Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
â5â Options for a 21st Century Vital Statistics Program I N THE FINAL SESSION OF THE WORKSHOP, Jennifer Madans (National Center for Health Statistics, NCHS) outlined a variety of conceptual challenges and possible future directions for the Vital Statistics Coop- erative Program (VSCP). These directionsâand vital statistics as a wholeâ were then brieï¬y addressed by each member of a panel of discussants; the timing of the workshop left some time, albeit not substantial, for general ï¬oor discussion. Madans noted that probably the biggest challenge in the VSCP is that neither NCHS nor the vital registration areas have, or can have, control over the original source of the data. The individual physicians, funeral directors, and others who complete the information on certiï¬cates do not work for the state registrars and, hence, the registrars and NCHS have only limited means to affect the quality of the information that is collected. The 2003 revision of the standard certiï¬cates and related materials sought to make concepts easier and to provide better guidance (e.g., through the development of separate worksheets), but, ultimately, there is no direct control over the interaction between the data providers and the collector. At the national level, NCHS can promulgate standards but cannot mandate them. Going forward, Madans suggested that some attention is needed on fun- damental questions, such as: Is the current system of collecting a wide va- riety of data items on the birth certiï¬cate the best or most efï¬cient method for getting that information? As Steven Schwartz and Ed Hunter observed in 65
66 VITAL STATISTICS their presentations, birth certiï¬cates are administrative and legal documents with well-deï¬ned civil registration and identiï¬cation purposes. Madans noted that early certiï¬cates were more focused on those needs but, over time, the number of items on the certiï¬cates grew. There were good reasons to add more public health data items to meet real public needs. Madans noted that the U.S. certiï¬cate may be unusual in the number of data items; by comparison, for example, Canadian birth certiï¬cates stick closely to the civil registration model and are limited to the most basic items (name, sex, and date and place of birth). The question is whether the U.S. certiï¬cate (with its 60 data items in the 2003 revision; see Appendix D) is now too large: What is known about the usefulness and quality of all that informa- tion, and what are the real cost implications of adding another item to the collection? To be clear, Madans said, this is not to say that the data items are unimportant by any meansâjust to question whether the birth certiï¬cate is the best vehicle for getting high-quality data efï¬ciently. More generally, Madans spoke of four issues for the future of vital statis- tics that all need to be dealt with, simultaneously, in choosing a future direc- tion: ⢠Infrastructure: Current investments at the state and local levels have been geared to the electronic infrastructure of data processing and col- lection: indeed, the objective in the general category of infrastructure is to use information technology to achieve a faster, more efï¬cient structure. Among the developments to date have been work on elec- tronic veriï¬cation of vital events, development of transmission stan- dards, and creation of web-based systems for local practitioners to use in completing certiï¬cate information. Madans noted that these are all important advances, and advances that have to continue, but it has to be recognized that the âpayoff â of these developments to date has not been as great as hoped. The phased nature of the imple- mentations has been disruptive for trend analysis and overall compa- rability; Madans commented that it is possible that there might have actually been more rapid adoption of the 2003 standard birth certiï¬- cate among the states if not for the difï¬culty of retooling electronic birth certiï¬cates. Madans also referred to statements made earlier in the workshop in commenting that infrastructure alone cannot solve all problems; even with the fastest and most efï¬cient infrastructure, vital statistics are only as fast as the slowest state. ⢠Content: Picking up on her earlier themeâand taking care to note that suggesting changes to data content is often unpopular but nonetheless necessaryâMadans said that the system needs to carefully consider what items are on the birth and death certiï¬cates and why. Is there some limited, core set of public health statistical data items that are
OPTIONS FOR A 21ST CENTURY VITAL STATISTICS PROGRAM 67 truly essential? Are there additional items beyond that core for which the vital record is the only possible source, the most accurate possible source, or the most efï¬cient source? If noncore items are placed on the certiï¬cates, is there a way of obtaining help from other agencies to pay for them? ⢠Short-term considerations: What are the most pressing current needs in the system and in terms of public health knowledge? Madans said that it has to be understood as a basic premise that current funding for the VSCP cannot sustain the current system and content into the future. Against this backdrop, the notion of a core of content and objectives to meet current needs becomes more critical. The current system is already making some sacriï¬ces, in terms of timeliness and (to a lesser extent) data quality, in directing resources to future infra- structure investments: Can the system make better choices about those sacriï¬ces to meet both short- and long-term needs? ⢠Long-term considerations: The goal of comprehensive electronic med- ical records is an elusive one. Madans commented that she was pes- simistic about it being fully implemented in the near future, but she noted that it is eventually coming and that the VSCP needs to start taking it into account because it could fundamentally change the col- lection of vital statistics. If a lot of the information that is currently col- lected on vital records is maintained on the electronic medical record attached to each individual, then vital statistics becomes less a data collection problem than a record linkage activity. Looking ahead, it will also be important to reconcile the quality of the data from admin- istrative and electronic medical records with the data from existing collection systems. In sketching out some proposals for the future of vital statistics in light of these constraints, Madans emphasized that none of these proposals is desirable; all involve real pain, very difï¬cult choices, and loss of data. She described three basic proposals: 1. NCHS could decide (with its stakeholders) that vital statistics are so crit- ical that it would sacriï¬ce all of its other data collections in order to fully fund and support the VSCP The obvious difï¬culty with this ap- . proach is that there are large and vocal constituencies for other NCHS programs, such as the National Health and Nutrition Examination Sur- vey (NHANES) and the National Health Interview Survey. Those con- stituencies would stridently argue the opposite: for example, the strong interest in biomarkers would suggest that NHANES should have primacy and everything else be terminated to fund it. Any major sacriï¬ce of the data collection system is going to be extremely unpopular with a different
68 VITAL STATISTICS constituencyâand for good reason, because these other data collection systems provide equally valuable information. 2. NCHS could completely revisit the cost structure of vital records and vital statistics. Madans said that it would help if funding for vital statistics had a more transparent cost model that could more readily convey the full costs, the different uses of different data items, and, accordingly, who should pay for them. These are not easy things to do because there are joint, overlapping uses. Just as NCHS and the states have worked to show the use and the quality of the data, a ï¬rst step in a future system would be to have a fuller, more informed discussion of the costs of data item additions and deletions and the way in which the costs of the system are distributed. 3. NCHS could implement some combination of major cost-saving mea- sures and reinvest those savings to develop the infrastructure. Such ex- tensive cost-saving measures might include ⢠switching to an effective 2-year cycle for national-level collection and reporting of vital statistics, alternating between birth records one year and death records the next; ⢠collecting only a sample of records; and ⢠reducing the content in the data ï¬les for all, or a large subset, of records, consistent with the idea of a core of essential data items, obtaining funding from other sources for any noncore items. As with the overall options, Madans argued that each of these cost-cutting measures is bad in different respects and would have deleterious effects for many data users (including several of the applications described at the work- shop). Still, the challenge is to consider the advantages and disadvantages of each option or combination of options. In discussion, Kenneth Prewitt added that the NCHS Board of Scientiï¬c Counselors had offered the agency one overriding piece of advice on cost- cutting measures: that across-the-board diminution of resources (and, con- sequently, quality) across all NCHS programs is not a viable option. Tough, if unpalatable, choices are preferable to a protracted death by a thousand cuts. Madans said that there is probably more agreement about what a future vital statistics program should look like, as a goal, than the means to get there. The goal for vital statistics should be a system that is based on both current and future needs; that is efï¬cient but adequately funded, drawing from a variety of sources if necessary; that optimizes data quality and time- liness; that takes advantages of information technology developments but is not limited by technology; and that informs research and decision making. The vital statistics program must also always be thought of as integrated with other data collection systems on health.
OPTIONS FOR A 21ST CENTURY VITAL STATISTICS PROGRAM 69 In reaction to these proposals, and the workshop topics in general, sev- eral invited experts provided their reactions. Howard Hogan (U.S. Census Bureau) prefaced his remarks by suggesting that a nationâs vital statistics system is a reï¬ection of its social concerns, an impression he ï¬rst derived from his dissertation work on infant mortality rates. For example, different countries emphasize details on nationality and parentage but nearly all countries, historically, detail legitimate or illegiti- mate births as an indicator of social conditions. To this extent, Hogan said, Madansâs suggestion that the birth certiï¬cate is dominated by data items on race and Hispanic origin and on public health issues actually says something about us as a nation. Moreover, Hogan argued that a nationâs vital statistics system is indicative of its overall statistical development. That Sweden was able to issue fairly good and detailed birth statistics says something about the social and statistical organization of Sweden in 1790; that alarms have been raised about funding crises and lags in data releases in U.S. vital statis- tics since at least 1975 says something else. Hogan said that vital statistics are, fundamentally and emotionally, a building block of a national statistics system and permitting them to wither away would say a great deal about the nationâs commitment to good statistics and the social problems they reï¬ect. Hogan suggested that the workshop discussionâlike conventional usage over the decadesâequates two things that are not quite the same. Though âvital statisticsâ have been taken to mean âstatistics tabulated from a vital events registration system,â it is possible to interpret âvital statisticsâ more broadly. Hogan noted that various workshop presentations had alluded to the fact that the United States does not have vital statistics on marriages and divorce; this is true if looking at data from registration systems but it is not true if data from different systems are considered. The Census Bureauâs American Community Survey (ACS) includes data items on marriage and di- vorce that appear to provide reasonably reliable results, including number of times married and year of last marriage. Divorce data are somewhat more problematic because people say they are divorced when they have only sep- arated or applied for divorce, but the data still provide useful information. Similarly, the ACS also queries for some items that correspond to birth cer- tiï¬cate dataâmotherâs age, race, Hispanic origin, and nativityâthat com- pare fairly well with the registration data. Although the registration data include much more detail on medical matters, the ACS arguably provides a richer array of social and economic data linked to the same births, permit- ting different kinds of analyses. Other data systems, such as the National Health Interview Survey, overlap the classic âvital statisticsâ data in impor- tant respects. Hogan said that his remarks are not meant to imply that the vital record system is unnecessary. To the contrary, although the surveys can provide data richness, they depend heavily on the âcensusâ of events, i.e., vital statistics
70 VITAL STATISTICS from the registration system, for calibration and benchmarking. Logically, deaths are an example of a vital event for which no survey is capable of pro- viding good and detailed information. In terms of the suggestions for cost savings in the vital statistics systems, Hogan suggested beginning by consid- eration of the vital statistics system as a whole and, from that perspective, think of what kind of data is needed for analysis of births, marriage, and other events. On the basis of this analysis, ideas such as Madansâ notion of a core of content need to be considered but, more importantly, the system needs to consider the ways in which the necessary information can be ob- tained from means other than the vital recordsâeither independent surveys or surveys using the registration system as a sampling frame. Garland Land (National Association of Public Health Statistics and In- formation Systems) agreed with several of Madansâ comments but suggested that the outlook for the current VSCP might be too pessimistic. He said that the impression that the current system is very costly and inefï¬cient and is, essentially, not viable ï¬nancially is too harsh: the current cost of obtaining birth and death record data (less than $2â3 per event) probably compares favorably with data collection in national surveys. Furthermore, he ques- tioned the impression that there are no controls over source data collection in the VSCP He said that the states do have controls in that they regulate . hospitals, funeral directors, and physicians, and the licensing process does give the states some leverage to address data quality concerns. Similarly, he said that NCHS has the ability to âcontrolâ the states through the contracts and purchase orders that underlie the VSCP; NCHS has the ability to effec- tively mandate that data be collected in a certain way or a state will not be paid. He acknowledged that the issues are more complex than his summary allowed, but concluded that there are controls at both the national and state levels that can be used much more efï¬ciently than they are at present. In terms of data content and quality, Land said that the 2003 revision process did go somewhat awry in the major expansion of data items: in meeting demands for research, Land suggested that the result was a cumber- some birth certiï¬cate. Moreover, the additions were not grounded in a full evaluation of the items already on the certiï¬cate. He said that the individual data items vary widely in terms of their quality and require an item-by-item assessment; the current system lacks a means of removing underperforming items. Land found the argument for a more transparent cost model uncom- pelling because of developments in recent decades. The cost models devel- oped in the early 1980s were effective in attaching prices to parts of the VSCP system and were used to form the cost-sharing formula that was used for many years. That work suggested the appropriateness of a 50-50 model, under which the federal government and the states would split the costs. Land asserted that the problem is that the federal government never met
OPTIONS FOR A 21ST CENTURY VITAL STATISTICS PROGRAM 71 that level; the largest federal share over the years has been about 30 percent, and Land estimated that the federal share is currently down to only about 6 percent of costs. Land challenged the cost-saving options currently on the table as suf- fering from the same fundamental ï¬aw: data collection may be staggered or data content reduced, which may produce some savings at the national level, but the states will still have to collect 100 percent of the data for the birth and death records. If states receive less money to collect the same amount of records, timeliness and quality are likely to suffer; sooner or later, sim- ilar to Schwartzâs description of New York Cityâs experience when funds for abortion reporting were eliminated (see Section 3âA in Chapter 3), the states and localities will be unable to afford coding clerks, trainers, or other support personnel, and data quality will decline. Land expressed particular concern that so much is made of the small and steady or decreasing pool of resources available to NCHS as the sole possible source of funding for the VSCP He said that there are major partnersâfor . instance, the Social Security Administration and the Census Bureauâwho might be looked to as sources of funding rather than only as receivers or users of the statistics. Land argued that the same kind of âmarketingâ of the data that is done at the state level is not adequately done at the national level: for instance, states have been active in ï¬nding new opportunities for use of the National Death Index. Land suggested that the amount of ad- ditional funding needed to pay for the current system is actually relatively modestâperhaps some $10 million annually, with $20 million in start-up costs for reengineering efforts in the states most in need. Land concurred with Schwartz that the VSCP would beneï¬t from greater attention to demon- stration of value and return on investment; the system needs to ï¬gure out how to market its products to a broader array of partners, so that NCHS is not alone in carrying the load at the federal level. Nancy Krieger (Harvard School of Public Health) suggested that two concerns be kept in mind, regardless of which type of decision is pursued. One is that equity has to be at the core of any considerations; the second is that it is important to view the current situation in a broad historical perspective as well as in an international perspective. Krieger said that reduction of health inequities is only possible if the nationâs health data systems collect the data needed to understand the prob- lem. Concerns about inequities in health along the lines of race, ethnicity, and socioeconomic position are central to the public good; hence, it is crit- ical that changes to the vital statistics system not affect the ability to obtain those data. She noted that the idea of going to a sample-based approach for records collection would be a serious impairment to understanding in- equities, undersampling or excluding some populations and oversampling others.
72 VITAL STATISTICS With regard to broad historical perspective, Krieger said that she was thinking about the debates that occurred in the 1930s when ofï¬cials began to see the use of having not only vital statistics, but also more general data on health status. With that concern came the ï¬rst national health survey, originally performed to assess the impact of the Great Depression. During that period, the nation came to realize that it needed different types of data for different purposes, and that different data streams can be mutually infor- mative. For this reason, Krieger expressed concern about seeing NCHS data collections (including the vital statistics) pitted against each other in terms of funding needs or priorities. The VSCP is sometimes criticized for having a somewhat 19th century approach with its emphasis on certiï¬cate processing, but Krieger suggested that there is much to be learned from the perspective that drove the devel- opment of statistical science in the 19th century. During that era, disciplines had not yet fractured to the extent they have nowâthere was no real distinc- tion between statisticians, epidemiologists, and population demographersâ so statistical work tended to bring together disparate aspects of government and research. Krieger argued that, going forward in vital statistics, there needs to be an emphasis on interdisciplinary research and intersectoral work; this amounts to returning to historical roots in the development of statistics writ large, as well as the emergence of the population health ï¬eld. Though not part of the ï¬nal workshop session, a thread of discussion from an earlier workshop panel is particularly relevant to the notion of fu- ture discussions. At the end of the opening session on uses of vital statistics data, NCHS director Ed Sondik challenged the speakers to pick the one area for research or the one aspect of the VSCP that they felt was most in need of remedy, going forward. The speakers responded: ⢠Nancy Krieger (Harvard School of Public Health) commented that col- laborative research involving academic researchers and staff of the rel- evant government agencies, together with staff in the state and local health departments, would be greatly beneï¬cial for understanding key data issues. She particularly noted the status of population estimate methodologyâunderstanding the denominators of vital statistics rates. She added a cautionary note that this kind of methodological research is not always fully appreciated, observing that her geocoding project work (see Chapter 2) was originally funded by the National Institute of Child Health and Human Development at the National Institutes of Health (NIH). When the program was up for renewal, NIH was less receptive to methodological work on public health surveillance. Though funding for such collaborative work may be difï¬cult to se- cure, it could provide a better foundation for the system in the future.
OPTIONS FOR A 21ST CENTURY VITAL STATISTICS PROGRAM 73 Krieger also noted infrastructure developments currently being implemented in the Canadian province of Manitoba, along the lines of integrated systems being developed in Indiana. These systems come closer to integrating data systems across the life course, following peo- ple over time and permitting linkage of records as people go through different health systems. Work to ï¬gure out how to do the same thing more broadly in the United States would be a useful goal for infras- tructure development, akin to Madansâ list of priorities for future di- rections. ⢠Peter Van Dyck (U.S. Department of Health and Human Services) com- mented that the pressing concern is timeliness. For infant deaths, for example, the credibility of the system is undermined when the depart- ment is always reporting 3- and 4-year old data. Resources and efforts speciï¬cally geared to curbing data-release lag times should be the top priority, he said. ⢠Consistent with Kriegerâs broad historical and international perspec- tive, Richard Rogers (University of Colorado) suggested that the sys- tem would beneï¬t from serious attention to international comparisons, as well as state and local comparisons. He also noted that interfaces between the vital records reporting systems and systems for reporting deaths in special circumstancesâmilitary deaths and victims of natural calamities and other catastrophesâalso merit exploration. In summary comments on the workshop, Charles Rothwell (director of vital statistics, NCHS) commented that the VSCP faces great challenges, but expressed conï¬dence that they can be overcome. He agreed with the general gist of Landâs commentsâthat the nation gets a great deal from the current system that is not fully appreciatedâbut also agreed that there is still a set of major issues. Clearly, timeliness is a major issue for the system, both for the delivery of its current products and for the prospects of health surveillance or other new activities. He concurred that NCHS and the system are not doing enough to continually evaluate data quality, going back to check the accuracy and consistency of long-standing data items on the birth and death certiï¬cates, much less the new items in the 2003 revisions. Rothwell said that there has to be more to the vital statistics system than just funding the states and hoping they do well: Simply allocating more money to the VSCP is not necessarily the answer. He said that what needs to happen is to have programs in place that will help states that are weak in certain areas so that they can improve their systems; there must also be a commitment for VSCP ï¬nances to be appropriately and effectively used. Workshop planning committee chair Louise Ryan (then, Harvard Uni- versity Department of Biostatistics) closed the workshop by expressing the hope that this workshop and the discussions it spurred would be a catalyst
74 VITAL STATISTICS for future action. The vital statistics system needs to be understood as a critical part of the nationâs scientiï¬c infrastructure, and building awareness of the systemâs strengths and limitations is essential to continuing to provide vital data for national needs.