Study area
The study was conducted in Gamo Gofa Zone, which is one of the 14 Zones in the Southern Nations Nationalities and People’s Region (SNNPR). The Zone has 15 districts (woredas) and 2 town administrations. Arba Minch Town, the Capital of Gamo Gofa Zone, is 502 km south of Addis Ababa. Gamo Gofa Zone is a zone with two Lakes (Lake Chamo and Abaya). The Zone is known for its banana, apple and fish production which may impact child nutrition and survival. There were three hospitals and 68 health centers providing health services for the population during the study period. In 2014, the total population of the Zone was projected to be 1,901,953 (with 943,834 males and 958,119 female, 285,043 Urban (15 %) and 1,616,910 Rural (85 %) residents) [9].
Arba Minch Zuria District has been selected as study site for the current study, as it is the study site for the Arba Minch DSS which is relatively new site in the country and as the District has three climatic/geographic zones (Dega(high land), Woina dega (mid land) and Kolla(low land)); which is suitable to represent population of different agro ecological zones. The District lies on 168,712 square kilometers and constitutes 29 kebeles (lowest administrative units in Ethiopia). The total population of the district was projected to be 185,302 (with 92,680 males and 92,622 female) in 2014. Arba Minch Town, which is the capital of the Zone, is included to represent the urban population of the Zone. The total population of the Town was projected to be 135,452 (with 68,132 males and 67,320 female) [9]. The Town was divided in to 11 urban kebeles.
The Arba Minch DSS was established in 2009 in one of the districts in the Zone (Arba Minch Zuria District), which was part of the current study. Arba Minch DSS is based in 9 kebeles of the district. It was established by conducting base line survey/census during July 01-September 30, 2009. Since then, it has been tracking information on vital events (birth, death, migration etc.) continuously. The total population of the DSS was 59,875 with 12,241 female in the reproductive age (15-49), 9825 under-five and 2388 under one year of age children (2011 report of the DSS).
Study design and period
A cross-sectional study design was conducted to assess the magnitude of under-five mortality in 2014, as part of a doctoral thesis work “assessment of magnitude and determinants of under-five mortality and its effect on maternal mental health in Gamo Gofa Zone, Ethiopia”.
Source and study population
The source population was all under-five children in the study area whereas, the study population was all children born between September 01, 2007-September 30, 2014.
Inclusion criteria
All children (alive and dead) together with their respective mothers born between September 01, 2007-September 30, 2014 were included in the study.
Exclusion criteria
Those who were still births were excluded.
Sample size determination
The sample size was determined using single population proportion formula by considering the prevalence of under-five mortality to be 88/1000 [3]. By taking 95 % confidence level and 1.5 % margin of error, the minimum required sample size for the study was 1371. By applying a design effect of 1.5 and adding 5 % to compensate for non-response, a total of 2158 under five children were required. However, all children who had been identified during census of the selected kebeles were included in the analysis to increase the precision and able to estimate other categories of childhood mortality rates.
Sampling technique
Arba Minch Town and the Arba Minch Zuria District were selected purposively out of the 15 districts and 2 town administrations of the Zone. Then, all the 11 kebeles of Arba Minch Town and the 9 kebeles of the Arba Minch Zuria District which are part of the Arba Minch DSS were included (initially these kebeles were selected randomly out of 29 kebeles in the District) and additional 11 kebeles from those kebeles which were not part of the Arba Minch DSS were selected randomly. Accordingly, 31 kebeles from the two districts were included in this study (11 from Arba Minch Town and 21 from the Arba Minch Zuria District). This number was assumed to provide adequate number of sample for the subsequent studies.
Then, a census of the 11 non-DSS kebeles of the Arba Minch Zuria District and 11 kebeles of Arba Minch Town had been done in order to identify all children (alive and dead) born between September 01, 2007-September 30, 2014. The children were followed retrospectively by asking the respondent about whether the child was alive or dead at the time of the survey. If the child was dead, the date of death was recorded.
As the Arba Minch DSS has been tracking all births and deaths since its establishment in 2009, children born between August 01, 2009 and September 30, 2014 in Arba Minch DSS kebeles were tracked from the data base of the DSS. Therefore the data since 2009 were tracked from all the 31 kebeles and the data since 2007 were tracked only from the 22 kebeles (Fig. 1).
Data collection
A pre-tested Amharic questionnaire was utilized for data collection. The questionnaire was developed in English and translated to Amharic, then back translated to English to check for its consistency. Finally, the Amharic Version was used for data collection. Variables in the questionnaire include: sex of the child, date of birth of the child, whether the child is alive or dead, if dead date of death and other identifiers (identification number (for the child and the mother), district name, kebele name, house number). At least two data collectors (grade 10 or above) per kebele were recruited and trained on the procedure. Four master holders (in Public Health) supervised the data collection process. The principal investigator had been strictly following the data collection throughout the process. Besides, additional data were sought from the kebele admirations and health posts through reviewing documents and/or interviewing the kebele officials or the health extension workers (HEWs) to determine characteristics of the kebeles.
Data processing and management
The data were edited, coded, entered into computer and cleaned using Epi Info Version 3.5.1 and the analysis was performed by open-epi version 2.3, SPSS version 16 and STATA 11 as appropriate. The daily collected data were transferred to the Arba Minch University and locked in a secure cabinet which was arranged in the compound of College of Health Sciences of the Arba Minch University on daily basis. The data were entered into Epi info by two data encoders after having training/orientation on the template, the procedures for insuring the quality of the data during data entry and the importance of quality of data. They were also expected to identify incomplete and inappropriate data and communicate to the principal investigator at this stage too. This was strictly followed and checked by the principal investigator on daily basis.
Data analysis
Descriptive analyses with frequency and cross tabulation with the corresponding confidence interval with p-value were made. As we collected data on complete live birth histories of all mothers within the last 7 years before the survey, we have applied a direct method to estimate mortalities.
Accordingly, birth cohort method was applied to determine overall level of childhood mortalities (only deaths of children born during the study period were included in the numerator). Whereas, death cohort method was used for trend analysis (deaths of children born prior to the target year may be included in the numerator of that year). Extended Mantel-Haenszel chi square for linear trend was also performed to assess presence of linear trend through the study period using open-epi version 2.3.
Weighted analysis was conducted to account for the non-proportional allocation of the sample to urban and rural kebeles. The sampling weight was calculated using the following notion: by determining sampling probability at two stages (district and kebele levels), as a complete census/coverage of individuals in selected kebeles was made.
P(kth individual in jth kebele in ith district being selected) = P(ith district being selected)P(jth kebele selected| ith district is selected)P(kth individual selected| jth kebele is selected, which is one (as a complete census was made)). Then the weights were the reciprocals of these probabilities [10].
Accordingly, the sampling weight for urban was: As one out of 2 urban districts was included and all the kebeles in selected district were included. The probability of selection of individuals in urban kebele = 1/2*1*1 = 0.5. The corresponding weight calculated to be 2. For that of rural: as one out of 15 rural districts was included, twenty out of 29 kebeles of the district were included. The probability of selection of individual in rural = 1/15*20/29*1 = 0.046. The corresponding weight calculated to be 21.8.
Data quality assurance
The questionnaire was pre-tested and corrections were made accordingly. Two days training was given to data collectors and supervisors on the questionnaire and the procedures. The data collection process was strictly followed up. All collected data were checked every day for their completeness, clarity and consistency by supervisors and the principal investigator. Any unclear and ambiguous data were corrected by recollecting data from actual study population by going back to the field, while minor errors were corrected by the principal investigator as deemed necessary. About 5 % of the households were re-visited by the supervisors/principal investigator to check the validity of the information collected by the data collectors. Then, data were cleaned and checked before data entry and analysis again. Besides, double entry of 10 % of the questionnaire was made to monitor any discrepancies.
Ethical considerations
Ethical clearance and approval was obtained from the Institutional Review Board of the College of Health Sciences at Addis Ababa University. Letters were written to all concerned bodies (Gamo Gofa Zone Health Department, Arba Minch Zuria District and Arba Minch Town Health Office and administration of all kebeles) and permissions were secured at all levels. After explaining about the purpose of the study and confidentiality of the data, verbal consent was obtained from each respondent. To assure the confidentiality of the responses, anonymous interviews were conducted. Besides, the daily collected data were transferred to the Arba Minch University and locked in a secure cabinet on daily basis.