Multiple covariates each can cause differences in mean labor duration among women at individual hospitals by significantly greater than a clinically important period (e.g., 1 hour).1–3 For example, women homozygous for the β2 adrenoceptor Arg/Arg16 have longer duration labors than other women of the same parity.1 This genotype is commonly found in Asian populations,2 and labor might be longer in populations rich in women of Asian heritage.2 Women homozygous for “T” at catechol-O-methyltransferase rs4633 have longer duration of labors.3 Obese women have longer duration of labors.2 Babies with weight exceeding 3 kg are delivered after longer durations of labor.3 These examples1–3 show that there may be sources of biological variability unbalanced among hospitals and causing heterogeneity among hospitals in the mean durations of labor.
The expected duration of labor may also be dependent on obstetrical practices, some influenced by the wishes of the mother for her birth experience.4 At some hospitals, active management of labor is practiced with the intention of reducing the risk of infection and hemorrhage associated with prolonged labor.5 At other hospitals, less intrusive protocols are used with the intention of providing a more natural birth experience leading to longer labors.
We analyze data from the American Society of Anesthesiologists’ (ASA) Anesthesia Quality Institute (AQI) to report the distribution of the mean durations of labor analgesia among hospitals in the United States.a,6–8 Based on the earlier studies,1–5 we hypothesized significant heterogeneity among hospitals in the duration of labor analgesia for vaginal delivery.
The cost of managing a labor epidural is related to the time from placement to birth, as well as the intensity of activity during its management. Methods of payment of anesthesia providers (anesthesiologists and/or nurse anesthetists) for labor analgesia are heterogeneous,b because there is not a requirement for continuous attendance of the anesthesia provider with the parturient, unlike for anesthesia in operating rooms.9 For example, at some hospitals, the anesthesia provider may place the epidural for labor and delegate its supervision to a labor nurse.c At other hospitals, the anesthesia provider places the epidural and is immediately available to fine-tune its efficacy, treat adverse effects, and provide other anesthesia treatments if the course of labor changes. Determining whether there is large heterogeneity in the mean durations of labor analgesia among US hospitals is important because, if present, such heterogeneity would influence the equity of different methods of fee-for-service payment.9
Data for Figures
The durations of labor shown in Figure 1 are from an observational study, of the influence of genetics on preterm labor, performed at Columbia University. The original study was approved by the Columbia University IRB with written informed consent. Patients were enrolled between September 2006 and December 2009. Healthy women of mixed parity were enrolled early in the second trimester of pregnancy. The study did not result in alteration of usual care. Labor was managed by the obstetrical service according to institutional protocols and national guidelines. Data used were from the 858 women from this cohort who had a vaginal delivery with epidural. The private hospital had a contract with its state to care for local residents covered by Medicaid. The patients have a substantial range of self-declared race/ethnicity: 1% Asian, 12% African American, 46% Hispanic, 13% Caucasian, and 29% mixed.
The data for Figure 2 are from an observational study of the influence of genetics on term labor progress, performed at the University of California at San Francisco. The study was IRB approved with written informed consent. Parturients with singleton gestations who intended vaginal delivery were enrolled during the third trimester of pregnancy. Deliveries occurred between November 2011 and October 2013. The study did not result in alteration of usual care. There was no study-related intervention in the labor room. We do not believe that the labor room personnel were aware of the patients’ participation in the study. As at the preceding hospital, labor was managed by the obstetrical service according to institutional protocols and national guidelines; however, patients were given substantial latitude to refuse recommended care. Data used were from the 163 women from this cohort who had a vaginal delivery with epidural. Although the data were from a state hospital, there were many private patients (i.e., multiple payers). As at the preceding hospital, the patients have a substantial range of self-declared race/ethnicity: 57% Asian, 7% African American, 26% Hispanic, 7% Caucasian, and 4% mixed.
The current study used limited deidentified data from the original 2 studies (time from epidural placement to delivery and the day of the week), with no protected health information.
Data for Figures 3 to 7
The University of Iowa determined that the use of the national (AQI) data “does not meet the regulatory definition of human subjects research,” “because the activity used de-identified data.” Durations of labor analgesia reported to the AQI from January 1, 2010, through September 30, 2014, were studied. Durations were reported in minutes ranging from 5 minutes (i.e., minimum reported time for placement) to a maximum of 1439 minutes (i.e., 24.0 hours). Some hospitals provided data using anesthesia Current Procedural Terminology (CPT) Codes, and others reported the primary surgical code from which the anesthesia code is calculated.b We used either anesthesia code 01967 or its corresponding surgical CPT: 59400, 59409, 59410, 59610, 59612, or 59614. These are the CPT codes for vaginal delivery. Cases with anesthesia or surgical CPT for cesarean delivery were excluded: 01961, 01968, 59510, 59514, 59515, 59618, 59620, or 59622.
Statistical Explorations and Intermediate Results
Our detailed data showed marked heterogeneity in the durations of labor analgesia. For example, Figure 1 shows a histogram of durations from Columbia University, with a mean of 6.80 hours (SE, 0.16 hours). The mean of the durations at Duke University was indistinguishable, 6.87 hours (SE, 0.70).10 In contrast, Figure 2 shows durations from the University of California at San Francisco. The mean is 16.90 hours (SE, 0.98).
Figure 3 shows a histogram with all 925,730 of the individual durations of labor analgesia for vaginal delivery reported to the AQI from 517 hospitals. Figure 4 shows the data of Figure 3 limited to durations <2 hours. Many of the durations in Figure 4 are 15, 30, or 60 minutes. Figures 1 and 2 show that most such durations and distributions in Figure 4 are too brief to represent the natural progression of labor. At the hospitals of Figures 1 and 2, there were only 3.4% and 3.7% of labor epidurals lasting 60 minutes or less, respectively. Thus, approximately 3.5% of patients may have epidural placement solely for second stage analgesia and/or instrumental delivery. This compares with >35.0% of patients (i.e., 10 × 3.5%) having durations 60 minutes or less at 25.0% of the 320 hospitals reporting at least 200 cases (Fig. 5).
Among the 320 hospitals, 11.6% accounted for 50% of the durations that were 60 minutes or briefer and 26.9% accounted for 80% of those durations (Fig. 5). In contrast, approximately half (45.6%, n = 175/320) of the hospitals had no greater than 5.0% of the durations lasting 60 minutes or less. We limited our further analysis to the data from the 175 hospitals reporting to AQI and for which no greater than 5.0% of durations lasted 60 minutes or less.d
Successive durations of labor analgesia from the 175 hospitals may not be independent and identically distributed across time (e.g., due to staff scheduling and/or case scheduling for elective cesarean deliveries influencing availability of anesthesia providers). In other words, the usual SE of a hospital mean that is calculated subject to the incorrect assumptions of statistical independence and homogeneity of individual durations will be inaccurate (Appendix). This arises often in the analysis of operating room information system data. Statistical analyses are commonly done after batching (binning) the data into 4-week periods, because managerial covariates are often unmeasured, not only nationwide but at individual hospitals (e.g., staff schedules and correlations among patients caused by appropriate queue management based on threshold numbers of patients).11–20 For example, we used the Kruskal-Wallis test to compare the labor epidural durations among the 7 days of the week. At random, 1.0% of the 175 hospitals should be statistically significant at a P < 0.01 criterion, but 10.6% met that criterion. At random, 5.0% should be statistically significant at P < 0.05 but 18.9% met that criterion.e
Analyses of the labor epidural data were done by using 4-week periods determined relative to January 1, 2010. For each combination of hospital and 4-week period with a mean of at least 1 epidural every couple of days (i.e., n ≥ 14), the mean was calculated on the durations of labor analgesia. Consider a hospital with m such 4-week periods. The estimate for the hospital’s mean duration of labor analgesia was taken by calculating the mean of the m individual estimates. Among the 175 hospitals, there were 172 hospitals that satisfied the preceding criteria (among ≥200 AQI reported cases ≤5.0% lasting ≤60 minutes) and had at least 5 four-week periods with 14 or more labor epidurals for vaginal delivery. Figure 6 shows a histogram of the labor epidurals for vaginal delivery per week at these 172 hospitals. The Results, detailed later, are limited to data from these hospitals.f
The 172 hospitals provided for n = 5671 combinations of hospital and 4-week period, with the numbers per period averaging 97.3 ± 67.4 epidurals (mean ± SD).g The 551,707 labor epidurals had a mean duration of 6.12 hours. The SE was <0.001 hours whether calculated from the 551,707 individual durations or the n = 5671 combinations of hospital and 4-week period. Our objective was to study the heterogeneity among hospitals from that overall mean value of 6.12 hours. The sample size for the test of each hospital ranged from 5 to 61 four-week periods, as described earlier. Statistical analyses were performed by using the Student 2-sided 1-group t test for each of the 172 hospitals.
The Lilliefors test was performed, for each of the 172 hospitals, to assess the assumption that the means of the 4-week periods follow a normal distribution (i.e., that the calculated SEs are accurate). If there were many 4-week periods and many measurements per period, the means of the 4-week periods would be expected to follow normal distributions by the central limit theorem. There were 2 hospitals for which this was not so, one with P = 0.0006 (n = 21 periods) and the other P = 0.0007 (n = 60). The distributions had greater kurtosis than normal distributions (i.e., peaked, not skewed). Both hospitals had means (6.50 and 6.19 hours, respectively) close to the overall national mean. In other words, the lack of a normal distribution had no consequence on our reported results. Nevertheless, to be conservative from a distributional perspective, and because there were multiple (172) comparisons, P < 0.001 was treated as significant. In the Appendix, we describe the methodology of our 2 sensitivity analyses providing different methods of analysis.
Using AQI Data to Test Primary Hypothesis
55.2% of the 172 hospitals had mean durations of labor analgesia for vaginal delivery that differed (P < 0.001) from the overall mean of 6.12 hours (Fig. 7). Among those, 55.2% were the 9.9% of hospitals with means ≤5.12 hours. Those mean durations ranged from 2.68 (SE, 0.17) to 5.10 (SE, 0.07) hours. Also among the 55.2% were the 12.2% of hospitals with means ≥7.12 hours. Those mean durations ranged from 12.03 (SE, 0.23) to 7.13 (SE, 0.08) hours. In the Appendix, we repeat the analyses 2 different ways and obtain the same results.
In this article, we analyzed data from AQI and report large heterogeneity in the mean durations of labor analgesia for vaginal delivery among hospitals reporting data to the AQI. This was even though we limited consideration to hospitals for which <5.0% of the labor epidurals were ≤60 minutes. In addition, the maximum reported durations were 1339 minutes. Had all hospitals and data been included, the heterogeneity in the mean durations among hospitals would have been greater than the substantive differences that we detected. Thus, our result of large heterogeneity in means among hospitals was obtained even though we underestimated the actual heterogeneity among hospitals.
Economic Implications of Our Results
One implication of there being large heterogeneity in the mean durations of labor analgesia among hospitals is the lack of validity of quantifying anesthesia productivity based on numbers of labor epidurals (e.g., mean daily number per anesthesia provider). However, for many hospitals, numbers of epidurals is unrelated to productivity (i.e., production per provider scheduled period), regardless of heterogeneity in the mean durations of labor analgesia. The reason is that there cannot be fractions of anesthesiologists or nurse anesthetists present. For most hospitals with few annual deliveries, regardless of whether their mean durations are less or greater than the overall US national mean, there will be just one anesthesia provider caring for the parturients.10,19,21 The mean, median, and mode labor epidurals for vaginal delivery was <4 per day (Fig. 6).21 In contrast, suppose that heterogeneity in durations of labor analgesia were to reflect heterogeneity in the durations of labor or the duration of hospitalized labor. Then, our findings may indicate differences in obstetrical nurse productivity because there is a direct relationship between the mean duration of labor and the mean obstetrical ward census, and between the mean census and the mean number of staffed obstetric beds needed.19
The principal implication of our results is for payment to US anesthesia groups. Payment can be based on face-to-face time with the patient (i.e., like a surgical anesthetic).b,9 There are initial (“base”) units (e.g., for placement of the labor epidural) and time units, 1 per 15 minutes of face-to-face time. Abouleish et al.9 previously analyzed Texas Medicaid data to evaluate the validity and equality of applying this practice to labor epidurals. They did not have data by hospital, but by provider. Their striking findings have substantially influenced policy.h Some providers spent nearly the entire period of labor with the patient, resulting in payment as if the patient were undergoing a general anesthetic.9 The results of Abouleish et al.9 show major limitations with payment based on face-to-face time.
Payment for labor analgesia can be a fixed amount of money.b For example, before December 2008, Iowa Medicaid paid anesthesiologists and nurse anesthetists a fixed amount of money for code 01967.i Although the payment was changed to the use of a time-based system as for operating room anesthesia, the payment was changed back to a fixed-payment system effective December 2014.i Our findings show major weaknesses to fixed payment, making payment based on time rational, as for surgery.22
First, there was large heterogeneity in the mean durations of labor analgesia among hospitals. In other words, although some individual women have much briefer durations of labor analgesia than do other women, the means among many women (i.e., total payment) will not be equal among hospitals (i.e., among anesthesiologists and nurse anesthetists working at different hospitals). Such heterogeneity can influence anesthesia workload because, after initial adequate analgesia from placement of the labor epidural, 6.8% (SE 0.2%) of patients subsequently have inadequate analgesia.23 Duration of labor analgesia >6.0 hours increases the odds of insufficient pain relief during labor (odds ratio, 9.1; P = 0.001).24 Durations are significantly greater among women with epidural catheter dislodgment (6.42 [SE, 0.83] vs 4.93 [SE, 0.13] hours; P = 0.002).25
Second, there were many hospitals with many durations of anesthesia providers’ billed time ≤60 minutes. We do not know why this was so. Most likely, an anesthesiologist was available, but management (evaluation of the patient, changing of syringes, etc.) was done by a registered nurse.c Regardless of the reason, providing a fixed amount of money to the group for managing the labor epidural, regardless of the billed duration, does not result in equitable payment. It also results in an incentive for anesthesia providers to spend less time managing the patients’ analgesia and peripartum care.
From Abouleish et al.,9 the mean face-to-face time was 1.99 hours (SE, 0.06). At Duke, the mean was 1.50 hours (SE, 0.09). These compare with our findings that the mean duration of labor analgesia was 6.12 hours (SE, <0.001 hours) (and 6.87 hours [SE, 0.70] at Duke10). As nicely explained by Abouleish et al.,9 anesthesia care for labor analgesia requires “continuous availability,” rather than “continuous attendance.”9 Thus, there is suitability of payment using base units and time, but the time units calculated not as 1 unit per 15 minutes (like for surgery) but a longer period such as 60 minutes.b,j Using the preceding numbers, the smallest suitable ratio would be 1 unit for each 45 minutes, since (6.12 hours for labor analgesia)/(1.99 hours face-to-face time9) ≅ 3. The largest suitable ratio would be 1 unit for each 75 minutes, since (6.87 hours for labor analgesia10)/(1.50 hours face-to-face time10) ≅ 5.
Our conclusions have equal relevance to a bundled payment for obstetrical care (i.e., a single fee-for-service payment to hospital, physicians, and advanced practice nurses). If each hospital were to receive the same bundled payment for routine obstetrical care, payment among hospitals would not be equitable even when averaged among many women. Equitable payment is the foundational principle of relative value unit payment, which for anesthesia in the United States means use of the American Society of Anesthesiologists’ Relative Value Guide.b
We determined that the mean durations of labor analgesia for vaginal delivery differ by a managerially important amount (>1 hour) among hospitals. There were many hospitals with data not analyzed. However, adding more hospitals cannot change that conclusion. The very large data set of AQI was sufficient to answer this question of relevance to anesthesia fee-for-service payment.k
We could not address quantification of the heterogeneity among hospitals in the variability of durations of analgesia. Because the durations follow gamma distributions (Figs. 1 and 2), such variability would be quantified using the coefficient of variation. Such variability may be important clinically and managerially.22 However, national analyses would need to be performed with batching (binning) or an equivalent method (e.g., sparse sampling, as for Figs. 1 and 2) because of correlations among patients. The sample sizes are insufficient for many 4-week periods and hospitals to perform such modeling (Fig. 6). Perhaps future research could approach this problem by combining models of biological variability1–4 both with clinician and patient decision making and administrative processes (e.g., case scheduling and staff assignment).
We could not address what is the US national overall mean duration of labor analgesia. The mean of 6.12 hours in Figure 7 is the mean among the hospitals studied.
Finally, we have not studied what causes the heterogeneity of the mean durations of labor analgesia among hospitals. In the Introduction, we discuss several biological factors1–3 that are unbalanced among hospitals (e.g., those of Figs. 1 and 2). However, there may also be a cultural explanation for the heterogeneity. Active management of labor with membrane rupture and augmentation with oxytocin is practiced at many but not all centers in the United States.5 Parturients who wish to have a more natural, less painful, labor experience may reject active management of labor.4 Although the AQI currently lacks both patient factor data and relevant data on obstetric management, American Society of Anesthesiologists and American College of Obstetricians and Gynecologists have recently announced the Maternal Quality improvement Project to address this deficiency.l
Using data from the American Society of Anesthesiologists’ AQI, we showed that the mean durations of labor analgesia for vaginal delivery have large heterogeneity among US hospitals. This is unlikely to reflect solely differences in the physiology of labor but also differences in local obstetrical practice among hospitals (e.g., when labor epidurals are requested and/or supplied) and/or heterogeneity of women’s demographics among the hospitals (e.g., different incidences of Asian women or obese women).1–5 An implication is that the number of labor epidurals alone is not a valid measure to be used as part of the quantification of the productivity of an anesthesia department. Equitable payment would not be based solely on the number of labor epidurals for vaginal delivery. Our results complement the prior work of Abouleish et al.9 that showed that payment for labor analgesia based on face-to-face time also is neither valid nor equitable. The use of base and time units,b with one time unit per hour, is a suitable payment system.
One sensitivity analysis that we performed was to treat each hospital’s mean duration not as the mean among each of the hospital’s 4-week periods’ means (i.e., the mean of means) but as the unweighted mean of the individual durations of labor analgesia for vaginal delivery at the hospital. The SE of the mean and the 99.9% confidence interval still was calculated using the SD among the means of each 4-week periods. Because the sample sizes are relatively homogeneous among 4-week periods, we expected the findings to be the same, and they were. Slightly more (56.4%) of the 172 hospitals had mean durations that differed (P < 0.001) from the overall mean of 6.12 hours. However, among those 56.4%, there were the same 9.9% of hospitals with means ≤5.12 hours. There also were the same 12.2% of hospitals with means ≥7.12 hours. The mean percentage difference of the estimates of each hospital’s mean was 0.20% (SE, 0.07%). The mean percentage absolute difference was 0.54% (SE, 0.06%).
The second sensitivity analysis was to use the central limit theorem by case, rather than batching among 4-week periods, when calculating the SE of the mean for each hospital. In other words, we improve the precision of estimates by having a greater effective sample size but may cause bias by neglecting heterogeneity in durations within hospitals (e.g., because of the day of the week). The data from Figure 1 follow a Gamma distribution.m Using SYSTAT 13.1 (Systat Software, Inc.), we performed 100,000 Monte-Carlo simulations, each drawing samples of size 200 from the gamma distribution shown in red in Figure 1. The reason for using sample sizes of 200 was that this was the minimum number of labor epidurals, as above. The percentage of simulations for which the asymptotic confidence interval did not overlap with the distribution’s true mean was 5.46% (SE, 0.07%) for P < 0.05, 1.30% (SE, 0.04%) for P < 0.01, and 0.21% (SE, 0.01%) for P < 0.001. The agreement between the expected and the actual coverage was quite close. Applying the central limit theorem to the Anesthesia Quality Institute data, more (64.5%) of the hospitals had mean durations that differed (P < 0.001) from the overall mean than in the Results.n Among these, 64.5% of hospitals were the same 9.9% with means ≤5.12 hours. Among the 64.5% were 13.3% of hospitals with means ≥7.12 hours.
Name: Pamela Flood, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript and is the archival author for the data from Columbia University and the University of California at San Francisco.
Attestation: Pamela Flood has approved the final manuscript.
Name: Franklin Dexter, MD, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Franklin Dexter has approved the final manuscript.
Name: Johannes Ledolter, PhD.
Contribution: This author helped conduct the study.
Attestation: Johannes Ledolter has approved the final manuscript.
Name: Richard P. Dutton, MD, MBA.
Contribution: This author helped conduct the study and is the archival author for the data from AQI.
Attestation: Richard P. Dutton approved the final manuscript.
The authors appreciate the assistance of Antonio Sepulveda in preparing some of the data.
Dr. Pamela Flood is the wife of Dr. Steven Shafer, Editor-in-Chief of Anesthesia & Analgesia. This manuscript was handled by Dr. James G. Bovill, Guest Editor-in-Chief, and Dr. Shafer was not involved in any way with the editorial process or decision. Dr. Dexter is the Section Editor for Economics, Education, and Policy for Anesthesia & Analgesia. This manuscript was handled by Dr. James G. Bovill, Guest Editor-in-Chief, and Dr. Dexter was not involved in any way with the editorial process or decision.
a The AQI is a nonprofit corporation created by the ASA. One of its data warehouses is the National Anesthesia Clinical Outcomes Registry.6 Electronic data specific to anesthesia cases are captured automatically (e.g., through monthly reports from nearly all major US anesthesia billing companies, quality management systems, and anesthesia information management systems). Anesthesia practices participate under business associate agreements permitting the transfer of deidentified health care information for facilitation of quality improvement.
b American Society of Anesthesiologists, 2014 Relative Value Guide, A Guide for Anesthesia Values, page xii, “Unlike operative anesthesia services, there is no single, widely accepted method of accounting for time for neuraxial labor anesthesia services … Methods … include: Base units plus time reported in minutes (insertion through delivery) subject to a reasonable cap …; Base units plus one unit per hour … for neuraxial anesthesia service management plus direct patient contact time (insertion … removal); … Single fee.” The base units represent initial activity, including evaluation and setup time, for placement of the labor epidural.
c http://FDshort.com/AWHONNstatement, accessed November 3, 2014.
d The means were negligibly related to the number of labor epidurals. By hospital (n = 172), Kendall’s tau = 0.073 (SE, 0.052). By hospital and 4-week period (n = 5671), Kendall’s tau = 0.034 (SE, 0.001).
e There were 320 hospitals each with at least 200 durations. There was significant (Kruskal-Wallis) heterogeneity among days of the week for 10.6% of hospitals at P < 0.01 and 21.6% at P < 0.05. There were 172 hospitals each with at least 200 durations, at most 5.0% of durations lasting 60 minutes or less, and at least 5 four-week periods with 14 durations. There was heterogeneity among weekdays for 11.0% at P < 0.01 and 19.2% at P < 0.05.
f The times of the labor epidurals have associated time zones.7,8 Among the n = 172 hospitals, there are 41.9% in the Eastern time zone, 33.1% Central, 8.7% Mountain, and 16.3% Pacific or Hawaii-Aleutian.
g The 10th, 20th, 25th, 50th, 75th, 80th, and 90th percentiles are 7.0, 9.5, 10.5, 20.8, 34.3, 37.3, and 45.8 per week.
h Texas Medicaid requires the use of 01960, 01961, 01963, 01967, 01968, and 01969 for obstetrical anesthesia. “Procedure codes 01960 and 01967 … are reimbursed a flat fee. The time reported must be in minutes and must represent the total minutes between the start and stop times for these procedures, regardless of the time actually spent with the client. Providers are not required to report actual face-to-face minutes with the client for these procedure codes … Procedure code 01968 or 01969 may be considered for reimbursement when submitted with procedure code 01967. For a Cesarean delivery following a planned vaginal delivery, the anesthesia administered during labor must be billed with …,” http://FDshort.com/TexasMedicaidLabor, accessed November 3, 2014.
i http://FDshort.com/IowaMedicaid01967 and http://FDshort.com/IowaMedicaidLetter1444, accessed November 15, 2014.
j http://FDshort.com/BlueCrossBlueShieldLouisiana, accessed November 23, 2014.
k Formally, for heterogeneity in the mean durations of labor analgesia among hospitals to cause inequality in payment, there needs to be lack of relationship between payment per minute for labor analgesia and mean duration. There are no data suggesting any such relationship including from the American Society of Anesthesiologists (B), 2 states (H, I), or commercial insurer (J).
l http://FDshort.com/AQIannouncement and http://FDshort.com/ACOGannouncement, accessed January 24, 2015.
m http://en.wikipedia.org/wiki/Gamma_distribution, accessed November 9, 2014.
n We trust the more conservative estimates with SEs calculated using 4-week periods as in the Results, as those SEs have been adjusted for temporal heterogeneity in the data.
1. Miller RS, Smiley RM, Daniel D, Weng C, Emala CW, Blouin JL, Flood PD. Beta-2 adrenoceptor genotype and progress in term and late preterm active labor. Am J Obstet Gynecol. 2011;205:137
2. Debiec J, Conell-Price J, Evansmith J, Shafer SL, Flood P. Mathematical modeling of the pain and progress of the first stage of nulliparous labor. Anesthesiology. 2009;111:1093–110
3. Terkawi AS, Jackson WM, Thiet MP, Hansoti S, Tabassum R, Flood P. Oxytocin and catechol-O-methyltransferase receptor genotype predict the length of the first stage of labor. Am J Obstet Gynecol. 2012;207:184.e1–8
4. Conell-Price J, Evans JB, Hong D, Shafer S, Flood P. The development and validation of a dynamic model to account for the progress of labor in the assessment of pain. Anesth Analg. 2008;106:1509–15
5. Pates JA, Satin AJ. Active management of labor. Obstet Gynecol Clin North Am. 2005;32:221–30
6. Dutton RP, Dukatz A. Quality improvement using automated data sources: the anesthesia quality institute. Anesthesiol Clin. 2011;29:439–54
7. Dexter F, Rosenberg H, Epstein RH, Semo JJ, Litman RS. Implications of national anesthesia workload on the staffing of a call center: the malignant hyperthermia consultant hotline. A&A Case Reports. 2015;5:43–6
8. Dexter F, Dutton RP, Kordylewski H, Epstein RH. Anesthesia workload nationally during regular workdays and weekends. Anesth Analg. 2015 [epub ahead of print]
9. Abouleish AE, Prough DS, Vadhera RB. Influence of the type of anesthesia provider on costs of labor analgesia to the Texas Medicaid program. Anesthesiology. 2004;101:991–8
10. Bell ED, Penning DH, Cousineau EF, White WD, Hartle AJ, Gilbert WC, Lubarsky DA. How much labor is in a labor epidural? Manpower cost and reimbursement for an obstetric analgesia service in a teaching institution. Anesthesiology. 2000;92:851–8
11. Law AM, Kelton WD Simulation Modeling and Analysis. 19912nd ed. New York, NY McGraw-Hill, Inc.:551–3
12. Dexter F, Macario A, Qian F, Traub RD. Forecasting surgical groups’ total hours of elective cases for allocation of block time: application of time series analysis to operating room management. Anesthesiology. 1999;91:1501–8
13. Dexter F, Marcon E, Epstein RH, Ledolter J. Validation of statistical methods to compare cancellation rates on the day of surgery. Anesth Analg. 2005;101:465–73
14. Dexter F, Epstein RH, Marcon E, Ledolter J. Estimating the incidence of prolonged turnover times and delays by time of day. Anesthesiology. 2005;102:1242–8
15. Dexter F, Marcon E, Aker J, Epstein RH. Numbers of simultaneous turnovers calculated from anesthesia or operating room information management system data. Anesth Analg. 2009;109:900–5
16. Ledolter J, Dexter F, Epstein RH. Analysis of variance of communication latencies in anesthesia: comparing means of multiple log-normal distributions. Anesth Analg. 2011;113:888–96
17. Dexter F. High-quality operating room management research. J Clin Anesth. 2014;26:341–2
18. Austin TM, Lam HV, Shin NS, Daily BJ, Dunn PF, Sandberg WS. Elective change of surgeon during the OR day has an operationally negligible impact on turnover time. J Clin Anesth. 2014;26:343–9
19. Dexter F, Macario A. Optimal number of beds and occupancy to minimize staffing costs in an obstetrical unit? Can J Anaesth. 2001;48:295–301
20. Dexter F, Epstein RH, Traub RD, Xiao Y. Making management decisions on the day of surgery based on operating room efficiency and patient waiting times. Anesthesiology. 2004;101:1444–53
21. Birnbach DJ, Bucklin BA, Dexter F. Impact of anesthesiologists on the incidence of vaginal birth after cesarean in the United States: role of anesthesia availability, productivity, guidelines, and patient safety. Semin Perinatol. 2010;34:318–24
22. Dexter F, Epstein RH, Bayman EO, Ledolter J. Estimating surgical case durations and making comparisons among facilities: identifying facilities with lower anesthesia professional fees. Anesth Analg. 2013;116:1103–15
23. Pan PH, Bogard TD, Owen MD. Incidence and characteristics of failures in obstetric neuraxial analgesia and anesthesia: a retrospective analysis of 19,259 deliveries. Int J Obstet Anesth. 2004;13:227–33
24. Le Coq G, Ducot B, Benhamou D. Risk factors of inadequate pain relief during epidural analgesia for labour and delivery. Can J Anaesth. 1998;45:719–23
25. D’Angelo R, Berkebile BL, Gerancher JC. Prospective examination of epidural catheter insertion. Anesthesiology. 1996;84:88–93