Meeting NewsPerspective

Sarcoma coding often inaccurate

WAILEA, Hawaii — Many sarcoma cases had inaccurate ICD-9/ICD-10 coding, which could affect the accuracy of the National Cancer Data Base, National Inpatient Sample and other national databases, according to study results presented at the Connective Tissue Oncology Society Annual Meeting.

“Because sarcomas are rare malignancies, large tumor registries and population-based databases are being used more frequently to answer clinical questions,” Heather Lyu, MD, postdoctoral fellow at Brigham and Woman’s Hospital, said during her presentation. “Research has cited inaccuracies in large databases, but there are no previous studies on the accuracy of these large databases for sarcoma research.”

Sarcomas may be coded by surgical coders and tumor registrars by tumor location rather than type of malignancy.

Because inaccurate coding can have an impact on the accuracy of national databases, Lyu and colleagues sought to characterize inaccuracies in coding practices to identify errors at a single high-volume sarcoma center.

Researchers confirmed sarcoma diagnoses from operative pathology reports. They excluded nonsarcoma cases and operations on patients with sarcoma that were unrelated to their sarcoma diagnosis.

They then compared operative diagnostic codes — ICD-9 or ICD-10 — and tumor registrar diagnosis codes, ICD-O-3, for each to determine the variation in coding.

In total, five surgical and orthopedic oncologists performed 2,715 individual soft tissue and bone oncologic surgeries between 2012 and 2016. Of them, 1,237 had a histologic diagnosis of sarcoma with pathology confirmation.

Sixty-three percent of cases (n = 764) appeared accurately coded as sarcoma resections based on ICD-9/ICD-10 codes.

However, 16% (n = 180) were coded as a nononcologic diagnosis, and 22% (n = 260) were coded by the organ, such as gastrointestinal, breast, hematologic, gynecologic, thoracic, cutaneous or “other” cancers. For example, 26% of gastric gastrointestinal stromal tumors were coded as gastric cancer, and 52% of breast angiosarcomas were coded as breast cancer.

Researchers found 72% of orthopedic oncology cases and 57% of surgical oncology sarcoma cases were accurately coded.

When researchers evaluated ICD-O-3 codes, they found 631 of 1,054 patients (60%) had accurate coding, whereas 26 (2%) were coded as “other cancer” and 397 (38%) were not included on the tumor registry. Forty-six percent of gastric gastrointestinal stromal tumors and 72% of breast angiosarcomas also were coded incorrectly.

“Potential solutions include multidisciplinary discussions to create recommendations for future ICD-9 or ICD-10 coding, and we want to streamline the sarcoma coding process at institutions as well,” Lyu said. “We hope that our study will encourage other sarcoma centers and specialty societies to scrutinize their own data collection.” – by Alexandra Todak

Reference:

Lyu H, et al. Abstract 001. Presented at: CTOS Annual Meeting; Nov. 8-11, 2017; Maui, Hawaii.

Disclosures: The authors report no relevant financial disclosures.

WAILEA, Hawaii — Many sarcoma cases had inaccurate ICD-9/ICD-10 coding, which could affect the accuracy of the National Cancer Data Base, National Inpatient Sample and other national databases, according to study results presented at the Connective Tissue Oncology Society Annual Meeting.

“Because sarcomas are rare malignancies, large tumor registries and population-based databases are being used more frequently to answer clinical questions,” Heather Lyu, MD, postdoctoral fellow at Brigham and Woman’s Hospital, said during her presentation. “Research has cited inaccuracies in large databases, but there are no previous studies on the accuracy of these large databases for sarcoma research.”

Sarcomas may be coded by surgical coders and tumor registrars by tumor location rather than type of malignancy.

Because inaccurate coding can have an impact on the accuracy of national databases, Lyu and colleagues sought to characterize inaccuracies in coding practices to identify errors at a single high-volume sarcoma center.

Researchers confirmed sarcoma diagnoses from operative pathology reports. They excluded nonsarcoma cases and operations on patients with sarcoma that were unrelated to their sarcoma diagnosis.

They then compared operative diagnostic codes — ICD-9 or ICD-10 — and tumor registrar diagnosis codes, ICD-O-3, for each to determine the variation in coding.

In total, five surgical and orthopedic oncologists performed 2,715 individual soft tissue and bone oncologic surgeries between 2012 and 2016. Of them, 1,237 had a histologic diagnosis of sarcoma with pathology confirmation.

Sixty-three percent of cases (n = 764) appeared accurately coded as sarcoma resections based on ICD-9/ICD-10 codes.

However, 16% (n = 180) were coded as a nononcologic diagnosis, and 22% (n = 260) were coded by the organ, such as gastrointestinal, breast, hematologic, gynecologic, thoracic, cutaneous or “other” cancers. For example, 26% of gastric gastrointestinal stromal tumors were coded as gastric cancer, and 52% of breast angiosarcomas were coded as breast cancer.

Researchers found 72% of orthopedic oncology cases and 57% of surgical oncology sarcoma cases were accurately coded.

When researchers evaluated ICD-O-3 codes, they found 631 of 1,054 patients (60%) had accurate coding, whereas 26 (2%) were coded as “other cancer” and 397 (38%) were not included on the tumor registry. Forty-six percent of gastric gastrointestinal stromal tumors and 72% of breast angiosarcomas also were coded incorrectly.

“Potential solutions include multidisciplinary discussions to create recommendations for future ICD-9 or ICD-10 coding, and we want to streamline the sarcoma coding process at institutions as well,” Lyu said. “We hope that our study will encourage other sarcoma centers and specialty societies to scrutinize their own data collection.” – by Alexandra Todak

Reference:

Lyu H, et al. Abstract 001. Presented at: CTOS Annual Meeting; Nov. 8-11, 2017; Maui, Hawaii.

Disclosures: The authors report no relevant financial disclosures.

    Perspective
    Photo of Carolyn Nessim

    Carolyn Nessim

    This study shows that there is an underrepresentation of sarcoma in the database and, thus, large database research does not represent a true picture. If some sarcomas were coded as gastrointestinal malignancy or breast malignancy and National Cancer Data Base studies are conducted about breast or gastrointestinal cancers, these publications may be incorrect. What was evaluated as a breast cancer was in fact not breast cancer, but sarcoma. These large database studies may guide our practice, so this is an important issue to address. This is a common pitfall of research based on large databases, as often the information about each patient is limited.

    I agree with today’s speaker that each sarcoma center should evaluate their coding and try to improve the coding to fix this issue. One other solution is to have sarcoma codes like melanoma and lymphoma by cancer diagnosis, as opposed to site of disease, to improve the cancer registry, which will, in turn, improve the quality of large database research.

    • Carolyn Nessim, MD, MSc, FRCSC, FACS
    • The Ottawa Hospital, University of Ottawa

    Disclosures: Nessim reports no relevant financial disclosures.

    See more from Connective Tissue Oncology Society Annual Meeting