Study setting, design and period
This hospital based cross-sectional study was conducted at Hawassa University Comprehensive Specialized Hospital (HU-CSH) and Adare General Hospitals which are both found at Hawassa City, Southern Ethiopia. Hawassa City is the capital city of southern nation’s nationalities and people’s region (SNNPR) from May 9, 2019 to June 7, 2019.HU-CSH is a teaching hospital for medicine and health sciences students. It has more than 350 beds, of which 84 are reserved for maternity beds, and perform more than 4378 deliveries per year. Likewise, Adare General Hospital has more than 126 beds, of which around 18 are used for maternity beds, and perform over 4238 deliveries per year. Both Hospitals are used as a referral where most complicated cases from the southern region and neighbor region zones served. Additionally, the city health centers are also referred the complicated cases to Adare General Hospital. These two hospitals are the only one giving ICU service for neonates at Zonal level for the catchment area.
Study participants, inclusion and exclusion criteria
All live births during the data collection period at the selected Hospitals were included in the study, whereas multiple pregnancies, neonatal deaths, and neonates who were referred from other health care institutions that were out of the study hospitals were excluded.
Sample size determination and sampling procedure
Sample size for first objective, assessing Neonatal Near Misses (NNM) was calculated using single population proportion formula. The specifications made during the computation were: Prevalence of NNM 36.7% (17), 95% confidence level, 4% margin of error and 10% compensation for possible missing values. The ultimate sample size was calculated as 614. Sample size for second objective was computed by Epi info7 Statcalc version 7.1.4.0 software by the assumptions of, 95% level of confidence, power of 80%, the ratio of exposed to unexposed 1:1 and percent of outcome in unexposed group 15.9 and AOR of 2. The percent of outcome in unexposed group and AOR were taken from the study conducted in Southern Ethiopia; the determinate variable was premature rupture of membrane [17]. By substituting the above values in to software the estimated sample size was 432. By comparing the two sample size calculated, the first sample size was larger than the second as a result we took 614 as the final calculated sample size for the study. Subsequently, the calculated sample size was allocated for both hospitals proportionally based on their prior annual delivery report. Subsequently, due to the rare cases of NNM, all the consecutive live births were included in the study during the study period.
Data collection and quality assurance
Data were collected by a face-to-face-interviewer administered structured questionnaire and standard data extractions checklist from medical record were used to collect the data. The standard data extraction was prepared by reviewing prior literatures [12, 16, 20] and WHO recommended information [11]. As there are different languages in the study area, Hawassa City, the questionnaire was primarily prepared in English and translated to the regional working language, Amharic, during the interview. Four diploma nurses and two Degree holder Nurses were recruited as the data collectors and supervisors respectively. All questions in the questionnaire were clarified to each data collectors before the data collection period. Likewise, the data collectors were trained on how to ask questions exactly as stated in the questionnaire and provide only non-directive guidance. Following to three days training, data collectors started the data collection. To reduce information bias the questionnaire was pretested on Bushulo Maternity Health Center prior to the actual data collection period. Primary data, socio-demographic and economic characteristics of mothers, were collected through face-to-face-interview and secondary data, obstetrics and medical history of mothers and neonatal characteristics, were extracted from maternal and neonate medical records by standard checklist.
The dependent variable of the study was NNM and coded as 0 for “no” and 1 for “yes”. Neonates who sustained NNM cases were identified by well trained and experienced data collectors using the standard WHO recommended pragmatic or management severity criteria’s (Table 1). Independent variables were: Socio-economic and demographic characteristics (Age, income, Household size, maternal and paternal educational status, place of residence, maternal occupational status, marital status), maternal obstetric history (ANC, frequency of ANC, parity, gravidity, gestational age at first ANC visit , abortion history, Premature Rupture of Membrane, Mode of delivery) and maternal medical history (Diabetic mellitus, Pregnancy induced diabetic mellitus, Anemia, Hypertension, Pregnancy induced hypertension, syphilis).
Data processing and analysis
Data were checked for completeness and consistencies, coded and entered into Epi data version 3.1 then exported to Statistical Package for Social Science (SPSS) version 25 for analysis. Continuous variable, maternal age was summarized by median with IQR because the data were not normally distributed and presented using frequency tables, figures and charts. The bi-variable and multivariable logistic regression was used to identify the possible factors of neonatal near miss at Hawassa city governmental hospitals. A variable with p value ≤ 0.2 during bivariate analysis was entered in to multivariable logistic regression for further analysis so as to control the confounding variables. Multi co-linearity was checked by collinearity statistics (Variance inflation factor). Finally, Adjusted Odds Ratio (AOR) and 95% Confidence Intervals (CIs) were used to declare statsticall significance.