Article, Surgery

Receiver operating characteristic analysis of leukocyte counts in operations for suspected appendicitis

Original Contribution

Receiver operating characteristic analysis of leukocyte counts in operations for suspected appendicitis

Mehmet Keskek MDa,?, Mesut Tez MDb, Omer Yoldas MDa, Atahan Acar MDa,

Ozgur Akgul MDa, Erdal Gocmen MDa, Mahmut Koc MDa

aFifth Department of Surgery, Ankara Numune Training and Research Hospital, 06500 Ankara, Turkey

bGazi University School of Medicine Department of General Surgery, 06500 Ankara, Turkey

Received 5 October 2007; revised 26 October 2007; accepted 27 October 2007

Abstract

Study Objective: There has been no clear-cut value of the white blood cell count in differential diagnosis of acute appendicitis in emergency medicine. Therefore, the aim of this study was to assess the preoperative WBC counts in 3 groups of patients operated on for a clinical suspicion of acute appendicitis with different findings at appendectomy: uninflamed appendix, uncomplicated acute appendicitis, or complicated acute appendicitis.

Methods: The medical records of 540 patients who underwent appendectomy for suspected acute appendicitis during a 17-month period were retrospectively reviewed. Data for 3 groups of patients were analyzed to calculate the sensitivity and specificity of the WBC count in the diagnosis of acute appendicitis, and we calculated likelihood ratios for 2 cut-points with either high sensitivity or high specificity. Receiver operating characteristic curves were used to evaluate the WBC count in relation to the true diagnosis and severity of acute appendicitis.

Results: We were able to identify patients with appendicitis on a statistically significant level using receiver operating characteristic curves of WBC counts (area under the curve = 0.756, P b .001), but the WBC count had no diagnostic value in differentiating between uncomplicated and complicated groups (area under the curve =0.55, P = .086). Likelihood ratio ranged from 1.79 (95% confidence interval, 1.17-2.3) for WBC count of more than 10500 cells/mm3 to 3.20 (95% confidence interval, 2.72-3.24) for WBC count of more than 14300 cells/mm3.

Conclusion: White blood cell count is helpful in the diagnosis and exclusion of appendicitis. However, there is no value to differentiate advanced appendicitis.

(C) 2008

Introduction

Acute appendicitis is one of the most common Abdominal emergencies. The clinical diagnosis is often difficult even for experienced surgeons, however, as evidenced by the high

* Corresponding author. Mesut Tez, 5.cad.10/3 Bahcelievler, 06500 Ankara, Turkey. Tel.: +90 312 215 38 34; fax: +90 312 310 34 60.

E-mail address: [email protected] (M. Keskek).

rate of negative explorations, which commonly reaches 20% to 30% [1]. A Delay in diagnosis of Acute appendicitis is associated with increased risk of perforation and further complications. Many attempts have been made to determine ways of decreasing the negative laparotomy rate after a clinical suspicion of AA. However, despite complete clinical history, physical examination, and the usual laboratory studies, clear decision aids for detection of early AA are lacking. Laboratory measurements such as WBC count,

0735-6757/$ – see front matter (C) 2008 doi:10.1016/j.ajem.2007.10.036

neutrophil percentage, and C-reactive protein (CRP) con- centration are commonly used as diagnostic aids in patients with suspected acute appendicitis. Patients with acute appendicitis often present with a raised WBC count, but owing to a considerable overlap with other Inflammatory processes mimicking acute appendicitis, the diagnostic value of the WBC count remains controversial [2].

Therefore, the aim of the present study was to assess the preoperative WBC counts in 3 groups of patients operated on for a clinical suspicion of acute appendicitis with different findings at appendectomy: uninflamed appendix, uncompli- cated acute appendicitis, or complicated acute appendicitis.

Patients and methods

Between August 2005 and December 2006, a total of

540 patients underwent appendectomy with a clinical diagnosis of acute appendicitis at the Emergency Depart- ment, Ankara Numune Training and Research Hospital (Ankara, Turkey). The clinical diagnosis was established preoperatively by clinical history, physical examination, and laboratory tests including WBC count and neutrophil percentage. Demographic, surgical, and histopathologic variables were recorded retrospectively. Patients who underwent appendectomy with the preoperative diagnosis of acute appendicitis were divided into 3 groups: group 1, 102 consecutive patients who had uninflamed appendix at appendectomy; group 2, 284 consecutive patients who had uncomplicated acute appendicitis at appendectomy; group 3, 154 consecutive patients who had complicated acute appendicitis (perforated appendix or periappendicular abscess) at appendectomy.

Laboratory tests were carried out on blood samples obtained from the patients on admission to hospital. In those patients who had repetitive tests before surgery, the final test data were used for statistical analysis. The WBC count and differential Neutrophil count were measured by an automated hematology analyzer (Coulter Hmx; Beckman Coulter [UK] Ltd, Buckinghamshire, England). The upper limits of the Reference interval for WBC count and neutrophil percentage were 10 x 109/L and 74%, respectively. All the excised appendices were sent for histopathologic examination, and the final diagnosis of acute appendicitis was based on histologic examination of the appendix.

    1. Statistical analysis

The data were expressed as mean +- SD. Statistical analysis was carried out using SPSS 11.0 for Windows (SPSS Inc, Chicago, Ill). Statistical analysis was done with 1-way analysis of variance to compare clinical data and laboratory values between the groups. For comparison of 2 groups an unpaired, 2-sided Student t test was used to further evaluate WBC count in studied groups. We measured the Clinical performance of WBC count using receiver operating characteristic curves and calculated likelihood ratios for 2 cut-points with either high sensitivity or high specificity.

A P value of b.05 was considered to be statistically significant.

Results

Demographic characteristics of patients were identical for all 3 groups (Table 1). The mean WBC counts were also shown in Table 1. The mean WBC count was higher in patients with complicated appendicitis than in those with uncomplicated and uninflamed appendix. The patients in groups 2 and 3 had a higher WBC count than group 1 (P b

.001), but the difference between groups 2 and 3 was not significant (P = .156). There was no difference between the neutrophil percentages of groups (P N .05).

Receiver operating characteristic curves of WBC counts was used to identify patients with appendicitis (group 1 vs 2 and 3) on a statistically significant level (area under the curve [AUC] = 0.756, P b .001) (Fig. 1).

Likelihood ratio ranged from 1.79 (95% confidence interval [CI], 1.17-2.3) for WBC count of more than 10500 cells/mm3 to 3.20 (95% CI, 2.72-3.24) for WBC count of more than 14300 cells/mm3 (Table 2).

On the other hand, the WBC count had no diagnostic value in differentiating between uncomplicated and compli- cated groups (AUC = 0.55, P = .086).

Discussion

Although the incidence of acute appendicitis appears to have been waning slightly over the past few decades, it

Table 1 Demographic characteristics and leukocyte counts of patient cohorts with preoperative diagnosis of acute appendicitis

Characteristics

Group 1

(n = 102)

Group 2

(n = 284)

Group 3

(n = 154)

P

Sex (female/male)

63/39

91/193

48/106

NS

Age (y)

31.8 +- 12.4

29.8 +- 11.9

31.8 +- 14.2

NS

WBC count (per mm3)

11441 +- 3,830

15246 +- 4746

15896 +- 4712

b.001

Data are presented as mean +- SD. NS indicates nonsignificant.

Fig. 1 Receiver operating characteristic curves for the WBC counts to identify patients with appendicitis (group 1 vs groups 2 and 3) (AUC = 0.756, P b .001).

remains a frequent cause of acute abdominal pain and urgent operative intervention. If untreated, acute appendi- citis will progress from inflammation to perforation with abscess formation or diffuse peritonitis, making timely operative intervention imperative. Generally, the morbidity and mortality of missing a case of acute appendicitis with subsequent peritonitis or abscess formation far outweigh the complications associated with a negative appendicect- omy [3].

The analysis of a patient with possible appendicitis can be divided into 3 parts: history, physical examination, and routine laboratory and x-ray tests. The leukocyte count is the test probably most often used to diagnose acute appendicitis. Several reports suggest that an elevated leukocyte count is usually the earliest laboratory test to indicate appendiceal inflammation, and most of the patients with acute appendicitis present with leukocytosis [4-6]. Both leukocyte count and neutrophil percentage are not specific for acute appendicitis. Therefore, surgeons

Table 2 The optimal cutoff value of WBC count in appendicitis

still have difficulty in diagnosing appendicitis accurately, and controversy exists regarding the relative usefulness of laboratory tests in the diagnosis of acute appendicitis. Leukocytosis is a nonspecific reaction caused by acute or chronic inflammation, acute physical or emotional stress, and several other conditions. This is reflected in numerous reports by an acceptable sensitivity (79%-93%) but a rather low specificity for AA [5]. Accordingly, the optimal diagnostic test for acute appendicitis should at least be highly sensitive to avoid false-negative results. The diagnostic accuracy of a test can be improved by changing the cutoff point above which the test result is considered positive. However, an arbitrarily chosen cutoff value, unless having been widely accepted by clinicians, would be of little significance in clinical practice.

Contrary to descriptive and comparing statistical meth- ods, analysis of ROC curves allows the estimation and verification of diagnostic suitability of diagnostic para- meters. The ROC curves provide an alternative to sensitivity and specificity that allows the examination of a test’s ability to discriminate between 2 populations regardless the cutoff level selected [7]. Therefore, we included this analysis to assess the overall diagnostic value of select parameters in clinical practice.

Part of the difficulty in comparing results arises from the lack of a definite end point in patient classification. The current study population was restricted to a defined group of patients who underwent surgery with a clinical diagnosis of acute appendicitis to eliminate this problem. The diagnostic end point was the presence or absence of histologically verified inflammation of the appendix, and presence or absence of complicated appendicitis. Using ROC curves, we tried to find a WBC count that has high sensitivity and specificity in the diagnosis of acute appendicitis. Our results showed elevated WBC count in all 3 groups of patients. Patients with appendicitis, either complicated or not, had higher WBC counts than those with negative laparatomies. On the other hand, the current study did not show strong evidence to support the use of WBC counts in differentiating between uncompli- cated and complicated cases. This was also in accordance with the previous reports that leukocyte count does not increase any more during protracted inflammation such as in the case of appendiceal perforation or abscess formation [8]. Recently, Okamoto et al [9] studied

WBC count

Sensitivity (95% CI)

Specificity (95% CI)

LR+ (95% CI)

LR- (95% CI)

N10500

84.00 (77.1-89.5)

53.13 (34.8-70.9)

1.79 (1.17-2.3)

0.30 (0.19-0.4)

N11500

78.00 (70.5-84.3)

62.50 (43.7-78.9)

2.13 (1.92-2.30)

0.32 (0.21-0.39)

N12850

68.00 (59.9-75.4)

75.00 (56.6-88.5)

2.72 (2.39-3.01)

0.43 (0.32-0.44)

N13400

61.33 (53.0-69.2)

78.12 (60.0-90.7)

2.86 (2.48-2.968)

0.45 (0.34-0.52)

N14300

56.67 (48.3-64.7)

81.25 (63.6-92.7)

3.20 (2.72-3.24)

0.49 (0.38-0.55)

LR+ indicates positive likelihood ratio; LR-, negative likelihood ratio.

WBC count for the differential diagnosis of simple and advanced appendicitis in pediatric patients using the ROC curve analysis. Their result was very similar to ours in that they found the WBC count had no diagnostic value in differentiating between the 2 groups (AUC = 0.563, P = .124).

A few studies evaluated the role of elevated WBC count in the diagnosis of AA; these studies had high sensitivity (71.4%-85.8%) but low specificity (31.9%-62.1%) for WBC count in the diagnosis of AA because they have used no specific cutoff value [1,10,11]. Korner et al [12] studied the role of elevated WBC count in the diagnosis of AA using ROC curves analysis and found 12300 cells/mm3 as the cutoff value of WBC count.

C-reactive protein is also an important serum inflamma- tory marker in the diagnosis of appendicitis in the pediatric age group. After 6 to 12 hours of inflammation, the concentration begins to rise and may increase a hundredfold [13]. It has been demonstrated that in patients whose symptoms had lasted less than 24 hours, WBC count had a high sensitivity, whereas in those in whom the symptoms had lasted more than 24 hours, CRP had a high sensitivity [6]. However, in a meta-analysis, CRP has been shown to have a medium sensitivity (53%-88%) and specificity (46%-82%) for appendicitis [14]. Because our patients are adults and most of them applied to a hospital in the first 24 hours of symptoms, we did not evaluate the CRP levels in this study.

One feature of the study is the selection of only patients who went to the OR. There is good internal validity with respect to the diagnosis. A negative aspect is that there is selection bias and a threat to generalizability in that there was a large subset, probably with low WBC count, that did not go to the OR because the WBC count was low (workup bias).

In conclusion, we think that although elevated WBC count is a nonspecific inflammatory marker in most of inflammatory diseases, high WBC count is helpful in the diagnosis and exclusion of appendicitis.

References

  1. Yang HR, Wang YC, Chung PK, Chen WK, Jeng LB, Chen RJ. Laboratory tests in patients with acute appendicitis. ANZ J Surg 2006; 76:71-4.
  2. Raftery AT. The value of the white blood cell count in the diagnosis of acute appendicitis. Br J Surg 1976;63:143-4.
  3. Jess P, Bjerregaard B, Brynitz S, Holst-Christensen J, Kalaja E, Lund- Kristensen J. Acute appendicitis. Prospective trial concerning diagnostic accuracy and complications. Am J Surg 1981;141:232-4.
  4. Dueholm S, Bagi P, Bud M. Laboratory aid in the diagnosis of acute appendicitis. A blinded, prospective trial concerning diagnostic value of leukocyte count, neutrophil differential count, and C-reactive protein. Dis Colon Rectum 1989;32:855-9.
  5. Hallan S, Asberg A, Edna TH. Additional value of Biochemical tests in suspected acute appendicitis. Eur J Surg 1997;163:533-8.
  6. Eriksson S, Granstrom L, Carlstrom A. The diagnostic value of repetitive preoperative analyses of C-reactive protein and total Leucocyte count in patients with suspected acute appendicitis. Scand J Gastroenterol 1994;29:1145-9.
  7. Beck JR, Shultz EK. The use of relative operating characteristic (ROC) curves in test performance evaluation. Arch Pathol Lab Med 1986;110: 13-20.
  8. Gronroos JM, Gronroos P. Leucocyte count and C-reactive protein in the diagnosis of acute appendicitis. Br J Surg 1999;86:501-4.
  9. Okamoto T, Sano K, Ogasahara K. Receiver-operating characteristic analysis of leukocyte counts and Serum C-reactive protein levels in children with advanced appendicitis. Surg Today 2006;36:515-8.
  10. Yang HR, Wang YC, Chung PK, Chen WK, Jeng LB, Chen RJ. Role of leukocyte count, neutrophil percentage, and C-reactive protein in the diagnosis of acute appendicitis in the elderly. Am Surg 2005;71: 344-7.
  11. Khan MN, Davie E, Irshad K. The role of white cell count and C-reactive protein in the diagnosis of acute appendicitis. J Ayub Med Coll Abbottabad 2004;16:17-9.
  12. Korner H, Soreide JA, Sondenaa K. Diagnostic accuracy of inflammatory markers in patients operated on for suspected acute appendicitis: a receiver operating characteristic curve analysis. Eur J Surg 1999;165:679-85.
  13. Peltola H, Ahlqvist J, Rapola J, Rasanen J, Louhimo I, Saarinen M. C-reactive protein compared with white blood cell count and erythrocyte sedimentation rate in the diagnosis of acute appendicitis in children. Acta Chir Scand 1986;152:55-8.
  14. Hallan S, Asberg A. The accuracy of C-reactive protein in diagnosing acute appendicitis–a meta-analysis. Scand J Clin Lab Invest 1997;57: 373-80.