# IMCI service percentage

This question was posted the Assessment and Surveillance forum area and has 6 replies. You can also reply via email – be sure to leave the subject unchanged.

### Tammam Ali

H&N Project Manager/Relief International

Normal user

22 Aug 2015, 10:49

### Mark Myatt

Frequent user

25 Aug 2015, 09:26

Caseload estimates can be made from program data or surveys (menage SMART type surveys estimate period prevalence of common childhood diseases). This will give prevalence but you want incidence. You can convert (quite roughly) between prevalence and incidence. You can get incidence from surveillance systems but these can be tricky to set up and keep running.

Coverage can be estimated from surveys (even KAP surveys).

### Mark Myatt

Frequent user

25 Aug 2015, 11:25

I suppose you are interested in

**direct**beneficiaries (i.e. children with the disease that come into contact with the program and receive effective treatment).

I think the first thing you will need is to make some estimate of need. Here I take the example of a disease X (it could be diarrhoea). You want to estimate how many cases of X that you expect in your population in a given period (let us assume this is one year). This is an incidence figure. You might get to this using a small sample survey by (e.g.) asking mothers how often in the previous three months did their child have an episode of X (you'll probably need to ask about dead children too as many IMCI disease are killers). You could add all these together. That would give you the number of episodes in your sample in the previous three months. If we ignore seasonality then we can estimate incidence in a population over a whole year as:

cases = (sum(episodes) * 4) / sample size * 12600

The "4" is to adjust 3 months to 1 year.

If (e.g.) your survey sample size was 100 and you found that there were 15 episodes of X in three months then:

cases = (sum(episodes) * 4) / sample size * 12600

cases = (15 * 4) / 100 * 12600

cases = 7560

This is a rough-and-ready method.

You will need to take care about the length of the recall period (three months is probably too long) and you'd probably want to correct for seasonality using disease calendars (i.e. what diseases occur at what times of year). It can get pretty complicated. It will always be very approximate.

You can seek more advice on this forum as you need it. You may be able to use existing data to make your needs estimate. I am sure there are quite a few people here who have done this for CMAM programming using HDS, MICS, or SMART data and will be able to advise you. There will probably be IMCI planning tools that will guide you through this process. I know (e.g.) that UNICEF have guides and tools for CMAM planning.

Moving on with the cases = 7560 example (above) ... that is your estimate of need.

What will be your contact coverage? That is, how many of these cases will access care? There are methods of measuring this but, in a new program, you will have to make a guess (or ask about this in teh small sample survey). Let us assume that contact coverage is 50%. Your number of beneficiaries will then be:

contact cases = 7560 * 50% = 3780

If you have 90% adherence to treatment and 90% effective treatment then:

effectively treated cases = 7560 * 50% * 90% * 90% = 3062.

This is perhaps not what you want as this is episodes not children. We can get to children using an adjustment calculated from your survey data.

You can do this for all of your common IMCI disease and sum the numbers of children for each disease. This will give you a number of children treated which you will have to adjust down because sometimes you treat a child for more than one disease.

BTW : Notice how the main determinants of your program's effectiveness is coverage. Mobile clinics should help you achieve good coverage.

I hope it this is of some use.