Now we will complete an exercise evaluating the mortality data for countries in the Americas. The Pan American Health Organization, or PAHO, and the World Health Organization, or the WHO, have a quality index to assess mortality data from countries in the Americas. The score has two components. First, the ability of the registration system to capture deaths, and second, the quality of the underlying cause of death. Please open the spreadsheet that accompanies this exercise. These data are from a publicly available report titled Health Situation in the Americas Basic Indicators. The column labeled under-registered deaths contains estimates of the proportions of deaths that remain unregistered for each country. The column labeled ill-defined COD, or cause of death, contains estimates of the proportion of reported deaths that a sign, symptom, or other ill-defined cause is listed in the underlying cause of death. The quality of mortality statistics index score is estimated as shown in this equation. The score gives greater weight to the proportion of under-registered deaths as compared to the proportion of deaths due to ill-defined causes. Once you have the estimated score, you can assign the country to a data quality category as shown here. There are two technical notes to this equation. First, the proportion of ill-defined causes of death or the proportion of under-registered deaths will be missing for some countries. If either value is missing, please do not replace the value with a zero. Rather, for those countries, they should be placed in the no-data category. Second, here we are using a single year of data for a collection. However, in order to generate more stable estimates, we encourage you to use data from multiple years to more accurately reflect the performance of the vital registration system. The first question for you, why do you think the data quality index places a greater weight on under-registered deaths as compared with ill-defined underlying cause of death? Take a minute to work out your answer and click continue when you are ready. The greater weighting of under-registered deaths is based on the idea that under-counting the number of deaths is going to have a greater impact on the mortality rate than having an ill-defined cause of death for a death that was counted. Remembering that a mortality rate can most simply be calculated as the number of deaths in the population divided by the number of persons in the population, underestimating the number of deaths will underestimate the mortality rate overall. The greater weight on under-registered deaths means that we are more concerned about the impact of under-registered deaths on the mortality rate than we are about the impact of an ill-defined cause of death on other morbidity and mortality statistics. Question 2, calculate the score for all the countries. Which countries received a score of "good quality"? Which countries have the 3 lowest quality of data scores? I'll give you a hint to use the sort function in the spreadsheet so that you can easily locate the countries with a good score and those with the lowest scores. Take a minute to work out your answer, and click continue when you are ready. Question 2, here are the 18 countries with a good quality rating. The three countries with the lowest quality of data include Nicaragua, with a score of 25.2%, the Turks and Caicos Islands, with a score of 26.5%, and Paraguay with a score of 27.0%. Question 3, consider the quality scores of Argentina and the Bahamas. Both countries have good quality mortality data, however, they have different proportions of ill-defined COD and under-registered deaths. How is your interpretation of the mortality data from these two countries different? Take a minute to work out your answer, and click continue when you are ready. For question 3, Argentina and the Bahamas both have good quality data scores below 10%. There is a greater proportion of under-registered deaths in the Bahamas, compared with Argentina. The mortality rates will be underestimated to a larger magnitude than in Argentina. Conversely, there's a greater proportion of ill-defined causes of death in Argentina compared with the Bahamas. Cause-specific mortality rates will be underestimated to a larger magnitude among the 95.3% of registered deaths in Argentina. Although both of these countries have good quality data, it will be important to remember that the mortality rates in the Bahamas will be a bit more underestimated than in Argentina. And conversely, among the reported deaths, the cause-specific mortality rates will be underestimated a bit more in Argentina as compared with the Bahamas. Question 4, think of at least one reason why cause of death could be under-reported or misreported on a death certificate for HIV, homicide, and cancer deaths. Take a minute to work out your answer, and click continue when you are ready. For question 4, when thinking about HIV, under-reporting can occur when there's a lack of testing resources. People may die from HIV and never know they were infected. Misreporting can also occur when a cause of death other than HIV is given on the death certificate to protect the deceased or their family members from HIV stigma. For homicide, there is evidence from Bangladesh that some of the young women brought to the hospital unconscious with a report of self-inflicted poisoning following disputes with spouses or in-laws are not suicide attempts, but rather homicides, meaning they were poisoned by someone. In this situation, homicide would be misreported. And for cancer, similar to HIV, in settings where testing and treatment are not universal, an individual could die from cancer without knowing their diagnosis, resulting in an under-reporting of cancer deaths. The three final questions in this exercise will be based on the scenario described here. Imagine that there's a new initiative to reduce the infant mortality rate in the countries where it is estimated to be the highest: Belize, with an infant mortality rate of 17.9 per 1,000 live births, El Salvador, with a rate of 21.5 per 1,000 live births, and Guatemala, with a rate of 30 per 1,000 live births. Your role will be to evaluate the success of the new initiative to reduce the infant mortality rate. Question 5, will data supplied by the vital registration systems from Belize, El Salvador, and Guatemala accurately count the number of infant deaths, specifically, that numerator of the infant mortality rate? Take a minute to work out your answer. Click continue when you are ready. Going back to our spreadsheet, we can identify the proportion of under-registered deaths, the quality score, and the data quality category for each of these countries. Notice that none of these countries are in the good quality data category. Also notice that the proportion of under-registered deaths is quite high in all three of these countries, which means if we rely on the vital registration system's data above, we will likely underestimate the infant mortality rate. In doing so, we may inadvertently suggest that infant mortality interventions were more successful in reducing the infant mortality rate than what was truly happening. Vital registration systems are often poorest in countries with the greatest public health problems. This results in a familiar situation for public health epidemiologists. Data that we want the most are the data that we are least likely to have. Question 6, what other sources of data might you use to evaluate whether or not the infant mortality initiative reduced the infant mortality rate? Take a minute to work out your answer, and click continue when you are ready. As we discussed, vital registration systems come in many shapes and sizes across the spectrum. Even in countries with full registration systems, they may also have verbal autopsies in more rural areas of the country, or demographic surveillance systems. For the particular scenario that we are discussing, the vital registration systems in Belize, El Salvador, and Guatemala, were poor and medium quality. I would suggest looking at the verbal autopsy studies or sample registration systems as additional sources of data that could be used to evaluate whether or not the infant mortality reduction initiative reduced the infant mortality rate. Question 7, how will improvements and changes in the vital registration system impact your assessment of the infant mortality rate over time? Take a minute to work out your answer, and click continue when you are ready. Changes in the vital registration system may result in changes to the estimate of infant mortality over time. These changes may impact the infant mortality rate resulting in fluctuations in the infant mortality rate that may not be real, but rather an artifact due to the changes in the vital registration system. For example, as you can see in the figure, if immediately following the infant mortality reduction initiative there was an initiative to improve the vital registration system, this would cause a problem. If the improvement in the vital registration system resulted in a reduction of the under-registered deaths, the infant mortality rate may increase because now the system is doing a better job of capturing the occurrence of infant death. If the changes to the vital registration system occurred at the same time as the initiative to reduce infant mortality, you may see an increase in the infant mortality rate. And it would be inappropriate to suggest that the infant mortality reduction initiative had anything to do with this change in infant mortality rate. Great job on completing this exercise.
