Reference Reviews

Péter's Digital Reference Shelf

October 2008

Title: State Health Facts
Publisher: Kaiser Family Foundation
Cost: free
URL: http://www.statehealthfacts.org/
Reviewed: August – September, 2008

THE CONTEXT

The U.S. government agencies do provide health-related statistics, but not always in the ideal form. Undoubtedly, the U.S Department of Health and Human Services has the most comprehensive information services for the public. Its flagship organization is the CDC, the Centers for Disease Control and Prevention. From the perspective of statistics, the National Center for Health Statistics is the central organization.

The Web version of the most recent annual report, Health, United States, 2007, is a useful and well-organized resource, offering a chartbook in PDF, PPT and Excel format for the figures and tables (in addition to the full text of the annual report). The example of cigarette smoking trends between 1965 and 2005 for men, women, high school students and pregnant women illustrates well the best segments of the CDC’s statistical information. Still on the good side, some statistics are very current, presented as early releases. This is best illustrated by the Early Release of Selected Estimates Based on Data from the January–March 2008 National Health Interview Survey released at the end of September. It is laudable, that many of the charts clearly indicate if the digital version has more current data than the print publication.

One of the most current and state-of-the-art statistical systems originating from the CDC’s National Center for Chronic Disease Prevention and Health Promotion is BRFFS, the Behavioral Risk Factor Surveillance System, based on regular phone surveys. Many of its results are based on intuitive, geographic information systems tools to visualize the results, and select the depth of information, such as the findings of the survey about the 2007 Health Status of the nation, presented on a dynamically and extensively customizable map.

However, this exemplary presentation of a set of statistics is not common. The selective information services where users can customize the scope of the measures and the population reported are limited. It does not help the propagation of CDC’s worthy services that in many cases the promising charts fail to display more than an error message. When this happens time and again, users give up using the information service. This happened when I tried the potentially useful, although topically limited Web–based Injury Statistics Query and Reporting System (WISQARS) service.

SAMSHA, the Substance Abuse and Mental Health Services Administration of the U.S. Department of Health & Human Services, provides statistics at the national and state level about prevalence and treatments within the scope implied by its name. It also allows online analysis of data selected by users and provides good tutorials.

THE CONTENT

The Kaiser Family Foundation (KFF) is possibly the most comprehensive information source about matters related to health policy. While the army of talking heads on TV spinned a lot, talked a lot but said very little without obvious political bias during the presidential campaign about the depressing situation of healthcare services and health status in the U.S., the KFF provided the clearest, fact–based summaries of the health of the nation, and the agenda of the candidates, in its special Web database, Health08.

It could do that because it has been collecting, analyzing and reporting various health statistics of the nation through its excellent State Health Facts factographic database (as well as of the world through the Global Health Facts database). It must be emphasized, that KFF does not have any stake in the Kaiser Permanente network of clinics and other medical service facilities.

The majority of the raw data are gathered from a variety of censuses, surveys of U.S. federal and state government agencies and from polls and surveys of KFF or its partners – offering an outstanding site for one–stop shopping for well–structured, consolidated and very well-presented information free.

The scope and structure of the database

There are statistics on hundreds of health and health–related topics in this database. KFF claims more than 500 topics, but if you scroll down the list of topics it seems that there are more like 600 topics for the 50 states plus the District of Columbia, and the three U.S. territories: Guam, Puerto Rico and the U.S. Virgin Islands.

In some surveys, other dependencies and independent nations in free association with the U.S. are also included, such as Palau, the Marshal islands and the Federated States of Micronesia. This explains why there are more than 54 responses to some of the survey questions, such as the confidentiality of the HIV test. These don’t appear on the statistical maps, but the three territories’ data appear in the tables.

Not surprisingly, not all states provided data for all the indicators in the Current Population Surveys of the Census Bureau, neither to all the questions of the surveys conducted by other government agencies and/or KFF. These are clearly marked on the statistical maps, although the answer types vary. Sometimes they are strange, such as the DK (Don’t Know) answer about the requirement for pre–test counseling for HIV testing from the state of Minnesota, which has one of the best computerized health information systems in the country.

The roughly 610 indicators are grouped into about 130 sub–categories, and 11 main categories. Some of the indicators are redundant as they appear in two or more categories, and there are some other oddities discussed below. One of the main categories, Demographics and the Economy, is not health–oriented, but as we know and learn even better from this illuminating resource, practically all the health–services issues are very much related to the economic situation of the population, of the states and of the nation. There is an additional “featured” category about children, which pulls together from the other categories the indicators related directly or indirectly to children from newborns to teenagers. (I have not counted these duplicates in my count of indicators).

Some of the subcategories and indicators rarely appear in the typical health statistics, but they are highly informative (and saddening), such as the minimum in–patient mastectomy stay within the sub–category of Mandated Benefits in Private Insurance. Only 20 states mandate it, and even in those states that mandate a minimum stay, only four states require that the length of stay be determined by the surgeon in consultation with the patient. In 13 states, the mandate is 24 hours, in five states it is 48 hours, in Connecticut it can be longer if medically necessary. In Virginia, the insurance companies apparently lobbied out a distinction between radical/modified radical mastectomy (48 hours) versus full/partial mastectomy (24 hours). To see the details through State Health Facts takes much less than it takes to read this paragraph, because it paints an informative map in addition to displaying the traditional tabular data to show which states are the Scrooges. I am not qualified medically to judge this compromise, but I am ethically appalled by envisaging insurance agents arguing with the patient about the need for the type of mastectomy in an effort to reduce the reimbursement expenses of the insurance companies. Banks, insurance companies and financial institutions operate at around 20% profit rate, but as we have seen recently greed, incompetence and criminal neglect can drive them into bankruptcy faster than they can utter bail–out.

The breadth of coverage

It varies widely how many indicators are listed within one sub–category. There are a few that have a single indicator which is functionally the same as the sub–category, such as unemployment, gross state product, percent of immunized children, occupational fatalities, violent crimes and motor–vehicle death rate.

On the other extreme are the indicators of the AIDS Drug Assistance Program (ADAP) subcategory with 19 indicators. The AIDS/HIV main category itself gets the most extensive coverage in the database with 13 sub–categories, and a total of 68 indicators. This may be because one of the top goals of KFF is the reduction of AIDS cases (not only in the U.S. but around the world). Noble as this goal is, from the perspective of the U.S. its prevalence is orders of magnitude lower than that of other diseases/conditions that are the main causes of death, such as cancer, cardiovascular and cerebrovascular diseases.

The types and number of indicators related to insurance, forms and budget of healthcare services and providers are the most impressive.

The only indicators that are sorely missing are the prevalence and/or death rate statistics about prostate cancer, the highest frequency cause of cancer death for males, and maternal mortality rate, where the U.S. has a poor position. It is in the same league as the economically far less developed Bulgaria, Lithuania and Portugal with 11 deaths per 100,000 live births. It is behind Cyprus, Latvia and Macedonia and is much behind –among others– Japan, Hungary, the Netherlands, Slovakia and Slovenia, where the rate is six. It is more than remarkable that maternal mortality rate is more than three times as high in the U.S. as in Greece, Italy and, figure this, Bosnia & Herzegovina – according to estimates consolidated by the World Bank and various UN agencies, such as WHO, UNICEF and UNFPA, reported in World Health Statistics 2008.

Otherwise, women's health statistics are excellent in every regard, in a main category of its own, with nine subcategories, and 45 indicators. Not all of them are gender–specific, let alone unique for this category as they appear in at least one other category, such as osteoporosis screening and mandated generic depression treatment coverage (i.e. not restricted to postpartum depression. (I found one confusing indicator about state-mandated coverage that is probably a software glitch to be covered in the software segment.).

The other side of the coin is –as I mentioned above– the omission of indicators (except for prostate screening) about the widely prevalent prostate cancer (with a five–year survival rate similar to breast cancer), and some non–gender specific indicators for the male population. It is sort of painting the lily to have an indicator “Percent of Women Who Had Dental Visit in the Past Year” with a value of 71.8, when there is another indicator in the Health Status main category “Percent of Adults Who Visited the Dentist in the Past Year” with a value of 70.3 % for the same year. You can figure out what percent of males did so, but it would be more efficient to have one indicator that shows the percentage by gender breakdown. The same could be applied to the population with arthritis. Data only appears in the Women’s Health category, even though there are between 17–18 million male adults known to have arthritis – versus the 24–25 million women. I can’t fathom why there is an indicator about colorectal screening in the Women’s Health category, and why there is no data for men, when it ranks third for both men and women for cause of death according to the National Cancer Instittute. At the same time, it is reasonable that the eating disorders indicator appears only in the Women’s Health category, as about 90% of the known population afflicted by bulimia and anorexia is female.

The Minority Health main category is not really needed. It has three sub–categories and eight indicators. Three of the population distribution sub–category indicators are already available in the Demographics & Economy main category, and that one should break down its Other group to Asian–American/Pacific Islanders, Native Americans and Two or More Races groups. The Minority Nonfederal Physicians indicator is already part of the Distribution of Nonfederal Physicians by Race indicator’s table with more details in the Physicians sub–category of the Providers & Service Use main category. Similarly, the Distribution of Medical School Graduates by Race/Ethnicity and by Gender should be merged into the Total Number of Medical School Graduates indicator table. FKK does cover well minorities across the categories (where such information is available), and does not need this extra category for merely political correctness. Having the race/ethnicity/gender breakdown and total in one table provides a better context at–a glance than such scattered listing. It is another question that many of the race and ethnicity indicators in the source documents specifies only White. Black and Hispanic, and relegates Asians, Pacific Islanders and Native Americans to the Other category. It is worse when government source documents lump Hispanics in the Other category as happens with essential indicators such as death rates for several diseases that are very different among the races and major ethnic groups.

THE SOFTWARE (and “KNOWARE”)

The software makes browsing and displaying intuitive and very smooth. There are some glitches with searching, but searching is not a critical step in the State Health Facts database. The dominant access mode is browsing as the expandable and collapsible topic tree is one of the prominent starting points beyond looking up information by state through clicking on the country map on the homepage. Beyond the elimination and consolidation of the modest and redundant Minority Health main category (which may not help to reveal the many race and ethnicity–specific indicators), the only feature that should be improved is the naming of the indicators.

It looks bad that about 10 of the indicators that provide data for males and females, the qualifier “by sex” is used, for about two dozen others the qualifier is “by gender”), such as Death Rates by Gender versus Adult Smoking Rate by Sex. These should be consolidated even if the source itself may use the non–preferred term. It should be changed to the preferred, standard term in the topic tree – after all this is not cataloging where AACR2 rules would apply. It makes no sense to rename the widely known and internationally standard term of Infant Mortality Rate as Infant Death Rate – especially as the source document use the much better know term.

In those cases where there are two or more indicators from surveys in different years, the year should appear in the name of the indicator on the tree. I was flabbergasted when I found different data values for the same indicator when I corroborated my notes for this review and re–ran some of the searches. The name of the indicator selected is highlighted on the tree menu – a smart idea. However, when I searched for "contraception" no indicator was highlighted and the State Mandated Benefits: Contraceptives, as of August 1, 2007, appeared as the indicator title on the screen, and its values differed from the ones in my notes.

The difference was very significant as this table reports that contraceptive coverage is mandated by 33 states. My note said 24 states mandate the coverage. The tree showed the indicator as Contraceptive Coverage and when I searched for "contraceptives" it got highlighted and showed a different table (with the values in my notes, and with the title of the table as State Mandated Benefits; Contraceptives, 2008). The layout of the table was not identical either, but more unnervingly the more current values showed a big step back in this matter, which is possible. It seems like a combination of human and software error, and must be corrected. There must be a reason why the 2007 data may not be accessible through the tree menu. Hopefully, KFF will trace down this problem and fix it, and check the few other cases where data from different years, with different indicator values, layout and content may be there.

The results can be displayed as tables, bar charts and maps. The bar charts could be made much more compact by using the traditional bar chart format and showing more state data without scrolling. There is a lovely option to display both the map and the tabular information in one fell swoop. It adds to the pleasure that the tabular data may be sorted by states or by values, and the tables can be downloaded as tab files for further processing. Surprisingly, even the notes fields may be sorted. Thanks to the consistently composed notes, this can be useful to see the ifs and buts for the variation of above mentioned mastectomy hospital stay issue.

The search module uses Google search Appliance. Google would not be Google if it reported the actual number of records that it found. As always, it inflates its hit counts by reporting the same item two or more times. The search for "newborn" (in the singular) reports to have found five hits. Actually, there are two hits, but the same table about Newborn Cystic Fibrosis Screening is brought up no matter which entry you click on in the result list. Right clicking on the entries in the hit list and displaying their properties show the identical URLs – without making the unnecessary revisit. Sometimes, however, the seemingly identical indicator titles take users to different tables.

For the indicator Newborn Hearing Screening, Google reports to have found three hits: two in the Health Status Main category, and one –oddly– in the Managed Care & Health Insurance main category. Actually, they lead to the very same indicator tables, and on the topic tree this indicator – rightly– does not appear in the latter main category. Probably it was originally a human error – later corrected but not reflected by Google’s hit counts. As for duplicates within the same main category – it is just an enigmatic bonus that shows up often but not always. This confuses and irritates the users, but another glitch deprives them from important data. The term newborns (in the plural form) does not retrieve anything through the Google Search Appliance even though there is an important (and once again, saddening] table for the indicator about HIV Testing for Mothers and Newborns – offered only by 10 states.

In spite of some of the shortcomings in content and software (which are easy to fix), this is a very good and very current resource for researching many dimensions of a quintessential issue. The software leads the users to discover the differences among states (although not among districts of the states) and to see the whole picture by allowing side–by–side comparison of two states or one state and the U.S.–wide data. This is very useful. It would further enhance the system if it would offer scattergrams to examine the correlations between indicator values selected by the users – across the country at a glance. This would be particularly effective for managing better the ever–increasing costs of healthcare services and producing more impressive results in the health of the nation.