• To be able to compare death rates, you need deed to standardise death rates, to reflect differences in age distribution between countries
• For example the median age in Sudan is 19 years and the median age in England is nearly 41 years. Clearly simply comparing crude mortality rates without standardisation will not give you a fair and unbiased view of the health status or mortality experience of the population.
• Use standard population with a fixed age structure as comparator. This standard population can be hypothetical.

• Standardisation of rates – approaches:
  • Direct –> comparative mortality ratio
  • Indirect –> standardised mortality ratio

	Country A			Country B
Age Group	No. of deaths	Population	Rate / 1000 people	No. of deaths	Population	Rate / 1000 people
0- 29	500	400,000	1.25	300	900,000	0.33
30-59	700	600,000	1.16	250	500,000	0.5
60	1500	2,000,000	0.75	400	200,000	2
Total	2700	3,000,000	3.16	950	1,600,000	2.83

	Standard Population  (hypothetical)	No. of deaths	Mortality rate	Mortality rate/1000 persons
Age Group
0- 29	50,000	50	0.001	1
30-59	75,000	80	0.00106	1.06
60	100,000	100	0.001	1
TOTAL	225,000

• Direct –> comparative mortality ratio (CMR) (if age range associated deaths known)
  • Age specific mortality rates in both populations need to be known.
  • Compare expected deaths for Country A and Country B when age specific mortality rates are applied to a standard population structure.

Age wise Mortality Rate (no of deaths/total population) x Standard Population for given age bracket (Mortality rate is converted back to basic rate, not per 1000 people.)

• CMR is ratio of standardised rates
  • Results: <1 lower death rate; >1 higher death rate

COMPARITIVE MORTALITY RATIO (B/A) = 1.13 

Thus after controlling for age, the mortality in country B is 13% higher than in country A.

• Indirect –> standardised mortality ratio (SMR) (if age range associated death rates not known but total number of deaths are)
  • Compare age-wise mortality rates of a standard population against population of each age group in given countries.
  • This will give you the expected deaths in a population.
  • SMR is ratio of observed over expected deaths, comparing each country to the standard population.
  • –> percentage/proportion relative to 100%
  • This statistic will tell you how many deaths would be expected in each country if they had the same age specific mortality ratios as the standard population. 

	Mortality rate for standard hypothetical population	Country A	Country B	Country A expected deaths	Country B expected deaths
Age Group		Population	Population	Age specific mortality rate for Country A x Population A	Age specific mortality rate for Country A x Population B
0- 29	0.001	400,000	900,000	400	900
30-59	0.0016	600,000	500,000	960	800
60	0.001	2,000,000	200,000	2000	200
Total		3,000,000	1,600,000	3360	1900

Thus IF age specific mortality rates in the 2 countries were the same as the standard population then,

Country A  : 

Observed deaths 2700

Expected deaths 3360

SMR Country A  = 0/E x100 =  80.35%

Country A would have just over 80 deaths for every 100 deaths in the standard population. Or it would have 20% less deaths than the standard population.

Country B

Observed deaths = 950

Expected deaths = 1900

SMR Country B = 0/E x100   = 50%

Country B would have 50 deaths for every 100 deaths in the standard population. Or country B would have 50% less deaths than the standard population.

NOTES:

In most cases the indirectly standardised mortality rates will not be directly comparable especially when the age structures are different. The SMRs can only be legitimately compared with the standard population and not with each other and and in most cases cannot be ranked.

As the SMR is calculated with reference to a standard population, the results will be influenced by the standard population and as such inferences can only be made if the reference standard population is truly representative or ‘standard’. Changing this standard will change the SMR.

In our case the age wise mortality rates in the two countries are non homogenous (increases with age in Country B and decreases with age in country A) and the age structures are widely different. Thus there is significant risk of bias with standardisation. In fact the choice of a representative standard population is probably the most important thing if using standardisation in real life.