Data-driven optimization of version 9 American Joint Committee on Cancer staging system for anal cancer.

INTRODUCTION: The American Joint Committee on Cancer (AJCC) staging system undergoes periodic revisions to maintain contemporary survival outcomes related to stage. Recently, the AJCC has developed a novel, systematic approach incorporating survival data to refine stage groupings. The objective of this study was to demonstrate data-driven optimization of the version 9 AJCC staging system for anal cancer assessed through a defined validation approach. METHODS: The National Cancer Database was queried for patients diagnosed with anal cancer in 2012 through 2017. Kaplan-Meier methods analyzed 5-year survival by individual clinical T category, N category, M category, and overall stage. Cox proportional hazards models validated overall survival of the revised TNM stage groupings. RESULTS: Overall, 24,328 cases of anal cancer were included. Evaluation of the 8th edition AJCC stage groups demonstrated a lack of hierarchical prognostic order. Survival at 5 years for stage I was 84.4%, 77.4% for stage IIA, and 63.7% for stage IIB; however, stage IIIA disease demonstrated a 73.0% survival, followed by 58.4% for stage IIIB, 59.9% for stage IIIC, and 22.5% for stage IV (p <.001). Thus, stage IIB was redefined as T1-2N1M0, whereas Stage IIIA was redefined as T3N0-1M0. Reevaluation of 5-year survival based on data-informed stage groupings now demonstrates hierarchical prognostic order and validated via Cox proportional hazards models. CONCLUSION: The 8th edition AJCC survival data demonstrated a lack of hierarchical prognostic order and informed revised stage groupings in the version 9 AJCC staging system for anal cancer. Thus, a validated data-driven optimization approach can be implemented for staging revisions across all disease sites moving forward.

The National Cancer Database Conforms to the Standardized Framework for Registry and Data Quality.

BACKGROUND: Standardization of procedures for data abstraction by cancer registries is fundamental for cancer surveillance, clinical and policy decision-making, hospital benchmarking, and research efforts. The objective of the current study was to evaluate adherence to the four components (completeness, comparability, timeliness, and validity) defined by Bray and Parkin that determine registries' ability to carry out these activities to the hospital-based National Cancer Database (NCDB). METHODS: Tbis study used data from U.S. Cancer Statistics, the official federal cancer statistics and joint effort between the Centers for Disease Control and Prevention (CDC) and the National Cancer Institute (NCI), which includes data from National Program of Cancer Registries (NPCR) and Surveillance, Epidemiology, and End Results (SEER) to evaluate NCDB completeness between 2016 and 2020. The study evaluated comparability of case identification and coding procedures. It used Commission on Cancer (CoC) standards from 2022 to assess timeliness and validity. RESULTS: Completeness was demonstrated with a total of 6,828,507 cases identified within the NCDB, representing 73.7% of all cancer cases nationwide. Comparability was followed using standardized and international guidelines on coding and classification procedures. For timeliness, hospital compliance with timely data submission was 92.7%. Validity criteria for re-abstracting, recording, and reliability procedures across hospitals demonstrated 94.2% compliance. Additionally, data validity was shown by a 99.1% compliance with histologic verification standards, a 93.6% assessment of pathologic synoptic reporting, and a 99.1% internal consistency of staff credentials. CONCLUSION: The NCDB is characterized by a high level of case completeness and comparability with uniform standards for data collection, and by hospitals with high compliance, timely data submission, and high rates of compliance with validity standards for registry and data quality evaluation.

Evaluation of the Impact of the COVID-19 Pandemic on Reliability of Cancer Surveillance Data in the National Cancer Database.

PURPOSE: This study evaluated the reliability of cancer cases reported to the National Cancer Database (NCDB) during 2020, the first year of the COVID-19 pandemic. METHODS: Total number of cancer cases reported to the NCDB between January 2018 and December 2020 were calculated for all cancers and 21 selected cancer sites. The additive outlier method was used to identify structural breaks in trends compared with previous years. The difference between expected (estimated using the vector autoregressive method) and observed number of cases diagnosed in 2020 was estimated using generalized estimating equation under assumptions of the Poisson distribution for count data. Interrupted time series analysis was used to compare changes in the number of records processed by registrars each month of 2020. All models accounted for seasonality, regional variation, and random error. RESULTS: There was a statistically significant decrease (structural break) in the number of cases diagnosed in April 2020, with no recovery in number of cases during subsequent months, leading to a 12.4% deficit in the number of cases diagnosed during the first year of the pandemic. While the number of cancer records initiated by cancer registrars also decreased, the number of records marked completed increased during the first months of the pandemic. CONCLUSION: There was a significant deficit in the number of cancer diagnoses in 2020 that was not due to cancer registrars' inability to extract data during the pandemic. Future studies can use NCDB data to evaluate the impact of the pandemic on cancer care and outcomes.

Twenty-Five Years of Cancer Follow-Up; Is the Data Worth the Effort?

BACKGROUND: Substantial resources are dedicated to long-term follow-up within cancer registries; however, the completeness of these data is poorly characterized. Our objectives were to quantify long-term cancer follow-up data completeness and the effort required to collect these data using the National Cancer Database (NCDB). METHODS: To quantify data completeness, patients diagnosed with cancer in 1989 were identified in the NCDB and loss to follow-up rates were assessed for 25 years after diagnosis. To quantify data collection effort, patients diagnosed from 1989 to 2014 who were alive and eligible for follow-up in 2014 were identified and the effort to perform patient follow-up was obtained via a survey of tumor registrars. The effort to perform follow-up beyond various intervals after diagnosis was calculated. RESULTS: In total, 484,201 patients at 958 hospitals were diagnosed with cancer in 1989. After 5 years, 6.5% of patients were lost to follow-up (13.1% of living patients), 50.3% were deceased, and 43.2% had ongoing follow-up. After 15 years, 22.9% were lost to follow-up (68.7% of living patients), 66.7% were deceased, and 10.5% had ongoing follow-up. By 25 years, loss to follow-up increased to 28.6% (93.7% of living patients), 69.5% were deceased, and 1.9% had ongoing follow-up. In 2014, 522,838 h were spent performing follow-up for 2,091,353 patients at 1456 hospitals who were >15 years from their initial cancer diagnosis. CONCLUSIONS: While 5-year follow-up is excellent in the NCDB, loss to follow-up increases over time. The impact of curtailing data collection is under investigation and follow-up duration requirements will be re-evaluated.

Evaluating information loss in the National Cancer Database from cases lost to follow-up.

BACKGROUND AND OBJECTIVES: Cancer registries must focus on data capture which returns value while reducing resource burden with minimal loss of data. Identifying the optimum length of follow-up data collection for patients with cancer achieves this goal. METHODS: A two-step analysis using entropy calculations to assess information gain for each follow-up year, and second-order differences to compare survival outcomes between the defined follow-up periods and lifetime follow-up. A total of 391 567 adult cases, deidentified in the National Cancer Database and diagnosed in 1989. Comparisons examined a subset of 61 908 lung cancer cases, 48 387 colon and rectal cancer cases, and 64 134 breast cancer cases in adults. A total of 4133 pediatric cases were diagnosed in 1989 examining 1065 leukemia cases and 494 lymphoma cases. RESULTS: Annual increases in information gain fell below 1% after 16 years of follow-up for adult cases and 9 years for pediatric cases. Comparison of second-order differences showed 62% of the comparisons were similar between 15 years and lifetime follow-up when examining restricted mean survival time. In addition, 90% of the comparisons were statistically similar when comparing hazard ratios. CONCLUSIONS: Survival analysis using 15 years postdiagnosis follow-up showed minimal differences in information gain compared to lifetime follow-up.

Compliance with Cancer Quality Measures Over Time and Their Association with Survival Outcomes: The Commission on Cancer's Experience with the Quality Measure Requiring at Least 12 Regional Lymph Nodes to be Removed and Analyzed with Colon Cancer Resections.

BACKGROUND: Many quality measures in cancer care are process measures. The rates of compliance for these measures over time have not been well described, and the relationships between measure compliance and survival are not well understood. METHODS: The National Cancer Database, representing cancer registry data from approximately 1500 Commission on Cancer (CoC) cancer programs, was queried to determine the rates of compliance, with the CoC's colon cancer quality measure requiring 12 regional lymph nodes be removed at resection. Data were assessed in 2003, before the measure was reported to programs, through 2015. Measure compliance and risk-adjusted survival were examined by hospital type. RESULTS: From 2003 to 2015, 544,018 cases of colon cancer were analyzed for number of nodes removed. In 2003, compliance was 52.8% and National Cancer Institute (NCI) centers had the highest compliance rate (69.0%), followed by academic cancer centers (61.9%), comprehensive community hospitals (50.9%), and community hospitals (44.0%). Between 2003 and 2015, compliance improved for all hospital types, although differences remained. Risk-adjusted survival in 2009 was better at NCI centers [hazard ratio (HR) 0.76] than at academic cancer centers (HR 0.90), which had better survivals than comprehensive community programs (HR 0.93) when compared with patients treated at community hospitals. CONCLUSION: After introduction of this quality measure, performance at CoC-accredited hospitals improved over the subsequent 13 years, and survival by hospital type paralleled measure compliance by hospital type. This demonstrated measurement may be associated with improvements in performance, and that there are differences in performance and outcome by hospital type.

Incident Cases Captured in the National Cancer Database Compared with Those in U.S. Population Based Central Cancer Registries in 2012-2014.

BACKGROUND: The National Cancer Database (NCDB) is a hospital-based cancer registry that includes diagnostic, staging, treatment, and outcomes data for newly diagnosed cancer patients in the United States. The NCDB data include 31 million records for patients diagnosed between 1985-2015. A Participant User File based on a subset of these data has been available to researchers at facilities accredited by the Commission on Cancer since 2010. This study aimed to compare the number of incident cancer cases in the NCDB with a national population cancer registry. METHODS: Incident cancer cases in the NCDB in 2012-2014 were compared with the number of cancer cases in the United States Cancer Statistics data for the 2012-2014 diagnosis years. Comparisons were made by primary site and other factors. RESULTS: In 2012-2014, the NCDB captured 72% of the cancer cases in the United States, which was slightly higher than the 67% and 69% reported respectively in two prior assessments. Among the top 10 major cancer sites, the highest coverage (80%) was found for breast cancer, and the lowest was found for melanoma of the skin (52%) and prostate (58%). Colon, bladder, and kidney and renal pelvis cancers had relatively high coverage of 71%, 70% and 78%, respectively, whereas lung and bronchus had slightly lower coverage (65%). CONCLUSIONS: The NCDB coverage of U.S. cancer cases has remained relatively high (72%), but differences remain by cancer site and other factors that should be taken into account by users of the NCDB data.

The newly proposed clinical and post-neoadjuvant treatment staging classifications for gastric adenocarcinoma for the American Joint Committee on Cancer (AJCC) staging.

PURPOSE: New stage grouping classifications for clinical (cStage) and post-neoadjuvant treatment (ypStage) stage for gastric adenocarcinoma have been proposed for the eighth edition of the AJCC manual. This article summarizes the analysis for these stages. METHODS: Gastric adenocarcinoma patients diagnosed in 2004-2009 were identified from the National Cancer Database (NCDB). The cStage cohort included both surgical and nonsurgical cases, and the ypStage cohort included only patients who had chemotherapy or radiation therapy before surgery. Survival differences between the stage groups were determined by the log-rank test and prognostic accuracy was assessed by concordance index. Analysis was performed using SAS 9.4 (SAS, Cary, NC, USA). RESULTS: Five strata for cStage and four strata for ypStage were developed. The 5-year survival rates for cStages were 56.77%, 47.39%, 33.1%, 25.9%, and 5.0% for stages I, IIa, IIb, III, and IV, respectively, and the rates for ypStage were 74.2%, 46.3%, 19.2%, and 11.6% for stages I, II, III, and IV, respectively. The log-rank test showed that survival differences were well stratified and stage groupings were ordered and distinct (p < 0.0001). The proposed cStage and ypStage classification was sensitive and specific and had high prognostic accuracy (cStage: c index = 0.81, 95% CI, 0.79-0.83; ypStage: c index = 0.80, 95% CI, 0.73-0.87). CONCLUSION: The proposed eighth edition establishes two new staging schemata that provide essential prognostic data for patients before treatment and for patients who have undergone surgery following neoadjuvant therapy. These additions are a significant advance to the AJCC staging manual and will provide critical guidance to clinicians in making informed decisions throughout the treatment course.

Doubling of 30-Day Mortality by 90 Days After Esophagectomy: A Critical Measure of Outcomes for Quality Improvement.

OBJECTIVES: Our objectives were to (1) compare 30- and 90-day mortality rates after esophagectomy, (2) compare drivers of 30- and 90-day mortality, and (3) examine whether 90-day mortality affects hospital rankings. BACKGROUND: Operative mortality has traditionally been assessed at 30 days. Ninety-day mortality has been suggested as a more appropriate indicator of quality, particularly after complex cancer surgery. METHODS: Esophagectomies for nonmetastatic esophageal cancer patients diagnosed between 2007 and 2011 were identified in the National Cancer Data Base. Mortality rates were examined by patient demographics, tumor characteristics, and hospital procedural volume. Risk-adjusted hierarchical logistic regression models examined hospital performance for mortality. RESULTS: A total of 15,796 esophagectomy patients at 977 hospitals were available for analysis. Ninety-day overall mortality was more than double the 30-day mortality (8.9% vs 4.2%; P < 0.0001). In multivariate analysis, while both 30- and 90-day mortality were associated with patient factors such as age, comorbidity, and hospital volume, only 90-day mortality was influenced by tumor- and management-related variables such as stage, tumor location, and receipt of neoadjuvant therapy. Hospital performance was examined as top 10%, middle 10% to 90%, and lowest 10% as ranked using risk-adjusted odds of mortality. There was moderate correlation between ranking based on 30- and 90-day mortality [weighted κ = 0.45 (95% confidence interval, 0.39-0.52)]. Compared with 30-day mortality rankings, nearly 20% of hospitals changed their ranking category when 90-day mortality rankings were used. CONCLUSIONS: Examination of 90-day mortality after esophagectomy reflects cancer patient management decisions and may provide actionable targets for quality improvement.

National Evaluation of Hospital Performance on the New Commission on Cancer Melanoma Quality Measures.

INTRODUCTION: To increase adherence to cancer management guidelines, the Commission on Cancer (CoC) developed and approved five melanoma quality measures in 2015. Our objectives were to evaluate formally the national performance of these melanoma measures and to examine patient, tumor, and hospital characteristics associated with adherence. METHODS: From the National Cancer Data Base (2012), patients with invasive, nonmetastatic melanoma were identified. Inclusion and exclusion criteria were based on the CoC definition for each measure. Patient-level and hospital-level adherence rates were calculated for the five measures. A hospital was deemed "compliant" if it met the CoC standard, which requires 80 % of patients to receive the measure-specific recommended care. Patient, tumor, and hospital characteristics potentially associated with higher likelihood of adherence at the patient-level were estimated using hierarchical random-effects logistic regression models. RESULTS: A total of 31,598 patients from 1343 hospitals were examined. Patient-level adherence rates varied from 31.6 % (Measure 5: ≥10 axillary lymph nodes removed/examined) to 72.6 % (Measure 1: sentinel lymph node biopsy (SLNB) appropriateness measure). Hospital-level adherence rates, ranged from 19.3 % of hospitals (N = 538 hospitals for Measure 5) to 44.8 % of hospitals (N = 1090 hospitals for Measure 3: completion lymph node dissection after positive SLNB). No hospital-level factors (e.g., teaching status) were consistently associated with better adherence. CONCLUSIONS: National adherence rates to the five new CoC melanoma quality metrics are low, and most hospitals would not meet the CoC requirement of 80 % adherence. Feedback for performance of these measures to hospitals, decisions support tools, and educational initiatives are needed to improve guideline adherence.

Effect of the USPSTF Grade D Recommendation against Screening for Prostate Cancer on Incident Prostate Cancer Diagnoses in the United States.

PURPOSE: In October 2011 the USPSTF (U.S. Preventive Services Task Force) issued a draft guideline discouraging prostate specific antigen based screening for prostate cancer (grade D recommendation). We evaluated the effect of the USPSTF guideline on the number and distribution of new prostate cancer diagnoses in the United States. MATERIALS AND METHODS: We identified incident cancers diagnosed between January 2010 and December 2012 in NCDB (National Cancer Database). We performed an interrupted time series to evaluate the trend of new prostate cancers diagnosed each month before and after the draft guideline with colon cancer as a comparator. RESULTS: Incident monthly prostate cancer diagnoses decreased by -1,363 cases (12.2%, p<0.01) in the month after the USPSTF draft guideline and continued to decrease by 164 cases per month relative to baseline (-1.8%, p<0.01). In contrast monthly colon cancer diagnoses remained stable. Diagnoses of low, intermediate and high risk prostate cancers decreased significantly but new diagnoses of nonlocalized disease did not change. Subgroups of age, comorbidity, race, income and insurance showed comparable decreases in incident prostate cancer following the draft guideline. CONCLUSIONS: There was a 28% decrease in incident diagnoses of prostate cancer in the year after the USPSTF draft recommendation against prostate specific antigen screening. This study helps quantify the potential benefits (reduced harms of over diagnosis and overtreatment of low risk disease and disease found in elderly men) and potential harms (missed opportunities to diagnose important cancers in men who may benefit from treatment) of this guideline.

Parathyroid Carcinoma: An Update on Treatment Outcomes and Prognostic Factors from the National Cancer Data Base (NCDB).

BACKGROUND: Parathyroid carcinoma is a rare disease. Conflicting results on prognostic factors and extent of surgical resection for patients with parathyroid carcinoma have been made based on small sample sizes. A large, robust dataset is needed to help address some of the controversies. METHODS: A retrospective review of patients with parathyroid carcinoma in the National Cancer Data Base from 1985 to 2006 was performed. Characteristics of the cohort and type of treatment were evaluated. Prognostic factors were assessed with Cox proportional hazards regression models and 5- and 10-year OS rates were determined. RESULTS: There were 733 evaluable patients with a mean age of 56.1 ± 15.3 years (median 57, range 15-89) and mean tumor size of 29.6 ± 18.4 mm (median 25.0 mm, range 10.0-150.0). Tumor size, age at diagnosis, male sex, positive nodal status, and complete tumor resection had hazard ratios for death of 1.02 (1.01-1.02, p < 0.0001), 1.06 (1.05-1.07, p < 0.0001), 1.67 (1.24-2.25, p = 0.0008), 1.25 (0.57-2.76, p = 0.6), and 0.42 (0.22-0.81, p = 0.01), respectively, on multivariable analysis. Patients who had removal of the parathyroid tumor with concomitant resection of adjacent organs had HR for death of 0.70 (0.35-1.41, p = 0.3). The 5- and 10-year OS rates were 82.3 and 66 % respectively. CONCLUSIONS: Patient age, tumor size, and sex have modest effects on survival in patients with parathyroid carcinoma. A staging system with prognostic value for parathyroid carcinoma should include at least these pertinent prognostic factors.

Association of hospital volume with conditional 90-day mortality after cystectomy: an analysis of the National Cancer Data Base.

OBJECTIVE: To examine the association of hospital volume and 90-day mortality after cystectomy, conditional on survival for 30 days. PATIENTS AND METHODS: The National Cancer Data Base was used to evaluate 30- and 90-day mortality for 35,055 patients who underwent cystectomy for bladder cancer at one of 1118 hospitals. Patient data were aggregated into hospital volume categories based on the mean annual number of procedures (low-volume hospital: <10 procedures; intermediate-volume hospital: 10-19 procedures; high-volume hospital: ≥20 procedures). Associations between mortality and clinical, demographic and hospital characteristics were analysed using hierarchical logistic regression models. To assess the association between hospital volume and 90-day mortality independently of shorter-term mortality, 90-day mortality conditional on 30-day survival was assessed in the multivariate modelling. RESULTS: Unadjusted 30- and 90-day mortality rates were 2.7 and 7.2% overall, 1.9 and 5.7% among high-volume hospitals, and 3.2 and 8.0% among low-volume hospitals, respectively. Compared with high-volume hospitals, the adjusted risks among low-volume hospitals (odds ratio [95% CI]) of 30- and 90-day mortality, conditional on having survived for 30 days, from the hierarchical models were 1.5 (1.3-1.9), and 1.2 (1.0-1.4), respectively. CONCLUSIONS: A low hospital volume was associated with greater 30- and 90-day mortality. These data support the need for further research to better understand the relatively high mortality rates seen between 30 and 90 days, which are high and less variable across hospital volume strata. The stronger association between volume and 30-day mortality suggests that quality-reporting efforts should focus on shorter-term outcomes.

Survival Among Patients With High-Risk Gastrointestinal Cancers During the COVID-19 Pandemic.

Importance: Prior reports demonstrated that patients with cancer experienced worse outcomes from pandemic-related stressors and COVID-19 infection. Patients with certain malignant neoplasms, such as high-risk gastrointestinal (HRGI) cancers, may have been particularly affected. Objective: To evaluate disruptions in care and outcomes among patients with HRGI cancers during the COVID-19 pandemic, assessing for signs of long-term changes in populations and survival. Design, Setting, and Participants: This retrospective cohort study used data from the National Cancer Database to identify patients with HRGI cancer (esophageal, gastric, primary liver, or pancreatic) diagnosed between January 1, 2018, and December 31, 2020. Data were analyzed between August 23 and September 4, 2023. Main Outcome and Measures: Trends in monthly new cases and proportions by stage in 2020 were compared with the prior 2 years. Kaplan-Meier curves and Cox regression were used to assess 1-year mortality in 2020 compared with 2018 to 2019. Proportional monthly trends and multivariable logistic regression were used to evaluate 30-day and 90-day mortality in 2020 compared with prior years. Results: Of the 156 937 patients included in this study, 54 994 (35.0%) were aged 60 to 69 years and 100 050 (63.8%) were men. There was a substantial decrease in newly diagnosed HRGI cancers in March to May 2020, which returned to prepandemic levels by July 2020. For stage, there was a proportional decrease in the diagnosis of stage I (-3.9%) and stage II (-2.3%) disease, with an increase in stage IV disease (7.1%) during the early months of the pandemic. Despite a slight decrease in 1-year survival rates in 2020 (50.7% in 2018 and 2019 vs 47.4% in 2020), survival curves remained unchanged between years (all P > .05). After adjusting for confounders, diagnosis in 2020 was not associated with increased 1-year mortality compared with 2018 to 2019 (hazard ratio, 0.99; 95% CI, 0.97-1.01). The rates of 30-day (2.1% in 2018, 2.0% in 2019, and 2.1% in 2020) and 90-day (4.3% in 2018, 4.4% in 2019, and 4.6% in 2020) operative mortality also remained similar. Conclusions and Relevance: In this retrospective cohort study, a period of underdiagnosis and increase in stage IV disease was observed for HRGI cancers during the pandemic; however, there was no change in 1-year survival or operative mortality. These results demonstrate the risks associated with gaps in care and the tremendous efforts of the cancer community to ensure quality care delivery during the pandemic. Future research should investigate long-term survival changes among all cancer types as additional follow-up data are accrued.

Disruption of National Cancer Database Data Models in the First Year of the COVID-19 Pandemic.

Importance: Each year, the National Cancer Database (NCDB) collects and analyzes data used in reports to support research, quality measures, and Commission on Cancer program accreditation. Because data models used to generate these reports have been historically stable, year-to-year variances have been attributed to changes within the cancer program rather than data modeling. Cancer submissions in 2020 were anticipated to be significantly different from prior years because of the COVID-19 pandemic. This study involved a validation analysis of the variances in observed to expected 2020 NCDB cancer data in comparison with 2019 and 2018. Observations: The NCDB captured a total of 1 223 221 overall cancer cases in 2020, a decrease of 14.4% (Δ = -206 099) compared with 2019. The early months of the COVID-19 pandemic (March-May 2020) coincided with a nadir of cancer cases in April 2020 that did not recover to overall prepandemic levels through the remainder of 2020. In the early months of the COVID-19 pandemic, the proportion of early-stage disease decreased sharply overall, while the proportion of late-stage disease increased. However, differences in observed to expected stage distribution in 2020 varied by primary disease site. Statistically significant differences in the overall observed to expected proportions of race and ethnicity, sex, insurance type, geographic location, education, and income were identified, but consistent patterns were not evident. Conclusions and Relevance: Historically stable NCDB data models used for research, administrative, and quality improvement purposes were disrupted during the first year of the COVID-19 pandemic. NCDB data users will need to carefully interpret disease- and program-specific findings for years to come to account for pandemic year aberrations when running models that include 2020.