Pregnancy is a normal physiological process, one of whose distinctive features is the presence of the utero-placental complex. Physiological changes associated with pregnancy have been observed to occur in all major systems of the body. These changes may result in strain to organ systems, in particular for those with limited reserve, often accompanied by a deterioration of any pre-existing medical conditions [1]. A pregnant woman may require ICU admission because of diseases specific to pregnancy, diseases that are worsened by pregnancy, as well as diseases co-incidental to pregnancy [1]. The occurrence of disease-related complications during pregnancy or postpartum may significantly increase the risk of maternal mortality. Furthermore, the assessment and management decisions made during pregnancy must also consider that which might be harmful to the fetus. Owing to the numerous factors affecting medical management of pregnancy, critically ill obstetric patients present a particularly great challenge to ICU physicians.

Worldwide, maternal mortality has decreased from 500,000 to 300,000 deaths/year over the past 30 years [2]. Strikingly, these deaths occurred almost exclusively in developing countries, with only 1% occurring in developed countries [2]. In addition, it is estimated that as much as 70% of these deaths occur in hospitals, mainly in ICUs, rather than outside medical care [3].

A first and vital step towards addressing this problem is the collection and analysis of relevant clinical data that could help shed light the high mortality of these patients. Accordingly, to help improve obstetric care in pregnant/postpartum patients requiring ICU admission in the developing world setting, we retrospectively evaluated clinical data collected from obstetric patients admitted to ICU in our hospital.

After approval by the institutional review board of First Affiliated Hospital of Guangxi Medical University, we retrospectively reviewed the medical records of all obstetric patients admitted to ICU either during pregnancy or within 42 days of delivery between January 1, 2014, and December 31, 2018. For each patient, the following data were collected: demographic data, residence, parity, medical and obstetric history, standard prenatal care (at least five visits for term pregnancies) [4], gestational age at ICU admission or at end of pregnancy, mode of delivery, admission diagnosis, cause of admission obstetric (only occurring in pregnant/postpartum patients) versus non-obstetric (also occurring in nonpregnant patients) [5], source of admission to ICU (emergency department/ward/operating room/other hospital), Acute Physiology and Chronic Health Evaluation (APACHE II) score [6] at admission, ICU management (mechanical ventilation, hemodialysis, central catheter, plasmapheresis, and blood transfusions), and length of stay in the ICU and the hospital. The mortality of all the mothers and fetal-neonatal losses were recorded. Any ICU cases in which multiple-organ dysfunction syndrome (MODS) (dysfunction $\ge$ 2 organs according to Sequential Organ Failure Assessment (SOFA)) developed was recorded [7]. ICU admissions were categorized into two main groups based on their place of residence as either urban or non-urban.

Data were analyzed using SPSS version 20.0 (IBM SPSS Statistics for Windows, version 20.0, Armonk, N.Y., USA). Continuous variables are presented as mean $\pm$ SD or median (interquartile range), according to their distribution. Categorical variables are presented using the composition ratio. Two group comparisons were made using t-tests or chi-squared tests, where appropriate. P-values 0.05 were considered significant.

This is a retrospective study of obstetric patients requiring ICU admission to a tertiary care hospital in a developing country. Critically ill obstetric patients accounted for a small proportion of our ICU admissions. Our data were consistent with other published studies [8, 9] of the two most common obstetric causes of admission (obstetric hemorrhage and hypertensive diseases from pregnancy). Among the underlying causes of obstetric hemorrhage recorded in our study, placental abnormality was the most common. Notably, placenta previa accounted for as much as 86.4% of the recorded placental abnormalities; our observation is consistent with that reported by Leung et al., 2010 [8]. We believe that more attention to the management of this population is needed in order to prevent and better treat obstetric hemorrhage. For example, pre-operative ultrasound should be performed to identify the location of the placenta for the purpose of choosing an appropriate uterine incision. Meanwhile, in order to reduce the bleeding of the placental exfoliation surface, oxytocin or massage of uterus could be applied quickly after the delivery of the fetus to promote the uterine contraction; for those with bleeding who do not respond to oxytocin and uterine massage, intrauterine gauze packing, B-lynch suture and uterine artery embolization are effective methods of treatment.

In our study, cardiac disorder was the most common reason for non-obstetric causes of ICU admission, in contrast with the findings of other studies in which sepsis was reported to be the predominant non-obstetric reason for ICU admission [8, 10, 11]. In Belgium, however, pre-existing and acquired cardiopathies were the most common cause for ICU admission and surpassed hypertensive disorders and hemorrhage [12]. In our hospital, the cardiac critical care unit (CCU) is a relatively independent intensive care unit, and this study did not include critically ill obstetric women in the CCU. If included, the proportion of ICU admission due to cardiac events might be higher. Maternal cardiac load increases significantly with increasing gestational age, especially in late pregnancy and the perinatal period, potentially leading to acute left heart failure, acute exacerbation of chronic heart failure, and even cardiogenic shock, particularly in women of advanced age and those with heart-related diseases. It is necessary that obstetricians and cardiologists work together to evaluate the appropriateness of pregnancy for women with cardiac disease, and to strengthen prenatal and postnatal management in these women. Furthermore, more intensive screening, especially for cardiac conditions, may improve pre- or early pregnancy identification of high-risk women and trigger early referral to more experienced medical facilities [12].

In this study, non-urban patients were sicker upon admission, showed a higher incidence of MODS and required more interventions, compared with urban patients. Although not statistically significant, non-urban patients accounted for 90.9% of maternal mortality. According to a recent study, protective factors for maternal mortality included prenatal care and admission to tertiary hospitals, while risk variables included high APACHE II and MODS [4]. While it is unsurprising that disease severity scores are associated with poor outcome, what is perhaps more relevant is that the detection of severe disease in such cases affords an opportunity to intervene [4].

Appropriate prenatal care has been shown to reduce the maternal death rate [13, 14]. In this study, 108 patients (81.2%) received at least one maternal health checkup. Standard prenatal care was followed in 70.4% of all patients (76/108), by 90.3% (28/31) of patients within the urban group and 62.3% (48/77) of patients within the non-urban group (P = 0.004; denominators inconsistent between comparisons due to missing data). Being cared for in a tertiary center improves access to better medical resources, such as well-trained physicians and advanced medical equipment that are essential to preventing maternal death [15, 16].

According to Zeeman G.G. (2006), maternal outcomes are expected to be adversely affected by a delay of more than 24 hours between onset of disease and ICU admission. The delays may be due to (1) a time-lag in seeking care, (2) the ability to access services or (3) the time required to receive appropriate care once the medical facility has been reached [17, 18]. The first such factor may explain our finding that non-urban patients were younger than urban patients (29.24 $\pm$ 0.57 versus 31.62 $\pm$ 0.87, P = 0.031), and there were more primipara pregnancies among urban patients (77.1% versus 53.1%, P = 0.013). The second cause of delay is related to accessibility barriers, such as healthcare cost, distance and transport [4]. The third possible cause of delay might explain why a higher percentage of non-urban patients had been referred from other less-specialized centers in this study [5].

Non-obstetric causes for admission may be associated with a higher prevalence of comorbidity [19] and greater illness severity [5, 20], which could explain why patients belonging to the non-obstetric group had higher APACHE II scores at admission, experienced longer hospital stays and ICU stays, developed more MODS and suffered higher maternal deaths than did patients in the obstetric group. Moreover, non-obstetric causes of admission were more frequent among non-urban patients, potentially accounting for the worse state of non-urban patients observed in this study. Cesarean section rates are higher among obstetric patients that require admission to ICU during their pregnancy [21]. The rate of cesarean sections of critically ill patients admitted to ICU in our study was 78.4%, in line with most previously reported studies of critically ill obstetric patients [22]. One such study reported that risk factors for requiring cesarean section were prenatal care and presenting with pregnancy-associated hypertension as the reason of admission. This was consistent with our finding that cesarean section was more frequent among urban group patients, who had received more frequent standard prenatal care, than non-urban group patients. For patients with hypertensive disorders, there were also differences between urban and non-urban groups (11/35 versus 14/98, 31.4% versus 14.3%, P = 0.026). Overall maternal mortality in this study was 8.3%. These figures were higher than in studies conducted in other countries [4, 12]. Similar to other studies, non-obstetric causes of death prevailed [11, 23], with eight deaths out of 11 (72.7%) due to non-obstetric reasons. Liver failure was the leading cause of death, followed by cerebral hemorrhage. However, pneumonia was the underlying primary cause of death in another study [11].

The main limitations of this study are its retrospective design, the small number of patients, and possible bias in information available from medical records. We were unable to retrieve birth outcomes including Apgar scores or Neonatal intensive care unit (NICU) admission, and the multiple organ dysfunction (MOD) score. Furthermore, patients admitted to the CCU during the study were not included. As, this is a single-center study, any extrapolation and generalization of the findings of these results should be carried out with caution.

In summary, we speculate that multidisciplinary team involvement is essential in the management of postpartum hemorrhages, hypertensive disorders and cardiac events. In our study, health disparities existed between patients from non-urban areas and urban areas: the former were sicker upon admission, were more frequently transferred from other hospitals, received less standard prenatal care, received more interventions, developed more complications, and their maternal mortality was higher. To help mitigate these disparities, we suggest that greater focus and resources be invested in obstetric healthcare in non-urban areas.

MZ conceived and designed the study; MZ and XG wrote the manuscript; MZ and YW collected the data; XG and YW analysed and interpreted the data; ZT and HW provided critical revisions; and XG, ZT and HW approved the final version of the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

The Medical Ethics Committee of First Affiliated Hospital of Guangxi Medical University had approved the protocol [approval number: 2020(KY-E-021)].

This work was supported by Natural Science Foundation of Guangxi Province [grant number: 2017GXNSFAA198045].

The authors declare no competing interests.