LOS CACHORROS

One of the most used methods for comparing disease patterns in different regions or between different populations in the same area, or for the same population over time involves the calculation of incidence rates. However, simply using the crude rate, where the observed number of cases over a given period of time is divided by the population and the time period, may give a misleading result, because of failing to consider the difference in the age structure (Boyle and Parkin, 1991). The comparisons of crude rates between populations with different age and gender structures seem to make no sense. Although it is possible to look at the incidence rate for the specific age to eliminate the age effects, it is a rather cumbersome procedure when comparing several populations. And the results are usually difficult to present (Bland, 1995).

For the purpose of comparing different incidence rates of different populations, a single summary from the age specific rates, which is accomplished by age standardisation, is often useful and calculated. There are two basic methods of age standardisation, direct and indirect methods. (Bland, 1995, Boyle and Parkin, 1991) Both of the two methods use a study population and a reference population, known as

Standard Population, to generate weighted average age-specific rates, but based on different weighting schemes. And each method has advantages and disadvantages.

(Higham et al., 2005)

In direct standardisation, the directly standardised rate (DSR) is the incidence rate expected in a standard population calculated by multiplying the observed age-specific rates of the study population to the proportion of the population in the standard population and summing the results for all age specific groups. In most circumstances, rather than DSR, the comparative mortality (morbidity) figure (CMF) is used to measure death or incidence of disease. The CMF is the ratio of the number of deaths (or incidences) expected in the standard population when applying the age-specific rates of the study population to the standard population to the number of deaths observed in the standard population.

1 k i i i p i N d DSR N n  



Expected incidence in standard population Observed incidence in standard population

CMF  (3.2) 1 k i i i i d N n CMF D  



N : The number of people in the ithgroup of the standard population.

p

N : The total number of population in the standard population.

i

d : The number of cases in the ithgroup of study population.

i

n : The total number of population in the ithgroup of the study population. D: The total number of cases in the standard population.

k: The number of groups. (Julious et al., 2001, Higham et al., 2005)

The main advantage of the direct standardisation method is that this method allows the comparisons of groups with different age structure since the rates (or ratios) are standardised to the standard population. So the direct method is generally preferred

for comparing different study groups (Julious et al., 2001). However, when calculating the local age-specific rates, if the incidence is very low, the age specific rate will be poorly estimated (Bland, 1995). The standardised rates and ratios will be relatively unstable (Sorlie et al., 1999). This is the case in this study. Minority ethnic groups usually have a relatively small population as well as a relatively small number of cardiovascular disease cases. When disaggregated at local area levels, the numbers of cases for minority ethnic groups will be even smaller, which will result in unreliable local age-specific rates and directly standardised rates.

Compared with direct standardisation, indirect standardisation does not require calculations of local age-specific rates. When dealing with a small number of cases, indirectly standardised rates are less variant and more precise than directly standardised rates. Thus indirect standardisation has advantages in measuring disease with a small number of cases. Furthermore, in most circumstances, incidence data are not available at local area level for calculating local age-specific rates, and indirect standardisation is the only option (Higham et al., 2005). Therefore the indirect standardisation method is the most commonly used technique to compare deaths or incidences of disease between different geographical areas (Julious et al., 2001). The indirect age standardisation method is a comparison or a ratio between the number of cases observed to the number of cases expected in the study population. The expected number is calculated by multiplying the standard age-specific rate by the study population of that age group and summing the results for all specific age groups in the study population, assuming that the risk of disease in the study population would be the same as the standard population. (Bland, 1995) Applied to mortality data, it is known as the standardised mortality ratio (SMR). And when applied to the incidence data, it is commonly known as the standardised incidence ratio (SIR) (Boyle and Parkin, 1991). The ratio is usually multiplied by 100 to get rid of the decimal point. (Bland, 1995) The standardised ratio is expressed as (Julious et al., 2001):

Observed Number

1 1 K i i i K i i i i Standardized Rat Observed Number Standardized Ratio = n R Observed Number D n i N o   





n : The total number of population in the ithgroup of the study population.

i

R: The standard rate in the ithgroup of the standard population.

i

D : The total number of cases in the ithgroup of standard population.

i

N : The total number of population in the th

i group of standard population.

However, indirect ratios are not standardised to the standard population but to the study populations with different age structures. For this reason, indirectly standardised ratios are not directly comparable with each other unless the age structures of study populations are similar. And the indirect ratios can only be compared with that of the standard population, which is 100. (Higham et al., 2005) If equal to 100, the ratio implies the rate is the same as the standard rate. A number higher than 100 indicates that there is an excess rate or higher risk of a particular disease than the standard population whereas a number below 100 implies the condition of that disease within the population of interest is better than the standard population.

Given the number of events is large enough, more than 10, the approximated 95% confidence interval is calculated as: (Bland, 1995)

100 *O 1.96 *100 * O Lower Limit E E   (3.6) pper 100 *O 1.96 *100 * O U Limit E E   (3.7)