The Maryland Lesson: Conducting Effective HIV Surveillance with Unique Identifiers
THE MARYLAND LESSON
Conducting Effective HIV Surveillance
with Unique Identifiers
An American Civil Liberties Union Report
Table of Contents
DESCRIPTION OF THE PROGRAM
SUCCESS OF MARYLAND'S UI PROGRAM
Table 1: Completeness of UI Elements
Table 2: Completeness of UI Number From Confidential Testing Sites
Table 3: Completeness of Reporting UI Status For Persons in CTS Excluding Results from Cited Lab
CONVERSATION WITH DR. LIZA SOLOMON
Director, Maryland AIDS Administration
See also HIV SURVEILLANCE AND NAME REPORTING: A Public Health Case for Protecting Civil Liberties, October 1997
THE MARYLAND LESSON
CONDUCTING EFFECTIVE HIV SURVEILLANCE WITH UNIQUE IDENTIFIERS1
Since June 1, 1994, the State of Maryland has utilized a Unique Identifier system to conduct HIV surveillance. As part of the ongoing public discussion about how to best expand HIV surveillance across the country to understand the changing epidemic, it is useful to evaluate Maryland's experience to determine whether Unique Identifiers form a viable surveillance option.
While some have suggested that it is not presently possible to implement a Unique Identifier system that works, Maryland's experience suggests that this is not true. In 1994 Maryland first began its HIV surveillance system using a non-names based Unique Identifier system. Since then the program has undergone constant evaluation and refinement. While more work remains to be done to perfect the program, at this time Maryland public health officials are satisfied that the program is running well and is adequately meeting Maryland's HIV surveillance needs. Maryland's increasingly successful experiment with Unique Identifiers suggests that effective HIV case reporting can occur without name reporting.
Maryland's Unique Identifier system was implemented after attempts to institute HIV name reporting were defeated in the Maryland legislature in 1992 and 1994. The objectives of Maryland's UI system are: (1) to monitor HIV infection; and (2) to provide supplemental case identification for AIDS and HIV+ symptomatic cases. Under the UI program, a numerical code is assigned to each new case of HIV infection and to lab reports of CD4 counts of 200 or less.2 The code consists of a 12-digit number including, from left to right, the last four digits of the individual's Social Security number, the individual's date of birth, race/ethnicity and gender.
Race and gender are coded by using the same coding scheme used in the CDC's HIV/AIDS Case Reporting Form. So, for example, an African-American male born on February 4, 1952 whose Social Security number is 678-01-1234 would have a UI of 123402045221.
Maryland's UI system proceeds through the following steps:
1) The provider orders the laboratory test and creates the Unique Identifier, which is sent with the laboratory requisition.
2) The lab sends the UI Report Form for positive HIV tests and CD4 counts of less than 200 to the State AIDS Administration office or to the local health department. (If the forms are sent to the local health department, they will be forwarded to the state AIDS Administration.)
3) The AIDS Administration matches each UI received against the State AIDS Registry (HARS), which has been coded with UI's using the same 12-digit numbering system.
4) The AIDS Administration generates a list of non-matches to the State AIDS Registry, thus creating a list of cases of HIV infection that are not yet reportable as AIDS or HIV+ symptomatic cases in Maryland.
5) Surveillance staff call physicians as necessary to obtain additional information on the patient, including the clinical status and risk categories. Patient names are not given to the surveillance staff.
6) If the patient is found to have an AIDS-defining illness or is HIV symptomatic (a reportable condition in Maryland), then the surveillance staff will obtain information, including the patient's name, which is used to create an AIDS case report.
In Maryland, it is the provider who is required to construct the UI and then transmit the number to the lab with the lab request form. The provider is required to maintain some means by which a surveillance officer can backtrack in the event that a UI number is incomplete and to provide additional clinical data and risk information. This is often referred to as the provider log, although providers use a variety of systems. Logs or other systems maintained by providers do not necessarily include the names of the individuals who test positive for HIV3. A log or other tracking mechanism might contain the UI and a corresponding medical record number or other non-name identifying information. The Johns Hopkins Medical Center, for example, maintains two separate databases, one containing the UI and the other additional patient information. The two databases are merged for the purpose of obtaining information on clinical status and risk practices.
Maryland recognizes that every system of HIV surveillance has its challenges. When Maryland's UI system was implemented in 1994, the challenge was to get complete UI numbers for as many lab reports of new cases of HIV infection as possible. An examination of Maryland's experience shows that the completeness of UI numbers has increased substantially over time, as providers have become more used to the UI system. While only 61% of UI's reported in the first six months of the program were complete, approximately 77% of UI's reported in the last six months of 1996 were complete. (See Table 1, "Completeness of UI Elements.") The major problem with UI code completeness lies in missing Social Security numbers. In the last evaluated five month period in 1996, Social Security numbers were missing from 15% of the reported UI numbers. (See Table 1.)
The Maryland AIDS Administration has reason to believe they can improve on the 77% completion rate. In a pilot program, training on UI reporting was provided to staff at all Confidential Testing Sites in the State. After the training was completed, the completeness of UI numbers in reports from Confidential Testing Sites increased to 96.6%. (See Table 2, "Completeness Of UI Number From Confidential Testing Sites," next page.) While it may be unrealistic to assume that such a high rate of completeness can be achieved in all reported UI numbers, officials at the Maryland AIDS Administration are confident that with further work the completeness rate in the State's UI program will exceed 80%.
The success of a UI program is measured not just by the completeness of UI numbers reported, but also by the ability to match the UI's of persons listed in the UI Registry with the UI's of persons listed in the State's AIDS Registry and consequently to be able to distinguish new cases of HIV infection from previously reported AIDS cases. For example, in a perfect system, all persons who were tested for HIV in 1996 and were included in the AIDS Registry because of an AIDS diagnosis would also be included in the UI Registry. This would be a match rate of 100%. An analysis of Maryland's program indicates a system that is not perfect but is functioning reasonably well. After removing specimens sent from one lab that was cited for non-reporting, the match rate was 76.5%. (See Table 3, "Completeness Of Reporting UI Status For Persons in CTS, Excluding Results From Cited Lab," next page.) CDC data from Alabama and Arizona, both of which use HIV name reporting, suggests that the match rate in those states is 79-90%.4
Is the success rate of Maryland's UI system good enough? That depends on what level of statistical precision is considered good enough to conduct adequate HIV surveillance. The Maryland AIDS Administration reports that its UI system fulfills all of the functions that can reasonably be expected of an HIV surveillance system. The UI program gives Maryland the data it needs to track HIV infection in the State and to guide decisions about AIDS funding and prevention efforts. Officials at the Maryland AIDS Administration are pleased that the Unique Identifier system has enjoyed considerable community support and is thought to create an environment in which individuals are more willing to be tested and learn their HIV status.
Maryland's confidence in this regard is bolstered by a comparison of the demographic characteristics of new cases reported in the State's AIDS registry and cases reported in the UI registry. The UI registry indicates an increasing number of HIV infections in Maryland among women (up to 34.5% in UI registry from 25.3% in AIDS registry) and people aged 13 to 29 (up to 19.4% in UI registry from 14.1% in AIDS registry). These trends are consistent with our current understanding of trends in HIV infection, and thus suggest that Maryland's UI system is effectively and accurately tracking the spread of HIV in that State.
Maryland public health officials readily acknowledge that there are some things that the State's UI system cannot do. For example, in the absence of good provider logs or their equivalent it will be difficult to obtain detailed information on clinical status and risk categories. Maryland has embarked on an education campaign to train clinicians in the importance of maintaining logs and is contemplating adding information on risk categories to the Unique Identifier in order to improve the information available to state health officials.
An examination of Maryland's experience indicates that Unique Identifiers, while not problem-free, can provide a sound basis for HIV surveillance. Maryland is now effectively using UI's to fulfill the basic functions of HIV surveillance: to monitor HIV infection trends and to identify cases of HIV and AIDS. It is worth noting that Maryland has achieved a level of success using Unique Identifiers despite the fact that the federal Centers for Disease Control has failed to provide any funding to support the program5. Requests by Maryland to use funds from its surveillance cooperative agreement to support this program have been repeatedly denied by the Centers for Disease Control. Maryland is committed to continuing and improving its UI program, despite this continued lack of federal funding.
That Maryland's UI system does not accomplish all public health goals related to HIV should not be surprising, because every disease surveillance system is limited in what it can accomplish. Disease surveillance systems are not health care delivery systems. Nor are they means of conducting in-depth epidemiological investigations of individual cases of infection.6 Thus, suggesting that a UI system does not work because, for example, it does not link individuals into health care, makes no more sense than suggesting that prevention education does not work because it does not increase AIDS funding.
In light of Maryland's increasingly successful experiment with Unique Identifiers, UI's must be considered as a viable alternative to names-based reporting as we expand HIV surveillance in response to the changing nature of the epidemic.
1 This report was authored by Michael Adams, staff counsel for the ACLU AIDS/HIV Project. The Maryland AIDS Administration verified all of the statistical data and analysis of that data contained in the report.
2 CD4 cell counts are the most commonly used surrogate markers for assessing the state of the immune system. As the CD4 count declines, the risk of developing opportunistic infections increases. The normal range for CD4 counts is 500 to 1500 per cubic millimeter of blood. A CD4 count of 200 or less triggers an AIDS diagnosis.
3 Some concerns have been raised that Maryland's UI system may in fact provide greater possibility for breaches of confidentiality because of the provider logs. The possibility of confidentiality breaches is substantially decreased when providers do not include names in their logs or databases. In addition, a significant concern in HIV surveillance is testing deterrence as a result of testee concern about confidentiality breaches. In the realm of deterrence, the perception of the testee is as important as the actual possibility of breaches of confidentiality. The available data indicates that members of communities most affected by HIV are less trusting of government agencies than they are of their own doctors. See, e.g., Woodrow Jones, "An Overview of Health Care Issues in Black America," Black Health Care, Jones and Rice, eds., 1987.
4 Alabama and Arizona are low seroprevalence states, reporting 4,266 AIDS cases and 5,036 AIDS cases, respectively. Maryland, by contrast, is a high seroprevalence state and reports 15,298 AIDS cases. (CDC HIV/AIDS Surveillance Report, Vol. 8, 1997). The lighter caseloads of Alabama and Arizona may help to explain why they have been able to obtain slightly higher match rates than Maryland.
5 By contrast, the CDC has provided funding to states to set up HIV names-based reporting systems.
6 This does not mean that we should ignore the possibilities for epidemiologic investigation that UI systems offer. For example, Maryland has participated with the federal Centers for Disease Control in surveillance of sub-types of HIV. In fact, it was through Maryland's UI system that the first case of sub-type G HIV was reported. P.S. Sullivan, Do., An.N., Robbins, K., et. al. "Surveillance for Variant Strains of HIV: Subtype G and Group O HIV-1," Letter, JAMA, 278:292 (1997).
ACLU: What should we look at to determine whether an HIV surveillance system like Unique Identifiers or names-based reporting is working?
L.S.: In order to assess whether a disease surveillance system is working, one must first clarify what are reasonable expectations for the system and what information is needed to accomplish the important public health goals of surveillance. If one looks to a surveillance system to give basic epidemiological information on populations that are affected and the risk practices involved, then some slight over or under-counting will not have a significant impact on the data. For example, it may only be important for epidemiological purposes to know that young heterosexual women are increasingly becoming infected with HIV. Knowing that there are 200 HIV positive women as compared to 220 will not impact service or prevention planning.
ACLU: What level of statistical precision is necessary in order to accomplish the public health goals we have for HIV surveillance?
L.S.: This question again relates to the goal of the surveillance activities. In order to plan services, prevention activities, and target programs, it is important to have a good, although not exact, idea of the number of individuals involved. Public health programs are usually targeted at populations and affected communities. Knowing each person's identity is not necessary to accomplish these goals.
ACLU: Would a surveillance system that included names be more helpful in insuring that individuals who need medical treatment have access to health care?
L.S.: Surveillance systems are not the most effective vehicle for insuring that individuals receive health care. Referring individuals into care and insuring that they have access to care is a critical public health objective. Disease registries, such as cancer registries or AIDS registries, have rarely been the means to identify individuals for referrals into care. Estimates from the Centers for Disease Control suggest that there is a significant lapse in time before an individual with AIDS is reported to the state registry. CDC estimates that 50% of people with AIDS are reported within 3 months of diagnosis while 20% are not reported to AIDS registries for more than one year. (CDC HIV/AIDS Surveillance Report, Vol. 8, 1997). Similar delays can be expected in any HIV case reporting system. Guaranteeing care and insuring that there are needed linkages to care should occur at the point of contact with the patient Ð at the testing site or clinician's office.
ACLU: Has Maryland been able to use its UI system to get accurate data for purposes of funding?
L.S.: Maryland has used data from its UI system to help inform decisions concerning allocation of resources and to provide information used in funding jurisdictions.
ACLU: Do you believe that it is possible to monitor recent trends in HIV infection in Maryland using Unique Identifiers?
L.S.: UI information is helpful in providing a picture of early infection and can provide information on new groups that may be affected by HIV. In a comparison of demographic information obtained from the UI registry and the AIDS registry in Maryland, we have seen that individuals with HIV are younger and more likely to be female than individuals in the AIDS registry. Using the UI and AIDS registries we have found that there is little difference by race among those with HIV and AIDS in Maryland.
ACLU: Has Maryland had difficulty providing services for people with HIV and AIDS because the state does not have a names-based surveillance system?
L.S.: Maryland has a comprehensive system of service delivery to insure that individuals needing services have access to them. In addition, Maryland has an AIDS Drug Assistance Program (ADAP) which provides medications to individuals with HIV and AIDS who have medical needs but inadequate resources to obtain medications. Maryland's ADAP provides unlimited access to all protease inhibitors and antivirals for individuals in the program. There are no waiting lists and no clinical restrictions other than a diagnosis of HIV.
ACLU: Would name reporting allow for more effective partner notification?
L.S.: Partner notification is an important component of Maryland's HIV prevention program. The contact point for insuring that partner notification is implemented is at the time that a person learns his/her HIV status. As part of Maryland's counseling procedure, when individuals receive HIV positive test results counselors inform them about the importance of partner notification and assist individuals in notifying. This process exists regardless of the surveillance system or the venue of the test, including anonymous test sites.
ACLU: Do we need name reporting in order to track down those individuals who are tested for HIV and do not return for their test results?
L.S.: It is critically important that individuals, both HIV positive and negative, learn their test results. In confidential HIV testing facilities, the patient name is known to the health care provider who orders the test. Follow-up with that patient can take place even though the State does not have a State-sponsored names-based registry of HIV. In anonymous testing facilities, the patient's name is not known to the provider. However, research shows that individuals who are tested anonymously are more likely to retrieve their test results than individuals who provide their name at the time of testing.
ACLU: Doesn't Maryland's requirement that providers maintain a log of those who are tested mean that the possibilities for confidentiality breaches are even greater?
L.S.: In Maryland, providers are required to maintain some means for surveillance staff to backtrack to obtain additional information. Providers don't necessarily keep logs. Some keep computer databases. And logs and computer databases don't need to include names. They can instead include medical record numbers or other non-name identifying information. Also, we're worried about the perception of the person being tested and whether they will be deterred from testing. In our experience, individuals most at risk for HIV trust their doctors more than they trust government agencies. They are thus less likely to be deterred from being tested by provider logs or their equivalent.
ACLU: Why do you think Texas has a negative evaluation of its UI program, given the success of Maryland's program?
L.S.: Although Texas and Maryland have both used the same 12 digit UI number, the two states have differences in their programs. In Texas, both physicians and labs have the responsibility to report. This may have created some difficulty with the volume of reports. Also, Texas' system does not require that physicians keep a log or equivalent. Thus when questions arose there was no way to return to the physicians and get additional data. Ultimately, any new surveillance system requires considerable work with the physician community to insure that it is well-implemented. It appears that additional education efforts with labs and physicians would be helpful.
ACLU: Has Maryland received any money from the CDC to develop or implement its UI program? úHas the state requested funds?
L.S.: Maryland has not received funds from the Centers for Disease Control to implement its HIV surveillance system. Maryland requested funds to support this program in our 1995, 1996 and 1997 surveillance cooperative agreement. All requests were denied. Maryland and Texas did receive funds from the CDC to evaluate the UI program.
ACLU: Do you think UI systems are more expensive than names-based surveillance systems?
L.S.: Maryland has implemented its UI system without additional funds. However, any expanded surveillance system, names-based or UI-based, does require additional resources. Although we have implemented our system without additional funds, we estimate that it would cost the State of Maryland $100,000 to more fully implement the UI surveillance system and have it reach the level of accuracy that is realized by our AIDS surveillance system.
ACLU: Does Maryland plan any improvements in its UI program?
L.S.: We are considering two revisions to our UI system that we believe will give us additional data and reduce errors. First, we are considering having the provider ascribe risk category and include this as part of the UI number. This would allow us to have complete risk categories for all individuals in our UI data base without needing to call the physician. In addition, we are considering changing the layout of our number. Currently the race/ethnicity categories (a 1 to 5 classification) is right next to the gender category (1,2 category). We believe this has caused transcription errors which may be reduced by changing to an alphanumeric code. We are also expanding our education and training for clinicians to help them comply with the law.
ACLU: Why has Maryland resisted conducting HIV surveillance through name reporting?
L.S.: Maryland adopted HIV surveillance by Unique Identifier after full discussion and debate within our General Assembly and our community. Bills promoting names reporting were introduced and defeated in the 1992 and 1994 General Assembly sessions. After a full discussion of these issues, legislation was passed which authorized the creation of the Unique Identifier system. This legislation has enjoyed considerable support from the HIV community. The AIDS Administration is committed to working in partnership with affected communities in all our programs.
ACLU: Does Maryland intend to continue using UI's, or to switch to names-based reporting?
L.S.: Maryland is continuing to refine and improve upon our HIV surveillance system. We have no plans to change to a names-based system.
AMERICAN CIVIL LIBERTIES UNION
Washington National Office
Kenneth B. Clark
Chair, National Advisory Council
This report has been prepared by the American Civil Liberties Union, a nationwide, nonpartisan organization of 275,000 members dedicated to preserving and defending the principles set forth in the Bill of Rights
This paper can be accessed via the ACLU's website <url: archive.aclu.org>
Copies may also be obtained for $1.00 each through ACLU Publications at 1-800-775-ACLU or P.O.B. 186, Wye Mills, Maryland, 21679
THE MARYLAND LESSON:
Conducting Effective HIV Surveillance with Unique Identifiers
Director, AIDS Project
Founded in 1986, the AIDS Project of the ACLU