January 15, 2005
National Coordinator for Health Information Technology Department of Health and Human Services
ATTN: NHIN RFI Responses
Re: NHIN RFI Responses
To Whom It May Concern:
The Association of American Physicians and Surgeons was founded in 1943 to preserve private medicine. We represent thousands of physicians in all specialties nationwide, and the millions of patients that they serve.
We are very concerned about the detrimental effects of imposing a government-dictated format for medical recordkeeping and also about the violation of patients’ rights by using their medical information without their knowledge or consent.
We would first like to discuss some general concerns and then address the specific questions that are presented with regard to the proposed National Health Information Network (NHIN).
1. The requirements for interoperability of all patient records could impede medical care.
Experienced clinicians find that electronic medical records are the most difficult to review. They contain a huge volume of information that is not useful for the patient’s particular problem and often lack details that are of importance in certain conditions but are not part of the routine examination. The rigid format treats all bits of information as being of equal importance. The traditional record has information filtered and organized by an intelligent clinician focused on the patient. The electronic record is a collection of data of variable reliability and relevance assembled by protocol. The traditional record at its best is comparable to a classic portrait that captures the essence of the subject’s personality as well as faithfully recording his physical features. The electronic record is more like a Picasso or other abstract work with a jumble of distorted, malproportioned, disjointed pieces.
The production of the electronic record also distracts the clinician’s attention from the process of clinical reasoning, thus impeding rather than facilitating diagnosis.
The idea that having the complete medical record instantly accessible will be life-saving in emergencies reflects a lack of understanding of how the medical record is actually used in practice. The clinician first obtains the pertinent history from the patient, family, or (in an emergency) bystanders or first responders, and attends to life-threatening problems. Only then does he have the opportunity to review the past clinical record–much of which may have just become irrelevant, and part of which may contain dangerous errors.
The medical record is first and foremost a tool for clinicians to aid in diagnosis and treatment. This purpose must not be subordinated to other objectives such as research, consumer information, biosurveillance, or “quality” measurement. Practicing clinicians–not academics or managers or planners–must determine whether a particular record format is acceptable. There is no uniform format that will be suitable for all specialties or practice settings. No format can be deemed acceptable before it has been extensively used in the field.
2. The NHIN would not be suitable for research or public health monitoring.
For research, there is no substitute for a standardized set of data designed to address a specific, narrowly defined question. Validating the data collection device is an indispensable part of any research protocol.
Public health monitoring does need improvement. State and local agencies are often very far behind on analyzing the data that they already have. Giving them access to vastly expanded data sets is likely to complicate their work. Better intercommunication within the public sector is very desirable, and needs to be developed independently of an NHIN. The sequence of starting with an NHIN, rather than with carefully tested component parts, is backwards, and is far more likely to lead to total system breakdown than to improvement. In any event, the optimal system for public health reporting is probably very different from the optimal system for routine medical care.
If disease surveillance depends on medical records, then it can only be as good as the records themselves (also see #3 below). Detection based on medical records cannot be early detection; for this, monitoring technology is essential. This might include ways of detecting feverish persons passing through public places such as airports or bus stations, as reportedly was done in China and perhaps other Asian countries at the time of the SARS outbreak. Waiting for cases of smallpox to appear, and be recognized and diagnosed, may delay public health response beyond the point where it could be of much use. The goal should be to detect the virus dispersed in the air or excreted by individuals on their breath before they are manifestly ill. The fact that many infectious diseases in the past were diagnosed by their odor suggests that they caused the afflicted patients to excrete characteristic metabolic products that with today’s analytic chemistry could be identified in very low concentrations, providing an effective early warning system.
3. Records in the NHIN will be unreliable.
When clinicians have to write a few sentences to describe the essence of a patient’s illness, they generally do so with some thoughtfulness. When they are required to fill out multiple-page checklists with all types of findings that they consider trivial or irrelevant, the inclination is just to put in check marks without giving it much thought. The reliability of any such findings is dubious and certainly has never been validated.
Research is needed on the interobserver and intraobserver variability of findings in the clinical record. The ability of clinicians to use standardized terminology or numeric codes consistently must also be tested. Based on experience with the relatively simple matter of CPT coding, it is likely that accuracy will be extremely poor.
An interesting experiment would be to convert a traditional clinical summary into the format suitable for the electronic medical record, and then having someone else translate the EMR back into an English summary. Would the result be better or worse than translating an English poem into Chinese and back again?
It is impossible to devise a standardized system of numeric codes or nomenclature that can have a one-to-one mapping with a summary in a natural language. There cannot be enough codes to capture nuance. Yet forcing the representation of a vague and ambiguous concept by a code that is supposed to be correct to five significant digits gives a dangerously misleading appearance of accuracy.
When, inevitably, errors occur, the correction of all pertinent entries in a huge data base is extremely difficult or impossible. Inaccurate information is worse than no information in most clinical settings. 4. The benefits of the NHIN are hypothetical and unproved.
Even if the data in the NHIN were shown to be accurate, there is no evidence that the imposition of a rigid, standardized form will produce any overall health benefits. While some limited benefits may be demonstrated in controlled settings, an adverse impact on outcomes that are not the subject of study is highly likely and must be ruled out. For example, one may demonstrate that diabetics receive more “preventive maintenance” in a clinic that uses electronic records to monitor compliance. But what is the effect on other conditions that may be of much higher priority to the patient and also have an important effect on long-term morbidity and mortality? And what is the effect of similar record-keeping requirements in facilities that cannot afford extra personnel and must divert funds from other uses? Long-term studies of all-cause mortality, as well as other objective health outcomes in the entire affected population are needed.
The economic benefits of the electronic medical record are also hypothetical. To prevent duplication of tests, a review of the current style of medical records, combined with a patient history, ought to be adequate. In many instances, however, duplication of the tests is exactly what is necessary because the patient’s situation may have changed since the last test was done, and findings that previously were obscure may become apparent.
Even if a positive economic impact may be shown overall, the entities receiving most of the benefit will be large institutions that can absorb tremendous upfront costs. Smaller facilities such as individual physicians’ offices will be forced to bear a disproportionate burden, which could drive many of them out of business. Thus, any analysis of costs and benefits must consider different types of facilities separately rather than providing overall averages.
5. All patients will be forced to sacrifice confidentiality.
Even if the ambitious goals of the NHIN could be achieved at a population level, the actual impact upon the health of the individual patient remains to be demonstrated. In a compulsory tradeoff for some hypothetical societal benefit, patients will be forced to relinquish all claims to the confidentiality of their medical information. Once entered into a networked computer, data cannot be kept secure. The requirements of interoperability and security are mutually contradictory. Punishment of persons who violate confidentiality does nothing to make the victim whole if his privacy has been destroyed and he has suffered consequences as far as his employability, financial status, or personal life are concerned.
No one should permitted to (much less forced to) enter data into such a database without fully informed consent. Technical methods are needed to track information releases and to hold individuals responsible for the misuse of information.
Whether the information network is centrally located, or simply centrally brokered with the ability to access some material on a remote computer, the effect on patient confidentiality is destructive.
We would also like to comment on some specific questions:
Question 1. An interoperable National Health Information Network, as we understand it, would eventually encompass all health records in the nation, whether centralized in one location or dispersed in many locations. This would require the standardization of data that are collected, the terminology that may be used, and the media and electronic formats in which data are recorded. If numerical codes are used, hundreds of thousands of codes would be required, with an entire new industry to develop them and to instruct physicians in their use.
The key barrier to interoperability is that it is incompatible with efficient clinical functioning, which requires flexibility and optimization to the specialized requirements of the setting and clinical conditions of the patient population served.
Another barrier is the wide variety of clinical systems now in use and the scarcity of funding. Will all clinicians and medical facilities be required to replace or upgrade their existing systems, or to computerize an all-paper office? How will old records be handled? How many existing enterprises will be driven out of business?
Question 2. There is no known model that would allow widely available access while protecting individual identifiable health information. As testimony at the National Committee on Vital and Health Statistics has indicated, access and security are two mutually exclusive requirements.
As HIPAA has shown, it is possible both to impede communication for legitimate purposes and facilitate uses that are not in a patient’s best interest. Problems in a system for processing claims would be amplified by orders of magnitude if such a model were imposed on the entire medical record.
As HIPAA “privacy notification” forms make clear, identifiable patient data can be used by both governmental and private organizations for all types of activities considered “treatment, payment, and health care operations” without the patient’s knowledge or consent. Yet simply to get an itemized statement, a patient may have to make a special telephone call to the medical facility (and figure out for himself that a phone call is needed), as HIPAA is said to require issuing a completely uninformative bill. Numerous absurdities can be cited that result either from the HIPAA regulations themselves or from misunderstanding by compliance personnel.
Question 3. If nationwide interoperability is required, a uniform national system would appear to be essential. And if the cooperation of all entities is to be demanded, there is no possible means other than by federal law enforcement to compel its use. This amounts to intrusive federal regulation of the practice of medicine.
HIPAA has already shown the nearly insurmountable difficulty in attempting to comply with both federal law and a multitude of state laws with contradictory requirements.
Apparently, a unique national identifier is envisioned to prevent duplication of records and to permit integration of all of an individual’s medical information–despite the widespread political opposition to the idea that has prevented funding of any efforts to develop it. Assuming this opposition is eventually overcome, how will identifiers be assigned? Will physicians be required to make positive identification of all individuals they treat? How else could the integrity of the record be protected?
Question 4a. A framework that could be used to develop a voluntary (non-comprehensive) NHIN is the Internet, which permits sharing of all types of information in a way that is very economical and efficient. The processes involved in Internet communications are not exclusive to defined “stakeholders.” Everyone who wishes can participate. In fact, because of the Internet, medical records already have a tremendous degree of interoperability in the sense that they can be transmitted by e-mail. Instantaneous transmission of legible information that is precisely what human beings decide is most needed is admirably suited to serve most of the achievable purposes that are sought in electronic medical records–except statistical surveillance to measure compliance with regulations and protocols. Attempting to add the latter function alters the record in a way that has many detrimental effects on the practice of medicine. Moreover, it forces patients to relinquish privacy as a condition of obtaining medical help.
To the extent that health information should be shared, private enterprise should develop the method, without government dictating the shape of the product.
b. Any role of the federal government in setting policy and standards should be very restricted. Federal policy has the potential to suppress many types of beneficial and voluntary information sharing, while forcing patients and clinicians to participate in a system that is destructive of efficient medical care as well as privacy.
c. To accelerate the adoption and use of voluntary information sharing in the mainstream delivery of care, any regulatory barriers should be removed, for example, to physician communication with patients across state lines.
If technology is advantageous, there is no need for artificial acceleration of its adoption (as shown with sophisticated computerized imaging techniques). Resistance is to be expected if the technology is costly, burdensome, and destructive of values such as trust and confidentiality (see question 15). d. Concerning the operation of an NHIN, hospitals and other institutions have already figured out how to communicate with each other (and to use the Internet). A significant barrier to such communications is the HIPAA Privacy Rule, which has complicated the sharing of information needed for patient care while expanding nonconsented access by nonclinical parties with no fiduciary responsibility to patients.
Question 5. Concerning the financial model required to build an NHIN, if a system is to be imposed it will have to be financed by taxes, either a hidden tax on the medical system or an expressly legislated tax on the general public to subsidize the excess costs. On the other hand, if the methods truly offer a benefit either in cost savings or improved quality, then it is expected that the private sector will adopt them quickly. The fact that something needs to be subsidized should be a signal that its benefit, if it exists at all, is not apparent and remains to be proved.
Question 6. To operate and sustain a functioning NHIN probably requires the same characteristics that already sustain an effective method of distributing groceries and other valuable products: freedom for institutions to manage their own affairs. Those who provide medical services already maintain their own information systems just as they maintain their x-ray machines and physical plant.
Question 7. The privacy and security considerations as alluded to above should be determined at the individual level by patients who control access to their own encrypted information. A set of rigid bureaucratic rules and penalties, as in HIPAA, does not effectively protect privacy. Prior to the electronic medical record, records were protected to a significant extent simply by being inaccessible. Inaccessibility may be the most reliable protection. Encryption and password requirements are the high-tech equivalent.
Question 8. The framework for NHIN should depend certainly on what the public policy objectives are and whether they can even be achieved through such a network. To date, clear definition is lacking for the public policy objectives, the concept of an NHIN, and any objective criteria for determining whether the objectives can possibly be met in the manner proposed.
Questions 9 and 10. Competition is a natural state of affairs that occurs if it is not punished and suppressed. The imposition of a government standard, especially since no obvious standard has emerged from the marketplace, would simply freeze into place a method that is probably far from optimal and destroy the market forces that would produce a better one. Free competition and a centrally planned system are incompatible.
Questions 11 and 12. For developing systems that gain wide acceptance, the evolution of computer operating systems might be studied. Somehow customer demand for interoperability drove the industry to enable it. One way of encouraging standardization would be to refrain from prosecuting or otherwise threatening companies such as Microsoft that might be developing the needed tools. And established technologies must be allowed to die and be replaced. Fortunately, the MUMPS programming language for medical records, which was under consideration by some systems, was not forced on the whole country 20 years ago. Card readers, FORTRAN, and COBOL have also largely disappeared, having persisted longest in academia and government.
An NHIN will be voluntarily utilized if and only if it is economical and clinically efficient and capable of preserving privacy.
Question 13. We expect the imposition of NHIN to have a destructive effect on the health information technology market if the industry is forced to adopt a uniform, rigid system. NHIN would block possible alternate paths, but no one would ever know what might have happened in its absence. Ten years before cellular telephones started to develop, no one could have foreseen the results.
Question 14. There are two possible ways for interoperability standards to develop: by top-down bureaucratic command and control, or by the natural process of give and take in the marketplace. We have Medicare and the personal computer as respective examples of each.
Question 15. In a free market, some vendors gain an advantage because they serve their customers better. To prevent “undue influence,” it is important for users to choose their own products, including the specific technology that is most advantageous to them. Government restrictions on marketplace entry or attempts to dictate standards inevitably carries the risk of “undue influence” or corruption of the process.
The only way to protect privacy is to assure that information entry is voluntary on the part of both physicians and the patients. Data should be encrypted with the patient in control of the key. The patient could then decide who should have access to the data and could even segregate portions of the data for privately determined purposes. A patient might want his attending physician to have access to all data, but might want to be sure that financial institutions and employers could access only a limited portion. Reliable technologic protection of privacy, rather than cumbersome bureaucratic process, would speed acceptance of the EMR as well as protecting patients’ rights.
Questions 16 and 17. If interoperability means a rigid, centrally specified system, or a governmentally imposed coding system with hundreds of thousands of codes, there is probably no imaginable type of management or business rules organization that could successfully promote widespread diffusion. It would have to be imposed by force. On the other hand, transfer of free text is already accomplished with great efficacy.
Other Questions. To the extent that the NHIN is desirable in the private sector, it will continue to develop. In such a rapidly changing field, it is imprudent to attempt to foresee what might happen, much less to impose a system, particularly one that has not been field tested.
The best way to develop this system, if the government wished to play a role, is to build some model systems as in public health agencies or the VA hospital, and then make the results available for scrutiny by the entire industry so that those aspects that are advantageous can be broadly adopted and those which are disadvantageous can be discarded. Special incentives to adopt technology that improves productivity and decreases costs are not needed. Special “incentives,” such as subsidies or penalties, are only required to force people to adopt systems that are disadvantageous.
One major barrier is HIPAA, which should be repealed. Its cumbersome and meaningless pretenses at privacy protection, when the need for consent to release information for a wide range of purposes has been abolished, have no redeeming features. The traditional practice of simply requiring patient consent for the release of information should be employed instead.
The NHIN has so far been described only in vague terms. It is impossible to even attempt a precise answer to the questions posed in the Federal Register, or to design a “potential technical architecture,” without having a concrete example of what the agency has in mind. Institutions should be demanding that government “show us a working model.”
Before proceeding with the implementation of a nationwide system, extensive field testing of a variety of proposed models is in order, with monitoring of long-term clinical results.
Even in the doubtful event that some clinical benefit could be demonstrated, it would not outweigh basic patients’ rights. The right to withhold informed consent for release of private data must be assured by a technically reliable mechanism. And no patient or physician should be forced to use an NHIN as a condition of receiving or rendering medical treatment.
Jane M. Orient, M.D.