Our results represent an investigation of maternal mortality and maternal near-miss based on the prospective data collection in both community and local hospitals in the Lao PDR. The presence of organ system dysfunction criteria recommended by the WHO (respiratory, cardiovascular, renal, hepatic, neurologic dysfunctions) were used for the identification of near-miss cases. We found MNM ratio of 9.8 (95% CI: 4.9 -17.5) per 1,000 live births, MMR of 179 (95% CI: 50–650) per 100,000 live birth, both of which appear to be surprisingly low.
For the MMR our finding was quite comparable to the recent study conducted in the North-Western Border of Thailand where the MMR was 184 (95% CI:150–230) per 100,000 live births . The other comparable result of MMR of 158 per 1,000 live births was reported in the study of maternal mortality estimation at the sub-national level in Bangladesh . Both studies were done in Asian countries similar to Lao PDR.
However, our finding was much lower than those reported in previous studies conducted in other developing countries in other regions. One population based survey of maternal mortality in West Africa found MMR of 311 per 1,000 live births . Another retrospective hospital based study found the MMR of 432 per 1,000 live births in a rural hospital in Sudan . The present study was population based, with prospective data collection in both the community and local hospitals. Retrospective data collection in a central or general hospital (tertiary care) in the previous studies would give higher numbers of maternal death, because of differences in the study population; mothers who gave birth in those hospitals came from unlimited catchment areas and higher proportion of complicated cases [16, 17].
To some extent, our findings can be explained in the context of the present conditions of MCH and other health services in Sayaboury Province, where the environment was different from the previous studies. Altogether, 629 village health volunteers and 573 traditional birth attendants have been trained in maternal and child care for all villages, and they are regularly supervised and attend refresher training . In addition, the health system has been receiving both technical and financial support from international agencies since 1984for the MCH services at all levels. Our findings are consistent with the trend in MMR of Lao PDR reported from 1990 to 2010 by WHO, et al. . The estimates MMR per 100,000 live births were 1,600 in 1990 and 470 in 2010 with a 70% reduction between the two years. The trend is progress towards improving maternal health according to the MDG 5 .
Although there were difficulties in transferring patients to the provincial hospital during the wet season, the geography of the province is such that all its districts share a border with Thailand and some with Vientiane Province. When an emergency occurs, there were options for seeking care in a neighboring country or in Vientiane Capital City. The above reasons may account for the low maternal mortality found in our study compared to those suggested by the national statistics [4, 5] and previous studies in developing countries [11, 16, 17].
The present study also found a maternal near-miss ratio of only 9.8 per1, 000 live births which was much lower than those in previous studies. Souza et al. reported the MNM ratio of 34.3 per 1,000 live births in the WHO global survey on maternal and perinatal health done in 120 of eight Latin American countries . Almerie et al. reviewed cases of maternal near- miss and maternal mortality in a maternity university hospital, Damascus, in Syria  and reported the MNM ratio of 32.9 per 1,000 live births. A prospective study in Ghana showed the MNM ratio of 28.6 per 1000 live births in a teaching Hospital (KBTH) . Ali et al. reported the MNM ratio of 21.1 per 1,000 live births in a rural hospital in Sudan . The reasons for our lower ratio may be because the previous studies were hospital-based and tend to have higher ratios because of the higher proportion of complicated cases than our community-based study. However our result was quite similar to the report of the WHO Multicounty Survey on Maternal and Newborn Health  which presented the ratio of 8.3 per 1, 000 live births. In addition, it was little higher than that of Jabir et al. reported the ratio of 5.06 per 1,000 live births in Baghdad, Iraq .
The rates of maternal near-miss may differ when using different identification criteria. A one year retrospective chart review study of medical records of 1,163 obstetric hospital admissions of a tertiary maternity hospital in Rio de Janeiro, Brazil was conducted in 2013 . It underlined that different approaches entail different estimates of MNM. The maternal near-miss ratio was 27 per 1,000 live births using WHO maternal near-miss criteria compared with 123 per 1,000 live births of using the Waterston criteria. Furthermore a systematic review of the prevalence of maternal near miss by Tunçalp, et al.  reported the rates of maternal near-miss varied between 0.04% and 4.54% for management-based criteria and between 0.14% and 0.92% for organ-based dysfunction based criteria. The prevalence using different criteria were higher in low-income and middle-income countries of Latin America and Africa [22, 23].
Our finding of maternal near-miss detected may be underestimated. There are several possible reasons for this. Firstly, maternal near-miss was a new marker of health and obstetric care in our setting. The local co-coordinators (VHV, TBA, MCH health center staff) might have difficulties in understanding the concept and process of MNM identification and led to miss some MNM in the communities. Secondly, limitation of the MCH staff and facilities at the district hospitals might be the important factors that did not allow investigators to apply full WHO criteria to detect a maternal near-miss. This was due to a number of factors: (a) the laboratories at the district hospitals had a limited capacity to confirm near-miss cases; (b) the supply of donated blood was inadequate; in the four study hospitals only the provincial hospital had a blood bank, although two district hospitals did have blood stores; and (c) there was a lack of medical equipment and alternative facilities for the care of severely ill patients at the district hospitals. Only the provincial hospital had an intensive care unit. As a consequence, some cases were probably not classified as near-misses because the severity of their condition was unable to be confirmed by the existing facilities.
One significant finding in this study is the very high proportion (7 from 11) of MNMs occurred early in pregnancy, 2 cases related to abortion complication and 5 caused of ectopic pregnancy. Maternal and child health care providers and policy makers should be informed about this important cause of MNM.
They were 9% (8/90 abortions) of all abortion and ectopic pregnancies. This figure was lower than that of 11.1% (61/549 abortions) found in a surveillance network study to identify severe maternal complications associated with abortion in 27 referral obstetrics units across regions of Brazil .
Requirement of the complexity of care could be estimated from the MNM incidence and SMO ratio. The higher values, the greater the extent to which women require high complexity care Our findings show the mortality index (MI) was 15% compared to 8.5%of Cecatti et al. who evaluated maternal near miss using WHO criteria and maternal death related to organ failure in Brazil . van den Akker et al. reported a MI of 12% in their study evaluating maternal near miss using organ-failure based criteria as a tool in Southern Malawi .The mortality index (MI) represents an estimate of performance. It is recommended that an institute with high index (>20%) may have inadequate quality of obstetric care for the severe cases . Although we had small maternal index in our setting, we could not be sure that there is adequate quality of obstetric care for the severe cases in Sayabury Province. There is lack of resources for obstetric care, such as blood bank, operation room, etc. in many district hospitals of the province. The maternal near-miss: mortality ratio (MNM: 1 MD) was 5.5:1.The ratio can provide direct information on problems related to care of severity obstetric complication and could be used as marker of obstetric care .
Strength of the study
This was the first population-based descriptive study involving a follow-up of pregnant women since the date of their last menstrual period until within 42 days after termination of pregnancy in order to identifying the incidence of MNM and MD in both the community and local hospitals in Lao PDR. The WHO clinical criteria of organ systems dysfunction were used for the identification of near-miss cases, allowing potential international comparisons. Its main strengths were the use of trained PHC workers (TBAs and VHVs) and health center staff to regularly screen eligible pregnant women in the villages and carry out the follow-up of complication cases to detect maternal near-misses and record maternal mortality. The early detection of maternal complication cases in the community and their referral to hospitals for prompt treatment were used to the maximum possible extent to reduce the number of maternal deaths. The follow-up model used in our study is expected to have implications, which will be benefit to women in developing countries.
The study was designed to identify near-misses and maternal deaths in both the community and the local hospitals. However, the WHO standard criteria for the identification of maternal near-misses could not be used in the community, TBAs in the village and local coordinator at the health centers might have limited knowledge to confirm MNMs. In addition, the district hospitals had limited resources for assessing organ dysfunctions, such as limited supply of blood and laboratory facilities. These factors might cause an underestimation of near-miss rates. The maternal near-misses could have been underestimated by the application of the WHO definition of a near-miss which relies on good laboratory and management-based criteria. These potential limitations are consistent with those presented in study of Cecatti et al. . van den Akker et al. used WHO MNM approach studied at Malawi  and the study of Nelissen et al. by applying WHO maternal near-miss criteria in low resources setting in rural hospital Tansania .
The future development of maternal health services (MHS) should include an upgrading of the capacities of MCH services in local hospitals and ways of improving these services in local health centers so that local MHS staff and midwives are better able to recognize and respond to the important factors associated with SMOs. This study was, in essence, a pilot project for the detection of maternal near-misses and maternal deaths in Lao PDR. Using the lessons learnt from this study, it is proposed that the Lao Ministry of Health should consider a nation-wide study of maternal near-misses, for the future development of maternal health services and also for further health research on MCH data collection in poorly resourced areas, especially in the remote provinces of Lao PDR.