In recent years, medical record review has been used to alert patients who have received drugs or treatments that have newly discovered side-effects. The experience of the national cooperative Diethylstilbestrol-Adenosis (DESAD) Project in identifying and notifying women exposed in utero to diethylstilbestrol (DES) shows this to be a difficult task. In order to identify 4,830 exposed women, 221,245 charts were reviewed. Detailed tracing data for one of the centers participating in the DESAD Project indicated that only 85 per cent of the 690 DES-exposed women identified at that center could be notified of exposure. The DESAD Project experience has led to recommendations for standardized prenatal records and drug lists, long-term storage of medical records, new legal guidelines, and improved recording of follow-up information, taking into account issues of privacy.
Medical chart review has long provided a major source of information for retrospective epidemiologic studies. In recent years chart review has also been used to identify and alert patients who received drugs or treatments that have newly discovered side effects. For example, attempts have been made to alert adults to their increased risk of thyroid cancer following head and neck irradiation in childhood and to alert women to their increased risk of clear cell adenocarcinoma of the lower genital tract following in utero exposure to diethystilbestrol (DES). In the case of DES, the National Cancer Institute in the 1978 Task Force Summary Report recommended that all women who received DES in pregnancy, as well as exposed daughters and sons, be informed of their exposure by a public information campaign, and notification of exposed individuals by their physicians. Although the cost of a public information campaign can be estimated from other such endeavors, limited information exists on the cost or problems involved in identifying a population from medical records and notifying these individuals of a hazardous exposure. The experience of the Diethylstilbestrol-Adenosis (DESAD) Project in obtaining access to records for review and in locating study participants gives insight into the difficulties of this task.
Synthetic estrogens including diethylstilbestrol (DES) were used in some pregnancies between the years 1940 and 1971. Exposure of the fetus to DES has now been associated with genital tract cancer in daughters and with benign genital tract anomalies in both daughters and sons. To date there is little evidence of risk for mothers who ingested the drug. The DESAD Project is an observational follow-up study of the daughters exposed in utero to DES. The study design has been reported previously. Participants enrolled at these institutions gained entry by referral from physicians, by self-referral, or through the process of prenatal record review carried out at each DESAD center for the purpose of identifying DES-exposed participants and unexposed controls. It is the prenatal record review group that is the focus of this paper.
In most communities, obstetrical records are maintained by the private physician, with labor and delivery records in separate hospital files. Prior to the beginning of the DESAD Project, physicians in the Boston, Houston, and Los Angeles DESAD centers had been examining DESexposed daughters referred by private physicians. Through these contacts and through knowledge of their local medical communities, Project physicians were able to locate private physicians or clinics known to have administered DES to pregnant women and obtain permission to review office records. Some records had been in storage due to the physician’s retirement or death. In these private practices, because records of both obstetrical and gynecological patients were combined, all charts required review. At the Mayo and Gundersen Clinics, the prenatal and obstetrical records were combined in a unit record and were retrievable through a diagnostic index. Thus, all obstetrical records could be identified and reviewed without reading through the gynecological records.
Exposure was defined as any chart notation of DES or synthetic estrogens prescribed between conception and the week before delivery. Centers reviewed all available obstetrical records covering the years 1940, when the first DES-exposed birth could occur, through 1972, the year after the US Food and Drug Administration warned against DES for use in pregnancy.
Record abstractors were trained in obstetrical terminology; sex hormone identification was facilitated by the generic and drug trade name list prepared by the Coordinating Center and used during record review. Abstractors were reminded of the confidential nature of the chart contents and the importance of correctly ascertaining exposure status (that is, DES-exposed or unexposed) for each live birth.
The training period for a record abstractor varied from one day to one week, depending largely on the reviewer’s previous experience with medical records and ability to read the physician’s handwriting. The majority of record abstractors required no more than eight hours of training. The information abstracted from prenatal records by the centers is given in the Appendix.
A variety of techniques similar to those described by Boice, Chung, and Edwards were used by centers to trace and locate those identified. Daughters had to be located by tracing their parents, since usually only sex and date of birth were recorded in the obstetrical chart. Because procedures were especially well documented at the Massachusetts General Hospital (MGH) center with regard to staff time and the number of contacts involved, the step-by-step tracing process as applied in Boston will be presented here.
At the MGH center, every name was first checked in the appropriate telephone directory. If not found, directories of surrounding cities were used to search the greater Boston area. If this failed or if multiple listings precluded positive identification, the name was checked through an occupational directory, alumni office, or out-of-state telephone directory when clues were available from the chart. Next, a search was made for the family’s name through the records of the Massachusetts Registry of Motor Vehicles (MRMV). Computer listings of licensed drivers within the Commonwealth of Massachusetts were reviewed at the Registry’s main office. The alphabetized listings provided name, date of birth, address, and city, as well as license identification number for all current license holders. DESAD personnel began by looking for the mother’s name and birth date, if available. Next, the father’s name was checked. His date of birth was seldom indicated in the mother’s record and this impeded the process of locating those persons with commonly occurring names. For those fathers with commonly occurring names, an attempt was made to match the mother’s name with the father’s name by looking for a common address. If these steps failed, the birth dates under the appropriate last name were scanned to locate a female with the daughter’s date of birth, which was then checked for an address match with the mother’s or father’s name.
When a family was located, a letter was sent to the parents explaining the DESAD Project, stating DES exposure status and inviting the daughter’s participation in the Project. Letters were sent from the delivering physician, if possible, or from the local DESAD center. If a letter was returned by the post office as “unable to forward” and all other methods of tracing were exhausted, the family was considered lost to follow-up.
Similar tracing procedures were used at the other centers. In addition, at the Mayo Clinic, babies were assigned clinic numbers at birth. These numbers are recorded in the mothers’ charts. Thus, it was possible to obtain daughters’ charts where the daughters’ names and occasionally current addresses were recorded. Because daughters’ names were available, birth certificates could be obtained if additional information such as father’s birthdate was needed for tracing.
The three centers using private physicians’ records encountered many physicians who refused to allow their records to be reviewed. Concerns expressed by those physicians were: lack of adequate space in which to conduct the review; patient’s privacy and confidentiality; and possible lawsuits from DES-exposed individuals.
In Boston, Houston, and Los Angeles, a total of 73 physicians and hospitals initially allowed review of their records. Five of these physicians refused permission to contact individuals after the review was completed, while one physician terminated the review without explanation after one-third of the charts were examined. The records of one physician were largely illegible and too incomplete to pursue. Those of another physician were only partially reviewed before he became ill, after which his office was moved and records, including those not yet reviewed, were destroyed.
For all centers, available space in the physician’s offices limited the number of staff able to review records at any one time. In most offices, abstractors were allowed to work with the records only a few hours per week. Thus, the entire reviewing process in each office necessitated many visits.
The time required to review and abstract each chart varied widely among centers (Table 1). The tedious nature of record reviewing cannot be overemphasized. Abstractors at all centers found that frequent rest periods were requiredat least 15 minutes every two hours-and that six hours of reviewing per day was the maximum possible.
DES-Exposed Females Identified
Records reviewed by all centers as of 1979 totaled 221,245 and represent the years 1940 through 1972. Table I presents the numbers of DES-exposed females identified by each center. The proportion of charts with DES-exposed mothers varies greatly among centers because of variations in the obstetrical-gynecological patient mix, and variations in the use of DES by physicians.
Figure 1 presents a histogram showing the number of DES-exposed female live births by year identified among the charts of the six physicians (A-F) whose records were reviewed by MGH personnel. This is consistent with known patterns of DES exposure. 16 The MGH center also wished to estimate the proportion of pregnant women who were prescribed DES. Because of the cost, records of only two physicians were chosen for this special review. For these two physicians, a count was kept of all unexposed live born females between the years 1940 and 1966. Between the years 1946 (when the first DES-exposed female was identified in each practice) and 1966, DES was used by one physician in 8 per cent of female live births and by the other physician in 13 per cent of female live births. Since only live births are represented in these calculations, and since the two physicians administered DES primarily to women with vaginal bleeding, cramping, and threatened miscarriage, these are most likely low estimates of actual use.
During the 1945 to 1966 time period 2 per cent of all deliveries at the Gundersen Clinic were DES-exposed births; 9 per cent of deliveries were DES-exposed during the peak year of use in 1964. At the Mayo Clinic during the years 1943 to 1959, approximately 7 per cent of births were estrogenexposed. Peak usage at the Mayo Clinic reached 19 per cent in 1947.
Table 2 summarizes results of tracing methods used at MGH to locate mothers and daughters through the mothers. For 39 out of 167 families not found through the initial search of local telephone directories, an out-of-state move was indicated in the chart. Tracing through out-of-state telephone books doubled the search time per name. In addition, personnel time and travel expenses were incurred in order to use the out-of-state telephone directories at the New England Telephone Company. Eleven families were traced through other sources such as occupational or university alumni directories.
For the 147 mothers not located by previous methods, a search of MRMV records was instituted. Of these mothers, 31 per cent were classified as “correctly identified,” 19 per cent were classified as “possibly identified.” Overall, 84 per cent of mothers (85 per cent of the daughters) were located at MGH.
Table 3 summarizes the time required for record review and tracking by MGH personnel. The tracking process could not be considered completed upon determination of vital status since, at a minimum, mothers had to be located and informed of their daughter’s exposure. If, in addition to notifying patients of exposure, the patients are to be enrolled in a study, time is also required to negotiate with cooperating physicians and explain the study and issues to prospective participants.
Costs in dollars are not given since these vary both geographically and temporally. Also, because the DESAD Project was designed to locate, enroll, examine and follow DES-exposed daughters, it was difficult to separate location and contact costs. Appropriate local salary and administrative costs can be applied to the time required for record review to give an estimate of the cost of identifying and contacting a similar group having a hazardous exposure.
Tracing those identified is most successful and inexpensive when using telephone directories, but success depends largely on circumstances beyond one’s control such as geographic stability or availability of information in medical records on professional or other affiliations. The MGH study cohort showed exceptional geographic stability, which partially accounted for the success in tracing. Boice, in a retrospective study in Massachusetts, used lists of marriages as well as town lists, but for DES daughters the lists of marriages were not useful because the full names of daughters most often were not known until the mother was located. Also, town lists were not available to study personnel. When tracing by the most obvious method failed, subsequent tracing involved considerably greater costs, often with only marginally increased success. It was common to review a chart where the only tracing information consist ed of the first and last name of the mother and her husband, which virtually precluded locating their daughter by any means. Even well-organized file systems, such as those at the Gundersen Clinic or the Mayo Clinic, occasionally produced names of exposed patients who were impossible to trace. The use of a commercial credit investigation company was considered and rejected because offinancial constraints.
On a nationwide basis, implementing the DES Task Force recommendation that physicians identify and notify DES-exposed individuals from their practices is an undertaking of monumental difficulty. Even at the Mayo and Gundersen Clinics where prenatal and obstetrical records are combined, chart review is exceedingly time consuming. The process becomes even more difficult if the records are on microfilm. Edwards estimated that there were 500 DESexposed women born in the Bluebird Sanitarium, and that it would take one person working full-time two years to identify and locate these women from the microfilmed records.
Streamlining the process by reviewing records selectively-based on year of DES administration, geographical area, or indications for use-may also prove impractical. Heinonen reviewed data on 51,000 pregnancies between 1959 and 1965 in 12 hospitals to estimate nationwide exposure. He found exposure varied between a low of zero at Columbia-Presbyterian Hospital and John Hopkins Hospital and a high of 1.5 per cent of pregnancies at the Boston Lying-In Hospital. However, that study was based on hospital records from pregnancies in the late 1950s, well after the two reports in 1953 that indicated DES had no beneficial effect on pregnancy outcome.
Targeting particular years for review of records, as a less expensive alternative to reviewing all records, may also result in overlooking many exposed persons. Although temporal trends in popularity of DES are evident, even from the somewhat limited experience of the DESAD Project, Mayo Clinic’s greatest use occurred in the late 1940s, while use at Gundersen Clinic peaked in 1964. In Boston, use peaked in the early 1950s, although, two of the practices reviewed by MGH were heavy users of DES in the 1960s.
A review of only hospital records also would not identify the majority of DES-exposed persons. In the experience of three DESAD Project centers, DES was administered most often by private physicians and documented only in the office chart. Usually, hospital and delivery records of women with known DES therapy will not mention prenatal hormones. Thus, it is likely that in most communities, only private office records would contain information on drug exposure.
Finally, identifying women with problem pregnancies as likely recipients of DES may not be effective. DES was one of the first drugs approved by the FDA for marketing under the 1938 Food, Drug and Cosmetic Act. Available by prescription only, it was used in the treatment of menopause, senile vaginitis, juvenile gonorrheal vaginitis, and lactation suppression. As with other drugs, DES was used for non-approved purposes, including treatment of pregnant women, and for a wide variety of indications, including routine administration.
Since the Project’s inception in 1974, DESAD personnel have found access to private physicians’ office records increasingly limited. More charts are being destroyed by retiring physicians in accordance with statute of limitation regulations. This destruction of records may reflect the increased cost of storage or the physician’s wish to limit liability.
However, according to Fay Saber, writing in the Journal of Legal Medicine, physicians would not be held liable for administering DES prior to 1971 since they prescribed the drug in good faith following standard medical practice at that time. Instead, they could be held liable if they do not make some effort to inform women of their increased cancer risk due to the DES exposure. As of Spring 1982, the only widely publicized suit against physicians was Mink vs the University of Chicago. In this case, three women were suing on the grounds that they were used in a controlled clinical trial of DES during the 1950s without their knowledge or consent. Other authors such as Noller, Decker, et al, see notification of exposed women as a moral responsibility of physicians. These two positions do not seem to influence the current availability of records.
Some physicians denied access to their records out of concern for patients’ privacy. Since the purpose of the review was to notify women of a potentially harmful exposure, it is difficult to justify this position. In other situations, where the purpose of the review is to identify patients for a case control study, privacy concerns are more difficult to address. This problem of obtaining permission to review records is also compounded when large institutional review boards are involved. Herrmann, et al, described their experiences with 47 hospitals in the Philadelphia metropolitan area. These investigators wished to enroll patients in a colon cancer case-control study. Within six months of contact by the investigators, only 79 per cent of the hospitals had agreed to participate. Of the 214 eligible attending physicians within the consenting hospitals, 14 per cent refused to let their patients be contacted for the study. Rothman, in an editorial on the problems of obtaining cooperation for epidemiologic studies, noted similar difficulties in obtaining permission. Funch and Marshall contacted breast cancer patients after they had participated in an interview study to determine if these patients were upset by the questioning or thought their privacy had been violated. Only 2 per cent were sorry that their physicians had given permission for them to be contacted.
If the DESAD Project were to be initiated today, the investigators would make a few changes based on their experience. First, because cooperation of local physicians is essential, there would be more formalized presentations to local medical societies to introduce the study before its onset. Second, some physicians withdrew permission to contact patients after all the DES-exposed were identified. To limit this problem, tracing efforts would be initiated and letters sent to patients as they were identified rather than waiting for the full cohort to be identified. Where possible, written agreement from participating physicians would also be obtained.
The experience of the DESAD Project has demonstrated that tracing and notifying individuals of exposure is more difficult than carrying out a study that determines only vital status. The process is complicated by the lack of standardized information to facilitate locating an individual many years after an exposure and by the lack of systematically available information on actual drug exposures.
Each year new treatments and new medications are added to standard medical practice. Although many safety precautions are taken, not all hazardous effects can be foreseen and many years may pass before such an effect can be noted. The DES-adenocarcinoma link was not discovered until more than 30 years after the drug was introduced. Thus, it is apparent that standardized record keeping is essential, particularly as it pertains to medications, treatments, diagnoses, and locator information. Record keeping should be instituted with long-term storage and retrieval capabilities in mind. For example, physicians who are now using computers for business and billing purposes should perhaps consider adding key items of medical information to their data files.
Since obstetricians are becoming accustomed to standardized prenatal records, a standardized drug list could be introduced more easily in the private practice of obstetrics than in other fields. Much progress has been made in recent years with the use of commercially available combined prenatal and delivery record systems. Many hospitals now require pre-delivery registration, which includes the accumulation of prenatal history forms from the obstetrician. Also, the Joint Commission on Accreditation of Hospitals, in their 1980 manual, has recognized the need for complete information on patient medications. The Commission states that institutions must provide for a medication record or drug profile to be kept on each patient and must have written drug recall protocols that can be readily implemented. Thus, the resulting quality, quantity, and consistency of this medical information is becoming more uniform. Further improvement can be made by using standardized diagnostic nomenclature nationwide in both ambulatory and hospital settings.
Some physicians and institutions are starting to use a patient’s social security number for purposes of identification and billing. This would aid location of patients and record linkage. However, health care consumers, professionals, and legislators must also come to some consensus on the issue of privacy versus accessibility so that patient locator information from sources such as third-party payers, Social Security, health maintenance organizations, and commercial credit companies can be used to facilitate notifying individuals of hazardous exposures. Without such cooperation, those people who move frequently, who use hyphenated last names, or who change their last names will inadvertently prejudice their opportunity to be made aware of their exposure.
Given that better records are maintained, the DESAD experience and that of others indicate a continuing need to improve access to these records by increasing physicians’ awareness of the principles and importance of epidemiologic research. Herrmann’s data showing that younger physicians appear more likely to give consent for their patients to participate in studies substantiates this conclusion since younger physicians are also more likely to have had an introduction to epidemiology. To ensure access to records, a legal mechanism should be developed to limit a physician’s liability in situations similar to that of DES where a physician prescribed a drug or treatment based on the best available information at that time.
In conclusion, there is a continued use of drugs both for approved and unapproved indications. New technology spawns new treatments and diagnostic procedures. It is likely that despite current safeguards it will again be necessary to contact patients regarding newly discovered hazards from medical exposures. The experience of the DESAD Project should give full warning that improved follow-up procedures and new legal guidelines are necessary. Standardized and well-indexed medical records that include a medical exposure list and locator information must become widely used and accessible if those individuals exposed to hazards are to be found.
Click to download the full paper.