Advertisement

Rates of labor induction without medical indication are overestimated when derived from birth certificate data

  • Jennifer L. Bailit
    Affiliations
    Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, and the Center for Health Care Research and Policy, MetroHealth Medical Center, Case Western Reserve University, Cleveland, OH
    Search for articles by this author
  • Ohio Perinatal Quality Collaborative

      Objective

      The purpose of this study was to determine the rates of late preterm inductions without a medical indication from birth certificate data and to compare them with rates that were obtained from medical charts.

      Study Design

      The Ohio Perinatal Quality Collaborative, which comprises 20 hospitals in Ohio that came together in 2008 for the purpose of decreasing nonmedically indicated scheduled deliveries, abstracted data on all scheduled births between 36 weeks and 38 weeks 6 days of gestation. We compared labor inductions with “elective” documented or no indication documented in charts to birth certificate data for inductions with no maternal or fetal complications recorded.

      Results

      Birth certificates overestimate rates of induction without medical indication compared with chart abstraction (11% vs 1%; P < .0001). The monthly difference between chart abstraction and birth certificates averages 10.1%.

      Conclusion

      Birth certificates overestimate nonmedically indicated inductions by 11-fold. Until birth certificate data improve, nonmedically indicated induction rates that are calculated from birth certificates should be interpreted with caution.

      Key words

      Nonmedically indicated induction of labor, when labor is induced for the convenience of the patient or doctor, is rising.
      • Zhang J.
      • Yancey M.K.
      • Henderson C.E.
      US national trends in labor induction, 1989-1998.
      Estimates on the prevalence of nonmedically indicated induction vary greatly. A large series from a single center estimated that 6.3% of all inductions (or 2.7% of all deliveries) were not medically indicated.
      • Yeast J.D.
      • Jones A.
      • Poskin M.
      Induction of labor and the relationship to cesarean delivery: a review of 7001 consecutive inductions.
      However, a population-based study from Washington State showed that 33% of all women had an induction of labor; of those women, 85.7% had a standard reason for induction of labor, which left 14.3% (4.7% of the overall population) with either no reason listed or a nonstandard reason for induction.
      • Lydon-Rochelle M.T.
      • Caìrdenas V.
      • Nelson J.C.
      • Holt V.L.
      • Gardella C.
      • Easterling T.R.
      Induction of labor in the absence of standard medical indications: incidence and correlates.
      For Editors' Commentary, see Table of Contents
      See related editorial, page 190
      Review of medical records is the gold standard for the evaluation of whether a patient had an nonmedically indicated or an indicated induction. However, chart reviews are work intensive and are limited generally to 1 medical center, which minimizes the generalizability of the findings. Birth certificate data offer the advantage of being population based and thus more generalizable, but the quality of the clinical data is suspect. In general, complications on birth certificate data are thought to be specific but not sensitive.
      • Piper J.
      • Mitchel E.
      • Snowden M.
      • Hall C.
      • Asdams M.
      • Taylor P.
      Validation of 1989 Tennessee birth certificates using maternal and newborn hospital records.
      • Romano P.S.
      • Yasmeen S.
      • Schembri M.E.
      • Keyzer J.M.
      • Gilbert W.M.
      Coding of perineal lacerations and other complications of obstetric care in hospital discharge data.
      • Lydon-Rochelle M.T.
      • Holt V.L.
      • Nelson J.C.
      • et al.
      Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records.
      The Ohio Perinatal Quality Collaborative (OPQC) is a group of 20 hospitals in Ohio that came together in 2008 for the purpose of decreasing the rate of nonmedically indicated scheduled deliveries at <39 weeks of gestation.
      • Donovan E.F.
      • Lannon C.
      • et al.
      Ohio Perinatal Quality Collaborative Writing Committee
      A statewide initiative to reduce inappropriate scheduled births at 36(0/7)-38(6/7) weeks' gestation.
      As part of this effort, we performed chart abstractions on all scheduled deliveries between 360/7 and 386/7 weeks of gestation. Birth certificate data from these same hospitals were also tracked during this time. We sought to determine the magnitude of the difference between chart abstracted data and birth certificate data in estimating nonmedically indicated induction rates. If the magnitude of the difference is known, birth certificate data, which is cheap and easily available, can be better interpreted. This work is a secondary analysis of OPQC data.

      Materials and Methods

      OPQC collected data on all scheduled deliveries from 360/7 to 386/7 from the 20 hospitals in OPQC from August 2008 through June 2009. Hospitals that participated in OPQC included all of the level-3 centers (tertiary) in Ohio and many of the level-2 centers. Level-1 centers are not included in these data. These 20 OPQC hospitals represent 47% of the deliveries in Ohio during those years. This project significantly lowered the rate of nonmedically indicated scheduled deliveries from 25% to <5% during the course of the project.
      Scheduled deliveries were determined from labor and delivery schedules and review of charts. We limited our sample to those women who had a scheduled induction of labor. Because there is not a gold standard of legitimate medical reasons for the inducement of labor, to be conservative, we counted all indications as medically indicated, unless they had no reason that was documented for the induction or “elective” delivery was documented as a reason for the induction. Thus, some of the induction reasons that we considered indicated were not medically standard. An example of a nonmedically standard indication might be “lives far from hospital.” Although the definition of nonmedically indicated induction can be debated, by defining it in this way, our definition was as conservative as possible and gave the full benefit of the doubt to the provider's reasoning and judgment.
      Birth certificate data from OPQC hospitals was obtained from the state repository for women who underwent an induction of labor between 36 and 38 completed weeks of gestation. An induction for the purpose of documentation on the birth certificate is when contractions are started medically or mechanically before the onset of spontaneous labor. There is no specific birth certificate item for reason for induction. Thus, we categorized the birth certificates as a nonmedically indicated induction if the following were absent from the birth certificate: a baby with a birthweight <10th percentile for gestational age, diabetes mellitus, prepregnancy hypertension, gestational hypertension, hypertension eclampsia, history of poor pregnancy outcomes, premature rupture of membranes, augmentation of labor, chorioamnionitis, and fetal anomalies. Unfortunately, not all fields that we ideally would have liked to include as medical reasons for induction are listed on the birth certificate. These items were chosen from the available fields on the Ohio birth certificate.
      We compared rates of nonmedically indicated induction that were calculated by the OPQC data and compared them to rates of nonmedically indicated induction that were calculated by the birth certificate using a Student t test. The total number of monthly deliveries (the denominator) was obtained from birth certificate data for both rates because birth certificate data are known to be complete for the population. We also describe the differences by data type in the rates month by month. Last, we looked at the magnitude of the difference between the OPQC data and the birth certificate data within each of the 20 centers to assess the range of differences between the data types and to assess whether a few large centers were driving the findings.

      Results

      There were 234 nonmedically indicated inductions that were identified by OPQC data and 2086 nonmedically indicated inductions that were identified by birth certificate data. Of the 234 nonmedically indicated induction in OPQC data, 75% of inductions were documented as “elective,” and 25% of inductions had no documentation of induction reason. Birth certificate data identified 19,266 deliveries between 36 and 38 weeks of gestation during this time. When we used the average hospital rates for each, birth certificates overestimated nonmedically indicated induction rates, compared with chart abstraction (11% vs 1%; P < .0001).
      The monthly difference between chart abstraction and birth certificates averages 10.1% (Figure), with a range of 0–34.6% at individual centers. Eight of the hospitals had average differences between nonmedically indicated induction rates from chart abstraction and birth certificate data of ≥10%. Only 2 hospitals had average differences between chart abstraction and birth certificate data of <5%.
      Figure thumbnail gr1
      FIGURERates of nonmedically indicated induction of labor that were calculated by birth certificate data vs chart abstracted data
      Bailit. Induction rates derived from birth certificate data. Am J Obstet Gynecol 2010.

      Comment

      Birth certificates overestimate nonmedically indicated deliveries 11-fold compared with chart abstraction. This is likely because birth certificates under report complications and thus over estimate nonmedical inductions. This finding is consistent with previous studies of birth certificate accuracy that show medical and obstetrics data are specific but not sensitive.
      • Piper J.
      • Mitchel E.
      • Snowden M.
      • Hall C.
      • Asdams M.
      • Taylor P.
      Validation of 1989 Tennessee birth certificates using maternal and newborn hospital records.
      • Lydon-Rochelle M.T.
      • Holt V.L.
      • Nelson J.C.
      • et al.
      Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records.
      • Parrish K.
      • Holt F.
      • Connell B.
      • Williams B.
      • LoGerfo J.
      Variations in the accuracy of obstetric procedures and diagnoses on birth records in Washington State, 1989.
      • Kahn E.B.
      • Berg C.J.
      • Callaghan W.M.
      Cesarean delivery among women with low-risk pregnancies: a comparison of birth certificates and hospital discharge data.
      In other words, if a pregnancy complication is recorded, it is likely to have really occurred. However, if a complication is missing, one cannot be sure that the event did not take place. Because we counted nonmedical inductions as those without complications that were listed, the type of birth certificate inaccuracy biases towards higher nonmedical induction rates. Until birth certificate data improves, nonmedical induction rates that are calculated from birth certificates should be interpreted with caution.
      Birth certificate data quality varies substantially by maternal characteristics.
      • Reichman N.E.
      • Schwartz-Soicher O.
      Accuracy of birth certificate data by risk factors and outcomes: analysis of data from New Jersey.
      Birth certificate data are collected differently at different hospitals.
      • Smulian J.C.
      • Ananth C.V.
      • Hanley M.L.
      • Knuppel R.A.
      • Donlen J.
      • Kruse L.
      New Jersey's electronic birth certificate program: variations in data sources.
      Our study suggests that the quality of the data varies by hospital. It is also possible that data quality varies by state. We did not collect data on the method of birth certificate data collection at the participating hospitals and thus cannot comment on which methods of collection are more accurate. Future studies should be done about the most accurate methods of data collection both in the types of data sources and personnel training level.
      With the advent of electronic medical records, the possibility of birth certificate data improvement is greater than ever. The technical ability to extract data from medical records electronically and download it into a central repository from multiple sites has been done by researchers.
      • Bailit J.L.
      • Gregory K.D.
      • Reddy U.M.
      • et al.
      Maternal and neonatal outcomes by labor onset type and gestational age.
      Now that the technology exists, it is possible to envision a statewide system in which data from medical records are downloaded to the state vital statistics department. Although electronic medical records are not yet in all hospitals and states may not yet be set up to receive data in this manner, it is easy to speculate that this is what the future may hold.
      Until the time that birth certificates become more accurate, our study helps to establish the magnitude of the difference between medical records data and birth certificate data. Birth certificate data has the advantages of being cheap to use, much cheaper certainly that abstracting data from medical records by hand. Because birth certificate data are population based, it is an appealing source for the assessment of trends in obstetrics care. As birth certificate data becomes more widely used to monitor obstetric care, especially in states in which birth certificate data are used to measure hospital quality and are made available to the public and insurers by the State, hospitals will have the impetus to improve their data quickly. Thus, although birth certificate data are not currently an accurate measure of nonmedically indicated inductions, the data will probably improve over time. Studies such as ours need to be repeated to recalibrate the magnitude of difference between hand-collected data and vital statistics data as birth certificate data collection methods improve.
      Ohio birth certificate data overestimate nonmedical inductions 11-fold. Birth certificate data on nonmedical induction of labor should be interpreted with caution. Further studies are needed to determine whether the magnitude of difference is similar in other states and over time.

      Acknowledgments

      The Ohio Perinatal Quality Collaborative writing committee: Jennifer L. Bailit, MD, MPH, MetroHealth Medical Center, Cleveland, OH; Carol Lannon, MD, Barbara Rose, RN, and Edward F. Donovan, MD, MPH, Cincinnati Children's Medical Center, Cincinnati, OH; Jay Iams MD, The Ohio State University College of Medicine, Columbus, OH.

      References

        • Zhang J.
        • Yancey M.K.
        • Henderson C.E.
        US national trends in labor induction, 1989-1998.
        J Reprod Med. 2002; 47: 120-124
        • Yeast J.D.
        • Jones A.
        • Poskin M.
        Induction of labor and the relationship to cesarean delivery: a review of 7001 consecutive inductions.
        Am J Obstet Gynecol. 1999; 180: 628-633
        • Lydon-Rochelle M.T.
        • Caìrdenas V.
        • Nelson J.C.
        • Holt V.L.
        • Gardella C.
        • Easterling T.R.
        Induction of labor in the absence of standard medical indications: incidence and correlates.
        Med Care. 2007; 45: 505-512
        • Piper J.
        • Mitchel E.
        • Snowden M.
        • Hall C.
        • Asdams M.
        • Taylor P.
        Validation of 1989 Tennessee birth certificates using maternal and newborn hospital records.
        Am J Epidemiol. 1993; 137: 758-768
        • Romano P.S.
        • Yasmeen S.
        • Schembri M.E.
        • Keyzer J.M.
        • Gilbert W.M.
        Coding of perineal lacerations and other complications of obstetric care in hospital discharge data.
        Obstet Gynecol. 2005; 106: 717-725
        • Lydon-Rochelle M.T.
        • Holt V.L.
        • Nelson J.C.
        • et al.
        Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records.
        Pediatr Perinatal Epidemiol. 2005; 19: 460-471
        • Donovan E.F.
        • Lannon C.
        • et al.
        • Ohio Perinatal Quality Collaborative Writing Committee
        A statewide initiative to reduce inappropriate scheduled births at 36(0/7)-38(6/7) weeks' gestation.
        Am J Obstet Gynecol. 2010; 202: 243.e1-243.e8
        • Parrish K.
        • Holt F.
        • Connell B.
        • Williams B.
        • LoGerfo J.
        Variations in the accuracy of obstetric procedures and diagnoses on birth records in Washington State, 1989.
        Am J Epidemiol. 1993; 138: 119-127
        • Kahn E.B.
        • Berg C.J.
        • Callaghan W.M.
        Cesarean delivery among women with low-risk pregnancies: a comparison of birth certificates and hospital discharge data.
        Obstet Gynecol. 2009; 113: 33-40
        • Reichman N.E.
        • Schwartz-Soicher O.
        Accuracy of birth certificate data by risk factors and outcomes: analysis of data from New Jersey.
        Am J Obstet Gynecol. 2007; 197: 32.e1-32.e8
        • Smulian J.C.
        • Ananth C.V.
        • Hanley M.L.
        • Knuppel R.A.
        • Donlen J.
        • Kruse L.
        New Jersey's electronic birth certificate program: variations in data sources.
        Am J Public Health. 2001; 91: 814-816
        • Bailit J.L.
        • Gregory K.D.
        • Reddy U.M.
        • et al.
        Maternal and neonatal outcomes by labor onset type and gestational age.
        Am J Obstet Gynecol. 2010; 202: 245.e1-245.e12