Validity of Administrative Data in Identifying Cancer-related Events in Adolescents and Young Adults: A Population-based Study Using the IMPACT Cohort

Development and validation of thromboembolism diagnostic algorithms in children with cancer from real-world data

OBJECTIVE

To evaluate the accuracy of diagnostic algorithms developed using the International Classification of Diseases (ICD-9-CM and ICD-10-CA) diagnostic codes and physician billing codes for thromboembolism (TE) from health administrative data compared to chart review diagnoses of TE in children with cancer.

METHODS

Using data linkage between the Pediatric Oncology Group of Ontario Network Information System (Ontario pediatric cancer registry) and various administrative data housed at ICES, eight algorithms were developed including a single reference to one of the billing codes, multiple references with varying time intervals, and combinations of various billing codes during primary cancer therapy for the whole cohort and, for early (<04/2002) and later (≥04/2002, solely ICD-10 codes) periods. Reference standard was chart review data from prior studies (from 1990 to 2016) among children (≤19 years) with cancer and radiologically confirmed TE.

RESULTS

Records of 2056 patients diagnosed with cancer at two participating sites during study period were reviewed; 112 had radiologically confirmed TE. The algorithm with addition of anticoagulation utilization codes was the best performing algorithm (sensitivity = 0.76;specificity = 0.85). With use of ICD-10 only codes, sensitivity of the same algorithm improved to 0.84 with specificity of 0.80.

CONCLUSION

This study provides a valid approach for ascertaining pediatric TE using real-world data.

IMPACT

Research in pediatric thrombosis, especially cancer-related thrombosis, is limited mainly due to small-sized studies. Real-world data provide ready access to large and diverse populations. However, there are no validated algorithms for identifying thrombosis in real-world data for children. An algorithm based on combination of thrombosis and anticoagulation utilization codes had 76% sensitivity and 85% specificity to identify diagnosis of thrombosis in children in administrative data. This study provides a valid approach for ascertaining pediatric thrombosis using real-world data and offers a good avenue to advance pediatric thrombosis research.

Treatment patterns and outcomes in adolescents and young adults with nodular lymphocyte-predominant Hodgkin lymphoma: an IMPACT cohort study

We leveraged population-based clinical and healthcare data to identify treatment patterns and long-term outcomes among adolescents and young adults (AYA) with nodular lymphocyte-predominant Hodgkin lymphoma (NLPHL). All Ontario, Canada, AYA aged 15-21 years at diagnosis with NLPHL between 1992 and 2012 were identified, and their detailed clinical data were collected. Linkage to healthcare databases identified additional events (subsequent malignant neoplasms [SMN], relapses and deaths). Event-free survival (EFS) and overall survival (OS) were compared by locus of care (adult vs. paediatric) and predictors of outcomes determined. Of 1014 AYA with Hodgkin lymphoma, 54 (5.3%) had NLPHL; 15 (27.8%) were treated at a paediatric centre. No paediatric centre patient received radiation only versus 16 (41.0%) of adult centre patients. Excision only was more common in paediatric centres (p < 0.001). The 20-year EFS and OS rates were 82.9% ± 5.2% and 100% respectively. Advanced stage (hazard ratio: 4.9, 95% CI: 1.3-18.4; p = 0.02) was associated with inferior EFS. Although the 25-year cumulative incidence of SMN was 19.3% ± 9.6% for the entire cohort, there were no SMN among the patients treated with excision only. AYA with NLPHL have outstanding long-term survival. Resection alone was rare outside of paediatric institutions but associated with excellent outcomes. Given substantial SMN risks, chemotherapy-sparing and radiation-sparing strategies for appropriate subsets of patients are warranted.

Impact of locus of care on outcomes in adolescents and young adults with osteosarcoma and Ewing sarcoma treated at pediatric versus adult cancer centers: An IMPACT cohort study

BACKGROUND

Location of cancer care (LOC; pediatric vs. adult center) impacts outcomes in adolescents and young adults (AYA) with some cancer types. Data on the impact of LOC on survival in AYA with osteosarcoma (OSS) and Ewing sarcoma (EWS) are limited OBJECTIVES: To compare differences in demographics, disease/treatment characteristics, and survival in a population-based cohort of AYA with OSS or EWS treated at pediatric versus adult centers METHODS: The Initiative to Maximize Progress in Adolescent Cancer Therapy (IMPACT) cohort captured demographic, disease, and treatment data for all AYA (15-21 years old) diagnosed with OSS and EWS in Ontario, Canada between 1992 and 2012. Patients were linked to provincial administrative health care databases. Outcomes were compared between patients treated in pediatric versus adult centers.

RESULTS

One hundred thirty-seven AYA were diagnosed with OSS (LOC: 47 pediatric, 90 adult) and 84 with EWS (38 pediatric, 46 adult). AYA treated at pediatric centers were more likely to be enrolled in a clinical trial (OSS 55% vs. 1%, p < .001; EWS 53% vs. 2%, p < .001) and received higher cumulative chemotherapy doses. Five-year event-free survival (EFS ± standard error) in OSS and EWS were 47% ± 4% and 43% ± 5%, respectively. In multivariable analysis, the impact of LOC (pediatric vs. adult center) on EFS in OSS (adjusted hazard ratio [HR] 1.15, 95% confidence interval [CI]: 0.58-2.27, p = .69) and EWS (adjusted HR 1.82, 95% CI: 0.97-3.43, p = .06) was not statistically significant.

CONCLUSION

Despite disparities in trial participation and chemotherapy doses, outcomes did not differ by LOC suggesting that AYA with bone tumors can be treated at either pediatric or adult centers.

Superior outcomes with paediatric protocols in adolescents and young adults with aggressive B-cell non-Hodgkin lymphoma

Survival disparities by locus of care (LOC; paediatric versus adult) among adolescents and young adults (AYA) with acute lymphoblastic leukaemia (ALL) are well documented. Whether similar disparities exist among AYA with aggressive mature B-cell non-Hodgkin lymphoma (B-NHL) is unknown. We identified all Ontario, Canada AYA aged 15-21 years at diagnosis of B-NHL between 1992 and 2012. Demographic, disease, treatment and outcome data were chart abstracted. The impact of LOC on event-free (EFS) and overall survival (OS) were determined, adjusted for patient and disease covariates. Among 176 AYA with B-NHL, 62 (35·2%) received therapy at paediatric centres. The 5-year EFS and OS [± standard error (SE)] for the overall cohort were 72·2 [3·4]% and 76·1 [3·2]% respectively. Both EFS and OS were superior among paediatric centre AYA [EFS (± SE) 82·2 (4·9)% vs. 66·7 (4·4)%, P = 0·02; OS 85·5 (4·5)% vs. 71·1 (4·3)%, P = 0·03]. Adjusted for histology, stage and time period, adult centre AYA had inferior EFS [hazard ratio (HR) 2·4, 95% confidence interval (CI) 1·1-4·9, P = 0·02] and OS (HR 2·5, 95% CI 1·1-5·7, P = 0·03). Sensitivity analyses restricted to the latest time period, when most adult centre AYA received rituximab, demonstrated similar disparities. Similar to AYA with ALL, AYA with B-NHL may benefit from being treated with paediatric protocols. Studies prospectively validating these results are warranted.

Identifying relapses and stem cell transplants in pediatric acute lymphoblastic leukemia using administrative data: Capturing national outcomes irrespective of trial enrollment

INTRODUCTION

Our objectives were to design and validate methods to identify relapse and hematopoietic stem cell transplantation (HSCT) in children with acute lymphoblastic leukemia (ALL) using administrative data representing hospitalizations at US pediatric institutions.

METHODS

We developed daily billing and ICD-9 code definitions to identify relapses and HSCTs within a cohort of children with newly diagnosed ALL between January 1, 2004, and December 31, 2013, previously assembled from the Pediatric Health Information System (PHIS) database. Chart review for children with ALL at the Children's Hospital of Philadelphia (CHOP) and Texas Children's Hospital (TCH) was performed to establish relapse and HSCT gold standards for sensitivity and positive predictive value (PPV) calculations. We estimated incidences of relapse and HSCT in the PHIS ALL cohort.

RESULTS

We identified 362 CHOP and 314 TCH ALL patients in PHIS and established true positives by chart review. Sensitivity and PPV for identifying both relapse and HSCT in PHIS were > 90% at both hospitals. Five-year relapse incidence in the 10 150-patient PHIS cohort was 10.3% (95% CI 9.8%-10.9%) with 7.1% (6.6%-7.6%) of children underwent HSCTs. Patients in higher-risk demographic groups had higher relapse and HSCT rates. Our analysis also identified differences in incidences of relapse and HSCT by race, ethnicity, and insurance status.

CONCLUSIONS

Administrative data can be used to identify relapse and HSCT accurately in children with ALL whether they occur on- or off-therapy, in contrast with published approaches. This method has wide potential applicability for estimating these incidences in pediatric ALL, including patients not enrolled on clinical trials.

Cancer Informatics for Cancer Centers: Scientific Drivers for Informatics, Data Science, and Care in Pediatric, Adolescent, and Young Adult Cancer

Cancer Informatics for Cancer Centers (CI4CC) is a grassroots, nonprofit 501c3 organization intended to provide a focused national forum for engagement of senior cancer informatics leaders, primarily aimed at academic cancer centers anywhere in the world but with a special emphasis on the 70 National Cancer Institute-funded cancer centers. This consortium has regularly held topic-focused biannual face-to-face symposiums. These meetings are a place to review cancer informatics and data science priorities and initiatives, providing a forum for discussion of the strategic and pragmatic issues that we faced at our respective institutions and cancer centers. Here, we provide meeting highlights from the latest CI4CC Symposium, which was delayed from its original April 2020 schedule because of the COVID-19 pandemic and held virtually over three days (September 24, October 1, and October 8) in the fall of 2020. In addition to the content presented, we found that holding this event virtually once a week for 6 hours was a great way to keep the kind of deep engagement that a face-to-face meeting engenders. This is the second such publication of CI4CC Symposium highlights, the first covering the meeting that took place in Napa, California, from October 14-16, 2019. We conclude with some thoughts about using data science to learn from every child with cancer, focusing on emerging activities of the National Cancer Institute's Childhood Cancer Data Initiative.

Adolescents and young adult acute myeloid leukemia outcomes at pediatric versus adult centers: A population-based study

BACKGROUND

Adolescents and young adult (AYA) acute myeloid leukemia (AML) outcomes remain poor. The impact of locus of care (LOC; adult vs pediatric) in this population is unknown.

PROCEDURE

The IMPACT cohort comprises detailed data for all Ontario, Canada, AYA aged 15-21 years diagnosed with AML between 1992 and 2012, linked to population-based health administrative data. We determined the impact of LOC on event-free survival (EFS) and overall survival (OS), treatment-related mortality (TRM), and relapse/progression.

RESULTS

Among 140 AYA, 51 (36.4%) received therapy at pediatric centers. The five-year EFS and OS for the whole cohort were 35.0% ± 4.0% and 53.6% ± 4.2%. Cumulative doses of anthracycline were higher among pediatric center AYA [median 355 mg/m² , interquartile range (IQR) 135-492 vs 202 mg/m² , IQR 140-364; P = 0.003]. In multivariable analyses, LOC was not predictive of either EFS [adult vs pediatric center hazard ratio (HR) 1.3, 95% confidence interval (CI) 0.8-2.2, P = 0.27] or OS (HR 1.0, CI 0.6-1.6, P = 0.97). However, patterns of treatment failure varied; higher two-year incidence of TRM in pediatric centers (23.5% ± 6.0% vs.10.1% ± 3.2%; P = 0.046) was balanced by lower five-year incidence of relapse/progression (33.3% ± 6.7% vs 56.2% ± 5.3%; P = 0.002).

CONCLUSIONS

AYA AML survival outcomes did not vary between pediatric and adult settings. Causes of treatment failure were different, with higher intensity pediatric protocols associated with higher TRM but lower relapse/progression. Careful risk stratification and enhanced supportive care may be of substantial benefit to AYA with AML by allocating maximal treatment intensity to patients who most benefit while minimizing the risk of TRM.

A Systematic Review of Estimating Breast Cancer Recurrence at the Population Level With Administrative Data

BACKGROUND

Exact numbers of breast cancer recurrences are currently unknown at the population level, because they are challenging to actively collect. Previously, real-world data such as administrative claims have been used within expert- or data-driven (machine learning) algorithms for estimating cancer recurrence. We present the first systematic review and meta-analysis, to our knowledge, of publications estimating breast cancer recurrence at the population level using algorithms based on administrative data.

METHODS

The systematic literature search followed Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. We evaluated and compared sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy of algorithms. A random-effects meta-analysis was performed using a generalized linear mixed model to obtain a pooled estimate of accuracy.

RESULTS

Seventeen articles met the inclusion criteria. Most articles used information from medical files as the gold standard, defined as any recurrence. Two studies included bone metastases only in the definition of recurrence. Fewer studies used a model-based approach (decision trees or logistic regression) (41.2%) compared with studies using detection rules without specified model (58.8%). The generalized linear mixed model for all recurrence types reported an accuracy of 92.2% (95% confidence interval = 88.4% to 94.8%).

CONCLUSIONS

Publications reporting algorithms for detecting breast cancer recurrence are limited in number and heterogeneous. A thorough analysis of the existing algorithms demonstrated the need for more standardization and validation. The meta-analysis reported a high accuracy overall, which indicates algorithms as promising tools to identify breast cancer recurrence at the population level. The rule-based approach combined with emerging machine learning algorithms could be interesting to explore in the future.

Tubular injury and cell-cycle arrest biomarkers to predict acute kidney injury in noncritically ill children receiving aminoglycosides

Aim: NGAL, IL-18, KIM-1 as well as urinary TIMP2 and IGFBP7 and their mathematical product (TIMP2*IGFBP7) were evaluated for detecting pediatric aminoglycoside acute kidney injury (AG-AKI). Methods: In a prospective study, noncritically ill children received aminoglycosides (AG) ≥3 days. The area under the curve (AUC) for biomarkers to detect AKI was calculated by a) days before AKI onset; b) treatment days. Results: There were 113 AG episodes (68% febrile neutropenia). The AKI group had a higher proportion with febrile neutropenia. The AKI group had significantly lower NGAL 3 days before AKI, as patients with febrile neutropenia had a lower NGAL during AG treatment (p < 0.05). NGAL, IL-18 and TIMP2*IGFBP7 had AUC ≥0.73 at 3, 2 and 2 days before AKI onset. Conclusion: NGAL, IL-18 and TIMP2*IGFBP7 were modest early biomarkers of AG-AKI. Febrile neutropenia was associated with lower NGAL.

Long-Term Incidence and Predictors of Significant Hearing Loss Requiring Hearing Assistive Devices Among Childhood Cancer Survivors: A Population-Based Study

PURPOSE

Hearing loss is a significant late effect among childhood cancer survivors. Recent guidelines note insufficient evidence to quantify its natural history or risk associated with specific exposures. We examined the long-term incidence and predictors of hearing loss requiring hearing amplification devices (HADs) using population-based health care data.

METHODS

In Ontario, Canada, HAD costs are subsidized by the Assistive Devices Program (ADP). Ontario children < 18 years of age at cancer diagnosis between 1987 and 2016 were identified and linked to ADP claims. Cumulative HAD incidence was compared between cases and matched controls. Patient, disease, and treatment predictors of HAD were examined.

RESULTS

We identified 11,842 cases and 59,210 controls. Cases were at higher risk for HAD (hazard ratio [HR], 12.8; 95% CI, 9.8 to 16.7; P < .001). The cumulative incidence of HAD among survivors was 2.1% (95% CI, 1.7% to 2.5%) at 20 years and 6.4% (95% CI, 2.8% to 12.1%) at 30 years post-diagnosis. The 30-year incidence was highest in neuroblastoma (10.7%; 95% CI, 3.8% to 21.7%) and hepatoblastoma (16.2%; 95% CI, 8.6% to 26.0%) survivors. Predictors of HAD in multivariable analyses included age 0-4 years at diagnosis (v 5-9 years; HR, 2.2; 95% CI, 1.4-3.3; P < .001). Relative to no cisplatin exposure, patients receiving < 200 mg/m² were not at greater risk, unlike those receiving higher cumulative doses. Relative to no cranial or facial radiation, those who had received ≤ 32.00 Gy were at no higher risk, unlike those who had received > 32.00 Gy. Carboplatin exposure was not associated with HAD.

CONCLUSION

Childhood cancer survivors are at elevated risk for requiring HAD, which continues to increase between 20 and 30 years after diagnosis. Thresholds of cisplatin and radiation exposure exist, above which risk substantially increases. Prolonged monitoring and trials of otoprotective agents are warranted in high-risk populations.

Treatment patterns and outcomes in adolescents and young adults with Hodgkin lymphoma in pediatric versus adult centers: An IMPACT Cohort Study

Hodgkin lymphoma (HL) is a common adolescent and young adult (AYA) cancer. While outcome disparities between pediatric vs. adult centers [locus of care (LOC)] have been demonstrated in other AYA cancers such as acute lymphoblastic leukemia, they have not been well studied in HL. We therefore compared population‐based treatment patterns and outcomes in AYA HL by LOC. The IMPACT Cohort includes data on all Ontario, Canada AYA (15‐21 years) diagnosed with HL between 1992 and 2012. Linkage to population‐based health administrative data identified late effects. We examined LOC‐based differences in treatment modalities, cumulative doses, event‐free survival (EFS), overall survival (OS), and late effects. Among 954 AYA, 711 (74.5%) received therapy at adult centers. Pediatric center AYA experienced higher rates of radiation therapy but lower cumulative doses of doxorubicin and bleomycin. 10‐year EFS did not differ between pediatric vs. adult cancer vs. community centers (83.8% ± 2.4% vs. 82.8% ± 1.6% vs. 82.7%±3.0%; P = .71); LOC was not significantly associated with either EFS or OS in multivariable analyses. Higher incidences of second malignancies in pediatric center AYA and of cardiovascular events in adult center AYA were observed, but were not significant. In conclusion, while pediatric and adult centers used different treatment strategies, outcomes were equivalent. Differences in treatment exposures are however likely to result in different late‐effect risks. Protocol choice should be guided by individual late‐effect risk.

Impact of surgical approach on oncologic outcomes in women undergoing radical hysterectomy for cervical cancer

BACKGROUND

Recent studies demonstrating shorter survival among cervical cancer patients undergoing minimally invasive versus open radical hysterectomy could not account for surgeon volume and require confirmation in other jurisdictions with larger sample sizes, longer follow-up, and data on disease recurrence.

OBJECTIVE

To determine if surgical approach is associated with oncologic outcomes in cervical cancer patients undergoing minimally invasive or open radical hysterectomy, while accounting for mechanistic factors including surgeon volume.

STUDY DESIGN

We performed a population-based retrospective cohort study of cervical cancer patients undergoing primary radical hysterectomy by a gynecologic oncologist from 2006 to 2017 in Ontario, Canada. A multivariable marginal Cox proportional hazards model and cause-specific hazards model were used to evaluate the association of surgical approach with all-cause death and recurrence respectively, clustering at the surgeon level. We tested for interactions between surgical approach and either pathologic stage or surgeon volume.

RESULTS

We identified 958 patients (minimally invasive 475; open 483) with mean age 45.9 and a median follow-up of 6 years. Of minimally invasive procedures, 89.6% were performed laparoscopically and 10.4% robotically. The unadjusted 5-year cumulative incidences of all-cause death (minimally invasive 12.5%; open 5.4%), cervical cancer death (minimally invasive 9.3%; open 3.3%), and recurrence (minimally invasive 16.2%; open 8.4%) were significantly increased for minimally invasive radical hysterectomy in patients with stage IB disease, but not the cohort overall. After adjusting for patient factors and surgeon volume, minimally invasive radical hysterectomy was associated with increased rates of death (hazard ratio [HR], 2.20; 95% confidence interval [CI], 1.15-4.19) and recurrence (HR, 1.97; 95% CI, 1.10-3.50) compared to open radical hysterectomy in patients with stage IB disease (n = 534), but not IA disease (n = 244; HR, 0.73; 95% CI, 0.13-4.01; HR, 0.34; 95% CI, 0.10-1.10).

CONCLUSION

Minimally invasive radical hysterectomy is associated with increased rates of death and recurrence in patients with stage IB cervical cancer even after controlling for surgeon volume; open radical hysterectomy should be the recommended approach in this population. Although there may be a subset of patients with microscopic early-stage disease for whom minimally invasive radical hysterectomy remains safe, additional studies are required.

Long-Term Risk of Infections After Treatment of Childhood Leukemia: A Population-Based Cohort Study Using Administrative Health Data

PURPOSE

Infections are a frequent complication during childhood leukemia treatment. Little is known about the infectious risk in survivors. We compared the relative rate (RR) of infections after treatment completion between pediatric leukemia survivors and the general population.

METHODS

We performed a retrospective, population-based cohort study of children diagnosed with leukemia between 1992 and 2015 in Ontario, Canada, who were alive and relapse free 30 days after treatment completion (index date). Leukemia survivors were matched 5:1 with the general population by year of birth, sex, and rural status and stratified by initial treatment, including and excluding hematopoietic stem-cell transplantation (HSCT). The primary outcome was time to infections, as identified using validated diagnostic codes from administrative databases. Individuals were censored at the earliest of death, first relapse, loss to follow-up, or end of study.

RESULTS

A total of 2,204 leukemia survivors were included and matched with 11,020 controls. The rate of infections was elevated after treatment completion compared with controls (RR, 1.51; 95% CI, 1.45 to 1.57) and at less than 1 year (RR, 1.77; 95% CI, 1.69 to 1.86); 1 to 4.99 years (RR, 1.66; 95% CI, 1.62 to 1.71), and 5 or more years (RR, 1.29; 95% CI, 1.22 to 1.36) from the index date. Among those whose initial treatment excluded HSCT, the rate remained elevated more than 5 years from the index date (RR, 1.29; 95% CI, 1.23 to 1.35). Infection-related death was significantly increased in leukemia survivors both among the entire cohort (hazard ratio, 149.3; 95% CI, 20.4 to 1,091.9) and among those without HSCT (hazard ratio, 92.7; 95% CI, 12.4 to 690.7).

CONCLUSION

A significant association was found between a history of leukemia therapy and an increased risk of infections. Additional study is needed to establish which exposures in patients with leukemia lead to late infections.

The effect of adopting pediatric protocols in adolescents and young adults with acute lymphoblastic leukemia in pediatric vs adult centers: An IMPACT Cohort study

BACKGROUND

Retrospective studies have shown adolescents and young adults (AYA) with acute lymphoblastic leukemia (ALL) have superior survival when treated in pediatric versus adult centers (locus of care; LOC). Several adult centers recently adopted pediatric protocols. Whether this has narrowed LOC disparities in real-world settings is unknown.

METHODS

The IMPACT Cohort is an Ontario population-based cohort that captured demographic, disease and treatment (treatment protocol, chemotherapy doses) data for all 15-21 year olds diagnosed with ALL 1992-2011. Cancer outcomes were determined by chart abstraction and linkage to provincial healthcare databases. Treatment protocols were classified as pediatric- or adult-based. We examined predictors of outcome, including LOC, protocol, disease biology, and time period.

RESULTS

Of 271 patients, 152 (56%) received therapy at adult centers. 5-year event-free survival (EFS ± SE) among AYA at pediatric vs adult centers was 72% ± 4% vs 56% ± 4% (P = 0.03); 5-year overall survival (OS) was 82% ± 4% vs 64% ± 4% (P < 0.001). After adjustment, OS remained inferior at adult centers (hazard ratio 2.5; 95% confidence interval 1.1-6.1; P = 0.04). In the most recent period (2006-2011), 39/59 (66%) AYA treated at adult centers received pediatric protocols. These AYA had outcomes superior to the 20 AYA treated on adult protocols, but inferior to the 44 AYA treated at pediatric centers (EFS 72% ± 5% vs 60% ± 9% vs 81% ± 6%; P = 0.02; OS 77% ± 7% vs 65% ± 11% vs 91% ± 4%; P = 0.004). Induction deaths and treatment-related mortality did not vary by LOC.

CONCLUSIONS

Survival disparities between AYA with ALL treated in pediatric vs adult centers have persisted over time, partially attributable to incomplete adoption of pediatric protocols by adult centers. Although pediatric protocol use has improved survival, residual disparities remain, perhaps due to other differences in care between adult and pediatric centers.