Investigators extracted all malignancy registrations (ICD-10 C00–C97) for individuals under 25 from publicly available National Disease Registration Service (NDRS) statistics for England. The team divided the cohort into children (0–14 years) and teenagers/young adults (TYAs, 15–24 years). Joinpoint regression (SEER v5.4.0) was used to calculate annual percentage change (APC) in age-specific incidence and to detect inflection points in trends. The authors explicitly excluded 2020 from trend models because COVID-19 related diagnostic disruption could bias longitudinal estimates; they nevertheless compared 2020 with 2019 using instant rate ratios (IRRs). The abstract reports overall incidence rates of 147 and 273 per million person-years in children and TYAs, respectively.
The distribution of tumour types differed sharply by age band. In children, almost half of diagnoses were concentrated in two groups: lymphoid leukaemia and brain/CNS tumours, which together accounted for 46% of childhood cancers in the study period. That concentration produced relatively stable overall childhood incidence despite shifts in rarer diagnoses. The TYA profile was broader: Hodgkin lymphoma comprised 15% of TYA cancers and malignant melanoma represented 9%. The study team also noted that one-quarter of male TYA cancers were testicular, indicating a sex-specific diagnostic concentration within the 15–24 age group. The source abstract provides these proportions but did not present detailed year-by-year counts or age-specific incidence tables in the summary.
Several site-specific trends stood out in the abstract. TYA colorectal cancer rose until 2015 and then fell from 2015 to 2019; the reported APC for the post-2015 decline was -37.59% (p < 0.001). The authors linked this abrupt change, in part, to reclassification of appendiceal carcinoid tumours, which can markedly influence small-number trends in younger populations. Thyroid cancer showed a continuous increase throughout the study period, with APC +5.49% (p < 0.001) reported in the abstract. Cervical cancer incidence among the cohort declined from 2015 to 2019 (APC -21.9%, p = 0.03). When 2020 was compared directly to 2019, cervical cancer incidence fell most sharply (IRR 0.38, p = 0.09), a change the authors suggest may reflect screening and diagnostic interruptions as well as effects from the national HPV vaccination programme introduced earlier in the decade.
The study team treated 2020 as an outlier year for trend estimation. They excluded 2020 from joinpoint trend models because pandemic-related disruption to primary care, diagnostic pathways and screening services could produce artefactual declines or delays in registration. To quantify the immediate change, the authors used IRRs to contrast 2020 with 2019; the abstract specifically reports the largest year-on-year drop for cervical cancer (IRR 0.38, p = 0.09). The authors proposed that some observed short-term reductions in 2020 may therefore represent diagnostic delay rather than true reductions in incidence.
The authors conclude that overall cancer incidence in children, teenagers and young adults remains low in England. They also emphasised that observed trend inflections may reflect a mix of real epidemiologic change, diagnostic or classification shifts (for example, appendiceal carcinoid reclassification), national preventive measures such as HPV vaccination, and COVID-19 related disruption to screening and diagnostic services. The PubMed abstract reports key APCs and IRRs but does not provide granular methodological detail, case counts by year, or full subgroup tables—those are available in the published article. The study team excluded 2020 from trend analyses because of pandemic effects; this decision is a documented analytic limitation that preserves longer-term trend interpretation but leaves immediate pandemic-era incidence patterns partially unresolved. The abstract does not report additional sensitivity analyses, age- or sex-stratified APCs beyond the headlines, or information on absolute numbers for rarer tumour sites; readers should consult the full text for those data and for the authors’ detailed discussion of attribution and registry completeness.