This survey documents self-reported health determinants in the Tibetan population of Surmang --poverty, low female education and literacy, lack of sanitation, transportation and infrastructure and lack of access to basic maternal, neonatal and child health care - which are often associated with poor health. Women everywhere are entitled to skilled care at delivery as a universal human right, and no woman should die giving life. The majority of women in Surmang were very worried about dying in childbirth, and the report of 3 recent maternal deaths in this small population suggests exceedingly high maternal mortality in this region.
Facility delivery rates <1% in Surmang in the 5 years between 1999 and 2004 can be compared with facility delivery rates of 92% for China overall , 98.5% facility delivery for eastern provinces , 66.6% for western regions  and 34% for Qinghai Province . Strong correlation between facility delivery rates and maternal mortality ratio (MMR) has been demonstrated throughout China, with highest official figures being MMR of 467 (TAR 2000) corresponding to 20% facility delivery . This survey was not intended to and is not powered to adequately estimate MMR in the Surmang area. However, the facility delivery rate of <1% reliably demonstrated in the current survey would likely produce a substantially higher MMR than the highest official MMR of 467 associated with 20% facility delivery in TAR. While the results of this survey cannot accurately estimate MMR in the Surmang subpopulation, they do suggest very high maternal mortality by global standards. Skilled care and cesarean delivery (both <1%) were also virtually inaccessible, which is unacceptable by global and Chinese standards. Population cesarean section rates less than 1% are associated with maternal death, as well as excess intrapartum fetal loss and neonatal morbidity and mortality .
The three classic delays in obstetric care which lead to maternal mortality are discussed below as they apply to the Surmang setting. As of 2004, the three delays in Surmang were not delays but nearly complete barriers to care, even before the recent earthquake destroyed the emergency obstetric referral system.
Barriers to Seeking Care
Female education and literacy rates well below Qinghai and national standards limit illness recognition and demand for health and maternity care. Cultural concepts of a safe delivery, such as avoiding contact with harmful spirits by staying home  and delivering in unclean areas , can only be countered by education. Improved education for girls and women would likely increase female empowerment, improve health care utilization and delay marriage and childbearing.
Despite this avoidance of facility delivery, over half of Surmang women reported seeking preventive care during their last pregnancies, about a third from a western medicine doctor. The vast majority said they would go for antenatal care if it were available in their village; cost and convenient access were limiting factors. Improving availability and affordability of these services is an appropriate next step toward improving uptake of facility delivery. In contrast to preventive antenatal care, the women overwhelmingly preferred Tibetan medicine to western medicine for problems or illness during pregnancy, possibly because of easier access at the village level. Incorporating local Tibetan medicine providers into the health system may improve utilization of services and encourage appropriate referral.
Barriers to Reaching Health Facility
Geographic factors and distance are strong predictors of health care utilization, and lack of transport is a common preventable cause of death for both mothers and newborns [14, 15]. The geographic isolation and rugged terrain, lack of reliable transportation for mothers in labor (most often by horse or motorcycle), poor road infrastructure and challenging weather of the Tibetan Highlands represent significant barriers to skilled delivery care in this population. None of the mothers in Surmang met WHO's maternal health access criterion of one hour to EmOC , and access has worsened since the earthquake. However, facility delivery rates were no better in another survey of Qinghai Tibetans where physical access to facility was much easier , potentially indicating cultural, political, social and economic factors as barriers to care in the population, as well as lack of physical access to delivery facilities.
Barriers to Adequate Care at Facility
Since <1% of Surmang women delivered in facility, the third delay of inadequate care at facility could not be evaluated by this report. The two government hospitals providing EmOC in Yushu were completely destroyed in the earthquake and have now been reestablished in temporary facilities. Outside Yushu town, some primary level health village clinics provide skilled care, but very few women access this care. In the NGO-run Surmang Clinic some increase in facility delivery has been observed in recent years .
The self-reported prevalence of symptoms commonly associated with maternal morbidity is high in this population, highlighting the need for skilled delivery care. Almost half of mothers subjectively experienced prolonged and possibly obstructed labor, while 17% felt they had excessive bleeding during delivery. Medical confirmation of these complications was not possible, and some problems may be over-reported due to varying definitions, poor recall and language challenges. For instance, the self-reported prevalence of seizures of 27% was clearly excessive, most likely reflecting misinterpretation of the term. However, almost half of women had antenatal complaints of dizziness/blurry vision and headache and over a quarter complained of swelling, all of which is concerning for undiagnosed and untreated preeclampsia, warranting further study.
Postpartum complaints were also frequent -- 56% reporting severe lower abdominal pain, 42% fever, 41% possible mastitis, 29% excessive bleeding, 27% foul discharge and 24% dysuria -- but only 38% of women with postpartum complaints sought care. These symptoms could indicate high rates of infectious complications, but less than 3% of women with self-assessed postpartum problems obtained western medical treatment, again highlighting the need for skilled care in the postpartum period as well. Again, the women strongly preferred local Tibetan medicine for these problems possibly because traveling out of the home soon after birth is proscribed in Tibetan culture, further limiting access to western medical treatment.
The poor sanitation-hygiene and lack of access to clean water may be contributing to the high rates of childhood diarrhea reported in the population. Lack of access to sufficient water is probably the main reason 95% of mothers did not bathe in the first week postpartum, and this could also contribute to maternal infectious morbidity. A clean instrument was used to cut the umbilical cord in only 6% of deliveries. In this setting, clean delivery is a priority, low-cost intervention that may be achieved with clean delivery kits and behavior change messages implemented by health workers and/or family members. The Surmang tetanus vaccination rate less than 14% requires urgent action. This is a low-cost highly effective intervention for reducing both neonatal  and maternal tetanus and may have substantial protective benefit for this population where women often deliver on yak dung or a dirt floor and choose the animal shed or another unclean part of the dwelling for delivery [6, 17].
Although breastfeeding rates and duration are high in all Tibetan population surveys, Surmang women reported delays in initiation of breastfeeding and high rates of non-exclusive breastfeeding, which have been associated with higher infant morbidity and mortality . Age of introduction of complementary feeding in Surmang was significantly delayed compared with other surveys [17, 21], and all available studies of the rural Tibetan population indicate complementary food is low in protein and vegetable variety compared with Chinese urban infants . These nutritional deficits and late weaning could contribute to the high prevalence of childhood stunting in the Tibetan Plateau population , which can, in turn, lead to obstructed labor .
The devastation of the recent earthquake provides an opportunity for both government and local NGOs to reevaluate the existing health system and consider ways of improving delivery of maternal and child health services in this marginalized Tibetan population. In addition to strengthening health facilities, innovative strategies need to be considered to improve demand for skilled delivery care and to link the poorest mothers and newborns with the health system. In Yushu, conditional cash transfers were instituted to pay mothers 300 yuan per hospital delivery; follow up evaluation of the impact of this program is needed. Community mobilization, and female empowerment via women's participatory groups can increase skilled care at delivery by as much as 50% as reported in a recent meta-analysis . Other effective strategies for improving uptake of skilled delivery care include community based loans, referral and transportation schemes or radio/cell phone communication systems, which can be used in remote low-resource settings for triage and coordination of transport .
In this setting, decades may be required to achieve universal facility access, and there may be a role for parallel community-level interventions alongside health facility strengthening. Community health workers (CHWs) can assist with tetanus and childhood vaccination, maternal nutrition, antenatal care, identification of danger signs during pregnancy and delivery, clean delivery, adequate care of the umbilical cord, timely initiation of breastfeeding and improved introduction of complementary feeding. CHW programs have been effective in reducing neonatal mortality from "birth asphyxia" and infections [25–27]. A package of services provided by CHWs can reduce early neonatal mortality by 36% . CHWs can also be trained to successfully assess and treat infant and childhood illnesses such as pneumonia, further reducing child mortality . For prevention of maternal mortality, evidence suggests that local provision of medications for postpartum hemorrhage and sepsis by CHWs can reduce maternal morbidity and mortality over and above health facility strengthening alone [30–32]. While community level interventions are controversial for maternal health, they need to be considered and evaluated as a bridging mechanism in populations such as Surmang that are not being reached by the health system.
The small sample size was not powered for estimating neonatal, child, or maternal mortality rates. Though care was taken to prepare and execute the survey using local language, there was misclassification of some birth outcomes (fetal deaths, stillbirths, neonatal deaths and miscarriage) that may have reflected translation issues, and thus it was difficult to interpret pregnancy outcome data. Furthermore, all data was based on self-report, which is subject to reporting and recall bias. Though care was taken to use the most appropriate local terminology for medical symptoms, understanding may have been limited by low education and varying definitions. No information was obtained on contraceptive prevalence and practices, one of the pillars of safe motherhood. Water quality and quantity was not directly assessed but likely contributes to the poor hygiene and infectious morbidity documented by the survey.