Time-to-death and predictors of mortality among preterm neonates diagnosed with neonatal sepsis, Southwest Ethiopia, 2022: prospective follow-up study

The study was conducted in government hospitals at the Southwest Region; officially the South West Ethiopia Peoples’ Region is a regional state in southwestern Ethiopia. It was split off from the Southern Nations, Nationalities, and Peoples’ Region (SNNPR) on November 23, 2021 after a successful referendum. It consists of Keffa, Sheka, Bench Sheko, Dawro, West Omo Zones, and Konta special woreda. The 3 hospitals are Mizan-Tepi University teaching hospital (MTUTH), Tepi general hospital, and G/tsaddik shewa general hospital. MTUTH is located 584 km far from Addis Ababa. The Tepi general hospital is in the Sheka administrative Zone, 622 km away from Addis Ababa. The G/tsaddik Shewa general hospital is in Keffa zone, 464 km away from Addis Ababa. Hospitals provide delivery and other maternal and newborn care including NICU services. The study was conducted between March 2021 and January 30, 2022 at the government hospitals.

A prospective cohort study was conducted among preterm neonates with neonatal sepsis admitted to the NICU wards of hospitals between March 2021 and January 30, 2022.

Power log rank estimated sample sizes for 2-sample comparison of survivor functions—Log-rank test; Freedman method was used to obtain representative samples of preterm neonates admitted to the NICU with diagnosed neonatal sepsis as follows: E=(z@12+ZB)2(lnHR)2p(1−P)=(z@12+ZB)2θ2π(1−π)E = {{{{\left( {z@{1 \over 2} + ZB} \right)}^2}} \over {(lnHR)2p(1 - P)}} = {{{{\left( {z@{1 \over 2} + ZB} \right)}^2}} \over {{\rm{\theta }}2\;\pi (1 - \pi )}}

Where:

z@1/2: z value at 95% confidence interval (CI) = 1.96;

ZB: the power of the study = 80%;

P = 50% cumulative mortality rate;

p = mortality rate;hazard ratio (HR) = 0.5;

Ɵ = ln (HR), i.e., HR = exp. (Ɵ) and then Ɵ = ln (HR).

The total sample size was calculated as n=Ep(E){\rm{n}} = {{\rm{E}} \over {p(E)}}.

Adding a 10%, for incomplete data/missing data the final minimum sample size required was 363.

From 7 government hospitals in the southwest region, MTUTH, Tepi general hospital, and G/tsaddik Shewa general hospital were selected using the lottery method.

Proportional allocation was performed for each hospital based on the previous number of preterm neonates diagnosed with neonatal sepsis and admitted to the NICU. Thereafter, we included all preterm neonates diagnosed with neonatal sepsis who were admitted to the NICU. The study participants were selected from each hospital using a consecutive sampling technique. All recruited study participants were followed until the outcome of interest (i.e., either death or censure) appeared.

Dependent variable: Time-to-death and dichotomized into death or censored.

Independent variables: Include maternal and neonatal socio-demographic variables, medical, obstetric, surgical, and neonatal characteristic variables.

Median survival time: The median survival of time of 50% of neonates.

Survival status: The outcome of premature neonates diagnosed with sepsis, either death or censored.

Death/Event: Refers to the neonates that have died while in the therapeutic units and declared as dead in the treatment logbook by the physician.

Censored: Preterm neonates diagnosed with neonatal sepsis and admitted in the NICU. Those alive beyond 28 days of neonatal age, those left against medical advice, or discharged with improvement and referred.

Time origin: Specific time of admission of preterm neonates at NICU with diagnosis of neonatal sepsis.

Time of exit: Specific time of observation outcome (death or censored).

After reviewing the literature, the tool was adapted and prepared in English and translated into the local language, Amharic, and re-translated back into English (Supplementary File 1). Data were collected through face-to-face interviews with the mothers and neonatal chart reviews (primary and secondary). Secondary data were used to obtain baseline information that the mothers could not address. After excluding all baseline data on different related factors, data on factors that can occur starting from the first day of follow-up were collected during follow-up (i.e., from the time of admission to 28 days of neonatal age) or at the occurrence of an event of interest (i.e., death or censored). Only the first episode was included if the patients were admitted with >1 episode of neonatal sepsis over 28 days.

After coding, editing, and cleaning data were entered into Epi-data manager version 4.4.2.1 (The EpiData Association, Odense, Denmark) and exported to Stata version 14 (StataCorp LLC, Texas, USA) for analysis. The median survival time was estimated using Kaplan-Meier survival estimates. Predictors of mortality were identified using the Cox proportional hazards (CPH) model. The assumptions of the CPH model were assessed using the Schoenfeld residual/global test (0.2194). Multi-collinearity was assessed using the variance inflation factor (VIF) (1.21). Based on bivariate analysis, variables with P-value <0.25 were transferred to the multivariable analysis. Variables with P-value <0.05 at 95% confidence level in multivariable analysis were considered as independent predictors for mortality.

Pretest was conducted on 5% of sample size at Chena General Hospital, before actual data collection period and then editions and wording were made. Training was provided to the data collectors and supervisors on data collection techniques and ethical issues. Supervisors regularly checked the completed questionnaires.

In this study, 178 neonates (50.28%) were male and 176 (49.72%) were female. A majority (97.46%) of the preterm neonates were in the age category of ≤7 days during admission. About 306 (86.44%) of women were found in the age range of 20–35 years with a median age of 25 years (interquartile range of 22–30) (Table 1).

A total of 116 (32.77%), 92 (25.99%), 164 (46.33%), and 42 (11.86%) preterm neonates with neonatal sepsis also had comorbidity of respiratory distress syndrome (RDS), hypoglycemia, hypothermia, and prenatal asphyxia, respectively (Table 2).

A total of 340 (96.05%) of the preterm neonates diagnosed with neonatal sepsis were born at health institutions. Most (85.59%) of the preterm neonates were born with low birth weight (<2500 g) with a median weight of 1950.00 g. A total of 303 (85.59%) preterm neonates were born with gestational age (GA) of 32–36 weeks with median GA of 34 weeks and interquartile range of 32–36 weeks; 238 (67.23%) preterm neonates had 1st-minute Apgar score range <7; 232 (65.54%) preterm neonates also had 5th-minute activity, pulse, grimace, appearance, and respiration (APGAR) of ₃7 and only 200 (56.50%) preterm neonates received kangaroo mother care (KMC) at the time of hospitalization in the NICU (Table 3).

A total of 330 women (93.5%) had a history of at least one antenatal care (ANC) follow-up during their current pregnancy. Only 42 (11.86%) women had a history of obstetric complications such as placental abruption, placenta previa, and pregnancy induced in current pregnancy. The majority (75.99%) were single tone type of pregnancy. Most preterm neonates were born from a multi-para mother, which was around 53.11% (Table 4).

This study was conducted by considering censoring and time to event using survival analysis and it was easy to establish a temporal relationship between the dependent and independent variables. The study used prospective data, and the better quality of data on the primary exposure and exposures are assessed before outcomes occur, hence there is less likelihood for bias. In addition, the study was conducted in multicenter settings using prospective follow-up, which increases the reliability and generalizability of data. Finally, the study has novel findings on an important topic that has not been explored before.

We had a relatively small sample size, which is difficult to generalize to all preterm neonates diagnosed with neonatal sepsis. Second, since the study was conducted among preterm neonates diagnosed with neonatal sepsis and were admitted to only government hospitals, it cannot be generalized to the preterm neonates admitted to private hospitals and healthcare centers.