NATIONAL CENTER Series 4 For HEALTH Number 7

VITAL and HEALTH STATISTICS

DOCUMENTS AND COMMITTEE REPORTS

use of Vital and Health Records

in EpidemiologicResearch

A Report of the United States National Committee on Vital and Health Statistics

An analysis of the changing needs for in epidemiologic research in relation to the present vital and health statistics system with recommendations for use of existing vital and health records and for developing new data for epidemiologic studies.

DHEW Publication No. (HSM) 73-1265

Washington, D. C. March 1968

U.S. DEPARTMENT OF HEALTH, , AND WELFARE Public Health Service John W. Gardner William H. Stewart Secretary Surgeon General Public Health Service Publication No. 1000-Series 4-No. 7 NATIONAL CENTER FOR HEALTH STATISTICS

THEODORE D. WOOLSEY, Director

PHILIP S.LAWRENCE, SC.D., Associate Director

OSWALD K. SAGEN, PH.D.,, Assistant Director for Health Statistics Development

WALT R. SIMMONS, M.A., Assistant Director for Research and Scientific Development

ALICE M. WATERHOUSE, M. D., Medical Consultant

JAMES E. ‘KELLY, D. D. S., Dental Advisor

LOUIS R. STOLCIS, M.A.,Executive O//icer

DONALD GREEN, lf@~afio~ of/icey

OFFICE OF HEALTH STATISTICS ANALYSIS

IWAOM.MORIYAMA,Ph.D.,Director DEAN E.KRUEGER, Deputy Director

Public Health Service Publication No. 1000-Series 4-No. 7

Library of Congress Catalog Card Iiurnber 67-62375 FOREWORD

This report, prepared under the auspices of the U.S. National Com- mittee on Vital and Health Statistics considers ways in which vital and health statistics systems can better serve the changing need for data in epidemiologic research. These needs arise from the increasing con- cern with chronic noninfectious diseases and the measurement problems related to the nature of these diseases, Recommendations are made for modifying the content of basic vital and health records, and for developing new ways of making data from these records available for epidemiologic research and other uses.

Robert L. Berg, M.D. Chairman U.S. National Committee on Vital and Health Statistics

... Ill U.S. NATIONAL COMMITTEE ON VITAL AND HEALTH STATISTICS

Robert L, Berg, M.D., Chairman Herbert E. Klarman, Ph.D. Professor and Chairman, Department of Department of Public Health Administration Preventive Medicine and Community Health School of Hygiene and Public Health The University of Rochester The Johns Hopkins University Medical Center Baltimore, Maryland Rochester, New York Iwao M. Moriyama, Ph. D., Executive Secretary Everett S. Lee, Ph.D. Director, Office of Health Statistics Analysis Department of Sociology and Anthropology National Center for Health Statistics University of Massachusetts Public Health Service* Amherst, Massachusetts Washington, D.C. Donald J. Davids John R. Philp, M.D. Chief, Records and Statistics Section Health Officer Colorado State Department of Public Health County of Orange Health Department Denver, Colorado P. O. Box 355 Santa Ana, California William Haenszel Chief, Biometry Branch Donovan J. Thompson, Ph.D. National Cancer Institute Department of Preventive Medicine National Institutes of Health Schgol of Medicine Public Health Service* University of Washington Bethesdaj Maryland Seattle, Washington Clyde V. Kiser, Ph.D. Senior Memker, Technical Staff Theodore D. Wmlsey, Ex officio Milbank Memorial Fund Director, National Center for Health Statistics 40 Wall Street Public Health Service* New York, New York Washington, D.C.

SUBCOMMITTEE ON USE OF VITAL AND HEALTH STATISTICS IN EPIDEMIOLOGIC RESEARCH

Brian MacMahon, M.D., Chairman Elmer A. Gardner, M.D., Director3 Professor of Community Mental Health Center Harvard School of Public Health Health Sciences Center Boston, Massachusetts Temple University Philadelphia, Pennsylvania Lillian Guralnick, Secretaryl Social Science Research Analyst Health Insurance Research Branch Robert W, Miller, M.D. Divison of Health Insurance Studies Chief, Epidemiology Branch Social Security Administration* National Cancer Institute Washington, D.C. National Institutes of Health Public Health Service* 1. M. Moriyama, Ph. D., Ex officio Bethesda, Maryland Executive Secretary U.S. National Committee on Vital and Health Statistics Donald L. Rucknagel, M.D. Washington, D.C. Department of Human Genetics John Cassel, M.D., M.P.jf. University of Michigan Medical Schcd Professor of Epidemiology Am Arbor, Michigan Schcol of Public Health University of North Carolina Colitt White, M.D. Chapel Hill, North Carolina Professor of Biometry Carl L. Erhardt, SC.D., Director2 Department of Epidemiology and Public Health Health Intelligence Statistics School of Medicine The City of New York Health Services Administration Yale University New York, New York New Haven, Connecticut * Department of Health, Education, and Welfare

lWhen appmn“ ted to the Suhcommittse, Miss Guralnick was a , Office of Health Statistics Analysis, National Center for Health Sta- tistics, Public Health Service, Department of Health, Education, and Welfare, Wasbingtcm, D.C. %fntil July 1, 1966, Dr. Erhardt was Associate Director, Office of Research, City of New York Deportment of Health.

3When appointsd to the Subcommittee, Dr. Gardner was Assistant Professor and Director, Division of Preventive Psyohia@, University of Rochester School of Medicine and Dentistty, Rochester, New York.

iv CONTENTS Page

.,. Foreword ------111

U.S. National Committee on Vital and Health Statistics ------iv

Subcommittee on Use of Vital and Health Statistics in Epidemiologic Research ------iv

Introduction ------1

Need for Review ------2

Procedure ------3

Findings and Recommendations ------3 Access to Vital and Health Records ------3 Record Linkage ------4 The Unique Number ------4 Ascertainment of Death ------5 Followup ------6 Information on Vital and Health Records ------6 Cause of Death ------6 Congenital Malformations ------7 Other Items ------8 Development of Other Resources ------8

Summary of Recommendations ------8

Appendix I. Selected Bibliography on Record Linkage ------10

Appendix II. The Underlying Cause Concept ------12

v IN THIS REPORT the changing needs for data in epidemiolo~”c re- search are examined in relation to the pvesent vital and health statis- tics systems. The ascendancy of chronic noninfectious diseases as major causes of movbidity and mortality requires new types ofdu$a and changes of em- phasis in existing types of data. New measurement problems are often encountered in these diseases, in which onset may be insidious, pro. gression slow, and the interval between causes and effects long. Each source of vital and health statistics is examined in terms of the information requested on the basic record, the present of access to the data on the vecord, and the tabulations now being prepared by the National Centev for Health Statistics.

Linkage of vcwious vital and health records is seen as a means of in- creasing their sepavate valwe, and fuvther exploitation of this device is recommended. Use of some ynivenal identity numbev fov this purpose should be explored along with the possibility of assigning such numbers at bivth, A National Death Index would be a great- asset in epidemiologic re- search. The National Center for Health Statistics is urged to explore the technical pvoblems in establishing this vesouvce. Extension of the scope of patholo~”cal conditions vecorded on death certificates beyond those contributing to death should be considered. Other recommendations are made for development of vital records in- formation on congenital ma~ormations and for inclusion of a sample of newbovn infants in the Health Examination Survey.

vi USE OF VITAL AND HEALTH RECORDS IN EPIDEMIOLOGIC RESEARCH t!~~~i% some t~ths, ad not commonly-believed opinions, to arise ~?om my meditations upon these neglected papevs (the Bills of Mortality), I proceeded j%rther, to consider what benefit the knowledge of the same would hying to the world; that I might not engage myself in idle, and useless speculations, but . . . present the world with some real fYuit Jrom those ayyie blossoms.”

John G~aunt, 1662

INTRODUCTION 2. For comparison of disease rates in dif- ferent populations, indifferent parts of the same population, and in similar groups The practice of epidemiology, or indeed of over a period of time, in order to develop public health, would today be inconceivable with- hypotheses regarding the etiology of dis- out access to vital and health records and the tabu- ease. Studies over time may also serve lations routinely assembled from them. The data to evaluate the effectiveness of preven- obtained by Graunt’s “antient matron” searchers tive or therapeutic measures. have been replaced by legally required certificates of birth and death. These documents, and the 3. For identification of groups of people at breadth of their application, have changed re- high risk of particular diseases. Such markably in the last 50 years, and today they groups may be of significance in disease probably constitute the single most important re- control or they may be particularly suit- source of epidemiologic investigation. In addition, able for studies of etiology and therapy. entirely new sources of health statistics have been 4. As the starting point for “follow-back” developed, notable among these, in the United studies in which a series of cases with States, being the several programs of the National particular characteristics (e.g., dying Health Survey. from a particular disease) is identified With respect to epidemiologic purposes, vital from the primary source (in this instance, and health records serve the following functions: the death certificate) and supplementary information is sought from other sources 1. As source data for the measurement of the (e.g., the certifying physician) relevant to incidence and prevalence of disease. etiology or other subjects of interest. These measurements serve administra- tive as well as epidemiologic purposes, 5., As the end point for studies in which sub- Because of the limitations of diagnostic sets of the population are selected because information derived from certificates of of their unusual characteristics or en- birth and death, it is to this general pur- vironmental exposures and followed to pose of providing source data that most identify diseases or other outcomes sus- of the newly introduced types of health pected of being related to the selected statistics have been addressed. factors.

1 Erhardt 1 and Moriyama~ have recently com- tential contribution of epidemiologic research. at piled some examples of the ways in which vital the present stage of knowledge of those chronic records are being utilized for epidemiologic diseases that now constitute the predominant research. Major projects in this country that are health problems in this country. dependent on vital records include the extensive This shift in emphasis of epidemiologic in- series of studies of veterans carried on by the vestigation has required great changes in method- Follow-Up Agency of the National Academy of ology, if not in philosophy. New or newly em- Sciences, National Research Council, the large- phasized methods are required to deal with scale studies of the effects of smoking on health problems such as: by the National Heart Institute and the American 1. Detection of causal associations in which Cancer Society, the followup programs of regis- decades or generations may lapse between tries of cancer and other specific diseases,, and cause and effect. a great many smaller scale studies aimed at specific problems. 2. Measurement of existence and stage of It is difficult to assess the extent of epi- diseases whose onset and progression are demiologic use of the statistics published in insidious and often unknown to the patient Vital Statistics of the United States or in the for many years. reports from the National Health Survey. Re- 3. Description of demographic distribution of ference to any journal of epidemiology or public conditions of low mortality for which the health will yield examples of such uses, but ex- quality of medical diagnosis may vary tensive use is also made of them by individual markedly between populations and popu- investigators in searching for ideas and in pre- lation subgroups. liminary testing of hypotheses-activities that largely pass unrecorded. 4. Scientific and administrative problems of cohort studies that require large numbers NEED FOR REVIEW and long-term followup. Detection of clustering of disease in time The emphasis of epidemiologic investigation 5. or place at a much lower level of intensity has shifted markedly in the last two decades. A than that observed in frank infections. decline in interest in the infectious diseases and increase in concern with the noninfectious dis- With these types of problems in minfi, “it has eases have resulted from the change in relative seemed useful to review existing health statis- importance of these categories of disease in tics and their sources to determine whether they many parts of the world, including the United should and could be modified to better serve States. It is also recognized that, although major epidemiologic needs. tasks still remain in the improvement of control The review has been further stimulated by over the infectious diseases, the contribution of recognition of the enormous increase that has epidemiology to the development of control occurred and will continue in the technical feasi- methods is largely past —new advances are being” bility of data reduction, retrieval, and analysis made predominantly as the result of work in the on a large scale. The potential of this improved experimental laboratory and. through the better technology may be unrealized if the basic records application of existing knowledge. On the other are not appropriately modified. hand, the identification of cigarette smoking as In the past, the nature of health statistics the major cause of this centuryis epidemic of and the usefulness of vital records have &en lung cancer has clearly demonstrated the po- limited by the fact that the records were es- tablished and continued for legal rather than health purposes. In the last 50 years, however, lErhardt, C. L. Vital records are useful. Harvard, Pub. Health Alumni Bull. 227-11, 1965. there has been increasing acceptance of the use 2Moriyama, I. M.: Use of vital records for epidemiological of vital records for public health and scientific research. J. Chronic W?. 17: 889-897, 1964. purposes. There is, rherefore, a good prospect that the demonstration of significant epidemio- the use and dissemination of data on individuals logic needs will be influential in producing as individuals and their use for describing the changes or applications in vital records and characteristics of groups without the identifi- statistics. While most of the information that cation of individuals has become widely recog- has been added to vital records and the new nized and accepted as a basis for statistical and sources that have developed have been based on epidemiological use of c, mfidential records. a request for voluntary response, rather than as Computers now provide new and even more ef- a legal requirement, there is no evidence that fective techniques for protection against the tpis- quality or completeness of information has been use of confidential information. compromised on this account. The subcommittee did not take up the ques- tion of infectious disease reporting, the national PROCEDURE component of which is now the responsibility of the National Communicable Disease Center The subcommittee has limited its considera- (NCDC), Atlanta, Ga. However, it is believed that tion to the uses of health records and statistics in a review of this area would be useful, and the investigating the etiology of disease. Although National Committee might consider the establish- this may imply a rather narrow definition of ment of a subcommittee with special competence “epidemiologic uses, “ it embraces the most im- in infectious disease and with representation from portant use of epidemiologic data. Other sub- NCDC and State Health Departments to review committees of the National Committee have, of this field. course, keen concerned with medical care, de- With respect to each of the types of records “epi- mography, and other areas in which reviewed, the subcommittee considered (a) the in- demiologic” data have application. formation existent, or requested, on the basic rec- The committee has reviewed the standard ord, (b) the tabulation now prepared from this certificates of birth, death, fetal death, marriage information by the National Center for Health Sta- and divorce, now in the final stages of a de- tistics (NCHS), and (c) the existing means of ac- cennial revision, as well as the tabulations regu- cess to the information contained on the records. larly developed from these sources and published By far the most important gap between in Vital Statistics of the United States. Programs, existing and potential usefulness seemed to lie data forms, and reports of the three main com- in connection with item c, access to records. ponents of the National Health SurveJ –The Health Since the major recommendations of the sub- Interview Survey, the Health Examination Survey, committee are in this area, it will be considered and the Health Records Survey— were considered. first. The purposes of and current activities in special disease registries, record linkage, and genetic uses of linked vital records were reviewed. FINDINGS AND RECOMMENDATIONS The subcommittee did not deal with im- portant issues of confidentiality of vital and Access to Vital and Health health records and protection of the privacy of Records individuals. While these issues underlie any use of vital records for epidemiologic purposes and The problem of access to vital records has might be important in the implementation of become acute with the revolutionary changes in several of its recommendations, the subcom- data processing techniques already referred to. mittee did not consider itself constituted with The potential for record utilization has been ex- the legal and other representation that would be panded almost beyond limit, provided the records required for a meaningful discussion of the issues. and their storage are in such forms as permit It should ke pointed out, however, that legal and machine access. Aggravating the problem is the procedural safeguards against the misuse of vital fragmentation of the U.S. vital record system record information have been part of the operation into more than 50 politically, geographically, and of Federal and State statistical systems for many administratively independent systems. The effects years, and that these safeguards have been phe- of this fragmentation are especially evident in nomenally successful. The distinction between studies th~ require that people be followed over

3 long periods of time, as is usually the case in the basic records themselves should already be investigations of the etiology of chronic diseases. in process of modification so that (a) record Two particularly urgent needs are apparent: linkage for the purpose of specific short-term (a) methods that will improve our ability to “link” studies, whether performed clerically or elec- records pertaining to the same person or family, tronically, is facilitated and improved in ac- and (b) facilities to improve our ability to trace curacy, and (b) record systems are prepared for death records of persons enrolled in investiga- the incorporation of routine machine linkage tions. when the time comes. Perhaps the most im- portant single component of this modification Record Linkage would be the acceptance and wider use of the The potential of the information to be ob- concept of a unique number to identify each in- tained by linkage of records is obvious. The dividual in a population. linkage of records pertaining to an individual allows the correlation of events and circumstances of his life extending over the long periods as- “The Unique Number sociated with the incubation and development of The idea that each individual would have a chronic disease. The linkage of records per- unique number and that this number would identify taining to members of the same family allows all his health (and other) records is an old one, studies of the genetic determinants of disease as The advantages of such a procedure for record well as the familial environment. No one seriously linkage are obvious. The widening use of the questions the premise that the linkage of health Social Security number in the United States puts records of as many varieties as possible would the concept within the realm of feasibility for be of great value for patient care and adminis- a large proportion of the population, perhaps for trative purposes as well as for the purposes of the first time. Several other numbering systems etiologic investigations. have been proposed, and even tried. A unique The practical aspects of the record linkage birth certificate numbering system, begun in problem are less clear. A number of experi- 1947, has subsequently been dropped in many mental record linkage systems have been set up areas, presumably because of lack of use. in recent years, and a bibliography of relevant The wider the use of a particular number, publications is given in Appendix I. Basic ques- the greater is the amount of information that can tions still at issue are (a) what identifying in- be linked. At the same time, the more frequently formation most accurately allows linkage of an individual is asked for that number the more records of the same individual or family, (b) what likely he is to remember it correctly or have it records can be linked with the greatest profit in available, which again encourages its use for terms of derived information, and (c) ‘what is the other purposes. Thus, there is a circular effect most appropriate population size to ke covered. operating-in favor of the more widespread number- Unfortunately, the answers to these questions are ing system. This consideration argues over- not generalizable to all times or all places, and whelmingly for the adoption of the Social Security although some of them have keen answered for number as the unique identifying number for investigations of particular problems in particular health record purposes. If there was doubt about places, it appears that available information is this previously, it was dispelled by the imple- insufficient to recommend inauguration of any mentation of the Medicare program, in which the particular record linkage system in the United Social Security number is the basic identifying States as part of routine record procedures. device. Two general recommendations do seem ap- The Social Security number is not ideal, A propriate however. First, exploration of individ- proportion of the population still has no number; ual and family record linkage should continue to a person may, by request, have more than one be pressed as a high priority research area to number; numbers are not usually assigned until identify which linkages are likely to be most prof- wage earning begins; the lack of any apparent itable and which methods most efficient. Second, pattern to the assigned number (i.e., a person

4 cannot directly relate his number to his birth the date and place of his death are known with date or any other readily remembered set of accuracy. The registration system is centralized figures) makes for frequent errors in recording only with respect to certain advisory functions and transcribing. Some of these problems will and the preparation of national statistical tabu- decrease as time passes and use of the numbers lations. After the national statistical tabulations spreads. Others may be mitigated by deliberate have been prepared, the centralized copies of the efforts. basic records are discarded. To locate the death In the latter context, for example, it is noted certificate of a person whose place of death is not that NCHS and representatives of State Health known might involve separate searches in more Departments are planning to confer with the than 50 different States and cities. There is con- Social Security Administration (SSA) to explore siderable variation across the country in the the possibility that a number assigned to an in- ease with which searches of these archives can dividual at birth by the State Health Department be conducted. In addition, there is an under- (from a list previously supplied by SSA) could standable reluctance on the part of some Regis- later become the Social Security number when trars to undertake searches for certificates the individual makes application for assignment which may not exist in their archives. It should of a number. be noted that, in some jurisdictions, while the Hospitals should be asked to incorporate cost of the search is borne by the Registrar’s Social Security numbers as part of their basic office, the reimbursement goes directly into the identifying information. (Presumably, this will State treasury and does not compensate the Reg- in any event be necessary for Medicare patient s.) istrar’s office for the additional labor involved. They should be asked to explore the feasibility As only one example of the type of problem of the use of the Social Security number in place that this fragmentation can provoke, it may be of their own record numbering system. This pointed out that in a current study of the mortality would greatly improve access to records of a among steelworkers undertaken by the National particular patient identified by means of his Cancer Institute, more than 30 clerical steps are Social Security number in some other context. involved in tracing the subjects from employment The U.S. Standard Certificate of Death now to death. In addition to the size of the task in- contains a place for the Social Security number. volved in tracing deaths, there is also a problem The States are urged to retain this item in their with completeness, since it is, in practice, never own certificates and to press for its completion practical to search for untraced cases in all 50 in all possible instances. Whether or not it sub- jurisdictions. Searches are usually restricted to sequently becomes possible to assign a “Social two or three States with the highest rates of im- Security number !! at birth, it would be of great migration from the area of the study. value to incorporate Social Security numbers of The idea of a National Death Index located both parents on the Certificate of Live Birth. in the National Center for Health Statistics has Family record linkages would be enormously been discussed, more or less casually, from facilitated by this procedure. time to time in the past. The index would permit identification of the fact, place and date of death, Ascertainment of Death and, probably, State death certificate number. Access to information on the death certificates U.S. death certificates contain a considerable themselves would, as now, be through individual amount of information regarding the decedent, and State registries. the presumed cause of his death. The quality of It is believed that the mounting volume of the information is high and the completeness of studies of the chronic diseases alone makes the registration is excellent. These certificates con- establishment of such as index an urgent ne- stitute a most valuable source of information for cessity. There are, in addition, needs for such epidemiologic studies, an index in connection with Medicare, and other The major problem arises in locating the medical care and commercial purposes. The cost death certificate of a specific individual unless of the index could be justified entirely by its

5 epidemiologic uses, but the additional medical incomplete, or when pathologic end points short care and commercial uses ‘appear to eliminate of death are being studied, it is usually necessary cost as a serious deterrent. to trace individuals over periods of years or In epidemiology, the index would be used decades. primarily for studies of prognosis or cause of In this country, in the presence of con- death among persons having particular diseases siderable migration and the absence of cent inuous or undergoing particular exposure, for example: population registration, a national health service or other national identification system, such a 1. Selected occupations, e.g., steelworkers, followup can be extremely difficult. Nevertheless, the rubber industry, hard-rock and ura- there are numerous resources and methods that nium miners, smelters, etc. can be utilized. Many workers appear to be un- 2. Professional societies whose members aware of the and potential of these re- have unusual laboratory exposure, e.g., sources. In connection with its own research American Chemical Society, American program the Division of Radiological Health Society of Immunologists. (DRH), USPHS, prepared in 1962 a roster of followup facilities.q More than 30 sources are 3. Groups undergoing health status evalua- described, together with the legal basis on which tions (as predictors of longevity or causes they may be utilized and the usual procedures of death) e.g., participants in the National involved. It is believed that it would be most Health Survey, Kaiser-Permanente mul- useful to make such a document readily available tiphasic examinations, studies of infant through publication, and it is recommended that development and health status, executive NCHS arrange for the preparation and publication health (preventive maintenance) pro- of a revised version of the DRH manuscript or a grams, and the American Cancer Society’s newly prepared document with similar intent. study of more than 1 million persons. Costs of various procedures should be included. 4. Other suspected special risk groups, e.g., children who received SV 40 or other Information on Vital and Health contaminated vaccines or drugs, persons Records blood exposed to ionizing radiation, Having in mind the undesirability of burden- cases of accidental poisoning, donors, ing routinely prepared documents with information etc. required only for ad hac and limited purposes, 5. Members of special disease registries— there are few recommendations to be made re- cancer, mental illness, congenital mal- garding the content of existing vital and health formation, records. However, some areas of concern were identified, It is therefore recommended, that a Nationul Death Index be established. Problems that re- Cause of Death quire detailed technical study include the amount The present standard certificate of death and nature of information that will be required for restricts information on pathologic conditions accurate identification, cost of the index, and the existing in the decedent to those which the certi- extent to which this cost can be met by com- fying physician considered as contributing to the mercial uses. death. In fact, however, with many chronic condi - Followup in some studies it is sufficient to identify 3Staff of the Cooperative Thyrotoxicosis Therapy Follow. the dead members of the study group--members Up Study: Resources for Locating Patients. Division of Radio. not so identified being assumed to be alive. How- logical Health, U.S. Public Health Service, 1962. (mimeo. ever, when death ascertainment is known to be graphed) tions it may be impossible to determine in individ- specific proposals for the next revision of the ual cases whether or not an existent condition standard certificate. contributed to the death; such a determination may be possible only in terms of statistical Congenital Malformations probability based on comparison of mortality As other causes decline, the congenital mal- rates in groups of persons with and without the formations become responsible for an increasing condition, In many instances, therefore, selection proportion of mortality and morbidity in infancy of pathologic conditions to appear on the certifi- and childhood. Because certain forms of mal- cate is arbitrary and inconsistent. A memoran- formation have been clearly shown to be pre- dum on this matter is reproduced in Appendix II ventable, and because of the relatively short as a basis for discussion. interval between cause and effect, this category The matter, while clearly a theoretical of diseases is a prime target for investigative problem, did not appear of great practical mo- effort. Birth certificates can provide a major ment so long as statistical tabulations, and, resource for this effort by providing access to generally speaking, access to certificates, (1) very large numbers of cases of common mal- were on the basis of a single pathologic condition formations, (2) large numbers of cases of com- selected as the “underlying” cause. This single mon malformations with characteristics of partic- selection made quite obvious the arbitrary nature ular value, e.g., being a twin. (3) series of of the information presented and the necessity cases of malformations of such rarity that series to restrict the use of diagnostic information to cannot be assembled from clinical sources, and pathologic conditions likely to be selected as the (4) population-based series that can be used to underlying cause if they were present. investigate geographic and temporal clustering. However, NCHS is now committed to early Birth certificates can never be expected to initiation of a program of coding and tabulating provide a complete ascertainment of malfor- multiple causes of death listed on individual mations—partly because of the impossibility of certificates. It then becomes much more crucial defining a malformation. However, descriptive which conditions are included on and which omitted features make it quite evident that different from the certificate, From two points of view-- etiologic factors are operative in different cate- (a) for investigating in statistical terms which gories of malformation, and, in some instances, conditions do in fact contribute to death and in within a single diagnostic category of malfor- what degree, and (b) improving the death certifi- mation. Birth certificates will be a useful re- cate as a general research tool— it seems im- source even if their information is limited to portant to shift the emphasis in certification from ‘those major, externally evident, anomalies which conditions supposedly contributing to death to could be expected to be fully reported in a conditions present at the time of death, or of reasonably competent system. which residua are present at the time of death. It is to be hoped, therefore, that all States How such additional information might best will retain the malformation item proposed on be collected has not been evaluated. Thus, in the standard live birth certificate, and make designing a possible Part III of the cause of efforts to improve reporting of this item. With death question on the death certificate, seeking respect to the maintenance of surveillance of information on other significant conditions pres- malformation reporting, in order to detect geo- ent, .it is not clear whether more complete graphic or temporal clustering at an early stage, certification would be obtained by means of an it is recommended that a comparison k made be- open- ended question or by a checklist of condi- tween reporting on birth certificates and systems, tions such as diabetes, hypertension, congenital such as have been set up in New Jersey and malformation, and so on. Such questions may be British Columbia, that are based on reporting approachable through follow- back studies start - independent of the birth certificate. ing from current certificates, and it is suggested It is further recommended that consideration that such studies be undertaken as a basis for be given to the possibility of examination of a

7 nationally selected sample of newborn infants Development (including late fetal deaths) as one of the cycles of Other Resources of the Health Examination Survey. With respect to congenital malformation specifically, such a It seems worthwhile to identify two resources survey would provide estimates of the prevalence that do not appear to have been utilized to their of congenital malformations detectable at birth— fullest potential for epidemiologic investigation, heretofore unavailable except with respect to although there are no specific recommendations the patients of a few highly specialized hospitals— as to how their utilization might be increased, and a standard against which to evaluate other The first consists of the records of deaths means of estimating malformation rates, Because coming under the purview of the medical ex- of the special problems of conducting examina- aminer. The medical examiners’ investigations tions immediately after birth, such a survey of such deaths are detailed and the records are would require different procedures public. Medical examiners in general appear to from those that have been used so far in the be favorably disposed to the utilization of the Health Examination Survey. However, sampling records for scientific investigation, and, in some of maternity facilities and of blocks of time instances, have initiated epidemiologic investi- within such facilities (rather than direct sampling gations based on their material. The records of individuals born) would seem to provide a offer extensive possibilities for analysis alone, relatively simple method for assembling a repre- or as the starting point for investigations in which sentative national sample of births. additional information is sought. Other Items The second resource consists of the records of health insurance schemes. The value of such Level of education of parents has replaced records is demonstrated by the series of reports occupation on the standard certificates of live based on the experience of the Health Insurance birth and fetal death. The replacement results Plan (HIP) of Greater New York. While it is true from the difficulty of interpreting and coding that HIP incorporates medical care as well as the occupation item, and belief that, as an index an insurance program, essentially similar rec- of social class (which is the main purpose of the ords are usually required as the basis for pay- occupation item), education is equally valid, In ment even when the medical service is not per- view of the very limited use of occupation as formed by the insuring company. Remarkably recorded on death certificates, an attempt might little use has been made, for example, of the also be made to collect education data on death voluminous material that must exist in Blue certificates in selected areas. The greater dif - Cross- Blue Shield plans relevant to studies of ficulty of obtaining education data on decedents, the familial clustering of disease, the association as contrasted with parents, is recognized, and in of diseases in individuals, and ‘the natural history the evaluation phase the item should probably be of chronic diseases. sought as a supplement to, rather than replace- Lastly, in connection with the major pro- ment for, information on occupation. grams of medical care now being initiated— In many studies of the effects of social and notably Medicare, the Title 19 provisions of psychologic stress it would be advantageous to Medicare as they relate to children, and the have the names (rather than simply the number) heart-cancer-stroke and community mental health of the children as an item on the certificate of centers—it is hoped that adequate statistical and divorce. It is understood, however, that there is epidemiologic consultation will be obtained at a reluctance on the part of Registrars to press this early stage to ensure the ultimate usefulness for this item. of the medical records. SUMMARY OF RECOMMENDATIONS on the amount and nature of the infor- mation required to identify decedents, 1, Linkage of various vital and health rec- and the computer technology most ap- ords should continue to be explored as a propriate to such an extensive operation. high priority research area, to identify It should be stressed that an index is which linkages are likely to be most meant, not a repository for the death cer- profitable and which methods most ef- tificates themselves. ficient. 6. A document should be prepared and pub- 2. The Social Security number should be lished for the assistance of investigators accepted as the most practical numeri- in followup studies, along the lines of that cal identification of individuals, and in- prepared for their own staff by the Divi- corporated into all vital and health rec- sion of Radiological Health. Sources of ords, Where possible, the Social Security followup, legal bases, usual procedures, number should become the actual identi- and costs should be included. fication number for the specific record of the individual. Hospitals should be asked 7. The basis for inclusion of pathologic con- to consider this last possibility. ditions on the death certificate may need to be extended beyond the present re- 3. NCHS, representatives of State Health striction to conditions presumed to have Departments, and the Social Security contributed to the death. Information might Administration should explore the possi- be sought on all significant conditions bility of assigning a Social Security num- present at the time of death, or of which ber to dn individual at birth. This recom- significant residua are present at the mendation supposes that the Social Se- time of death. The best way of obtaining curit y Administration is itself not already this information should be sought in considering the use of a birth numbering follow- back studies based on current system for its own purposes. certificates in preparation for the next 4. Social Security numbers of parents should revision of the standard certificate. be added to the certificates of live birth and fetal death. 8. Efforts should be made to extend, im- prove, and utilize information on con- 5. Because of the great epidemiologic use- genital malformations reported on vital fulness of a National Death Index, NCHS records, particularly the birth certificate. should explore with some urgency the technical problems involved in the es- 9. Consideration should be given to the tablishment of such a resource. Apart possibility of examination of a national from estimates of cost and the extent to sample of newborn infants (including fetal which this would be mitigated by com- deaths) as one of the cycles of the Health mercial use, there is need for information Examination Survey.

00

9 APPENDIX I

SELECTED BIBLIOGRAPHY ON RECORD LINKAGE

Acheson, E. D.: The oxford Record Linkage Study, Report Dunn, H. L.: A national identity registration system to on the Second Year’s Operations. Oxford Regional Hospital synthesize . Estadiatica, Journal of the Inter Board. Oxford, England 1963. American Statistical htstitute 11:605, Sept. 1953. Acheson, E. D.: The Oxford Record Linkage Study, a re- Harris, R.: Letter to Dr. Councell dated Mar. 8, 1965. view of the method with some preliminary results. F’rvc.Roy. Kennedy, J. M.: Linkage of birth and marriage records Soc.Med. 57:11, 1964. using a digital computer. Atomic Energy of Canada Ltd., Report No. 1258, Chalk River, Ontario, 1961. p. 18. Acheson, E. D.: The Oxford Record Linkage Study, a cen- tral file of morbidity. &it. J, Preo. &SocW Med. 18:8, Jan. 1964. Kennedy, J. M.: The use of a digital computer for record linkage. Proceeding of the Seminar on the Uee of Vital and Acheson, E. D., and Evans, J. G.: The Oxford Record Link- Health Statistics for Genetic and Radiation Studies, 1962, age Study. Biometrics 2:367, 1963. pp. 155-159. Acheson, E. D., Truelove, S. C., and Witts, L. J.: National Kennedy, J. M., Newcombe, H. B., Okazaki, E. A., and epidemiology. Brit.M.J. 1:668, 1961. Smith, M. E.: List processing methode for organizing filee of Bahn, A. K.: Methodological study of population of oub linked records. Atomic Energy of Canada Ltd., Report No. 2078, patient psychiatric clinics, Maryland 1958-59. Public Health 1964. Monograph No. 65. PHS Pub. No. 821. Washington, D.C., 1961. Kjelsberg, M., and Metzner, H.: Record system used in the Bahn, A. K.: The development of an effective statistical Tecumseh Community Health Study. Personal communication. system in mental illness. Arn.J.Psychiat. 116:798, 1960. Marshall, J. T.: Canada’s national vital statistics index. Bahn, A. K.: Person by electronic methods. Population Strtdies 1(2):204, Sept. 1947. (Printed in Great Communications of the Association for Computing Machinery Britain.) 5:404, 1962. Masi, A. T., Sartwell, P. E., and Shulman, L. E.: The use Babn, .4. K., and Bahn, R.: Considerationsin using social of record linkage to determine familial occurrence of diseaee security numbers on birth certificates for research purpoees. fromhospital records (Haehimoto’s disease). Am.J.Pub.Health

Pub. HeaZth Rep. 79:937, 1964. 54(11): 1887-1894, NOV. 1964. National Office of Vital Statistics: Some aspects of vital McKeown, T. and others:. Session VII, panel discussion, statistics registration in Canada, by A. E. .Bailey. Vital session on epidemiological studiee. Second international Con- Statiez!ice-SpeciaZ Rep&k Vol. 23, No. 22. Public Health ference on Congenital Malformations. The International Medical Service. Washington, D.C., 1947. pp. 252-61. Congress, Ltd., New York, 1964. pp. 341-351. National Center for Health Statistics: Summary report on Minet, P. L.: Fertilitd pr~coce d’une cohorte de mariages Tennessee’s “Record of Events,” by M, R, Baird, and E. H. clans une province canadienne. Biology Branch. Atomio Energy Halpin. The Public Health Conference on Records and StG- of Canada Ltd. Basel 14:186-196. Chalk River, Ontario, 1964. tie tics, PHS Dec. No. 473. Public Health Service. Washing- National Center for Health Statistics: “Studygroup on record ton, D.C., Dec. 1959 linkage. The Public Health Conference on Records and Statis- Central Bureau of Statistics: Population changes 1961. tics, PHS Dec. No. 603. Public Health Service. Washington, Official Statistics of Sweden (Population and Vital Statis- U.S. Government Printing Office, Nov. 1964. tics) Stockholm, 1963. National Centsr for Health ”Statistics: Planning session for Davidson, L.: Retrieval of mispelled names in an airline study group on record linkage. The PUWC Health Conference passenger records system. Communications of the i4ssoc& on Records and Statistics, PHS Dec. No. 603.1. Public Health ation for Compu$ing Machiney. 5:169, 1962. Service. Washington. U.S. Government Printing Office, Jan. Dunn, H. L.: Record linkage. Am. J. Pwb.Heatth. 36:1412, 1965. 1946. Newcombe, H. B.: Detection of genetic trends in public Newcombe, H. B., and Kennedy, J. h!.: Demographic analy- health. E/feet of Radiation on Human He?edity. World Health sis and computer programs. United Nations World Population Organization, Geneva, 1957. pp. 157-168. Conference, Belgrade, Yugoslavia, Aug. 30 to Sept. 10, 1965. Newcombe, H. B.: Record linkage as a means of studying Newcombe, H. B., Kennedy, J. M., Axford, S. J., and James the somatic effects of low doses of radiation. Science 130: A. P.: Automatic linkage of vital records. Science 130(3381): 994.959, 1959. 954-959, Oct. 1959. Newcombe, H. B.: Population genetics, population records. Newcombe, H. B., and Rbynas, P. O. W.: The cost of in- hlethodology in Human Genetics. Holden-Day, Inc., 1962. pp. dividual follow-up studies of large populations. Proceedings 92.113. of the International Population Conference- Paper No. 16. (Re- Newcombe H. B.: Untapped knowledge of human populations. printed as Atomic Ener~ of Canada Ltd., Report No. 1255, Transactions of the Royal Society of Canada. Vol. LVI, Series Sept. 1961. p. 8.) III, Section 111, June 1962. pp. 173-180. Newcombe, H. B., and Rhynas, P.O. W,: Family linkage Newcombe, H. B.: Risk of fetal death to mothers of differ- ofpopulation records. On the Use of VitaJrznd Health Statis- wrt ABO and RH blood types. Am. J. Human Genet. 15(4): tics fo? Genetic and Radiation St&ies. United Nations Pub., 449-464, Dec. 1963. Sales No. 61. XVII(8) :135-154, 1962. Newcornbe, H. B.: Pedigrees for population studies, a pro- Newcombe, H. B., and Rhynas, P.O. W.: Child spacing gress report. Cold Spring Harbor Symposia on Quantitative following stillbirth and infant death. Eugenics QuarteT?y 9(l): Biology, 29:21, 1964. ‘ 25-35, Mar. 1962. Newcombe, H. B.: Panel discussion, session on epidemio- Newcombe, H. B., and Tavendale, O. G.: Effectsof father’s logical studies. Second International Conference on Congenital ageontherisk ofchildhandicap or death. -4m.J.Human@~e~. Malformations. The International Medical Congress, Ltd., New 17(2):163-178, 1965. York, 1964. pp. 345-349. National Office of Vital Statistics: A coordinated system Newcombe, H. B.: Environmental versus genetic interpre- of vital records and statistics, the need for coordination, by tations of birth.order effects. Eugenios Quarterly, Nov. 1964. H. L. Dunn. Public Health Service. Washington, D. C., JUly p. 36. 1953. Nmvcombe, H. B.: Screening for effects of maternal age Phillips, W., Jr., and Bahn, A. K.: Experience with corn- and birth order, in a register of handicapped children. Ann. puter matcbingby names. Proceedings of the Som.al Statis- Human Genet. 27:367-382. London, 1964. tics Sections, American Statistical Asso&ation, 1963. Newcombe, H. B.: Use of vital . United Nations Phillips, W., Jr., Bahn, A. K., and Miyasaki, M.: Person- World Population Conference, Belgrade, Yugoslavia, Aug. 30 matching hyelectronic metbods. Communications of the Asso- to Sept. 10, 1965. ciation foT Computing Machiney 5:404, 1962. Newcombe, H. B., Axford, S. J., and James, A. P.: A plan Phillips, W., Jr., and Gorwitz, K., and Bahn, A. K.: Elec- for the study of fertility of relatives of children euffering from tronic maintenance of case registers. Pub. Health Rep. 77: hereditary rind other defects. Atomic Energy of Canada Ltd., 503, 1962. Report No. 511, Nov. 1957. p. 50. Schwartz, E. E.: Some observations on tbe Canadian fami- Newcombe, H. B., James, A. P., and Axford, S. J.: Family ly allowances program. The Social Serv4ce Review Vol. XX, linkage of vi tef and health recorde. Atomic Energy of Canada No. 4. Dec. 1946. Ltd., Report No, 470. Chalk River, Ontario, July 1957. (Re- Statistical Commission: Twelfth session, Item 9 of the printed Aug. 1961 and May 1964.) provisional agenda E/CN/3/293. United Nations Economic Newcombe, H. B., and Kennedy, J. M.: Record linkage, and SociaJ Council. Feb. 7, 1962. p. 38. reeking maximum use of tbe discriminating power of identify- Vinge, Margit, Head, Documentation Unit, Central Bureau ing i~formation. Communications of the Association for Com- of Statistics, Stockholm. Letter dated Mar. 9, 1965, tn Louis puting Machinery 5(11):563-566,1962. R. Stolcis with attachments.

000

11 APPENDIX II

THE UNDERLYING CAUSE CONCEPT

At the November 1961 Geneva meeting of the While it was never sound to use mortality data Subcommittee on Classification of Diseases of as an estimate of morbidity, this was commonly the Expert Committee on Health Statistics, it was done in the past. Such a use is no longer defen- suggested that the “concept of underlying cause sible. The function of the death certificate is still might be reexamined to see if a more satisfactory that of providing information on those diseases conceptual basis for classifying causes of death that cause death. These data augment those ob- could not be developed.” The concept can be re- tained on morbidity by supplying data for dis- viewed in the light of uses made of information eases that cause death without previous signs or obtained on death certificates, and from the symptoms; by evaluating the severity of disease viewpoint of the data available for decedents. through the fact that it has caused “premature” No matter how the medical certification is de- death; by supplying counts for diseases, that, signed, its source is the existing observations owing to their rare occurrence, or occurrence on the decedent, and its statistical function is to in a limited population (say, under 1 month of serve public health and medical needs. age) are not easily reached through survey Cause of death information is now used: (1) methods. In all of these cases, the concept of an to describe the most important current health underlying cause of death can provide valid problems; (2) to record of certain information, But if disease incidence were viewed rare, fatal diseases; (3) to serve as a starting as a continuum of information beginning with the point of epidemiologic studies of specific popu- ~ccurrence of signs and symptoms in the living lations characterized by the disease to which the population, followed by disabling illness and death was assigned; and (4) to assess the fre- finally, death, then the information collected on quency of conditions associated with death, but the death certificate should be restructured 10 which are not the direct cause. of death. parallel the “prevalence” concept used in the Does the underlying cause of death supply collection of morbidity data. The underlying cause data that are appropriate to these uses? If not, concept cannot provide prevalence data, or the what questions on the medical certification would counts needed for case fatality data. elicit a more valid response for these appli- The second use, to record the frequency of cations? Are there other uses not nov served rare disepses with high fatality rates is clearly that could be met with a change in the kind of met by the underlying cause concept. There should information requested? not be any difficulty in collecting such information The first use, to describe the most important under other definitions of the cause of death, current health problems, has changed, at least in The third use, as a starting point of epi- the United States, in the last decade. The National demiologic studies, is now practically limited to Health Survey can now provide data on the major studies of diseases selected as the underlying illnesses found among the living population. cause of death. Many investigators would pre-

12 fer to follow every occurrence of a disease in purposes, the requirements must be phrased to the population of decedents rather than thepopu- fit into the way death certificates are completed, lation of persons whose deaths were assigned to and what is known about decedents. The medical this disease. The larger population can be ap- and social information at hand for a decedent proached by coding all the diseases reported on may vary from that assembled for the person under the death certificate. The mention of these dis- systematic lifelong medical care to the one-time eases will still be limited to the physician’s in- posthumous observation made by the medical ex- terpretation of the response required by the aminer of an unidentified person. For the former present certificate. A thoughtful certifier may individual there may be a record of his physical select only those diseases related to the death, condition at regular intervals, each illness epi- while another physician may list all serious sode, his last illness, and perhaps an autopsy diseases present at death. If the epidemiologist protocol. On the basis of this information, what wishes to know for how many persons a partic- should be recorded on the death certificate? ular disease contributed to death, it may be How can the questions os instructions in the medi- possible to obtain the information from the cur- cal certification be phrased to elicit the data rent certificate when all conditions reported on needed? Can these questions be so phrased to the record are coded. If the epidemiologist needs serve also for the death in w!lich little information information on how many persons died with a dis- is known? ease, this count cannot be obtained with the pres- What approach should be taken to psrmit ent form of the medical certification. separation of data by qualit y—that based on com- The fourth use, to assess frequency of con- plete histories; on attending physician records ditions associated with death but not the direct for a terminal illness only, on autopsy only, etc? cause of death, has been cited as the chief rea- Some exploration of quality of diagnostic infor- son for coding all the conditions reported on the mation and its effect on medical certification has death certificate, rather than the underlying been reported in special studies, such as one by cause alone. As pointed out in the previous para- L M. Moriyama and others. 4 Thus far, there has graph, the wording on the present form does not not been any consideration of establishing a rou- elicit a report of all conditions present at death, tine collection of information on the quality of not even all serious conditions. Response to medical certification. Is it possible or desirable specific queries concerning diseases present at to make such measures part of the collection death has demonstrated that the entries on the system for m.,rtality statistics? certificate represent the physician’s judgment or If it can be decided what data should be col- understanding of the nature of the report ex- lected through the medical certifications on death pected from him. The form must be redesigned records, it m~.y become easily evident as to if conditions present at death are needed. whether these data can be obtained through the In summary, the underlying cause concept underlying cause approach, or through a new cannot generally provide prevalence or incidence concept. A new concept, and the questions or data. It can provide counts for diseases causing instructions needed to produce the desired an- death, or contributing to the event of death. Where swers would then need to be tested. An ex- there is more than one disease contributing to periment would need to be devised to learn the death, the present wording of the certificate whether or not the new form is producing the does not encourage a complete response. Where expected data. there is a serious disease present that is un- related to the sequence of events resulting in death, the certificate specifically discourages 4Moriyama,L M., Dawber, T. R., and Kannel, W. B.: Eval- such an entry. uation of Diagnostic Information Supporting Medical Certifi- Once a decision is reached on what data are cation of Deaths from Cardiovascular Diseases. National Can- cer fnstitute Monograph No. 19, Jan. 1966. pp. 405-419. needed about death for medical and public health

13 * U. S. GOVERNMENT PRINTUJG OFFICE : 1973 5 15-214/70 OUTLINE OF REPORT SERIES FOR VITAL AND HEALTH Statistics

Public Health Service Publication No. 1000

Sei’ies 1. l+og~ams and collection procedu~es.— Reports which describe the general programs of the National Center for Health Statistics and its offices and divisions, methods used, definitions, and other material necessary for understanding the data.

SeVies 2. Data evaluation and methods reseavch. —Studies of new statistical methodology including: experi- mental tests of new survey methods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, contributions to ,.

Series 3. Analytical studies. — Reports presenting analytical or interpretive studies based on vital and health statistics, carrying the analysis further than the expository types of reports in the other series.

Series 4. Documents and committee ~epovts. —Final reports of major committees concerned with vital and health statistics, and documents such as recommended model vital registration laws and revised b~h and death certificates.

Series 10. Data fvom the Health Intemiew Survey. —Statistics on illness, accidental injuries, disability, use of hospital, medical, dental, and other services, and other health-related topics, based on data collected in a continuing national household interview survey.

SeYies 11. Data fyom the Health Examination Suvvey. —Data from direct examination, testing, and measure- ment of national samples of the population provide the basis for two types of reports: (1) estimates of the medically defined prevalence of specific diseases in the United States and the distributions of the population with respect to physical, physiological, and psychological characteristics; and (2) analysis of relationships among the various measurements without reference to an explicit finite universe of persons.

Series 12. Data fvom the Institutional Population Surveys. —Statistics relating to the health characteristics of persons in institutions, and on medical, nursing, and personal care received, based on national samples of establishments providing these services and samples of the residents or patients.

SeVies 13. Data from the Hospital DischaYge .Swvey. —Statistics relating to discharged patients in short-stay hospitals, based on a sample of patient records in a national sample of hospitals.

Series 20. Data on mortality. —Various statistics on mortality other than as included in annual or monthly reports— special analyses by cause of death, age, and other demographic variables, also geographic and analyses.

Sevies 21. Data on natality, marriage, anddivoYce. -Various statistics on natality, marriage, and divorce other than as included in annual or monthly reports—special analyses by demographic variables, also geographic and time series analyses, studies of fertility.

SeTies 22. Data fyom the National Natality and Mortality surveys. —Statistics on characteristics of births and deaths not available from the vital records, based on sample surveys stemming from these records, including such topics as mortality by socioeconomic class, medical experience in the last year of life, characteristics of pregnancy, etc.

For a list of titles of reports published in these series, write to: office of Information National Center for Health Statistics U.S. Public Health Service Washington, D.C. 20201 $ I