Obstructed labour i.e. failure of descent of the fetal presenting part in spite of adequate uterine contractions [1, 2] is a major cause of maternal morbidity and mortality in low income countries and accounts for approximately 8% of maternal deaths globally [3]. Studies in low-income countries have reported the prevalence of obstructed labour to be between 2% to 8% of all institutional deliveries [2, 4, 5]. This is most likely an underestimate as the majority of maternal deaths due to obstructed labour as a primary cause of death are rarely documented, but instead document the terminal cause of death, hence are classified as sepsis, ruptured uterus or haemorrhage rather than the underlying cause [6]. The main obstetric causes of obstructed labour in low income countries include cephalopelvic disproportion, malposition and malpresentation.

According to Thaddeus and Maine [7] who presented the "Three Delays Model", the chain of factors affecting the outcome of obstructed labour in low-income settings include both cultural and socio-economic factors. Adherence to traditional childbirth practices and individual beliefs as well as poverty restricting the family's ability to pay for transport is directly related to delay in seeking skilled care at birth (Delay I) but also in reaching the facility once the decision has been taken (Delay II). The association between being poor and limited use of health care services is well established although the direction of this association may differ from context to context [8, 9].Factors pertaining to the health care system interact in a complex manner, where one set of factors influence the others. The resources and skills may, for example, be available but for various reasons not offered in time (Delay III) consequently discouraging families from seeking care another time due to perceived low quality of care (Delay I) [7].

In most sub-Saharan countries including Uganda, women are traditionally expected to give birth at home [10] and consequently delay their health care seeking in childbirth, even if complications arise. Moreover, women are often marginalized in decision making regarding where and when to seek care [11]. Unofficial financial demands from health workers prevent women of badly needed maternal health services [10, 12]. Inadequately developed health care systems including poor infrastructure, poor transportation and poor obstetric services are also major contributors to obstructed labour [4, 13].

In a community study conducted in Gulu district in the northern part of Uganda, Orach [14] reported that obstructed labour was responsible for 26% of maternal deaths. However a national assessment of Uganda's obstetric care services in 2006 concluded that obstructed labour was responsible for 22% of all maternal deaths due to direct causes [15]. Through implementation of programmes such as the Sector Wide Approaches, strengthening family planning programs and increasing access to emergency obstetric care, Uganda has only managed to reduce maternal mortality from 557 maternal deaths per 100,000 live births in 1995 to 435 maternal deaths per 100,000 live births in 2006 [16, 17]. It is therefore highly unlikely that Uganda will reduce maternal mortality to achieve the Millennium Development Goal (MDG) target of 131 maternal deaths per 100,000 live births by 2015 [18].

In order to increase the knowledge base for a more successful intervention against obstructed labour in low income settings, it is crucial to gain deeper insights in the roles of socio-economic factors and the health care system in the causal chain of the mentioned outcome. The aim of this study was to investigate the role of individual socio-demographic factors and factors related to the performance of the health system regarding the risk of obstructed labour and its adverse outcomes in a in a low-income sub-Saharan setting like south-western Uganda.

There were 1170 cases of obstructed labour in a sample of 11,180 records analysed, amounting to a prevalence of obstructed labour of 10.5%. Ibanda and Mbarara Regional Referral hospitals registered the highest prevalence of obstructed labour at 12.5% and 12.4% respectively. The causes of obstructed labour were cephalopelvic disproportion (63.7%), malpresentation (19.8%) and malposition (16.2%). Three women (0.3%) had babies with hydrocephalus. Caesarean section delivery was performed on 91.1% of the women with obstructed labour. The other methods of delivery included laparotomy (6.3%), vacuum extraction (0.8%), symphysiotomy (0.5%), forceps (0.4%) and craniotomy (0.2%).

The fetal outcome was known for 10,127 (90.6%) of the 11,180 women records analysed. The perinatal mortality rate was 141 per 1000 total births (165/1167) for women who had obstructed labour and 65 per 1000 total births (586/8960) among women in the non-obstructed group (OR 2.31: 1.92-2.78). Overall the perinatal mortality rate was 74 per 1000 total births.

The risk of maternal complications was 10.8% (126/1170) in women with obstructed labour compared to 1.4% (145/10,010) in women without obstructed labour. Thus the odds of complications were 8 times higher (95% CI of odds ratio: 6.32-10.45) among women with obstructed labour compared to those without.

Maternal complications observed among women with obstructed labour were ruptured uterus (7.1%), puerperal sepsis (3.4%), bladder injury (1.8%), postpartum haemorrhage (1.2%), fistulae (1.4%) and one case of disseminated intravascular coagulation. While in the non-obstructed group the main complications were postpartum haemorrhage (0.5%), retained placenta (0.4%) and puerperal sepsis (0.1%). Thirteen women (1.2%) died as a result of obstructed labour while 38 women (0.4%) died in the non-obstructed labour group.

The odds of perinatal death were about five and half times higher among women whose main occupation was of subsistence farming compared to women with salaried jobs or traders. Similarly housewives had 4 times higher odds of perinatal death compared to women with salaried income (see Table 6). However, the occupation variable could not be included in the multivariate model due to substantial missing data.

Partographs were found attached to 45.8% (536) of maternal records of women with obstructed labour. However only 19 (3.5%) partographs were satisfyingly documented; 13 in Comboni hospital, 4 in Mbarara regional referral hospital, 1 in Kitagata hospital and 1 in Rushere community hospital.

We determined the prevalence of obstructed labour in our sample to be 10.5%. Our findings showed that obstructed labour was associated with individual factors such as age, parity and socio-demographic factors such as occupation and place of residence and with particular health facility at which the woman gave birth. The association between low age, lower parity, district of residence and obstructed labour remained statistically significant even after multivariate regression analyses.

The proportion of women with obstructed labour who developed complications was 8 times that of those who did not have the condition. Thirteen women (1.2%) died from obstructed labour, compared to thirty eight (0.4%) in the non-obstructed labour group. We also found that adverse outcomes of obstructed labour, such as perinatal death, were associated with older age, high parity, occupation, district of residence and particular health facility. The association between old age, other districts and high risk of perinatal death remained strong after multivariate analysis.

The observed overall prevalence of obstructed labour of 10.5% for the six hospitals is high but, Mugerwa in an unpublished study reported a prevalence of 13% for Mbarara regional referral hospital for the years 1993 - 1995 while in our study the prevalence for the same hospital was 12.4%. To our knowledge, there are no previously published studies on the magnitude of obstructed labour in Uganda. However, studies conducted elsewhere in Africa have reported prevalence rates of between 0.9% - 7% [2, 5, 21, 22].

The main obstetric causes of obstructed labour in our study were cephalopelvic disproportion, malpresentation and malposition which have also been reported by other authors elsewhere [2, 4, 21, 22]. Cephalopelvic disproportion is mostly due to contracted pelvis and adolescent or early marriages. In south-western region of Uganda contracted pelvis could be attributed to malnutrition during childhood. Though rich in food crops, this region has consistently registered the highest levels of stunting among children in national demographic surveys conducted between 1995 and 2006 [23].

Women who are delivering for the first or second time may be young and as such be at a higher risk of obstructed labour than those who have delivered more times. Grand multiparous women may also be at a higher risk of obstructed labour from malposition and malpresentation. The finding that women whose district of residence was Isingiro had the highest risk for obstructed labour can be attributed to lack of a hospital which offers comprehensive obstetric services. Women diagnosed with obstructed labour therefore have to be transported on poor roads to Mbarara where they can access appropriate care. On the other hand, the low risk of obstructed labour for women resident in Mbarara and Ibanda districts found in our study could be related to the general good health services and transport infrastructure in the two districts. It should also be noted also that the two hospitals were the ones with resident Obstetricians.

Moreover, we found that adverse outcome of obstructed labour such as perinatal death was strongly associated with increasing parity, residence in other districts outside the study area. Women who have ever delivered may opt to deliver at home in subsequent pregnancies thereby exposing themselves to complications and a similar trend was reported in the Uganda demographic health survey of 2006 [23] where delivery by women in health facilities progressively decreased from 58% in nulliparous to 31% in those with parity 6 and above. In the same survey the south-western region was found to have only 31% of women delivering under the care of skilled birth attendants.

Our study showed that women whose occupation was subsistence farming or housewife were at a higher risk of having perinatal death as an adverse outcome compared to salaried women. Resource poor women may not be able to raise the money required in the event of an obstetric emergency and therefore face delay I and II as described by Thaddeus and Maine [7].The association between low socio-economic status and obstructed labour has also been reported by other authors [24–26]. Obstructed labour, if not well managed in the early stages, leads to severe or debilitating complications and sometimes death [27]. In our sample the main maternal complications or adverse outcomes observed were ruptured uterus, puerperal sepsis, bladder injury, postpartum haemorrhage, fistulae and are similar to what has been reported by other authors [2, 4, 5, 22].

The partograph, which is a low cost tool for monitoring labour, has been shown to reduce the occurrence of prolonged/obstructed labours [28]. The finding in our study that less than a half of all maternal records diagnosed with obstructed had partographs attached and that only 3.5% of these were completed may be an indication of poor monitoring of labour and child birth in the hospitals studied. A study on assessment of partograph use during labour conducted in Rukungiri district (neighbouring Bushenyi district) reported a 70% use but with only 2% of the partographs fulfilled the standard monitoring of fetal heart rate [29]. Studies conducted elsewhere in Africa have reported poor partograph utilization in lower health facilities [30, 31] and a high utilization rate in a tertiary hospital [32].

Our study findings point to the effect of socio-economic factors and the health system structure in the causal chain of obstructed labour. Since 2000 Uganda government through the Health Sector Strategic Plans [19] has been in a process of increasing availability and accessibility of emergency obstetric services in the whole country. One of the key interventions was upgrading of health centre level IV to offer comprehensive emergency obstetric services. It was hoped that since health centre IV are in easy reach of most of the population they would help to manage most obstetric emergencies. A study in 2003 by Mbonye et al. [15] indicated that only 13% and 9% of health centre IV could offer blood transfusion and caesarean section services respectively. Although there are no published studies on the current availability of emergency obstetric services, anecdotal information indicates that most health centre IVs in the country are not able to offer a full range of emergency obstetric services. This is evidenced by the MDG report, which shows Uganda's poor performance towards attaining the target of reducing maternal mortality by the year 2015 [18].

Our study provides baseline information on the individual socio-economic and health system factors associated with obstructed labour and can be used by policy makers and implementers to improve on management of obstructed labour in south-western Uganda.

Socio-demographic and health system factors are strongly associated with obstructed labour and its outcome in south-western Uganda. Our study provides baseline information on the individual socio-economic and health system factors associated with obstructed labour. The findings can be used by policy makers and implementers to improve on management of obstructed labour in the country as a whole. In order to address the problem of obstructed labour, interventions are needed both within the community and the health system level. Further research in the management of obstructed labour within health facilities is advocated for and it also pertinent to study the community's perception of prolonged/obstructed labour and actions taken when it occurs.

JK is a PhD student at Lund University, Sweden and a Lecturer of Public Health at Mbarara University of Science and Technology (MUST), Uganda. KOP is a Researcher/Lecturer of Social Medicine & Global Health at Lund University, Sweden.

P-OO is a Professor of Social Medicine at Lund University, Sweden. ET is Lecturer of Biostatistics/Epidemiology at MUST. PM is a Senior Medical Associate with EngenderHealth/Fistula Care Project, Kampala, Uganda and at the time of conducting the study, PM was a Lecturer of Obstetrics & Gynaecology at MUST.