Background: Ninety-six percent of the world’s 3 million neonatal deaths occur in developing countries where the majority of births occur outside of a facility. Community-based approaches to the identification and management of neonatal illness have reduced neonatal mortality over the last decade. To further expand life-saving services, improvements in access to quality facility-based neonatal care are required. Evaluation of rural neonatal intensive care unit referral centers provides opportunities to further understand determinants of neonatal mortality in developing countries. Our objective was to describe demographics, clinical characteristics and outcomes from a rural neonatal intensive care unit (NICU) in central Uganda from 2005-2008. Methods: The NICU at Kiwoko hospital serves as a referral center for three rural districts of central Uganda. For this cross sectional study we utilized a NICU clinical database that included admission information, demographics, and variables related to hospital course and discharge. Descriptive statistics are reported for all neonates (<28 days old) admitted to the NICU between December 2005 and September 2008, disaggregated by place of birth. Percentages reported are among neonates for which data on that indicator were available. Results: There were 809 neonates admitted during the study period, 68% (490/717) of whom were inborn. The most common admission diagnoses were infection (30%, 208/699), prematurity (30%, 206/699), respiratory distress (28%, 198/699) and asphyxia (22%, 154/699). Survival to discharge was 78% (578/745). Mortality was inversely proportional to birthweight and gestational age (P-value test for trend <0.01). This was true for both inborn and outborn infants (p < 0.01). Outborn infants were more likely to be preterm (44%, (86/192) vs. 33%, (130/400), P-value <0.01) and to be low birthweight (58%, (101/173) vs. 40%, (190/479), P-value <0.01) than inborn infants. Outborn neonates had almost twice the mortality (33%, 68/208) as inborn neonates (17%, 77/456) (P-value <0.01). Conclusions: Understanding determinants of neonatal survival in facilities is important for targeting improvements in facility based neonatal care and increasing survival in low and middle income countries.
Kiwoko hospital serves as a rural referral center for three districts of central Uganda (combined population: 600,000) where 31% of births take place at home and only 33% of births are registered with the government [18]. As of 2006, this region had a neonatal mortality rate consistent with other rural areas in Uganda (33 per 1000 live births) and an under-5 mortality rate of 129 per 1000 live births [19]. Kiwoko hospital has 4–8 doctors working at a time and has bed capacity for up to 300 patients. The NICU in Kiwoko hospital accepts inborn and outborn newborns with gestational ages greater than 24 weeks and up to a chronologic age of 3–6 months. It opened in 2001 as a 20-bed unit with 300 admissions per year and was staffed by one nurse and two midwives [20]. The most commonly reported admission diagnoses at that time were prematurity, neonatal tetanus, prematurity with respiratory distress, and birth asphyxia. By 2012, the NICU had 35 beds, admitted 600 newborns per year, the majority of which were born at the hospital, and was staffed by 23 nurses and an assigned physician. A prospective NICU database containing 203 demographic and clinical variables for all admissions was established in 2005, allowing for more in-depth analysis of patient characteristics and outcomes. During the period of this study, the NICU had capacity for thermoregulation, intravenous hydration, cup and nasogastric tube feedings, limited phototherapy equipment, limited number of oxygen concentrators for use with nasal cannula, and an intermittent electrical supply. Neonates requiring pediatric subspecialty care required transfer to the national referral center in Kampala, located two hours’ travel by car. (Figure 1) During the study period there was at least one other NICU in Uganda (at the national hospital in the capital city) that had similar capabilities to care for ill neonates. Location of Kiwoko Hospital, Uganda. Kiwoko Hospital in Nakaseke district shown with hut symbol (used with permission of the ISIS foundation). This study was conducted by retrospective review of all admissions to the NICU from December 2005 through September 2008. Prior to 2005, no medical information systems existed in the Kiwoko NICU. Beginning in October of 2005 nurses recorded relevant health information at the time of admission for each infant using a medical record form developed for this NICU. The form was then updated throughout the hospitalization by the nursing staff, and completed at discharge/death of the patient. The medical record form included: maternal and delivery characteristics (e.g., maternal age, type of birth attendant), neonatal clinical characteristics (e.g., birth weight, admission diagnosis), hospital course (e.g., blood transfusions, oxygen therapy) and discharge status. Data from all forms during the study period were manually entered in to an electronic database in SPSS. All analyses reported in this paper are restricted to infants less than 28 days of age at admission. Descriptive statistics were used to assess demographic and clinical characteristics as well as neonatal outcomes. Medians and ranges were calculated for continuous variables. As appropriate, continuous variables were converted to clinically meaningful categories and analyzed as categorical variables. Frequencies were calculated for categorical variables. All analyses were disaggregated by location of birth, with those born at Kiwoko Hospital referred to as “inborn”, and those born outside of the hospital in a surrounding village health post or at home referred to as “outborn”. The chi-square test was used to evaluate differences between inborn and outborn infants and linear regression test for trend was used to evaluate differences in mortality based on birth weight and gestational age group. Gestational age was determined (in order of preference) by the mother’s report of her due date as determined from prenatal care, her last menstrual period or using Ballard assessment at admission if available. When birth weight was not known, admission weight was used if the patient was admitted within 3 days of birth. Admission diagnoses were categorized (such as prematurity, respiratory distress, infection, etc.). A neonate could have multiple admission diagnoses recorded. For admission diagnoses, reported percentages reflect the number of neonates with a particular diagnosis recorded as a proportion of the total number of neonates included in the study. For all analyses the reported unit of analysis is the neonate. The University of Washington Human Subjects Division reviewed the protocol and designated this study Minimal Risk and granted a waiver of consent due to the nature of the retrospective review and inability to contact subjects that were included in the database (HSD Study #43072). Local approval was obtained from Kiwoko hospital. This research has adhered to the STROBE guidelines for cross sectional studies.
N/A