Objective
To evaluate a policy of routine versus selective postpartum complete blood count (CBC).
Study Design
Historic case control design with matched subjects from 1 year periods bracketing the policy change (n = 800). Our primary outcome was postpartum transfusion rate. Univariable and multivariable analyses were performed. Total hospital costs were estimated.
Results
Eliminating routine postpartum CBC testing was associated with decreased transfusion rates (5.5% vs 1.8%, P = .007) despite similar transfusion risks. CBC utilization decreased from 59% to 22.2% ( P < .0001). No adverse bleeding outcomes occurred. Multivariable modeling suggested that the occurrence of postpartum hemorrhage was the best clinical predictors of transfusion n risk. Tachycardia, oliguria, and symptoms were also effective at identifying transfusion candidates. Elimination of routine CBC was independently associated with a reduced risk of transfusion (odds ratio, 0.30; 95% confidence interval, 0.12–0.72). Annual cost savings were estimated at $58,000.
Conclusion
Targeted CBC testing results in fewer transfusions, lower costs and improved quality of patient care.
Routine assessment of maternal hemoglobin level at obstetric admission is common practice and is aligned with recommendations from the American Congress of Obstetricians and Gynecologists (ACOG). In contrast, the clinical utility of routine postpartum hemoglobin assessment is less clear; resulting in practice variations. Ostensibly, the benefits of routine evaluation would be to identify women with postpartum anemia without a clinical risk factor such as postpartum hemorrhage and without clinical symptoms of hypovolemia. The potential risks of routine evaluation include delaying evaluation until a scheduled phlebotomist’s visit in women who might benefit from earlier evaluation, and encouraging unnecessary transfusion in asymptomatic women. In 2011, attention was focused on excess transfusion in many specialties by the implementation of Blood Management Performance Measures by the Joint Commission. Similarly, policies of restrictive rather than liberal blood transfusion have been associated with improved patient outcomes in conditions such as acute gastrointestinal bleeding and others.
The primary objective of our study was to evaluate the clinical use and cost of routine postpartum hemoglobin evaluation. In 2010, in anticipation of the 2011 performance measures, and in response to a higher than expected rate of postpartum transfusion at our institution, our postpartum order set was changed; routine postpartum complete blood count (CBCs) were removed from the order set, and replaced with a targeted evaluation based on a clinical protocol. What follows is our evaluation of the impact of this institutional policy change on patient outcomes, primarily postpartum transfusion, and hospital costs. An important secondary objective was to estimate any potential adverse effects on patient safety.
Materials and Methods
A retrospective case control study was performed. In 2010, we changed our postpartum order set from routine postpartum CBCs to targeted CBCs. Targeted CBCs were performed for signs of intravascular depletion including: maternal tachycardia, oliguria, dizziness, or orthostasis. In addition, a CBC was performed at the discretion of the attending physician, a judgment that was assumed to be based on estimated intrapartum blood loss or other clinical factors. Allowing a 1 year adjustment during the year the policy change occurred, 200 vaginal and 200 cesarean deliveries were analyzed for 1 year before the policy change (2009, time period 1) and 1 year following the policy change (2011, time period 2) for a total of 800 subjects. Study subjects were selected at random and the sample size was chosen to provide a power of 0.8 to detect a 50% reduction in transfusion rates from our baseline rate of 12.5%. A P value of .05 was considered to be significant.
Subjects were identified using the Medical University of South Carolina perinatal validated research database. Data were obtained through a combination of database extraction (eg, demographic variables), electronic data warehouse data (CBC) and detailed chart review (eg, patient symptoms, delivery complications, estimated blood loss, and transfusion). Maternal adverse outcomes (intensive care unit admission, maternal death, reoperation) were carefully reviewed.
The χ 2 or Fisher exact 2-sided tests were used for proportions/rates; Student’s t test was used for means and Mann-Whitney U test for medians. Forward conditional multivariate modeling was used to evaluate whether study period remained significantly associated with transfusion rate after controlling for differences with respect to potential transfusion risk factors between time period 1 and 2 that were identified in univariable analyses or had face validity. Candidate variables included: study period, maternal age, number of prenatal care visits, preterm rupture of membranes, antepartum anemia, and postpartum hemorrhage (>500 cc for vaginal delivery and >1000 cc for cesarean delivery). Nonsignificant variables were eliminated in a stepwise fashion. A simple costsavings analysis was performed by extracting estimated costs of laboratory testing (CBC, blood type, and screening), transfusion premedication (diphenhydramine 50 mg and acetaminophen 650 mg), and packed red blood cells (per unit) from our hospital cost database.
Results
A total of 200 vaginal and 200 cesarean deliveries for each time period (2009 and 2011) were analyzed, yielding a final sample size of 800. Group comparisons for demographics are summarized in Table 1 . There were significant differences in maternal age and frequency of prenatal care visits between the groups. The cohort in time period 2 had a slightly older maternal age and less frequent prenatal care visits than the cohort in time period 1, which could have increased transfusion risk. Table 2 summarizes pregnancy complications seen in the 2 groups; overall this was a relatively high risk population, consistent with a tertiary care referral hospital. Pregnancy risk factors were similar between the 2 cohorts with the exception of significantly increased risk of preterm premature rupture of membranes in time period 1. Delivery complications are demonstrated in Table 3 . There were no significant differences in key risk factors for transfusion including maternal fever, postpartum hemorrhage, and placental abruption.
| Variable | Time 1 (2009) n = 400 | Time 2 (2011) n = 400 | P value |
|---|---|---|---|
| Age, y | 27.1 ± 5.0 | 28.1 ± 5.3 | .008 |
| Black race | 86.0% | 82.2% | .344 |
| BMI, kg/m 2 | 35.6 ± 9.6 | 35.2 ± 9.5 | .61 |
| Married | 15.5% | 19.5% | .16 |
| Insurance | .11 | ||
| Medicaid | 79.5% | 79.5% | |
| Medicare | 9.8% | 13.5% | |
| Private | 7.0% | 4.0% | |
| Self-pay | 3.8% | 3.0% | |
| Parity | .68 | ||
| 0 | 28.2% | 24.0% | |
| 1 | 26.0% | 26.2% | |
| 2 | 25.2% | 25.0% | |
| ≥5 | 4.0% | 4.8% | |
| Prenatal visits | 14.3 ± 7.4 | 13.1 ± 6.2 | .02 |
| Anesthesia | .63 | ||
| None | 12.0% | 10.5% | |
| Regional | 82.2% | 84.8% | |
| General | 5.8% | 4.8% | |
| Gestational age, wk | 36.8 ± 3.7 | 37.0 ± 3.8 | .96 |
| Birthweight, g | 2838 ± 837 | 2829 ± 829 | .40 |
| Median neonatal LOS, d | 3 (1-118) | 2 (1-174) | .05 |
| Variable | Time 1 (2009) | Time 2 (2011) | P value |
|---|---|---|---|
| Chronic hypertension | 16.2% | 16.2% | > .99 |
| Gestational HTN | 4.8% | 3.8% | .60 |
| Preeclampsia | 12.5% | 11.2% | .66 |
| Thrombocytopenia | 1.2% | 1.5% | > .99 |
| Gestational diabetes | 9.8% | 7.5% | .31 |
| IDDM | 6.5% | 6.5% | > .99 |
| Polyhydramnios | 1.0% | 1.2% | > .99 |
| Preterm labor | 5.0% | 2.8% | .14 |
| GBS positive | 29.2% | 32.5% | .36 |
| Tobacco use | 10.5% | 13.0% | .34 |
| PPROM | 15.0% | 9.8% | .03 |
| Antepartum bleeding | 1.2% | 2.8% | .21 |
| Variable | Time 1 | Time 2 | P value |
|---|---|---|---|
| Fever | 4.5% | 2.8% | .26 |
| Postpartum hemorrhage | 7.5% | 6.2% | .58 |
| Placental abruption | 2.2% | 1.2% | .42 |
| Uterine rupture | 1 | 0 | |
| Manual extraction of placenta | 1 | 1 | |
| Cervical laceration | 1 | 0 | |
| Operative vaginal delivery | .46 | ||
| Vacuum | 3.0% | 5.5% | |
| Forceps | 0.5% | 0.5% |
Elimination of a routine postpartum CBC was associated with a reduction in the rate of blood transfusion from 5.5% to 1.8%, ( P = .007, Table 4 ). In addition, the utilization of postpartum CBC testing was reduced by half (59% to 22.2%, P < .0001). There was also a nonstatistically significant trend toward a shorter hospital length of stay. There were no adverse bleeding outcomes in either time period (ie, intensive care unit admission, maternal death, reoperation). Table 5 summarizes the clinical symptoms that were used to identify transfusion candidates. Clinical symptoms were strongly associated with the need for transfusion and could be used to identify patients that required further evaluation and ultimately transfusion. The most important predictors of transfusion were tachycardia, oliguria, and patient complaint of dizziness. The trends in transfusion interval (delivery to transfusion) were compared between time period 1 and 2. There was a nonsignificant trend toward earlier targeted transfusions in time period 2 ( Figure ). The transfusions in time period 1 were clustered at 24 hours, coinciding with the results from the postpartum CBC.
| Variable | Time 1 (2009) | Time 2 (2011) | P value |
|---|---|---|---|
| Anemia | 29.5% | 23.5% | .07 |
| Admit hematocrit, % | 33.6 ± 3.7 | 33.6 ± 3.8 | > .99 |
| Estimated blood loss, median mL | 500 (20-3500) | 500 (100-2200) | .94 |
| Postpartum hematocrit, % | 28.7 ± 4.2 | 28.7 ± 4.5 | .96 |
| Postpartum CBC collected | 59% | 22.2% | < .0001 |
| Transfusion | 5.5% | 1.8% | .007 |
| Units | .12 | ||
| 1 | 18.2% | 42.9% | |
| 2 | 72.7% | 42.9% | |
| >2 | 9.1% | 14.3% | |
| Symptoms | 3.5% | 2.5% | .54 |
| Oliguria | 1.5% | 1.8% | .78 |
| Maximum pulse | 91.6 ± 16.0 | 89.2 ± 14.2 | .03 |
| Low SBP | 113.8 ± 14.6 | 117.8 ± 62.4 | .21 |
| Low DBP | 63.9 ± 10.3 | 66.5 ± 10.3 | < .001 |
| Orthostatics performed | 2.0% | 2.0% | > .99 |
| Length of stay, d | 3 (1-73) | 2 (0-31) | .16 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
