Today, we're going to be talking about Under-5 Mortality as a measure of impact and evaluation. As part of that we're going to do a quick overview of indicators, data sources and estimation modules that then we'll go in-depth later on with each lessons on each of those. Okay, so let's start with Under-5 Mortality and the indicators. The kind of classic indicator is UNDER-5 Mortality U5MR which is gives you the probability of dying between the age of birth and exactly five years of age per 1000 live births. So if you look back in 80 or 100 years ago, under-five mortality in many countries was 3 to 400. So basically means 30 to 40% of children died between birth and the age of five. Nowadays, fortunately for most countries under-five mortality is under 100. If you're a demographer it's express it as 5 to 0. The idea of its a five year appear it's starting at birth. Another way to think about under-five mortality is child mortality, which is probability of dying between the ages of 1 and 5. Again, expressed per 1000 live births. Next there's infant mortality, which is 1q0, which is basically the probability of dying between birth and 1 year of age. So it's the first year of life. What's the probability of dying? The other indicator that we're not, it's not on the slide and we're not going to talk about until the next lesson is neonatal mortality, which talks about the probability of dying between birth and 28 days. The first 28 days of life. This slide shows the world map where countries are color coded according to their current under-five mortality rates. As you can see, most of the kind of countries with high under-five mortality are in Africa or in Southern Asia. With the highest countries, there are still a few countries with mortality rates above 100, meaning, roughly 1 out of 10 children die before the age of 5. But most of the countries now have partly in relation to MDGs, and the effort to change under-five mortality have dropped from an average of 150 in sub Saharan Africa 20 years ago to now, roughly about 60. So many of the countries have much lower values. You're noticing in South America, there's only a few countries that still have relatively high rates of under-five mortality. Most of the developed world or the high-income countries you see have mortality rates below 25. Many countries have under-five mortality rates of less than 10. In fact, the SDG targets are trying to reduce under-five mortality to below 20 by the year 2030 and all countries in the world. Now, what I want to talk about and what the rest of this module is going to focus on is data sources for under-five mortality. So if you're running an evaluation, where can you get information? So pre-existing information or information you're going to use for your evaluation, both pre and post, related to under-five mortality. So the first thing I want to talk about a little bit is vital registration systems. The idea of vital registrations systems are sometimes referred to as CRVS. Is simply the government system where they record births and deaths and along with deck causes of death. The key is this is kind of the gold standard. It's what most high income countries would use and it's what all countries want to push towards. Unfortunately for most low and even middle income countries, vital registration systems are poor. In some countries there basically absent that they have almost no data. Generally what they do is they don't provide enough, they're not accurate enough or complete enough in their registration so that even if they do have numbers of reported deaths, it's not sufficient to give you an estimate of the rate of mortality. They may say, well, we had 10,000 under-five deaths reported in the last year, but that may be only half or 30 or 40 or 60% of the total deaths that occur. The idea is a few of the countries where you would actually be running an evaluation on one under-five mortality could use a vital registration system to actually measure impact. The second big places that nearly every country has its population censuses. Population censuses in most countries are done every 10 years. Their goal is not simply to measure mortality, that they have lots of questions that focus more on demographic issues, but most censuses do have a measure of mortality. The key is if it exists, you certainly can't run a population census and evaluation because they're large and heavy to do. And that when they're done by national governments, that only if you wanted to build on a census, you say I'm running an evaluation of the country and the current census is coming up. Really, most senses cannot accommodate adding additional questions. So the idea of you being able to leverage a census is going to occur roughly at the end point of your evaluation. That generally they will not allow you to add things simply because they have lots of demographic and economic information they try to collect. They do in general population since these use an indirect method to measure mortality. The key is that it's going to provide an estimate of mortality five years in the past, not at the current time when they were in the census. Therefore, it makes sense. This is really not an effective way to gather information for an evaluation, though they do provide information of national mortality in the past. What we're really going to focus on more in this class, you're going to see is sample household surveys where a population census does household surveys of all households or abodes within a country. Sample household surveys simply try to get a nationally representative or sometimes it's state or provincial level representation of households. That can be used to gather information just like the census on mortality, coverage of interventions, risk factors, other things related to public health. Usually, what happens in sample surveys? They can use two different approaches and we'll be discussing those in the next lesson. Ways to measure mortality. One, the one like the population census that use an indirect measure. That gives a good reliable measure, but further in the past where they can use a full birth history that allows you to measure mortality in the more recent past or within the last year. A key with the sample household surveys is that how reliable your estimate is and for what time period it can be gained, depends on sample size. And so you'll see that often. That's what drives do you have the budget to allow you to run a sample household survey that's large enough to let you measure accurately mortality at the beginning and the end of your project for evaluation? Also, other ways that people get information about current and past measures of mortality. A sample registration system is similar to vital registration except with the idea, instead of trying to record all deaths and birth within a country, they simply sample and set up intensive efforts within clusters or areas within the country that they sample those base to make it nationally representative. Currently there's two systems in lower middle income countries, India and china that used this approach. In India, there's India and China, both. There's 1000s of different locations where they do in-depth work to make an accurate representation of births and deaths. And then they can use that just like with the sample survey to estimate both at the national and at the state level. They can be very good, but unfortunately few countries have this. There are now efforts to develop these systems within two countries within sub Saharan Africa with discussions of expanding this further. The next approach that people often use or have data on data is data on mortality is demographic surveillance systems. Demographic surveillance systems are again kind of a registration system that they do very detailed household visits to follow up births and deaths and causes the death. And it provides gold standard data on mortality. The problem for demographic surveillance sites is they're not built to be representative of anything within a country. A country may have one or two surveillance sites and it gives very good information about their catchment area, but nothing else. So, a good example is that if you've heard of Math Lab in Bangladesh again geographic region that for years they've been doing very good DSS. They have good trends in mortality, but it's not representative of the other areas around it. One of the reasons they become less representative is because this data is, this collection is ongoing. DSS sites are often chosen for places to run randomized controlled trials or to roll out new interventions that they can evaluate the impact. Because mortality both at pre-intervention and post-intervention are being measured as part of the routine system. A final source of data that we'll discuss a little bit is of under-five mortality can be drawn from hospital or clinic records. The key about this is that most hospitals or clinics do track deaths and they track deaths by age. So it would give you a trend over time and under-five mortality. However, the problem is it's not complete. It doesn't capture a representative sample. For example, if you're looking at a hospital and changes, the deaths change over time, you don't know who came, who died outside the hospital, who died within the hospital or sought cure in another place. Still, it's often used in evaluations if you're looking at gross changes. So if you're expecting a huge change, you might look at trends within a clinic or a hospital. It's kind of an extra piece of information that could never be thought of as a primary measure of impact unless you have a complete birth or registration system. So in the rest of this module, what we're going to do is we're going to focus on the most likely ways you collect data for an evaluation on under-five mortality. We're going to talk specifically a lot about direct measurements that are used in surveys, primarily household surveys, MICs and DHS and how those are done. We're going to talk about indirect measures that are used in survey that are simpler, easier, but also produce good estimates of under-five mortality. We're going to talk about modeling from multiple measurements. This is a little bit based on what. What we're going to discuss is IGME, the Interagency Group of Mortality Estimation, their approach to combining information to get more reliable estimate mortality and in trends in mortality. And finally, we'll also discuss the lives saved tool, which it and there's it in similar models that allow you to estimate mortality change based on changes in intervention coverage.
