OBJECTIVE: To determine the error rate for discharge coding of anal sphincter laceration at vaginal delivery in a cohort of primiparous women.
METHODS: As part of the Childbirth and Pelvic Symptoms study performed by the National Institutes of Health Pelvic Floor Disorders Network, we assessed the relationship between perineal lacerations and corresponding discharge codes in three groups of primiparous women: 393 women with anal sphincter laceration after vaginal delivery, 383 without anal sphincter laceration after vaginal delivery, and 107 after cesarean delivery before labor. Discharge codes for perineal lacerations were compared with data abstracted directly from the medical record shortly after delivery. Patterns of coding and coding error rates were described.
RESULTS: The coding error rate varied by delivery group. Of 393 women with clinically recognized and repaired anal sphincter lacerations by medical record documentation, 92 (23.4%) were coded incorrectly (four as first- or second-degree perineal laceration and 88 with no code for perineal diagnosis or procedure). One (0.3%) of the 383 women who delivered vaginally without clinically reported anal sphincter laceration was coded with a sphincter tear. No women in the cesarean delivery group had a perineal laceration diagnostic code. Coding errors were not related to the number of deliveries at each clinical site.
CONCLUSION: Discharge coding errors are common after delivery-associated anal sphincter laceration, with omitted codes representing the largest source of errors. Before diagnostic coding can be used as a quality measure of obstetric care, the clinical events of interest must be appropriately defined and accurately coded.
LEVEL OF EVIDENCE: II
Hospital discharge coding did not identify nearly one quarter of women with anal sphincter lacerations at vaginal delivery.
From the 1Department of Obstetrics and Gynecology, Loyola University Medical Center, Maywood, Illinois; 2Department of Obstetrics and Gynecology, University of Iowa, Iowa City, Iowa; 3Department of Obstetrics and Gynecology, Johns Hopkins University, Baltimore, Maryland; 4Department of Obstetrics and Gynecology, University of Alabama at Birmingham, Birmingham, Alabama; 5Department of Obstetrics and Gynecology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; 6Department of Biostatistics, University of Michigan, Ann Arbor, Michigan;. and 7National Institute of Child Health and Human Development, Bethesda, Maryland.
Supported by grants from the National Institute of Child Health and Human Development (U01 HD41249, U10 HD41268, U10 HD41248, U10 HD41250, U10 HD41261, U10 HD41263, U10 HD41269, and U10 HD41267).
Corresponding author: Linda Brubaker, MD, MS, Loyola University Chicago, 2160 South First Avenue, Maywood, IL 60153; e-mail: Lbrubaker@lumc.edu.
Financial Disclosure The authors have no conflicts of interest relevant to this article.
Anal sphincter lacerations are reported in 5–10% of deliveries in the United States.1–3 Anal sphincter lacerations at vaginal delivery have become an indicator for patient safety and quality of care, with monitoring by the Agency for Healthcare Research and Quality (AHRQ), a federal agency within the Department of Health and Human Services that is charged with supporting health services research to improve the quality of health care and to promote evidence-based decision-making.4 Anal sphincter lacerations at vaginal delivery, along with rates of vaginal birth after cesarean and inpatient neonatal mortality, are important components for a new core measure of obstetric care quality adopted by the Joint Commission on Accreditation of Healthcare Organizations (JCAHO).5 The AHRQ and JCAHO rely on hospital discharge coding to obtain their data.
In addition to national monitoring of patient safety and quality of care, studies of anal sphincter lacerations often use hospital discharge coding to identify these lacerations and other obstetric events because of the potential morbidity and subsequent pelvic floor effect. However, the accuracy of administrative data sets is imperfect, as reported in studies of total hip replacement,6 hip fracture,7 acute myocardial infarction,8 cardiac bypass,9 and hospital-reported complications.10 Data based on diagnostic codes may be less accurate than procedure-based data.6 Coding accuracy probably varies between institutions; errors in coding occur as a result of incomplete documentation of clinical information on the chart, physician misdiagnosis, or coders' miscoding or incomplete coding of diagnoses and procedures.11
Although studies exist that examine the accuracy of coding for various medical conditions, only one publication describes the accuracy of discharge coding for anal sphincter laceration and repair at vaginal delivery.12 This study reported that discharge coding was more than 90% sensitive and 97% specific. However, the study obtained data from a single state, was limited by a small number of anal sphincter lacerations (n=64), and used recoding of hospital charts by expert coders as the standard for comparison. The objective of this report was to describe whether hospital-coded discharge diagnoses of anal sphincter injury accurately captured perineal lacerations, particularly anal sphincter injury, using data prospectively collected from a cohort study conducted by the Pelvic Floor Disorders Network.
MATERIALS AND METHODS
Between September 2002 and September 2004, primiparous women after delivery were enrolled into one of three groups for an National Institutes of Health (NIH)–sponsored cohort study to evaluate the relationship between childbirth and pelvic symptoms. The three groups included women with clinically diagnosed anal sphincter laceration (third- or fourth-degree perineal laceration) and repair after vaginal delivery (anal sphincter tear group); women without a clinically diagnosed anal sphincter laceration (no laceration, or first- or second-degree perineal laceration) after vaginal delivery (vaginal control group); and women delivered by cesarean without labor (cesarean control group). This prospective study was conducted at nine hospitals, including seven university hospitals and two community hospitals. institutional review board approval was obtained at all nine hospitals.
To identify the diagnosis of anal sphincter laceration, trained research nurses examined medical records, including delivery notes. All but one study hospital were using paper, not electronic, medical records during the study period. We defined anal sphincter laceration as any third- or fourth-degree perineal laceration diagnosed at delivery. Medical records were evaluated while patients were hospitalized. The clinical record was used as the standard, because this is the source document for hospital discharge coding.
This ancillary study to the Childbirth and Pelvic Symptoms study was designed after primary data collection from the hospitals had been completed. Therefore, the administrative discharge coding was completed by coders who were unaware of this study. Coders handled any discrepancies between nursing and physician notes per their usual process without input from research staff with the Childbirth and Pelvic Symptoms study. After all participants were discharged, we collected the hospital discharge codes from each medical record. We then compared the coded diagnoses to the clinical record regarding perineal lacerations, particularly anal sphincter laceration. Table 1 lists the diagnostic codes used in this study.
The error rates of coded diagnoses were calculated using the clinical documentation of anal sphincter laceration as the standard. With close to 400 subjects per group, the 95% confidence interval (CI) for the estimates of the error rates is no greater than ±5%. When the estimate is 10% or less, the 95% CI is no greater than 3%. Confidence intervals were estimated using the binomial distribution.
Diagnostic codes were obtained for 883 of the 907 subjects in the Childbirth and Pelvic Symptoms study: 393 women who had clinically diagnosed anal sphincter lacerations (third- or fourth-degree perineal tears), 383 women in the vaginal control group without clinically diagnosed anal sphincter tears, and 107 women in the cesarean control group. None of the women in the cesarean control group was coded as having an anal sphincter tear or any type of perineal laceration. This group was not used for further comparisons.
The majority of women with a clinically recognized anal sphincter tear (301 of 393, [76.6%]) had coding that included either the 664.21 (third degree) or 664.31 (fourth degree) diagnostic codes for perineal lacerations (Table 2). An additional four (1.0%) were coded as either a first- or second-degree perineal laceration. In the remaining 88 (22.4%) participants, no diagnostic code for perineal status or procedure was used. Therefore, the coding error rate for the anal sphincter tear group was 23.4% (92 of 393). In the sphincter tear group, the percent of coding errors for the eight hospitals that recruited more than 30 participants ranged from 0% to 62%; the error rate was not associated with the number of study deliveries. The one hospital with electronic records had an error rate of 18%.
Seventy-one of the 383 women in the vaginal control group were reported to have an intact perineum in the medical record; of these, 19 were coded as either first- or second-degree tear for an error rate of 26.8%. The remaining 312 women in the vaginal control group were classified in the medical record as having either a first- or second-degree tear. Of these 312, one was coded as having a third- or fourth-degree tear (0.3%), and 120 did not have any code, for an error rate of 38.5%.
Eleven subjects had more than one code associated with perineal laceration status. Duplicate coding occurred in six of 393 (1.5%) of the anal sphincter tear group and five of 383 (1.2%) of the vaginal control group. In the anal sphincter tear group, five subjects had codes for a third-degree tear (664.21) as well as a code for a first- or second-degree tear (664.01 or 664.11); one subject had codes for both third- and fourth-degree tears. In the vaginal control group, five subjects had codes for both first- and second-degree tears (664.01 and 664.11). In all cases, the higher code was used to classify the subject for data analyses.
The mean number of discharge diagnostic codes by site ranged from 2.9 to 7.8 for women with anal sphincter laceration compared with 2.5 to 7.2 for women without anal sphincter laceration. Coding error was not associated with the number of deliveries at each institution, or the number of hospital discharge codes for each individual.
Administrative discharge codes are increasingly used in health services research as a means for monitoring quality at local and national levels.4,5,13 With respect to pelvic floor disorders research, these databases can be used to identify a large representative sample of patients of interest (eg, patients with anal sphincter tear).14 Patient recall has been used in some studies of obstetric outcomes, although recall of obstetric events is poor,15 especially for anal sphincter injury. We found that nearly one of four hospital discharges associated with a third- or fourth-degree anal sphincter laceration was undercoded, suggesting that administrative data underestimates the occurrence of these sphincter lacerations, which limit the usefulness of these data as a quality-of-care metric.
Past studies have found other obstetric conditions are also inaccurately coded. One study reviewed medical records and administrative data related to indications for elective primary cesarean delivery in a single hospital and reported an overall sensitivity of 73% and specificity of 98.1% using International Classification of Diseases, 9th Revision–Clinical Modification diagnosis codes.16 Other studies that reviewed medical charts for cases identified through specific diagnosis codes have found that the codes' positive predictive values are low. In a study of discharge coding and uterine rupture, the positive predictive value of uterine rupture discharge codes was 39.8%.17 Discharge codes were not specific for uterine rupture and were not applied consistently over the 7 years studied. The positive predictive value of discharge codes for other pregnancy-related complications has also been reported to be low, for instance, 39% in the case of venous thromboembolism and 54% for preeclampsia, and to vary widely.18,19 In this study, we have not reported predictive values for anal sphincter laceration discharge codes, because predictive values vary by prevalence, and the prevalence of anal sphincter tears in our study was artificially high by design.
Beyond obstetrics, inaccurate coding has also been demonstrated in medical and surgical complication reporting.20 In contrast to our results, errors occurred frequently in coding complications for which there was no evidence in the medical record in 30% of medical admissions and 19% of surgical admissions.
Romano et al12 recently published the only other study investigating the accuracy of diagnosis and procedure codes for anal sphincter lacerations. These investigators reported high specificity and sensitivity. This study sampled cases from 52 hospitals in California, selected by a complex stratified random sampling procedure, and studied patient records and discharge data at each hospital over a 1–2-year period. Using two International Classification of Diseases, 9th Revision-Clinical Modification diagnosis codes (664.3x, any fourth-degree perineal laceration; and 664.2x, any third-degree perineal laceration) as well as one procedure code (75.62), they identified 64 cases of anal sphincter laceration; coding experts compared hospital discharge data with the medical records as the standard. Discrepant cases were then reviewed by two investigators. Hospital discharge data had a weighted sensitivity of 94% (90% CI, 83–98%) and specificity greater than 97% for anal sphincter laceration. These results led them to conclude that discharge data are an adequate substitute for medical record review in estimating the occurrence of anal sphincter lacerations. However, the error rate was not reported. The very wide CI is due to the small sample size; the lower bound of their CI is similar to the rate that was observed in our study.
Our study's strengths include the multicenter, multistate design and abstraction of data directly from the medical record. Due to the parent study (the Childbirth and Pelvic Symptoms study) design, we were able to include a large number of cases of third- or fourth-degree anal sphincter laceration, allowing us to estimate more precisely the coding error rates for the most clinically relevant patients. However, our results may not be generalizable to all hospitals providing obstetric care. It is possible that our results differ from other studies because of regional differences in coding training. As with many case-based quality-of-care studies, we used the medical record as a standard; however, it is possible that the medical record is also inaccurate with regard to anal sphincter lacerations and that some events may not be recorded in this manner or may have been missed by our abstraction process.
The occurrence of anal sphincter lacerations is one of the core quality indicators reviewed by AHRQ and JCAHO.4,5 Such quality indicators are typically abstracted from discharge codes without reviewing the clinical record. However, apparent differences in the rates of anal sphincter laceration could represent different coding practices between institutions, rather than a true difference in lacerations. Our data suggest that hospital discharge coding may underestimate the true occurrence of anal sphincter lacerations, although the coding error rate varied widely, from 0% to 62%, among the eight hospitals in our study.
The Joint Commission defines a third-degree perineal laceration as “a rupture or tear in the perineum involving the anal sphincter and rectovaginal septum”; a fourth-degree laceration is “a rupture or tear involving anal sphincter, rectovaginal septum and mucosa” (presumably anal or anorectal mucosa). These definitions may be consistent with definitions used by obstetric providers. Clinical terms such as “partial third degree lacerations” may not fall into such a coding scheme, leaving coding teams with uncertainty as to the correct diagnostic code. We suggest that recording of perineal status might be improved by implementing standardized definitions to describe the perineum after delivery.
In conclusion, our results call into question the validity of using hospital discharge data for the identification or monitoring of anal sphincter lacerations, because these data may underestimate the true occurrence of this complication. The importance of identifying anal sphincter tears has been clearly demonstrated, because these new mothers have higher rates of fecal and flatal incontinence.21 In keeping with the goal to recognize and reduce these adverse events, it is important to minimize discharge coding errors. Hospitals may have artificially low anal sphincter laceration rates secondary to coding errors, leading to erroneously favorable comparisons with institutions with more accurate coding practices.
1. Goldberg J, Hyslop T, Tolosa JE, Sultana C. Racial differences in severe perineal lacerations after vaginal delivery. Am J Obstet Gynecol 2003;188:1063–7.
2. Handa VL, Danielsen BH, Gilbert WM. Obstetric anal sphincter lacerations. Obstet Gynecol 2001;98:225–30.
3. Richter HE, Brumfield CG, Cliver SP, Burgio KL, Neely CL, Varner RE. Risk factors associated with anal sphincter tear: a comparison of primiparous patients, vaginal births after cesarean deliveries, and patients with previous vaginal delivery. Am J Obstet Gynecol 2002;187:1194–8.
4. AHRQ Quality Indicators: guide to patient safety indicators. Version 2.1, Revision 3. Rockville (MD): Agency for Healthcare Research and Quality; 2005. Available at http://www.ahrq.gov
. Retrieved March 2, 2007.
5. Specification manual for national hospital quality measures (2005). Oakbrook Terrace (IL): Joint Commission on Accreditation of Healthcare Organizations; 2004. Available at http://www.jointcomission.org
. Retrieved March 2, 2007.
6. Losina E, Barrett J, Baron JA, Katz JN. Accuracy of Medicare claims data for rheumatologic diagnoses in total hip replacement recipients. J Clin Epidemiol 2003;56:515–9.
7. Keeler EB, Khan KL, Bentow S. Assessing quality of care for hospitalized Medicare patients with hip fracture during coded diagnoses from the Medical Provider Analysis and Review files. Prepared for the Health Care Financing Administration. Santa Monica (CA): Rand Corporation; 1992. Available at http://www.hcfa.org
. Retrieved March 2, 2007.
8. Kiyota Y, Schneeweiss S, Glynn RJ, Cannuscio CC, Avorn J, Solomon DH. Accuracy of Medicare claims-based diagnosis of acute myocardial infarction: estimating positive predictive value on the basis of review of hospital records. Am Heart J 2004;148:99–104.
9. Hartz AJ, Kuhn EM. Comparing hospitals that perform coronary artery bypass surgery: the effect of outcome measures and data sources. Am J Public Health 1994;84:1609–14.
10. Lawthers AG, McCarthy EP, Davis RB, Peterson LE, Palmer RH, Iezzoni LI. Identification of in-hospital complications from claims data. Is it valid? Med Care 2000;38:785–95.
11. Quan H, Parsons GA, Ghali WA. Validity of procedure codes in International Classification of Diseases, 9th revision, clinical modification administrative data. Med Care 2004;42:801–9.
12. Romano PS, Yasmeen S, Schembri ME, Keyzer JM, Gilbert WM. Coding of perineal lacerations and other complications of obstetric care in hospital discharge data. Obstet Gynecol 2005;106:717–25.
13. Main EK, Blookfield L, Hunt G, for the Sutter Health First Pregnancy and Delivery Clinical Initiative Committee. Development of a large-scale obstetric quality improvement program that focused on the nulliparous patient at term. Am J Obstet Gynecol 2004;190:1747–56.
14. Mendelsohn AB, Whittle J. Accuracy of coding for cardiac catheterization and percutaneous transluminal coronary angioplasty at a Department of Veterans Affairs Medical Center. J AHIMA 1996;67:64–70.
15. Elkadry E, Kenton K, White P, Creech S, Brubaker L. Do mothers remember key events during labor? Am J Obstet Gynecol 2003;189:195–200.
16. Korst LM, Gregory KD, Gornbein JA. Elective primary caesarean delivery: accuracy of administrative data. Paediatr Perinat Epidemiol 2004;18:112–9.
17. From the Centers for Disease Control and Prevention. Use of hospital discharge data to monitor uterine rupture—Massachusetts, 1990-1997. JAMA 2000;283:2098–100.
18. White RH, Brickner LA, Scannell KA. ICD-9-CM codes poorly identified venous thromboembolism during pregnancy. J Clin Epidemiol 2004;57:985–8.
19. Geller SE, Ahmed S, Brown ML, Cox SM, Rosenberg D, Kilpatrick SJ. International Classification of Diseases-9th revision coding for preeclampsia: how accurate is it? Am J Ostet Gynecol 2004;190:1629–33
20. McCarthy EP, Iezzoni LI, Davis RB. Palmer RH, Cahalane M, Hamel MB, et al. Does clinical evidence support ICD-9-CM diagnosis coding of complications? Med Care 2000;38:868–76.
21. Borello-France D, Burgio KL, Richter HE, Zyczynski H, Fitzgerald MP, Whitehead W, et al. Fecal and urinary incontinence in primiparous women. Obstet Gynecol 2006;108:863–72.