12
Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Embed Size (px)

Citation preview

Page 1: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data

Colin Fox (NICR)Richard Middleton (NICR)Denise Lynd (BCH – COIS)

Page 2: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Clinical Oncology Information System (COIS)

Radiotherapy Data – treatment intent, machine parameters, fractions, tumour site, dates.

Clinical Staging Data – from year 2000 all cancers exc. NM Skins (TNM, site-specific classifications, overall stage).

Chemotherapy Data – regimes, number of sessions, dates, drug details, dosages.

Detailed Patient Annotation Notes. Approximately 4200 registrations

annually of which around 2500 involve XRT.

Page 3: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Methods 1999-2002 COIS Radiotherapy data (13,730

XRT sessions relating to approx 9400 patients).

Patient and tumour data from COIS was matched to NICR Cancer Registration Database.

NICR tumour data relating to site, behaviour and diagnosis date were extracted for matched COIS registrations for which no tumour match was obtained.

COIS site and behaviour data as well as XRT start dates compared with corresponding NICR registration data using SPSS.

Page 4: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Matching Results

Exact match at patient and tumour level. 12,331 XRT sessions (89.8%) 8348 registrations (89.0%)

Match at patient level but not tumour. 1232 XRT sessions (9.0%) 868 registrations (9.3%)

Not Matched to NICR 167 XRT sessions (1.2%) 167 registrations (1.8%)

Page 5: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

XRT Patients Not On NICR

Breast 37 (1% of annual incidence) NM Skin 35 (0.4% of annual incidence) Prostate 17 (0.8% of annual incidence) Lung 16 (0.45% of annual incidence) Other Sites 62 (0.35% of annual

incidence)

167 cases<0.5% of annual

incidence

COIS registrations for selected sites that were missing from the NICR database:

Page 6: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

No Tumour Match: Site Different

NM Skin 58 (0.42% of XRT cases) Site Unspecified 43 (0.31% of XRT cases) Lung 33 (0.24% of XRT cases) Breast 24 (0.17% of XRT cases)

Other Sites 158 (1.15% of XRT cases)

316 (2.3% of XRT cases)

Malignant sites differed between NICR and COIS in 316 XRT sessions. Number of COIS cases differing on site for selected sites:

Page 7: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

No Tumour Match:Near Site

Lung & Mesothelioma 85 (0.62% of XRT cases) Head & Neck 67 (0.49% of XRT cases) Colon & Rectum 38 (0.28% of XRT cases) Stomach & Oesophagus 34 (0.25% of XRT cases) Lymphomas 27 (0.2% of XRT cases) Cervix & Uterus 25 (0.18% of XRT cases) Other Sites 63 (0.46% of XRT cases)

339 (2.47% of XRT cases)

“Near” meaning the malignant site is similar between both NICR and COIS (339 XRT sessions). Number of COIS cases differing for selected sites:

Page 8: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

No Tumour Match:Behaviour Differences

Site Unspecified 138 (1% of XRT cases) Breast 133 (0.97% of XRT cases) Head & Neck 82 (0.6% of XRT cases) Lung 68 (0.5% of XRT cases) NM Skin 60 (0.44% of XRT cases) Brain 53 (0.39% of XRT cases)

Other Sites 55 (0.4% of XRT cases)

589 (4.3% of XRT cases)

Both NICR and COIS agree on the site but one of the sources has it as non-malignant or metastatic (589 XRT sessions). Numbers of COIS cases differing onbehaviour for selected sites:

Page 9: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Comparison of Diagnosis and XRT Start Dates

XRT began after diagnosis in majority of cases. XRT administered just before diagnosis in around

4.5% of cases, with 3% getting XRT well before NICR diagnosis date.

Main sites – lung, breast, NM skin, head/neck and prostate.

Reasons? Recurrent cases may account for some Metastases requiring early XRT before diagnosis of the

primary tumour Possible missed pathology

Requires further investigation.

Page 10: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Composition of NICR Tumour Registrations 1999-

2002

PAS Pathology

COIS

Deaths

2.5%

10.8%(8.7% dths)

23.6%(4.2% dths)

0.6%(<0.1% dths)

39.1%(14.9%)

18.4%(7.6%)2.2%

(1.8%)

2.7%(0.8%)

(83.8%)

(23.9%)

Page 11: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Summary and Conclusions

Over 90% of COIS data matches exactly Of the 8% where tumour site differed, only 2%

was significant (completely different tumour site to what was registered)

Highlighted missing registrations – approx 2% over the 4 years of COIS data (0.5% per year)

Helped reduce further an already low DCO rate Just over 3% of cases where radiotherapy

started significantly earlier than NICR diagnosis date.

Page 12: Evaluation of Electronic Radiotherapy Data for Quality Checking Cancer Registry Data Colin Fox (NICR) Richard Middleton (NICR) Denise Lynd (BCH – COIS)

Recommendations

COIS represents a clinically rich new source of electronic data for NICR and will provide for:

Verification of NICR tumour registrations Enhancement of treatment data on NICR Electronic clinical staging data

Follow-up on cases where: COIS tumour site differed completely to NICR and XRT start dates occurred significantly earlier than the diagnosis

date on NICR Revision of system update rule for diagnosis

date. Carry out a similar investigation with

chemotherapy data when available.