Objective
We sought to evaluate whether the presence of condition-specific obstetric protocols within a hospital was associated with better maternal and neonatal outcomes.
Study Design
This was a cohort study of a random sample of deliveries performed at 25 hospitals over 3 years. Condition-specific protocols were collected from all hospitals and categorized independently by 2 authors. Data on maternal and neonatal outcomes, as well as data necessary for risk adjustment were collected. Risk-adjusted outcomes were compared according to whether the patient delivered in a hospital with condition-specific obstetric protocols at the time of delivery.
Results
Hemorrhage-specific protocols were not associated with a lower rate of postpartum hemorrhage or with fewer cases of estimated blood loss >1000 mL. Similarly, in the presence of a shoulder dystocia protocol, there were no differences in the frequency of shoulder dystocia or number of shoulder dystocia maneuvers used. Conversely, preeclampsia-specific protocols were associated with fewer intensive care unit admissions (odds ratio, 0.28; 95% confidence interval, 0.18–0.44) and fewer cases of severe maternal hypertension (odds ratio, 0.86; 95% confidence interval, 0.77–0.96).
Conclusion
The presence of condition-specific obstetric protocols was not consistently shown to be associated with improved risk-adjusted outcomes. Our study would suggest that the presence or absence of a protocol does not matter and regulations to require protocols are not fruitful.
The importance of measuring and improving quality in obstetrics has been increasingly recognized as a priority. As a result, an increasing number of studies and opinions concerned with how to measure and improve quality of care in obstetrics have been published. In this literature, the concept of using protocols to improve care has been suggested as an important component of patient safety initiatives. Protocols are a plan of treatment that provide detailed instructions for the medical team on what to do once a specific complication, such as hemorrhage, shoulder dystocia, or preeclampsia, has arisen. Pressure to adapt protocols may also come from government regulation and from malpractice carriers. These regulations often do not have requirements on how the protocols are implemented, just that protocols are required. Correspondingly, many departments have started implementing condition-specific protocols.
Despite the increasing emphasis on the use of protocols to improve care, there is relatively little empiric evidence that obstetric outcomes are improved through their use. Also, studies that have demonstrated an improvement have tended to be performed at single centers, and as such the generalizability of the findings is uncertain. One further difficulty in determining whether protocols are associated with better outcomes is that preexisting patient characteristics also influence patient outcomes, and changes over time in these characteristics need to be accounted for in any longitudinal study. We hypothesized that hospitals with condition-specific obstetric protocols in place would have better risk-adjusted patient outcomes than hospitals without such protocols.
Materials and Methods
From 2008 through 2011, we performed a cohort study at 25 hospitals in the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. This study, the Assessment of Perinatal Excellence (APEX), was designed to develop quality measures for intrapartum obstetrical care. The APEX study was approved by the institutional review board at each participating institution under a waiver of informed consent. This is a planned secondary analysis of the APEX dataset.
Each hospital that participated in the study submitted all existing labor and delivery protocols each quarter that the study was ongoing. Research staff at participating hospitals were instructed to submit all potential protocols pertaining to obstetrics and to provide the dates that the individual protocols went into (or out of) effect. Two authors (J.L.B. and W.A.G.) then independently reviewed the protocols and determined whether the submissions were truly condition-specific protocols (eg, as opposed, for example, to instructions narrowly directed toward single-medication administration) and what specific topics they pertained to (ie, hemorrhage, preeclampsia, or shoulder dystocia). Protocols had to involve regulation of provider behavior as well as other hospital systems issues to be considered a protocol. The 2 reviewers determined whether a condition-specific protocol was present or absent for the particular center. The content and quality of the protocol as well as the steps taken to implement the protocol were not assessed. Disagreements were resolved by conversation between the 2 reviewers until consensus was reached.
Patients were then categorized by whether they delivered in an institution at a time when specific protocols were in effect. For example, if a hemorrhage and shoulder dystocia protocol went into effect at a given hospital in February, patients delivering in January of that year were categorized as not having delivered in the presence of a protocol, but patients from February on were categorized as having delivered in the presence of both of these protocols.
Methods for risk adjustment of patient characteristics, and the specific risk-adjustment characteristics for hemorrhage, have been described previously. The same methodologies were used to determine the patient characteristics that should be used for risk adjustment for preeclampsia and shoulder dystocia. Only women eligible for a given outcome were included (eg, women who did not deliver vaginally were not assessed with regard to shoulder dystocia). Variables used in the risk-adjustment models are shown in Table 1 .
| Variable | Postpartum hemorrhage | Shoulder dystocia | Preeclampsia |
|---|---|---|---|
| Denominator size for each outcome (with all nonmissing covariates) | 105,165 | 75,910 | 11,529 |
| Maternal characteristics | |||
| Age | ● | ● | |
| Body mass index at delivery | ● | ● | |
| Cigarette use during pregnancy | |||
| Cocaine or methamphetamine use during pregnancy | |||
| Insurance status | ● | ● | ● |
| >2 prenatal care visits | ● | ||
| Obstetric history | ● | ● | |
| Any hypertension | ● | ||
| Diabetes mellitus | ● | ● | |
| Anticoagulant use during pregnancy | ● | ||
| Multiple gestation | ● | ||
| Placenta previa | ● | ||
| Placenta accreta | ● | ||
| Placental abruption | ● | ||
| PROM/PPROM | |||
| GBS status | |||
| Type of labor | ● | ||
| Gestational age | ● | ● |
Analysis was performed to determine whether risk-adjusted outcomes relevant to each complication were associated with the presence of relevant protocols at the time of delivery. For example, we assessed whether the presence of a postpartum hemorrhage protocol was associated with less frequent postpartum hemorrhages (ie, estimated blood loss >1000 mL) or the severity of the hemorrhage (severe hemorrhage was defined as either estimated blood loss ≥1500 mL at delivery, a blood transfusion, or a hysterectomy performed for hemorrhage, placenta accreta, or uterine atony). The analysis of the hemorrhage outcomes excluded women whose reason for admission was bleeding or abruption, given the greater potential that their outcomes were related to events prior to admission. Only women whose hemorrhage started in the hospital were eligible so that hospitals were not held accountable for the severity of a hemorrhage when much of the blood loss was beyond their control before admission. For women who had been given the diagnosis of preeclampsia, we evaluated whether the presence of preeclampsia-specific protocols were associated with any of the following markers of severity: eclampsia, any persistent severe blood pressures defined as 2 maternal systolic blood pressures ≥160 mm Hg or any 2 diastolic blood pressures ≥110 mm Hg, admission to intensive care unit (ICU), pulmonary edema, creatinine >2.0 mg/dL (if no creatinine value was present in the medical record, we assumed it did not exceed this value). Shoulder dystocia protocols were assessed to determine whether their presence was associated with the frequency of shoulder dystocia or the severity of the shoulder dystocia as assessed by the number of maneuvers (≤2 and ≥3) used to relieve shoulder dystocia. While ideally we would track poor infant outcomes for shoulder dystocia, we were unable to examine brachial plexus injury, cord blood gases, or bones fractures due to low numbers of outcomes.
Univariable comparisons were performed with χ 2 tests. Hierarchal models were developed for the outcomes of hemorrhage and shoulder dystocia and logistic models for the preeclampsia outcomes, with adjustment for patient characteristics, to estimate the independent association between the presence of a condition-specific protocol and obstetric complications. Logistical models were used for the eclampsia outcomes because the low number of outcomes did not allow for use of a hierarchal technique. No imputation for missing data was performed. Hierarchal models take into account hospital of delivery and hospital was added to the logistics model for eclampsia.
Results
Data were collected from 115,502 women. Unadjusted outcome rates are shown in Table 2 . The frequencies of hemorrhage outcomes and ICU admission for preeclampsia were significantly different between hospitals with and without protocols in place. Hospitals with a hemorrhage protocol had higher rates of hemorrhage and hospitals that did not have a preeclampsia protocol in place had higher rates of ICU admissions.
| Descriptive table of outcomes by protocol status | Hospital did not have hemorrhage protocol at time of delivery (n = 85,291) | Hospital did have hemorrhage protocol at time of delivery (n = 26,236) | P value comparing these groups |
|---|---|---|---|
| EBL >1000 mL | 7381 (8.7%) | 4240 (16.2%) | < .01 |
| Severe PPH | 1772 (2.1%) | 679 (2.6%) | < .01 |
| Hospital did not have dystocia protocol at time of delivery (n = 59,852) | Hospital did have dystocia protocol at time of delivery (n = 18,503) | ||
| Shoulder dystocia | 1641 (2.7%) | 469 (2.5%) | .13 |
| No. of maneuvers ≥3 | 285 (0.5%) | 117 (0.6%) | .02 |
| Hospital did not have preeclampsia protocol at time of delivery (n = 9598) | Hospital did have preeclampsia protocol at time of delivery (n = 1931) | ||
| Eclampsia | 32 (0.3%) | 10 (0.5%) | .22 |
| Severe hypertension a | 4345 (45%) | 800 (41%) | .002 |
| Admission to ICU | 373 (3.9%) | 22 (1.1%) | < .0001 |
| Pulmonary edema | 104 (1.1%) | 19 (0.98%) | .81 |
| Creatinine ≥2 mg/dL (if missing assume <2) | 80 (0.83%) | 18 (0.93%) | .69 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
