Description and Analysis of the VA National Formulary (2000)

In this report, a committee of the Institute of Medicine (IOM) describes and analyzes the role of the Department of Veterans Affairs' (VA) National Formulary in veterans health care. In requesting this study, the Congress and the Department of Veterans Affairs (VA) asked that four aspects of the National Formulary be evaluated: its restrictiveness, its effects on both cost and quality, and how it compares with other formularies. After an introductory chapter that provides background and context, these evaluations are reported in Chapter 2, Chapter 3, Chapter 4 and Chapter 5, each of which covers one of these aspects. This final chapter is a narrative summary of the preceding chapters. The specific conclusions and recommendations of the committee are found in the Executive Summary of this report.

In 1998, House Report 105-610 noted that serious concerns had been raised about the impact of the VA National Formulary and directed the VA to contract with the IOM to conduct an independent analysis of its effects on quality of care and potential costs and to compare it with other public and private insurance formularies. The IOM and the VA entered into a contract to address the congressional concerns, effective April 12, 1999. Implementation of this contract was begun with the appointment of staff and a committee of independent experts to carry out the necessary work. This committee consisted of representatives of two veterans service organizations; health professionals knowledgeable in clinical pharmacology, pharmacy and therapeutics activities, and clinical medicine and geriatrics; and pharmacists and others with experience in managed care and pharmacy benefits management. The committee delivered an interim report on January 28, 2000, and this final report was scheduled for delivery not later than July 11, 2000.

The history of formularies dates back hundreds or even thousands of years in other parts of the world, and to the American Revolution in this country. Formularies began as simple lists of remedies and their formulas. In the United States, they developed along with changes in health care delivery and the science of pharmacology. Primarily used in hospitals, they gradually came to include purified and standardized drugs identified by generic nomenclature. More recently, they have focused on the cost-effectiveness as well as the quality of drugs that are included and controlled. Beginning in the 1950s, professional societies, government programs, and accreditation agencies began to define formularies and to require them in health facilities. The facilities and organizations using formularies evolved with changes in health care financing and delivery to include managed care plans, pharmacy benefit management organizations, all kinds of health care settings, and government programs such as Medicaid, the Department of Defense, and the Veterans Health Administration (VHA). Veterans health facilities have used formularies to control inventory and the cost-effectiveness and quality of drug treatment for veterans for the last 40 or 50 years.

Fundamentally, formularies are lists of drugs that may be more or less inclusive. They can be differentiated by the formulary system, that is, the restrictions, controls, or modifications that are employed in their management to achieve objectives for the pharmacy benefits of a health care system. At the simplest level, formularies may be open—that is, they list many drugs and place few limits on access or coverage; they may be closed, in which case, they list a limited number of drugs and place more limits on access or coverage; or they may be partially closed.

Aside from listing or not listing a drug in the formulary itself, the limits or controls that characterize a formulary and formulary system include generic prescribing, generic substitution, therapeutic interchange, use of step protocols, restrictions by certain specialties or clinical settings or conditions, nonformulary exceptions and prior approval or authorization processes, prescription copayments that vary in amounts and differ for generics and brands covered or not covered in the formulary, specific exclusions of drugs or drug classes, or closure of drug classes and designation of drugs or drug classes as preferred, among others. These are defined and discussed in the body of this report. Limits on prescription size, numbers, dollar values, or frequencies are restrictions or controls parallel to direct formulary management that are also discussed in this report, primarily in Chapter 2. Committees made up of practicing physicians, pharmacists, and some other professionals in a health system (pharmacy and therapeutics [P&T] committees), drug class reviews, and treatment guidelines are also important to decision making and management of formularies and formulary systems.

The VHA consists of 146 medical centers, including 172 hospitals, more than 600 ambulatory care and community-based clinics, 132 nursing homes, 40 domiciliaries, and a number of other programs. In 1995, a reorganization of the VA affected the status of all VA facilities and the relationships of the VA formularies and formulary systems, the control of drug use, and pharmacy operations. This reorganization created 22 Veterans Integrated Service Networks (VISNs), that is, essentially 22 managed care organizations with their own capitation-based budgets. Each VISN, on average, encompasses 7 to 10 hospitals, 25 to 30 ambulatory clinics, 4 to 7 nursing homes, 1 or 2 domiciliaries, and various other assets. The VHA is a unique health care system (and different from private-sector managed care) in terms of its size, cost, and budgeting; diversity of settings; geographic scope; role in the use and training of young physicians; and its permanently eligible patient population.

In November 1995, a VISN-level formulary was required in order to provide a uniform drug benefit in each network or region and to prepare for a national formulary. At about the same time, a VA Pharmacy Benefits Management Strategic Health Care Group (VA PBM), a central Medical Advisory Panel (MAP), VISN formulary committees, and a VISN Formulary Leaders Committee were created. VISN formularies were then merged into the VA National Formulary which was issued in May 1997. In October 1997, VHA Directive 97-047 required each VISN to develop a nonformulary exceptions process and specified criteria for granting exceptions. In December 1997, supplies were added to the National Formulary. Additional minor changes have been made periodically.

The VA National Formulary is controlled by the VA PBM, the MAP, and the VISN formulary leaders. These bodies issue drug class reviews and treatment guidelines, and make decisions on drug additions to and deletions from the formulary and the designation of closed and preferred classes. VISN formularies differ among themselves and from the National Formulary, but local formularies are usually the same as the VISN formulary. Specific restrictions by specialty, setting, or condition may differ among the formularies depending on local practice patterns, antibiotic resistance profiles, and the like. At present, there are four closed classes, that is, therapeutic classes in which a selection has been made among member drugs. Those selected must be listed and used systemwide. Those not selected cannot be listed on a formulary at any level and can be used only by nonformulary exception. Two classes are preferred; that is, a selection has been made among drugs in the class, and the drugs selected are preferred and subject to national contracts that provide for favorable prices. Other drugs can be listed on VISN or local formularies, however.

As described further in the cost sections of this report, drug prices depend on competition—the ability to choose among competitors, to purchase in volume, and to enforce compliance with market share agreements. By choosing among drugs in a class, the National Formulary and formulary system enable prudent purchasing of drugs. For the VA, the National Acquisition Center

(NAC), which administers the Federal Supply Schedule (FSS), uses National Formulary selections in closed or preferred classes as leverage to make market share or volume commitments and to negotiate prices that often are substantially lower than the already favorable prices for brand drugs on the FSS. National Formulary cost savings achieved through these selections and prices are supplemented by mandatory generic prescribing and substitution, negotiation of favorable prices among genetic manufacturers, and a substantial number of blanket purchase agreements that take advantage of various market conditions from time to time to get good prices on a range of products at the regional (or occasionally national) level.

Restrictiveness is a multifactorial attribute of a formulary and formulary system. It is a measure of the stringency of the controls on veterans' access to prescribed medicines at the appropriate times. If formulary structure or formulary system controls deny or significantly delay access to drugs that, in the reasonable judgment of medical experts, are clinically indicated, then the VA National Formulary meets the definition of overly restrictive. National Formulary elements of restrictiveness include formulary size, number of classes closed and number of drugs in closed classes, timeliness of addition of newly approved drugs, appropriateness and responsiveness of the nonformulary exceptions process, sensitivity of therapeutic interchange policies to patient risks, coverage of over-the-counter (OTC) drugs, and generic substitution. Other limits incorporated into restrictive designs include exclusions of drugs or drug classes, volume or quantity limits, high copayments, and prior approvals. Restrictiveness can be judged by comparison among public and private formularies and formulary systems, by comparison to reasonableness standards in the literature or in the informed judgment of the IOM committee, by comparison to objective standards where these exist, and by effects on satisfaction of patients and prescribers.

The VA National Formulary (July 1999 version) contains about 1,200 items, of which about 170 are OTC and 133 are medical-surgical supplies. Although there is no standard that specifies a particular formulary size for a given health care system, most managed care formularies contain less than 1,000 items. Also, before 1997, some VISNs were functioning, apparently satisfactorily, with formularies about 70% the size of the current National Formulary. The VA closed classes account for about 13 to 16% of VHA drug costs. Drugs in closed classes have important therapeutic effects. The conditions they treat are prevalent. Many prescriptions are written for these drugs. These classes are in the top five contributors to drug costs in managed care, and they are frequently closed in managed care and hospital formularies.

VA choices of classes to close and of drugs within closed or preferred classes are based on good-quality drug class reviews. There is no convincing evidence that choice is overly restricted in VA closed and preferred classes. Unlike Medicaid, managed care, or PBM plans, the National Formulary rarely

designates drugs or drug classes that are excluded from coverage. Sildenafil citrate (Viagra), which was excluded, is now available although with limitations. The effects of class closure and committed-use contracts on utilization of drugs in the VA are impressive. Preferred or formulary agent market shares in some classes are driven to 95% or more. The committee found no scientific evidence that this detracted from, or posed risks to, veterans' care.

Drugs newly approved by the Food and Drug Administration (FDA) are considered for addition to the VA National Formulary only after a 1-year delay, except in special cases of important new (FDA priority) 1P category drugs. The VA considers this a safety precaution, allowing evidence of adverse drug effects or studies on comparative safety, efficacy, and cost-effectiveness to accumulate during that interval. In practice, few comparative studies are published in the first year after market entry of a new drug. Most FDA recalls in the 1990s involved drugs that were not being actively marketed, occurred well after the first year on the market (and in one case, troglitazone, was available in some of the VISNs in the VHA), were not 1P drugs in the first place, or were, in fact, on the National Formulary (cisapride). Although most VISNs can add new drugs without delay, the blanket national policy of delay may occasionally protect veterans from exposure to a drug that will be recalled but has not always protected veterans from a problem drug. Given the higher cost of new drugs, costs will probably be avoided. However, veterans will not have the advantage of new treatments. Also, the VA policy is more restrictive than Medicaid and managed care policies on new drugs. The committee recommended that this policy be abandoned in favor of examining new drugs on their merits as they appear.

By the end of 1999, a net of 20 mostly existing drugs (that is, drugs already on the market) had been added to the National Formulary. VISN formularies, in the aggregate, had added 260 different, mostly existing drugs. In addition, VISN formularies differed from the National Formulary from the outset, sometimes by more than 100 items. VISN and local policies and procedures on drug additions, nonformulary exceptions, and therapeutic interchange—all important attributes of formulary systems—are different from the National Formulary and vary among themselves, as discussed further below. These differences and inconsistencies across the VHA potentially expose veterans to variable access to drugs and restrictions on drug treatment. The tension between national standardization and uniformity and local autonomy and preferences is difficult to balance, depending on time and place. Recalibration in favor of a more consistent, uniform national approach for these key attributes is desirable.

A nonformulary exceptions process should involve procedures for obtaining nonformulary drugs that are simple, fair, and reasonable and do not involve needless delays and complicated technicalities. In 1999, the VHA reported that 3.45% of total prescriptions, and 4 to 6% of prescriptions in closed classes, were filled with nonformulary drugs. Apparently, compliance with the National Formulary is excellent. According to a standard in the literature, hospital formularies were deemed restrictive if less than 5% of the pharmacy budget was spent on

nonformulary drugs, and in managed care, a survey reported 10% of prescriptions filled with nonformulary drugs.

The VA nonformulary exceptions process is often informal and variable across VISNs and facilities. Not all VISNs have standard forms. The procedures for assessing and acting on a request appear to require different amounts of time and impose different administrative burdens in different VISNs. Because some requests are unrecorded, program statistics are not reliable. The National Formulary would appear fairer and more responsive if the nonformulary system were revised and simplified, and reporting was accurate and consistent.

In closed or preferred classes, anticipated or promised volume of drug use supports better negotiated prices. There is, therefore, an expectation that VHA prescribers will discontinue prescribing nonformulary or noncontract drugs so that veterans are not started on them. Alternatively, veterans on nonformulary or noncontract drugs may be converted to the formulary therapeutic alternates. The VA PBM, MAP, and VISNs have left policy and procedure on such therapeutic interchanges to local facilities, although directives leading to them often originate at the national or VISN level. These interchanges are often made without the authorization of individual prescribers at the time of dispensing.

Therapeutic interchange is an accepted practice driven by price differentials and market share commitments in about half of managed care plans. Very often MCO interchanges involve the same drug classes as VA interchanges. They almost always are carried out with the permission of the prescriber at the time of dispensing. Therapeutic interchange is common in hospitals, generally also in the same drug classes. Since hospital medical staffs agree in advance to interchanges, individual prescriber permission at the time of dispensing is usually not sought. Therapeutic interchange is not practiced in Medicaid fee-for-service. It is uncertain how frequent it is in Medicaid managed care or Department of Defense (DOD) programs.

Although evaluations of VA therapeutic interchanges in the medical literature are generally reassuring, these studies frequently have methodological deficiencies; they can generate discontent, and occasional problems are reported. VA interchanges respond to national contract (and sometimes blanket purchase agreement) volume commitments and negotiated price differentials. They are left to local facilities without national VHA guidelines or written policies at the VISN level, however. Some veterans report not having received adequate, or any, information on the replacement drug, and some physicians have registered complaints. There should be consistency throughout the VHA in ensuring interchange program quality, patient and prescriber acceptance, adequate advance notice and education, and protection against risks to certain vulnerable patients. A responsive nonformulary process would also help provide this assurance.

Since VA contracts with drug manufacturers are renewable annually, they and the prices they set may change, obligating new interchanges. At some point, the VHA will have to evaluate how frequently veterans taking a drug chronically should be subjected to interchange or determine the total number of interchanges that is reasonable for an individual patient.

The IOM committee compared the coverage of OTC drugs and the use of generic substitution in the VA National Formulary with coverage and use in other formularies and formulary systems. Few health maintenance organizations (HMOs) offer OTC coverage as a specific pharmacy benefit, and the number of OTC drugs or drug classes covered in managed care is limited mostly to insulin and diabetic supplies with lesser coverage of antihistamines, histamine₂ receptor (H₂R) blockers, nonsteroidal anti-inflammatory drugs (NSAIDs), and cough and cold remedies. Medicaid OTC coverage is also less generous than that of the National Formulary. In comparison to other public and private-sector formularies the VA is less restrictive in this respect.

Generic substitution is an accepted medical practice. In hospitals, it is usually automatic. About half of the states require dispensing of the generic drug, if there is one, in their Medicaid programs. Managed care organizations also commonly mandate or encourage generic substitution. The VHA requires genetic substitution, and is more successful than the private sector in promoting the use of generics. This practice relies on FDA determination of the equivalence of a generic to the branded “innovator” product. It is unexceptional and has no implications for restrictiveness.

The committee examined information on veterans' complaints about the National Formulary and surveys of physician opinions of the formulary and formulary system. Complaints and dissatisfaction may be indicators of the restrictiveness of the formulary and may point to areas that have to be addressed. Veterans complained primarily about access to nonformulary drugs, in particular sildenafil citrate (Viagra). These complaints about the National Formulary comprised 0.4% (2,385 out of 570,937) of the total complaints recorded by official patients' advocates in VA facilities.

A number of surveys of physician attitudes about the National Formulary or other formularies have been reported. Of these, the most helpful was performed by the RAND Corporation for the VA. This survey had problems limiting the conclusions that could be drawn from it, but it did find that a minority of physicians complained about not being able to prescribe needed drugs. Other surveys suffered from more substantial problems of a similar nature, that is, very low response rates, survey design, and support by interested parties that made any conclusions highly suspect. The committee appreciates the normal human tendency to want unfettered access to available benefits without (even reasonable) economic or other restrictions. Nevertheless, complaints of patients and prescribers should be evaluated seriously, and their acceptance of the formulary should be a priority objective. Physician support, especially, is essential to the success of a formulary, and refusal of doctors to cooperate may greatly reduce anticipated savings.

The VHA is one of the larger purchasers of drugs in the United States. Prices for most brand drugs are set in the Federal Supply Schedule, which is

administered by the National Acquisition Center. The NAC also negotiates and administers contracts for drugs for the VHA. The NAC, using the leverage of a large purchaser and the selection among agents by the National Formulary, reports substantial savings resulting from implementation of the formulary. As of February 2000, these savings (that is, the difference between NAC estimates of actual expenditures on drugs and what would have been spent in the absence of the National Formulary and other contracting activities) were said to amount to at least $572,521,352. This included savings due to closed or preferred classes, selected blanket purchase agreements (BPAs), generic drug purchases under contract, bulk purchases of pharmacy-related supplies, and savings from patent expirations on brand drugs, among others, from FY 1996 through FY 2000. Using a limited scope (that is, only savings accruing from favorable prices negotiated for six closed or preferred National Formulary drug classes) and a more limited time period (that is, from the date of designation as a closed or preferred class to class opening or the end of available data [July 1999]), the IOM arrived at a conservative, lower estimate, approximating $100 million in the aggregate.

In economic terms, the objective of the VA National Formulary is to make the demand for specific prescription drugs more responsive to price than might otherwise have been the case. Formularies increase a buyer's bargaining power, enabling buyers to be more aggressive in price negotiation. By excluding certain products or by shifting demand significantly between competing products, the buyer presents a seller with a more elastic, or price-responsive, demand, thereby inducing a lower price. The greater the ability to direct the volume of prescriptions between competing products, the more elastic the demand and the greater the bargaining power of the buyer.

Because only drugs selected through drug class reviews by the National Formulary in a closed class are available throughout the VHA unless a nonformulary exception is approved, class closure exerts strong effects on prescribing behavior. Prescribing of preferred drugs in preferred classes is encouraged by drug usage criteria, information about the preferred drugs, and other administrative directives. Alternatives to preferred agents can be provided by VISN or local facility formularies, however. BPAs can also affect drug choices at the VISN, local, or, occasionally, national level. Other factors affecting drug use in the VHA include the shift to outpatient services (and outpatient pharmacy), greater numbers of veterans using the VHA (and the VHA pharmacy benefit), and the introduction of a number of effective, popular, but expensive new drugs.

To assess the effect of the VA National Formulary on VHA drug expenditures, ideally person-level data should be used to compare per-person expenditures for 1 to 2 years before and after formulary implementation. Such data would allow examination of changes in drug spending, in overall health care spending, and in shifts among VHA budgets associated with the National Formulary. Instead, VA data limitations required the IOM to employ aggregate drug use data per VISN per month for FY 1994 through FY 1999 for 14 classes of drugs, 6 of which were, or had been, closed or preferred (angiotensin-converting enzyme inhibitors [ACEIs], hydroxymethylglutaryl coenzyme A reductase in-

hibitors [HMG CoA RIs], alpha blockers, H₂R blockers, proton pump inhibitors [PPIs], and calcium channel blockers [CCBs]), and 8 of which were open. Price data by VISN by month, with some gaps, were obtained from the VA, and these prices were multiplied by drug use by VISN by month to get expenditures. Expenditures were controlled for changes in the VHA user population and age and gender distribution. To explore for cost shifting or other secondary effects, aggregate data on VA inpatient discharges for conditions treatable with drugs affected by class closure were also examined for any association with implementation of the National Formulary.

Several different approaches to estimating savings from formulary policy are described in the pharmacoeconomics literature. Many of these depend on being able to convert data into defined daily doses and to use detailed management information to predict purchasing patterns absent formulary changes. For example, some large private-sector health centers multiply total annual spending on a particular formulary agent by the percentage price decrease negotiated at the time of formulary selection and take this as a cost avoidance or savings for the year, unadjusted for possible coinciding factors affecting drug usage. The data limitations in this study precluded such approaches.

The IOM committee compared pre-National Formulary prices for the most commonly prescribed product in each affected class with post-National Formulary prices calculated as the average VISN prices over the 3 months after the class was closed. Price reductions varied from 16 to 41%. Market shares of selected drugs in these classes changed substantially in most cases, reaching 97% in one case. Price, utilization, and market share changes were generally not found or were consistent with existing trends in the open classes for which data were obtained. Trends in outpatient pharmacy spending varied considerably. In three closed classes (ACEIs, H₂R blockers, alpha blockers), implementation of the National Formulary was associated with a decreased spending level per user relative to previous trends. In two classes (PPIs, HMG CoA RIs), the spending increase was consistent with increasing numbers of prescriptions per outpatient user. In the sixth class (CCBs), spending did not change significantly. Regression analyses were performed to control for key variables, and coefficients of interest were reported to assess the spending effect for each closed or preferred class relative to what would have occurred absent the VA National Formulary.

Effects were reported as reductions in spending, nominal spending change, change adjusted to real present discounted value, and nominal spending adjusted for changes in the general level of prices in the economy as measured by the Consumer Price Index, that is, “real” savings. For ACEIs, reductions were 16.9% (lisinopril) and 8.5% (fosinopril); nominal savings, $17.6 million; and real savings, $16.9 million. For alpha blockers, reductions were 17.5%; and both nominal and real savings, about $1.8 million. For HMG CoA RIs, reductions were 8.1%; nominal savings, $14.4 million; and real savings, $13.8 million. For PPIs, reductions were 7.4%; nominal savings, $4.9 million; and real savings, $4.7 million. For H₂R blockers, reductions were 41%; nominal savings, $47.1 million; and real savings, $45.2 million. For CCBs, there was an increase of 7.9%, nominal extra

spending of $13.2 million, and real extra spending of $12.8 million. There was additional (BPA) contracting for drugs in this class for which data were not available, and these changes were not statistically significant. Because of this uncertainty, the effects in the CCB class should be interpreted with caution.

As noted earlier, the committee explored the possibility of changes in inpatient use associated with VA National Formulary activities. Trends in the number of discharges for selected heart- and ulcer-related diagnoses were plotted to explore possible associations with National Formulary policies on drugs for treatment of these conditions. The regression analyses showed an estimated decrease in heart disease-related discharges of 1.3% relative to the numbers expected in the absence of the National Formulary, and a decrease of 4.9% in ulcer-related discharges. Although neither of these changes was significant, they are very crude indicators of shifting utilization that might be caused by changes in the delivery system.

The committee also attempted to estimate the costs of developing and operating the National Formulary using data from the VA PBM. About half of the VA PBM budget was said to be associated with the formulary, that is, $400,000 for FY 1995 and FY 1996, and $900,000 for each year thereafter. Costs not captured in this estimate would include time spent by VISN leaders and personnel in implementing and managing the National Formulary at that level, time spent by the NAC on National Formulary contract work, procurement costs for the formulary's closed classes, and time spent administering National Formulary exceptions and therapeutic interchanges at the local level, among others. The committee had no information that would allow an assessment of these costs or an overall estimate of VA National Formulary operational expenditures.

The committee estimated the aggregate gross savings associated with five of the six examined classes to be $85.8 million in nominal terms and $82 million in real dollars. As noted, incomplete information —that is, the effect on expenditures of a BPA for diltiazem, the highest-volume drug in the class—prevented a complete estimate of cost trends for CCBs, the sixth class studied. The seventh closed or preferred class, luteinizing hormone-releasing hormones (LHRHs), could not be studied due to the unavailability of data. Overall, the committee found that the National Formulary's closed classes were associated with notable reductions in average outpatient pharmacy spending per outpatient VHA user. After controlling for secular trends in drug utilization not related to the formulary, time invariant unmeasured differences between VISNs, and the changing VA population, the committee estimated decreases in per-user pharmacy spending of between 7% and 41% for these classes. Given the fact that effects of BPAs and other activities of the VA PBM were not measured in the IOM assessment, it is likely that aggregate gross savings may have exceeded the previously cited figures. True gross savings approximating $100 million are quite likely to have been realized over the first 2 years of operation of the VA National Formulary. This figure is about 3% of total pharmacy or 15% of the six analyzed drug class expenditures over those 2 years.

The cost savings directly attributable to the VA National Formulary enable the VHA to provide additional services of other kinds to veterans. Continuation of the formulary is justified on the basis of these savings, especially since, as discussed below, there are no scientifically valid reports of adverse quality effects from the National Formulary. The committee has observed some deficiencies in the data needed to assess various effects of the National Formulary or to ensure knowledgeable formulary management. Improvements in cost data would help the VHA to assess per-person expenditures and cost shifts among various VHA budgets.

The committee has evaluated VA National Formulary and formulary system effects on quality of care using structural elements. Effects of the formulary and formulary system, using process and outcome criteria, have also been discussed. Process criteria include effects on utilization: whether veterans have access to the fight drug at the right time. Outcome criteria refer to whether clinical or human outcomes are affected. They would be the most persuasive indicators, as noted in the IOM definition of quality. By this definition, quality of care is the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge. In fact, there are few data on anything other than structural characteristics of the VA National Formulary. The committee, therefore, concentrated on these characteristics.

The committee looked predominantly at clinical pharmacy services, local facility pharmacy and therapeutics committees, VISN formulary committees, the VA PBM and MAP, the quality and availability of existing and new FDA-approved drugs on the National Formulary, drug class reviews and therapeutic guidelines, the nonformulary process, therapeutic interchange policy, and drug utilization review (DUR). The development of clinical pharmacy services and the redefinition of the scope of pharmacy practice in the VA that occurred in the 1990s are important structural elements. However, these changes began well before the VISN formulary and the National Formulary and are important in overall quality of drug therapy and management of the pharmacy benefit. They are also important to the quality-of-care effects of the National Formulary, but they cannot be said to be a specific structural element of the National Formulary itself.

Local facility pharmacy and therapeutics committees (P&T) predated VISN formularies and National Formulary by several decades. They have other than National Formulary functions. They also are important to the quality-of-care effects of formularies at the local level. They consider staff prescriber nonformulary requests, ensure proper contract drug use and generic prescribing, implement therapeutic interchanges, and monitor adverse drug events, among others. Their role in the design of local formularies is unclear since most of these are the same as the VISN formulary, but they can originate requests to add drugs to a VISN formulary or to the National Formulary.

VISN formulary committees usually have 15 members (range 6–28), and average 52% pharmacists (range 33– 100%) and 44% physicians (range 0– 65%). Some VISN formulary committees appear to focus on VISN and National Formulary management and budgetary matters, that is, to function as little PBMs. As such, they will affect quality of care, but there are no formal studies of this. Other formulary committees may be more like traditional pharmacy and therapeutics committees and may affect quality of care as these committees do. Formulary committees appear to be highly variable in membership, and the presence of no or very few physicians on some of them may have implications for physician acceptance of the VISN or National Formulary.

The VHA reorganization that authorized 22 VISNs also led to the creation of a Pharmacy Benefits Management Strategic Healthcare Group located in Washington, D.C., and Chicago; a Medical Advisory Panel made up of 1 DOD and 11 VA field physicians; and a VISN Formulary Leaders Committee, as described earlier. These bodies are central structural elements related to the National Formulary's effects on quality of care. The quality of their management and clinical decisions regarding the formulary and formulary systems is as important to the VA formulary as are similar management and decisions of private-sector PBMs to private formularies. The IOM committee assessed VA PBM performance by examining the quality of the formulary, additions of drugs, drug class reviews and therapeutic guidelines, the nonformulary process, therapeutic interchange policy, and other elements of the system that affect quality. The committee observed good quality discussions and decision making at a MAP/VA PBM meeting in August 1999.

Addition of existing and newly FDA-approved drugs to the National Formulary and the overall quality of drugs on the formulary are important factors in the availability of drugs and thus in the quality of care. The committee recommended more timely consideration of additions and a better balance between a uniform national entitlement and local preferences. The size of the National Formulary was found to be reasonable. The committee evaluated the quality of the drugs listed against available lists of questionable drugs, questionable combination products, and drugs considered inappropriate for the elderly by the General Accounting Office (GAO). The national list was found to contain few such products. Drugs included on the National Formulary appear to meet reasonable standards of numbers, variety, and quality based on IOM committee members' professional judgment and experience.

A drug class review is an important mechanism by which a formulary system evaluates and selects from among drugs and drug products those that are considered most useful in patient care. Choosing in this way has quality implications, but it also enables a formulary system to negotiate favorable prices for selected drugs based on anticipation or commitments of high-volume use, as observed elsewhere. Detailed standards and guidelines for performing drug class reviews have been published several times. The committee also has professional and institutional experience with drug class reviews at academic medical centers, managed care organizations, and PBMs. Using these standards and its expe-

rience, the committee examined the conclusions and recommendations of nine VA drug class reviews. In all cases the drugs recommended by the reviews were included on the National Formulary. Both intrinsically and in comparison to reviews in private-sector organizations, the reviews were of high professional quality and reached recommendations based on scientific evidence and sound interpretation of clinical data.

VA clinical guidelines are clearly written, and their recommendations are consistent with current recommendations of other organizations, such as the American College of Cardiology, the American Heart Association, and the Department of Health and Human Services. The guidelines are also tailored to the older VA patient population. These guidelines are based on current scientific and clinical research data and are equivalent to similar documents in the private sector. Evidence on the use of these guidelines is scanty, but some facilities or VISNs are monitoring their effect on clinical practices and following up with appropriate interventions.

The IOM committee surveyed pharmacy or medical personnel in all 22 VISNs and discovered a wide range of circumstances in which nonformulary prescriptions, requests, and decisions might not be recorded or appear in monthly statistics. As a result, reports of the percentage of requests submitted or approved are probably not reflective of the actual situation in the VISNs. The access of veterans to nonformulary drugs, other effects of the nonformulary process, and their implications for quality are therefore hard to assess or compare with other systems. Although the committee appreciates the advantages of some degree of informality, the VA system should be made more consistent and likely to be portrayed accurately in program statistics.

Current published reports of VA therapeutic interchanges, unpublished documents from VA facilities, and expressions of physician and patient dissatisfaction from surveys and patient advocate data have not made a compelling case for quality problems. They do reinforce that therapeutic interchanges are often viewed as problematic, and they raise questions for exploration and suggest possible responses that might be taken by the VHA. Some of these questions have been discussed earlier. If veterans receiving a drug on a long-term basis are subject to interchanges multiple times or too often because of changes in the formulary or committed-use contracting, there will be effects on treatment acceptance and compliance, and there may well be changes in the effectiveness of treatment. In short, quality-of-care and health outcomes may be affected. The VHA has to address the consistency and policy gaps in therapeutic interchange to ensure that quality is not affected.

The use of drugs, as part of the process of care, was examined by the committee. Class closure, committed-use contracts, preferred drug designations, drug usage criteria, and separate negotiations for drug prices, such as blanket purchase agreements, all may affect the utilization of drugs in certain classes. Dramatic changes in utilization may follow class closure, National Formulary selection, and contracting for drugs. These changes are indicated in Chapter 3 (Cost) and Chapter 4 (Quality) of this report. If some of these changes are not system-

wide, as is often the case due to BPAs, for example, veterans traveling from VISN to VISN may experience different preferred drugs. They will then have to depend on the uncertainties of the nonformulary process to ensure access to their current drug treatment. The committee did not find any scientifically valid evidence that changes in the number or variety of drugs by class closure were affecting the quality of drug treatment and health outcomes of veterans. To analyze such effects, person-specific tracking of drug utilization would be needed, as noted in Chapter 3.

Changes in the prevalence or types of adverse drug events associated with implementation of the National Formulary and formulary system would be important evidence of health outcomes affected by the formulary. The VHA has taken steps, such as electronic prescribing and machine-readable coding of prescriptions, to reduce adverse events. Other programs, such as the VA Patient Safety Event Registry, do not reliably collect adverse event data or are in early stages. These programs are not a part of the National Formulary. Statistics on adverse events during therapeutic interchange are not available in the VHA or nationally in other health care systems.

The IOM committee, as part of its analysis of cost effects of the National Formulary, collected hospital discharge data before and after formulary implementation in 1997. As reported earlier, no association was observed between changes in the distribution of discharge diagnoses for specific conditions and changes in National Formulary policies on drugs to treat these conditions. This analysis suggests only that there are no quality effects of sufficient magnitude to be demonstrated by this imprecise measurement tool.

Patient satisfaction is a generally accepted element of quality of care. Patient dissatisfaction discovered through significant levels of complaints to advocates or through surveys could be an important indicator of the need for a system response. Data from surveys are not conclusive, but physician acceptance of the formulary is important to its success and has to be addressed.

Based on what is known about its effects on quality of care, there is no reason to discontinue the National Formulary. Concerns exist in a number of areas, such as the addition of drugs, the nonformulary process, therapeutic interchange, and patient and physician acceptance. The absence of good data on quality effects is a particular concern, as is the need for better data to enable prudent management of the National Formulary. The VA should focus its health services research capacity on National Formulary and drug treatment issues and should improve operational data. It is the responsibility of an important national program to illuminate these issues, which have implications beyond the boundaries of the VHA.

Almost all managed care organizations offer pharmacy benefits and have formularies. PBMs frequently either manage these benefits, pay claims, or both. Most of the formularies are closed or partially closed, although it is not clear

precisely what is meant by this, since formulary systems are extremely varied in managed care plans and PBMs. Copayments are a common practice in managed care, often amounting to the entire cost of a brand drug if generics are available. Other formulary management strategies in managed care include generic substitution, therapeutic interchange, DUR, exclusion of certain specific drugs or drug classes, and prior approval. The committee reviewed a number of publicly available managed care or PBM formularies and described some classes that are closed in these formularies and also in the VA National Formulary. The numbers and varieties of drugs or drug classes that are excluded or subject to prior approval in some managed care formularies were listed. New information of this kind was collected by the Academy of Managed Care Pharmacy and is discussed in Chapter 2.

In general, managed care formularies employ exclusions, prior approvals, and copayments, which are not features of the VA National Formulary. Some of these—for example, prescription copayments—are insensitive to medical need. Reacting to data such as these and to public concerns, legislatures in many states have enacted legislation requiring public disclosure of formulary and nonformulary processes. Some states have imposed requirements on the nonformulary process, mandating that it be expeditious or that requests be approved within 24 hours and that a 72-hour supply of the drug be provided. Other states have set criteria for the approval of nonformulary requests. Many states have laws that set limits on therapeutic interchange, prohibit increased cost sharing or copayments, or prohibit eliminating or decreasing coverage during a contract year. These state requirements for managed care generally are not relevant to the National Formulary, which is already characterized by disclosure on the Internet (although the nonformulary process is in practice not transparent), low or no copayments, and relatively stable drug coverage.

Because the VA is a government health care program, the committee decided to review two public-sector programs with formularies and formulary systems. The Medicaid program has the largest government pharmacy benefit and has been functioning with formularies for several decades. The committee also decided to review the DOD formulary and formulary system. The DOD health system and pharmacy benefit are roughly the same size as those of the VA. Unlike Medicaid recipients who are primarily poor women and children, the DOD population consists of military personnel (that is, persons who are or will be veterans and dependents of veterans) with many similarities to the VHA population.

The Medicaid fee-for-service program provides coverage for prescription drugs in every state under Title XIX of the Social Security Act and applicable state law. In addition to Title XIX, the current Medicaid benefit is governed by provisions enacted in the Omnibus Budget Reconciliation Act (OBRA) of 1990 and 1993. Prior to OBRA 1990, states faced fewer limits on drug management strategies, such as formularies, prescription limits, generic substitution, prior approval systems, refill limits, and copayments. In OBRA 1990, Congress prohibited restrictive state Medicaid formularies, allowed prior approval under cer

tain conditions, required Medicaid programs to reimburse all new drugs for at least 6 months after FDA approval, and required that manufacturers rebate to Medicaid the lesser of an approximately 15% discount below average wholesale price or the best price offered to other purchasers.

OBRA 1993 softened the prohibition on restrictive formularies. Currently manufacturers are required to sign rebate agreements to qualify their products for Medicaid reimbursement. Medicaid may exclude only a list of drugs or drug classes specified in the legislation. Prior approval is allowed as long as the approval system provides a response within 24 hours and a 72-hour emergency supply of the drug under review. The limits on exclusions and the requirements for listing all drugs under rebate agreements have left prior approval as one of the last flexible drug management strategies available to state Medicaid programs. States may adopt their own clinical or nonclinical criteria for approving a prior approval request. States may restrict a particular drug to patients of a certain age, those with a specific diagnosis, or those who have been treated with another drug first. States also can impose small copayments ($0.50 to $3.00) on certain categories of recipients, and many states impose prescription limits.

The different formulary system controls and the types of formularies, open or closed, in each state are listed in Table 5.1. Most states use some combination of prior approval, exclusions, prescription limits, and copayments. The effects of these controls are not clear. They depend on the design and rigor of implementation. Since the states are so different, comparisons with the VA National Formulary are uncertain. The VA provides a standardized national entitlement through the National Formulary, but this uniformity is diluted by the variability among VISNs in their formularies and formulary system policies and procedures. The VA nonformulary system is perhaps more variable than Medicaid's, although data are limited for both systems. Exclusion of OBRA classes and some kinds of limits on the amount and frequency of dispensing by Medicaid are more restrictive than VA controls. The National Formulary provides fewer choices in some drug classes, particularly closed classes, although Medicaid prior approval in some of these classes may limit access.

The committee reviewed 15 studies of formulary system controls in Medicaid fee-for-service programs. Many of these studies had significant deficiencies, including short follow-up periods and a lack of control groups, which limited any conclusions that could be drawn from them. In spite of this, the committee suggested certain inferences. Programs to decrease utilization or costs through a particular control sometimes result in offsetting utilization or costs in another part of the pharmacy or other benefits of the health care system. Controls that do not allow for medical need-based exceptions sometimes cause undesirable changes in utilization, costs, or quality of care. Most of the restrictions or characteristics of Medicaid formularies and formulary systems studied in the 15 reports that the committee reviewed are not features of the VA National Formulary, however. Therefore, these reports are of limited usefulness in evaluating the restrictiveness or effects on cost and quality of the VA formulary system.

Medicaid managed care has expanded dramatically in the past 10 years. Currently, more than half of Medicaid recipients are enrolled in almost 600 different plans. Two states, Tennessee and Arizona, administer their entire program through managed care plans. Only two states do not use managed care (Alaska and Wyoming). Most plans are authorized through federal approval of waivers of Title XIX requirements—Section 1915b or Section 1115 waivers. The Balanced Budget Act of 1997 made it easier for states to enroll recipients in managed care. Most plans have enrolled women and children, not the more vulnerable and costly aged and disabled population which is more like the VHA population. The pharmacy benefit in Medicaid managed care varies with the plan, but is based on the contract between the state and the plan, the provisions of the waivers, and an underlying requirement that it should not be more restrictive than the fee-for-service benefit. At present, 13 states have carved out their pharmacy benefit from managed care and administer it separately. Insufficient information was available to the committee to make detailed comparisons between Medicaid managed care and the VA National Formulary. It is likely that findings would be similar to those for Medicaid fee-for-service programs.

The committee briefly reviewed the Department of Defense pharmacy benefit, which has a somewhat lesser cost than the VA benefit. The DOD formularies and formulary systems are variable across a number of facilities. The benefit varies because it is in part available at more than 500 military treatment facilities worldwide and in part available through contract mail order or thousands of retail pharmacies. Substantial deductibles and copayments are a part of the system for some non-active duty personnel. The DOD Basic Core Formulary, which is a national requirement at all military treatment facilities, is a very limited formulary with 159 listings.

The DOD benefit, formularies, and formulary systems are in transition under provisions of the National Defense Authorization Act for FY 2000 (P.L. 106-65, Section 701), which requires a new Uniform Formulary, data systems, and advisory committees, among others. The committee observed that the present DOD Basic Core Formulary, mail order formulary, and multiple treatment facility formularies are not comparable counterparts to the VA National Formulary and formulary system. They are in earlier stages of development. How they will compare after full implementation of the National Defense Authorization Act is unknown and presumably will depend on the details of changes.

In examining public and private-sector formularies in comparison to the VA, the committee concluded that some are more open. Medicaid programs are required to offer all drugs on the Federal Supply Schedule that manufacturers list for rebates. Some are more restricted. They require prior approvals and exclude some drugs. All are variable, some probably more so than the VA's (for example, DOD's). Some controls that are not part of the VA system, such as relatively costly deductibles and copayments, may present real barriers to needed drugs, especially for low-income patients. These controls are part of DOD requirements for some eligibles or employed by some managed care plans.

Other controls, such as generic substitution and therapeutic interchange, are in common use in many systems. The important element for quality and restrictiveness is timely availability of a safe and effective, medically necessary drug, if not listed, through an exceptions process. A good formulary supports this element and, through its capacity to make quality choices, enhances price negotiations and prudent purchasing.