Describe the measurements of morbidity and mortality used in epidemiology.

Rates, Ratios and Proportions

• Examples make the distinction clear — the sex ratio (males per 1000 females) is a ratio; the proportion of deaths due to cancer is a proportion; and the death rate per 1000 per year is a rate
• Every rate requires four elements — a numerator, a denominator, a specified time period, and usually a multiplier (1000, 100,000) to give a convenient whole number
• Crude rates apply to the whole population; specific rates apply to a defined subgroup (age-specific, sex-specific, cause-specific); and standardised (adjusted) rates are corrected for differences in composition, usually age
• Why standardisation matters — crude rates cannot be compared between populations of different age structure. A retirement town will have a higher crude death rate than a university town regardless of how healthy it is
• Direct standardisation applies the age-specific rates of the study populations to a standard population, giving the rates that would have occurred if both populations had the same age structure; it requires reliable age-specific rates
• Indirect standardisation applies the age-specific rates of a standard population to the study population, giving the expected number of events; the standardised mortality ratio (SMR) = (observed / expected) × 100
• An SMR of 100 means the mortality is as expected; above 100 means higher than expected. Indirect standardisation is used when age-specific rates in the study population are unreliable or the numbers small, which is common in occupational studies

Morbidity and Mortality Measures

• Incidence measures the number of new cases arising in a defined population during a specified period
• Incidence = new cases in a period / population AT risk during that period
• It measures the risk or rate of developing disease, and is therefore the measure used to study aetiology and to evaluate prevention
• Cumulative incidence (attack rate, risk) is a proportion, with the population at risk at the start as denominator; incidence density (incidence rate) uses person-time as denominator, which handles varying periods of observation
• Only those AT risk belong in the denominator — those already having the disease, or immune, or without the relevant organ, are excluded; and this is a common source of error
• Prevalence measures all cases, new and old, existing at a given time
• Point prevalence is measured at a point in time; period prevalence over an interval, and includes those who had the disease at the start plus those who developed it during the period
• Prevalence measures the burden of disease and is used for planning services and estimating workload and need
• The relationship: prevalence = incidence × duration (in a steady state)
• Its consequences must be reasoned through — prevalence rises if incidence rises, if duration lengthens (better survival without cure), or if cases immigrate; and falls with better cure, with higher case fatality, or with emigration of cases
• Therefore a falling prevalence may reflect either cure or increased deaths, and a rising prevalence may reflect successful treatment — which is why prevalence alone must never be used to judge a programme
• Prevalence is unsuitable for aetiological research, because it is influenced by survival: a study of prevalent cases over-represents those with long-lasting disease, which is survival or prevalence-incidence bias (NEYMAN bias)
• The mortality measures:
• Crude death rate = deaths in a year / mid-year population × 1000
• Specific death rates — by age, sex or cause
• Case fatality rate = deaths from a disease / total cases of that disease × 100. It measures the killing power (virulence) of the disease, and is used chiefly for acute, short-duration illness
• Proportional mortality rate = deaths from a particular cause / total deaths × 100
• The commonest error IN the subject IS TO confuse proportional mortality with cause-specific mortality — proportional mortality has total deaths as its denominator, cause-specific mortality has the population
• The consequence is that a rising proportional mortality from heart disease may mean nothing more than that fewer people are dying of infection, with no change whatever in the risk of heart disease — so proportional mortality cannot be used to compare risk between populations
• Survival rate — the proportion surviving for a specified period (commonly 5 years)
• The maternal and infant measures — infant mortality rate (deaths under 1 year per 1000 live births); neonatal (under 28 days) and post-neonatal (28 days to 1 year) rates; perinatal mortality rate (stillbirths from 28 weeks plus deaths under 7 days, per 1000 total births); stillbirth rate; under-five mortality rate; and the maternal mortality ratio (per 100,000 live births)
• Note that the perinatal rate uses total births as denominator, since stillbirths are in the numerator — a detail frequently got wrong
• The fertility measures — crude birth rate (live births per 1000 population); general fertility rate (live births per 1000 women aged 15 to 49); total fertility rate, the average number of children a woman would bear at current age-specific rates, with replacement level at about 2.1; and the net reproduction rate, which is 1.0 when a generation exactly replaces itself
• Proportional mortality is not cause-specific mortality; the denominators are total deaths and population respectively
• Rising proportional mortality may mean nothing except that fewer people are dying of something else
• Prevalence equals incidence times duration, and every consequence of that relationship should be reasoned rather than memorised
• A falling prevalence may mean cure or death, so prevalence alone can never judge a programme
• Incidence measures risk and studies aetiology, while prevalence measures burden and plans services
• Only those at risk belong in the denominator, excluding the immune and those already affected
• Person-time handles varying observation periods, which cumulative incidence cannot do
• Crude rates cannot be compared directly between populations of different age structure
• Direct standardisation needs reliable age-specific rates, and indirect is used when numbers are small
• An SMR above one hundred means worse than expected, and the measure is widely used in occupational studies
• Case fatality measures the killing power of a disease, and suits acute short illnesses rather than chronic ones
• The perinatal rate uses total births as denominator, since stillbirths appear in the numerator
• Prevalent cases over-represent long illness, which is Neyman bias and makes prevalence unsuitable for aetiology
• Replacement fertility is about two point one, and a net reproduction rate of one means a generation exactly replaces itself
• The mid-year population is the standard denominator for annual rates, approximating the average population at risk
• A rate needs numerator, denominator, time and multiplier, and omitting any of the four makes it uninterpretable
• A ratio has a numerator outside the denominator, a proportion inside it, and a rate adds time
• Total fertility rate is per woman, not per thousand, which is a common slip in written answers
• Point prevalence and period prevalence differ, the latter including those who developed disease during the interval
• Attack rate is a cumulative incidence, used in outbreaks over a short defined period
• Survival rate summarises prognosis, and is conventionally quoted at five years for malignancy
• Maternal mortality is expressed per hundred thousand, unlike infant mortality per thousand, which is easily muddled