CANCER IN YORKSHIRE 1989-93

STATISTICAL METHODS  

Definitions - Registrations and Deaths - Age-Standardised Rate -
Standardised Registration Ratio - Cumulative Rate/ Risk -
New Patients per GP - Health of the Nation Targets -
Socio-economic Measures - Socio-economic Profile Classification - Investigation, Treatment and Prognosis - Survival - Waiting Times - Formulae


Definitions

For the purposes of this report, 'Region' refers to the former Yorkshire Regional Health Authority, and District' refers to the district of residence, a nd corresponds to the District Health Authority of the period. Statistics are provided for patients who were resident within the Yorkshire Health region at the time of diagnosis. This includes patients treated both within the region and extra-regionally.

The statistical software package SAS was used to produce all the analyses included in the report.

All populations referred to in the methodology are the ONS mid-year population estimates based on the 1981 or 1991 censuses.


Registrations & Deaths

A registration is any new case of primary invasive cancer, identified by the Yorkshire Cancer Registry, arising in the population under study. The incidence rate gives the annual number of invasive registrations per 100,000 population.


Age-Standardised Rate

Age-standardised registration rates (ASRs) have been computed where the comparison of incidence between groups was of interest. This rate enables such comparisons to be made allowing for differences in their population structures, and is equivalent to the rate that would be seen if the standard population were subject to the same rates as that of the group. Where districts, purchasing authorities or socio-economic groups have been compared, the standard population is that of the Yorkshire region. ASRs have also been produced for Yorkshire, standardised against the European and World standard populations.

The graphs showing trends in incidence in Yorkshire have been standardised against the Yorkshire population in 1993.

To obtain the observed annual rate by five-year age groups for each district, the total number of registrations between 1989 and 1993 was divided by the district population for that period.

The ASR was then calculated by multiplying the Yorkshire population for the five-year period by the observed rate, within each age group. The result was summed and divided by the total regional population for the period (Formula 1). This is known as the direct method of age standardisation.

The charts show the ASR as a dot, with the 95% Confidence Interval for the ASR as either an error line (districts) or a box (socio-economic profile) around it. The Yorkshire rate is shown as a double line, the middle representing the rate and the line thicknesses depicting the confidence interval.


Standardised Registration Ratio

Standardised Registration Ratios (SRRs) provide an alternative method for comparing incidence in two populations, whilst allowing for differences in age structures. Here the purpose was to compare the registration rate in Yorkshire and that in other health regions in England and Wales over the period 1990-92.

The expected rate for a region is the rate which would be seen if the region were subject to the same rates as the standard population (England and Wales). To obtain the expected number of cases, the population of each region for the years 1990-92 was multiplied by the England and Wales rate of cancer in the same years, within five year age groups. The expected rate is found by summing the result and dividing by the total regional population for the three year period. (Formula 2). The SRR is the ratio of the observed and expected rates multiplied by 100. The SRRs were calculated using data supplied by ONS from each registry in England and Wales.

 


Cumulative Rate/ risk

The cumulative risk is defined as the risk that an individual would have of being registered with a cancer during a particular age span (in this case 0-74). The cumulative rate is calculated by summing the age-specific incidence rates (in 5 year intervals) and is expressed as a percentage (formula 3a). The cumulative risk is then estimated by formula 3b, expressed in the form '1 in x'.

 


New Patients per GP

There are approximately 2050 General Practitioners in the Yorkshire region, with an average list size of 1840 people. The number of cases of a cancer a GP could be expected to see have been expressed in the form '1 per x years', based on the average list size and the incidence rate of the disease.

 


Health of the Nation Targets

The Health of the Nation targets for lung cancer and breast cancer are to reduce the mortality or incidence rates in relation to the 1990 rates. The target for cervical cancer is to reduce the incidence rate of invasive disease in relation to the 1986 rate. The target for melanoma is to halt the year-on-year increase in incidence. This could be depicted by plotting the year-on-year difference in rates; the target then corresponding to a difference of zero. In this report, however, to be consistent with the plots for the other cancer targets, the trend in melanoma incidence has been shown.

It should be noted that the axes of these charts have been broken to make it easier to compare trends in incidence with the Health of the Nation targets.

 


Socio-economic Measures

Socio-economic factors can be examined in a number of ways. One method is to collect socio-economic information, such as occupation, on each individual with cancer. It has the advantage that the socio-economic status of each individual patient is taken into account in any analysis. However, registry data on occupation is incomplete and unreliable, especially for females.

The approach taken in this report was to use census information on the demographic characteristics of the areas in which patients were resident. This information was combined with registry data on the addresses of patients. Each patient was assigned to the socio-economic group (Super Profile) of their enumeration district of residence.

 


Socio-economic Profile Classification

This is based on an analysis of 120 original census variables, at Enumeration District level, many of which are highly correlated. A transformation is applied to these variables by Principal Components Analysis to create uncorrelated derived variables. The EDs are then grouped together using Cluster Analysis based on the new derived variables or principal component scores.

Initially there are 160 relatively homogenous profile groups. These have been further aggregated by cluster analysis into 40 groups and then into 10 groups, or 'Super Profiles'. The names attached to the different Super Profiles are an attempt to capture the wider characteristics of the groups in a name that can be easily referred to.

A summary and characteristic description of the 10 Super Profile groups and the 40 groups contained within them is given below.

  • Group I - 'Affluent Achievers'
    Very high income professionals in exclusive areas Mature families with large detached properties in 'stockbroker belts' Mature families in select suburban properties
  • Group II - 'Country Life'
    Prosperous and farming communities Small holders and rural workers (mainly Scotland)
  • Group III - 'Thriving Greys'
    High income households in genteel neighbourhoods Affluent aging couples, many in purchase property Older professionals in retirement areas Comfortably well off older owner occupiers Affluent aging couples in rural areas
  • Group IV - 'Settled Suburbans'
    White collar families in owner occupied suburban semis Mature white collar couples in established suburban semis White collar couples in mixed suburban housing
  • Group V - 'Nest Builders'
    Mortgaged commuting professionals with children in detached properties Double income young families in select properties Military families Young white collar families in small semis and terraces Young white collar families in smaller semis Young blue and white collar families in semis and terraces Young families in terraces, mainly council
  • Group VI - 'Producers'
    Older blue collar owner occupiers in semis Older workers established in semis and terraces Older and retired blue collar workers in small council properties
  • Group VII - 'Senior Citizens'
    Retired white collar workers in owner occupied flats Older residents and young transient singles many in seaside towns Old and young buying terraces and flats Retired blue collar workers in council flats, mainly in Scotland
  • Group VIII - 'Urban Venturers'
    High income young professionals mainly renting (mainly Greater London) Young white collar workers in multi-racial areas (mainly London) Young professionals buying property Young families buying terraces in multi-racial areas Young families renting basic accommodation Young white collar singles sharing city centre accommodation
  • Group IX - 'Hard-Pressed Families'
    Blue collar families in council properties Young blue collar families in council terraces Manufacturing workers in terraced housing.
  • Group X - 'Have-Nots'
    Families in council flats in multi-racial areas with high unemployment Blue collar young families in council properties with high unemployment Young families, many single parent, with high unemployment Young singles and pensioners in council flats with high unemployment

 


Investigation, Treatment & Prognosis

The treatment distributions were calculated using data which relates to the first nine weeks of definitive treatment only. In breast and prostate cancers only, hormone therapy was counted as definitive treatment. Diagnostic and investigative procedures are not included as surgery.

 


Survival

Survival distributions were estimated by the Kaplan-Meier method. Allowance was made for non-cancer deaths by adjustment using relative survival methodology. The expected survival of the patients was based on England and Wales life tables.

Death certificate only registrations were excluded, as they could not contribute to any survival analysis.

It should be noted that for those cancer sites where prognosis is very poor, the axes of the charts of 5-year survival have been broken to aid making comparisons.

Survival estimates have not been shown for categories with less than five cases.

 


Waiting Times

The intervals between the date of GP referral and date of first hospital visit, and between the date of first hospital visit and date of first treatment were calculated, where these dates were known to the Registry. Same day referrals have been regarded as emergency admissions and have been excluded from the analysis. Medians were calculated for each of these intervals and have been depicted as a stacked total interval for each of the districts. The overall Yorkshire intervals are shown as lines crossing the bars.

 


Formulae

1. Age-Standardised Registration Rate Age-Standardised Registration Rate


2. Expected Registration Rate Expected Registration Rate
3. Cumulative Risk

If the cumulative rate (CR) during a particular age span is given by: Cumulative Risk

a) then expressed as a percentage: Cumulative Risk

b) or in the form '1 in x': Cumulative Risk


4. Relative Survival Let the 'normal' population be a population with the same age and sex structure as the cohort of patients with oesophageal cancer in Yorkshire, but subject to the same survival rates as the general population of England and Wales then:

Relative Survival


contents - foreword - introduction - executive summary - acknowlegements -
activities of the registry - key site analysis (graphics) - more graphics - statistical tables - statistical methods - population tables - registration form