High proportions of obstetric referrals in Addis Ababa: the case of term premature rupture of membranes

Study settings

The study was conducted in Addis Ababa, the capital of Ethiopia as part of an intervention
project that intends to improve maternal and neonatal health outcomes through intensive
knowledge and skills training for midwives/nurses on basic EmONC. The city is home
for about 3.5 million people and is administratively divided in 10 sub-cities. Under
the City Administration, Health Bureau, there are over 90 public primary HCs, which
provide basic EmONC and four regional public hospitals providing comprehensive EmONC.
These hospitals are Zewditu Memorial, Ghandi Memorial, Tirunesh Beijing and Yekatit
12. Ghandi Memorial is a maternity hospital while the other three are general hospitals
that receive referrals from all over the city. Moreover, there are federal specialized
referral hospitals in the city, which also provide comprehensive EmONC. Generally
speaking availability, accessibility and acceptability of EmONC services are quite
high in the city 11], 12]. The numbers of basic and comprehensive EmONC facilities outnumber the WHO minimum
standard and the median distance to the nearest comprehensive EmONC facility is 5 km.
Eight five percent of mothers in the city give birth in health facilities with the
great majorities at public facilities.

Study design and sample selection

Using a mixed methods approach, this study collected quantitative and qualitative
data. The study employed a sequential explanatory design. Routine retrospective data
were collected from registers and preliminary analyses were made. This was followed
by qualitative interviews to explore the issues behind the numbers and come up with
plausible explanations.

The study used Consolidated Criteria for Reporting Qualitative Research (COREQ) guideline
for the qualitative findings 17] (see Additional file 1).

To have a representative sample, ten well-established primary public HCs were randomly
selected one from each sub-city using a lottery system. All the selected HCs are primary
care facilities providing basic EmONC and have a similar staffing profile. The HCs
provide delivery services free of charge and serve heterogeneous, low-income group
of women as many women from high-income strata opting for private facilities. Hospital
data were collected from two regional and one federal referral hospitals, which all
together attend about two-thirds of the deliveries in the public health facilities
in Addis Ababa. Zewditu Memorial and Ghandi Memorial hospitals were selected randomly
from the four hospitals under the Addis Ababa City Administration Health Bureau. The
third hospital was Tikur Anbessa, a federal referral hospital randomly selected from
the two federal referral hospitals, catering obstetric and neonatal care services
in Addis Ababa.

Quantitative data collection and analysis

This review retrieved retrospective data from labour, delivery and intrapartum referral
logbooks from the selected HCs. These included total number of women who sought care
during labour and delivery, number of total referrals, referrals due to term PROM
and number of full-time skilled providers from January 1st, 2012 to December 30, 2012.

To identify women who were referred with a diagnosis of PROM, routine hospital referral
registers were first checked. By using their unique patient identifiers, in total
227 individual patient records were retrieved and reviewed. Women referred for other
obstetric and neonatal complications were not included. The hospital data were collected
using checklists to assess the standards of care for term PROM and for assessing maternal
and neonatal outcomes. The checklist included maternal age, gravidity, parity, last
menstrual period, gestational age, referring facility, diagnosis made by the referring
primary health center, diagnosis after hospital arrival, prophylactic antibiotic,
time referred from the referring facility, arrival time in hospital, admission to
hospital, if a mother was not admitted to the hospital and referred again, reason
for second or third referral from a hospital, time of delivery/induction or caesarean
section, action at the hospital (induction, augmentation or caesarean section) and
mode of delivery. Mean and range values were calculated for continuous data, while
proportions and Chi square tests were calculated for categorical data.

Definitions of terms and synonyms

In this study a mother is said to have term PROM when she fulfils the following three
criteria 1], 2]: (1) she should be at 37 completed weeks of gestation or more, (2) the amniotic membranes
should have ruptured, (3) she should not be in labour. Due to the difficulty to ascertain
whether the woman was in labour or not upon referral, we used the time taken from
referral to delivery as a proxy indicator. By definition normal labour could take
12 h in multigravida and 18 h in primigravida mothers 18], 19]. Taking into consideration the travel time and logistic challenges to reach from
HCs to hospitals or from hospitals to hospitals, we set 9 h as a cutoff. Therefore,
a mother who had spontaneous labour and vaginal delivery within 9 h of referrals were
considered to be in labour by the time the referral was issued and the rupture of
amniotic membranes for this mother could have been a sign of labour. Those women who
had ruptured amniotic membranes and who had spontaneous labour and delivery after
9 h of referral were considered to have PROM.

In this study primary HCs are also referred as basic EmONC facilities or public HCs.
Tertiary hospitals are also referred as comprehensive EmONC facilities or hospitals.

Qualitative interviews

Key informant interviews were conducted. Ten head midwives, one from each HC were
approached for the interviews and all of them agreed to participate. Using a focused
interview guide, the interviews were conducted after obtaining informed verbal consent.
The guide explored experiences on (1) how PROM diagnosis was made? and (2) how mothers
with term PROM were managed at health center? (see Additional file 2). All the interviews were conducted in Amharic, the national language fluently spoken
by all the interviewees and the interviewer. The principal investigator (PI) did all
the interviews in the HCs. During the interviews, dialogues were made to continue
to the point where no new information was coming up and took on average 15 min. Notes
taken during the interviews were transcribed and then translated to English for analyses
by the PI. Doing the transcriptions and translations allowed the PI to get immersed
into the data for gaining an overall impression of the findings. The interviews were
analysed using content analysis. According to the principles in content analysis 20]–22], the interview transcripts were first read and re-read to have an overview of the
data. Then by using the two major questions from the interview guide, two themes were
identified. The first theme was ‘making PROM diagnosis’ and the second theme was ‘managing
PROM’. Then after the interview transcripts were sorted out and aligned with the respective
themes, which then followed quantitative interpretations of the findings as shown
in Table 1. In qualitative content analysis, the process of data analysis also involves interpretation
of findings and can be presented in the form of frequency 20]–22]. To ensure the validity of the findings, we did participant checking by calling up
the informants in a consultative meeting where we presented preliminary findings.
At the meeting, the informants gave us positive feedbacks that the findings presented
were the main issues addressed during the in-depth interviews.

Table 1. Shows variable diagnostic and management approaches for term PROM explored in key
informant interviews in Addis Ababa

Ethical considerations and study permits

The project obtained ethical approval from the Ethics Committee of Addis Ababa City
Administration, Health Bureau (AACAHB), and the Ethics Committee in Western Norway.
Study permits were obtained from the AACAHB, sub-city health bureaus, hospitals and
health centers. First, the AACAHB issued us a support letter for all sub-cities health
bureaus. Based on the support letter, we got permission from each sub-city health
bureau to access all the health facilities under them. Finally, in reference to the
support letter from the sub-cities, the head of each health facility granted us access
to the registers and to do the qualitative interviews. As stated in the protocol approved
by the ethics committees’, informed verbal consent to participate in the in-depth
interviews was secured from each informant after explaining the purpose of the study.
Prior to the interviews all the informants were informed about the interview procedure
and their right to opt out at any point during the interviews without any consequence.
For ethical reasons, instead of their names all the HCs were de-identified using capital
letters.