Study design and site
A cross sectional study was conducted between April - June 2011 in Hai and Siha districts of
Kilimanjaro Region, Northern Tanzania. Hai and Siha districts are among the 7 districts of
Kilimanjaro Region. Hai district had a projected population of about 197,000 by June 2010, based on 2002 national census. Hai district has a total of 57 health facilities where 46 are providing Reproductive and Child Health (RCH) services. These include; 2 hospitals, 6 health centres and 38 dispensaries both government/public and voluntary agencies. Siha district had projected population of about 120,000 with total of 22 health facilities, 8 owned by the government and 14 privately owned. Of the 22 health facilities of Siha district, 15 provide RCH services. A total of 36 of 61 facilities with RCH services in the 2 districts were randomly selected and included in the study (24 facilities from Hai district and 12 from Siha district).
Study population, sample size and sampling
The study population included health care workers (HCWs) from selected health facilities both government and private, providing RCH services in Hai and Siha districts. HCWs included nurses and clinicians providing services to mothers and children i.e. those working in maternity ward or labour ward, working in RCH clinics, and in paediatric wards. We excluded health workers that did not consent for participation, were administrators not supervising RCH services and on leave during the study.
Sample size was calculated using Epicalc 2000. The prevalence of knowledge of breastfeeding among health care providers of 19.2% was used (from a study conducted in Nigeria by Okolo and Ogbonna, in 2002), alpha error level was set at 5%, power of 80% and addition of 5% for non-response. Substituting the values, a minimum sample size of 250 was obtained.
A multistage sampling technique was used. The first stage involved selection of health facilities. The health facilities were selected to reflect all levels of health care provision in the district including hospitals, health centres and dispensaries both government and private. All hospitals (N = 3) were included in the study, while simple random sampling was used to select seven of nine health centres and 26 of 49 dispensaries. The second stage was selection of healthcare workers reflecting the proportion of HCWs at different levels. Proportionate sampling was used with 38% (95) of participants from hospitals, 27% (68) from health centres and 35% (87) from dispensaries, giving a total sample of 250 health care providers. Simple random sampling was then used to select providers at respective health facilities.
Data Collection
A questionnaire consisting of both closed and open-ended questions was used to collect information from HCWs providing RCH services. A checklist was used to assess if breastfeeding education is given to pregnant and post-partum women, and if nurses demonstrate skills to women in post-natal wards in initiating and maintaining breastfeeding.
Instruments and Measurements
The questionnaire included socio-demographic characteristics of the HCWs (age in years, sex, marital status, education, cadre (nurse or clinician), years of experience, in-service training on breastfeeding, and district of work and facility ownership). Other variables measured were knowledge, attitudes and supportive practices on EBF. The questionnaires were self-administered and made available in Kiswahili (the local language). The questionnaire was pretested in health facilities not included in the study to ensure clarity of interpretation and ease of completion.
A standardized checklist was used for observation of supportive practices on exclusive breastfeeding by health care providers in ten health facilities including the three hospitals, four health centres and three dispensaries with RCH and delivery services.
Knowledge on exclusive breastfeeding
Seven questions covering four themes were used to summarize the level of knowledge.
The themes were; knowledge of the WHO definition of EBF (had four items: knowledge on WHO definition and duration on EBF, knowledge on breastfeeding initiation (had three items: when to start initiation of breastfeeding, importance for breastfeeding initiation and colostrum giving), knowledge of frequency of breastfeeding (one item: demand feeding) and knowledge on benefits of exclusive breastfeeding (two items: for the mother and child). Correct knowledge of any of the item was given a score of one; giving a minimum score of 0 and a maximum score of 4. Those who scored 3 or above were graded as having sufficient knowledge on EBF and those scored below 3 were categorized as having poor knowledge.
Attitude towards exclusive breastfeeding
Likert type scales (ranging from 1–4) were used for attitudinal questions to allow for varying degrees of agreement or disagreement. A score of 4 implied strong agreement, 3 implied agreement, 2 implied disagreement and a score of 1 implied strong disagreement with positively framed attitude statements whereas; scoring were reversed for a negatively framed attitude statement. A score of 4 implied strong disagreement and a score of 1 implied strong agreement. There were 7 attitude questions with maximum score of 28 and minimum score of seven. All questions were given equal weight. Those scoring 17 and above were categorized as having positive attitudes towards EBF.
Skills and supportive practices on EBF
Skills/practice was measured by 3 questions with the aim of categorizing those with good practice and those with poor practice. Overall score on supportive practice for EBF was calculated based on 4 points of advice for positioning, 4 points on advice for attachment, and
4 points on advice to facilitate breast milk flow. The maximum score was 12 and a provider who scored 7 or more was considered to have “good supportive practice for EBF” and score below 7 to have “poor supportive practice on EBF”.
Data Analysis
Data was cleaned, checked for consistency and open-ended questions categorized and coded before analysis. Data was analysed using computer software SPSS version 16.0. Descriptive statistics were used to summarize data, proportion for categorical variables and mean or median with their respective measures of dispersion for continuous variables. Difference between groups was analysed by using Chi squire test for categorical data.