Peripartum referrals to korle bu teaching hospital, Ghana – a descriptive study


correspondence Dr K. Nkyekyer, Department of Obstetrics & Gynaecology, Ghana Medical School, Accra, Ghana. E-mail:


Summary A 6-week prospective study of women referred to Korle Bu Teaching Hospital during labour and delivery was conducted to determine the sources of and indications for referral, and to assess the adequacy of various aspects of the referral mechanisms. Of 396 women referred, 86% were referred from polyclinics (which are state-owned) and private midwives. The commonest indications were failure to progress (21.5%) and hypertensive disease (15.7%). A total of 35 (8.8%) women were referred with third-stage or immediate postpartum complications. Referrals constituted 17.6% of hospital deliveries. Significant deficiencies were identified in the referral mechanisms. 72.7% of the patients travelled by public or private means of transport and 54.2% were not accompanied by any staff during transfer. Only 16.7% of those expected to have partographs came with one and 55.6% of women with complications of vaginal bleeding did not have intravenous infusions sited. However, in almost 90% of patients, the general condition on arrival in the hospital was good, and some referrals could probably have been avoided. Efforts must be made to improve patient transport and to evaluate labour management practices in the referring centres.


The outcome of pregnancy, to both mother and baby, depends to a large extent on the quality of antenatal and intrapartum care. Several reports from developed countries affirm the fact that women with uncomplicated pregnancies may be cared for by midwives, and taken through labour and delivery at home or in birth centres with no increased risk of adverse maternal or perinatal outcome compared to similarly low-risk pregnant women who deliver in hospital (Rooks et al. 1989; Ackermann-Liebrich et al. 1996; David et al. 1999). There have been significantly fewer interventions in the former group (Rooks et al. 1989; David et al. 1999). The achievement of these comparable outcomes, however, has been predicated on early recognition of complications and prompt referral and transfer to hospitals, for further management, in the home and birth centre groups. Indeed, in one country in which the perinatal mortality rate in home births was higher than the national average, this was attributed to underestimation of the risks associated with such complications as post-term pregnancy, twin pregnancy, and breech presentation, and a lack of response to fetal distress (Bastian et al. 1998).

Maternal mortality continues to be a major problem in sub-Saharan Africa, with maternal mortality rates being among the highest in the world. In Ghana, for example, although the maternal mortality rate has been officially stated to be 250 per 100 000 live births (Ministry of Health Ghana 1999), figures of 734.4 and 1140 per 100 000 deliveries have been individually reported from the country's two teaching hospitals (Martey et al. 1993; Lassey & Wilson 1998). Many women in sub-Saharan Africa go through pregnancy, labour and delivery without access to health facilities. Where these are available there may be delay in seeking medical attention. Another contributory factor is the failure of health personnel to recognize the severity of the patient's condition, leading to delays in treatment and referral (Fawcus et al. 1996). Inefficient referral systems are further complicated by low compliance due to transportation difficulties and perceived or real negative attitudes of staff at the referral centre (Eades et al. 1993; MacLeod & Rhode 1998; Martey et al. 1998). The referral centre may not be able to manage the complication because of inadequate numbers of appropriately trained personnel or lack of facilities and drugs (Thaddeus & Maine 1994).

Korle Bu Teaching Hospital, the site of this study, is situated in Accra. It is the largest hospital in the country and serves as the teaching hospital for the University of Ghana Medical School. The obstetric unit of the Department of Obstetrics and Gynaecology handles about 12 000 deliveries a year. The unit receives referrals from various lower-level facilities within metropolitan Accra as well as surrounding areas. In most of these facilities the women are cared for almost solely by midwives. Accra city has a population of about 1.8 million people; it also serves as the administrative capital of the Greater Accra Region with a population of about 2.4 million. In addition to Korle Bu Teaching Hospital there are seven other government or quasi-government hospitals, seven polyclinics (which are state-owned), 37 registered private maternity homes and 35 private hospitals in the metropolis. The annual number of deliveries in the Greater Accra Region is about 39 000 (Ministry of Health Ghana 1999). In 3.3% of births in the urban areas of Ghana the mothers would not have received any antenatal care and in 18.1% they deliver at home (Tawiah 1998). Home births are not encouraged in the urban areas and therefore those that occur may be due to unforeseen circumstances or the mothers' decision to deliver at home for various reasons, usually without trained supervision.

The aims of the study were to determine the major sources of and indications for referral of patients during the labour and delivery periods and to assess the adequacy of various aspects of the referral mechanisms.

Materials and methods

The study period covered six consecutive weeks in April and May 1998. A period of 6 weeks was chosen because of logistic constraints; the particular period was selected because it covered part of the time of year (April to June) when the highest number of monthly deliveries are recorded in Korle Bu Teaching Hospital. It was presumed that this would yield the largest study sample within the limits of the logistic constraints. Included were women referred in labour or with immediate postpartum complications and women referred with antenatal complications requiring urgent delivery, although they may not have been in labour at the time of referral.

At the end of delivery or while the woman was in the postnatal ward, files were obtained and the following data extracted: age, parity, complications in past obstetric history, complications in index pregnancy, the facility from which the woman was referred, the indication for referral, time of arrival in Korle Bu Teaching Hospital and whether or not she had an intravenous infusion sited. From the referral letter the following were recorded: the time the woman reported at the referring facility, the time the decision was taken to refer her, the time when she left for Korle Bu, and whether a partograph accompanied the referral.

During the first half of the study the women were interviewed after delivery about the means of transport by which they came to the hospital and whether they were accompanied by any staff from the referring facility. In the case of the only woman who died (10 h after arrival), this information was gathered from the receiving midwife at Korle Bu Teaching Hospital. In the original plan of the study this information was to be provided by the receiving midwives at the outpatients' department (OPD) but in a pilot study it was found that they were too busy to fill in forms meant for that purpose for most women.

The woman's condition on arrival as noted by the receiving midwife, her blood pressure, pulse, temperature and findings on obstetric examination were obtained.

Data analysis was performed by personal computer using Epi-Info 6.04. The χ2-square test was used to compare proportions. Differences were considered significant if < 0.05.


During the period of study there were 396 peripartum referrals, 360 (90.9%) of whom delivered in Korle Bu; the total number of deliveries in the same period was 2041. Thus, referrals constituted 17.6% of deliveries. Thirty-five (8.8%) women were referred with various complications in the third stage or immediately post-delivery, whilst one (0.3%) woman delivered in the ambulance before reaching the hospital. The mean age was 26.9 (SD, 6.0) years. The mean parity was 1.5 (SD, 1.7) with the parity distribution as follows: nulliparous: 154 (38.9%); para 1: 90 (22.7%); para 2: 62 (15.7%), para 3: 38 (9.6%); para 4: 29 (7.3%) and para 5 and above: 23 (5.8%). A total of 214 (88.4%) of the 242 parous women had no adverse factors in their past obstetric history; 15 (6.2%) had had perinatal deaths, eight (3.3%) Caesarean sections, and five (2.1%) primary postpartum haemorrhage. Concerning the index pregnancy, 324 (81.8%) women had no complications, 62 (15.7%) were antenatal clinic non-attendants or poor attendants, eight (2.0) carried twin pregnancies and two (0.5%) had cervical cerclage sutures inserted in private hospitals for presumed cervical incompetence. 216 (54.5%) women were interviewed to determine means of transport to the hospital and whether or not they were accompanied by any staff from the referring facility.

Sources of and indications for referral

Of the referrals, 212 (53.5%) women were referred from polyclinics, 128 (32.3%) from private maternity homes, 24 (6.1%) from private hospitals, 16 (4.0%) from government hospitals and 16 (4.0%) from clinics outside Accra (Table 1). Of the 216 women interviewed 112 (51.8%) were from polyclinics, 73 (33.8%) from private maternity homes, 14 (6.5%) from private hospitals, 10 (4.6%) from government hospitals and seven (3.3%) from clinics outside Accra. Comparison of the proportions that the centres' referrals formed of the total and the proportions they formed of those interviewed showed no significant differences (χ2= 0.56, = 0.97). Thus it is unlikely that there is bias in the information gathered by this means in favour of one referral source or the other.

Table 1.  Number of patients referred, by indication and source of referral Thumbnail image of

A total of 334 (84.4%) women were referred in the first stage of labour, 15 (3.8%) in the second stage and 35 (8.8%) in the third stage or immediate postpartum period (less than 12 h). Twelve women (3.0%) were not in labour but had to be delivered because of the complications for which they had been referred. The commonest indications were failure to progress and hypertensive disease (including pre-eclampsia and eclampsia) in the first stage. Malpresentations were mainly breech presentations (18 cases), with one shoulder and one face presentation. At the time of the study the country was facing an acute energy crisis and therefore there was rationing of electricity; it is surprising that only three (0.8%) of the referrals were for reasons of power shortages at the referring facility. Under ‘others’ were indications such as low maternal age, elderly primigravid, maternal rhesus negative status, vaginal ring pessary in situ, big baby, maternal pyrexia, huge vulval warts, long interval since last delivery and past history of stillbirths, retained placenta or postpartum haemorrhage.

Means of transport and staff accompanying women to hospital

Of the 216 women interviewed 59 (27.3%) travelled in ambulances, 128 (59.3%) in taxis, 15 (6.9%) in family or friends' cars and 14 (6.5%) in tro-tros, which are small buses that carry between 15 and 20 passengers and are much less comfortable to travel in than taxis; moreover, they drop their passengers at their station which is about 400 m away from the maternity block. The majority of those who travelled in ambulances, i.e. 54 (91.5%) of 59, were referred from state-owned facilities (polyclinics and government hospitals) in metropolitan Accra. Although most patients travelled in taxis, those from the state-owned facilities within the city were significantly more likely than referrals from other facilities to travel by ambulance than by any other form of transportation (< 0.0001).

Of the women attending because of referral, 117 (54.2%) were not accompanied by any staff from the referring facility; 82 (38.0%) were accompanied by midwives, 15 (6.9%) by ward assistants, one (0.5%) by a nursing officer and one (0.5%) by a doctor. All but one of the women who travelled by ambulance were accompanied by staff from the referring facility; one woman, referred on account of failure to progress, came only with the ambulance driver. Of those who travelled by taxis 95 (74.2%) came without any accompanying staff as did nine (60%) of those who came by family or friends' cars and 12 (85.7%) of those who travelled by tro-tros. The indications for referral among unaccompanied women are shown in Table 2.

Table 2.  Indications for referral in women not accompanied by staff from referring facility Thumbnail image of

Information provided by referring facility

Almost all women who were antenatal clinic attendants had their attendance cards with them. Three items relating to time (time woman reported to the facility, T1, time decision was taken to refer, T2, and time woman actually left for Korle Bu Teaching Hospital, T3) were noted. T1 was indicated in 263 (66.4%) of referrals, T2 in 52 (13.1%) and T3 in 46 (11.6%). As regards the number of ‘time’ items indicated in each referral, 108 (27.3%) referrals had none, 220 (55.6%) had one, 64 (16.2%) had two and only four (1.0%) had all three. The duration of stay in the facility before the decision to refer was taken could be calculated for only 35 (8.8%) women; the mean was 7.7 h (SD = 6.9 h). The interval between the time the decision to refer was taken and the time when the woman actually left, could be determined for only seven women. The time taken to arrive at Korle Bu could be determined for 46 (11.6%) women; the mean was 77.7 min (SD = 66.3 min).

Sixty-four women came with partographs; this was 16.7% of the 384 women referred in labour or the immediate postpartum period. Fifty-seven came from polyclinics, six from government hospitals and one from a private maternity home. None came from private hospitals or the clinics outside Accra. Women referred from state-owned centres were significantly more likely to have partographs than those from privately owned centres (χ2= 44.73, < 0.0001).

Intravenous infusion during transfer

Forty women arrived in Korle Bu with intravenous infusions sited. It is estimated that 223 women required intravenous infusions sited before transfer: these comprise all those referred in the second and third stages of labour and in the immediate postpartum period, and those in the first stage with complications such as failure to progress, hypertensive disease, intrapartum haemorrhage, fetal distress, cord prolapse and chorioamnionitis. Thus only 17.9% of those expected to have intravenous lines in situ actually had them. Among 54 women referred with complications associated with bleeding or the potential to bleed during transfer (i.e. ante-/intrapartum haemorrhage, PPH, retained placenta, vulval haematoma), 30 (55.6%) were without intravenous infusions in situ on arrival.

Some pertinent findings on arrival

The patient's general condition was assessed on arrival to be good in 356 (89.9%) women, fair in 35 (8.8%) and poor in five (1.3%). Good condition described women who were not perceived to be in a life-threatening condition, fair condition described those where there was no immediate threat to life but intervention was necessary to prevent them falling into that state, and poor condition referred to those in whom there was an immediate threat to life and urgent measures needed to remove that threat. The main indications for referral in those in fair condition were retained placenta/PPH in 11 women, hypertensive disease in nine, antepartum haemorrhage in five and failure to progress in four. Four of the women in poor condition had retained placenta with PPH while the fifth had PPH only.

Of the 360 women who arrived before delivery, fetal presentation was vertex in 334 (92.8%), breech in 23 (6.4%), shoulder in two (0.6%) and face in one (0.3%). No referral on account of malpresentation was found to have a fetus in vertex presentation; five additional cases of breech and one of shoulder presentations were diagnosed on arrival in Korle Bu.

Fetal heart activity was confirmed in one of the cases in which it was thought to be absent. However, in 18 additional cases the fetal heart activity could not be detected, suggesting that the fetuses were either dead before referral or died between the time of referral and time of arrival in Korle Bu, where fetal heart activity was checked using the Pinard stethoscope. If it could not be detected ultrasonography was used to confirm or exclude fetal death.

Thirty-two (37.6%) of those referred on account of failure to progress in the first stage and one of those referred because of delayed second stage, had intact membranes. The fetuses were in vertex presentation with the head no more than three-fifths palpable; uterine contractions were less than three in 10 min, and the cervix was at least 3 cm dilated.

Seventy women had blood pressures of 140/100 mmHg or higher at the initial Korle Bu examination, the highest being 220/120 mmHg. Twenty-one (30%) had been referred with indications other than hypertensive disease (i.e. blood pressure 140/90 mmHg or higher, with or without proteinuria or convulsive seizure). Seven women had blood pressures of 80/50 mmHg or below (two unrecordable); they had all been referred on account of retained placenta with/or PPH.


On the basis of 12 000 annual deliveries at the Korle Bu Teaching Hospital, the average number of deliveries expected during the study period was 1385. The actual number of 2041 deliveries, an increase of 47.4%, gives an indication of the marked increase in workload in the obstetric unit of the hospital during that time of the year. Failure to progress in labour being the commonest indication for referral is consistent with what has been reported from both developed countries and the West African sub-region, although the 21.5% in this series is much lower than the 48.2–52.6% reported (Rooks et al. 1989; David et al. 1999; Diarra Nama et al. 1999). The second commonest indication in this study, hypertensive disease, hardly occurs in the indications in most developed countries. This may be a reflection of early recognition during the antenatal period with referral to hospital before the onset of labour in those countries. In fact in one study the commonest reason for transfer to hospital care in the antenatal period was hypertensive disease in pregnancy (MacVicar et al. 1993). Third stage and immediate postpartum complications forming 8.8% of referrals falls within the range of 3.0 to 16.6% reported in the series mentioned above (Rooks et al. 1989; MacVicar et al. 1993; David et al. 1999; Diarra Nama et al. 1999).

It is disturbing that more than 70% of the women referred travelled in vehicles other than ambulances and that more than half were not accompanied by any staff from the referring facility. Although it may be acknowledged that some of the indications (for example antenatal clinic poor-/nonattendance, power shortage, previous retained placenta) did not necessarily require ambulance transport, in general a woman referred implies the presence of a factor or factors that pose a threat to her life and/or to that of her unborn baby. It is therefore imperative that she travel in safe and comfortable conditions, with continuous care by qualified personnel. The fact that 30% of those with hypertension on arrival in Korle Bu had been referred with indications unrelated to hypertension indicates the need for close monitoring of all patients in transit so that appropriate therapeutic measures may be taken where necessary. It is unacceptable that women with complications such as failure to progress, hypertension, antepartum/intrapartum haemorrhage, cord prolapse, retained twin, fetal distress and retained placenta should travel without any accompanying staff. The potential dangers inherent in the situation are enormous. Obviously most of the private facilities do not have ambulances for transporting patients; the state-owned facilities do not have enough of them either. It is also a matter of great concern that most of those expected to have intravenous infusions in situ did not have them, particularly those referred on account of indications related to vaginal bleeding.

It appears that the mechanisms of referral and transport to hospital are such that an untoward event is waiting to happen. Urgent action needs to be taken to provide appropriate forms of transportation for peripartum referrals. Every referral must be accompanied by a trained midwife during transfer. While acknowledging that the staff at the referring facility should be allowed to exercise discretion over who should or should not have an intravenous line sited before transfer, guidelines need to be given. In patients with haemorrhage, failure to progress, fetal distress or hypertensive disease, intravenous infusion must be started before transfer.

This study did not look in detail at the appropriateness or otherwise of each referral. It is obvious, however, that some of the referrals may have been unnecessary or could have been avoided. For example, in 37.6% of those referred on account of failure to progress in the first stage of labour uterine contractions were inadequate while the membranes were intact. Timely artificial rupture of membranes in these situations may have been sufficient to augment labour and thereby lead to uneventful vaginal deliveries, thus avoiding the referrals. While considering ways of improving patient transport, obstetric care, particularly labour management practices in the referring facilities must also be evaluated. The workload in the teaching hospital might thus be reduced.

Most referring facilities provided scanty information regarding the ‘time’ items. It is therefore impossible to make any meaningful estimations on the interval between when the decision to refer was taken and when the patient left the facility. This would have given a good indication of the difficulties encountered in obtaining transport. The mean interval between the patient leaving the referring facility and arriving in Korle Bu was almost 78 min and is rather protracted when compared to one series in which only 13.7% of patients were in transit for more than 15 min (Rooks et al. 1989). This may be because most did not travel in ambulances. Some women may have spent some time standing by the roadside looking for public transport.

The very low percentage of referrals with partographs may result either because partographs for monitoring the progress of labour were not used in the referring facility or because the staff did not consider it necessary to send it along with the patient. In spite of the fact that most doctors and midwives in Ghana are familiar with the partograph, it cannot be assumed that every one of them uses it. A study in Enugu, Nigeria, revealed that although about 94% of a group of doctors and midwives thought that the partograph was useful, only 25% of them used it routinely (Umezulike et al. 1998). It would be helpful to determine what the situation is in Accra.

Fortunately, most of the women were referred early, rather than when it would have been too late; this is evidenced by the fact that almost 90% of the patients were in good general condition on arrival in Korle Bu. Only 1.3% were in poor condition, which is very low compared to 24% reported from another West African country (Samai & Sengeh 1997). It is significant to note that 16 (53.3%) of 30 women referred on account of retained placenta/PPH were described as being in fair or poor condition on arrival. The fact that most maternal deaths in developing countries occur in the postpartum period, with 45% of postpartum deaths occurring in the first 24 h, and haemorrhage and complications of pregnancy-induced hypertension being among the commonest causes (Li et al. 1996), should heighten the concern about how these patients are managed.

In Ghana patients pay for health care directly at the point of delivery. Costs in the teaching hospital are higher than in other state-owned centres and it is possible that some referrals defaulted because, among other things, they considered the fees unaffordable. It is not possible from this study to determine the numbers in this category. However, this is unlikely to include very ill patients since in such circumstances the perception of immediate threat to life was likely to override concerns over costs.

In conclusion, most peripartum referrals to Korle Bu Teaching hospital come from polyclinics and private midwives, and the commonest indications for referral are failure to progress and hypertensive disease. There are significant deficiencies in the mechanisms of referral and transport of peripartum patients to the Teaching Hospital: the information provided is less than adequate, intravenous infusions are not set up prior to transfer in most cases requiring them, and most referred patients travel by public transport and are not accompanied by any staff during transfer. Some of the referrals could probably have been avoided. Efforts must be made to improve patient transport, and an evaluation of labour management practices in the referring centres is necessary to improve upon appropriateness of referrals.


I would like to express my sincere gratitude to the staff of the labour wards and the antenatal/postnatal wards of Korle Bu Teaching Hospital for their invaluable help in retrieving patients' notes and preparing the patients for interview.