IX. IMPACT ANALYSIS

[Please label written and e-mailed comments about this section with the subject: Impact.]

A. Executive Summary

The costs of implementing the standards specified in the statute are primarily one-time or short-term costs related to conversion. These costs include system conversion/upgrade costs, start-up costs of automation, training costs, and costs associated with implementation problems. These costs will be incurred during the first three years of implementation. The benefits of EDI include reduction in manual data entry, elimination of postal service delays, elimination of the costs associated with the use of paper forms, and the enhanced ability of participants in the market to interact with each other.

In our analysis, we have used the most conservative figures available and have taken into account the effects of the existing trend toward electronic health care transactions. Based on this analysis, we have determined that the benefits attributable to the implementation of administrative simplification will accrue almost immediately but will not exceed costs for health care providers and health plans until after the third year of implementation. After the third year, the benefits will continue to accrue into fourth year and beyond. The total net savings for the period 1998-2002 will be $1.5 billion (a net savings of $1.7 billion for health plans, and a net cost of $.2 billion for health care providers). The single year net savings for the year 2002 will be $3.1 billion ($1.6 billion for plans and $1.5 billion for providers).

B. Introduction

We assessed several strategies for determining the impact of the various standards that the Secretary will designate under the statute. We could attempt to analyze the costs and savings of each individual standard independently or we could analyze the costs and savings of all the standards in the aggregate. We chose to base our analysis on the aggregate impact of all the standards. Assessing the cost of implementing each standard independently would yield inflated costs. The statute gives health care providers and health plans 24 months (36 months for small health plans) to implement each standard after it is designated. This will give the industry flexibility in determining the most cost- effective way of implementing the standards. A health plan may decide to implement more than one standard at a time or to combine implementation of a standard with other system changes dictated by its own business needs. As a result, overall estimates will be more accurate than individual estimates.

Assessing the benefits of implementing each standard independently would also be inaccurate. While each individual standard is beneficial, the standards as a whole have a synergistic effect on savings. For example, the combination of the standard health plan identifier and standard claim format would improve the coordination of benefits process to a much greater extent than either standard individually. Clearly, the costs and benefits described in this impact analysis are dependent upon all of the rules being published at roughly the same time.

It is difficult to assess the costs and benefits of such a sweeping change with no historical experience. Moreover, we do not yet know enough about the issues and options related to the standards that are still being developed to be able to discuss them here. Our analysis, as a result, will be primarily qualitative and somewhat general. In order to address that shortcoming, we have added a section discussing specific issues related to the provider identifier standard. In each subsequent regulation, we will, if appropriate, include a section discussing the specifics of the standard or standards being designated in the regulation. In addition, we will update this analysis to reflect any additional cost/benefit information that we receive from the public during the comment period for the proposed rule. We solicit comments on this approach and on our assumptions and conclusions.

C. Overall Cost/Benefit Analysis

In order to assess the impact of the HIPAA administrative simplification provisions, it is important to understand current industry practices. A 1993 study by Lewin-VHI (1, p. 4) estimated that administrative costs comprised 17 percent of total health expenditures. Paperwork inefficiencies are a component of those costs, as are the inefficiencies caused by the more than 400 different data transmission formats currently in use. Industry groups such as ANSI ASC X12N have developed standards for EDI transactions, which are used by some health plans and health care providers. However, migration to these recognized standards has been hampered by the inability to develop a concerted approach, and even “standard” formats such as the Uniform Bill (UB-92), the standard Medicare hospital claim form (which is used by most hospitals, skilled nursing facilities, and home health agencies for inpatient and outpatient claims) are customized by plans and health care providers.

Several reports have made estimates of the costs and/or benefits of implementing electronic data interchange (EDI) standards. In assessing the impact of the HIPAA administrative simplification provisions, the Congressional Budget Office reported that:

“The direct cost of the mandates in Title II of the bill would be negligible. Health plans (and those providers who choose to submit claims electronically) would be required to modify their computer software to incorporate new standards as they are adopted or modified...Uniform standards would generate offsetting savings for plans and providers by simplifying the claims process and coordination of benefits.” (page 4 of the Estimate of Costs of Private Sector Mandates)

The most extensive industry analysis of the effects of EDI standards was developed by WEDI in 1993, which built upon a similar 1992 report. The WEDI report used an extensive amount of information and analysis to develop its estimates, including data from a number of EDI pilot projects. The report included a number of electronic transactions that are not covered by HIPAA, such as materials management. The report projected implementation costs ranging between $5.3 billion and $17.3 billion (3, p. 9-4) and annual savings for the transactions covered by HIPAA ranging from $8.9 billion and $20.5 billion (3, pp. 9-5 and 9-6). Lewin estimated that the data standards proposed in the Healthcare Simplification and Uniformity Act of 1993 would save from 2.0 to 3.9 percent of administrative costs annually ($2.6 to $5.2 billion based on 1991 costs) (1, p.12). A 1995 study commissioned by the New Jersey Legislature estimated yearly savings of $760 million in New Jersey alone, related to EDI claims processing, reducing claims rejection, performing eligibility checks, decreasing accounts receivable, and other potential EDI applications (4, p.316)

We have drawn heavily on the WEDI report for many of our estimates. However, our conclusions differ, especially in the area of savings, for a number of reasons. The WEDI report was intended to assess the savings from a totally EDI environment, which HIPAA does not mandate. Health care providers may still choose to conduct HIPAA transactions on paper. In addition, a significant amount of movement toward EDI has been made (especially in the claims area) since 1993, and it is reasonable to assume that EDI would have continued to grow at some rate even without HIPAA. In order to assess the true impact of the legislation and these regulations, we cannot claim that all subsequent benefits are attributable to HIPAA.

D. Implementation Costs

The costs of implementing the standards specified in the statute are primarily one-time or short-term costs related to conversion. They can be characterized as follows:

1. System Conversion/Upgrade -- Health care providers and health plans will incur costs to convert existing software to utilize the standards. Health plans and large health care providers generally have their own information systems, which they maintain with in-house or contract support. Small health care providers are more likely to use off-the-shelf software developed and maintained by a vendor. Examples of software changes include the ability to generate and accept transactions using the standard (for example, claims, remittance advices) and converting or crosswalking current provider files and medical code sets to chosen standards. However, health care providers have considerable flexibility in determining how and when to accomplish these changes. One alternative to a complete system redesign would be to purchase a translator that reformats existing system outputs into standard transaction formats. A health plan or health care provider could also decide to implement two or more related standards at once or to implement one or more standards during a software upgrade. We expect that each health care provider’s and health plan’s situation will differ and that each will select a cost-effective implementation scheme. Many health care providers use billing agents or claims clearinghouses to facilitate EDI. (Although we discuss billing agents and claims clearinghouses as separate entities in this impact analysis, billing agents are considered to be the same as clearinghouses for purposes of administrative simplification). Those entities would also have to reprogram to accommodate standards. We would expect these costs to be passed on to health care providers in the form of fee increases or to be absorbed as a cost of doing business.
2. Start-up Cost of Automation -- The legislation does not require health care providers to conduct transactions electronically. Those who do not currently have electronic capabilities would have to purchase and implement hardware and software and train staff to use it in order to benefit from EDI. However, this is likely to be less costly once standards are in place, because there will be more vendors supporting the standard.
3. Training -- Health care provider and health plan personnel will require training on use of the various standard identifiers, formats, and code sets. For the most part this will be directed toward administrative personnel, but training in new code sets would be required for clinical staff as well.
4. Implementation problems -- The implementation of any industry-wide standards will inevitably introduce additional complexity as health plans and health care providers struggle to re-establish communication and process transactions using the new formats, identifiers, and code sets. This is likely to result in a temporary increase in rejected transactions, manual exception processing, payment delays, and requests for additional information.

While the majority of costs are one-time costs related to implementation, there are also on-going costs associated with administrative simplification. Health care providers and health plans may incur on-going costs to subscribe to or purchase documentation and implementation guides related to code sets and standard formats as well as health plan and provider identifier directories or data files. These entities may already be incurring some of these costs, and the costs under HIPAA would be incremental. We will be pursuing low-cost distribution options to keep these costs as low as possible.

In addition, EDI could affect cash flow throughout the health insurance industry. Electronic claims reach the health plan faster and can be processed faster. This has the potential to improve health care providers’ cash flow situations while decreasing health plans’ earnings on cash reserves.

The only known impact on individuals and employers (other than those that function as health plans) is the need to obtain an identifier.

E. Benefits of Increased Use of EDI for Health Care Transactions

Some of the benefits attributable to increased EDI can be readily quantified, while others are more intangible. For example, it is easy to compute the savings in postage from EDI claims, but attributing a dollar value to processing efficiencies is difficult. In fact, the latter may not result in lower costs to health care providers or health plans but may be categorized as cost avoidance, rather than savings. For example, a health care provider may find that its billing office staff can be reduced from four clerks to three after standards are implemented. The health care provider could decide to reduce the staff size, to reduce the billing office staff and hire additional clinical personnel, or to retain the staff and assign new duties to them. Only the first option results in a “savings” (i.e., fewer total dollars spent) for the health care provider or the health care industry. However, all three options allow health care providers to reduce administrative costs associated with billing. We are considering these to be benefits for purposes of this analysis because it is consistent with the way the industry views them. The benefits of EDI to industry in general are well documented in the literature. One of the most significant benefits of EDI is the reduction in manual data entry. The paper processing of business transactions requires manual data entry at the point in which the data are received and entered into a system. For example, the data on a paper health care transaction from a health care provider to a health plan have to be manually entered into the health plan’s business system. If the patient has more than one health plan, the second health plan would also have to manually enter the data into its system if it cannot receive the information electronically. The potential for repeated keying of information transmitted via paper results in increased labor as well as significant opportunities for keying errors. EDI allows for direct data transmission between computer systems, which reduces the need to rekey data.

Another problem with paper-based transactions is that these documents are mostly mailed. Normal delivery times of mailings can vary anywhere from one to several days for normal first class mail. To ship paper documents more quickly can be expensive. While bulk mailings can reduce some costs, paper mailings remain costly. Using postal services can also lead to some uncertainty as to whether the transaction was received, unless more expensive certified mail options are pursued. A benefit of EDI is that the capability exists for the sender of the transaction to receive an electronic acknowledgment once the data is opened by the recipient. Also, because EDI involves direct computer to computer data transmission, the associated delays with postal services are eliminated. With EDI, communication service providers such as value added networks function as electronic post offices and provide 24-hour service. Value added networks deliver data instantaneously to the receiver’s electronic mailbox.

In addition to mailing time delays, there are other significant costs in using paper forms. These include the costs of maintaining an inventory of forms, typing data onto forms, addressing envelopes, and the cost of postage. The use of paper also requires significant staff resources to receive and store the paper during normal processing. The paper must be organized to permit easy retrieval if necessary.

F. The Role of Standards in Increasing the Efficiency of EDI

There has been a steady increase in use of EDI in the health care market since 1993, and we predict that there would be some continued growth, even without national standards. However, we believe the upward trend in EDI health care transactions will be enhanced by having national standards in place. Because national standards are not in place today, there continues to be a proliferation of proprietary formats in the health care industry. Proprietary formats are those that are unique to an individual business. Due to proprietary formats, business partners that wish to exchange information via EDI must agree on which formats to use. Since most health care providers do business with a number of plans, they must produce EDI transactions in many different formats. For small health care providers, this is a significant disincentive for converting to EDI.

National standards would allow for common formats and translations of electronic information that would be understandable to both the sender and receiver. If national standards were in place, there would be no need to determine what format a trading partner was using. Standards also reduce software development and maintenance costs that are required for converting proprietary formats. The basic costs of maintaining unique formats are the human resources spent converting data or in personally contacting entities to gather the data because of incompatible formats. These costs are reflected in increased office overhead, and a reliance on paper and third party vendors as well as communication delays and general administrative hassle. Health care transaction standards will improve the efficiency of the EDI market and will help further persuade reluctant industry partners to choose EDI over traditional mail services.

The statute directs the Secretary to establish standards and sets out the timetable for doing so. The Secretary must designate a standard for each of the specified transactions and identifiers but does have the discretion to designate alternate standards (for example, both a flat file and X12N format for a particular transaction). We have chosen to designate a single standard for each identifier and transaction. On the surface, allowing alternate standards would seem to be a more flexible approach, permitting health care providers and health plans to choose which standard best fits their business needs. In reality, health plans and health care providers generally conduct EDI with multiple partners. Since the choice of a standard transaction format is a bilateral decision between the sender and receiver, most health plans and health care providers would need to support all of the designated standards for the transaction in order to meet the needs of all of their trading partners. Single standards will maximize net benefits and minimize ongoing confusion.

Health care providers and health plans have a great deal of flexibility in how and when they will implement standards. The statute specifies dates by which health plans will have adopted standards, but within that time period health plans can determine when and in which order they will implement standards. Health care providers have the flexibility to determine when it is cost- effective for them to convert to EDI. Health plans and health care providers have a wide range of vendors and technologies from which to choose in implementing standards and can choose to utilize a health care clearinghouse to produce standard transactions. Implementation options for transactions will be the subject of more detailed analysis in a subsequent regulation.

G. Cost/Benefit Tables

The tables below illustrate the costs for health plans and health care providers to implement the standards and the savings that will occur over time as a result of the HIPAA administrative simplification provisions. All estimates are stated in 1998 dollars -- no adjustment has been made for present value.

The tables are extracted from a report prepared by our actuaries, who analyzed the impact of the HIPAA administrative simplification provisions. Using standard actuarial principles, they utilized data from a wide range of industry sources as a base for their estimates but revised them as needed to precisely reflect the impact of the legislation. For example, the number of health care providers and percentage of EDI transactions were adjusted to reflect expected 1998 levels. Where data were not available (for example, the percentage of EDI billing for hospices), estimates were developed based on assumptions. Where data from multiple sources were in conflict, the various sources were considered in developing an independent estimate. These processes are complex and are described in detail in the actuaries’ report, both in narrative form and in footnotes to tables. The report is too voluminous to publish here, and it is not feasible to describe the processes used to arrive at each and every number. We are presenting here the data that are most critical to assessing the impact of HIPAA administrative simplification provisions and a general description of the processes used to develop those data. The full actuarial report is available for inspection at the HCFA document room and at the following web site: http://aspe.hhs.gov/admnsimp/.

The costs are based on estimates for the cost of a moderately complex set of software upgrades. The range of costs that health plans and health care providers will incur is quite large and is based on such factors as the size and complexity of the existing systems, ability to implement using existing low-cost translator software, and reliance on health care clearinghouses to create standard transactions. The cost of a moderately complex upgrade represents a reasonable midpoint in this range. In addition, we assume that health plans and health care providers with existing EDI systems will incur implementation costs related to manual operations to make those processes compatible with the EDI systems. For example, manual processes may be converted to recognize standard identifiers or to produce paper remittance advices that contain the same data elements as the EDI standard transaction. We have estimated those costs to equal 50 percent of the upgrade cost. Health care providers that do not have existing EDI systems will also incur some costs due to HIPAA, even if they choose not to implement EDI for all of the HIPAA transactions. For example, a health care provider may have to change accounting practices in order to process the revised paper remittance advice discussed above. Health plans must accept HIPAA transactions via EDI, but not all health plans will be called upon to accept all HIPAA transactions. For example, some health plans process only dental claims, while others process claims for institutional and noninstitutional services. We have assumed the average cost for non-EDI health care providers and health plans to be half that of already-automated health care providers and health plans.

Savings are based on the estimated increase in EDI attributable to the HIPAA administrative simplification provisions, multiplied by a per transaction savings for each type of transaction. Our estimates are much lower than those included in the WEDI report, primarily because we only recognize savings that would not have occurred without the legislation. While some industry estimates of gross savings (not net of costs) have been as high as $32.8 billion over five years, we believed it was important to utilize the most conservative assumptions possible. It is important to view these estimates as an attempt to furnish a realistic context rather than as precise budgetary predictions. Our estimates also do not include any benefits attributable to qualitative aspects of Administrative simplification, because of the lack of reliable data. (For example, we do not attempt to put a dollar value on improved public health practices that will result from implementation of standard identifiers.) We strongly encourage comments on how to quantitatively and qualitatively measure the efficiencies realized as a result of the HIPAA administrative simplification standards.

More detailed information regarding data sources and assumptions is provided in the explanations for the specific tables.

Table 1 below shows estimated costs and savings for health plans. The number of entities is based on the WEDI report, Department of Labor data, and various trade publications trended forward to 1998. The cost per health plan for software upgrades is based on the WEDI report, which estimated a range of costs required to implement a fully capable EDI environment. The high- end estimates ranged from two to ten times higher than the low-end estimates. We have used the lower end of the estimates in most cases because, as explained above, HIPAA does not require as extensive changes as envisioned by WEDI. The estimated percentages of health plans that accept electronic billing are based on reports in the 1997 edition of Faulkner & Gray’s Health Data Directory (5). The total cost for each type of health plan is the sum of the cost for EDI and non-EDI plans. Cost for EDI plans is computed as follows:

Total Entities x EDI % x Average Upgrade Cost x 1.5 (NOTE: As described above, the cost of changing manual processes is estimated to be half the cost of system changes.)

Cost for non-EDI plans is computed as follows:

Total entities x (1 - EDI %) x Average Upgrade Cost x .5 (NOTE: As described above, cost to non-EDI health care providers is assumed to be half the cost of systems changes.)

The $3.9 billion in savings is derived from Table 4, and represents savings to health plans for the first five years of implementation. The assumptions related to these savings are contained in the explanation to Table 4. The savings have been apportioned to each type of health plan based on the ratio of that health plan type’s cost to the cost to all health plans. For example, a plan type that incurs ten percent of the costs would be assigned ten percent of the savings. We acknowledge that this is an imprecise method for allocating savings. We have not been able to identify a reliable method for allocating savings to specific types of health plans but nonetheless believed that it was important to present costs and savings together in order to provide a sense of how the HIPAA administrative simplification provisions would affect various entities.

Table 1 - Health Plan Implementation Costs and Savings (in Millions) (1998-2002)

Table 2 illustrates the costs and savings attributable to various types of health care providers.

The number of entities (practices, not individual health care providers) is based on the 1992 Census of Services, the 1996 Statistical Abstract of the United States, and the American Medical Association survey of group practices trended forward to 1998. Estimated percentages of EDI billing are based on the 1997 edition of Faulkner & Gray’s Health Data Directory or are actuarial estimates.

The cost of software upgrades for personal computers (PCS) is based on reports on the cost of software upgrades to translate and communicate standardized claims forms. The low end is used for smaller practices and the high end for larger practices with PCS. The estimate for mainframe upgrade packages is twice the upper end for PCS. The cost per upgrade for facilities is ours after considering estimates by WEDI and estimates of the cost of new software packages in the literature. The estimates fall within the range of the WEDI estimates, but that range is quite large. For example, WEDI estimates the cost for a large hospital upgrade would be from $50,000 to $500,000. For an explanation of the method for computing Total Cost, see the explanation for Table 1.

The $3.4 billion in savings is derived from Table 4 and represents savings to health care providers for the first five years of implementation. We have included them here to provide a sense of how the HIPAA administrative simplification provisions would affect various entities. As in Table 1, the savings have been apportioned to each type of health care provider based on the ratio of that health care provider type’s cost to the cost to all health care providers.

Table 2 - Health Care Provider Implementation Costs and Savings (in Millions) (1998-2002)

Table 3 shows the estimates we used to determine the portion of EDI increase attributable to the HIPAA administrative simplification provisions. The proportion of claims that would be processed electronically even without HIPAA is assumed to grow at the same rate from 1998 through 2002 as it did from 1992 to 1996, except that the rate for hospitals, which is already high, is assumed to grow at one percent annually instead of the two percent that was observed from 1992-1996. The proportion of “other” provider claims is high because it includes pharmacies that generate large volumes of claims and have a high rate of electronic billing.

The increase attributable to HIPAA is highly uncertain and is critical to the savings estimate. Our actuary arrived at these estimates based on an analysis of the current EDI environment. Because the rate of growth in electronic billing is already high, there is not much room for added growth. On the other hand, much of the increase that has already occurred is attributable to Medicare and Medicaid; private insurers and third party administrators still have fairly low rates of electronic billing and may benefit significantly from standardization.

Table 3 - Percent Growth in EDI Claims Attributable to HIPAA AS Provisions (Cumulative)

Table 4 shows the annual costs, savings, and net savings over a five-year implementation period. We assume that the costs will be incurred within the first three years, since the statute requires health plans other than small health plans to implement within 24 months and small health plans to implement within 36 months. As each health plan implements a standard, health care providers that conduct electronic transactions with that health plan would also implement the standard. We assume that no savings would accrue in the first year, because not enough health plans and health care providers would have implemented the standards. Savings would increase as more health plans and health care providers implement, exceeding costs in the fourth year. At that point, the majority of health plans and health care providers will have implemented the standards, and costs will decrease and benefits will increase as a result.

The savings per claim processed electronically instead of manually is based on the lower end of the range estimated by WEDI. We have used $1 per claim for health plans and physicians, and $.75 per claim for hospitals and other health care providers. These estimates are based on surveys of health care providers and health plans. Savings per EDI claim are computed by multiplying the per claim savings times the number of EDI claims attributed to HIPAA. The total number of EDI claims is used in computing the savings to health plans, while the savings for specific health care provider groups is computed using only the number of EDI claims generated by that group (for example, savings to physicians is computed using only physician EDI claims).

WEDI also estimated savings resulting from other HIPAA transactions. The savings per transaction was higher than the savings from electronic billing, but the number of transactions was much smaller. Our estimates for transactions other than claims were derived by assuming a number of transactions and a savings per transaction relative to those assumed for the savings for electronic billing (see table 4a). In general our assumptions are close to those used by WEDI. One major difference is that we derived the number of enrollment/disenrollment transactions from Department of Labor statistics. We used their estimate of the number of events requiring a certificate to be issued, which includes such actions as starting or leaving a firm, children “aging out” of coverage and death of policyholder. That estimate is about 45 million events. We used WEDI’s estimate that the savings per transaction is about half that of billing transactions.

We also assumed that savings could be expected from simplifications in manual claims. The basic assumption is that the savings are ten percent (per transaction) of those that are projected for conversion to electronic billing. However, it is also assumed that the standards only gradually allow health care providers and health plans to abandon old forms and identifiers because of the many relationships that have been established with other entities that will require a period of overlap.

Table 4 - Five-Year Net Savings (in Billions of Dollars)

NOTE: Figures do not total due to rounding.

Table 4a shows the savings per nonclaim transaction as a multiple of claims savings per transaction and the ratio of transactions to number of claims. These values were used to determine the savings for nonclaims transactions.

Table 4a - Relative Savings and Volume of Other Transactions

H. Qualitative Impacts of Administrative Simplification

Administration simplification produces more than hard-dollar savings. There are also qualitative benefits that are less tangible, but nevertheless important. These changes become possible when data can be more easily integrated across entities. WEDI suggests in its 1993 report that there will be a “ripple-effect” of implementing an EDI infrastructure on the whole health care delivery system in that there would be a reduction in duplicate medical procedures and processes as a patient is handled by a continuum of health care providers during an episode of care. WEDI also suggests that there will be a reduction in the exposure to health care fraud as security controls on electronic transactions will prevent unauthorized access to financial data.

We also believe that having standards in place would reduce administrative burden and improve job satisfaction. For example, fewer administrative staff would be required to translate procedural codes, since a common set of codes would be used. All codes used in these transactions will be standardized, eliminating different values for data elements (for example, place of service).

Administrative simplification would promote the accuracy, reliability and usefulness of the information shared. For example, today there are any number of claims formats and identifiers in use. We estimate that there are over 400 variations of electronic formats for claims transactions alone. As we noted earlier, these variations make it difficult for parties to exchange information electronically. At a minimum, it requires data to be translated from the sender’s own format to the different formats specified by each intended receiver. Also, since industry has taken different approaches to uniquely identifying patients, health care providers and health plans (based on their individual business needs and preferences), it has become difficult to develop methods to compare services across health care providers and health plans. This mixed approach to enumeration has made it extremely difficult for health care researchers to do comparative analysis across settings and over time, and complicates identification of individuals for public health and epidemiologic purposes.

Administrative simplification greatly enhances the sharing of data both within entities and across entities. It facilitates the coordination of benefit information by having in place a standardized set of data that is known to all parties, along with standardized name and address information that tells where to route transactions. Today, health care providers are reluctant to file claims to multiple health plans on the behalf of the patient because information about a patient’s eligibility in a health plan is difficult to verify. Additionally, identifying information about health plans is not standardized or centralized for easy access. Most claims filed by patients today are submitted in hardcopy. We anticipate that more health care providers will file claims and coordinate benefits on the patient’s behalf once standard identifiers are adopted and this information is made available electronically.

I. Regulatory Flexibility Analysis

The Regulatory Flexibility Act (RFA) of 1980, Public Law 96- 354, requires us to prepare a regulatory flexibility analysis if the Secretary certifies that a proposed regulation would have a significant economic impact on a substantial number of small entities. In the health care sector, a small entity is one with less than $5 million in annual revenues. Nonprofit organizations are considered small entities; however, individuals and States are not included in the definition of a small entity. We have attempted to estimate the number of small entities and provide a general discussion of the effects of the statute. We request comments and additional information about our estimates and discussion.

All nonprofit Blue Cross-Blue Shield Plans are considered small entities. Two percent of the approximately 3.9 million employer health plans are considered small businesses. All doctors of osteopathy, dentists, podiatrists, chiropractors, and solo and group physicians’ offices with fewer than three physicians are considered small entities. Forty percent of group practices with 3 or more physicians and 90 percent of optometrist practices are considered small entities. Seventy-five percent of all pharmacies, medical laboratories, dental laboratories and durable medical equipment suppliers are assumed to be small entities.

We found the best source for information about the health data information industry to be Faulkner & Gray’s Health Data Dictionary. This publication is the most comprehensive we found of its kind. The information in this directory is gathered by Faulkner & Gray editors and researchers who called all of the more than 3,000 organizations that are listed in the book to elicit information about their operations. It is important to note that some businesses are listed as more than one type of business entity. That is because in reporting the information, companies could list themselves as up to three different types of entities. For example, some businesses listed themselves as both practice management vendors as well as claims software vendors because their practice management software was “EDI enabled.”

All the statistics referencing Faulkner & Gray’s come from the 1996 edition of its Health Data Dictionary. It lists 100 third party claims processors, which includes health care clearinghouses (5-33). Faulkner & Gray define third party claims processors as entities under contract that take electronic and paper health care claims data from health care providers and billing companies that prepare bills on a health care provider’s behalf. The third party claims processor acts as a conduit to health plans; it batches claims and routes transactions to the appropriate health plan in a form that expedites payment.

Of the 100 third party processors/clearinghouses listed in this publication, seven processed more that 20 million electronic transactions per month. Another 14 handled 2 million or more transactions per month and another 29 handled over a million electronic transactions per month. The remaining 50 entities listed processed less than a million electronic transactions per month. We believe that almost all of these entities have annual revenues of under $5 million and would therefore be considered small entities by our definition.

Another entity that is involved in the electronic transmission of health care transactions is the value added network. Value added networks are involved in the electronic transmission of data over telecommunication lines. We include value added networks in the definition of a health care clearinghouse. Faulkner & Gray list 23 value added networks that handle health care transactions (5, p. 544). After further discussion, the editors clarified that only 8 of the 23 would be considered “pure” value added networks. We believe that all of these companies have annual revenues of over $5 million.

A billing company is another entity involved in the electronic routing of health care transactions. It works primarily with physicians either in office or hospital-based settings. Billing companies, in effect, take over the office administrative functions for a physician; they take information such as copies of medical notes and records and prepare claim forms that are then forwarded to an insurer for payment. Billing companies may also handle the receipt of payments, including posting payment to the patient’s record on behalf of the health care provider. They can be located within or outside of the physician’s practice setting.

The International Billing Association is a trade association representing billing companies. The International Billing Association estimated that there are approximately 4500 billing companies currently in business in the United States. The International Billing Association’s estimates are based on the name and address of actual billing companies that it compiled in developing its mailing list. We believe all of the 4500 billing companies known to be in business have revenues under $5 million annually.

Software system vendors provide computer software applications support to health care clearinghouses, billing companies, and health care providers. They particularly work with health care providers’ practice management and health information systems. These businesses provide integrated software applications for such services as accounts receivable management, electronic claims submission (patient billing), record keeping, patient charting, practice analysis and patient scheduling. Some software vendors are also involved in providing applications for translating paper and nonstandard computer documents into standardized formats that are acceptable to health plans.

Faulkner & Gray list 104 physician practice management vendors and suppliers (5, p. 520), 105 hospital information systems vendors and suppliers (5, p. 444), 134 software vendors and suppliers for claims-related transactions (5, p. 486), and 28 translation vendors (5, p. 534). We were unable to determine the number of these entities with revenues over $5 million, but we assume most of these businesses would be considered small entities under our definition.

As discussed earlier in this analysis, the cost of implementing the standards specified in the statute are primarily one-time or short-term costs related to conversion. They were characterized as follows: software conversion, cost of automation, training, implementation problems, and cost of documentation and implementation guides. Rather than repeat that information here, we refer you to the beginning of this impact analysis.

1. Health care Providers and Health Plans

As a result of standard data format and content, health care providers and health plans that wish to do business electronically could do so knowing that whatever capital outlays they make are worthwhile, with some certainty of return on investment. This is because entities that exchange electronic health care transactions would be required to receive and send transactions in the same standard formats using the same health care provider and health plan identifiers. We believe this will be an incentive to small physicians’ offices to convert from paper to EDI. In a 1996 Office of the Inspector General study entitled “Encouraging Physicians to Use Paperless Claims,” the Office of the Inspector General and HCFA agreed that over $36 million in annual Medicare claims processing savings could be achieved if all health care providers submitting 50 or more Medicare claims per month submitted them electronically. Establishment of EDI standards will make it financially beneficial for many small health care providers to convert to electronic claim submissions, because all health plans would accept the same formats.

Additionally, we believe that those health care providers that currently use health care clearinghouses and billing agencies will see costs stabilize and potentially some cost reduction. This would result from the increased efficiency that health care clearinghouses and billing companies will realize from being able to more easily link with health care industry business partners.

2. Third Party Vendors

Third party vendors include third party processors/clearinghouses (including value added networks), billing companies, and software system vendors. While the market for third party vendors will change as a result of standardization, these changes will be positive to the industry and its customers over the long term. However, the short term/one time costs discussed above will apply to the third party vendor community.

a. Clearinghouses and Billing Companies

As noted above, health care clearinghouses are entities that take health care transactions, convert them into standardized formats acceptable to the receiver, and forward them on to the insurer. Billing companies take on the administrative functions of a physician’s office. The market for clearinghouse and billing company services will definitely be affected by the HIPAA administrative simplification provisions; however there appears to be some debate on how the market for these services will be affected.

It is likely that competition among health care clearinghouses and billing companies will increase over time. This is because standards would reduce some of the technical limitations that currently inhibit health care providers from conducting their own EDI. For example, by eliminating the requirement to maintain several different claims standards for different trading partners, health care providers will be able to more easily link themselves directly to health plans. This could negatively affect the market for health care clearinghouses and system vendors that do translation services; however, standards should increase the efficiency in which health care clearinghouses operate by allowing them to more easily link to multiple health plans. The increased efficiency in operations resulting from standards could, in effect, lower their overhead costs as well as attract new health care clearinghouse customers to offset any loss in market share that they might experience.

Another potential area of change is that brought about through standardized code sets. Standards would lower costs and break down logistical barriers that discouraged some health care providers from doing their own coding and billing. As a result, some health care providers may choose an in-house transaction system rather than using a billing company as a means of exercising more control over information. Conversely, health care clearinghouses may acquire some short-term increase in business from those health care providers that are automated but do not use the selected standards. These health care providers would hire health care clearinghouses to take data from the nonstandard formats they are using and convert them into the appropriate standards. Generally, we would also expect health care clearinghouses to identify opportunities to add value to transaction processing and to find new business opportunities, either in marketing promotional materials or in training health care providers on the new transaction sets. Standards would increase the efficiency of health care clearinghouses, which could in turn drive costs for these services down. Health care clearinghouses may be able to operate more efficiently or at a lower cost based on their ability to gain market share. Some small billing companies may be consumed by health care clearinghouses that may begin offering billing services to augment their health care clearinghouse activities. However, most health care providers that use billing companies would probably continue to do so because of the comprehensive and personalized services these companies offer.

Value added networks do not manipulate data but rather transmit data in its native form over telecommunication lines. We anticipate that the demand for value added network services would increase as additional health care providers and health plans move to electronic data exchange. Standards would eliminate the need for data to be reformatted, which would allow health care providers to purchase value added network services individually rather than as a component of the full range of clearinghouse services.

b. Software Vendors

As noted above, software vendors provide computer software applications support to health care clearinghouses and health care providers. They particularly work with health care providers’ practice management and health information systems. We believe these entities would be affected positively, at least in the short term. The implementation of administrative simplification would enhance their business opportunities as they would be involved in developing computerized software solutions that would allow for health care providers and other entities that exchange health care data to integrate the new transaction set into their existing systems. They may also be involved in developing software solutions to manage the crosswalk of existing health care provider and health plan identifiers to the national provider identifier and health plan identifier (PAYERID) until such time as all entities have implemented the identifiers.

J. Unfunded mandates

We have identified costs to the private sector to implement these standards. Although these costs are unfunded, we expect that they will be offset by subsequent savings as detailed in this impact analysis.

Most costs will occur in the first 3 years following the adoption of the HIPAA standards, with savings to health care providers and health plans exceeding costs in the fourth year. Five-year costs of implementing the HIPAA standards are estimated at $ 5.8 billion for health care providers and health plans combined. Savings to these entities over the same period in electronic claims processing, other electronic transactions (e.g., enrollments and disenrollments), and manual transactions are estimated at $ 7.3 billion, for a net savings of $ 1.5 billion in 5 years.

The costs to State and local governments and tribal organizations are also unfunded, but we do not have sufficient information to provide estimates of the impact of these standards on those entities. Several State Medicaid agencies have estimated that it would cost $1 million per state to implement all the HIPAA standards. However, the Congressional Budget Office analysis stated that “States are already in the forefront in administering the Medicaid program electronically; the only costs--which should not be significant--would involve bringing the software and computer systems for the Medicaid programs into compliance with the new standards.” The report went on to point out that Medicaid State agencies have the option to compensate by reducing other expenditures and that other State and local government agencies are likely to incur less in the way of costs since most of them will have fewer enrollees. Moreover, the Federal government pays a portion of the cost of converting State Medicaid Management Information Systems (MMIS) as Federal Financial Participation -- 75 percent for system maintenance changes and 90 percent for new software (if approved). Many States are in the process of changing systems as they convert many of the current functions in the move to enroll Medicaid beneficiaries in managed care.

K. Specific Impact of Provider Identifier

This is the portion of the impact analysis that relates specifically to the standard that is the subject of this regulation -- the health care provider identifier. This section describes specific impacts that relate to the provider identifiers. However, as we indicated in the introduction to this impact analysis, we do not intend to associate costs and savings to specific standards. In addition, this section assesses the relative cost impact of the various identifier options and implementation options set out in the regulation.

Although we cannot determine the specific economic impact of the standard being proposed in this rule (and individually each standard may not have a significant impact), the overall impact analysis makes clear that, collectively, all the standards will have a significant impact of over $100 million on the economy. Also, while each standard may not have a significant impact on a substantial number of small entities, the combined effects of all the proposed standards may have a significant effect on a substantial number of small entities. Therefore, the following impact analysis should be read in conjunction with the overall impact analysis.

In accordance with the provisions of Executive Order 12866, this proposed rule was reviewed by the Office of Management and Budget.

1. Affected entities.

a. Health care providers.

Health care providers that conduct electronic transactions with health plans would have to begin to use the NPI in those transactions. Health care providers that are indirectly involved in electronic transactions (for example, by submitting a paper claim that the health plan transmits electronically to a secondary payer) may also use the NPI. Any negative impact on these health care providers generally would be related to the initial implementation period. They would incur implementation costs for converting systems, especially those that generate electronic claims, from current provider identifiers to the NPI. Some health care providers would incur those costs directly and others would incur them in the form of fee increases from billing agents and health care clearinghouses.

Health care providers not only would have to include their own NPI on claims, but they would also have to obtain and use NPIs of other health care providers (for example, for referring and ordering). This would be a more significant implementation workload for larger institutional health care providers, such as hospitals, that would have to obtain the NPIs for each physician practicing in the hospital. However, these health care providers are accustomed to maintaining these types of data. There would also be a potential for disruption of claims processes and timely payments during a particular health plan’s transition to the NPI. Some health care providers that do not do business with government programs may be resistant to obtaining an NPI and providing data about themselves that would be stored in a national database.

Health care providers would also have to obtain an NPI and report changes in pertinent data. Under one of the enumeration options presented in this preamble, current Medicare providers will receive their NPIs automatically, and other health care providers may be enumerated in this manner to the extent that appropriate valid data files are available. New health care providers would have to apply for an NPI. This does not impose a new burden on health care providers. The vast majority of health plans issue identifiers to the health care providers with whom they transact business in order to facilitate the electronic processing of claims and other transactions. The information that health care providers must supply in order to receive an NPI is significantly less than the information most health plans require to enroll a health care provider. There would be no new cost burden; the statute does not support our charging health care providers to receive an NPI.

After implementation, health care providers would no longer have to keep track of and use different identifiers for different insurers. This would simplify provider billing systems and processes and reduce administrative expenses. A standard identifier would facilitate and simplify coordination of benefits, resulting in faster, more accurate payments. Under option 2 of the enumeration options, (see section IX.K.2.d. of this preamble, on enumerators), many health care providers (all those doing business with Medicare) would receive their NPIs automatically and would be able to report changes in the data contained in the NPS to a single place and have the changes made available to many health plans.

b. Health plans.

Health plans that engage in electronic commerce would have to modify their systems to use the NPI. This conversion would have a one-time cost impact on Federal, State, and private health plans alike and is likely to be more costly for health plans with complex systems that rely on intelligent provider numbers. Disruption of claims processing and payment delays could result. However, health plans would be able to schedule their implementation of the NPI and other standards in a manner that best fits their needs, as long as they meet the deadlines specified in the legislation.

Once the NPI has been implemented, health plans’ coordination of benefits activities would be greatly simplified because all health plans would use the same health care provider identifier. In addition, utilization review and other payment safeguard activities would be facilitated, since health care providers would not be able to use multiple identifiers and could be easily tracked over time and across geographic areas. Health plans currently assign their own identification numbers to health care providers as part of their enrollment procedures, and this would no longer be necessary. Existing enumeration systems maintained by Federal health programs would be phased out, and savings would result.

c. Health care clearinghouses.

Health care clearinghouses would face impacts (both positive and negative) similar to those experienced by health plans. However, implementation would likely be more complex, because health care clearinghouses deal with many health care providers and health plans and would have to accommodate both old and new health care provider identifiers until all health plans with which they deal have converted.

2. Effects of Various Options

a. Guiding Principles for Standard Selection

The implementation teams charged with designating standards under the statute have defined, with significant input from the health care industry, a set of common criteria for evaluating potential standards. These criteria are based on direct specifications in the HIPAA, the purpose of the law, and principles that support the regulatory philosophy set forth in Executive Order 12866 of September 30, 1993, and the Paperwork Reduction Act of 1995. These criteria also support and are consistent with the principles of the Paperwork Reduction Act of 1995. In order to be designated as a standard, a proposed standard should:

• Improve the efficiency and effectiveness of the health care system by leading to cost reductions for or improvements in benefits from electronic HIPAA health care transactions. This principle supports the regulatory goals of cost-effectiveness and avoidance of burden.
• Meet the needs of the health data standards user community, particularly health care providers, health plans, and health care clearinghouses. This principle supports the regulatory goal of cost-effectiveness.
• Be consistent and uniform with the other HIPAA standards -- their data element definitions and codes and their privacy and security requirements -- and, secondarily, with other private and public sector health data standards. This principle supports the regulatory goals of consistency and avoidance of incompatibility, and it establishes a performance objective for the standard.
• Have low additional development and implementation costs relative to the benefits of using the standard. This principle supports the regulatory goals of cost-effectiveness and avoidance of burden.
• Be supported by an ANSI-accredited standards developing organization or other private or public organization that will ensure continuity and efficient updating of the standard over time. This principle supports the regulatory goal of predictability.
• Have timely development, testing, implementation, and updating procedures to achieve administrative simplification benefits faster. This principle establishes a performance objective for the standard.
• Be technologically independent of the computer platforms and transmission protocols used in HIPAA health transactions, except when they are explicitly part of the standard. This principle establishes a performance objective for the standard and supports the regulatory goal of flexibility.
• Be precise and unambiguous, but as simple as possible. This principle supports the regulatory goals of predictability and simplicity.
• Keep data collection and paperwork burdens on users as low as is feasible. This principle supports the regulatory goals of cost-effectiveness and avoidance of duplication and burden.
• Incorporate flexibility to adapt more easily to changes in the health care infrastructure (such as new services, organizations, and provider types) and information technology. This principle supports the regulatory goals of flexibility and encouragement of innovation.

We assessed the various candidates for a provider identifier against the principles listed above, with the overall goal of achieving the maximum benefit for the least cost. We found that the NPI met all the principles, but no other candidate identifier met all the principles, or even those principles supporting the regulatory goal of cost-effectiveness. We are assessing the costs and benefits of the NPI, but we did not assess the costs and benefits of other identifier candidates, because they did not meet the guiding principles. We invite your comments on the costs and benefits of the alternative candidate NPI options for the various market segments.

b. Need to Convert

Because there is no standard provider identifier in widespread use throughout the industry, adopting any of the candidate identifiers would require most health care providers, health plans and health care clearinghouses to convert to the new standard. In the case of the NPI, all health care providers would have to convert because this identifier is not in use presently. As we pointed out in our analysis of the candidates, even the identifiers that are in use are not used for all purposes or for all provider types. The selection of the NPI does not impose a greater burden on the industry than the nonselected candidates, and presents significant advantages in terms of cost- effectiveness, universality, uniqueness and flexibility.

c. Complexity of Conversion

Some existing provider identifier systems assign multiple identifiers to a single health care provider in order to distinguish the multiple identities the health care provider has in the system. For example, in these systems, the health care provider may have a different identifier to represent each “pay- to” identity, contract or provider agreement, practice location, and specialty or provider type. Since the NPI is a unique identifier for each health care provider, it would not distinguish these multiple identities. Systems that need to distinguish these identities would need to use data other than the NPI to do so. The change to use other data would add complexity to the conversion to the NPI or to any other standard provider identifier, but it is necessary in order to achieve the goal of unique identification of the health care provider.

The complexity of the conversion would also be significantly affected by the degree to which health plans’ processing systems currently rely on intelligent identifiers. For example, a health plan may route claims to different processing routines based on the type of health care provider by keying on a provider type code included in the identifier. Converting from one unintelligent identifier to another is less complex than modifying software logic to obtain needed information from other data elements. However, the use of an unintelligent identifier is required in order to meet the guiding principle of assuring flexibility.

Specific technology limitations of existing systems could affect the complexity of conversion. For example, some existing provider data systems use a telephone keypad to enter data. Data entry of alpha characters is inconvenient in these systems. In order to mitigate this inconvenience, we would implement the NPI by initially assigning numeric NPIs. After all numeric possibilities have been exhausted, we would introduce alpha characters in one position at a time. This implementation strategy would allow additional time for systems with technology limitations to overcome conversion difficulties.

In general, the shorter the identifier, the easier it is to implement. It is more likely that a shorter identifier, such as the NPI, would fit into existing data formats.

The selection of the NPI does not impose a greater burden on the industry than the nonselected candidates.

d. Enumerators

Based on the analysis discussed earlier in the preamble, we assess the two most viable combinations of choices for the entities that would enumerate health care providers. We do not assess choices that permit large numbers of enumerators (for example, all health plans, educational institutions, professional associations) because these choices do not satisfy the critical programmatic requirements of maintaining a high degree of data quality and consistency and minimizing confusion for health care providers.

No matter which of the two enumeration options is chosen, certain costs and impacts would not vary.

• We assume that the NPS would be used in both options to generate NPIs and serve as the central enumeration system and database. We began to develop the NPS for Medicare use, and this effort, which was funded by HCFA, is now nearing completion. As the NPS becomes national in scope, we estimate that the cost of maintaining the NPS software, hardware, and telecommunications, and operating a Help Desk to deal with user questions, would cost approximately $10.4 million over the first three years of operation and approximately $2.9 million per year thereafter. Roughly half of these costs are attributable to telecommunications expenses. This analysis presumes the availability of Federal funds to support the development and operations of the NPS. However, we are seeking comments on how the NPS could be funded once it becomes national.
• We further assume that, in both options, the same implementation strategy of loading the NPS database using health plans’ existing prevalidated files will be utilized to the extent possible. This would reduce costs by not repeating the process of soliciting, receiving, controlling, validating and keying applications from health care providers that have already been enumerated by a trusted source. For example, we would use existing Medicare provider files to initially load the NPS database. The majority of work to reformat and edit these files has already been completed.

We estimate that approximately 1.2 million current health care providers and 30,000 new health care providers annually would require NPIs because they conduct HIPAA transactions.

An additional 3 million health care providers (120,000 new health care providers annually) do not conduct HIPAA transactions, but they may choose to be enumerated at some future time. We refer to these health care providers as “non-HIPAA-transaction health care providers” (see section 4. Enumeration Phases of this preamble). These health care providers would be primarily individual practitioners such as registered nurses and pharmacists who perform services in institutions and whose services are not billed by the institution. More research is required on the time frame and process for enumerating these health care providers.

Based on Medicare carriers’ costs, we have estimated that the average cost to enumerate a health care provider should not exceed $50. Enumeration activities would include assisting health care providers and answering questions, accepting the application for an NPI; validating as many of the data elements as possible at the point of application to assure the submitted data are accurate and the application is authentic; entering the data into the NPS to obtain an NPI for the health care provider; researching cases where there is a possible match to a health care provider already enumerated; notifying the health care provider of the assigned NPI; and entering updated data into the NPS when notified by the health care provider. The cost of processing a data update is not known, and for purposes of this analysis we are assuming an average cost of $10 per update transaction, and that 5 percent per year of these health care providers on file would have updated data. However, we estimate that approximately 15 percent of health care providers that do not conduct business with Federal health plans or Medicaid would require updates each year. These health care providers may be unfamiliar with the terminology for some of the information they need to provide in order to be enumerated; thus, they may need to correct errors they could have made in completing the applications for NPIs or may have a need to change some of that information for other reasons. The per transaction cost would be lower if practice location addresses and membership in groups were not collected (see section IV., Data, and section IX.E., Maintenance of the Database, of this preamble) and if enumerators were already validating data as part of their own enrollment processes. The number of updates would also be affected by the practice location and group membership issues because these data are more volatile than demographic data (see IV., Data, and IX.E., Maintenance of the Database, of this preamble).

For a similarly sized commercial numbering system that uniquely identifies corporations and assigns unique identifiers, we have received independent estimates from Dun & Bradstreet (D&B) of $7 per enumeration and $3 per update. The D&B estimates are based on the cost of assigning and maintaining the Data Universal Numbering System (D-U-N-S) number. The D-U-N-S number is a nine- digit, non-indicative number assigned to each record in D&B’s file. It uses a modulus 10 check digit in the ninth position. Over 47 million D-U-N-S numbers have been assigned, worldwide, with 22 million attributed to locations in the United States. D&B uses the D-U-N-S number to enumerate businesses, including commercial sites, sole proprietorships, cottage industries, educational institutions, not-for-profits, and government entities, but does not maintain records on private individuals. D&B estimates an average cost of $7 to add a record to its database and assign it a unique record identifier. To establish a record and ensure uniqueness, D&B requires the entity's legal name, any "doing business as" names, physical address, telephone number, chief executive, date started, line of business, number of employees and relationship(s) with other business entities. D&B runs a daily computer process to audit all records added during the day and extracts any that may be duplicates for research by an analyst. Updates to each record are estimated at approximately $3 but can run as high as $30 per year for very robust database entries, some of which contain 1500 different data elements. The D&B estimates may be understated for our purposes because the four to six data elements used to uniquely identify the enumerated corporations do not require verification. We welcome comments on which data elements are required to uniquely identify health care providers (individuals, groups, and organizations), on whether verification of the data is necessary for purposes of enumeration, and on estimates of the cost to enumerate and update that minimum data set. We understand that the cost would be lower if the number and complexity of the data elements were reduced, but this cost must be balanced against the level of confidence that can be placed in the uniqueness of the health care providers identified. Specific consideration of these tradeoffs in submitted comments will be very helpful.

The $50 estimated average cost to enumerate a health care provider is an upper limit. The cost would decrease significantly if the second data alternative is selected (see section IV.B., Practice Addresses and Group/Organization Options, of this preamble). Under this alternative, the NPS would capture only one practice address for an individual or organization provider. It would not assign location codes. The NPS would not link the NPI of a group provider to the NPIs of individuals who are members of the group. Costs would decrease because we would collect significantly less data at the time of enumeration, and the data that would be collected would not need to be updated very frequently. Recent consultations with the industry reveal a growing consensus for this alternative.

Table 5 below provides estimates as to the cost of each enumeration option for start-up and outyear, with Federal, State, and private costs, for HIPAA-transaction and non-HIPAA-transaction health care providers, and the Federal costs of the NPS. We define “start-up” as the first 3 years during which the NPS becomes operational nationally and the bulk of the health care providers requiring NPIs are enumerated. “Outyear” would be each subsequent year, in which the majority of actions would be enumerations of new health care providers and provider updates. Assumptions follow the table.

Table 5 - Enumeration Costs: Federal, State, and Private

ENUMERATION COSTS: Federal, State, and Private

ASSUMPTIONS:

1. Definitions:

   a. “HIPAA-transaction health care provider” means a health care provider that we would require to have an NPI; that is, a health care provider that must be identified in the transactions specified in HIPAA.

   b. “Non-HIPAA-transaction health care provider” means a health care provider that we would not require to have an NPI.

   c. “Start-up” means the first 3 years in which the NPS becomes operational nationally and the bulk of the health care providers requiring NPIs are enumerated. It is the sum of the cost of enumerating existing health care providers in the first year plus the annual cost of enumerating new and updating existing health care providers for the 2 subsequent years.

   d. “Outyear” means each subsequent year in which the majority of actions would be enumerating new health care providers and updating existing ones. It is the sum of the cost of enumerating new health care providers plus the cost of updating existing health care providers.

2. The cost to enumerate a health care provider that is not enrolled or enrolling in a Federal health plan (e.g., Medicare, CHAMPUS) or Medicaid is estimated to be $50. (See Assumption 4.)
3. The cost to update information on a health care provider that is not enrolled or enrolling in a Federal health plan (e.g., Medicare, CHAMPUS) or Medicaid is estimated to be $10. (See Assumption 4.)
4. The cost to Federal health plans (e.g., Medicare, CHAMPUS) and Medicaid to enumerate or update their own health care providers is relatively small as these health plans must collect the same information to enroll or update the health care providers in their own programs. Possible up-front costs to these health plans and Medicaid would be offset by simpler, more efficient coordination of benefits, elimination of the need to maintain multiple enumeration systems, and elimination of the need to maintain other provider numbers. The Federal Government pays 75 percent of Medicaid State agencies’ costs to enumerate and update health care providers. Because all of these costs are relatively small and would be offset by savings, they are considered to be $0 (zero).
5. This analysis presumes the availability of Federal funds to support the registry.
6. It is estimated that 5 percent of existing HIPAA-transaction health care providers that conduct business with Federal health plans or Medicaid require updates annually; 15 percent of the remaining HIPAA-transaction health care providers require updates annually.
7. It is estimated that 5 percent of Medicaid State agencies may decline to participate in enumerating/updating their health care providers. The registry would enumerate/update that 5 percent.
8. Non-HIPAA-transaction health care providers would not be enumerated in the initial phases of enumeration. These costs are estimated to be $165,000,000 for start-up and $7,500,000 for outyear. The registry would enumerate/update these health care providers only if funds are available.

Option 1 calls for all 1.2 million HIPAA-transaction health care providers to be enumerated by a Federally-directed registry. The one-time cost for the registry to assign NPIs to existing HIPAA-transaction health care providers would depend on the extent to which existing files could be used. The cost could be as high as $60 million (1.2 million health care providers x $50) or as low as $9 million (see option 2). The low estimate assumes that prevalidated provider files are available for 100 percent of all Federal and Medicaid providers. The annual outyear cost would be $2.1 million (30,000 new health care providers x $50 plus 60,000 updates x $10). The Federal health plans and Medicaid State agencies would no longer have to assign their own identifiers, which would result in some savings, but they would still incur costs related to provider enrollment activities that would duplicate Federally-directed registry functions (for example, duplicate collection and verification of some information).

Option 2 calls for enumeration of HIPAA-transaction health care providers to be performed by a combination of Federal programs named as health plans, Medicaid State agencies, and a Federally-directed registry. This registry would enumerate non- Federal, non-Medicaid providers. All enumerators would receive, validate, and enter application data into the NPS and would communicate with health care providers. Data files would be available from a central source. The registry would utilize the NPS and would be operated under Federal oversight but could, if appropriate, be contracted out.

Medicare, Medicaid, CHAMPUS, and the Department of Veterans Affairs already assign identifiers to health care providers with whom they conduct business. They would simply begin to use the NPS to issue NPIs instead of using their own systems to assign the identifiers they now use. Initially, these Federal health plans and Medicaid may incur up-front costs in issuing NPIs; however, these additional costs would be offset by savings from the fact that each health care provider would only have to be enumerated once; multiple enumeration systems would not have to be maintained; other provider numbers would not have to be maintained; and coordination of benefits would be simpler and more efficient. We estimate that approximately 5 percent of Medicaid State agencies may decline to participate (that is, they would not enumerate and update their health care providers). These health care providers would need to be enumerated and updated by the Federally-directed registry; however, that cost would be offset by savings realized by the discontinuance of UPIN assignment and maintenance of the UPIN registry. We estimate that approximately 85 percent of the health care providers that conduct HIPAA transactions would be enumerated in this manner (75 percent by Federal health plans, 10 percent by Medicaid). Additional costs, if any, to enumerate these health care providers or update their data would be insignificant.

The remaining 15 percent of health care providers that conduct HIPAA transactions (180,000) would be enumerated by a Federally-directed registry. The one-time cost of enumerating these health care providers would be $9 million (180,000 health care providers x $50). The cost of enumerating 4,500 new health care providers would be $225,000 per year, and the cost to process 27,000 updates would be $270,000, for a total registry cost of $495,000 per outyear.

Based on the cost estimates in this analysis, option 1 is considerably more expensive than option 2. We believe option 2 to be preferable to option 1 in that Federal programs and Medicaid State agencies would enumerate and update their own health care providers. The enumeration functions of the 5 percent of Medicaid State agencies that may decline to enumerate and update their own health care providers would fall to the Federally-directed registry.

The initial and ongoing cost of developing, implementing and operating the NPS would be borne by the Federal government, depending on the availability of funds; some of this cost could be offset by ceasing current enumeration systems like Medicare’s UPIN registry.

The previous analysis relates only to health care providers that are required to have an NPI to perform HIPAA transactions. The remaining health care providers would not be required to obtain an NPI but could do so if they wished to have one for other reasons. We indicated in the Implementation section of this preamble that we would not issue NPIs to these health care providers until the health care providers that needed NPIs to conduct any of the electronic transactions specified in HIPAA had been enumerated. The cost of enumerating the approximately 3 million non-HIPAA-transaction health care providers could be as high as $150 million (3 million health care providers x $50). We are soliciting comments on sources of information on non-HIPAA-transaction health care providers. We cannot provide a realistic estimate of the cost of enumerating these health care providers without this additional input.

e. Maintenance of the Database

Another cost implication is the maintenance of the database being developed by the NPS. (We discuss this cost implication in more detail in section IV. Data but believe the general discussion should be repeated here in the impact analysis as well.) That database, known as the National Provider File (NPF), is currently being designed to contain the data elements shown in the table entitled, “National Provider File Data Elements” in section IV. Data, A. Data Elements, earlier in this preamble. The majority of the information is used to uniquely identify a health care provider; other information is used for administrative purposes. A few of the data elements are collected at the request of potential users that have been working with HCFA in designing the database prior to the passage of HIPAA. All of these data elements represent only a fraction of the information that would comprise a provider enrollment file. The data elements shown in the “National Provider File Data Elements” table earlier in the preamble, plus cease/effective/termination dates, switches (yes/no), indicators, and history, are being considered as those that would form the NPF. The table includes appropriate comments. The table does not display systems maintenance or similar fields, or health care provider cease/effective/termination dates.

We need to consider the benefits of retaining all of the data elements shown in the table versus lowering the cost of maintaining the database by keeping only the minimum number of data elements needed for unique provider identification. We solicit input on the composition of the minimum set of data elements needed to uniquely identify each type of health care provider. In order to consider the inclusion or exclusion of data elements, we need to assess their purpose and use.

The data elements in the table with a purpose of “I” are being proposed to identify a health care provider, either in the search process (which is electronic) or in the investigation of health care providers designated as possible matches by the search process. These data elements are critical because unique identification is the keystone of the NPS.

The data elements in the table with a purpose of “A” are not essential to the identification processes mentioned above, but they nonetheless are valuable. Certain “A” data elements can be used to contact a health care provider for clarification of information or resolution of issues encountered in the enumeration process and for sending written communications; other “A” data elements (e.g., Provider Enumerate Date, Provider Update Date, Establishing Enumerator/Agent Number) are used to organize and manage the data.

The data elements in the table with a purpose of “U” are collected at the request of potential users of the information in the system. While not used by the system’s search process to uniquely identify a health care provider, Race (with a purpose of “U”) is nevertheless valuable in the investigation of health care providers designated as possible matches as a result of that process. In addition, Race is important to the utility of the NPS as a statistical sampling frame. Race is collected “as reported”; that is, it is not validated. It is not maintained, only stored. The cost of keeping this data element is virtually nil. Other data elements (Resident/Intern Code, Provider Certification Code and Number, and Organization Type Control Code) with a purpose of “U”, while not used for enumeration of a health care provider, have been requested to be included by some members of the health care industry for reports and statistics. These data elements are optional and do not require validation; many remain constant by their nature; and the cost to store them is negligible.

The data elements that we judge will be expensive to either validate or maintain (or both) are the license information, provider practice location addresses, and membership in groups. We solicit comments on whether these data elements are necessary for the unique enumeration of health care providers and whether validation or maintenance is required for that purpose.

Licenses may be critical in determining uniqueness of a health care provider (particularly in resolving identifies involving compound surnames) and are, therefore, considered to be essential by some. License information is expensive to validate initially, but it is not expensive to maintain because it does not change frequently.

The practice location addresses can be used to aid in investigating possible provider matches, in converting existing provider numbers to NPIs, and in research involving fraud or epidemiology. Location codes, which are discussed in detail in section B. Practice Addresses and Group/Organization Options of this preamble, could be assigned by the NPS to point to and identify practice locations of individuals and groups. Some potential users felt that practice addresses changed too frequently to be maintained efficiently at the national level. The average Medicare physician has two to three addresses at which he or she practices. Group providers may have many more practice locations. We estimate that 5 percent of health care providers require updates annually and that addresses are one of the most frequently changing attributes. As a result, maintaining more than one practice address for an individual provider on a national scale could be burdensome and time consuming. Many potential users believe that practice addresses could more adequately be maintained at local, health-plan specific levels.

Some potential users felt that membership in groups was useful in identifying health care providers. Many others, however, felt that these data are highly volatile and costly to maintain. These users felt it was unlikely that membership in groups could be satisfactorily maintained at the national level.

We welcome comments on the data elements proposed for the NPF and input as to the potential usefulness and tradeoffs for these elements such as those discussed above.

References:

1. Dobson, Allen, Ph.D. and Bergheiser, Matthew; “Reducing Administrative Costs in a Pluralistic Delivery System through Automation;” Lewin-VHI Report prepared for the Healthcare Financial Management Association; 1993.
2. Congressional Budget Office; “Federal Cost Estimate for H.R. 3070;” 1996.
3. Workgroup for Electronic Data Interchange; “Report,” 1993.
4. “Electronic Network Solution for Rising Healthcare Costs;” New Jersey Institute of Technology and Thomas Edison State College, 1995.
5. Faulkner & Gray’s Health Data Directory, 1997 Edition; Kurt T. Peters, Publisher (also earlier editions).