In this presentation I'll be covering considerations for surveillance of chronic disease or non communicable disease. The learning objectives for this presentation are in the context of chronic disease surveillance describe the justification for surveillance list surveillance targets describe uses of surveillance describe key surveillance design components including key data sources. Be able to look up I.C.T. nine and I.C.T. ten codes in their descriptions using an online site access online peer F.S.S. data describe analysis strategies and list of dissemination strategies for surveillance data. So why do we even want to conduct chronic disease surveillance well due to the epidemiologic shift chronic disease is the main cause of mortality in developed countries. And because we have a longer life expectancy that leads to greater incidence of chronic disease so there's a greater burden of chronic disease especially in developed countries. And with the greater burden and the longer life expectancy then we have chronic disease as a major source of disease cost. So in order to address these issues the greater burden and greater cost of chronic disease public health programs require accurate and timely surveillance data so that they can target those interventions and have an impact on these through these areas. Chronic disease surveillance is used in similar ways to that for communicable disease surveillance but there are some unique features. Of course we always use our surveillance data to estimate the magnitude of a particular health problem the prevalence of a condition or. Or the incident of a condition. We also will use it to identify populations at highest risk what populations experience the greatest burden of the health problem. When we compare measures of surveillance measures from multiple chronic diseases we may decide how to prioritize our public health programs and interventions. And then use that same data or additional data to evaluate public health programs and interventions. Especially in the case of chronic disease we can use our surveillance data to detect changes in health care practice whether or not we see improvements or differences in how persons with a chronic disease are being treated whether or not they're following guidelines of care that have been established by public professional health organizations. And lastly we might use the surveillance data to determine allocation of resources. This goes again with the prior to zation of the public health programs and interventions we may evaluate these chronic disease or valence measures across multiple conditions and then decide how to allocate our resources. While some of the same tenets exist for chronic disease surveillance as exists in infectious disease we have some unique needs in a chronic disease surveillance system. It means a little something different when we talk about comparable data or standardized methods in a chronic disease surveillance system than infectious in chronic disease we don't have the backbone of reportable conditions that we can rely on. So when we talk about having comparable data that means we need similar data sets between jurisdictions that we're all drawing from similar. Types of data that we have case definitions case definitions don't always exist especially when we're using certain types of data like claims data or having a clear. Question on a survey that would indicate that a person has a certain condition so developing a case definition is sometimes required and that needs to be standardized across jurisdictions and lastly and have the metrics that we're talking about related to that condition could be incidence and prevalence or it could also be features of the condition like their symptom frequency or or how often they see their house their healthcare provider so we need comparable data between jurisdictions but usually that requires more work in a chronic disease surveillance system than an infectious disease because many of those systems and standardized methods have not been established. For chronic disease we often require multifaceted data leveraging many data sources now in infectious disease we use multiple data sources as well but in chronic disease they they can actually come from five six seven different data sets the data sources to pull together the entirety of information we might want to know about a chronic disease because we're not just interested in the prevalence or incidence of the diseases self but we also want to know how well it's managed and it's mediators we also want to know it's outcomes how frequent are hospitalizations or mortality rate related to the condition and also we want to track the causal factors of the disease and you can you can start to think about how this would require multiple datasets and leveraging many many partners and collaborators across the jurisdiction to try and pull these things together. This diagram depicts a conceptual model for a chronic disease surveillance system it. Kind of alliance with the data sources and types that you may want to have in place in order to achieve a comprehensive surveillance system. We'll go through it piece by piece. First the causal factors these are those identified risk factors that have been demonstrated to be causally associated with the chronic disease perhaps tobacco smoking for long cancer or certain chemicals for work related asthma. By monitoring these causal factors we may be able to predict future prevalent incidence of the condition they also might service as a target for remediation efforts or public health programs. Next we have incidence and prevalence of the condition of interest these are our simple metrics of risks and prevalence and rates of new cases or existing cases of a condition these are your core metrics for any surveillance system. Next in these green boxes are the mediating factors or factors related to the chronic disease triggers management symptoms quality of life and access by triggers I mean those external factors that may make the chronic disease worse when you think about asthma triggers might be the environmental pollutants or the pet dander that give rise to asthma symptoms it could also be triggers for a diabetic event or an inflammation so think about data that may offer sort of depth into what are the triggers of more severe disease if we can track those in our surveillance system that would be adding today to our understanding of the condition for the population. Chronic disease management is a an area of much focus these days in our health care system and metrics that allow. As to monitor management whether that be clinical management from the part of the health care system or self management how a person might manages their own condition we can and we can sort of think of metrics that might allow us to measure how well its chronic disease is being managed over time and might also give us some ideas for targeting interventions if we find that clinical management is is falling short of what the guidelines would suggest we may do an educational intervention to physicians or allied health staff so that we can improve the clinical management for a population. Next we have symptoms those are the reactions to the triggers this may be symptoms about blood sugar for diabetes or symptoms about pain and other affiliated conditions with cancer. It could also be symptoms related to respiratory conditions when we're talking C.-O. P.D. It could be shortness of breath or having a C O P T episode. Next we have quality of life there are many existing quality of life metrics that exist usually survey based so for able to collect information on the quality of life that a person with a chronic disease is experiencing that may be a feature for the surveillance system that that's quite advantageous it can help tailor our public health interventions and hopefully improve the quality of life that they are experiencing. Access here generally means access to care but it may also be access to points of service such as pharmacies so access as there is do they have health insurance are they close enough proximity to a hospital or to a primary care provider or to specialists improving access to care will have an impact on the clinical management of their chronic disease and potentially trickle down to symptoms and Chua. Day of Life so by improving access we improve other things on this diagram so it also is a good thing to measure for members for for people with the chronic disease of interest. So all of these things are are part of the disease process and on and now will focus on these areas up above where we start with pharmacy claims claims data provides a very rich resource for surveillance purposes pharmacy claims can tell us how frequently a person with a chronic disease is filling their appropriate meds medications and how often they're filling them whether or not they're potentially adherent to a regimen that's prescribe by a physician so pharmacy claims can be used to measure self management of a condition as well as clinical magic management in determining whether or not they're being prescribed the appropriate regimen. Then we can look at scheduled and unscheduled office visits you this is also using claims data for the most part although you could collect information about office visits from an individual by self report oftentimes it's not possible to distinguish between scheduled an unscheduled office visits especially from claims data however if you're talking to an individual and collecting information from them you can make that distinction. I'll go back actually I'm going to go back to scheduled an unscheduled office visits and why we might want to know about that guidelines clinical treatment guidelines would suggest that certain you know there's a certain frequency at which a person with a chronic to disease should should talk to a specialist or to a primary care provider so we can measure scheduled an unscheduled office visits as a measure of clinical management and self management of a condition. OK then E.G. an urgent care and A here stands for emergency department this is more of the more so. Fear out one of the more severe outcomes related to a chronic condition so we measure this is an outcome that we hope to reduce over time sometimes it's more appropriate for certain conditions than others emergency department visits are infrequent for people with cancer however they are more frequent for people with respiratory conditions. So we can measure the frequency at which an emergency department visit occurs as a severe outcome. Next we have hospitalization again another severe outcome related to a chronic disease. And lastly mortality. Mortality related to the chronic disease and this would be mortality rates now these are ordered these arrows kind of a grayish tone arrows are ordered from left to right from it with increasing number of events. So the the pharmacy claims have the greatest number of data points compared to mortality mortality is going to be more rare than pharmacy claims. And in increasing severity so the mortality obviously is the most severe outcome. Followed by hospitalization just just less severe so order from left to right in decreasing frequency but increasing severity. And that is depicted by the the size of the arrows. And lastly all along this continuum between causal factors as well as these outcomes in the healthcare system we have cost and cost could be measured to the individual the individual with the chronic disease you can also measure cost affiliated with their extended family their caretakers their support system you can also measure cost to society costs to the health care system costs to the public health sector cost to the community so cost falls all along. This continuum so you can imagine if you have measures related to all of these factors for a particular chronic disease you would have a pretty darn comprehensive surveillance system for that chronic disease you'd be able to measure everything related to it and hopefully pinpoint specific targets for public health intervention. When developing a surveillance system for chronic disease just like with infectious disease a clear case definition is essential but the style of case definition is somewhat different from that that we would use for an infectious disease surveillance system. We might use a self report with a positive answer to a survey question as a definition of having a condition. We might use of mission of a new case to a registry that meet clear criteria for inclusion as evidence of a new case or we might use a series of administrative claims data that together are considered evidence to suggest a case of the house problem. Case definitions in chronic disease may be slightly different than we would expect to see in an infectious disease situation because infectious disease rely on lab results and not all chronic disease. Can be mapped back to a particular lab test result some can but not all. When developing a surveillance system for chronic conditions you want to in value it evaluate and incorporate what the existing available data sets might be and in the event that those available data sets don't give you all the types of data that you want to have resources to develop you might collect some data of your own. This means that you'll need to evaluate the existing data nationally or from other jurisdictions to. Leverage standard methods whether it be specific survey questions or types of administrative data. Here's just an example of a couple of survey response criteria for defining an asthma case these questions were drawn from the Behavioral Risk Factor Surveillance survey So these are case definition criteria that come from a survey. The first is that for lifetime asthma a person is considered to have had lifetime asthma if they respond yes to the question has a doctor nurse or other health professional ever told you that you have asthma if they answer yes to that question well then they are considered a person or respondent on the survey that has asthma we don't try to back that up with a call to their physician or a look at their clinical record we take their report their self report as evidence that they have the condition. That they are fast also asks this question Do you still have asthma so among the population of respondents if they answer yes to both of these questions they're considered to currently have asthma. Using data from these two survey questions we generate a lifetime asthma prevalence and current asthma prevalence there are similar questions for diabetes or having had a heart attack or stroke or even some mental illnesses on the behavioral respect or surveillance survey to determine prevalence of those conditions and they've they're very similar in that they're rooted in having a doctor or nurse or other health professional diagnosing the condition. Here's a case definition for asthma based on administrative data. We saw this before in the presentation on case definitions it's the definition of persistent as based on the health care effectiveness data and information set definition that's published from the National Committee for Quality assure. That's and C. Q. and A It uses administrative claims data to determine whether or not a person has asthma. If a person has at least four dispensing event of an asthma medication or at least one emergency department visit where the primary diagnosis was asthma or at least one hospitalization where the primary diagnosis was asthma or at least for office visits where there was an asthma diagnosis plus at least two dispensing events of an asthma medication that the person is considered to have asthma. You can see the difference between using these administrative claims data to define a person as having a condition first says the survey. A survey asks a question of the person and they respond this asks the question of basically their claim that gets submitted to their health insurance company. Now will go through just some of the major sources of data that are used in chronic disease surveillance. Here's a list of the essential oils the key data sources used in chronic disease surveillance I'll go through each one of these in more detail in the coming slots census Vital Records registries surveys and administrative data for health care utilization. Census data provides information on the population size and its composition when you census data as the denominator for many of our our data measures and the public health setting it's run every ten years as you might recall we not too long ago completed our census in two thousand and ten and in between the every ten year data collection time points. For us. Census runs the American Community Survey this provides up to date estimates of the population in between the descending census data collection point. Vital Records are those records that are required by law to be reported being clued death records birth records a special focus on fetal death marriage and divorce records. Let's take a moment and talk about the international classification of disease system or I.C.T. I bring it up here because we're talking about but vital records data and the causes of death marked on a death certificate are coded and stored as the international classification of gc's this I.C.D. system is used for coding health outcomes such as health care utilization and death is published by The World Health Organization and every so often it is revised we are looking down downstream very shortly here to have I.C.D. ten available to us in the healthcare setting I.C.D. ten has been used for quite some time for mortality records but will be converting our healthcare utilization classification of diseases to I.C.D. ten in the very near future and it by using this system it helps us come ashore compare ability across state and across countries if everybody's using the same coding system in the same diagnose diagnostic approach we can make comparisons between regions or municipalities. There are sixteen chapters within the I.C.T. of the current I.C.T. system the I.C.T. nine and they are broken up by disease areas Chapter seventeen is a supplementary chapter on external causes of injuries and poisoning. I.C.D. codes. Or international classification of disease codes are really an essential element of any chronic disease surveillance system since we so heavily rely on them for mortality and hospitalization another other administrative data set this little tutorial will show you how to navigate a site that has all the information for I.C.T. nine codes it gives the codes and the description that comes directly from the I.C.T. nine publication there are many other sites out there that are similar to this one although this is the one I tend to go to most frequently I like the layout of the information so this is the I.C.T. nine code look up you'll see here over to the left hand side they have the data stratified in a couple of different ways we can look at the disease injuries disease and injuries either by a tabular index or alphabetic index these are your I.C.D. mean I.C.D. nine diagnosis codes the procedure codes are also arranged in a tabular index or alphabetic index Let's quickly go to the tabular index so this is how you would see it in let's say a table of contents. All right and they're listed here in areas topic areas one through seventeen and then there's some supplementary classifications down here your V. codes and E. codes but one through seventeen are diagnosis codes and there are as I said stratified by subject area so you'll see infectious and parasitic diseases that have codes ranging from zero zero one two one three nine topic area two is Neo Plaza their codes range from one hundred forty through two hundred thirty nine at and these areas of topic areas go on from there their circulatory system respiratory die just of pregnancy and childbirth etc So let's click one of these. They are all hyperlinks So if you are interested in the asthma codes for instance you may click diseases of the respiratory system here with the codes ranging from four six zero to five one nine. And it puts us here to this higher level list of diseases of the respiratory system you can see each of these are also hyperlink so with every link you'll get to more specificity in the codes acute respiratory infections other diseases of the upper respiratory tract pneumonia an influenza chronic obstructive pulmonary disease and allied conditions. I don't know if I can pronounce that and other diseases of the respiratory system. So asthma falls under chronic obstructive pulmonary disease and allied condition so we're going to click that and we'll find out there. And there it is on the list for nine three you also notice that this is a hyperlink I.C.D. nine codes are up to five digits two of to two of those digits fall after a decimal place so here's asthma for nine three. And there's the decimal point point zero point one point two point eight and point nine notice that four nine three point eight is also a hyperlink that means it has another level of specification and that's where we see that fifth digit for nine three point eight one exercise induced Bronco spasm and four nine three point eight two cough variant asthma so all of this information about asthma if we were trying to use this to determine a case definition or use it in hospitalization frequencies emergency department visit frequencies these would be the diagnosis codes we'd want to include for asthma. The other way we could get to this information is just search there is a search here with the dropdown if you see there's you can select which data set to search diseases and injuries. Procedures drugs and chemicals a medical dictionary or hex picks codes are also in here so this this serves as a you know source of information for a variety of codes but if we wanted to search the disease diagnosis codes for asthma we can just. Put this in the search box and list search there click Search. And here we have the information so there's the four and three point zero point one point nine point eight. Four nine three point two and all the different all the different listings for asthma you'll notice here we're using is here even though this wasn't a four nine three code but that's because it's listed as an exclusion so we using without asthma. Is this code seventy six point zero seven so it will list every one of its exclusions. If you want to view this for more detail you can just click a view and it'll take you to that place in the code and you can see that this is just Nia and respiratory abnormalities seven eight six with those subcategory of wheezing that's the point zero seven and it excludes asthma so when they're making a diagnosis on a particular claim or administrative record they'll list this one only if it excludes as well so there was wheezing without the presence of asthma among just click the back button because there was something else I wanted to show you here after you've searched this was something I missed. The first few hundred times I looked at this the search worse search result list this is showing results one through ten but it doesn't tell you how many results there were and so if you come down here to the bottom and you see the next button click that next button and it gives you a whole lot more information and the only time that you are at the end of the list is when that next button no longer appears at the bottom so make sure you don't. Thank you just get ten results and that's all there is check to see if that next button is there. OK the alphabetic list is a bit more complicated to use I prefer the tabular index if I'm looking for a particular condition but you can click to the it's the. Alphabetic list of all the conditions listed in the in the I.C.D. nine code book and they are in Elf order you just have to go searching for it I think it's probably easier to use the keyword search for the tabular index. And I'll just show you the procedures here too often rudely look at procedure codes on an administrative claim or possible discharge record these procedures are the procedures that went on during the visit or during the facility event like a hospitalization or a need to visit. Their listed here also by topic area they start at zero and go to sixteen most of these are operations although there are other miscellaneous diagnostic and therapeutic procedures here in three A. There's obstetrical procedures here and thirteen and more miscellaneous diagnostic and therapeutic procedures listed here. You can click any one of these and just sort of explore Alyse operations on the respiratory system system there's the explosion of the larynx exigence other operations on the long and bronchitis and again this follows a similar process as with the diagnosis codes that the there's a decimal place and up to four digits on these. And you can just keep clicking to exhaust all the possible hyperlinks and that's the extent of the codes and you can see what these these codes actually stand for. Go back to the tabular index you can of course search for the procedure codes as you could for the diagnosis code you just have to click the procedure in the dropdown here so that you are searching the right. The right data set. All right that's I see nine diagnoses I want to show you a little bit about I.C.D. ten so we are in the midst of converting our health care system to reporting I.C.D. ten instead of I.C.T. nine I.C.T. nine is still in existence and as I understand that the government just pushed back the date when it needs to move to I.C.T. ten again it's been pushed back well I think that I when I first started in public health it was just right around the corner and that was about over ten years ago now so it's been a long process to convert to I.C.D. ten but when we do finally get there it will be quite a change mortality records adopted I.C.D. ten diagnosis codes it back in one nine hundred ninety nine so we have been using I.C.T. ten diagnoses on mortality records for over ten years I'm sorry not over ten years not quite yet but for quite some time. That has been over ten years excuse me definitely over ten years. All right so let's click this new site this is a site that I just found actually it's not one that I've used very much most of the work that I do isn't in healthcare utilization data that still uses I.C.T. nine but this this provides I.C.D. ten this is a find a code link and it has other code schemes here not just the city ten there's others that does have I.C.T. nine has D.R.G. codes diagnosis related groups C.P.T. codes and Hicks picks codes I just wanted to show you the I.C.D. ten code said I think it's pretty pretty useful let's look for asthma again they give you lots of options for how to find it and they have coding guidelines for how you would do coding and billing but the ones that I have you focus on here are the I.C.D. ten clinical modification that C.M. clinical modification codes click the codes all of these are live links although they don't appear in blue as they did on the other on the other page but these are all lived. Links and they are the disease categories again so here js arrows zero to J ninety nine are the diseases of the respiratory system recall that on the other site when we were looking at asthma it was four ninety three so they see ten diagnosis codes have become alphanumeric So I clicked diseases of the respiratory system. Scroll down and now I'm going to look for for asthma. And here's the chronic lower respiratory diseases which is the disease category where asthma falls so I'm in a click this hyperlink that the range of J. for zero two J forty seven and we click down a little bit further and we see that there finally as my has appeared J forty five the range is J forty five point two zero two J forty five point nine nine eight well go ahead and click this. Just you can see the specificity in the codes. And there they are as mild intermittent asthma mild persistent moderate persistent severe persistent and other and unspecified and the code ranges are here so we can click this one. And we see all the variants of those uncomplicated with acute exacerbation with statuses matter because so the one of the complaints of I.C.D. ten is is it's degree of specificity and I think that's probably what the does that the designers have approached it this way they've gone to the degree of specificity so that we have a clear understanding of what the diagnosis code was and that there's no ambiguity However with this level of specificity and complexity it's making the health care system have to do so a major overhaul to be able to adopt and use these codes effectively So that's that's one of the challenges that is that we faced in converting from I.C.T. nine to I.C.T. ten is to adopt this complexity. All right. So that I.C.D. ten. We can go back to those codes that take a look at I.C.D. ten for all the variety of codes that they have. Most of the categories the topic areas are similar although there are additional codes in each one of these categories that you won't find in I.C.T. nine so the mapping of I.C.T. nine codes to I.C.D. ten codes has been quite an exercise as well. All right that's an overview of how to find I.C.T. nine and I.C.D. ten codes. The limitations of vital records data are rooted in the variability of reporting with multiple persons reporting information the completeness and accuracy of the data can be compromised for example on the death certificate information is contributed by the physician the funeral director and the medical records are with all of those sources contributing information there could be issues with reporting and completeness some of the measures like maternal smoking during pregnancy and birth certificate have poor validity given by a cs in reporting this information also there are potential issues in identifying mortality trends due to trends in assigning causes of death when more is learned about a certain disease outcome like all timers the cause of death can be assigned with greater accuracy this may in turn affect it it's a fine meant as a cause of death rates of all Simers deaths may be increasing just because we're better at identifying and all timers death. Registries contain a complete collection of health offense they can be for a disease like cancer or a procedure like immunizations or even an exposure like hazardous substances they can be on a very large scale pot. Scale This includes things like the statewide cancer registry that into even a National Cancer Registry or they can be on a very small scale like a particular doctor's office or clinic and their patient population registry. The general limitations of data include timeliness completeness and accuracy timeliness can be an issue when there may be delays in reporting to the registry from a health care provider or but occupational setting where exposure to a hazardous substance has taken place they need to complete forms and submit them to the central registry and that a very mation has to be converted into a into an electronic record in some cases if it's a paper based system so all of this can result in delays and availability of data for completeness a couple of examples in the cancer registry it's often missing the treatment information beyond the first course of treatment so you only really see what their their initial course of treatment is and they don't track it over time. Another piece that may result in a lack of completeness is under-reporting for certain conditions again the cancer registry certain conditions may be under reported like melanoma skin cancer whereas the the other more severe invasive cancers are are more reliably reported. There are a variety of state based population surveys I'll go through most of the recognizable ones here the Behavioral Risk Factor Surveillance survey is a survey of adults eighteen years and older to generate the frequency of risk factors or health behaviors and in some cases actual disease status such as asthma or diabetes. Prey arms or the pregnancy risk assessment monitoring system and. A survey of women with a life Perth to assess risk factors related to birth outcomes. The why R.B.S. or the youth risk behavior survey has a survey that measures high school use it's very similar to the B R A fast survey in adults but instead focuses on the high school population it measures frequencies of risk factors and health behaviors and some disease some disease status indicators. If you are working in public health and let's say a state health department the best way to access B.R. F.S.S. data for your state is to get to know the be R.F.'s us ordinator in your state that way you're able to learn directly from them and their standard reporting for your state however if you're interested in looking at data from other states or maybe you don't work at the Health Department and you don't necessarily have access to that resource you can take a look at the Behavioral Risk Factor Surveillance System data and methods available on the C.D.C. website and so that's what this video will show you how to do it how to navigate the C.D.C. website for some information about the behavior of with Factor Surveillance System. This is at the C.D.C. website forward slash B R F S S And I'll just scroll down here so you can see what some of your options are. This area here gives the survey data and documentation and here are the prevalence data and data analysis tools these are the two main features that I want to show you in this presentation there are fact sheets and publications related to the findings out of the bureau fess However I just wanted to focus on these elements today OK So let's take a look at the survey data and documentation of first so let's say you're interested in doing some data collection for your own jurisdiction that you want to be comparable to the B. or F.S.S. data this is where you come to do that you could get the questions draw. Plea from serving them use them and then you know your data will be you know if collected in the same manner will be comparable to the to the national data or even your state level the R.F.S. US data so at that link you'll see a. Little description of what the survey is and how long it's been running it will give you a data user guide if you were to download data itself which it is publicly available off of the C.D.C. website if you were to download the national data set you would need this user guide I never encourage people to use these publicly available datasets without first getting very familiar with their data user guide and how to how they would recommend doing the analysis so here we go I'm skin to scroll down just a little bit more so you can see a little how it's done. Or. With the how they have it framed up here's here's the piece where you'd want to go to find the documentation and survey data and click that this has all data years available. And they've broken it out into ten year increments Let's go to the most recent area twenty eleven and twenty twelve no notice twenty thirteen is not there yet once it's ready it will be posted here. This is where you'll actually find the survey data itself so if we were to click this link it'll describe for you how to go about to download in the dataset Here's the survey questions that will open up a P.D.F. document with the questions in the format of the survey click that will take a quick look there's the two thousand and twelve questionnaire header and if we scroll down I scroll down to a quick thinking back up there the sections are here not only for the core sections but the optional modules. And they are topics specific and if we go down just little bit this is the interviewer script how they start the survey and collect some information and we get right into the core sections Here's the very first question on the survey how would you say. That your general health is excellent very good good fair or poor. And as the the telephone interviewer reads this question they collect this information in a standard computer format but here it gives you the exact wording of the question. And this P.D.F. goes on for quite a few pages where you can collect all of those survey questions and maybe pick and choose the ones that are are the ones you're really interested in or if you have the resources to implement the the entire course survey that would be fantastic and then you can compare to your state if you're in a local health jurisdiction OK I'm going to hit back here now there's the overview of the two thousand and twelve survey the code book calculated variables so once they can collect all the information from individual questions they may create new variables either from expressing questions or multiple questions and these are what's termed a calculated variable so here it will show you how they did the calculated variables by clicking this P.D.F. document then there's a quality report and the module information modules the optional modules within the be R.F.'s. And then they have the modules by state and by category certain states opt for certain modules they don't have to do all of the optional modules usually they don't because they don't have the resources to support it so certain states will pick option optional modules to implement and so it's there and lastly the waiting for me on this describes how to go about doing the weighting of the final dataset. If we scroll down a little bit further here's where you can actually download the data itself they have it in ASCII file or in a SAS available format. And you could download those data import them into your favorite software analytics software program and start doing analysis on the offense data it is publicly available as I mentioned. But I wouldn't encourage doing that without having thoroughly read the data User's Guide and the methods here in these in these documents so that you are appropriately implementing the analysis. OK So let's say you're not interested in collecting your own information from your jurisdiction and you don't want to collect the survey data off the Web site and do some analysis well then we can go back a couple clicks here. And get back to. Where we can use the tools available on the website. And click this. And here's the analytic tools that they have embedded on the C.D.C. website for you to take advantage of. They have prevalence and trends data interactive maps that utilize us. A web enabled analysis tool so you can create custom Cross have tables M.M.W. our surveillance of surveillance reports that are published annually out of the bureau FS data. Then they're selected Metro and Michael Politan area risk trends this project is an ongoing surveillance project in smaller areas than state level and lastly the chronic disease indicators. Let's just go to the most simple approach but you can you can see there's lots to do here here's the prevalence and trends data let's say we're interested in Michigan scroll down to Michigan here there it is click it will go for two thousand and twelve which is the most recent data a year and I'm interested in a particular category. Let's go with. Health status. And I could go. Now it'll bring up the two questions that fall under that category. How was your general health and health status. So I choose the question how is your general health and here we have Michigan's. Twenty Twelve data. So sixteen point two percent of Michigan adults reported that their general health was excellent thirty four point two percent reported that their general health was very good and etc You see down the line the confidence intervals are provided below that point as the minutes and the sample size is reported below that in this table. If we scroll down a little bit further we see the distribution of the data. I'm not a fan of their three D. Graph. Because I think that it's unnecessary for the state of but that's what they've chosen to use they have a three D. column graph here and maybe we're interested in making a comparison to the Michigan let's say Ohio. And select Ohio. And say go. Now or looking at Ohio and their results they look very similar to Michigan and there's PIOs distribution of responses. Now you may be interested in grouping the data. And by that they mean Sub Pop you lay should analyses so grouped by gender age race income and education and these are the standard five categories by which by Behavioral Risk Factor Surveillance data is reported at the state level. For Let's say I want to take a look at general health status in Ohio by gender I'll say group by gender and push go and now I have the point us to minutes for males versus females Let's scroll down and take a look at the graph and see if we see any different patterns by male and female. Wow We sure don't the pattern looks very similar. Comparing males to females in Ohio. OK so that's how we would do some some very high level analysis by state using the behavioral. Risk Factor data on the C.D.C. website let me see if I can go to Yep here we go select another question. Maybe you would like to do. So maybe we're also interested in over overweight and obesity in Ohio. And will click weight classification by body mass index our body mass index would be a calculated variable using the two items of weight and height on the survey so here we are in Ohio two thousand and twelve results and we have lost our grouping but we do have underweight normal weight or grouping would be listed up here we did underweight normal weight overweight and obese Let's take a look at the graph. Distribution shows that we have. The majority of adults in Ohio are considered overweight. The graph didn't. Produce the axis labels very well so this is normal weight and this is. Obese but this graphic did not produce that very well let's take a look at this by gender and see if we see differences there. We go. Now we have male and female and I'll scroll down and see if we see a different pattern and we do males are might have a much greater prevalence of overweight. They're about the same for obese. And normal weight have a higher prevalence a higher prevalence of normal weight among women. You know. So there we go. All right I think that does it this is a spend an overview of how to access the national data or the state level data on the beer FS website within the Centers for Disease Control and Prevention website. The National Center for Health Statistics coordinates and collects data out of. Very robust survey program these are just a few of the national surveys. The National Health and Nutrition examination survey or and. The National Health Interview Survey or N.H.L. Yes the National Survey of Family Growth and us. The medical expenditure panel survey or maps the state and local area telephone survey or slates and the National Center I'm sorry the National Health Care surveys or an H C S. Here are just some words of caution when using national or state surveys to provide information in the public health setting many surveys conducted at the state level or national level cannot provide information act a more local level this is simply because they don't have an adequate sample size to support estimates at a more local level for example the survey is a state level survey. And provides annually state level estimates of its items on the survey in order to get information at a more local level let's say at a county level multiple data years are combined to provide enough sample size to have reliable estimates. Information from surveys are often self reported and this may be biased information that's provided on the survey that results in this classification either differential or non differential to be aware of that when you're looking at your at your data and what that might mean it is cross-sectional and that longitudinal in nature means we're collecting it in a single point in time many are based on very complex sampling designs. And others are complex designs that require some analytic gymnastics to make sure that the estimates are truly representing the population from one. They're drawn so whenever borrowing from surveys either pulling data off of publicly available data sets make sure you grab the technical specifications as well as any guidance to analyzing the data so you make sure you're. Handling the complex sampling designs imbedded in these surveys in your analysis and there are other sources of bias from just simply the modes of data collection the survey is a telephone based survey if you do not have a telephone you cannot answer the survey so it is biased to a population who has a telephone prams the pregnancy survey for new moms live birth is a paper based survey and requires a certain literacy level and persons with a low literacy level may have difficult a difficult time reporting why are B.S.. Focuses on a sampling of schools and may under represent. Schools because it may focus on just urban area schools and their sampling design may actually bias their samples toward a more urban of urban population. Administrative data offers a fantastic source for public health data about health care utilization as either a risk factor or an outcome. Sources for these include facilities the hospitals themselves or hospital systems health plans or even national data set of administrative data many data types are potentially available. Some of which include pharmacy data office visits lab tests and procedures emergency department or hospital discharge the major caveat to this data is that it is not collected for the purpose of public health surveillance or program planning it's a stablished to pay claims or. Monitor costs or to manage patterns of care so it is a secondary dataset and has some limitations inherent to it because of that. Hospital discharge data is the most common administrative dataset available for public health purposes generally their statewide debt is sad but they can provide information at the local level there's geographic information attached to each claim for each billing record and so you can actually drill down to a zip code level hospitalization rate for a particular cause. We use I.C.D. codes for the discharge diagnoses as well as procedure codes and so we can categorize our hospital discharge into diagnostic groups all right now there I see nine Cotto over soon we should be converting to I.C.T. ten for hospital billing data there are demographic data available such as age race sex J graphic location etc But some of the information on the record like race is poorly recorded and some of them absolutely unavailable for use like total cost because it's so under reported there is limited information it is what puts on the billing record is what's recorded and in some cases the state wide data set limits the number of variables available for public health use so you don't have all the information that you might want about that particular admission it is event based not person based Some states have available to them in their hospital discharge data a unique identifier in Michigan we do not have a unique identifier available for our use in the public health setting so I can tell you from our dataset the number of hospitalizations but I can't tell you the number of people hospitalized. The analysis strategies for chronic disease surveillance are similar to that for the communicable disease surveillance the types of frequency measures you might use include counts prevalent and incident prevalence might be for the actual chronic diseases south the percent of persons in the population with diabetes or it could be for management or mediators for the cost condition for example the percent of persons with diabetes who have their hemoglobin A one C. tested in the last year. Again incidence may be for new cases of a condition like the number of new cases or the rate of new cases of diabetes over time. Or it could be the incidence of disease outcomes like hospitalizations this would be than the rate of hospitalizations for diabetes for a given year. And chronic disease surveillance you'll often see cost measures associated with that it could be the cost of health care utilization or it could be cost relative to the the person with the with the chronic condition or their family. And just like in communicable disease surveillance which often per perform these analyses by person place and time features. The dissemination strategies for chronic disease surveillance data are very similar to that for communicable disease surveillance data we have several jurisdiction level reports they may be web based tables or queries systems and documents and reports these tend to be updated more frequently at least on an annual basis peer reviewed publications like M.M.W. are articles or surveillance summaries and we also will find chronic disease surveillance data reported in other clinical journals these are obviously reported less frequently than the federal or state level report. This concludes the presentation on considerations for surveillance of chronic disease.
Considerchronicdiseasesurveillance
From pblhlth Program in Public Health March 30th, 2017
7 plays
0 comments
|
From pblhlth Program in Public Health September 22nd, 2015 |
Add a comment