Breast cancer – 1 year absolute survival rates for breast excision for cancer

Metric

Adjusted 1 year survival rates for breast surgery for cancer.

N.B. This will be presented as adjusted survival rates but will be calculated as adjusted death rates and simply reversed for presentation

Numerator

Deaths in or out of hospital between -1 and 365 days inclusive, for any cause, after the index operation date (or admission date if missing); the ONS date of death will be used and can be a day before the date of discharge for patients dying in hospital, for administrative reasons

Denominator

Any inpatient having a breast excision procedure  of one of the following OPCS codes:

• B27 Total excision of breast
• B28 Other excision of breast (excluding B284)

And a primary diagnosis with the following ICD10 diagnostic code:

•  C50 Malignant neoplasm of breast

If they had more than one such admission during the time period of interest, then their first will be taken. Patients are excluded if they have had a similar admission in the previous three full years (i.e. 3×365 days before the index admission date)

 Exclusions

• OPCS Procedure Code: B284
• Surgical subgroup 4 will be excluded. The denominator procedure groups “Excision of breast” and “Excision of breast lump” will be organised into four subgroups within the DFI tools and for the purposes of risk adjustment; see Notes below for details.
• Patients with a secondary tumour in any organ apart from axillary and upper limb lymph nodes (i.e. patients with ICD codes C770-C772, C774-C779 or C78-C79 in any of the diagnosis fields apart from DIAG1),
• Those receiving palliative care (if any episode in the spell has the treatment function code 315 or contains ICD10 code Z515 in any of the diagnoses fields)

Data Source

Linked HES-ONS

Time frame

April 2005 – March 2008 (to allow one-year follow-up with ONS data). Three years of data are required to ensure sufficient power.

Basis

Acute Trust

Statistical methods used

Adjust for the following factors using logistic regression:

• age group
• sex
• emergency admission
• deprivation (exclude those with missing deprivation codes)
• Procedure subgroup
• Charlson score
• number of emergency admissions in prior 12 months
• source of admission
• year
• Age group – Charlson score (interaction terms)

Categorical variables which include small group sizes will need to be re-categorised into larger categories to ensure that the model converges (this is particularly likely for age groups, ethnic groups and source of admission groups.)

Include provider as factor in the model to obtain provider-specific odds ratios (parameterise so that each but one is compared with the national average; the log odds ratio for the omitted one is obtained by summing the log odds ratios for all the others). Multiply provider-specific odds ratios by crude overall (national) odds of death to get adjusted odds for each provider.

Convert these adjusted odds of death into adjusted survival rates.

Bring the 95% CIs from the regression and convert them in the same way as the odds ratios.

Present as tables and binomial funnel plots (with 99.8% confidence limits).

Notes

 ICD10 contains no TNM staging information therefore we have tried to reproduce the essence using primary and secondary diagnosis fields. ‘Local spread’ requires the presence of a code for a local lymph node AND the absence of codes for distant nodes or organs. ‘Distant spread’ requires the presence of a code for distant nodes or organs (there may also be codes for local node or organ involvement present).

The denominator procedure groups “Excision of breast” and “Excision of breast lump” will both be organised to contain the following procedure subgroups within the DFI tools and for the purposes of risk adjustment. This indicator is a collation of the first 3 subgroups.

1. Breast, no nodes
2. Breast, local spread
3. Breast, distant spread
4. All other disease (includes ca in situ in breast)

Subgroup title	Primary diagnosis codes	Secondary diagnosis codes
1. Breast, no nodes	C50	Any excl C77-C79
2. Breast, local spread	C50	C773 (axillary and upper limb lymph nodes) if no other C77-C79
3. Breast, distant spread	C50	Any C excl C773
4. All other diseases (includes ca in situ in breast)	Any except C50	Any