Participant Characteristics of acute appendicitis during pregnancy
A total of 118 patients who underwent appendectomy during pregnancy were included for observation, of which 27 patients were excluded based on the exclusion criteria (Figure. 1). Of these, 91 patients who met the inclusion criteria were enrolled, and 21and 70 patients were divided into complicated appendicitis and uncomplicated appendicitis cohorts, respectively. The baseline characteristics of acute appendicitis during pregnancy are summarized in eTable.1 in the Supplement.
Identification Of Independent Risk Factors
Univariate regression analysis found four variables to be statistically significant. However, in this multivariate regression analysis, data from the uncomplicated and complicated appendicitis groups were compared using stepwise backward logistic regression analysis, and the results suggested that gestational weeks, CRP, and NEUT% were statistically significant independent risk factors (eTable.2 in the Supplement). To improve the clinical utility, the gestational weeks were divided into three periods (first trimesters, second trimesters, and third trimesters), while the optimal cut-offs for CRP level and NEUT% were found to be 34.82 mg/L and 85.35%, by ROC curves, respectively.
Further multivariate regression analysis showed that third trimester (P = 0.013, OR = 16.81), CRP level ≥ 34.82 mg/L (P = 0.007, OR = 6.24) and NEUT% ≥85.35% (P = 0.011, OR = 18.05) were independent risk factors for complicated appendicitis (Table.1).
Table.1 Multivariate Logistic Regression Analysis of the Association between Variables and Complicated Appendicitis during Pregnancy.
Variables
|
Multivariate analysis
|
OR (95% CI)
|
P value
|
Trimesters
|
|
|
First
|
1.0 (Reference)
|
|
Second
|
4.64(0.85–38.99)
|
0.104
|
Third
|
16.81 (2.15-201.76)
|
0.013
|
NEUT%
|
|
|
<85.35%
|
1.0 (Reference)
|
|
≥ 85.35%
|
18.05 (2.86-364.82)
|
0.011
|
CRP(mg/L)
|
|
|
<34.82
|
1.0 (Reference)
|
|
≥ 34.82
|
6.24(1.75–25.84)
|
0.007
|
NEUT%, neutrophil percentage; CRP, C-reactive protein.
Nomogram And Model Performance
Backward stepwise regression analysis using AIC in a logistic regression analysis model eventually revealed a significant association between the following 3 variables and complicated appendicitis: third trimesters, CRP ≥ 34.82mg/L, and NEUT% ≥85.35%. Nomogram to predict complicated appendicitis during pregnancy are shown in Figure. 2A.
The nomogram clearly displays that each predictor corresponds to a different score. The total point was the sum of the points of three predictors for each patient. The bottom of the nomogram displays the relationship between the total score and the probability of complicated appendicitis. The higher the total score based on the sum of the assigned points for each factor in the nomogram, the more likely complicated appendicitis was indicated.
The area under the ROC curve was used to assess the discriminatory ability of the nomogram, and the results showed that the AUC of the nomogram was 0.872 (95% CI: 0.803–0.942), which indicates moderately high performance (eFigure.2 in the Supplement). A calibration plot and Hosmer–Lemeshow test were adopted to calibrate the predictive model. The analysis showed good consistency between predicted and actual probabilities (p = 0.869) (Figure. 2B). In addition, DCA was used to evaluate the clinical utility of the nomogram. As depicted in Figure.2C, if the threshold probability is 0–64%, patients with complicated appendicitis during pregnancy would benefit more from the nomogram than with treat all or no treatment conditions. Based on the results of DCA, CIC was further developed to assess the clinical utility of the nomograms. CIC revealed that the predicted probabilities matched well with the actual probabilities (Figure. 2D).
Diagnostic Performance Of The Scoring System
We modified the nomogram to a scoring system with integer points to facilitate better use in clinical practice: First trimesters (0 points), Second trimesters (4 points), Third trimesters (8 points), NEUT% <85.35% (0 points), NEUT% ≥85.35% (10 points), CRP<34.82 mg/L (0 points), CRP ≥ 34.82 mg/L (6 points) (Table.2). Based on the scores of the above variables, the total score for each patient was calculated and then the ROC curve was plotted (Figure. 3A). The results show that the scoring system has shown excellent ability in differentiating complicated and uncomplicated appendicitis (AUC: 0.869, 95% CI: 0.799–0.939). When the cut-off is set at 12, the likelihood that the patient will be diagnosed with complicated appendicitis is greatly increased if the total score exceeds 12. However, when the total score is below 12, patients are less likely to be diagnosed with complicated appendicitis. When the optimal cut-off point was set at 12, the corresponding sensitivity, specificity, PLR, NLR, PPV, and NPV values were 100%, 58.60%, 2.41, 0, 42%, and 100%, respectively (Table.3). At the same time, the scoring system also shows good calibration, which means that the predicted probabilities of the scoring system are highly consistent with the actual probabilities (p = 0.909) (Figure. 3B).
Table 2
Scoring system for Complicated Appendicitis during Pregnancy.
Variables
|
scores
|
Trimesters
|
First
|
0
|
Second
|
4
|
Third
|
8
|
NEUT%
|
<85.35%
|
0
|
≥ 85.35%
|
10
|
CRP (mg/L)
|
<34.82
|
0
|
≥ 34.82
|
6
|
Table.3 ROC analysis of the scoring system for identifying Complicated Appendicitis during Pregnancy.
Cut-off
score
|
Youden
index
|
Sensitivity
|
95% CI
|
Specificity
|
95% CI
|
PLR
|
> 7
|
0.31
|
1.00
|
0.85–1.00
|
0.31
|
0.22–0.43
|
1.46
|
> 9
|
0.33
|
1.00
|
0.85–1.00
|
0.33
|
0.23–0.45
|
1.49
|
> 12
|
0.59
|
1.00
|
0.85–1.00
|
0.59
|
0.47–0.69
|
2.41
|
> 15
|
0.57
|
0.76
|
0.55–0.89
|
0.81
|
0.71–0.89
|
4.10
|
> 17
|
0.51
|
0.67
|
0.45–0.83
|
0.84
|
0.74–0.91
|
4.24
|
CI = confidence interval; PLR = positive likelihood ratio |
Restricted Cubic Splines Of Complicated Appendicitis During Pregnancy
To further clarify the relationship between predictors and complicated appendicitis, we performed a restricted cubic splines analysis. As shown in Figure. 4, the relationship between all these predictors and complicated appendicitis in pregnancy was linear (P for non-linear > 0.05), which further proves that these predictors of continuous variables have excellent predictive performance.