Large variation in assessment and outcome definitions to describe the burden of long-term morbidity in childhood cancer survivors: A systematic review.

We systematically reviewed outcome assessment methods, outcome classification, and severity grading of reported outcomes in studies investigating the burden of physical long-term morbidity in childhood cancer survivors (CCS). A MEDLINE and EMBASE search identified 56 studies reporting on three or more types of health conditions in 5-year CCS, for which information was extracted on outcome types and classification, methods of outcome ascertainment, and severity grading. There was substantial variability in classification and types of health conditions reported and in methods of outcome ascertainment. Only 59% of the included studies applied severity grading, mainly the common terminology criteria of adverse events. This large variation in assessment and definition of the burden of physical long-term morbidity in CCS challenges interpretation, comparison, and pooling data across studies. Global collaboration is needed to standardize assessments and harmonize definitions of long-term physical morbidity and associated outcomes in childhood cancer survivorship research.

Risk and associated risk factors of hospitalization for specific health problems over time in childhood cancer survivors: a medical record linkage study.

Childhood cancer survivors (CCS) experience higher hospitalization rates compared to the general population for neoplasms, circulatory diseases, endocrine/nutritional/metabolic diseases and eye disorders. We studied trends in hospitalization rates and associated patient and treatment-specific risk factors for diagnosis subgroups among these four diseases. We performed medical record linkage of a ≥5-year CCS cohort with national registers, and obtained a random reference sample matched on age, gender and calendar year per CCS. For each diagnosis subgroup we compared hospitalization rates and trends over time in CCS and the reference population. Further, we analyzed risk factors for hospitalizations within the four CCS diagnosis groups. We used multivariate Poisson regression for all models. We retrieved hospitalization data from 1382 CCS and 26,583 reference persons. CCS had increased hospitalization rates for almost all diagnosis subgroups examined. Hospitalization rates for endocrine/nutritional/metabolic diseases appeared to increase with longer time since primary cancer diagnosis up to 30 years after primary cancer diagnosis. Survivors initially treated with radiotherapy had increased hospitalization rates for neoplasms (P < 0.001), those initially treated with anthracyclines (2.5 [1.1-5.5]) and radiotherapy to thorax and/or abdomen (9.3 [2.4-36.6]) had increased hospitalization rates for diseases of the circulatory system, and those initially treated with radiotherapy to head and/or neck had increased hospitalization rates for endocrine/nutritional/metabolic diseases (6.7 [3.5-12.7]) and diseases of the eye (3.6 [1.5-8.9]). Our study highlights that long-term health problems resulting in hospitalizations are still clinically relevant later in life of CCS. The identified treatment-related risk factors associated with hospitalizations support targeted follow-up care for these risk groups of CCS.

A new method to facilitate valid and consistent grading cardiac events in childhood cancer survivors using medical records.

BACKGROUND: Cardiac events (CEs) are among the most serious late effects following childhood cancer treatment. To establish accurate risk estimates for the occurrence of CEs it is essential that they are graded in a valid and consistent manner, especially for international studies. We therefore developed a data-extraction form and a set of flowcharts to grade CEs and tested the validity and consistency of this approach in a series of patients. METHODS: The Common Terminology Criteria for Adverse Events version 3.0 and 4.0 were used to define the CEs. Forty patients were randomly selected from a cohort of 72 subjects with known CEs that had been graded by a physician for an earlier study. To establish whether the new method was valid for appropriate grading, a non-physician graded the CEs by using the new method. To evaluate consistency of the grading, the same charts were graded again by two other non-physicians, one with receiving brief introduction and one with receiving extensive training on the new method. We calculated weighted Kappa statistics to quantify inter-observer agreement. RESULTS: The inter-observer agreement was 0.92 (95% CI 0.80-1.00) for validity, and 0.88 (0.79-0.98) and 0.99 (0.96-1.00) for consistency with the outcome assessors who had the brief introduction and the extensive training, respectively. CONCLUSIONS: The newly developed standardized method to grade CEs using data from medical records has shown excellent validity and consistency. The study showed that the method can be correctly applied by researchers without a medical background, provided that they receive adequate training.