If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Hospital Corporation of America, Nashville, TNDepartment of Obstetrics and Gynecology, University of Utah, Salt Lake City, UTDepartment of Maternal-Fetal Medicine, St. Mark's Hospital, Salt Lake City, UT
The purpose of this study was to analyze reasons for postpartum readmission.
We conducted a database analysis of readmissions within 6 weeks after delivery during 2007, with extended (180 day) analysis for pneumonia, appendicitis, and cholecystitis. Linear regression analysis, survival curve fitting, and Gehan-Breslow statistic with Holm-Sidak all-pairwise analysis for multiple comparisons were used. Probability values of < .05 were considered significant.
Of 222,751 women delivered, 2655 women (1.2%) were readmitted within 6 weeks (0.83% vaginal delivery and 1.8% cesarean section delivery; P < .001). A high percentage of these readmittances occurred within the first 6 weeks: pneumonia (84%), appendicitis (43%), or cholecystitis (46%). Cumulative readmission rates were higher in the first 6 weeks after delivery than in the next 20 weeks (pneumonia curve gradient, 3.7 vs 0.11; appendicitis curve gradient, 1.1 vs 0.36; cholecystitis curve gradient, 6.6 vs 1.7).
The cause of postpartum readmission is primarily infectious in origin. A recent pregnancy appears to increase the risk of pneumonia, appendicitis, and cholecystitis.
The postpartum period is associated with significant physiologic changes that include major alterations in the immune system. Despite the recognized importance of the puerperium period in obstetric care, few data exist that describe puerperal morbidity beyond the period of the initial hospitalization for delivery. The few articles that have addressed postpartum readmission have focused primarily on the relationship between the route of delivery and the need for subsequent readmission rather than a comprehensive analysis of the conditions that necessitate readmission.
This limited focus has prevented the acquisition of data that could facilitate the development of programs to prevent readmission. Such data may also allow a critical examination of the contribution of infectious or inflammatory processes that are not commonly associated with pregnancy to the overall readmission rate.
For Editors' Commentary, see Table of Contents
See related editorial, page 1
The prevention of morbidity is the primary goal of any such evaluation. However, hospital readmissions have critical economic implications, particularly in an era of health care reform, and may account for up to 20% of all health care costs.
Indeed, the White House Office of Management and Budget has proposed a series of incentives and penalties to reduce readmission costs that include bundling of payments for hospitalization to include postacute care costs that are incurred within 30 days of discharge.
It was against this background that we sought to examine the indications for readmission after delivery. We hypothesized that both conditions, which are related directly to pregnancy and morbidity that traditionally have been believed to be unrelated to pregnancy, would contribute to readmission rates. Secondarily, we examined the incidence of readmission on the basis of delivery method.
Materials and Methods
We initially performed an internal quality-driven audit of ICD-9 discharge diagnosis codes for all women who delivered during 2007 who were readmitted within 6 weeks of the original delivery discharge date. This analysis involved 114 Hospital Corporation of America (HCA) hospitals in 21 states across the United States. In terms of delivery volume, these facilities ranged from <1000/year to >10,000/year (1 hospital). This population has been shown in previous publications to be representative of the US population as a whole.
Because some women who delivered late in 2007 were not admitted until the first few weeks of 2008, readmission analysis for such women included some 2008 data. Readmissions were stratified according to readmission diagnosis and mode of delivery (TABLE 1, TABLE 2).
TABLE 1Readmissions within 42 days of delivery, 2007
Patient data were analyzed according to diagnosis code groupings that involved similar pathophysiologic conditions. Most patients had >1 discharge code. In such cases, the primary diagnosis was used in the analysis, unless a secondary diagnosis was judged to be more specific or to represent the principal underlying clinical problem. Examples would include a primary diagnosis of “major puerperal infection-postpartum” (code 67004) and a secondary diagnosis of “abscess of appendix” (code 5401) or a primary diagnosis of “acute pancreatitis” (code 5770) with a secondary diagnosis of cholelithiasis with cholecystitis (code 57410). In these and similar cases, such patients were grouped according to the more specific or underlying secondary diagnosis. HCA has implemented policies, procedures, educational requirements, and regular audit activities to facilitate accurate coding practices.
Our preliminary internal quality audit suggested a far higher rate of readmission within 6 weeks of initial delivery discharge for 3 medical conditions (pneumonia, cholecystitis, and acute appendicitis) than might be expected. To further investigate this issue, we then obtained institutional review board approval for a broader, temporally stratified analysis of hospital admission rates for these 3 indications for all women who delivered in 2007 and who were readmitted during the 180-day period after their delivery. Cumulative admission data were collated for patients with a diagnostic code for appendicitis, cholecystitis, and/or pneumonia (Table 2). In addition to evaluating the primary study group of women who delivered in 2007 for 6 months after delivery discharge, we studied 2 control groups of nonpregnant women who were admitted to HCA hospitals (the first from January 1 to June 30, 2007, and the second from July 1 to December 31, 2007). The control patients were chosen in the following manner: specification of women from the system database who were similarly aged (14-45 years) to the postpartum sample without pregnancy-related codes within 6 months of their admission for 1 of the 3 conditions in question. Two control groups were used to control for any seasonal variation in disease incidence.
For all women who delivered in 2007, readmission data for pneumonia, appendicitis, and cholecystitis on each postdischarge day from 1-180 were recorded in a cumulative fashion, with date of discharge being day 0 for the pregnant women. The data were plotted as a scatter plot, and linear and polynomial regression lines were fitted, as appropriate, to determine the best fit, with the initial use of 6 weeks as the breakpoint. In the pneumonia analysis, it was obvious from simply looking at the best fit curve that 3 linear regression lines were required to best define the curve. We plotted separate curves for vaginal delivery and cesarean delivery. The slopes of the regression lines were compared with the use of the Gehan-Breslow test. The same analysis was done for each mode of delivery and for the nonpregnant control groups. When proportions were used, they were compared with a χ2 test. The Gehan-Breslow statistic was used to compare the curves in FIGURE 1, FIGURE 2, FIGURE 3, FIGURE 4. The Holm-Sidak all-pairwise analysis was used for multiple comparisons (Sigmastat for Windows, version 3.11; Systat Software, Inc, Point Richmond, CA). Significance was set at a probability value of < .05.
In 2007, there were 222,751 deliveries in facilities of the HCA. There were 2655 women readmitted within 6 weeks of their delivery discharge date, for a total readmission rate of 1.2%. A breakdown of readmission in the first 6 weeks after delivery by mode of delivery is shown in Table 1. Indications for readmission are detailed in Table 2. The probability values represent a χ2 comparison of the readmission rates. Readmission within the first 6 weeks after primary (2.1%) and repeat (1.4%) cesarean delivery was significantly (P < .001) more common than after vaginal delivery (0.83%; Table 1). Odds ratios and 95% confidence intervals are presented in Table 1.
Table 2 shows the data regarding readmission in the first 6 weeks after delivery discharge. Endometritis, sepsis/bacteremia, cellulitis, and necrotizing fasciitis were more common in women who had a cesarean section delivery, compared with those who were delivered vaginally (Table 2). In addition, readmissions for cholecystitis and pneumonia, but not appendicitis, were more common in women who had been delivered by cesarean section delivery.
An extended longitudinal analysis of readmission rates for pneumonia, appendicitis, and cholecystitis was then undertaken. With the use of a cumulative daily total from postpartum day 1 until 6 months, there was a clear breakpoint in the readmission rate at 6 weeks for appendicitis and cholecystitis; for pneumonia, there were breakpoints at 2 weeks and 6 weeks (FIGURE 1, FIGURE 2, FIGURE 3). Sixty percent (58/96), 57% (309/539), and 46% (46/100) of women with a readmission diagnosis of appendicitis, cholecystitis, or pneumonia, respectively, were delivered vaginally. The curve for the percentage of total admissions for pneumonia after cesarean section delivery was significantly different than that for vaginal delivery (Figure 4) and showed an earlier and higher percentage of readmissions in the first 6 weeks after delivery than thereafter (up to 6 months). This difference was not seen for appendicitis or cholecystitis. The rate of admission for these 3 conditions in nonpregnant women remained stable throughout the 1-year period that was examined (2007) and was significantly lower than that seen in the first 6 postpartum weeks in the primary study group (Figure 4). There was no difference seen in the rate of admission for these conditions between the 2 control groups (data not shown), which indicated that there was no seasonal variation present.
The 6 weeks after delivery traditionally has been referred to as the postpartum period; complications that occur within this time frame commonly are assumed to be related, or potentially related, to the pregnancy or events of delivery. Although this designation is arbitrary and lacks true scientific foundation, convention dictates that we use this time frame for analysis, and accordingly we did so. Our data confirm that, although readmissions in the first 6 weeks after delivery are uncommon, cesarean delivery carries with it roughly twice the risk for readmission as does vaginal birth. These data also confirm that the increased morbidity, which has been previously described with primary as opposed to repeat cesarean section delivery, extends at least into the period of 6 weeks after discharge. Potential explanations as to the reason that a primary cesarean section delivery represents a greater risk for readmission than a repeat cesarean delivery may include such factors as increased risk of infection from prolonged labor (repeated vaginal examinations, chorioamnionitis), increased blood loss with emergency surgery, and higher use of general anesthesia. The rates of readmission after delivery in our study (when comparable) are similar to those reported in the literature (cesarean section vs vaginal delivery: odds ratio, 2.18 [our study] vs 1.81 or relative risk, 1.92). Most studies have used a cutoff of 8 weeks,
Hypertension and uterine and wound infections were the most common reason for readmission in both the vaginal and cesarean delivery groups. The frequency of infectious causes for readmission is not surprising, given the invariable clean-contaminated nature of all delivery procedures and the known frequency of postpartum endomyometritis, particularly in women who have undergone cesarean delivery. The administration of perioperative antibiotics has been demonstrated to reduce the rate of immediate postcesarean delivery infectious complications, including endomyometritis and wound infection.
During the study period, the administration of cefazolin after cord clamping was the most common practice in our facilities. Recent data suggest that the administration of such antibiotics before the surgical incision should replace the older practice of delaying such administration until the umbilical cord has been clamped.
The frequency of these conditions in our population suggests that such measures, in addition to standard techniques of surgical asepsis, may also be important in the reduction of later readmissions (Table 2). Our data, which demonstrated an increased rate of readmission for sepsis or bacteremia after cesarean delivery compared with vaginal birth, likely reflect the devitalization of tissue that invariably accompanies cesarean delivery. This alludes to the potential for the systemic spread of infection from microscopic pockets of infarcted tissue, which allows seeding of bacteria into the surrounding circulatory system. Bacteremia is known to complicate up to 14% of cesarean deliveries after failed labor, especially in preterm labor cases and those complicated by chorioamnionitis.
Our finding that readmission for hypertension was common was unexpected. A detailed, case-by-case analysis of avoidable antecedents that predispose to readmission is beyond the scope of this article. Our data do suggest, however, that studies that focus on peripartum management of hypertension, with the aim of decreasing readmissions, may have significant impact on maternal health and health care costs.
We were surprised that there was no difference in the rate of readmission for urinary tract infection between the cesarean and vaginally delivered groups. Given the almost universal placement of an indwelling catheter in women who undergo cesarean delivery, the expectation was that there would be a higher rate of urinary tract infection in these women than in those who gave birth vaginally. Much of this lack of difference is probably due to the common administration of prophylactic antibiotics for endomyometritis in women who undergo cesarean delivery; these are antibiotics that also cover most urinary pathogens. Prevention of urinary tract infections is therefore a secondary benefit of such prophylaxis. The relatively high rate of urinary tract infection in patients who did have a vaginal birth probably reflects the increasing use of epidural anesthesia and the common nursing practice of intermittent or indwelling catheterization during labor for such women. Although our data do not justify a recommendation for routine antibiotic prophylaxis in this group of women, they do emphasize the importance of the minimization of such invasive procedures and of strict adherence to aseptic technique when catheterization is necessary. We believe that, in women who have undergone multiple intermittent catheterizations or more prolonged indwelling catheter placement during the course of vaginal delivery, a single dose of oral microbial suppressant therapy may be an appropriate consideration.
Perhaps of most interest was our observation of a significantly higher rate of hospital readmission for cholecystitis, appendicitis, and pneumonia in the first few postpartum weeks than would be expected by chance. None of these conditions has ever before been linked causally to pregnancy or delivery. There are several possible explanations for our observations. First, the nature of our data does not allow distinction between true- or false-positive diagnoses of these conditions. Thus, it is possible that nonspecific postoperative pain is attributed falsely to cholecystitis or appendicitis during the puerperium period. We believe this explanation to be unlikely, however, given that this diagnosis was also more frequent in the puerperium period in women who underwent vaginal birth. Further, although an erroneous discharge diagnosis of cholecystitis might be made on the basis of less than certain clinical findings, such an erroneous discharge diagnosis would likely be quite uncommon in the case of appendicitis or pneumonia. Second, it is possible that such diagnoses are made more commonly in the puerperium period simply because of increased patient access to health care during this period of time. Thus, gallbladder disease that might have gone unreported during late pregnancy might be brought to the attention of a health care provider in the early puerperium period. However, because appendicitis and pneumonia are likely to be fatal if untreated, similar findings with these conditions make this explanation less likely. Although the cause of these findings requires both additional confirmation and investigation, we believe that the release of inflammatory mediators during the process of birth could potentially aggravate an underlying subclinical process and result in frank disease expression in the early puerperium period. In this same vein, physiologic immune reconstitution (as opposed to the release of inflammatory mediators from the uterus and placenta) could lead to the emergence of quiescent conditions or decreased resistance to new infection.
During pregnancy, T-helper 2 (eg, interleuken-10) and T-helper 3 responses (eg, transforming growth factor β) are enhanced, although T-helper 1 cytokines (ie, interleuken-12 and interferon-γ) are suppressed.
In addition, cortisol, progesterone, norepinephrine, and 1,25-dihydroxyvitamin, which are involved in the modulation of immune responses, change significantly during pregnancy and in the postpartum period.
The increased risk of pneumonia after cesarean section delivery, compared with vaginal delivery, in the first 6 weeks after delivery is also extremely interesting. Although data are few, there may be a link between cesarean section delivery and the increased risk for pneumonia after a pregnancy delivered with the use of regional anesthesia. There are data to suggest that spinal anesthesia may decrease the sensitivity of the laryngeal airway protective reflexes.
In addition, spinal anesthesia, when compared with endotracheal intubation or laryngeal mask airway placement, does not stimulate an increase in salivary secretion concentration of immunoglobulin M or increased secretion of immunologic host defenses.
Women who have had a spinal anesthetic for a cesarean section delivery may have a suppressed cough reflex for up to 4 hours after delivery, which may increase the risk of aspiration during and after delivery.
Combining this diminished ability to protect the airways with a reduced immunologic response to infection may be 1 mechanism to explain the findings in our study. Regardless of the pathophysiologic conditions underlying these observations, our findings of an increased rate of cholecystitis, appendicitis, and pneumonia in the immediate puerperium period, compared with the first 6 months after delivery, have important clinical implications and suggest the need for enhanced vigilance and a lower threshold for performance of definitive diagnostic tests when signs or symptoms of these conditions appear in the weeks after birth.
This study has the benefit of a large sample size from a representative US population that was managed in a wide variety of hospital settings. The weaknesses are those which accompany the interrogation of any retrospective discharge database (eg, accuracy of data, lack of adjustment for comorbidities, or case-mix differences). In addition, we cannot be sure that all patients were captured because some may have been readmitted to a non-HCA facility, and we were unable to take into account any patients who may have died after delivery or discharge.
In conclusion, our data also highlight the need to include readmission for nonobstetric reasons in the puerperal period when the risks and benefits of cesarean vs vaginal delivery were evaluated. The fact that the risk of pneumonia is increased after cesarean section vs vaginal delivery may be another factor to be considered when elective cesarean section delivery is being considered.
Risk of maternal postpartum readmission associated with mode of delivery.
Authorship and contribution to the article is limited to the 10 authors indicated. There was no outside funding or technical assistance with the production of this article.
Cite this article as: Belfort MA, Clark SL, Saade GR, et al. Hospital readmission after delivery: evidence for an increased incidence of nonurogenital infection in the immediate postpartum period. Am J Obstet Gynecol 2010;202:35.e1-7.
Extensive physiologic, endocrine, and immunologic adaptations occur in the mother in response to the demands of pregnancy, which includes support of the fetus, preparation of the uterus for labor, and protection of the mother from potential cardiovascular injury at delivery. Although these demands increase incrementally over many months, resumption of the prepregnancy physiologic state after childbirth occurs at a far more rapid pace. The postpartum period, which is also referred to as the postnatal period (Latin for “after birth,” from post meaning “after” and natalis meaning “of birth'”) or puerperium (Latin for “after childbirth,” from puerperal meaning “a woman in childbed”), refers to the period that begins immediately after delivery and continues for the next 6 weeks.