- Email: [email protected]
Correlating surgical and pathological diagnoses in pediatric appendicitis
Journal of Pediatric Surgery 50 (2015) 638–641
Contents lists available at ScienceDirect
Journal of Pediatric Surgery journal homepage: www.elsevier.com/locate/jpedsurg
Correlating surgical and pathological diagnoses in pediatric appendicitis☆ Sara C. Fallon a, Michael E. Kim a, Charlene A. Hallmark b, Jennifer L. Carpenter a, Karen W. Eldin c, Monica E. Lopez a, David E. Wesson a, Mary L. Brandt a, J. Ruben Rodriguez a,⁎ a b c
Texas Children’s Hospital, Division of Pediatric Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine Texas Children’s Hospital, Surgical Outcomes Center, Division of Pediatric Surgery Texas Children’s Hospital, Department of Pathology, Baylor College of Medicine
a r t i c l e
i n f o
Article history: Received 17 May 2014 Received in revised form 30 October 2014 Accepted 3 November 2014 Key words: Pediatrics Appendicitis Pathology Quality improvement
a b s t r a c t Background: The stratification of appendicitis into simple and complex variants has far-reaching implications. While the operative diagnosis made by the surgeon dictates clinical management, the pathologic diagnosis often differs and is frequently used for coding and reimbursement. The purpose of this study was to examine discrepancies between the operative and pathologic diagnoses with subsequent correlation to clinical outcomes. Methods: Patients with acute appendicitis from July 2011 to July 2012 were identified. Diagnoses included simple (normal, acute, and suppurative) and complex (gangrenous and perforated). We evaluated the inter-rater reliability between pathologic and operative diagnoses in the five appendicitis categories. Clinical outcomes of deep and superficial surgical site infections were evaluated according to the pathologic and surgical diagnosis. Results: During the study period, we identified 1166 patients with acute appendicitis. The surgeon and pathologist agreed on the specific diagnosis (acute, suppurative, gangrenous, perforated, normal) in 48% of patients (kappa 0.289, 95% CI 0.259–0.324, p = 0.001). Agreement on disease severity (simple vs. complex) improved to 82%. The operative diagnosis more accurately predicted infectious complications than the pathologic diagnoses. Conclusion: Significant discordance exists between surgical and pathologic diagnoses. While the relevance of this discordance to clinical outcomes is still not clear, a potential for incorrect hospital coding and subsequent reimbursement exists. Future quality improvement projects should focus on standardizing the surgical and pathologic diagnoses. © 2015 Elsevier Inc. All rights reserved.
1. Background Appendicitis is frequently cited as the most common indication for an abdominal operation in children. As such, children with appendicitis are generally treated under institution-specific management protocols at high volume centers [1–3]. These protocols typically dichotomize patient management into two treatment arms. Patients with simple, nonperforated appendicitis receive recovery care and are discharged within a day, whereas those with perforated, complex appendicitis frequently have more protracted hospital courses with intravenous antibiotic therapy. Furthermore, discharge protocols (ie, antibiotic use after hospitalization) are often dependent on these treatment arms. The decision for the treatment pathway is made at the time of operation. The criteria for which patients should be treated in the “simple” and “complex” categories differ at each institution [4]. Some treatment protocols define “complex, or perforated” appendicitis as those with a visible hole in the appendix. Others include patients with gangrene of the
☆ Disclosures: The authors have no sources of support or funding to disclose with regards to the preparation of this manuscript. ⁎ Corresponding author at: 6701 Fannin, Suite 1210, Houston, TX 77030. Tel.: +1 832 822 3135; fax: +1 832 825 3141. E-mail address: [email protected] (J. Ruben Rodriguez). http://dx.doi.org/10.1016/j.jpedsurg.2014.11.001 0022-3468/© 2015 Elsevier Inc. All rights reserved.
appendix (microperforation), but no visible transmural defect, in the complex treatment pathway [2]. The diagnosis made at the time of the pathologic examination is typically not used to make subsequent treatment decisions. At Texas Children’s Hospital, gangrenous appendicitis and perforated appendicitis are included in the “complex” disease category. During the hospital stay, a number of diagnoses reflecting the severity of the appendicitis are documented in the patient’s chart. The patient is assigned an intraoperative, surgeon-determined diagnosis and placed in a treatment protocol. The appendix is routinely submitted for pathologic examination and diagnosis. The post operative diagnosis is documented in the discharge summary. And finally, hospital coders assign an official ICD-9 code for billing purposes [C. Hallmark, November 2012]. At our institution, an 8%–10% error rate was noted between the surgeon’s operative diagnosis and the official, hospital ICD-9 diagnosis. This was observed during the development of internal cohorts of simple and complex appendicitis patients for the purposes of tracking patient data and outcomes in real time through an electronic data warehouse. When explanations for this discrepancy were investigated, the hospital coding team indicated that they review the pathologic diagnosis to help assign the final ICD-9 code for billing purposes [C. Hallmark, November 2012]. It appeared as though the use of the pathologic diagnosis in the coding of appendicitis was contributing to this discrepancy between the patient’s treated severity of appendicitis, and the ICD-9 code that was being used
S.C. Fallon et al. / Journal of Pediatric Surgery 50 (2015) 638–641
for reimbursement. Therefore, the purpose of this study was to investigate the discrepancies between the pathologic and operative diagnoses, and determine how these diagnoses correlated with clinical outcomes. 2. Methods
639
wound infection in the 30-day post-operative period and the patient’s hospital length of stay. IAA was defined as a post-operative fluid collection detected on imaging that was treated with either percutaneous drainage or intravenous antibiotic therapy. A superficial surgical site infection was defined as cellulitis and/or purulence that required incision and drainage or antibiotic therapy.
2.1. Texas Children’s Hospital Appendicitis Management Algorithm 2.3. Statistical analysis Patients with acute appendicitis are managed according to an institutional protocol that was developed in 2009 and has been under study since. The details of the protocol have been previously reported [1,5]. Briefly, patients with simple appendicitis are placed in 23-h observation without further antibiotic therapy and remain in the hospital while recovering from the operation. Those with complex disease are placed on piperacillin–tazobactam until clinical criteria for discharge are met. At that point, further management is based on the presence of leukocytosis. For the purposes of determining the management algorithm, the operative diagnosis is used, with the following definitions: Acute appendicitis is red, swollen, and inflamed without evidence of surrounding fibropurulent exudate. Suppurative appendicitis has fibropurulent exudate without odor; cloudy fluid may be present. Gangrenous appendicitis has an ischemic, discolored wall without evidence of a hole or frank pus. Perforated appendicitis includes those with a hole, frank pus, or a fecalith. The pathologic definitions of appendicitis used at our institution are as follows: Acute appendicitis is transmural (muscularis propria) acute inflammation with or without mucosal ulceration without periappendicitis or serositis. Suppurative appendicitis is acute appendicitis with serositis or periappendicitis. Acute necrotizing/gangrenous appendicitis is all the criteria for acute appendicitis but also demonstrates any focus of transmural myonecrosis of the muscularis propria with an intact serosa. Perforations can be gross (macroscopic) and characterized by transmural defect (hole). The defect may be obvious (large defect or fragmented appendix) or only appreciated by squeezing the appendix and noting extrusion of luminal contents (pinpoint hole). A microscopic perforation is transmural myonecrosis with serosal disruption in an appendix that didn’t have a macroscopic perforation. 2.2. Study design Patients with acute appendicitis from July 2011 to July 2012 were identified through an electronic data warehouse, based on ICD-9 code. IRB approval was obtained. Patients treated with an interval appendectomy were excluded. Patients were categorized as having either a normal appendix, or acute, suppurative, gangrenous, or perforated appendicitis. The diagnosis was confirmed by review of the operative note. Patients were also stratified into having simple (normal, acute, or suppurative) or complex (gangrenous or perforated) disease. Patients with pathologic findings other than a normal appendix or a variant of appendicitis (eg, lymphoid hyperplasia, carcinoid tumor) were excluded from further analysis. In these cases, the appendix specimen was abnormal, but was not amenable to the classification scheme used intraoperatively. A chart review was performed to determine if the patient had a post-operative intraabdominal abscess (IAA) or
The inter-rater reliability between pathologic and operative diagnoses in the five appendicitis categories was calculated using the kappa statistic. The incidence of superficial and deep surgical site infections was calculated according to the pathologic and surgical diagnosis and compared using a chi-square test. The length of stay was also reported based on the operative and pathologic diagnoses. 3. Results We identified 1166 patients meeting the study inclusion criteria. The pathologic and operative diagnoses were identical in 560/1166 cases (48%) (Table 1). Thirty-four patients had an alternative finding during pathologic examination that was not a variant of appendicitis (Table 2). Using the intraoperative diagnosis as the gold standard, agreement was the highest in patients with perforated disease (282/ 378, 75%), and lowest in those with gangrenous appendicitis (13/78, 17%). The negative appendectomy rate based on the operative diagnosis was 0.4%, and based on the pathologic diagnosis was 4%. The kappa statistic for inter-rater reliability using these five diagnostic categories indicated a poor–fair agreement between the pathologist and surgeon (k = 0.289, 95% CI 0.259–0.324, p = 0.001). When evaluating concordance between the pathologic and intraoperative diagnoses of either simple versus complex disease, agreement increased to 82%. Of the 18% that had discordant findings, in 7% the surgeon found an increased severity compared to the pathologist, and in 11%, the surgeon found less severe disease than the pathologist. The clinical outcomes (IAA and wound infection) were then determined based on the pathologic and intraoperative diagnoses (Table 3). In cases where both the pathologist and surgeon agreed that the disease severity was simple, a 0.8% SSI rate was noted, which is consistent with known institutional outcomes of SSI rates of 1%–2% in simple disease. When both agreed that the disease severity was complex, a 17% complication rate was found, which is consistent with known SSI rates of 15%– 20% in complex disease. When the pathologic diagnosis underestimated disease severity compared to the operative diagnosis, the SSI rate was 10%, which is more consistent with complex disease. Conversely, when the pathologic diagnosis overestimated the disease severity, the SSI rate was 2.3%, which is more consistent with simple disease. Viewed another way, when the operative diagnosis is used as the gold standard, when simple or complex disease is found, there is no statistically significant difference in SSI rate whether the pathologic disease is simple or complex. Conversely, when the pathologic diagnosis is used as the gold standard, a significant difference does exist in the SSI rate, indicating that the operative diagnosis is more predictive of infection rate outcomes in simple, compared to complex, disease.
Table 1 Comparison of intraoperative and pathologic diagnoses. Pathologic Diagnosis
Operative Diagnosis
Normal Acute Suppurative Gangrenous Perforated Total
Normal
Acute
Suppurative
Gangrenous
Perforated
Total
4 38 0 0 2 44
1 171 39 6 17 234
0 236 90 21 34 381
0 36 22 13 43 114
0 44 29 38 282 393
5 525 180 78 378 1166
640
S.C. Fallon et al. / Journal of Pediatric Surgery 50 (2015) 638–641
Table 2 Pathologic findings other than a normal appendix or acute appendicitis (n = 34).
Operative Simple (n, % with SSI) LOS (Median, Range) Operative Complex (n, % with SSI) LOS (Median, Range) Total
Pathology Simple
Pathology Complex
N, (% with SSI)
N, (% with SSI)
LOS (Median, Range)
LOS (Median, Range)
579 (n = 5, 0.8%) 1 (1–31)
131 (n = 3, 2.3%) 2 (1–16)
80 (n = 8, 10%) 5 (1–22) 659 (n = 13, 2.0%)
376 (n = 63, 17%) 7 (1–29) 507 (13%)
The median length of stay in patients with simple disease by both operative and pathologic diagnosis was one day (Table 3). When the operative diagnosis was simple and the pathologic diagnosis was complex, this only increased to 2 days. If complex disease was reported by both the surgeon and pathologist, the median length of stay was 7 days; this decreased to 5 days when the pathologic diagnosis was simple, and the operative diagnosis was complex. 4. Discussion Determining and classifying appendicitis severity remain difficult to standardize, as there can be significant provider and institutional differences. One study by St. Peter et al. recommends that patients are defined as having perforated appendicitis if there is a visible hole in the appendix, or if a fecalith is present in the abdomen [4]. The study found that this classification scheme was predictive of abscess formation, and likely led to accurate identification of which patients would benefit from additional antibiotic therapy [4]. These authors continue to recommend a dichotomous approach to appendicitis treatment. Another study by Emil et al. prospectively evaluated outcomes in patients with gangrenous appendicitis, using the definition of a black or gray discoloration of the appendix without a visible hole, fecalith, pus, or appendiceal leak [6]. Based on these findings, the authors recommend that gangrenous appendicitis should perhaps be treated with a third management algorithm that has a similar length of stay to simple appendicitis but includes post-operative antibiotic administration [6]. At our own institution, we have adopted similar intraoperative definitions for gangrenous and perforated appendicitis, and at this time both are treated under the advanced appendicitis treatment algorithm, including intravenous piperacillin–tazobactam, which is continued until clinical discharge criteria are met and leukocytosis (b 12 k) normalizes [1,5]. This study highlights discrepancies in the assignment of diagnoses that may have implications beyond clinical outcomes. The findings in this study differ somewhat from other pediatric studies of surgeon and pathologist congruence regarding the appendicitis diagnosis [7]. Bliss et al. found that the concordance rates between the surgeon and pathologist were higher in those with acute appendicitis (~90%), compared to perforated appendicitis (~55%) [7]. We found that the concordance was actually much higher in patients with perforated disease (75%) compared to simple (33%), or even simple and suppurative disease (37%). Additionally, their study found that when the pathologist and surgeon were discordant, the pathologist’s diagnosis was more
Total
710 (n = 8, 1.1%) 2 (1–31) 456 (n = 15.5%) 6 (1–29)
predictive of the clinical outcome (complication rate) than the surgeon’s diagnosis. In adult studies, alternative categorization schemes have been proposed that are based on a combination of the appendix appearance and the associated intraabdominal extension of inflammation and infection. One study also found fair to moderate concordance between the operative and pathologic diagnoses (k = 0.39, 95% CI 0.23–0.55) [8,9]. In our study, the operative diagnosis appeared to better predict surgical site infection rates than the pathologic diagnosis. When the pathologist’s diagnosis is used as the gold standard for clinical management, there is a significant difference in outcomes between those the surgeon determined had simple or perforated disease. When the operative diagnosis is used as the gold standard, there is no significant difference in the clinical outcomes based on the pathologist’s diagnosis. Therefore, this supports our practice to use the surgeon’s diagnosis as the gold standard for determining clinical management. Of note, we did find that the surgeon reporting of a normal appendix at the time of appendectomy was lower than the pathologist rate (b1% vs. 4%). We assume this a combination of surrounding peritoneal inflammation not assessed in the specimen or surgeon difficulty with diagnosing a normal appendix versus early appendicitis on gross examination. This also may be the result of the limits of a retrospective review, where the surgeon was not asked directly to comment on the disease severity. Additionally, our findings differed from those reported by Emil et al. with respect to gangrenous appendicitis [2]. Gangrenous appendicitis is diagnosed at a much lower rate in our study by the surgeon (6.8% vs. 17%). They report a 62% histopathologic agreement rate between the surgeon and the pathologist in cases of gangrenous appendicitis, which is much improved over our own 13% [2]. With this high rate of concordance and reproducibility, these authors recommend a separate treatment pathway for those with gangrenous appendicitis. Due to concerns over the reproducibility of these findings at our own institution, as well as concerns for disruption of the integrity of the appendiceal wall by bacteria with subsequent intraperitoneal contamination requiring antibiotics, we continue to use a dichotomous treatment pathway with gangrenous patients treated in the “complex” arm, understanding that these patients will likely have slightly short hospital courses and lower complication rates than their grossly perforated counterparts. We recognize that our treatment of gangrenous appendicitis in the same protocol as the perforated group is based on observation from an internal review of our own institutional data, which demonstrated that these patients have a return of bowel function and surgical site infection rate that is more similar to those with perforated disease. To
Table 3 Correlating pathologic and intraoperative diagnoses to clinical outcomes. Pathologic Finding
Number of Patients
Intraoperative Diagnosis
Lymphoid Hyperplasia Acute or Chronic Serositis Fibrous Luminal Obliteration Chronic inflammation/appendicitis Carcinoid Tumor Lympoplasmacytic infiltrate with mucosal erosion and crypt abscesses
21 5 3 2 2 1
All simple appendicitis 2 simple, 1 suppurative, 2 perforated All simple appendicitis 1 suppurative, 1 perforated 1 simple, 1 perforated Suppurative
S.C. Fallon et al. / Journal of Pediatric Surgery 50 (2015) 638–641
address this, we have an ongoing prospective observational study to assess these clinical questions and determine the optimal treatment algorithm for patients with gangrenous appendicitis [2,5]. The 34 (or 2.8%) patients with other findings at the time of pathologic examination remain a challenge to assess. While the majority appeared to have a pathology that did warrant appendectomy, the clinical course of the patient had the appendix not been removed remains uncertain. These patients did not have any post-operative complications, and generally had a clinical course similar to those with simple, acute appendicitis. The two patients who were found to have a carcinoid tumor support the continued, routine examination of the appendix specimen. This study highlights the need for quality improvement at a hospital level with regards to the diagnosis and categorization of appendicitis. A 10% error rate in the coding for the severity of appendicitis could potentially have reimbursement consequences, particularly as bundled care payments based on disease severity become the standard. However, we were unable to measure the direct cost effect of this discrepancy in the current study. The results of our study have prompted a reexamination of our coding practice, as it does not accurately predict the clinical course. In order to assist with diagnostic clarity in the operative note for the purposes of correct assignation of ICD-9 codes, we have implemented an operative note template that allows the surgeon to specify the correct ICD-9 code for the procedure, and to have the operative diagnosis (normal, acute, suppurative, gangrenous, or perforated) clearly stated within the text of the note. The effect of this standardization is the subject of an ongoing quality improvement study to enhance the homogeneity of the appendicitis cohort used to track outcomes within our electronic data warehouse.
641
In conclusion, the operative diagnosis at the time of appendectomy appears to be more predictive of clinical outcome than the pathologic diagnosis. Pathologic examination of the appendix remains important, as we did identify two cases of carcinoid tumor in a one-year time frame and a negative appendectomy rate of 4%. And finally, attention to detail regarding provider and hospital-level diagnostic coding can reveal errors that require intervention for research, treatment, and reimbursement purposes. References [1] Fallon SC, Brandt ML, Hassan SF, et al. Evaluating the effectiveness of a discharge protocol for children with advanced appendicitis. J Surg Res 2013;184(1):347–51. [2] Emil S, Laberge JM, Mikhail P, et al. Appendicitis in children: a ten-year update of therapeutic recommendations. J Pediatr Surg 2003;38(2):236–42. [3] Warner BW, Rich KA, Atherton H, et al. The sustained impact of an evidenced-based clinical pathway for acute appendicitis. Semin Pediatr Surg 2002;11(1):29–35. [4] St Peter SD, Sharp SW, Holcomb III GW, et al. An evidence-based definition for perforated appendicitis derived from a prospective randomized trial. J Pediatr Surg 2008; 43(12):2242–5. [5] Fallon SC, Hassan SF, Larimer EL, et al. Modification of an evidence-based protocol for advanced appendicitis in children. J Surg Res 2013;185(1):273–7. [6] Emil S, Gaied F, Lo A, et al. Gangrenous appendicitis in children: a prospective evaluation of definition, bacteriology, histopathology, and outcomes. J Surg Res 2012; 177(1):123–6. [7] Bliss D, McKee J, Cho D, et al. Discordance of the pediatric surgeon's intraoperative assessment of pediatric appendicitis with the pathologists report. J Pediatr Surg 2010; 45(7):1398–403. [8] Gomes CA, Nunes TA, Fonseca Chebli JM, et al. Laparoscopy grading system of acute appendicitis: new insight for future trials. Surg Laparosc Endosc Percutan Tech 2012;22(5):463–6. [9] Garst GC, Moore EE, Banerjee MN, et al. Acute appendicitis: a disease severity score for the acute care surgeon. J Trauma Acute Care Surg 2013;74(1):32–6.
Contents lists available at ScienceDirect
Journal of Pediatric Surgery journal homepage: www.elsevier.com/locate/jpedsurg
Correlating surgical and pathological diagnoses in pediatric appendicitis☆ Sara C. Fallon a, Michael E. Kim a, Charlene A. Hallmark b, Jennifer L. Carpenter a, Karen W. Eldin c, Monica E. Lopez a, David E. Wesson a, Mary L. Brandt a, J. Ruben Rodriguez a,⁎ a b c
Texas Children’s Hospital, Division of Pediatric Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine Texas Children’s Hospital, Surgical Outcomes Center, Division of Pediatric Surgery Texas Children’s Hospital, Department of Pathology, Baylor College of Medicine
a r t i c l e
i n f o
Article history: Received 17 May 2014 Received in revised form 30 October 2014 Accepted 3 November 2014 Key words: Pediatrics Appendicitis Pathology Quality improvement
a b s t r a c t Background: The stratification of appendicitis into simple and complex variants has far-reaching implications. While the operative diagnosis made by the surgeon dictates clinical management, the pathologic diagnosis often differs and is frequently used for coding and reimbursement. The purpose of this study was to examine discrepancies between the operative and pathologic diagnoses with subsequent correlation to clinical outcomes. Methods: Patients with acute appendicitis from July 2011 to July 2012 were identified. Diagnoses included simple (normal, acute, and suppurative) and complex (gangrenous and perforated). We evaluated the inter-rater reliability between pathologic and operative diagnoses in the five appendicitis categories. Clinical outcomes of deep and superficial surgical site infections were evaluated according to the pathologic and surgical diagnosis. Results: During the study period, we identified 1166 patients with acute appendicitis. The surgeon and pathologist agreed on the specific diagnosis (acute, suppurative, gangrenous, perforated, normal) in 48% of patients (kappa 0.289, 95% CI 0.259–0.324, p = 0.001). Agreement on disease severity (simple vs. complex) improved to 82%. The operative diagnosis more accurately predicted infectious complications than the pathologic diagnoses. Conclusion: Significant discordance exists between surgical and pathologic diagnoses. While the relevance of this discordance to clinical outcomes is still not clear, a potential for incorrect hospital coding and subsequent reimbursement exists. Future quality improvement projects should focus on standardizing the surgical and pathologic diagnoses. © 2015 Elsevier Inc. All rights reserved.
1. Background Appendicitis is frequently cited as the most common indication for an abdominal operation in children. As such, children with appendicitis are generally treated under institution-specific management protocols at high volume centers [1–3]. These protocols typically dichotomize patient management into two treatment arms. Patients with simple, nonperforated appendicitis receive recovery care and are discharged within a day, whereas those with perforated, complex appendicitis frequently have more protracted hospital courses with intravenous antibiotic therapy. Furthermore, discharge protocols (ie, antibiotic use after hospitalization) are often dependent on these treatment arms. The decision for the treatment pathway is made at the time of operation. The criteria for which patients should be treated in the “simple” and “complex” categories differ at each institution [4]. Some treatment protocols define “complex, or perforated” appendicitis as those with a visible hole in the appendix. Others include patients with gangrene of the
☆ Disclosures: The authors have no sources of support or funding to disclose with regards to the preparation of this manuscript. ⁎ Corresponding author at: 6701 Fannin, Suite 1210, Houston, TX 77030. Tel.: +1 832 822 3135; fax: +1 832 825 3141. E-mail address: [email protected] (J. Ruben Rodriguez). http://dx.doi.org/10.1016/j.jpedsurg.2014.11.001 0022-3468/© 2015 Elsevier Inc. All rights reserved.
appendix (microperforation), but no visible transmural defect, in the complex treatment pathway [2]. The diagnosis made at the time of the pathologic examination is typically not used to make subsequent treatment decisions. At Texas Children’s Hospital, gangrenous appendicitis and perforated appendicitis are included in the “complex” disease category. During the hospital stay, a number of diagnoses reflecting the severity of the appendicitis are documented in the patient’s chart. The patient is assigned an intraoperative, surgeon-determined diagnosis and placed in a treatment protocol. The appendix is routinely submitted for pathologic examination and diagnosis. The post operative diagnosis is documented in the discharge summary. And finally, hospital coders assign an official ICD-9 code for billing purposes [C. Hallmark, November 2012]. At our institution, an 8%–10% error rate was noted between the surgeon’s operative diagnosis and the official, hospital ICD-9 diagnosis. This was observed during the development of internal cohorts of simple and complex appendicitis patients for the purposes of tracking patient data and outcomes in real time through an electronic data warehouse. When explanations for this discrepancy were investigated, the hospital coding team indicated that they review the pathologic diagnosis to help assign the final ICD-9 code for billing purposes [C. Hallmark, November 2012]. It appeared as though the use of the pathologic diagnosis in the coding of appendicitis was contributing to this discrepancy between the patient’s treated severity of appendicitis, and the ICD-9 code that was being used
S.C. Fallon et al. / Journal of Pediatric Surgery 50 (2015) 638–641
for reimbursement. Therefore, the purpose of this study was to investigate the discrepancies between the pathologic and operative diagnoses, and determine how these diagnoses correlated with clinical outcomes. 2. Methods
639
wound infection in the 30-day post-operative period and the patient’s hospital length of stay. IAA was defined as a post-operative fluid collection detected on imaging that was treated with either percutaneous drainage or intravenous antibiotic therapy. A superficial surgical site infection was defined as cellulitis and/or purulence that required incision and drainage or antibiotic therapy.
2.1. Texas Children’s Hospital Appendicitis Management Algorithm 2.3. Statistical analysis Patients with acute appendicitis are managed according to an institutional protocol that was developed in 2009 and has been under study since. The details of the protocol have been previously reported [1,5]. Briefly, patients with simple appendicitis are placed in 23-h observation without further antibiotic therapy and remain in the hospital while recovering from the operation. Those with complex disease are placed on piperacillin–tazobactam until clinical criteria for discharge are met. At that point, further management is based on the presence of leukocytosis. For the purposes of determining the management algorithm, the operative diagnosis is used, with the following definitions: Acute appendicitis is red, swollen, and inflamed without evidence of surrounding fibropurulent exudate. Suppurative appendicitis has fibropurulent exudate without odor; cloudy fluid may be present. Gangrenous appendicitis has an ischemic, discolored wall without evidence of a hole or frank pus. Perforated appendicitis includes those with a hole, frank pus, or a fecalith. The pathologic definitions of appendicitis used at our institution are as follows: Acute appendicitis is transmural (muscularis propria) acute inflammation with or without mucosal ulceration without periappendicitis or serositis. Suppurative appendicitis is acute appendicitis with serositis or periappendicitis. Acute necrotizing/gangrenous appendicitis is all the criteria for acute appendicitis but also demonstrates any focus of transmural myonecrosis of the muscularis propria with an intact serosa. Perforations can be gross (macroscopic) and characterized by transmural defect (hole). The defect may be obvious (large defect or fragmented appendix) or only appreciated by squeezing the appendix and noting extrusion of luminal contents (pinpoint hole). A microscopic perforation is transmural myonecrosis with serosal disruption in an appendix that didn’t have a macroscopic perforation. 2.2. Study design Patients with acute appendicitis from July 2011 to July 2012 were identified through an electronic data warehouse, based on ICD-9 code. IRB approval was obtained. Patients treated with an interval appendectomy were excluded. Patients were categorized as having either a normal appendix, or acute, suppurative, gangrenous, or perforated appendicitis. The diagnosis was confirmed by review of the operative note. Patients were also stratified into having simple (normal, acute, or suppurative) or complex (gangrenous or perforated) disease. Patients with pathologic findings other than a normal appendix or a variant of appendicitis (eg, lymphoid hyperplasia, carcinoid tumor) were excluded from further analysis. In these cases, the appendix specimen was abnormal, but was not amenable to the classification scheme used intraoperatively. A chart review was performed to determine if the patient had a post-operative intraabdominal abscess (IAA) or
The inter-rater reliability between pathologic and operative diagnoses in the five appendicitis categories was calculated using the kappa statistic. The incidence of superficial and deep surgical site infections was calculated according to the pathologic and surgical diagnosis and compared using a chi-square test. The length of stay was also reported based on the operative and pathologic diagnoses. 3. Results We identified 1166 patients meeting the study inclusion criteria. The pathologic and operative diagnoses were identical in 560/1166 cases (48%) (Table 1). Thirty-four patients had an alternative finding during pathologic examination that was not a variant of appendicitis (Table 2). Using the intraoperative diagnosis as the gold standard, agreement was the highest in patients with perforated disease (282/ 378, 75%), and lowest in those with gangrenous appendicitis (13/78, 17%). The negative appendectomy rate based on the operative diagnosis was 0.4%, and based on the pathologic diagnosis was 4%. The kappa statistic for inter-rater reliability using these five diagnostic categories indicated a poor–fair agreement between the pathologist and surgeon (k = 0.289, 95% CI 0.259–0.324, p = 0.001). When evaluating concordance between the pathologic and intraoperative diagnoses of either simple versus complex disease, agreement increased to 82%. Of the 18% that had discordant findings, in 7% the surgeon found an increased severity compared to the pathologist, and in 11%, the surgeon found less severe disease than the pathologist. The clinical outcomes (IAA and wound infection) were then determined based on the pathologic and intraoperative diagnoses (Table 3). In cases where both the pathologist and surgeon agreed that the disease severity was simple, a 0.8% SSI rate was noted, which is consistent with known institutional outcomes of SSI rates of 1%–2% in simple disease. When both agreed that the disease severity was complex, a 17% complication rate was found, which is consistent with known SSI rates of 15%– 20% in complex disease. When the pathologic diagnosis underestimated disease severity compared to the operative diagnosis, the SSI rate was 10%, which is more consistent with complex disease. Conversely, when the pathologic diagnosis overestimated the disease severity, the SSI rate was 2.3%, which is more consistent with simple disease. Viewed another way, when the operative diagnosis is used as the gold standard, when simple or complex disease is found, there is no statistically significant difference in SSI rate whether the pathologic disease is simple or complex. Conversely, when the pathologic diagnosis is used as the gold standard, a significant difference does exist in the SSI rate, indicating that the operative diagnosis is more predictive of infection rate outcomes in simple, compared to complex, disease.
Table 1 Comparison of intraoperative and pathologic diagnoses. Pathologic Diagnosis
Operative Diagnosis
Normal Acute Suppurative Gangrenous Perforated Total
Normal
Acute
Suppurative
Gangrenous
Perforated
Total
4 38 0 0 2 44
1 171 39 6 17 234
0 236 90 21 34 381
0 36 22 13 43 114
0 44 29 38 282 393
5 525 180 78 378 1166
640
S.C. Fallon et al. / Journal of Pediatric Surgery 50 (2015) 638–641
Table 2 Pathologic findings other than a normal appendix or acute appendicitis (n = 34).
Operative Simple (n, % with SSI) LOS (Median, Range) Operative Complex (n, % with SSI) LOS (Median, Range) Total
Pathology Simple
Pathology Complex
N, (% with SSI)
N, (% with SSI)
LOS (Median, Range)
LOS (Median, Range)
579 (n = 5, 0.8%) 1 (1–31)
131 (n = 3, 2.3%) 2 (1–16)
80 (n = 8, 10%) 5 (1–22) 659 (n = 13, 2.0%)
376 (n = 63, 17%) 7 (1–29) 507 (13%)
The median length of stay in patients with simple disease by both operative and pathologic diagnosis was one day (Table 3). When the operative diagnosis was simple and the pathologic diagnosis was complex, this only increased to 2 days. If complex disease was reported by both the surgeon and pathologist, the median length of stay was 7 days; this decreased to 5 days when the pathologic diagnosis was simple, and the operative diagnosis was complex. 4. Discussion Determining and classifying appendicitis severity remain difficult to standardize, as there can be significant provider and institutional differences. One study by St. Peter et al. recommends that patients are defined as having perforated appendicitis if there is a visible hole in the appendix, or if a fecalith is present in the abdomen [4]. The study found that this classification scheme was predictive of abscess formation, and likely led to accurate identification of which patients would benefit from additional antibiotic therapy [4]. These authors continue to recommend a dichotomous approach to appendicitis treatment. Another study by Emil et al. prospectively evaluated outcomes in patients with gangrenous appendicitis, using the definition of a black or gray discoloration of the appendix without a visible hole, fecalith, pus, or appendiceal leak [6]. Based on these findings, the authors recommend that gangrenous appendicitis should perhaps be treated with a third management algorithm that has a similar length of stay to simple appendicitis but includes post-operative antibiotic administration [6]. At our own institution, we have adopted similar intraoperative definitions for gangrenous and perforated appendicitis, and at this time both are treated under the advanced appendicitis treatment algorithm, including intravenous piperacillin–tazobactam, which is continued until clinical discharge criteria are met and leukocytosis (b 12 k) normalizes [1,5]. This study highlights discrepancies in the assignment of diagnoses that may have implications beyond clinical outcomes. The findings in this study differ somewhat from other pediatric studies of surgeon and pathologist congruence regarding the appendicitis diagnosis [7]. Bliss et al. found that the concordance rates between the surgeon and pathologist were higher in those with acute appendicitis (~90%), compared to perforated appendicitis (~55%) [7]. We found that the concordance was actually much higher in patients with perforated disease (75%) compared to simple (33%), or even simple and suppurative disease (37%). Additionally, their study found that when the pathologist and surgeon were discordant, the pathologist’s diagnosis was more
Total
710 (n = 8, 1.1%) 2 (1–31) 456 (n = 15.5%) 6 (1–29)
predictive of the clinical outcome (complication rate) than the surgeon’s diagnosis. In adult studies, alternative categorization schemes have been proposed that are based on a combination of the appendix appearance and the associated intraabdominal extension of inflammation and infection. One study also found fair to moderate concordance between the operative and pathologic diagnoses (k = 0.39, 95% CI 0.23–0.55) [8,9]. In our study, the operative diagnosis appeared to better predict surgical site infection rates than the pathologic diagnosis. When the pathologist’s diagnosis is used as the gold standard for clinical management, there is a significant difference in outcomes between those the surgeon determined had simple or perforated disease. When the operative diagnosis is used as the gold standard, there is no significant difference in the clinical outcomes based on the pathologist’s diagnosis. Therefore, this supports our practice to use the surgeon’s diagnosis as the gold standard for determining clinical management. Of note, we did find that the surgeon reporting of a normal appendix at the time of appendectomy was lower than the pathologist rate (b1% vs. 4%). We assume this a combination of surrounding peritoneal inflammation not assessed in the specimen or surgeon difficulty with diagnosing a normal appendix versus early appendicitis on gross examination. This also may be the result of the limits of a retrospective review, where the surgeon was not asked directly to comment on the disease severity. Additionally, our findings differed from those reported by Emil et al. with respect to gangrenous appendicitis [2]. Gangrenous appendicitis is diagnosed at a much lower rate in our study by the surgeon (6.8% vs. 17%). They report a 62% histopathologic agreement rate between the surgeon and the pathologist in cases of gangrenous appendicitis, which is much improved over our own 13% [2]. With this high rate of concordance and reproducibility, these authors recommend a separate treatment pathway for those with gangrenous appendicitis. Due to concerns over the reproducibility of these findings at our own institution, as well as concerns for disruption of the integrity of the appendiceal wall by bacteria with subsequent intraperitoneal contamination requiring antibiotics, we continue to use a dichotomous treatment pathway with gangrenous patients treated in the “complex” arm, understanding that these patients will likely have slightly short hospital courses and lower complication rates than their grossly perforated counterparts. We recognize that our treatment of gangrenous appendicitis in the same protocol as the perforated group is based on observation from an internal review of our own institutional data, which demonstrated that these patients have a return of bowel function and surgical site infection rate that is more similar to those with perforated disease. To
Table 3 Correlating pathologic and intraoperative diagnoses to clinical outcomes. Pathologic Finding
Number of Patients
Intraoperative Diagnosis
Lymphoid Hyperplasia Acute or Chronic Serositis Fibrous Luminal Obliteration Chronic inflammation/appendicitis Carcinoid Tumor Lympoplasmacytic infiltrate with mucosal erosion and crypt abscesses
21 5 3 2 2 1
All simple appendicitis 2 simple, 1 suppurative, 2 perforated All simple appendicitis 1 suppurative, 1 perforated 1 simple, 1 perforated Suppurative
S.C. Fallon et al. / Journal of Pediatric Surgery 50 (2015) 638–641
address this, we have an ongoing prospective observational study to assess these clinical questions and determine the optimal treatment algorithm for patients with gangrenous appendicitis [2,5]. The 34 (or 2.8%) patients with other findings at the time of pathologic examination remain a challenge to assess. While the majority appeared to have a pathology that did warrant appendectomy, the clinical course of the patient had the appendix not been removed remains uncertain. These patients did not have any post-operative complications, and generally had a clinical course similar to those with simple, acute appendicitis. The two patients who were found to have a carcinoid tumor support the continued, routine examination of the appendix specimen. This study highlights the need for quality improvement at a hospital level with regards to the diagnosis and categorization of appendicitis. A 10% error rate in the coding for the severity of appendicitis could potentially have reimbursement consequences, particularly as bundled care payments based on disease severity become the standard. However, we were unable to measure the direct cost effect of this discrepancy in the current study. The results of our study have prompted a reexamination of our coding practice, as it does not accurately predict the clinical course. In order to assist with diagnostic clarity in the operative note for the purposes of correct assignation of ICD-9 codes, we have implemented an operative note template that allows the surgeon to specify the correct ICD-9 code for the procedure, and to have the operative diagnosis (normal, acute, suppurative, gangrenous, or perforated) clearly stated within the text of the note. The effect of this standardization is the subject of an ongoing quality improvement study to enhance the homogeneity of the appendicitis cohort used to track outcomes within our electronic data warehouse.
641
In conclusion, the operative diagnosis at the time of appendectomy appears to be more predictive of clinical outcome than the pathologic diagnosis. Pathologic examination of the appendix remains important, as we did identify two cases of carcinoid tumor in a one-year time frame and a negative appendectomy rate of 4%. And finally, attention to detail regarding provider and hospital-level diagnostic coding can reveal errors that require intervention for research, treatment, and reimbursement purposes. References [1] Fallon SC, Brandt ML, Hassan SF, et al. Evaluating the effectiveness of a discharge protocol for children with advanced appendicitis. J Surg Res 2013;184(1):347–51. [2] Emil S, Laberge JM, Mikhail P, et al. Appendicitis in children: a ten-year update of therapeutic recommendations. J Pediatr Surg 2003;38(2):236–42. [3] Warner BW, Rich KA, Atherton H, et al. The sustained impact of an evidenced-based clinical pathway for acute appendicitis. Semin Pediatr Surg 2002;11(1):29–35. [4] St Peter SD, Sharp SW, Holcomb III GW, et al. An evidence-based definition for perforated appendicitis derived from a prospective randomized trial. J Pediatr Surg 2008; 43(12):2242–5. [5] Fallon SC, Hassan SF, Larimer EL, et al. Modification of an evidence-based protocol for advanced appendicitis in children. J Surg Res 2013;185(1):273–7. [6] Emil S, Gaied F, Lo A, et al. Gangrenous appendicitis in children: a prospective evaluation of definition, bacteriology, histopathology, and outcomes. J Surg Res 2012; 177(1):123–6. [7] Bliss D, McKee J, Cho D, et al. Discordance of the pediatric surgeon's intraoperative assessment of pediatric appendicitis with the pathologists report. J Pediatr Surg 2010; 45(7):1398–403. [8] Gomes CA, Nunes TA, Fonseca Chebli JM, et al. Laparoscopy grading system of acute appendicitis: new insight for future trials. Surg Laparosc Endosc Percutan Tech 2012;22(5):463–6. [9] Garst GC, Moore EE, Banerjee MN, et al. Acute appendicitis: a disease severity score for the acute care surgeon. J Trauma Acute Care Surg 2013;74(1):32–6.