Original Contribution

Clinical prediction rule to distinguish Pelvic inflammatory disease from acute appendicitis in women of childbearing age

Koji Morishita MD^a,*, Masanori Gushimiyagi MD^b, Mikio Hashiguchi MD^c,

Gerald H. Stein MD^d,e, Yasuharu Tokuda MD, MPH^f

^aDepartment of Surgery, Okinawa Hokubu Hospital, Okinawa 905-8512, Japan

^bDepartment of Surgery, Okinawa Chubu Hospital, Okinawa 904-2293, Japan

^cDepartment of Obstetrics and Gynecology, Okinawa Chubu Hospital, Okinawa 904-2293, Japan ^dDepartment of Medicine, College of Medicine, University of Florida, Gainesville, FL 32611, USA ^eDepartment of Medicine, John A. Burns School of Medicine, University of Hawaii, Honolulu, HI 96813, USA ^fClinical Practice Evaluation and Research Center, St. Luke’s Life Science Institute, Tokyo 104-8560, Japan

Received 7 May 2006; revised 8 June 2006; accepted 21 June 2006

Abstract

Objective: We aimed to develop a clinical prediction rule to distinguish pelvic inflammatory disease (PID) from acute appendicitis in women of childbearing age.

Methods: We reviewed medical records over a 4-year period of female patients of childbearing age who had presented with abdominal pain at an urban emergency department and had either appendicitis (n = 109) or PID (n = 72). A prediction rule was developed by use of recursive partitioning based on significant factors for the discrimination.

Results: The significant factors to favor PID over appendicitis were (1) no migration of pain (odds ratio [OR], 4.2; 95% confidence interval [CI], 1.5-11.5), (2) bilateral abdominal tenderness (OR, 16.7; 95% CI, 5.3-50.0), and (3) absence of nausea and vomiting (OR, 8.4; 95% CI, 2.8-24.8). The prediction rule could rule out appendicitis from PID with sensitivity of 99% (95% CI, 94-100%) when classified as a low-risk group by the following factors: (1) no migration of pain, (2) bilateral abdominal tenderness, and

(3) no nausea and vomiting.

Conclusion: We developed a prediction rule for childbearing-aged women presenting with acute abdominal pain to distinguish acute appendicitis from PID based on 3 simple, clinical features: migration of pain, bilateral abdominal tenderness, and nausea and vomiting. Prospective validation is needed in other settings.

D 2007

* Corresponding author. Tel.: +81 980 52 2719; fax: +81 980 54 2298.

E-mail address: [email protected] (K. Morishita).

Introduction

It is often difficult to accurately separate the diagnosis of acute appendicitis from pelvic inflammatory disease (PID)

0735-6757/$ - see front matter D 2007 doi:10.1016/j.ajem.2006.06.013

in women of childbearing age [1-5]. The clinical presenta- tion of PID from inflammation originating at the uterus, the right ovary, and the fallopian tube is similar to that of acute appendicitis because of the proximity of the appendix to these organs [6-8]. Consequently, female patients aged between 15 and 25 years had about a 60% incidence of negative laparotomies compared with patients aged between 36 and 45 years with an incidence of about 20% [9].

early diagnosis and treatment of acute appendicitis are extremely important. A delay in the diagnosis can lead to appendiceal perforation with an increased rate of Wound infection and Intra-abdominal abscess [5,10]. In addition, the relative risk of subsequent tubal infertility is increased to about 5 from appendectomy for a ruptured appendix [11].

Although several previous studies have shown discrim- inant factors in the differential diagnosis of acute appendi- citis and PID in women of childbearing age, they presented with slightly different results. One study revealed the factors were duration of symptoms, nausea and vomiting, prior PID, cervical motion tenderness, bilateral abdominal ten- derness, and rebound tenderness at the right lower quadrant [1]. Other studies revealed that the indicators favoring appendicitis included the presence of anorexia and onset of pain later than day 14 of the menstrual cycle; those favoring PID included a history of vaginal discharge, urinary symptoms, prior PID, tenderness outside the right lower quadrant, cervical motion tenderness, and abnormal urinal- ysis findings [2,12]. Consequently, physicians may find it difficult to make the precise diagnosis of acute appendicitis or PID in women of this age group. Expensive studies such as computed tomography, magnetic resonance imaging scans, and laparoscopy are often needed for many patients. [13-18]. Finally, a high false-positive rate for appendicitis was noted in women of childbearing age, although the modified Alvarado score was used as a diagnostic aid for acute appendicitis in patients in general including men and children [19].

Recursive partitioning analysis is well suited for the construction of a simple decision rule. This method can maximize the overall sensitivity to reduce the misclassifica- tion of a particular diagnosis by using its significant clinical factors based on multivariable-adjusted logistic regression analysis. Because early diagnosis and treatment of acute appendicitis are important, we aimed to develop a simple and robust prediction rule for excluding acute appendicitis with high sensitivity in the differential diagnosis of acute appendicitis and PID in women of childbearing age.

Methods

Patient enrollment and data collection

Consecutive female patients of childbearing age who presented with acute abdominal pain to an urban emergency department (ED) and who were found to have either

appendicitis or PID during the 4-year study period (from January 2000 to June 2003) were included in the study. We considered the age range of 12 to 58 years for the childbearing age based on the study method of Bongard et al [1]. The hospital is a 550-bed community teaching hospital. The ED provides primary to tertiary care to a population of approximately 400000 and treats more than 50000 patients annually.

The enrollment criteria and data collection were based on the study method of Webster et al [2]. We retrospectively reviewed the medical records of all female patients with a final diagnosis of appendicitis and PID. Patients were excluded if they had undergone a prior appendectomy, hysterectomy, salpingectomy, or oophorectomy. Pregnant patients were also excluded. The inclusion criterion for the diagnosis of appendicitis was histopathology indicating inflammatory cell infiltrates in the walls of the removed appendix. PID was diagnosed based on the presence of bminimum criteria for clinical diagnosisQ of the Centers for Disease Control (CDC): lower abdominal tenderness, adnexal tenderness, and cervical motion tenderness. Addi- tional criteria of the CDC guideline for PID were selectively used to assist in making the diagnosis: an oral temperature greater than 38.38C, abnormal cervical or vaginal mucopur- ulent discharge, the presence of white blood cells on saline microscopy of vaginal secretions, elevated erythrocyte sedimentation rate or C-reactive protein, laboratory docu- mentation of a cervical infection with Neisseria gonor- rhoeae or Chlamydia trachomatis, histopathologic evidence of endometritis on an endometrial biopsy, Transvaginal sonography or magnetic resonance imaging showing thick- ened fluid-filled tubes with or without free pelvic fluid or tubo-ovarian complex, and laparoscopic abnormalities consistent with PID [20].

Data collection from clinical history included age, onset symptoms, migration of pain, and Gastrointestinal symptoms (appetite loss, nausea, vomiting, and diarrhea). Data from physical examination were body temperature, location of abdominal tenderness, presence of rebound tenderness, and cervical motion tenderness. Laboratory data included blood tests for total leukocyte count, C-reactive protein, and urinary sediment for pyuria. Serum C-reactive protein concentrations were measured by the quick test kit. Pyuria was defined as urinalysis with greater than 5 leukocytes per high-power field. The study received prior approval from the institutional review board of the Okinawa Chubu Hospital.

Statistical analysis

Univariate analysis involved Fisher exact test for the nominal variables and Student t test for the continuous variables. We identified significant variables for discrimi- nation between acute appendicitis and PID using a multivariable-adjusted logistic regression analysis. All P values were two-sided and considered statistically significant at less than .05. We used CART version 5.0

(Salford Systems, San Diego, CA) for recursive partitioning analysis and SPSS version 12.0 (SPSS, Chicago, IL) for general statistical calculation.

We used a recursive partitioning analysis to build a prediction rule. This technique generates a classification tree with series of binary splits in which the patients are assigned to mutually exclusive subgroups according to a set of predictors. When applied to our cohort data, each binary split in a tree produces 2 subgroups, one that contains a relatively high proportion of the patients with acute appendicitis and the other that contains a relatively high proportion of the PID patients. The construction of the rule considered the significant predictor variables based on the multivariable-adjusted logistic regression analysis.

A Gini index, a measure of impurity of a node, was used as the node splitting criterion to identify the cutoff point for the best separation of the 2 subgroups. This index selects the partition with the greatest purity by measuring the amount of variance in the proportion of patients with acute appendicitis between each potential pair from the partition. The partitioning starts after evaluating each predictor for its potential to separate subgroups and selects the best predictor with the most pure division for the first split. This procedure is then repeated for each of the subsequent subgroups that are generated from the split, again evaluating all potential cutoff points of each variable to identify the predictor that provides the best separation.

Because an initial large tree model using the large number of parameters usually has an overfitting problem, a pruning technique is used to eliminate several branches of the initial tree and to form a sequence of nested candidate subtrees. A cross-validation technique then identifies the subtree to minimize the misclassification rate by a sample reuse method. Because we used 10-fold cross-validation, the entire sample is randomly partitioned into 5 samples, each containing 20% of all patients. With the first 20% sample set aside, using the other 80% samples selects the best subtree. The 20% sample is then run down this subtree as a validation sample and the misclassification rate is calculat- ed. This process is repeated for all 10 validation partitions and identifies the most complex subtree while minimizing the misclassification rate.

We grouped the several terminal nodes into 3 different risk levels (high-, intermediate-, and low-risk group) based on the prevalence of acute appendicitis within each terminal node. The Gini index can assign costs of misclassifying each category. Because our goal was to generate a prediction rule that would accurately exclude the patients with acute appendicitis, we set the misclassification cost for labeling the patients with acute appendicitis as PID at 20 times higher than for labeling the PID patients as acute appendi- citis. Varying misclassification costs in this manner tends to result in a final model with higher sensitivity in the recursive partitioning analysis.

We evaluated the classification performance of the final rule by calculating the prevalence with 95% confidence

intervals (CIs) within each classified group. We also assessed the classification characteristics of the final rule by calculating the sensitivity, specificity, and positive and negative predictive value with 95% CIs within each group. We used a cutoff point for the high-intermediate vs the low-risk group classifications and calculated the classifica- tion characteristics because our goal was to generate a prediction rule that would accurately exclude the patients with acute appendicitis by identifying the low-risk group.

We also calculated positive and negative likelihood ratios with 95% CIs to assess the clinical usefulness of the prediction rule. The positive likelihood ratio is defined as sensitivity/(1 — specificity), indicating the probability of a positive test result in the patients with acute appendicitis divided by the probability of a positive test result in the PID patients. The negative likelihood ratio is defined as (1 — sensitivity)/specificity, indicating the probability of a negative test in the patients with acute appendicitis divided by the probability of a negative test result in the PID

Table 1 Clinical characteristics of patients with appendicitis

and patients with PID

Clinical characteristics Appendicitis PID P*

PMN indicates polymorphonuclear leukocytes; HPF, high-power field.

* P values were calculated using Student t test or Fisher exact test.

+ 80 patients with appendicitis did not receive rectal examination.

(n = 109)		(n = 72)
Age (y) (mean F SD)	30.6 F 13.1	26.1 F 7.0	.003
Temperature (8C) (mean F SD)	37.152 F 0.6101	37.425 F 0.8105	.016
Onset symptoms (%)
Epigastric or	64 (58.7)	8 (11.11)	b.001
midgastric pain
lower abdominal pain	23 (21.1)	58 (80.56)	b.001
Diffuse abdominal pain	12 (11.0)	1 (1.39)	.016
Other symptoms	0 (0)	2 (2.78)	.167
Migration of pain	7.5 (75.8)	9 (13.0)	b.001
Gastrointestinal
symptoms (%)
Appetite loss	55 (50.5)	11 (15.28)	b.001
Nausea or	77 (72.0)	21 (29.2)	b.001
vomiting or both
Diarrhea	15 (13.8)	12 (16.67)	.832
Abdominal
tenderness (%)
Right	92 (84.4)	17 (23.61)	b.001
Left	0 (0)	7 (9.72)	.001
Bilateral	11 (10.1)	38 (52.78)	b.001
Rebound tenderness	86 (80.4)	54 (75.0)	.461
Cervical motion	9 (8.3)+	61 (84.72)	b.001
tenderness (%)
Laboratory tests (%)
Leukocyte count	13421 F	13200 F	.808
(/mm3) (mean F SD)	5556	6218
C-reactive protein	3.2 F 4.4	3.5 F 4.7	.662
(mg/dL) (mean F SD)
Pyuria (N5 PMNs/HPF)	27 (24.8)	32 (44.4)	.038

patients. A likelihood ratio greater than 1 indicates an increased probability that appendicitis is present, and a likelihood ratio less than 1 indicates a decreased probability that appendicitis is present. The magnitude of likelihood ratios correlates with the magnitude of decrease (or increase) in the likelihood of appendicitis.

Results

We reviewed 181 consecutive patients (109 patients with acute appendicitis and 72 patients with PID). Table 1 shows the clinical characteristics of patients with acute appendicitis and with PID, and the P values of univariate analyses. In these univariate analyses, the significant variables for the discrimination were age ( P = .003); body temperature ( P = .016); onset symptoms including epigastric or midgastric pain ( P b .001), lower abdominal pain ( P b .001), and diffuse abdominal pain ( P = .016); migration of pain ( P b .001); gastrointestinal symptoms including appetite loss ( P b .001) and nausea or vomiting ( P b.001); location of abdominal tenderness including right ( P b.001), left ( P = .001), and bilateral ( P b .001); cervical motion tenderness ( P b .001); and pyuria ( P = .038). Although univariate analysis showed cervical motion tenderness as significant, there were missing data for this finding in 80 patients with appendicitis as some pa- tients with appendicitis did not receive pelvic examination due to practical and personal reasons. Accordingly, this physical sign was not included in the following multivar- iable analysis.

Multivariable adjusted logistic regression analysis showed that the 3 significant variables to favor PID over acute appendicitis were the following: (1) no migration of pain indicating odds ratio (OR) of 16.8 with 95%

confidence interval (CI) of 5.7-49.4; (2) bilateral abdominal tenderness with OR of 16.7 (95% CI, 5.3-50.0); (3) no nausea or vomiting with OR of 3.6 (95% CI, 1.4 -9.5). There was no significant difference of any laboratory data between patients with acute appendicitis and patients with PID.

Fig. 1 shows the prediction tree based on a recursive partitioning analysis that consisted of the 5 classification splits and generated the 6 terminal nodes. These terminal nodes were then grouped as the high-, intermediate-, and low-risk groups. Patients in the high-risk group were

(1) those with migration of pain and without bilateral abdominal tenderness and (2) those without migration of pain and bilateral abdominal tenderness but with nausea or vomiting. Patients in the intermediate-risk group were

(1) those with migration of pain and bilateral abdominal tenderness; (2) those without migration of pain, bilateral abdominal tenderness, and nausea or vomiting; and (3) those without migration of pain and with bilateral abdominal tenderness and nausea or vomiting. Finally, patients in the low-risk group were those without migration of pain and with bilateral abdominal tenderness but without nausea or vomiting.

Misclassification rate was low in the low-risk group. There was only 1 patient with acute appendicitis (4.6%; 95% CI, 0.12%-23.8%) of 21 patients classified as the low- risk group. In the high-risk groups, on the other hand, there were 81 patients with acute appendicitis (91.0%; 95% CI, 83.1%-96.0%) of 89 patients. Using a cutoff point for identifying the low-risk group vs other groups (high and intermediate risk), we obtained a sensitivity of 99.0% (95% CI, 94.4%-99.9%) and a specificity of 33.9% (95% CI, 23.1%-46.6%). Furthermore, using the same cutoff point for identifying the low-risk group vs other groups, the positive predictive value was 71.3% (95% CI, 63.2%-

78.3%) and the negative predictive value was 95.2% (95%

Fig. 1 The classification algorithm for clinical prediction rule for differential diagnosis of acute appendicitis and PID in female patients of childbearing age.

CI, 77.3%-99.8%). Likewise, the positive likelihood ratio was 1.50 (95% CI, 1.25-1.80) and the negative likelihood

ratio was 0.03 (95% CI, 0.004-0.22).

Discussion

We developed a simple and sensitive clinical prediction rule for distinguishing acute appendicitis from PID by using recursive partitioning with 10-fold cross-validation. This rule consists of 3 clinical criteria for ruling out acute appendicitis from PID with 99% sensitivity and 95% negative predictive value when classified as a low-risk group by the following factors: (1) no migration of pain,

(2) presence of bilateral abdominal tenderness, and (3) no nausea and vomiting. Although each of these 3 factors cannot rule out acute appendicitis as a single criterion, the combination of these 3 findings could be used as a quick guide for the clinical decision making.

This risk-aversive prediction rule seems clinically rea- sonable, as it is crucial to avoid misclassifying patients with acute appendicitis as PID because of the increased risk of overall adverse events from appendiceal perforation by misdiagnosis of acute appendicitis compared with that of PID. Consequently, using this rule for identifying low-risk patients may reduce the need for imaging tests and laparotomy with possible cost saving. Nevertheless, we should also consider other clinical factors in estimating individual risk of acute appendicitis.

Although univariate analyses showed a large number of variables significant for the discrimination between acute appendicitis and PID, our multivariable-adjusted logistic regression analysis identified only 3 significant variables. We need to be cautious when interpreting the results of multiple procedures of univariate analyses because performing multiple tests can provide a greater risk of obtaining statistically false positive results. Therefore, we may need to be conservative for identifying significant factors from multiple clinical variables. It may be safer to use significant factors based on a multivariable-adjusted analysis for developing a robust clinical prediction rule.

Migration of pain, typically from the epigastric or midgastric area to the right lower quadrant, is a common finding seen in patients with acute appendicitis [10]. The onset of pain of acute appendicitis is generally related to a visceral referred pain. The pain radiates to the part of the body supplied by somatic sensory fibers associated with the same segment of spinal cord that receives visceral sensory fiber from the associated visceral organ [21]. Afferent sensory nerve fibers from the appendix accompany the sympathetic nerve fibers to the T10 segment of the spinal cord [21]. Thus, the area of onset pain tends to be the epigastric region. In contrast, afferent sensory nerve fibers from the salpinx have 2 pathways. One is the pelvic splanchnic nerve, which is a parasympathetic nerve and enters spinal cord L5 and S1-3 [21]. Another pathway is the

hypogastric nerve, which is a sympathetic nerve and enters the spinal cord from T12 to L2 [21]. This is considered the reason why the onset pain of acute appendicitis tends to be in the epigastric area, whereas that of PID is in the lower abdomen.

A physical examination of acute appendicitis generally shows tenderness at the site of the appendix (ie, McBurney point), whereas the tenderness in PID tends to be more often the bilateral lower abdomen or diffuse abdomen [1]. The serosa of the appendix becomes inflamed in the progress of appendicitis, thus causing irritation of the parietal peritone- um and stimulation of the somatic pain receptors in the abdominal wall [22]. On the other hand, the inflammatory route of PID is from the uterocervical through the corpus uteri to the fallopian tubes, causing bilateral salpingitis and is associated with uterine inflammation. For this reason, PID patients tend to show diffuse bilateral lower abdominal tenderness.

The presence of nausea or vomiting or both also proved to be useful in making the differential diagnosis, which was suggested by a previous study [1]. Acute appendicitis pathologically involves a part of the gastrointestinal tract and thus tends to show gastrointestinal symptoms such as nausea and vomiting more frequently than PID. However, we found about 30% of PID patients in our study had nausea or vomiting or both. So the sole presence of nausea or vomiting could not rule in the diagnosis of acute appendi- citis. Rather, the absence of nausea or vomiting combined with the other 2 factors in the prediction rule seems clinically useful.

To our knowledge, this study is the first to generate a clinical prediction rule using a recursive partitioning analysis for differential diagnosis of acute appendicitis and PID in female patients of childbearing age. The previous studies analyzed various clinical variables and identified several significant predictors [1,2,23]. However, few pre- sented the prediction rule based on statistically sound contemporary methods. The performance of our rule seems clinically reasonable to guide clinical decision making. Moreover, many physicians are now using various predic- tion rules if they are simple and sensitive. Nevertheless, we interpret our results in the context of its clinical application. We performed the study and generated the prediction rule retrospectively in a single institution. Thus, our prediction rule would need external prospective validation for its generalizability to other institutional settings. When exter- nally validated, the rule could be used in combination with individualized clinical judgment.

Conclusion

We developed a prediction rule for childbearing-aged women presenting with acute abdominal pain to exclude acute appendicitis from PID, based on 3 clinical simple features: no migration of pain, presence of bilateral

abdominal tenderness, and no nausea and vomiting. Combination of these features suggests PID as the more likely cause of the lower abdominal pain.

Acknowledgments

We thank Mr Riki Tanaka for his excellent secretarial support. Drs Morishita and Tokuda created the study design. Drs Morishita, Gushimiyagi, and Hashiguchi collected and input the data. Drs Morishita, Hashiguchi, and Tokuda performed statistical analysis and wrote the manuscript.

References

1. Bongard F, Landers DV, Lewis F. Differential diagnosis of appendi- citis and pelvic inflammatory disease A Prospective analysis. Am J Surg 1985;150(1):90 - 6.
2. Webster DP, Schneider CN, Cheche S, Daar AA, Miller G. Differen- tiating acute appendicitis from pelvic inflammatory disease in women of childbearing age. Am J Emerg Med 1993;11(6):569 - 72.
3. Dahlberg DL, Lee C, Fenlon T, Willoughby D. Differential diagnosis of abdominal pain in women of childbearing age. Appendicitis or pelvic inflammatory disease? Adv Nurse Pract 2004;12(1):40 - 5 [quiz 45-6].
4. Jacobson L, Westrom L. Objectivized diagnosis of acute pelvic inflammatory disease. Diagnostic and prognostic value of routine laparoscopy. Am J Obstet Gynecol 1969;105(7):1088 - 98.
5. Rothrock SG, Green SM, Dobson M, Colucciello SA, Simmons CM. Misdiagnosis of appendicitis in nonpregnant women of childbearing age. J Emerg Med 1995;13(1):1 - 8.
6. Chang FC, Hogle HH, Welling DR. The fate of the negative appendix. Am J Surg 1973;126(6):752 - 4.
7. Gilmore OJ, Browett JP, Griffin PH, et al. Appendicitis and mimicking conditions. A prospective study. Lancet 1975;2(7932): 421 - 4.
8. Kahn JG, Walker CK, Washington AE, Landers DV, Sweet RL. Diagnosing pelvic inflammatory disease. A comprehensive analysis and considerations for developing a new model. JAMA 1991;266(18): 2594 - 604.
9. Nakhgevany KB, Clarke LE. Acute appendicitis in women of childbearing age. Arch Surg 1986;121(9):1053 - 5.
10. Lewis FR, Holcroft JW, Boey J, Dunphy E. Appendicitis. A critical review of diagnosis and treatment in 1,000 cases. Arch Surg 1975; 110(5):677 - 84.
11. Mueller BA, Daling JR, Moore DE, et al. Appendectomy and the risk of tubal infertility. N Engl J Med 1986;315(24):1506 - 8.
12. Arnbjornsson E. Varying frequency of acute appendicitis in dif- ferent phases of the menstrual cycle. Surg Gynecol Obstet 1982; 155(5):709 - 11.
13. White JJ, Santillana M, Haller Jr JA. Intensive in-hospital observation: a safe way to decrease unnecessary appendectomy. Am Surg 1975; 41(12):793 - 8.
14. Leape LL, Ramenofsky ML. Laparoscopy for questionable appendi- citis: can it reduce the Negative appendectomy rate? Ann Surg 1980; 191(4):410 - 3.
15. Deutsch AA, Zelikovsky A, Reiss R. Laparoscopy in the prevention of unnecessary appendicectomies: a prospective study. Br J Surg 1982;69(6):336 - 7.
16. Incesu L, Coskun A, Selcuk MB, Akan H, Sozubir S, Bernay F. Acute appendicitis: MR imaging and sonographic correlation. AJR Am J Roentgenol 1997;168(3):669 - 74.
17. Nitta N, Takahashi M, Furukawa A, Murata K, Mori M, Fukushima

M. MR imaging of the normal appendix and acute appendicitis. J Magn Reson Imaging 2005;21(2):156 - 65.

1. Togawa A, Kimura F, Chiku T, Sano W, Tashiro T, Miyazaki M. Simple way to improve accuracy in diagnosis of quadrant inflamma- tory disease: how to avoid adverse laparotomy by using Plain CT. Hepatogastroenterology 2005;52(61):135 - 8.
2. Malik AA, Wani NA. Continuing diagnostic challenge of acute appendicitis: evaluation through modified Alvarado score. Aust N Z J Surg 1998;68(7):504 - 5.
3. Ratelle S, Yokoe D, Blejan C, et al. Predictive value of clinical Diagnostic codes for the CDC case definition of pelvic inflammatory disease (PID): implications for surveillance. Sex Transm Dis 2003;30(11):866 - 70.
4. Gray H, Standring S, Ellis H, Berkovitz BKB. Gray’s anatomy: the anatomical basis of clinical practice. 39th ed. Edinburgh7 Elsevier Churchill Livingstone; 2005.
5. Doherty GM, Lewis Jr FR. Appendicitis: continuing diagnostic challenge. Emerg Med Clin North Am 1989;7(3):537 - 53.
6. Robinson JA, Burch BH. An assessment of the value of the menstrual history in differentiating acute appendicitis from pelvic inflammatory disease. Surg Gynecol Obstet 1984;159(2):149 - 52.