Cover Image
close this book12. Referral in Pregnancy and Childbirth: Concepts and Strategies
View the document(introduction...)
View the documentSummary
View the documentIntroduction
View the documentReferral and Its Function in District Health Systems
View the documentReferral situation and interventions at community level
View the documentReferral Situation and Interventions on the Level of First Line Health Services
View the documentReferral Situation and Interventions on First Referral Level
Open this folder and view contentsThe Crucial Issues
View the documentSteps in Improving the Referral System
View the documentReferences

Referral and Its Function in District Health Systems

The term referral is used in different ways: For instance, it is used to indicate the advice of a health worker to attend a higher-level health unit, whether followed or not. Here we use the term referral for any upwards movement of health care seeking individuals in the health system ( Figure ). There are many ways to do this with respect to pathway, timing and urgency. Thus, we can categorise referrals in pregnancy and childbirth as (1) institutional or self-referral, depending on the involvement of first line services; (2) antenatal, delivery or postnatal referral; and (3) elective or emergency referral.

The following data on pregnancy-related referral in Tanzania provide an overview of levels and categories of referral in a rural African district (Jahn et al . 1998). Based on the analysis of 415 hospital maternity in-patients, the following referral pattern was observed (values as percentage of all maternity admissions; percentage of expected birth in the catchment area in brackets):

· Self-referral 70% of all maternity admissions (15% of all expected births) vs. institutional referral 30% (6%)

· Referral for delivery 84% (18%) vs. antenatal referral 16% (3%)

· Elective referral (including referral for general safety reasons) 98,8% (20,8%) vs. emergency referrals 1,2% (0,3%).

Figure 1. The health care pyramid at district level [Adapted from the Mother-Baby Package (WHO 1994)]

Thus, self-referral for delivery - often without specific medical reason - is the most common mode of referral, while institutional referral is less frequent and emergency referral is very rare. Similar observations have been reported from a rural district in Nepal (Jahn et al . 2000) and Burkina Faso (Falkenhorst & Jahn 1997) with population-based rates of emergency referrals of 0,4% and 0,7% respectively. Nkyekyer 2000 reports from a teaching hospital in Ghana 82% self referrals and 2% emergencies among hospital deliveries. A high proportion of self referrals (80%) has also been observed in Kenya (1996). From a professional point of view, this skewed referral pattern results in an inappropriate use of referral level care by by-passers of first line services. Measures such as disincentives (e.g. fees) for self-referrals and incentives for institutional referrals have been suggested (1988), but are problematic as discussed later in the context of informed decision making.

As referral is a dynamic process, we will analyse the current referral pattern and discuss potential interventions along the following model of the referral chain (Figure 2).

Figure 2. The referral chain

While this model identifies the series of actors and links, the current debate on referral is often restricted to the issue of lack of transport and communication. Without denying the importance of this aspect, we want to emphasise the role of all system components (Kowalewski et al . 2000). Despite relatively good accessibility of obstetric referral facilities many cities have a high maternal mortality, such as 148 in St Louis (Senegal) to 852 in Kaolack (Senegal) as shown in table 1.

Table 1. Maternal mortality ratios in selected African cities

City (country)

Maternal Mortality Ratio per 100,000 live births (C. I. 95%)


Dar Es Salaam (Tanzania)


(Urassa et al ., 1995)

Conakry (Guinea)


(Toure et al . 1992)

Brazzaville (Congo)


(Le Coeur et al . 1998)

Bamako (Mali)


(Etard et al. 1996)

Bamako (Mali)

275 (126-521)

(Bouvier-Colle et al. 1998)

Abidjan (Côte d’Ivoire)

428 (228-732)

(Bouvier-Colle et al. 1998)

Niamey (Niger)

371 (149-764)

(Bouvier-Colle et al. 1998)

Nouakchott (Mauritania)

161 (52-376)

(Bouvier-Colle et al. 1998)

Ouagadougou (Burkina Faso)

318 (146-604)

(Bouvier-Colle et al. 1998)

St Louis (Senegal)

148 (31-433)

(Bouvier-Colle et al. 1998)

Kaolack (Senegal)

852 (453-1,457)

(Bouvier-Colle et al. 1998)

A study from a large referral hospital in Karachi found the 118 mothers brought dead to the hospital maternity had all been residing within a 8 km range. Social and cultural factors played the most significant role (Jafarey & Korejo 1993). A recent maternal death enquiry from South Africa related 18% of avoidable deaths to problems of transport but 57% to problems of the in-service management of emergencies (National Committee on Confidential Enquiries into Maternal death 2000). Similar findings are reported from Ghana (Walraven et al . 2000).