MALARIA PREVENTION AND CONTROL STRATEGIES

Get Complete Project Material File(s) Now! »

CHAPTER 3 RESEARCH METHODOLOGY

INTRODUCTION

Research experiences regarding malaria prevention and control and a theoretical model (HBM) that guided the study have been addressed in the second chapter. This chapter presents the research methodology used in this study. The research method employed data collection in two phases. The sources of data for phase I included households in malaria endemic districts of the Sidama Zone in Ethiopia. The second phase of data collection was accomplished at the health posts (the smallest grass roots level health service facilities in Ethiopia). In each phase, the study design and specific research objectives addressed by the study are subsequently presented. The study population, sample size estimation and procedures followed to select representative samples from the target populations are described. Furthermore, data collection instruments, ethical considerations, data analysis plans and efforts to minimise systematic and random errors are discussed.

PHASE I: HOUSEHOLD SURVEY

Phase I of this study aimed at examining the implementation of malaria prevention and control strategies at the household level. The research procedures pertinent to phase I are discussed in the following sections.

Research paradigm

Polit and Beck (2008:761) define a paradigm as a way of looking at a natural phenomenon that encompasses a set of philosophical assumptions that guide a researcher’s approach to addressing research question(s). Paradigms for researchers involving humans are characterised by the ways in which they respond to philosophical research questions. One of these ways of epidemiological investigations is a positivist world view (Bruce, Pope & Stanistreet 2008:3). Positivism, as a philosophical view, was developed in the 18th century and initially dominated the scientific world views (Bruce et al 2008:6).
The current research approach applied a positivist paradigm, using a quantitative design. This paradigm assumes that there exists an observable reality that can be measured quantitatively independently of human observation (Bruce et al 2008). A quantitative approach is a precise measurement and quantification, often involving a rigorously controlled design (Polit & Beck 2008:763). This type of research is a more appropriate choice for collecting empirical data through the senses (Polit & Beck 2008:16). Since this design addressed a range of confirmatory questions which could lead to inferences enhancing the RBM programme, it is used in this study.

Research design

Elston and Johnson (2008:31) defined a research design as ‘the conceptual structure within which research is conducted’. It can be classified in different ways. Among them are qualitative or quantitative studies, applied versus basic, deductive versus inductive, experimental and non-experimental categories, and process versus content research (Brink & Wood 1998:4). During the past decade, several scholars classified research designs into experimental and non-experimental (Aschengrau & Seage III 2008:139-140; Fathalla 2004:44; Rothman et al 2008:87-88). Regardless of the ways in which research designs are classified, all the designs generally, constitute the plan for the collection, measurement and analysis of data and for reaching conclusions based on the study’s findings, in relation to the objectives stated for conducting the study. The latter design is also called observational and may be descriptive or analytical. It is more appropriate when the researcher does not intend to assign interventions, but is rather, an observer of the problem (Rothman et al 2008:87). A descriptive design provides a detailed description of the characteristics of interest while the analytical design describes associations in terms of possible cause and effect relationships (Bruce et al 2008:422-425).
Quantitative research is relevant, according to Zegeye, Worku, Tefera, Getu and Sileshi (2009:35), when:
• the investigator wants to answer a research question which is best addressed by a quantitative answer, describing the measurements numerically;
• the variable of interest changes and the rate of that change is accurately estimated by numerical evaluation;
• it is useful to quantify opinions, attitudes and behaviours and to find out how the whole population feels about a certain issue.
Furthermore, quantitative research is suitable for explaining some phenomena or identifying factors associated with the outcome event of interest. When the researcher intends to justify a hypothesis, it is possible by adopting the quantitative approach.
The objectives of phase I, aimed to:
• assess the knowledge, attitudes and practices of households in the study area about the prevention and control methods of malaria;
• identify factors that could be associated with malaria prevention and control activities in the study areas;
• describe the malaria preventive measures implemented at household level; and
• ascertain which persons are most affected by malaria at household levels in the study areas
Each research objective is addressed by a quantitative method. For instance, the three behaviour-related variables in the first objective were described by computing the proportion of respondents across each category used to measure knowledge, attitudes and practices of respondents regarding malaria prevention and control (details are discussed under section 3.2.5.). As Polit and Beck (2008:16) recommended, the remaining three research objectives are effectively measured in quantitative terms. As a result, the quantitative research design was selected as an appropriate approach for this study.
While the quantitative design was used, the researcher recognised and tried to minimise the effect of limitations arising from the quantitative design, such as a lack of representation of local constituencies due to the researcher’s response categories. The possibility of missing contextual details and structural bias were recognised as a limitation of this method.

Research method

Research methodology is the overall approach to studying the research topic of interest including issues such as the constraints, dilemmas and ethical choices. They are the techniques an investigator uses to structure a study to collect data relevant to the research questions (Polit & Beck 2008:765).

Description of the study area

This study was conducted in one of the 13 zones found in the SNNPR. It is located in the north eastern part of the region and 275 kilometres south of Addis Ababa, the capital city of Ethiopia. It is bordered by Oromia regional state in the east and north, by Wolayta and Gedeo Zones in the west and south, respectively. The zone has a total area of 6 981.8 square kilometres and contains a total of 19 districts. Sidama Zone has three ecological zones namely the low land, middle land and highland constituting 30%, 54% and 16% of the geographical area, respectively. The zone is home to a population of 3 288 083 with a gender ratio of 1:1 and is one of the most densely populated zones in the region (471 persons per square kilometre). According to the zonal health department, 94.3% of the population lives in rural settings. There were a total of 597 public health facilities during the data collection period. These include three hospitals, 94 health centres and 500 health posts (Sidama Zonal Health Department 2011:2).

Research population

Bruce et al (2008:133) and Fathlla (2004:50) define a population as “group of people in whom the investigators are interested, and to whom the results of the study are applicable”. Polit and Beck (2008:761) also delineated a population (sometimes called a universe) as referring to the entire set of individuals or objects having some common characteristics which the researcher(s) wish to investigate. According to the Ethiopian Public Health Association (2009:40), the target population in a research context, is defined as a group of individuals who comply with the characteristics of the population from which the study sample is to be drawn.
In the context of these definitions, this study was conducted in nine malaria endemic districts of the Sidama zone in Southern Ethiopia. Many community-based studies (except those limited to reproductive issues) conducted in Ethiopia include household heads (which mostly include men) due to their decision making powers. However, the main disease prevention and health promotion activities at the household level, involve women, as active role players in the family. Although malaria prevention and control efforts are the responsibilities of all household members, the role of women is significant to prevent, diagnose and get treatment for malaria. Because of this, the research population for the first phase comprised households in the malaria endemic districts and mothers or caregivers from the selected households were the study units. In Ethiopia, mothers (housewives or employed women) are the prominent caregivers in the family and have experience related to sickness from malaria and its prevention and control at the household level. This includes knowledge of malaria transmission, signs and symptoms of malaria, breeding and resting sites of mosquitoes, utilisation of nets/ITNs, application of IRS, use of prophylaxes (IPT) and health seeking behaviours. Thus, households in the nine malaria endemic areas of the study zone were considered as the target population for phase 1 of the study to which the findings were generalised.

Sample size

According to Bruce et al (2008:133), a sample is a group of individuals taken from a larger population of interest. Sample size refers to the number of individuals or items to be selected from the universe to constitute a sample, which should neither be excessively large, nor excessively small (Kothari 2004:56; Polit & Beck 2008:348). The sample size must be optimum and fulfil the requirements of efficiency, representativeness, reliability and flexibility (Kothari 2004:56). In estimating the sample size for population proportion, Bruce et al (2008:158) suggested that researchers need to specify the value of the anticipated population proportion, level of error to accept and level of confidence. These principles were applied to estimate the optimum sample size required for this study. One of the objectives, this study aimed to address was to identify factors associated with malaria prevention and control, which requires a moderately large sample. Examination of the existing data indicates that 70% of the population at risk of malaria infection used ITNs nationally in Ethiopia (FMoH 2007a).
In another study that assessed factors associated with ITN use by individuals in households owning ITNs in Ethiopia, individuals with good (highest knowledge score of 5 and above in a scale categorised into 0-1, 2-4 and >5) level of knowledge about malaria were found to be 2.4 times more likely to use ITNs compared to those with less knowledge (Graves, Ngondi, Hwang, Getachew, Gebre, Mosher, Patterson, Shargie, Tadesse, Wolkon, Reithinger, Emerson & Richard 2011:356).The researcher, therefore, assumed that the rate of ITN utilisation was 70% and 55% among those with good and poor knowledge, respectively. That provided an estimated odds ratio (OR) of 2.0. Considering the odds ratio of 2.0 between individuals with good and poor knowledge about malaria, 95% level of confidence, a power of 80%, Epi-Info (version 3.2.2.) yielded a sample size of 201 for each category and 402 for both (those with good and poor level of knowledge about malaria). A ratio of 1:1 for those with good to poor malaria knowledge was considered. Allowing a designing effect of 2.0 for cluster sampling and non-response rate (10%) produced a total sample of 885 respondents. Sample estimated for the second objective, which provided the largest sample, was used as a minimum size for the study. Table 3.1 shows the summary of the sample calculated for each objective.

Selection of respondents

Respondents were mothers from randomly selected households permanently living in the nine participating malaria endemic districts in Sidama Zone, Southern Ethiopia. To be specific, a multistage sampling technique was employed. Multistage sampling is a sampling strategy that proceeds through a set of stages from larger to smaller sampling units (Polit & Beck 2008:758-759). This sampling approach is economical and practical, especially when the sampling unit involves clusters and is from a large population distributed over a wide geographical area (Polit & Beck 2008:347). In this study, the primary and secondary sampling units included districts and kebeles (smallest administrative units) respectively (figure 3.1). There was no sampling at the district level, because all nine malaria endemic districts were included. Selection of clusters (2 rural and 1 urban kebeles from each of the nine districts) was employed using a simple random sampling technique. This sampling technique provides a known chance of being selected for every kebele in the district (Bruce et al 2008:137). Finally, clusters of households (locally named ‘gots’) were randomly picked from a list of all ‘gots’ in each kebele and all households in the selected got were assessed. One cluster (got) usually contains 30-60 households. See figure 3.1 for details of sampling procedures.
Once the kebeles and ‘gots’ were selected, the leaders of the respective kebeles and ‘gots’ were contacted in advance to inform them about the purpose of the study. They were asked for permission to enter their villages. Permission letters from the districts were also submitted to these leaders.

Exclusion and inclusion criteria

Exclusion criteria are defined as the characteristics that a defined population does not have (Polit &Beck 2008:753) whereas the inclusion criteria are the characteristics that the eligible population possesses. This study aimed to generalise the findings to the target population in the Sidama Zone. Therefore, respondents who were not permanent inhabitants of the selected ‘gots’ such as visitors at the time of data collection were excluded. Women with mental disorders, people who were ill, and those who could not properly communicate were excluded from participating in the study.

Research tool/Interview schedule

An interview schedule is a list of questions that are read out or verbalised by an interviewer to a respondent (WHO 2001:171). It is the instrument that specifies the wording of all questions to be asked of the respondents in structured self-reported studies (Polit & Beck 2008:756). Interview schedules are usually designed as unstructured with open-ended items, semi-structured or structured with close-ended items (Bruce et al 2008:167).
The structured interview is one in which the researcher specifies in advance all possible answers from which the respondents choose the most appropriate option. This type of tool ensures consistency in the range of answers and facilitates data analysis. Interviews and self-completion structured questionnaires are the most commonly used methods of collecting information during quantitative studies (Bowling & Ebrahim 2005:204). The questions were presented to the respondents in a similar way in terms of wording with responses presented in pre-coded response options. It enables the researcher to collect relatively unambiguous data collected from a large sample. The use of the structured interview schedule provides an opportunity of ensuring that the respondent has understood the questions.
Structured interviews might have some drawbacks. Pre-coded response choices might not accommodate all potential choices in a comprehensive manner (Bowling & Ebrahim 2005:204). Respondents might not always share the same perspectives and concepts that the researchers had when they compiled the response options. Questions might include behavioural attributes and, therefore, might lead to potential social desirability bias.
The population for this study comprised rural communities and the majority were expected to be uneducated. Because of this, the researcher utilised purposively constructed, translated into the local language (Sidama) and pre-tested structured interview schedules that were prepared based on the research objectives and the review of the literature. The tools were developed mainly using closed-ended questions with a list of possible responses and very few open-ended questions. Closed-ended questions ensure consistency in the range of responses provided and are suitable for coding and statistical analysis (Bruce et al 2008:165). The instrument was structured as follows (with the list of questions shown in parentheses as shown in annexure 1):
• Section I: geographical location (1-4)
• Section II: socio-economic and demographic variables (5-19)
• Section III: housing condition (20-28)
• Section IV: knowledge level of respondents (29-53)
• Section V: environmental factors (54-61)
• Section VI-X: domains of the HBM (62-106)
• Section XI: practices of the respondents regarding malaria prevention (107-124).
• Section XII: reported prevalence of malaria (125-130)
• Section XIII: observation checklist (131-138).
The interview schedules were prepared in English and translated into the local language, known as ‘Sidama’. That was retranslated back to English by another expert in the field and checked for consistency. A bilingual (English-Sidama) expert translator checked these translations (see annexure 12 for the translator’s letter approving the translations and specifying his qualifications). The instrument was pretested during the training of the data collectors to identify limitations, such as vague questions, phrases or terms and also the estimated average time required to complete each interview. Necessary adjustments and corrections were made accordingly to improve the reliability of the instrument. During the pre test of the tool, the following limitations were identified and corrected:
• typographic errors or missing words (Items:37,95,100,102,104,105, 109,123, and 136)
• multiple responses were not included for some questions (Items: 39, 40, 41, 109, 111 and 124)
• conducting the interview before 10:00 a.m. was inappropriate as most women were busy with several activities (cleaning rooms, cooking, fetching water).
• each interview took a long time to be completed.
Corrections were made to improve the reliability of the tool. The data collection time was revised and the starting time was set to be 10:00 a.m. At this time, most women finished their activities and were willing to be interviewed. In order to reduce the effect of the long duration of each interview, every interviewer became familiar with the questions. Every interviewee was also told the approximate time the interview would take so that she could decide whether to participate or not.

Selection and training of interviewers

Eighteen data collectors were recruited from the selected respective kebeles. Kebele offices were contacted to identify potential female data collectors that satisfied the following set criteria:
• those who at least completed high school education
• had no permanent job (as the data collection was conducted during the regular working hours.
• those who knew the village
• had experiences of data collection (optional)
The use of the local persons as data collectors is useful in that they know the physical setup of the village and also understand the culture. Females who had completed grade 12 but who did not have permanent jobs were given priority. Data collectors with this level of education can easily understand concepts and terms used in the instrument during the training. Moreover, female data collectors were preferred for easy communication with female interviewees. Data collectors were contacted and selected during the preliminary visits to the study sites and they were approached through the kebele and ‘got’ leaders.
The training of the data collectors was conducted centrally at zone level for two days. During the training, emphasis was given to important points such as brief overviews of the study, study objectives, features of face-to-face interviews and techniques of interviewing were discussed and practised by means of role play sessions. Ethical aspects of research were also addressed. The interviewers signed an agreement with the researcher that they did not write any identifiers of the respondents such as the identification numbers and names on a questionnaire and also properly kept the filled questionnaires in a folder till they submitted to the researcher. The instruments were discussed item by item. That was followed by interviews conducted among the trainees during role play sessions facilitated by the researcher. The last day of the training was used for pretesting the tool in one of the rural kebeles (Alamura) around Hawassa town, which did not participate in the actual study. Every interviewer signed an agreement of confidentiality with the researcher that no information obtained during the interviews would be divulged to anybody except the researcher.

Data collection

Data collection for this study was carried out during March to May 2013. The range of data collected included geographical location of the respondents, background variables such as socioeconomic and demographic characteristics, and housing conditions. Moreover, the study assessed variables related to the level of awareness and knowledge about malaria, perceptions and practices relevant to the prevention and control of and episodes of malaria and its treatment measures. The interviews were conducted by trained female research assistants. The data collectors then interviewed the mothers/caregivers from all households in the selected clusters (gots) through house-to-house visits.
The procedures of the interview schedule included that:
• the data collector moved from house to house only during working hours but starting after 10am;
• the interviewer greeted the family and introduced herself to the household members;
• she explained the purpose of the study to the respondents as stated in the consent form (see Annexure 3);
• the interviewer asked whether the respondent was willing to participate in the study;
• those who were not willing were thanked and the household was skipped;
• once the respondent has expressed her willingness, the data collector asked the respondent to choose a convenient place for the interview;
• the data collector built rapport with the respondent and told her to ask any questions or clarifications at any time during the course of the interview;
• the data collector then read each question to the interviewee and properly recorded every response.
The investigator supervised the process of data collection to check that the interview schedules had been completed accurately and was available in case any queries or crises arose.

Socioeconomic and demographic variables

Fifteen items (questions 5-19) assessing the respondents’ background information, using nominal variables, were developed considering previous similar studies. These included: gender, age, marital status, religion, ethnic group, level of education, household size, occupation and income.

Housing condition

Prevention of malaria is partly affected by the condition of the house such as type of the house and number of rooms, family size, whether the house is shared with domestic animals, the presence of a kitchen, toilet and window and door screens. The researcher designed nine items (questions 20-28) to assess these housing conditions. Besides the structured interview, an observation checklist was used to confirm some of the responses. Eight checklist items (questions 131-138) were used to observe the condition of the house’s walls (smooth/roughness) presence of ITNs, window screens, mosquito breeding sites, openings in the upper wall parts that serve as entrances for mosquitoes. Data on distance of the house from health posts, availability of transportation services, and vector control activities such as IRS were recorded

TABLE OF CONTENTS
CHAPTER 1 ORIENTATION TO THE STUDY
1.1 INTRODUCTION
1.2 BACKGROUND INFORMATION
1.3 OVERARCHING ELEMENTS OF THE RESEARCH
1.4 PHASE I: HOUSEHOLD SURVEY
1.5 PHASE II: HEALTH POST LEVEL SURVEY
1.6 ETHICAL CONSIDERATIONS
1.7 DEFINITIONS OF KEY CONCEPTS
1.8 SIGNIFICANCE OF THE STUDY
1.9 STRUCTURE OF THE DISSERTATION
1.10 SUMMARY
CHAPTER 2 LITERATURE REVIEW
2.1 INTRODUCTION
2.2 THE GLOBAL BURDEN OF MALARIA
2.3 THE BURDEN OF MALARIA IN ETHIOPIA
2.4 MALARIA PREVENTION AND CONTROL STRATEGIES
2.5 CHALLENGES IMPACTING ON MALARIA PREVENTION AND CONTROL ACTIVITIES
2.6 THE THEORETICAL ORIENTATION
2.7 APPLICATION OF THE HEALTH BELIEF MODEL TO THE STUDY
2.8 SUMMARY
CHAPTER 3 RESEARCH METHODOLOGY
3.1 INTRODUCTION
3.2 PHASE I: HOUSEHOLD SURVEY
3.3 PHASE II: POST LEVEL SURVEY
3.4 SUMMARY
CHAPTER 4 ANALYSIS AND DESCRIPTION OF THE STUDY’S FINDINGS
4.1 INTRODUCTION
4.2 RESEARCH RESULTS
4.3 SUMMARY
CHAPTER 5 ANALYSIS AND DISCUSSION OF THE RESEARCH FINDINGS (PHASE II)
5.1 INTRODUCTION
5.2 RESEARCH FINDINGS
5.3 SUMMARY
CHAPTER 6 CONCLUSIONS, RECOMMENDATIONS AND LIMITATIONS OF THE STUDY
6.1 INTRODUCTION
6.2 COMPARISON OF THE FINDINGS OF PHASE 1 AND PHASE II OF THE STUDY
6.3 LIMITATIONS OF THE STUDY
6.4 CONCLUSIONS
6.5 RECOMMENDATIONS BASED ON THE MAIN FINDINGS OF THE STUDY
6.6 THE IMPLICATIONS OF THE STUDY’S FINDINGS FOR MALARIARELATED ISSUES IN ETHIOPIA
Programme
GET THE COMPLETE PROJECT