At this point, perhaps you're confident in the role of routine health information systems in providing vital statistics of births and deaths in a population. But maybe you're wondering if there are other data collection systems that are developed specifically to address public health problems. Such systems are often classified as public health information systems or PHIS. Like routine health information systems, public health information systems also attempt to produce timely and quality information. Three traditional sources of PHIS include: population based surveys, registries of specific health outcomes, and surveillance systems. Public health information systems provide data for exposures, risk factors, and health outcomes beyond mortality. These data can be used to calculate other important health statistics. Because these systems are established to specifically collect public health data, and oftentimes have a specific public health purpose. These data are often richer than data from routine health information systems. Let's start with some examples of population based surveys. Population based means that there is a sampling strategy put into place in order to attempt to select a specific population to complete the survey. Oftentimes, the goal is to select a population that is representative of the entire population of interest. For example, we will talk about the Baltimore City Community Health Survey. There are not enough resources to seek every Baltimore citizen's response to these survey questions. So, the sampling strategy attempts to select a population that is representative of the citizens of Baltimore. The Baltimore City Community Health Survey is an example of a local population based survey. The National Health and Nutrition Examination Survey (NHANES) is one of the richest sources of public health information that can be generalized to the population of the United States. At the international level, the World Health Survey, and the Demographic and Health Surveys (DHS) are surveys used in countries all over the world. An important feature of these population based surveys is that they measure self-reported health behaviors. Such information, is typically not available in routine health information systems. Some of these population based surveys even collect specimens for testing. These are examples of what I mean when I say the data are collected for public health purposes, and that the data are richer than the data from routine health information systems. Let's start with the local example, the Baltimore City Community Health Survey. This is a cross-sectional population-based survey that was administered in 2009 and in 2014. The 2014 survey objectives included; assessing the health status and needs of the residents, identifying gaps in access to health services, assessing the use and perception of the Affordable Care Act and the Maryland Health Exchange, and assessing the use and perception of city health services. The Affordable Care Act was a dramatic federal level change in the U.S, that required everyone to have health insurance. The Maryland Health Exchange was the resource provided by Maryland to its residents to help them purchase health insurance. So, the purpose of this survey was not only to solve public health problems but also, evaluate changes in policies and services. This was a cross-sectional study. Meaning, they did not follow up with individuals that completed the survey. The survey was conducted by a telephone. The sampling design was a random digit dial of both landlines and cell phones. A random digit dialing strategy is often employed when attempting to generate a sample that is representative of the population of interest. The Baltimore City Community Health Survey was restricted to adults, and they did not over-sample any specific population. 1,722 Baltimore City residents participated. Sampling weights were then estimated and applied. Sampling weights correct for things like unequal sampling, and the bias that may enter this study from people not wanting to participate in the study. Sampling weights are an analytic methodology. The weights reflect their age, sex and race/ ethnicity characteristics of the city. So that the statistics estimated using data from the survey better reflect the entire Baltimore City population, not just those who participated in the survey. NHANES is a population based survey used to assess the health and nutritional status of adults and children in the US. It started in the early 1970s, and is considered a gold standard for population based surveys in the US. This survey is also cross-sectional. There have been many cross-sectional cycles of NHANES since its inception in 1970s. Beyond self-reported responses obtained via interview, a clinic examination is conducted and laboratory tests are run on collected specimens. Five thousand people from 15 selected counties participate. The sampling design for NHANES starts with the 15 participating counties. Within each county, clusters of households are selected. Every person in a household is screened for demographic characteristics. One or more persons are then selected from each household. Just like the Baltimore City Community Health Study, sampling weights are used to correct for unequal probabilities of selection and non-response adjustments, so that the statistics generated from NHANES can better reflect the population of the United States. NHANES data are publicly available. NHANES encourages investigators of their data to use two, two-year cycles or four years worth of data in any analysis to improve the stability of the estimates. The World Health Organization (WHO) invites all of its member states to participate in the World Health Survey. The goal of the World Health Survey is to strengthen the national capacity to monitor critical health outcomes and health systems through the fielding of a valid, reliable, and comparable household survey instrument. The first round of the World Health survey began in 2002. To date, 70 countries have participated resulting in data from a total of 300,000 participants. The data for the World Health Survey are collected by a face-to-face interviews or computer assisted telephone interviews based on a country's capacity. The WHO has clear instructions for sampling stating that a minimum sample is 5000 individuals. The sampling frame must start with 100% of the eligible population, and it cannot be restricted by ethnicity, geography, etc.. This can be a challenge for many countries with rugged terrain and populations that live cut off from the rest of the country. The WHO specifies instructions for both a single-stage random sample but also a multi-stage clustering sample if resources or capacity do not allow for this single stage random sample. Instructions for constructing the sampling weights are included so that results can better reflect the entire eligible population. For more information on the World Health Survey, we encourage you to go to the WHO website. The final population-based survey we will discuss are the Demographic and Health Surveys (DHS). The DHS take about 18 to 20 months to complete and are broken into four stages. Survey preparation, and questionnaire design which includes the sampling design. Training and fieldwork, data processing and the final report which is the primary mode of dissemination of the results. The DHS has a fantastic website and works very hard to make sure that their surveys can be used in all countries, even countries with little resources or other challenges. The DHS posts on online the data sets that are collected along with the tools for data collection. One example of a completed DHS report in a country in need of the public health information generated from this survey is Cambodia. Their final report included too many statistics to cover here. But examples of statistics that were generated from the DHS data are listed here. The total fertility rate is often calculated. In Cambodia it was 2.7 children per woman. 56% of married women are using contraception. The infant mortality rate is quite high at 28 per 1000 live births. The data from the DHS include an estimate of the proportion with registered births in the vital registration system, and 73 percent of children aged one to two years are fully immunized. Please go online to see the complete report from Cambodia as you may be as fascinated as I am with the breath of public health statistics that are presented in the report. The basic Demographic and Health Survey is the bread and butter of the DHS administration. But, the DHS has also created surveys to collect more information about AIDS and malaria, service provision assessments of health care facilities in organizations, and they also have key health indicators surveys. For a complete listing of surveys, please go online to learn more. Population-based surveys are a rich source of public health data, but there are limitations. Notably, almost all of these types of surveys are cross-sectional, meaning that individuals are not followed over time. Following individuals over time allow for investigations into more specific relationships that may impact public health problems. Measurement issues include cultural sensitivities, recall bias and social desirability bias. There can be missing data in these surveys. And finally, it's nearly impossible for confirmation of the self-reported health outcomes. Most of the design challenges to population-based surveys reside in the sampling methodology that can quickly get complicated. Biases result when groups are left out. Like people who do not have landlines, or those who are institutionalized, those who are ill, or if groups are excluded based on specific characteristics such as race or ethnicity. Finally, weights must be appropriately calculated and applied to the results so that they can better reflect what is going on in the population of interest. Let's take a quick break, and we'll move on to our second example of public health information systems.