全文

Abstract

Objectives

To examine the clinical risk factors and complications of shoulder dystocia today and to evaluate ultrasound methods predicting it.

 

Study design

Retrospective, matched case-control study at a University Hospital with 5000 annual deliveries. The study population consisted of 152 deliveries complicated by shoulder dystocia over a period of 8.5 years (January 2004–June 2012) and 152 controls matched for gestational age and parity. The data was collected from the medical records of mothers and children and analyzed by conditional logistic regression. Incidences and odds ratios were calculated for risk factors and complications. Antenatal ultrasound data was analyzed when available by conditional logistic regression to test for significant differences between study groups.

 

Results

Birthweight (OR 12.1 for ≥4000 g; 95% CI 4.18–35.0) and vacuum extraction (OR 3.98; 95% CI 1.25–12.7) remained the most significant clinical risk factors. Only a trend of an association of pregestational or gestational diabetes was noticed (OR 1.87; 95% CI 0.997–3.495, probability of type II error 51%). Of the complications of shoulder dystocia the incidence of brachial plexus palsies was high (40%). Antenatal ultrasound method based on the difference between abdominal and biparietal diameters had a significant difference between cases and controls.

 

Conclusions

The impact of diabetes as a risk factor has diminished, which may reflect improved screening and treatment. Antenatal ultrasound methods are showing some promise, but the predictive value of ultrasound alone is probably low.

 

Keywords

Shoulder dystocia; Diabetes; Gestational; Paralysis; Obstetric; Ultrasonography

 

Introduction

Despite decades of research and numerous published works combining known risk factors for predicting shoulder dystocia (SD), none of these have been shown to provide reasonably high predictive values when considering the amount of morbidity caused by unnecessary cesarean sections, and this appears to be the situation with antenatal ultrasound (US) methods as well [1], [2] and [3]. The most consistent and significant historical risk factors of SD have been fetal macrosomy (usually defined as birthweight of ≥4500 g) and operative vaginal delivery. Other factors such as previous SD, maternal obesity, increased parity and induction of labor have also been reported [4], [5] and [6]. Maternal diabetes has been repeatedly recognized as a major risk factor of SD [4] and [7], but there is evidence suggesting that active pharmacological treatment of elevated maternal blood glucose levels can prevent SD [8] and [9]. Neonatal encephalopathy and fetal death due to delayed delivery and asphyxia are the most devastating but fortunately rare complications caused by SD. The most common significant complication is obstetrical brachial plexus palsy (BPP) of the fetus. Maternal complications include perineal lacerations and increased blood loss [4].

 

Antenatal US examination has some value in identifying pregnancies that have too high a risk for vaginal delivery due to estimated fetal macrosomy or more specifically unfavorable proportions of the fetus [6], [10] and [11]. Customizing the fetal weight estimation to local population and the sex and the size of the fetus may improve the accuracy of the estimate [12]. Calculating the difference between fetal abdominal diameter (AD) and biparietal diameter (BP) as described by Cohen et al. [10] was applied clinically in our study hospital. Abdominal circumference (AC) alone has also been applied for the same purpose [13].

 

The main objective of this study was to examine retrospectively the consequences of SD and the significance of the classic clinical risk factors as a predictor or causative agent of SD in a present-day tertiary care hospital population. Screening and treatment of gestational diabetes within Finnish antenatal care system follows a national guideline, which currently defines the criteria for insulin treatment as plasma fasting glucose of ≥5.5 mmol/l or postprandial glucose at one hour of ≥7.8 mmol/l. We hypothesized that recent improvement in screening, more active treatment of impaired glucose tolerance and the selection of many diabetic mothers with large fetuses for cesarean section in Finland may have diminished the impact of diabetes as a risk factor. In our clinic the proportion of cesarean sections in gestational diabetes has slightly increased during the past years. In 1995 the rate of cesarean sections was 16% and it has increased to 18% in 2003 and 24% in 2010. Other risk factors may therefore now have more significance than previous studies indicate. The incidence of BPPs has been decreasing at our hospital and is currently relatively low at 0.1–0.2% of all deliveries. Substantial changes have therefore recently taken place regarding the key risk factors and consequences of SD. Another objective was to provide a preliminary estimate on the usefulness of one novel (AC divided by biparietal diameter; AC:BP) and some previously published US-based methods to predict SD [10] and [14].

 

Material and methods

We searched for SD cases from 1st of January 2004 to August 2012 using the ICD-10 codes O66.0 (obstructed labor due to shoulder dystocia) for parturients and P14 (birth injury to peripheral nervous system) for children as search criteria. In all cases and controls the correctness of the diagnosis was ascertained by checking carefully the patient files. Deliveries found using the P14 diagnosis code for BPP were included in the study only when SD could be confirmed. One twin delivery and one fetus mortus were excluded. For each SD case we allocated one control delivery by going through the delivery diaries starting from the SD case and searching forward for first subsequent vaginal delivery fulfilling matching criteria; parity (up to four previous deliveries) and gestational age at birth (±7 days). Breech or twin deliveries were not accepted as controls.

 

Antenatal fetal US examinations done within 14 days preceding the delivery were included in the study when available. For fetal weight estimations BP, AC and femur length (FL) were measured and the Hadlock 2 formula was used. Abdominal diameter (AD) was calculated by dividing the measured AC by 3.14. The substraction of AD − BP described previously [10] and their ratio (AC:BP) were then derived.

 

Analysis of data was performed with SPSS 19.0 software (IBM SPSS Statistics for Windows, Version 19.0. Armonk, NY: IBM Corp. Released 2010). Differences between cases and matched controls for risk factors and complications of SD and US parameters were tested with conditional logistic regression. Risk factors with significant univariate p values were included in a multivariate conditional regression to identify independent risk factors and calculate their adjusted odds ratios (OR) with 95% confidence intervals (CI). All p values were two-sided. No meaningful ORs or p values could be calculated for birthweight exceeding 4500 g, pregestational diabetes and several complications of SD due to low n in the control group. Wilcoxon test was used for AC:BP. Post hoc power calculations for the effect of diabetes were made with PS: power and sample size calculation software to estimate the probability of type II error. The study protocol was approved in the Ethics Committee of Tampere University Hospital.

 

Results

There was a total of 42,964 deliveries in our hospital in January 2004–June 2012, 83% of which were vaginal. 152 deliveries were eventually included in the analysis as SD cases. One woman had two deliveries complicated by SD during the study period. The incidence of SD was 0.35% of all deliveries and 0.42% of vaginal deliveries. The proportion of nulliparous women was slightly more frequent in deliveries complicated by SD (48%) than in the total child birthing population in our hospital (44%).

 

Table 1 concerns the risk factors and their univariate p values for differences between groups. Results of multivariate conditional logistic regression analysis of risk factors of SD are reported in Table 2. There was no significant difference between groups in the incidence of maternal diabetes, but a post hoc power calculation showed a probability of 51% for a false negative result. Children born with SD weighed more than their controls, but there was significant overlap between groups ( Fig. 1).

 

 

 

 

US data on different measurements was available for 83–95 cases and 57–80 controls depending on the measurement. Of the basic US parameters, only AC was significantly different between groups, whereas BP (p = 0.734) or FL (p = 0.128) were not ( Table 3). There was a significant difference between cases and controls for AC, AC:BP, AD − BP and estimated fetal weight (p = <0.001–0.005), but only AD − BP remained significant in a multivariate model. The median delay from US examination to labor was 3 days for SD cases and 2.5 days for controls (interquartile range 5 days in both groups), but the difference was not statistically significant (p = 0.439). In our data a specificity of 98% for SD could be achieved with a sensitivity of 12–25% for different cutoffs for the US parameters. We experimented with an estimation of the ratio of true positive US test results (SD cases detected correctly by US) to false positive results (assuming the ratio of false positive US examinations to be the same in all pregnancies as in our control group). In our material this suggests that at least 30 unnecessary cesarean sections (ratio of false positives to true positives) would be required to prevent one SD when using an optimal cutoff for the most accurate US parameter (AD − BP ≥ 25 mm) when reasonable sensitivity is to be maintained.

 

 

SD was associated with complications for both parturients and newborns (Table 4). During the study period 86 BPPs were diagnosed, and 69% of these (n = 59) were associated with SD. There were no fetal deaths, but one serious brain injury occurred in the SD group. BPP was diagnosed in 39% of children born with SD, but only ten of the BPPs (6.6% of SD cases) necessitated a surgical intervention. Other fetal birth complications in the SD group included two adrenal hematomas and 26 clavicular fractures (17%) but no humeral fractures.

 

 

Comment

Although the population of this study represents a tertiary care hospital, majority of all women in the region deliver in our hospital regardless of their need for specialized obstetric care. The study sample thus probably represents well the general population of the region, which is Caucasian with only rare exceptions. As antenatal US was not done routinely to all parturients, the women who underwent an US examination probably had more complicated pregnancies than those who did not, which may have reduced the observed capability of US to identify women in risk.

 

The incidence of SD was within the range reported in previous studies (0.15–7%), [6], [15], [15], [16], [17], [18], [19] and [20] and although there was annual variation between 0.21–0.57%, no obvious trend of change could be observed. The highest figures for incidence have previously been reported from prospective studies and are probably more representative of the true incidence [6]. Our retrospective study design was likely to underestimate the incidence of milder cases of SD, as the diagnosis code is probably not recorded every time there are difficulties in extracting the shoulders. The incidence of BPPs after SD (39%) was high compared to previous studies with incidences of usually less than 20% [2], [4], [17], [21] and [22], and BPPs were surprisingly often (17% of BPP cases) permanent, as previously reported incidences have usually not exceeded 10% of BPP cases [2], [4], [7], [22], [23] and [24]. This may once more suggest that in our study material some milder cases had been omitted. Including BPP as a search criterion to find as complete material as possible evidently causes overestimation of the incidence of BPP and slight selection bias towards more complicated cases. The incidence of clavicular fractures in our study (17%) was within the previously reported range of 0–21% [4] and [7]. As expected in the light of previous studies [4], [16], [17], [20] and [21], the neonates born with SD had more often signs of poor oxygenation during the delivery compared to controls, although the difference was not prominent.

 

Increasing maternal height appeared to be slightly protective from SD (OR 0.90 for an increase of 1 cm compared to SD cases), which has been noted before and could represent the effect of a disparity between the relative sizes of mothers and their fetuses [5] and [18]. We found no difference in maternal body mass index (BMI) while previous studies have demonstrated variable results [5], [7], [20], [22] and [25]. Although there was a trend towards higher incidence of both pregestational and gestational diabetes in the SD group, interestingly the differences in the incidence of diabetes between groups did not reach statistical significance, no matter if insulin-treated or not. Based on the observed OR of 1.87, a sample size of 313 matched case-control pairs would have been required to achieve statistical power of 80% for detecting such a difference. Our study was probably underpowered to demonstrate the effect of insulin treatment in gestational diabetes as well, but the lack of association between insulin treatment and SD could also interestingly signify the success of therapy for this otherwise risky pregnancy disorder. Nevertheless, the difference between groups appears to be smaller than noted in most previous studies which have reported ORs for maternal diabetes ranging between 1.7 and 28 [5], [16], [19], [20] and [23]. There are however also previous studies in which maternal diabetes has not appeared as an independent risk factor [1] and [15]. Improved blood glucose control and increased diagnoses of even milder cases of gestational diabetes probably contribute to this [8], and a remarkable proportion of diabetic pregnancies with large fetuses are ending in elective cesarean sections in the modern obstetrics. In fact, since the introduction of efficient home monitoring of blood glucose in gestational diabetes from 2002 and strict criteria for insulin therapy we have seen a remarkable decline in the proportion of large fetuses in our clinic. In 1995, 36% of newborns after a pregnancy with gestational diabetes weighed more than 4 kg, while the proportions were 22% and 20% in 2003 and 2010, respectively. The results of this study could be interpreted as an additional incentive to focus on efficient management of gestational diabetes. Contrary to some studies showing a risk as high as 12–20% for recurrent SD [4], [6] and [23], previous SD was not a risk factor in our population, probably reflecting the successful selection for cesarean section of many cases with a history of previous SD.

 

We did not find any studies showing previous vacuum extraction as a risk factor, although SD has been shown to recur more often when the index delivery with SD was operational [18] and [26]. Although in our study the effect of previous vacuum delivery did not appear as an independent risk factor after controlling for confounders, it could reflect some feature of the parturients predisposing them to repeated difficult deliveries and vacuum extractions, which could be causative agents of SD. Vacuum extraction was indeed one of the most significant independent risk factors in this study as well as most previous ones, and the OR (3.98) is quite similar compared to previously reported ORs or relative risks for instrumental vaginal delivery with a range of 1.6–28 [4], [5], [16], [19], [21], [22] and [23]. It would thus appear feasible to consider the possibility that instrumental delivery in fact does increase the incidence of SD. In our study there was a small increase in the risk of SD when the 1st stage of labor was prolonged even after controlling for confounders. We noticed no difference in the duration of the 2nd stage, whereas previous studies have variably noticed associations with prolonged or sometimes precipitous labor, often neither [4], [6], [7], [16], [17], [18], [19], [20] and [21]. Concerning the choice of obstetric analgesia, previous studies have focused on the role of epidural anesthesia, with negative results in a vast majority of studies [15], [17], [18], [19], [20], [21] and [22]. None of the different anesthetic methods could be identified as an independent risk factor in our material either. After controlling for confounders, induced labor was not a significant risk factor in our study, whereas previous reports have been conflicting [1], [5], [16], [19], [20], [21] and [22]. In our material fetal macrosomy was clearly the strongest risk factor with an adjusted OR of 12.1 for ≥4000 g within the wide range of ORs or relative risks (4.8–24.3) presented previously [1], [15], [16], [20] and [23]. Nevertheless, birthweight was less than 4000 g in 35% of our SD cases.

 

There were statistically significant differences between cases and controls for all the examined US parameters, which should provide an incentive to investigate the diagnostic value of antenatal US in a prognostic setting. Sensitivities and specificities were subject to spectrum bias in our study setting and thus overoptimistic Moreover, future studies could be designed to find out, which of the US algorithms have predictive value independent of fetal weight estimation or simple AC, how they perform when applied in the presence of significant clinical risk factors and whether they recognize different subgroups of risk pregnancies. Due to infrequency of SD, any method would have to reach extremely high positive predictive value to justify a cesarean section due to SD risk considering that most SDs are resolved without complications. AC:BP was examined here as a novel parameter, however showing diagnostic value probably similar to AD − BP.

 

Conflicts of interest

None.

 

Funding and sponsors

None.

 

Acknowledgements

None.