Traditional teaching states that there is an increased risk of miscarriage, stillbirth, and preterm labor when a pregnant woman undergoes surgery under general anesthesia during pregnancy.1 This risk has however never been quantified and such risks may have decreased with the use of modern anesthetic agents and techniques. At present, when counseling a pregnant woman on this risk, the information provided is not robustly evidence-based.
Previous research suggests that 1% to 2 % of women undergo nonobstetric surgery during pregnancy,2 common operations being appendicectomy, cancer surgery, and orthopedic procedures. Research has also suggested that nonobstetric surgery during pregnancy is associated with adverse birth outcomes: a Canadian study of 2565 pregnant women reported an increased risk of spontaneous abortion after a general anesthetic [relative risk (RR) = 1.58, 95% confidence interval (95% CI) 1.19–2.09];3 and a Swedish registry study showed associations between nonobstetric operations during pregnancy and low birth weight.4 A systematic review brought these and many smaller studies together.5 In this analysis of 54 papers dominated by abdominal surgery, no association was shown between nonobstetric surgery and major birth defects, but strong associations were observed between acute appendicitis with peritonitis and fetal loss. However, much of the available evidence was between 20 and 40 years old and therefore unlikely to be representative of the current situation; studies were generally small and poorly controlled and overall gave conflicting results.
We aimed to estimate the risk of adverse birth outcomes, including spontaneous abortion associated with hospitalization (miscarriage), stillbirth, preterm delivery, low birth weight, caesarean section, and maternal death, for women who underwent nonobstetric surgery during pregnancy in England compared with those who did not, using more recent routinely collected hospital administrative data.
We used the Hospital Episode Statistics (HES) database, which includes records of all admissions to English NHS hospital trusts. Each record contains data on patient demographics (eg, age, ethnicity, and socioeconomic deprivation based on postcode of residence), the episode of care (eg, hospital name, date of admission, and discharge), and clinical information.6,7 Diagnoses for each patient are recorded using the International Classification of Diseases, 10th edition (ICD-10). Procedures performed during an episode are coded using the Office of Population, Censuses, and Surveys Classification of Surgical Operations and Procedures, 4th revision (OPCS).8 Each episode relating to the delivery of a baby contains details about the obstetric history, labor, and delivery (eg, parity, mode of delivery, gestational age, birth weight) in supplementary data fields known as the HES “maternity tail.” In HES, each patient is assigned a unique identifier (HESID). This makes it possible to link historical medical records. We used all records collected between April 2002 and March 2012 inclusive. We searched for pregnancy records with procedure (OPCS 4) codes or “maternity tail” fields relating to delivery or with diagnosis (ICD-10) codes relating to spontaneous abortion associated with hospitalization. We restricted our sample to women aged between 15 and 49 years.9–12 If pregnancy ended in delivery, we estimated the beginning of pregnancy as admission day minus gestational age (if available) OR admission day minus 36 weeks (if gestational age not available, but diagnosis field indicates preterm delivery), OR admission day minus 40 weeks (otherwise). As there was no gestational age in the dataset on episodes that ended in spontaneous abortion, we looked back only 3 months for surgery to avoid inclusion of surgery occurring before pregnancy. Nonobstetric surgery during pregnancy was identified using a refined operating room procedure code list based on OPCS procedure codes defined in previous research.13
We determined 7 adverse birth outcomes based on both mothers’ and infants’ records. Spontaneous abortion associated with hospitalization was defined as a pregnancy episode for which one of the diagnoses was recorded as ICD-10 code of spontaneous abortion (O03) or other abortion (O05), or unspecified abortion (O06); this definition excluded planned surgical or medical termination of pregnancy. The definition is based on coding, but to all intents and purposes means miscarriage, and therefore, from now on, we will refer to it as miscarriage associated with hospitalization. Still birth was defined as a pregnancy episode for which one of the diagnoses was recorded as ICD-10 code of single or multiple births with all stillborn (Z31.1, Z37.4, Z37.7), multiple births with some stillborn (Z37.3, Z37.6), or birth status field in the maternity tail coded as “still birth.” We defined preterm delivery as a pregnancy episode for which one of the diagnoses was recorded as ICD-10 code of preterm delivery (O60) or the length of the gestation field in the HES maternity tail was lower than 37 weeks. We classified birth weight into 3 groups: one or more new-born with low birth weight (<2500 g), all new-born weighing 2500 g or more, and unknown. We identified cesarean section as a pregnancy episode wherein codes indicated an elective (R17) or other (R18) cesarean delivery or cesarean hysterectomy (R25.1), or delivery method field in the maternity tail coded as “elective caesarean section” or “emergency caesarean section.” Long inpatient stay was defined as the upper quartile of length of stay plus 1.5 times the interquartile range (with separate values for cesarean and vaginal delivery; 9.5 and 3.5 days, respectively).14 Maternal death was defined as a delivery or miscarriage associated with a hospitalization episode with method of discharge field coded as a death within the same admission stay.
We adjusted the risk of all outcomes of interest for maternal age, socioeconomic deprivation (Carstairs quintile),15 parity, previous cesarean section, previous emergency admission, comorbidities (Charlson score),16 multiple gestation, operations on amniotic cavity (including amniocentesis, sampling of chorionic villus), obstetric surgery, and for the presence of cardiac disease, hypertension/preeclampsia, and gestational diabetes17–24 (Supplement Table 1, http://links.lww.com/SLA/B89).
We used the adjusted odds ratio (ORs) obtained directly from a logistic regression model to estimate the RRs and its 95% CIs of adverse birth outcomes in pregnancies wherein surgery occurred compared with pregnancies wherein surgery did not occur.25,26 All pregnancy records were included in the analysis when miscarriage associated with hospitalization and maternal death were the outcomes of interest. For the remaining outcomes, only pregnancy records that ended in delivery were included. We used the logistic regression model to estimate the marginal probabilities of each outcome of interest (Austin's method).27 This method overcomes the issue in studies of common outcomes wherein the OR can overestimate the RR. This permitted us to estimate the adjusted RR, the attributable risk (AR), and the number needed to harm (NNH).28,29 The AR is the difference in risk between mothers undergoing nonobstetric surgery and all other mothers, and represents the additional risk experienced by mothers undergoing nonobstetric surgery, above and beyond the baseline risk. NNH is a measure that indicates how many patients on average need to be exposed to a risk factor to cause harm in an average of 1 patient who would not otherwise have been harmed. CIs for each measure of effect were estimated using a nonparametric bootstrap method. The endpoints of 95% CIs were defined as the 2.5th and 97.5th percentiles of that measured across the bootstrap samples.30 We also independently examined adverse birth outcomes in the common group of procedures, and by trimester. The OPCS codes were grouped according to specialty in order to provide a framework for analysis that was clinically meaningful. Operations on the abdomen were placed in the “abdominal” group, but this was then subdivided into “laparoscopic,” “open,” and “appendix.” “Orthopedic” was not only used for operations on bone and tendons but also for example carpal tunnel release and ganglion excision wherein the surgeon was not necessarily orthopedic. “ENT” procedures did not include head and neck cases or thyroid/parathyroid surgery, as they are coded separately under OPCS.
Sensitivity analyses were carried out to evaluate the potential effect of missing data in the stillbirth and low birth weight variables. All analyses were carried out with SAS 9.2 software package (SAS Institute, Cary, NC).
A total of 6,484,280 pregnancies were identified from the period April 2002 to March 2012. Two-thirds of study population were aged between 20 and 34 years (73.4%) and the majority were multiparous (55.5%) (Table 1). Twenty-seven percent of women were living in the most socio-economically deprived area. More than 10% of our population had an emergency admission to hospital a year before pregnancy and nearly 10% previously had a cesarean delivery.
Miscarriage associated with hospitalization occurred in 5.8% of these pregnancies. Among our cohort, there were more than 450,000 (7.5%) preterm deliveries and nearly 1.5 million (23.9%) elective or emergency cesarean deliveries. Two hundred thirty-five maternal deaths were identified and this corresponds to a maternal death rate of 4 per 100,000 (Table 2). We identified 47,628 pregnancies wherein nonobstetric surgery occurred (0.7% of all pregnancies).
The most common surgical group was abdominal (12,493, 26.2%) followed by dental (5365, 11.3), nail-skin (4762, 10.0), orthopedic (4563, 9.6%), ENT (3060, 6.4%), perianal (2977, 6.3%), and breast (1884, 4.0%) (Table 3). There were 3062 cases of appendicectomy and 1306 cases of cholecystectomy. Only 4% of all operations occurred in the same admission as the delivery or miscarriage associated with hospitalization. Furthermore, only 1.8% of miscarriage associated with hospitalization occurred in the same admission as the procedure. A closer examination of timing between delivery (or miscarriage) and surgery suggested that fewer than 6% of operations occurred within 1 week of the end of pregnancy.
After adjusting for potential confounders, the AR for miscarriage associated with hospitalization was 0.7% (95% CI 0.4–0.9) (NNH = 143), for still birth 0.4% (95% CI 0.3–0.4) (NNH = 287), for preterm delivery 3.2% (95% CI 2.9–3.4) (NNH = 31), for low birth weight 2.6% (95% CI 2.5–2.9) (NNH = 39), for cesarean section 4.0% (95% CI 3.6–4.5) (NNH = 25), and for long inpatient stay 2.0% (95% CI 1.7–2.4) (NNH = 50) in those women who had nonobstetric surgery compared with those who did not (Table 4). The AR for maternal death was 0.013% (95% CI 0.003–0.028) (NNH = 7692). Due to missing data, we carried out sensitivity analyses for stillbirth and low birth weight, excluding missing values and found little difference in our results.
The AR of adverse birth outcomes (excluding maternal death) for women who underwent abdominal surgery was higher than women who did not have surgery during their pregnancy (Table 5). For miscarriage associated with hospitalization, the AR was 5.0% with an NNH of 20. Laparoscopic abdominal surgery was associated with a high AR for miscarriage (AR = 8.2%) and seemed to contribute to most of the risk for abdominal surgery, as open abdominal surgery alone was not associated with an increased risk of miscarriage (RR = 0.65, 95% CI 0.56–0.76). Nonabdominal surgical groups, including breast surgery, ENT, and orthopedic surgery, were also associated with significantly raised risk of some outcomes.
Approximately two-thirds of all abdominal operations were laparoscopic. Where gestational age was known, nearly 5 times as many abdominal operations in the first trimester were laparoscopic (3102) versus open (643). This ratio was reversed in the second trimester (606 laparoscopic vs 995 open), and by the third trimester, 2.5 times as many procedures were open (698) versus laparoscopic (284). Our analysis comparing outcomes between laparoscopic versus open abdominal surgery (RR = 3.82, 95% CI 3.29–4.41) shows an increased risk of a miscarriage associated with hospitalization for laparoscopic procedures.
The RR for preterm delivery and cesarean section for pregnancies wherein surgery occurred during the third trimester was 20% higher than the RR for pregnancies wherein surgery occurred during the first trimester (Fig. 1). There was a little change for the RR for long inpatient stay.
This is the first study to report the risk of adverse birth outcomes following nonobstetric surgery during pregnancy across NHS hospitals in England. We extracted data from nearly 6.5 million pregnancies between April 2002 and March 2012, 10 times the total in the published literature to date. Furthermore, our data are more recent and therefore better represent current outcomes.
Previous studies have tended to present relative measures of risk, which we also report in Tables 4 and 5. However, we believe that AR and NNHs are more useful measures of risk when attempting to gauge the additional risk of harm following nonobstetric surgery. However, to discuss how our results compare with other studies, we have drawn together the literature, and made some comparisons with published measures of association, including RR and ORs. For example, Duncan et al3 found a higher RR (1.58) of spontaneous abortion associated with a general anesthetic in their 1986 Canadian study of 2565 pregnant women. For the purpose of comparison, we found a RR of 1.13 for miscarriage associated with hospitalization for surgery during pregnancy. Our study is larger, and more recent, perhaps reflecting improvements in surgery, but (like Duncan's) is unable to disassociate the risk of surgery from the risk of anesthesia or the underlying condition for which the procedure was carried out. Unlike Duncan's, our data include all surgical patients, and the dataset did not allow us to determine whether they had a general anesthetic, although it is likely that the majority did so. For abdominal surgery, we found an even higher risk of miscarriage (associated with a hospital admission) with a RR = 1.90 (95% CI 1.81–1.99), and of preterm delivery RR = 1.62 (95% CI 1.54–1.70) than women who had no surgery. In contrast, Gerstenfield et al31 found no significant difference in preterm delivery rate when comparing women who had abdominal surgery to women who did not have surgery (OR 1.13, 95% CI 0.56–2.25, P = 0.84), and only 2 cases of spontaneous abortion following surgery. Their study was small, and included only 106 women who underwent surgery. However, in our study, we are likely missing a large number of pregnancies ending in miscarriage, as many are never hospitalized. We found a significant difference between laparoscopic and open abdominal surgery for risk of miscarriage (RR 3.82, 95% CI 3.29–4.41). There may be a reluctance by surgeons to perform laparoscopic surgery beyond 26 to 28-week gestation because of perceived or real practical difficulties and previous evidence,32 and although this has more recently been refuted33 may still reflect common practice. As we are unable to determine gestational age for miscarriage, the estimated risk for laparoscopic surgery may be biased, given the timing of procedures, relative with open procedures.
In our study, we identified 235 maternal fatalities, which corresponded to a maternal death rate of 4 per 100,000. Our estimate is much lower than national estimates (10 per 100,000, report “Saving Lives, Improving Mothers’ Care,” December 9, 2014) because we only identified maternal deaths occurring in the same admission as the delivery or miscarriage. We were not able to capture maternal deaths wherein no delivery occurred (deaths wherein mother died before delivery), or deaths outside hospital. In relation to nonobstetric surgery during pregnancy, we estimated a RR of 4.72 (95% CI 2.61–8.52). The baseline risk of maternal death is fortunately very low, and translates into NNHs of 7692 procedures associated with 1 additional maternal death. This was based on only 12 women who died in our cohort during or after surgery during pregnancy, and so the CIs are wide. There is only 1 previous study reporting the maternal death of a woman undergoing laparoscopic cholecystectomy and a maternal death of 0.006% was estimated.34 Contrary to our study, the latter study was very small and presented only crude risk.
Our study revealed that in pregnancies wherein nonobstetric surgery occurred has an increased risk of stillbirth and low birth weight compared with those pregnancies wherein surgery did not occur. We found that pregnant women who underwent nonobstetric surgery had an AR of 0.4% for stillbirth and 2.6% for low birth weight compared with women who did not have surgery. However, there were no significant differences in the risk of low birth weight and stillbirth in women undergoing laparoscopic versus open abdominal surgery. Contrary to our findings, a Swedish study4 concluded that the incidence of stillbirth was not increased in women having an operation. However, the incidence of very low and low birth weight infants was increased. Mazze and Kallen35 investigated appendicectomy and laparoscopic surgery, and did not observe a statistically significant increase in stillbirth. But the authors concluded that the mean birth weight in the operated group was on average 78 g less than the expected birth weight. In both studies, analysis was conducted on the data collected between 1973 and 1981, which is now over 30 years old and is therefore unlikely to be representative of current situation.
Interestingly, the point estimates demonstrated in Fig. 1 (except for miscarriage) seem to show an increasing risk with successive trimesters. However, this may reflect that the severity of the condition necessitating surgical intervention is greater, given the natural reluctance to intervene in women at an advanced stage of pregnancy.
There are a number of limitations that need to be considered when interpreting the findings. The first is around data quality and completeness. Submission of HES records is mandatory and in general coverage is very high. Most debate around HES data quality concerns the primary and secondary diagnostic and procedure field.36–39 Nonetheless, a recent systematic review of discharge coding accuracy in routine UK data found that primary diagnosis accuracy has improved from 73.8% to 96.0% in the 10 years since the introduction of Payment by Results.40 Not all delivery records contain supplementary information, although the percentage of records with a complete maternity tail has improved over time.41 HES data have more general limitations in the recording other potential confounders. It is well known that body mass index (BMI), smoking status, and environmental factors such as air pollution are important risk factors for adverse outcomes in pregnancy; however, these variables are not recorded.
A further limitation to our study is that we treat surgery during pregnancy as a homogenous intervention, and are unable to disentangle the independent effects of anesthesia, the various indications for surgery, as well as the surgery itself, all of which may influence the outcome of pregnancy. Thus, we are not able to estimate the independent contribution to the risks of an adverse outcome during pregnancy of each component of this complex intervention.
A specific limitation is around miscarriage associated with hospitalization. Our definition of miscarriage associated with hospitalization represents only a proportion of all miscarriages. It may be more likely to be reported if it occurs in same hospital admission as a procedure, and this could account for the associated increased risk with surgery during pregnancy. However, only 1.8% of miscarriage associated with hospitalization actually occurred in same admission as the procedure. Because of the potential for ascertainment bias, temporal issues around recording of risk factors, and other unaccounted for confounding related to gestational age, we urge caution when interpreting the risk of miscarriage associated with nonobstetric surgery during pregnancy.
A strength of this study is the size of the database, allowing for adjustment of multiple obvious confounders, so some estimates of risk can be dependably derived. This type of clinical area does not lend itself to clinical trials and is too complex for plausible propensity matching as an alternative.
Our study is much more recent than other studies and reflects current risks in light of improvements to anesthesia and surgical procedures. General principles regarding the care of pregnant women requiring nonobstetric surgery have not changed significantly in the past 20 years. The anesthetic drugs used however have changed over time and there has been an increased use of laparoscopic surgery, as surgical experience has demonstrated that this is safe in pregnancy.42,43
As far as we are aware, this is the most recent study to report the risk of adverse birth outcomes following nonobstetric surgery during pregnancy. We found that the overall AR of an adverse birth outcome in women having surgery during their pregnancy compared with women who did not was generally low. A key limitation is that we are unable to separate the risk of the underlying condition (for which surgery was carried out) from the surgery itself, and the limitations of our data for defining miscarriages and maternal deaths in hospital mean that these indicators do not adequately reflect the whole risk for these outcomes. However, we still believe that in particular, the NNHs for the other indicators will be helpful to women and their surgeons to make better individual decisions based on better information. Clearly, surgery during pregnancy should only be considered if absolutely necessary; however, we believe that for those whose optimal treatment includes surgery, our analysis may provide some reassurance to mothers concerning the risks to their pregnancy.
We thank our patient representative, Emma Cooper, for her contribution to the study.
1. Rosen MA. Management of anesthesia for the pregnant surgical patient. Anesthesiology
2. Mhuireachtaigh RN, O’Grman DA. Anesthesia in pregnant patients for nonobstetric surgery
. J Clin Anesth
3. Duncan P, Pope W, Cohen M, Greer N. Fetal risk of anesthesia and surgery
4. Mazze RI, Kallen B. Reproductive outcome after anesthesia and operation during pregnancy
: a registry study of 5405 cases. Am J Obstet Gynecol
5. Cohen-Kerem R, Railton C, Oren D, et al. Pregnancy
outcome following non-obstetric
surgical intervention. Am J Surg
6. Aylin P, Williams S, Bottle A, Jarman B. Counting hospital activity: spells or episodes? BMJ (Clinical Research Ed)
2004; 329 7476:1207.
8. NHS Connecting for Health. OPCS Classification of Intervention and Procedures. London: TSO. 2012.
9. Knight H, Cromwell D, van der Meulen J, et al. Patterns of Maternity Care in English NHS Hospitals. London: Royal College of Obstetricians and Gynaecologists; 2013.
10. Fantini MP, Stivanello E, Frammartino B, et al. Risk adjustment for inter-hospital comparison of primary caesarean section rates: need, validity and parsimony. BMS Health Serv Res
11. Erekson EA, Brousseau EC, Dick-Biascoechea MA, et al. Maternal postoperative complications
after nonobstetric antenatal surgery
. J Matern Fetal Med
12. Main EK, Moore D, Farrel B, et al. Is there a useful caesarean birth measure? Assessment of the nulliparous term singleton vertex caesarean birth rate as a tool for obstetric
quality improvement. Am J Obstet Gynecol
13. Bottle A, Aylin P. Application of AHRQ patient safety
indicators to English hospital data. Qual Saf Health Care
14. Epstein D, Mason A. Costs and prices for inpatient care in England: mirror twins or distant cousins? Health Care Manag Sci
15. Carstairs V, Morris R. Deprivation and Health in Scotland. Aberdeen: Aberdeen University Press; 1991.
16. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis
17. Arck PC, Rücke M, Rose M, et al. Early risk factors miscarriage: a prospective cohort study in pregnant women. Reprod Biomed Online
18. Valero de Bernabé JV, Soriano T, Albaladejo R, et al. Risk factors for low birth weight: a review. Eur J Obstet Gynecol Reprod Biol
19. Dempsey JC, Ashiny Z, Qju C-F, et al. Maternal pre-pregnancy
overweight status and obesity as risk factors for caesarean delivery. J Matern Fetal Med
20. Garcia-Enguidanos A, Calle M, Valero J, et al. Risk factors in miscarriage: a review. Eur J Obstet Gynecol Reprod Biol
21. Flenady V, Koopmans L, Middleton P, et al. Major risk factors for stillbirth in high-income countries: a systematic review and meta-analysis. Lancet
22. Gardosi J, Madurasinghe V, Williams M, et al. Maternal and fetal risk factors for stillbirth: population based study. BMJ
23. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet
24. Flores PL, Gonzalez Perez G, Trejo FJ, et al. Risk factors in caesarean section. Gynecol Obstet Mex
25. Zhang J, Kai FY. What's the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA
26. Daly LE. Confidence limits made easy: interval estimation using a substitution method. Am J Epidemiol
27. Austin PC. Absolute risk reductions, relative risk, relative risk reductions, and numbers needed to treat can be obtained from a logistic regression model. J Clin Epidemiol
28. Schechtman E. Odds ratio, relative risk, absolute risk reduction, and the number needed to treat: which of these we should use? Value Health
29. Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment effect. BMJ
30. Efron B, Tibshirani RJ. An Introduction to the Bootstrap. Boca Raton: CRC Press; 1994.
31. Gerstenfeld TS, Chang DT, Pliego AR, Wing DA. Nonobstetrical abdominal surgery
in Women's Hospital. J Matern Fetal Med
32. Fatum M, Rojansky N. Laparoscopic surgery
. Obstet Gynecol Survey
33. Pearl J, Price R, Richardson W, Fanelli R. Guidelines for diagnosis, treatment, and use of laparoscopy for surgical problems during pregnancy
. Surg Endosc
34. Barone JE, Bears S, Chen S, et al. Outcome study of cholecystectomy during pregnancy
. Am J Surg
35. Mazze RI, Kallen B. Appendectomy during pregnancy
: a Swedish registry study of 778 cases. Obstet Gynecol
36. Knight HE, Gurol-Urganci I, Mahmood TA, et al. Evaluating maternity care using national administrative health dataset: how are statistics affected by the quality of data on method of delivery? BMS Health Serv Res
37. Evans R, Zorlu G, Boseley S. Flaws in hospital episode statistics revealed by Fol requests. Guardian (London)
38. Kirkman MA, Mahattanakul W, Gregson BA, Mendelow AD. The accuracy of hospital discharge coding for hemorrhagic stroke. Acta Neurol Belg
39. Haliasos N, Rezajooi K, O’neil K, et al. Financial and clinical governance implications of clinical coding accuracy in neurosurgery: a multidisciplinary audit. Br J Neurosurg
40. Burns E, Rigby E, Mamidanna R, et al. Systematic review of discharge coding accuracy. J Public Health
41. Murray J, Saxena S, Modi N, et al. Quality of routine hospital birth records and the feasibility of their use for creating birth cohorts. J Public Health
42. Crowhurst JA. Anaesthesia for non-obstetric surgery
. Acta Anaesthesiol Belg
43. ACOG committee on obstetric
Practice. ACG committee opinion No. 474: nonobstetric surgery
. Obstetrics & Gynecology