The relationship between severe maternal morbidity and a risk of postpartum readmission among Korean women: a nationwide population-based cohort study

Compared with the maternal mortality ratio, severe maternal morbidity (SMM) is a useful indicator for the evaluation and improvement of maternal health services [1], and it use as such an indicator has increased. SMM is defined as unintended outcomes of the process of labor and delivery that result in significant short-term or long-term consequences to a woman’s health [2]. Although SMM is rare, it is associated with high medical costs, extended length of hospital stay, and long-term rehabilitation [3]. To improve maternal health care, it is necessary to determine a relevant measurement of SMM, find preventable factors for SMM, and monitor the quality of maternal care during postpartum. Unplanned hospital readmission is used as a quality metric in medical and surgical health care [4], and it focuses on quality improvement and the reduction of medical costs [5]. Although the readmission rate is used as a quality indicator, postpartum readmission rates are not used as a quality metric in obstetric care [5]. Postpartum readmission was used as one of the indicators of postpartum maternal morbidity [6], but few researchers suggested that it could be a quality indicator in obstetric care [5].

Previous studies identified the risk factors for SMM including maternal age [7], family income [8], residing in disadvantaged areas [9], employment status [10], comorbidities [11], multiple births [12], cesarean section (CS) delivery [11, 13, 14], and hospital volume [15]. Although several studies identified which factors affected postpartum readmission [5, 16‐18], most of them focused on the relationship between postpartum readmission and the mode of delivery, and CS was especially associate with postpartum readmission [16‐18]. CS has been commonly considered as a safe and low-risk procedure [18] with a high incidence rate. In the United States (US), the rate of CS slightly declined in 2016 (31.9%), but in the recent 7 years, it accounted for more than 32% of all deliveries [19]. In South Korea, CS was the fourth most frequent type of total surgery in 2017 [20]; approximately 45% of women had CS, and it use has been increasing [21]. However, CS was significantly related to the risk of SMM [11, 13, 14], double-fold higher medical costs [22], and a high risk of postpartum readmission [16]. The evidence of the association between SMM and postpartum readmission is almost non-existent. This study aimed to determine whether SMM is related to the risk of postpartum readmission.

Table 1 shows the general characteristics of the study participants according to SMM (see supplementary Table S1). Of the 90,035 women included in this study, 2041 (2.27%) had SMM during the delivery hospitalization. The overall postpartum readmission was 0.95%. Among women with SMM, postpartum readmission was recorded in 2.45, and 0.92% of those without SMM had postpartum readmission.

Figure 1 presents the Kaplan Meier curve for the comparison of the unadjusted postpartum readmission rates according to SMM. The incidence of postpartum readmission was significantly higher in women with SMM than in those without SMM (P < 0.0001).

Table 2 presents the results of the Cox proportional hazard analysis (see supplementary Table S2 for full model). The risk of postpartum readmission was higher in women with SMM than in those without SMM (HR: 2.36, 95% CI: 1.75–3.19). In addition, compared with the reference group, women who in their 20s (20–24 years: HR 1.38, 95% CI 1.02–1.88; 25–29 years: HR: 1.29, 95% CI: 1.10–1.51), those who were nulliparous (HR: 1.59, 95% CI: 1.04–2.45), and those who delivered in a tertiary hospital (HR: 1.55, 95% CI: 1.19–2.02) were at high risk of postpartum readmission.

Figure 2 shows the results of the subgroup analysis between postpartum readmission and SMM by the mode of delivery. Women with SMM and spontaneous vaginal (SV) delivery or CS were more likely to have a high risk of postpartum readmission than those without SMM (SV delivery, HR: 3.31, 95% CI: 1.71–5.79; CS, HR: 1.86, 95% CI: 1.26–2.77). There was marginally no statistical significant association between women with SMM and instrumental delivery.

The current study found that SMM increases the risk of postpartum readmission. Additionally, women in their 20s, nulliparous women, and women who delivered in a tertiary hospital were associated with the risk of postpartum readmission. Although there was no statistically significant association between the type of delivery and postpartum readmission, women with SMM and SV delivery or CS were at high risk of postpartum readmission, which was a statistically significant association. To our knowledge, this population-based cohort study is the first one to determine the relationship between SMM and postpartum readmission using nationwide cohort data.

Herein, the overall incidence of SMM was 2.27%, which was similar and different compared to that reported previously. Previous studies reported an SMM incidence of 2–2.5% [9, 25, 26], whereas others reported that of 0.3–1.7% [11, 14, 27]. This difference could be because participants had different maternal health conditions in their communities or countries, and different indicators for SMM were used. Several US studies have shown that the incidence of SMM is 2–2.5% because the study populations included individuals of the same nationality, although their races were different, and they used the same SMM indicators, such as CDC’s algorithm [9, 25, 26]. In contrast, other studies that have shown a low incidence of SMM used clinical indicators of SMM such as blood cell transfusion units [11, 27], ICU admission [11, 27], or opinion of the treating obstetrician [14, 27]. Moreover, this may explain the difference in the participants’ nationality, which affects maternal health conditions based on various cultures, communities, and countries’ healthcare systems. However, CDC’s algorithm for SMM was used in this study; the incidence of SMM was similar to that reported in previous studies that used CDC’s algorithm, even though most of our study population was Asian. Therefore, the used indicators might affect the incidence rate of SMM. Moreover, the overall incidence of postpartum readmission was 0.95%, which was less than that in previous studies. One Canadian study reported that the incidence of postpartum readmission was 1.8% within 60 days after initial discharge [18], whereas other studies in the US noted that 1.2% or 1.72–2.16% of women experienced postpartum readmission within 6 weeks after delivery from 2004 to 2011 [5, 16, 17]. This difference might be because of the different health care systems or obstetric care, or the different evaluation of quality of care could affect the incidence of postpartum readmission. Further, our methods of calculating postpartum readmission might have underestimated the incidence.

This study confirmed that SMM and the risk of postpartum readmission were significantly related. Women with SMM had a 2.35 times higher risk of postpartum readmission than those without SMM. Additionally, mothers aged 20–24 and 25–29 years had a higher risk of postpartum readmission than those aged 30–34 years, and nulliparous women had a higher risk of postpartum readmission than did multiparous women. There was no evidence of the cause of this relationship, but we can hypothesize the etiology. Previous studies have revealed that the relationship between age and the risk of SMM is J-shaped, which means that young and older women have a high risk of SMM [28, 29]. These studies also showed that nulliparous women have an increased risk of SMM. In the current study, younger women had several risks of SMM, especially mothers in their teens or early twenties who are more likely to be nulliparous, have a low income level [28], and have less adequate prenatal care access [29]; that is, young women might have significant risks of SMM, and therefore, younger age might be associated with the incidence of postpartum readmission. Although women with SMM recover their symptoms of SMM, it is necessary to continuously monitor them during postpartum. However, the health care system tended to avoid long-term hospitalization. When patients do not have severe complications, early hospital discharge is recommended, and new patients are hospitalized because of the health care system’s profit structure. Another possible reason for the lack of evidence is the reduction of human resources. Staff deficiency, a lack of resources and preparedness for emergent situations, and less experienced physicians during off-hour periods are expected to be associated with poorer quality and safety [30, 31]. Deficient human resources might cause poor quality of maternal care and lead to adverse health outcomes [30], as well as increasing readmission during postpartum.

For the mode of delivery, we could not find a statistical significant association with postpartum readmission. Liu and colleagues showed that total CS had a 2.18 times higher risk of postpartum readmission than vaginal delivery [18]. Clapp et al. reported that CS had a 1.43 times higher risk of all-cause postpartum readmission than vaginal delivery [5], but this finding resulted from unadjusted covariates. In another study, CS had a statistically significant association with postpartum readmission when adjusted by year, patient demographics, and hospital characteristics, but when maternal comorbidities were adjusted, the relationship between CS and postpartum readmission became insignificant [16]. Herein, CS did not have a statistically significant association with postpartum readmission compared with SV delivery because we adjusted SMM and all other control variables. Women with SMM had a high risk of CS [11, 13, 14]; if we controlled for SMM, the risky health condition of maternity would be uniform and only women with a relatively healthy pregnancy might have remained. Furthermore, we excluded women with an extremely long-term delivery admission and those who died during delivery hospitalization. CS might have had an insignificant association with postpartum readmission because we did not include high-risk women. However, in the subgroup analysis, women who with SMM had a 3.13 times and 1.86 times higher risk of postpartum readmission by SV delivery and CS, respectively. Regardless of the mode of delivery, SMM was a high-risk factor for postpartum readmission; especially, if women with SMM had SV delivery, the risk was significantly increased compared with other modes of delivery.

Several limitations of this study must be discussed. First, the incidence of postpartum readmission can change depending on the method of calculating readmission based on the administrative data. Although our readmission rate was underestimated, it might more accurately reflect the real-world readmission rate. Second, we did not estimate cause-specific postpartum readmission but included all-cause postpartum readmission. The aspects of cause-specific readmission might be overestimated. It might have been better to determine which factors were more likely to affect the risk of postpartum readmission. Further research is needed to determine the specific causes of postpartum readmission. Third, we used administrative data that did not include relevant clinical data on the severity of disease; therefore, we did not identify the severity of SMM. Moreover, there was a problem with conversion of the ICD-9 procedural codes. The ICD-10 codes did not include the procedural codes; therefore, we converted the ICD-9 procedural codes to match the EDI codes. During this process, some codes did not match the EDI codes; hence, some procedure-based SMM indicators might have been less precise.

Nevertheless, our study has several strengths. First, this is a unique study of SMM and postpartum readmission. Until recently, there has been little evidence regarding SMM and postpartum readmission with any maternal, clinical, and provisional indicators. Our research might provide important evidence for use in future maternal health care. Second, this study had a population-based design and included long-term follow-up and nationwide data; therefore, the results could be representative of the nation. Lastly, this database was continuously updated for 12 years without periods of interruption.

This study founded that severe maternal morbidity was related to the risk of postpartum readmission. Policy makers should consider introducing policies to improve the quality of maternity care during the intrapartum and postpartum periods to minimize the incidence of SMM and reduce postpartum readmission. Since we did not determine the causes for postpartum readmission, more research on this subject and whether sub-indicators of SMM affect specific causes of postpartum readmission is needed. This information would provide further evidence that could support individual-level and population-level policies for improving maternal health outcomes and quality of care for childbirth.