Audit of Cancer Registrations Notified by National Health Service Central Register in England and Wales

Audit of cancer registrations notified by National Health Service Central Register in England & Wales Dr Heather O Dickinson My web page Department of Child Health University of Newcastle I work on the epidemiology of children’s cancer, mainly on cohort studies in relation to exposures around birth:    population mixing infections father’s preconceptional irradiation Heather Dickinson  Learning objectives - understand: importance of ascertainment of cases from multiple sources importance of diagnostic review age - cohort effects Performance objectives - able to: use capture-recapture methods organise a diagnostic review  Cancer registration system in England & Wales 12 regional cancer registries National Health Service Central Register (NHSCR) Cancers notified by NHSCR No. of cancers 100 80 60 40 Notified by NHSCR Expected 20 0 1971 1975 1980 1985 1990 1993 Year of diagnosis Objective For NHSCR cancer registrations, 1971-89, to assess:  Accuracy of diagnoses Were site and type codes correct?  Completeness of ascertainment Did they tell us about ALL the people who got cancer? Accuracy of diagnostic codes from NHSCR Identify hospital of diagnosis Request biopsy, pathology or post-mortem report No Material found and satisfactory? Yes No Hospital ID no. known? Request clinical records No Clinical records obtained? Yes Review Review cancer or death registration complete Yes Pathology review of NHSCR cancer registrations METHOD OF REVIEW n Biopsy 413 Pathology report 68 Post-mortem report 13 Clinical records only 128 Cancer/death registration 66 TOTAL 688 (%) (60) (10) (2) (19) (10) (100) Diagnostic groups We grouped the cancers into diagnostic groups:  we used 13 standard groups for children’s cancer Kramarova et al., 1996  we added 3 groups for adult cancer Level of agreement between diagnostic codes of NHSCR and review LEVEL OF AGREEMENT n (%) Complete agreement in site 262 (38) and type codes Agreement in diagnostic group 378 (55) but not in site and type codes Disagreement in diagnostic 47 (7) group TOTAL 688 (100) Reasons for disagreement REASONS FOR DISAGREEMENT Failures in data capture and coding by NHSCR More or less specific diagnosis than NHSCR Different diagnosis TOTAL DISAGREEMENTS n (%) 19 (3) 5 (1) 23 (3) 47 (7) Reasons for disagreement We were interested in whether the accuracy of diagnosis depended on:  age  how recently the original diagnosis was made Total no. in cohort The Cumbrian birth cohort 100,000 50,000 0-14 yrs 15-24 yrs over 25 yrs 1975 1980 1985 1989 0 1971 Year NHSCR and review diagnoses in different groups % of cases 12 8 4 0 0-14 15-24 over 25 Age (in years) at diagnosis p for heterogeneity = 0.03 NHSCR and review diagnoses in different groups % of cases 12 8 4 0 1971-74 1975-79 1980-84 1985-89 Time period of diagnosis p for trend = 0.07 Completeness of ascertainment by NHSCR Did NHSCR tell us about all the people in our cohort who got cancer? Completeness of ascertainment by NHSCR SOURCE OF NOTIFICATION Age (yrs) N* Children’s Cancer Registries: Region A 0-24 34 Region B 0-14 1 National 0-14 18 Cancer Bureaux: Region A 15+ 18 Region B 15+ 1 NHSCR death registrations All ages 14 *N = No. of cancers not notified by NHSCR Completeness of ascertainment by NHSCR 720 cases notified by NHSCR  694 confirmed as malignant 740 malignancies notified from all sources Ascertainment = 694/740 = 94% Completeness of ascertainment by NHSCR % of cases missed 12 8 4 0 Age (in yrs) at diagnosis p for heterogeneity = 0.02 0-14 15-24 over 25 Completeness of ascertainment by NHSCR % of cases missed 8 6 4 2 0 1971-74 1975-79 1980-84 1985-89 Time period of diagnosis p for trend = 0.83 Completeness of ascertainment by NHSCR Other cases?  Not notified by NHSCR AND  Not notified by any other source Cases not notified by NHSCR or other sources Over 25 years Notified by NHSCR? Notified Yes by other sources? No Yes 215 86 No 11 ? Cases not notified by NHSCR or other sources 15 - 24 years Notified by NHSCR? Notified Yes by other sources? No Yes 194 24 No 15 ? Cases not notified by NHSCR or other sources 0 - 14 years Notified by NHSCR? Notified Yes by other sources? No Yes 175 86 No 0 ? Cancers missed by NHSCR Notified by Age other sources? (in years) NO at diagnosis YES 0-14 15-24 over 25 ALL AGES 20 15 11 46 0.0 1.9 4.4 6.3 TOTAL n (%) 20.0 (10%) 16.9 14.4 51.3 (7%) (5%) (7%) Summary: completeness of ascertainment by NHSCR  NHSCR definitely missed 46 (6%) out of 740 cases notified by all sources. probably missed more cases which were not notified to us by any source.  NHSCR Summary: completeness of ascertainment by NHSCR  NHSCR missed 10% of cases in the youngest age group. NHSCR probably missed at least 10% of cases in other age groups too.  Conclusions Ascertainment from multiple sources &  Diagnostic review  are essential for epidemiological studies of cancer