So in this lesson we're going to be talking about deaths around the time of birth, the indicators, data sources and estimation options. And try to explain to you hy we kind of carved out a separate category when you're talking about mortality. So first we're trying to show a timeline here that moves from conception on the left to up to one year of life, and talk about deaths both after birth, but also fetal death. So the idea is you talk about between 1 week and 28 weeks they talk about, miscarriage and abortion. So basically early pregnancy loss, and then the next category is still births that comes between 28 weeks and birth. So basically children that develop well until 28 weeks of life, often they could be born and they would be premature births, but die in the last two months of pregnancy. And then you have neonatal deaths, neonatal refers to deaths that occur between birth and the first 28 days of afterbirth. They've now started trying to break that down into earlier neonatal, so the first seven days and then, and then kind of the late neonatal between, the next 21 days of the period. And then finally, you kind of broken this down into sometimes they talk about post neonatal deaths and infant deaths. But the idea is we're trying to capture mortality or end of life events that occur through this whole time period. Why split mortality among children under five into smaller age groups? The reason for that is because you have a very skewed distribution of when deaths occur. So if we look at a whole series of deaths, this was compiled you can see the reference down below from IGME, of when deaths occur. Well what you see is in the first day of life, the huge majority of all deaths in the first 28 days occurred on day one, that you get a very steep drop offs. So most of the neonatal deaths occur in day one, and it's still fairly high in the first seven days of life, but then very rapidly tapers off. And it continues to taper off not only for the first 28 days, but for the full period up till five years of age. So the idea is that there is more of a focus now when trying to identify when the deaths occur, and have our models and measurement focus on those specific things. A good example of how one of the things also is, this has changed over time. If you go to the IGME website, the childmortality.org and look at the differences between what percent of under five deaths occurred in the neonatal period between 1990 and 2015. You'll see for example in Bangladesh, it went from 40 % of all under five deaths eere in the first 28 days of life neonatal, to 60% by 2015. In Burkina Faso it's slightly different but again, neonatal has become much more an important part of under five mortality in terms of the absolute number, or the percentage of the deaths that occur during that period. Some of the common indicators I've got this whole list really I want want to focus on a few of these stillbirth rates refers to the number of fetal deaths that occur in the third trimester of pregnancy, 28 weeks to birth per 100,000 live births. Okay, so that you'll see more and more, that's kind of a common metric they talk about countries of what their stillbirth rates. Early neonatal mortality rate is the probability of dying between birth and seven days, per ,1000 live births. And then there's a third indicator that was used before and you'll see there's some advantage to it erinatal mortality rate. It's the probability of a stillbirth or an early neonatal death per 1,000 live births. We're going to discuss that later in a later slide, because there's often confusion it's hard to identify was the first born stillborn, or did they die after birth? And that characterization can often cause problems in perinatal, mortality rate is a way to combine those two things together to eliminate that. Then there's neonatal mortality rate, probability of dying between birth and the first 28 days of life post-neonatal mortality, which is almost never used but sometimes. It talks about deaths in the first year of life excluding neonatal and infant mortality rate, which is the probability of dying between birth and the end of the first year of life. Data sources for this, it's a little tricky, the common data sources are really identical to what we talked about in under five. Final registration systems again, they're not going to do a good job of this even in good systems, they're not going to capture stillbirths. Population censuses again, using indirect measures which most do you're not going to get any information specific to neonatal, and certainly nothing one stillbirths because they don't ask questions about pregnancies. Sample surveys the household surveys like DHS and MICS, even if you use a full birth history, that's good for estimating neonatal mortality because they ask more specific questions. When you say my child died, they ask if say they were not one than they ask for what month if it's in the first month they ask for what day. So it actually does give you a measure of neonatal mortality, but it gives you no information related to stillbirths. To do that, if you're actually interested in stillbirths there is a slightly modified version of the full birth history called a pregnancy history. Here women are asked the same set of questions not about all live births, but about all pregnancies. And this has been used in a few countries to allow you to actually get a measure of stillbirths, and abortions and miscarriages as well as neonatal mortality. The key is it's really hard to measure even if you have a pregnancy history. One of the things is the issue about cultural practices, some cultures don't name children until they get older and part if they die in the first day of life they're treated as a stillbirth, not a live birth. Classifications even if you have kind of a medical classification, they're difficult to identify. And we're going to discuss that on the next slide simply because, it's very complex to say if the child dies of asphyxia, they're not breathing well is that a live birth that the kid couldn't breathe or was the child actually dead at birth? And finally dating errors especially if you're using sample surveys with pregnancy history, the timing, asking the mothers, when did the child die? When did you have a miscarriage? When was there a pregnancy lost, especially for longer periods of time, becomes very complex. In this slide what we're trying to do is give you data on a study that was done that tried to look at classification of stillbirths or misclassifications of stillbirths and neonatal deaths. And so what happens is they looked and said they took data where actually all births had been classified into neonatal or stillbirths, and then said, were they correctly classified? So what they found is you can see on the right in two districts, a huge number of misclassifications most of the misclassifications occurred into neonatal deaths being shifted, into stillbirths. The idea there's often common symptoms like asphyxia, prematurity that sometimes it's called perinatal mortality again kind of dodging the question. But in one of the reasons why perinatal mortality is actually a good indicator because it doesn't require correct classification. And the idea is that babies born with limited signs of life or who die shortly after, often can be mistakenly classified as a stillbirth or intentionally reported based on cultural practices. Or they may say, well the mother will feel better that we report it was a stillbirth as opposed to the child was alive and did die. You see here in the recent analysis from Malawi compared a full birth history and verbal autopsy data and found one-fifth of the neonatal deaths had been classified as stillbirths. So suggesting that if you use a full birth history to measure neonatal mortality, you very likely are underestimating that simply because deaths that occur after birth or early after birth are often classified as stillbirths. A little bit of measurements of death that occurred, what happens is because of this difficulty the UN estimates. If you go look at IGME's estimates of neonatal mortality, are not really based on actually taking the data that they get from full birth histories. Now remember indirect methods don't get you neonatal mortality just thet full birth histories. But even there they don't use it they instead to a proportional model where they've taken lots of high quality data and look at the relationship between under five mortality, which we do measure more accurately in full birth histories. And say, given that there's a proportional relationship between under five mortality and neonatal, that changes. So as the examples I gave earlier, went under five mortality is quite high above 100 neonatal mortality may only account for 30 or 40 of all under five deaths. When under five mortality drops, let's say it gets to 50 or 35, then neonatal mortality is going to o account for close to 60% of all under five deaths. The other difficulty with this is that gestational age and birth weight and timing, are not standard indicators that people collecting household surveys. Birth weight sometime is reported a mother's memory of birth weight, but even there it's very small, average, somewhat large, large. There's really no birth weight for most children are not measured at birth or not weighed at birth. Gestational age is something that is very tricky, that unless you've had a sonogram in the first trimester of life, in fact gestational age has to be guesstimated by the mothers, reporting of when she thinks she was pregnant. It's not routinely collected and therefore it's not a good indicator for sorting out when these deaths occur. If we kind of do a summary of what's the difficulty about measuring mortality around births, is it any more complex than under five mortality? Well the clear answer is yes, it's extremely challenging to measure deaths around the time of birth. Primarily because the things we've already mentioned that often they're misclassified stillbirths and neonatal are misclassified, the time periods that mothers have to accurately recall are very short. Did the child I immediately right after birth? What was her date or when did she think she got pregnant? Kind of gestational age, all of that information makes it much more complicated. Generally what happens with most of the outcomes but stillbirths, especially into some degree, neonatal modeling is the alternative people used to measuring mortality during this time period. But as with all modeling, you've gotta be careful about the quality of the model and the methods that are used to make the new estimates.