Cohort studycited by the Tayebat archivefull text checked★ key study A-CHM-11 in-depth review
Late deaths among childhood-cancer survivors have fallen as treatment became gentler
Reduction in Late Mortality among 5-Year Survivors of Childhood Cancer.
Overview
Armstrong GT et al.. N Engl J Med 2016 · PMID 26761625 · DOI 10.1056/NEJMoa1510795
15-year death rate after surviving 5 years dropped from 12.4% to 6.0% between the 1970s and 1990s.
Does the study support the specific claim? Partly · Methodological quality: high
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Track late mortality across eras.
- How was the study done?
Cohort, 34,033 survivors.
- How many people, animals or samples were studied?
34,033.
- What was actually found?
41% of late deaths health-related; mortality halved over time.
- How large was the observed effect?
12.4% -> 6.0%.
- What do the results mean in everyday life?
Late effects are real and being reduced.
- What are the study's limitations?
Observational era comparison; cannot measure improvements in medical care; treatment data imputed for 29%; shorter follow-up for the 1990s.
- What does it say about Elawady's specific claim?
Supports burden and shows progress.
- Do the results apply to Germany and Europe?
Moderate.
- What can explicitly not be concluded from this study?
That treatment should be avoided.
Scientific in-depth review
1. Full reference
- Authors
- Armstrong GT, Chen Y, Yasui Y, Leisenring W, Gibson TM, Mertens AC, Stovall M, Oeffinger KC, Bhatia S, Krull KR, Nathan PC, Neglia JP, Green DM, Hudson MM, Robison LL
- Journal
- N Engl J Med
- Year
- 2016
- Identifier
- PMID 26761625 DOI 10.1056/NEJMoa1510795 · PMC4786452
- Verification
- verified – PubMed E-utilities efetch, 2026-10-25
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
Funding: National Cancer Institute (CA55727), Cancer Center Support (CORE) grant CA21765, American Lebanese-Syrian Associated Charities (ALSAC). Disclosure forms available at NEJM.org (not reproduced in the full-text file).
2. Research question
Has late mortality among 5-year childhood-cancer survivors declined as treatments were modified?
3. Study design and population
- Study design
- Cohort study – Retrospective cohort (CCSS), late mortality by treatment era
- Population
Five-year survivors of childhood cancer diagnosed <21 years with leukaemia, CNS malignancy, Hodgkin lymphoma, non-Hodgkin lymphoma, Wilms tumour, neuroblastoma, soft tissue sarcoma or bone tumour, initially treated 1970-1999 at 31 US/Canadian institutions (about 20% of US children diagnosed in the period)
- Country
USA/Canada (CCSS)
- Period
Diagnosis 1970-1999; mortality follow-up from 5 years after diagnosis to 31 Dec 2007 (National Death Index); median follow-up 21 years (range 5-38); 705,806 person-years
- Sample size
34,033 eligible five-year survivors (over 9,000 from the 1970s, over 13,000 from the 1980s, over 11,000 from the 1990s); treatment data abstracted for 24,243, imputed for 9,790
- Inclusion and exclusion criteria
CCSS eligible diagnoses
4. Methods and endpoints
- Exposure / intervention
Treatment era (5-year or decade intervals) and treatment exposures abstracted from medical records (cranial, abdominal, chest radiotherapy dose; anthracycline dose; other chemotherapy)
- Comparator
Earlier eras
- Endpoints
Late mortality (>=5 years after diagnosis) from National Death Index/death certificates, classified as recurrence/progression, external causes, and health-related causes (subsequent neoplasm, cardiac, pulmonary, other)
- Statistical methods
Cumulative incidence by era and diagnosis; SMRs vs US population; multivariable piecewise exponential models (RR per 5-year era) adjusted for sex, age at diagnosis, attained age, diagnosis; era effect compared with and without adjustment for treatment variables (mediation-type comparison); multiple imputation for missing treatment (9,790), cause of death (440 Canadian cases) and time of death (27); US-only sensitivity analysis; p-values for cancer- and cause-specific differences declared exploratory.
5. Actual results
- Results
3,958 deaths: 2,002 recurrence/progression, 338 external, 1,618 (41%) health-related (746 subsequent neoplasms, 241 cardiac, 137 pulmonary). 15-year all-cause mortality by decade 10.7% (1970s), 7.9% (1980s), 5.8% (1990s) (by 5-year era 12.4% in 1970-74 to 6.0% in 1990-94); recurrence/progression 7.1%, 4.9%, 3.4%; health-related 3.1%, 2.4%, 1.9% (by 5-year era 3.5% to 2.1%) (all P<0.001). Adjusted RR per 5-year era: health-related death 0.86 (95% CI 0.82-0.89), subsequent neoplasm 0.83 (0.78-0.88), cardiac 0.77 (0.68-0.86), pulmonary 0.77 (0.66-0.89). Health-related mortality fell in ALL (3.2% to 2.1%), Hodgkin lymphoma (5.3% to 2.6%), Wilms tumour (2.6% to 0.4%) and astrocytoma (4.7% to 1.8%). Adding therapy to models attenuated the era effect for ALL (RR 0.88 to 1.02) and Wilms (0.68 to 0.80) but not Hodgkin lymphoma (0.79 both) or astrocytoma (0.81 to 0.82). Radiotherapy use fell from 77% (1970s) to 41% (1990s). Neuroblastoma late mortality increased in recent decades.
- Absolute risks
15-year cumulative all-cause late mortality 10.7% (1970s) vs 5.8% (1990s) by decade; health-related 3.1% vs 1.9%; cardiac 0.47% (1970-74) vs 0.13% (1990-94); subsequent neoplasm 1.8% vs 1.0%. 30,075 (88.4%) of 34,033 survivors alive at end of follow-up.
- Authors' conclusion
The strategy of lowering therapeutic exposure has translated into an observed decline in late mortality among 5-year survivors; the all-cause decline is mainly due to fewer deaths from recurrence/progression, but deaths from subsequent neoplasms and cardiopulmonary late effects also fell, extending lifespan.
6. Strengths
- Very large cohort (34,033) spanning three treatment decades
- Long follow-up (median 21 years; 705,806 person-years)
- National Death Index linkage (objective mortality outcome)
- Medical-record treatment data with organ-specific dosimetry
- Explicit test whether treatment changes explain era trends; exploratory p-values acknowledged
7. Weaknesses
- Observational era comparison; changes in supportive care, screening and medical care cannot be measured (authors)
- Health-related deaths cannot be directly attributed to treatment sequelae (authors)
- Treatment data imputed for 9,790 (29%) survivors; cause of death imputed for 440 Canadian cases
- Shorter follow-up for the 1990s cohort (authors)
- Only 5-year survivors; early deaths and treatment efficacy not addressed
- For Hodgkin lymphoma and astrocytoma, the decline was not explained by reduced therapy
- Many comparisons; cancer- and cause-specific p-values exploratory
8. Risk of bias
- Orienting tool
- ROBINS-E
- Type of assessment
- structured assessment based on the full text by a single reviewer (no formal double review)
- Full formal assessment
- no
- Overall risk of bias
- some concerns
- Rationale
Unchanged overall judgement. The causal attribution of declining late mortality to reduced therapy is supported only for ALL and Wilms tumour. Single-reviewer structured judgement from the full text; a formal assessment requires two independent reviewers and a protocol.
| Domain | Judgement | Rationale |
|---|---|---|
| Confounding | some concerns | Era effects adjusted for sex, age at diagnosis, attained age and diagnosis, but concurrent improvements in supportive care, screening and treatment of late effects are unmeasured (authors). |
| Measurement of exposure | some concerns | Treatment abstracted from records with standardised protocols for 24,243; imputed for 9,790 survivors (sensitivity analysis in appendix). |
| Selection of participants | low | All eligible five-year survivors included in the National Death Index search regardless of questionnaire participation. |
| Post-exposure interventions | some concerns | Improved later medical care may delay or prevent death from late effects; cannot be measured. |
| Missing data | some concerns | Multiple imputation for treatment (29%), cause of death (n=440) and time of death (n=27); US-only sensitivity analysis showed no appreciable differences. |
| Measurement of outcome | low | Death and cause of death from National Death Index/death certificates with ICD coding; cause-of-death misclassification possible but likely non-differential by era. |
| Selection of reported result | low | Prespecified mortality categories reported for all diagnoses; detailed supplementary tables. |
Full-text review (25 October 2026): Added decade-level cumulative mortality (10.7/7.9/5.8%), cause-specific absolute rates, era RRs per 5 years (health-related 0.86, 0.82-0.89; cardiac 0.77), treatment-adjusted era models (attenuation only in ALL and Wilms), 88.4% alive, imputation of treatment for 9,790 and authors' limitations; full funding. ROBINS-E overall unchanged (some concerns).
9. Assessment of the statistical interpretation
Shows a real but modest absolute burden of treatment-related late mortality (health-related 15-year mortality 3.1% in 1970s survivors) and its decline (1.9% in 1990s); the larger part of the all-cause decline comes from fewer relapse deaths. The era-treatment mediation analysis is only partly supportive (ALL, Wilms).
10. Relevance to the specific claim
Documents burden and improvement; shows oncology adapts treatment to reduce late effects.
11. Independent contrary findings
Itself.
12. Transferability to Germany and Europe
Moderate.
13. Overall methodological judgement
Key study, correctly quoted in English; the Arabic headline ('remains high for decades') reverses the main message of a marked decline. Full text adds that most all-cause late deaths were from recurrence and that 88.4% of survivors were alive at end of follow-up.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | high |
| Risk of bias | some concerns |
| Precision of results | high |
| Directness | direct |
| Reproducibility | partly replicated |
| Consistency with the overall evidence | consistent |
| Regional transferability | moderate |
| Adequacy of the interpretation (by the archive) | essentially correct but incomplete |
14. Open questions
Late mortality under protocols after 1999; why Hodgkin lymphoma and astrocytoma mortality fell independently of treatment reduction; rising late mortality in neuroblastoma with intensified therapy.
15. Linked fact checks
- Childhood-cancer survivors carry a large burden of chronic disease caused by treatment ('severe chronic conditions in 62%'). – Studies with mixed or unclear results
- The archive's 19 chemo/radiation studies are presented on the homepage as 'scientific studies confirming what Dr. Diaa proposed'; the harm-only study selection implies that the harms of chemotherapy and radiotherapy outweigh their benefits. – Studies with mixed or unclear results
- Oncology only buys time: 'survival rates' are meaningless, patients die during treatment, and early diagnosis only makes you 'die early'. – Studies contradicting or limiting the claim
Use by the Tayebat archive
Note: according to the archive, the studies were compiled by the archive operators. Whether Dr. Elawady himself cited this study is not established.
- What the archive says about the study
Armstrong 2016 NEJM: among 34,033 survivors, 41% of deaths beyond 5 years were from health-related late effects; reduction in treatment exposure was associated with reduced late mortality (EN). AR headline: late mortality from late effects 'remains high for decades'.
- A. What was actually studied?
Late mortality trends.
- B. What did the authors find?
Late mortality halved over three decades as treatment intensity was reduced.
- C. What conclusions do the authors themselves draw?
Lowering therapeutic exposure helped reduce late mortality.
- D. What does the archive derive from it?
Late effects kill many survivors.
- E. Is this derivation scientifically justified?
The English card is accurate (41% of 3,958 late deaths were health-related). The Arabic headline ('remains high for decades') reverses the study's main message: 15-year late mortality nearly halved (10.7% to 5.8% by decade), health-related late mortality fell (3.1% to 1.9%), and 88.4% of the 34,033 survivors were alive at the end of follow-up.
- Adequacy of the interpretation (by the archive)
- essentially correct but incomplete