The observed facility delivery prevalence of 85% is comparable to the estimate of 81% in 2010 provided by the District Public Health Office of Kaski , but much higher than the 35% facility delivery reported by the National Demographic Health Survey . A similar rate of facility delivery (81%) was obtained in previous studies conducted in the urban Kathmandu valley  and Pokhara city . Moreover, health system surveillance data revealed an average 43% facility delivery rate with a wide range (7% to 93%) across districts . Such geographical differences in utilisation should be taken into account when planning for service expansion and allocation of resources by the government.
Women’s education level, distance to the nearest health facility and frequency of antenatal care visits were found to be significantly associated with the likelihood of facility delivery. Our results are consistent with the literature [9, 22]. Previous studies in Nepal had also shown that a distance of more than one hour to the facility could exert a negative impact on delivery service use , while mothers with no education were more likely to deliver at home .
Birth preparedness was high in the study district and almost all women made at least one antenatal visits. Health workers at the birth centres and hospitals counselled women about preparation activities and danger signs of pregnancy and delivery. Female community health volunteers provided information to pregnant women in the communities and encouraged them to use the nearest health facility for delivery .
Apparently, demographic factors such as age, parity, household wealth and caste became non-significant in the stepwise regression model, even though they appeared to be plausible predictors of facility delivery in the univariate analysis as well as previous studies in Nepal [15, 16] or other countries [24–29]. Firstly, these factors are highly correlated with the women’s education level. Secondly, the discrepancy in results might be due to the introduction of incentive scheme and the provision of free delivery care after implementation of the safer mother programme. In the past few years, maternity services have reached out to different social strata with regards to age, wealth and caste. Similar effects have been reported by studies in Burkina Faso  and Ghana  after the reduction of user fees. Indeed, data from the District Public Health Office of Kaski showed that facility utilisation increased from 49% in 2008 to 72% in 2009 and 81% in 2010. Increases in facility delivery have also been observed in other districts of Nepal .
Because of the hilly landscape, poor or non-existent roads and the absence of systematic transport in rural areas, distance remains as the major obstacle to use a health facility. Even in urban areas, the distance and availability of transport affect the timely utilisation of delivery services. The main reason behind unplanned home birth was quick precipitation of labour, followed by lack of transport or nearby facility, similar to other studies in Nepal [16, 21] and India .
There are several options to be considered to further increase the rate of facility delivery. One possibility is to establish more birth centres. However, building new or upgrading existing facilities that can provide comprehensive emergency obstetric care within easy reach of every rural woman is unlikely to be a feasible solution for Nepal in the foreseeable future. An arrangement for rapid transportation in rural areas, which help women to reach a rural birth centre or a nearby hospital quickly, also poses as a challenging logistic problem due to the difficult terrain and resource constraints.
Maternity waiting homes at or near the health facility may offer a viable option particularly for women residing in remote areas, whereby they could arrive early before the due date and wait for their delivery. Maternity waiting homes have been used in other countries but they generally differ in structure and provision of services, resulting in varying degrees of success . Although maternity waiting homes had been constructed and introduced in rural west Nepal, they were not utilised effectively by pregnant women, possibly due to unawareness of their availability and/or lack of adequate facilities . Improving and increasing maternity waiting homes may be an acceptable and affordable way to enhance the facility delivery rate and should be further investigated.
Several issues should be considered when interpreting the findings. The present study was conducted on mostly literate, young, nulliparous women, in a district of Nepal which has a relatively high adult literacy rate of 66.8% and ranks third in terms of the human development index amongst the 75 districts in Nepal . This might limit the generalizability of our findings to the whole country. Selection bias could not be ruled out because all participants were voluntary. Our recruitment process, nevertheless, ensured that they were representative of the pregnant women population in the entire catchment region of Kaski district. Moreover, to improve the accuracy of their responses, face-to-face interviews were conducted by female data enumerators who were residents of the selected areas and thus aware of the local context and issues.