers and research assistants will be trained on the conduct of ethical procedures and will be monitored during field work by Population Council.
Informed consent will be obtained separately for each study participant for each component. All participants will be given detailed information about the study including: aims/methods of study; institutional affiliations of the research; anticipated benefits, risks/discomfort it may cause (expected to be minimal) and follow-up of the study; the time the questionnaire or interview will take; the fact that they may choose not to answer any questions and that they have the right to abstain from participating in the study, or to withdraw from it at any time, without reprisal; measures that will be taken to ensure confidentiality and anonymity of information provided; the conduct of interviews in places of the participant’s choosing and which maximize audio privacy; contact details of the study coordinator for any questions or concerns.
All data will be stored in password protected computer files. Hard copies of questionnaires, anonymised transcriptions and tapes of the group discussions will be stored securely in a locked cabinet, in accordance with the Population Council policy and the Kenya Data Protection Policy.
The research protocol has been reviewed by key stakeholders and ethical clearance has been granted by the Kenya Medical Research Institute (KEMRI) Ethical Review Board (approval number SCC No 288), the Population Council’s Institutional Review Board (No.517), and the Division of Reproductive Health, Ministry of Public Health and Sanitation and the Ministry of Medical Services.