Classifying diseases and measuring their patterns

1Sixty-six cases, and one adult in six

On 27 September 2024 Rwanda confirmed its first outbreak of Marburg virus disease, a severe haemorrhagic fever. Within three months the country recorded 66 confirmed cases and 15 deaths, and the outbreak was declared over on 20 December 2024. It made headlines around the world. In the same years a far quieter condition affected far more people: in Rwanda's 2021–22 national survey of adults aged 18 to 69, about one person in six had raised blood pressure, and more than three-quarters of them did not know it.

Which is the bigger problem for public health? Counting cases is not enough to answer. You need to classify the diseases and measure them in ways that can be compared: how many new cases, how many people living with the disease, how many deaths, and where, when and in whom they occur.

2Classifying diseases

The first split is by cause: is a pathogen involved? Non-infectious diseases are then grouped by their main cause.

The categories overlap. Cervical cancer is a cancer, but most cases follow infection with human papillomavirus (HPV), which is why Rwanda vaccinates girls against HPV. Hypertension has inherited, lifestyle and age-related causes at once. Classify by the main cause, and say when more than one applies.

Worldwide, non-communicable diseases are now the biggest killers: WHO estimates that they killed at least 43 million people in 2021, 75 % of deaths not caused by the COVID-19 pandemic, and that 73 % of NCD deaths occur in low- and middle-income countries.

3Measuring disease

All the measures below are rates: a count divided by the size of the group it came from. Only rates can be compared between populations of different sizes.

$$\text{incidence rate}=\frac{\text{number of new cases during the period}}{\text{population at risk during the period}}\times 1000\backslash text\{incidence\; rate\}\; =\; \backslash frac\{\backslash text\{number\; of\; new\; cases\; during\; the\; period\}\}\{\backslash text\{population\; at\; risk\; during\; the\; period\}\}\; \backslash times\; 1000$$ $$\text{prevalence}=\frac{\text{number of existing cases at one time}}{\text{total population at that time}}\times 100\mathrm{\%}\backslash text\{prevalence\}\; =\; \backslash frac\{\backslash text\{number\; of\; existing\; cases\; at\; one\; time\}\}\{\backslash text\{total\; population\; at\; that\; time\}\}\; \backslash times\; 100\backslash ,\backslash \%$$ $$\text{case fatality rate}=\frac{\text{deaths from the disease}}{\text{number of cases of the disease}}\times 100\mathrm{\%}\backslash text\{case\; fatality\; rate\}\; =\; \backslash frac\{\backslash text\{deaths\; from\; the\; disease\}\}\{\backslash text\{number\; of\; cases\; of\; the\; disease\}\}\; \backslash times\; 100\backslash ,\backslash \%$$ $$\text{mortality rate}=\frac{\text{deaths from the disease during the period}}{\text{total population}}\times 100000\backslash text\{mortality\; rate\}\; =\; \backslash frac\{\backslash text\{deaths\; from\; the\; disease\; during\; the\; period\}\}\{\backslash text\{total\; population\}\}\; \backslash times\; 100\backslash ,000$$

The multiplier (1000, 100 000 or 100 %) is chosen to give a convenient number; always state it, and state the period for incidence and mortality ("per 1000 per year").

4Why prevalence depends on how long a disease lasts

Think of prevalence as water in a tank. New cases (incidence) flow in; recovery and death drain it out. When the flow in and out are steady and the disease is not too common:

$$P\approx I\times DP\; \backslash approx\; I\; \backslash times\; D$$

where $PP$ is prevalence, $II$ the incidence rate and $DD$ the average duration of the disease. So a short, acute disease such as cholera can have a high incidence during an epidemic yet a low prevalence at any moment, because each case lasts only days. A chronic disease such as hypertension or HIV infection builds up a high prevalence even from a modest incidence, because people live with it for decades. Good treatment that keeps people alive longer, such as antiretroviral therapy for HIV, raises prevalence while it lowers mortality.

5Patterns: person, place and time

Epidemiologists describe every disease by asking who gets it, where and when. Figure 2 shows two real Rwandan patterns.

• Place. In the 2019–20 Rwanda Demographic and Health Survey, about 1 % of children aged 6 to 59 months carried malaria parasites, but the rate was 1.5 % in the West and 0.3 % in the North. Malaria also followed wealth: 1.9 % of children in the poorest fifth of households, 0.2 % in the richest.
• Person. In the 2021–22 STEPS survey, raised blood pressure rose steadily with age, from 8.5 % of adults aged 18 to 29 to 43.2 % of those aged 60 to 69 (16.8 % of all adults aged 18 to 69).
• Time. Malaria transmission in Rwanda peaks in April and May, after the long rains. The survey report warns that the South and East were surveyed in June and July, after the peak, so their measured prevalence is lower than it would be at the peak. A prevalence figure is a snapshot: always ask when it was taken.

Over the course of a single outbreak, time is shown by an epidemic curve: a bar chart of new cases by day or week of onset. Its shape reveals how the disease is spreading.

A sharp single peak (A) points to a common source, such as one contaminated well, and investigators look for what all the cases shared. Successive, growing waves (B) point to person-to-person spread, and the response is to find and isolate cases and vaccinate or protect their contacts. In the Marburg outbreak, almost 80 % of cases were health workers infected while caring for patients and colleagues, so protecting health workers and isolating patients broke the chain.