An Investigation into Childhood Leukaemia in Northampton

Return to the Index Page
 

Childhood leukaemia in Northamptonshire - 1995
4.1

How many cases are there in the Pembroke Road area? 

4.1.1

What the Health Authority were told initially: 

We were initially alerted to 5 cases of childhood leukaemia in the Pembroke Road area. The children all lived in Northampton in the NN5 postal area and the smallest part of the post-code the five children had in common was NN5 7**.   One child lived in Countess Road which is adjacent to Pembroke Road (see map)
4.1.2

Were there any other children with leukaemia? 

We looked for other cases of childhood leukaemia in Northamptonshire, looking in detail at NN5 and the NN5 7** post-code sector, using various methods including: 
  • checking with the Oxford Cancer Intelligence Unit who keep a register of all people diagnosed with cancer in the region
  • reviewing the Northamptonshire database of deaths
  • contacting the hospital doctors who would be treating a child if they had leukaemia
  • checking with the Childhood Cancer Research Group who keep a register of all children with cancer in the country
We found no other children with leukaemia from the area of the identified cluster.  One of the children we had listed as living in NN5 7** had a different post-code at diagnosis recorded on both the cancer registers and does not live in NN5 7**.   We sought and obtained confirmation that the child did not live in NN5 7** and had not done so when diagnosed.  This is child 5 in Table 1 - Children with leukaemia in the post-code sector NN5 7** from 1962 to date below. 

Since 1993 there has been one more child in the NN5 7** post-code sector diagnosed with leukaemia (child 6).   This child does not live on the Spencer Estate. 
 

Table 1 - Children with leukaemia 
in the post-code sector NN5 7** from 1962 to date
Child  Year of Diagnosis Year of Birth
1 1986 1982
2 1989 1979
3 1992 1988
4 1993 1985
(5) (1990) (1989)
6 1994 1989
 

We also plotted cases of childhood leukaemia on maps of the county to look at the geographical distribution.  A typical map is shown below.  The cases are not as clustered as they appear at first glance because the solid areas are areas where a lot of people live while the white areas are largely rural.  We expect there to be more cases in the areas where more people live. (Click on the map to see a larger version.) 
 

 
4.2

Is this just a coincidence?

 
4.2.1

Rates of childhood leukaemia using standard geographic boundaries:

How can we work out the wider implications of these five cases?  Is this a statistically significant cluster or not?  The way we do this is by comparing local figures with what we would expect if national rates applied locally and seeing whether there are more cases than expected.  For reasons that will be explained in section 5 below, this should be done using areas that are defined in advance rather than areas that are identified by drawing as tight a line as possible around the cases.  There are 7 geographic districts that we normally use in Northamptonshire to review health data.  Table 2 below gives both the number of cases and the age-standardised rates of leukaemia in each of these standard geographic districts for both the 15 year period 1969-88 and for the last 6 years.  The NN5 7** post-code sector is in Northampton district. 
 
 
Table 2 - Age-standardised rates of childhood leukaemia in Northamptonshire by District for 1969-1988 and 1989-94*
District   Number of children with leukaemia  Age-standardised rates of  leukaemia in children under 15 per million children under 15 
1969 - 88 1989 - 94 1969 - 88 1989 - 94*
Corby 12 2 44.3 31.4
Daventry 10 3 38.3 39.5
East N/hants 16 5 60.0 62.3
Kettering 8 3 26.1 32.6
Northampton 26 9 38.2 38.1
South N/hants 19 5 64.0 58.6
Wellingborough 15 2 52.4 24.1
Northants 106 29 44.8 39.9
England 7431 N/A 37.8 N/A
*Note: 1994 Population estimates not yet available, so 1993 figures used for that year 
¶ Figures supplied by CA Stiller from the Childhood Cancer Research Group 

The data show that in Northampton the annual rate of childhood leukaemia, at 38 cases per million children under 15, is not unexpectedly high. We think these are the appropriate areas to use to calculate the rates of childhood leukaemia if we are to be able to make unbiased comparisons with the rest of the country.  The reasons for this are explained in section 5.3.
4.2.2  

What are the rates of childhood leukaemia in the Pembroke Road area, itself? 

The Health Authority believe that using narrow post hoc boundaries (that is, geographic areas that are defined after a possible cluster is identified) to calculate rates and comparative figures is inappropriate.  This approach is methodologically flawed and biased and tends to produce apparently significant clusters where there are none.  Why this happens is explained in detail in Sections 5.2 and 5.3 below.  Nonetheless we give the rates based on narrow post hoc boundaries below, as they have been requested by the parents.  These figures need to be interpreted with caution for reasons which will become apparent later.  Tables 3 - 6 below give comparative figures for both NN5 and NN5 7**.  (The figures have been supplied by the Childhood Cancer Research Group from the National Registry of Childhood Tumours 5). 

We do not know what causes leukaemia.  However, we know that different cancers may share common risk factors, so we have also provided information about all childhood cancers as well as leukaemia. 
 
 

Table 3 - All Childhood Cancers in NN5 from 1962 - 1994
 Number *Annual rate per million children under 15
1962 - 1974 11 152.0
1975 - 1984  4 71.9
1985 - 1994 8 143.8
1962 - 1994  23 125.2
 
 
Table 4 - All Childhood Cancers in NN5 7** from 1962 - 1994
 Number *Annual rate per million children under 15
1962 - 1974 5 173.5
1975 - 1984 1 45.1
1985 - 1994 5 225.5
1962 - 1994 11 150.4
 
 
Table 5 - Childhood Leukaemia in NN5 from 1962 - 1994
 Number *Annual rate per million children under 15
1962 - 1974 2 27.6
1975 - 1984 1 18.0
1985 - 1994 6 107.8
1962 - 1994 9 49.0
 
 
Table 6 - Childhood Leukaemia in NN5 7** from 1962 - 1994
 Number *Annual rate per million children under 15
1962 - 1974 0 0
1975 - 1984 0 0
1985 - 1994 5 225.5
1962 - 1994 5 68.3
* Rates are crude rates for the population under 15, that is, they are not age-standardised because age grouped data was not available. They are calculated on the basis of the 1991 census population because sufficiently detailed information broken down by post-code is not available before this.
4.2.3

 What do these rates mean? 

As we said before, to interpret these rates we need to compare them with the expected rates derived from elsewhere.  There are all sorts of ways we can compare rates.  Below we give the observed and the expected number of cases and calculate the probability that the observed number (or an even higher number) of cases might occur simply by chance if the true rate in the Pembroke Road area were the same as elsewhere.  The relative risks are also given. 

These figures are calculated for the period 1985-1994 when there was a peak incidence in cases of leukaemia, that is to say, we have taken the "worst possible case scenario". 

Childhood Leukaemia in NN5 1985-1994: 
 

  • Annual rate per million children under 15 = 107.8
  • Number of cases observed = 6
  • Expected annual rate per million children under 15 = 40.5 
  • Average number of cases expected  = 2
  • The probability of seeing six or more cases by chance if the true rate is the same as the national rate is about 1 in 40
  • Relative Risk = 2.7
Childhood Leukaemia in NN5 7** 1985-1994:  
  • Annual rate per million children under 15 = 225.5
  • Number of cases observed = 5
  • Expected annual rate per million children under 15 = 40.5 
  • Average number of cases expected  = 1
  • The probability of seeing five or more cases by chance if the true rate is the same as the national rate is about 1 in 400
  • Relative Risk = 5.6
NB This figure is the crude rate/million children under 15 for Northamptonshire for the same period (national figures are not yet available but these will be very similar).  

Another way of looking at the figures for NN5 7**, is that we would expect there to be at least 15 clusters of childhood leukaemia in the UK with the same or a larger number of cases of childhood leukaemia as seen in the Pembroke Road area (where there are the same number of children we have) purely due to chance 6. 

When interpreting these figures it must also be remembered that the calculated probability of a particular cluster occurring (if the true rate is the same as the national rate) is much smaller that the true probability of any cluster of this size or larger having occurred.  This is explained in section 5.2 below. 

All Childhood Cancers 

Childhood cancers other than leukaemia are not higher than expected in the Pembroke Road area.. 

The annual rate for all childhood cancers for the period 1962 -1994 is 125 per million children under 15 for NN5 and 150 per million children under 15 for NN5 7**.  The national annual rate for all childhood cancers is 115 per million children under 15.  At first sight this might suggest that other childhood cancers are also higher than expected in these areas.  However, this is not true.  The excess is simply due to the raised leukaemia rates (as we noted above leukaemia accounts for over 1/3 of childhood cancers).  The rate for all childhood cancers other than leukaemia in NN5 is 76 per million children under 15 which is exactly the same as the national rate 5.  The figure for NN5 7** is 82 per million children under 15.  This is not significantly different from the national rate.
 
 
Back to Section 3 Return to the Index Page Forward to Section 5
 
 

© Northamptonshire Health Authority, reproduced by kind permission of Dr Amanda Burls, Sen Reg in Public Health Medicine.

Created using: Lightning HTML Editor Version 2.10.1997