The Pharmacotherapy of Acute Anxiety

Richard I. Shader and David J. Greenblatt

Panic attacks are a prototypical presentation for acute anxiety. Rapid in onset, they quickly build to a crescendo of symptoms of sympathetic overactivity and typically last from a few minutes to a few hours. Experientially, panic attacks differ from other attack-like episodes of intense anxiety mostly because of their rapidly mounting pattern of symptoms and because patients usually feel they must flee from their current location or fear they will suffocate, "go crazy," or even die during the height of the attack. Panic attacks can occur without obvious triggers and are not simply acute bouts of fear or intense worry. Approximately 10% of the adult population reports having experienced at least one panic attack. Fewer have had recurrent attacks, and estimates of the lifetime prevalence for those meeting criteria for panic disorder range from 0.1% to 2.3%. Panic attacks can occur in conjunction with other disorders or as a component of the symptomatology of other disorders. When seen with other disorders, their response to treatment may differ. For example, patients with schizophrenia may worsen rather than improve when tricyclic antidepressants are used for their anti-panic effects (5). About one-half of patients who present with panic disorder will eventually develop depressive symptoms, and about 20% of patients with recurrent major depressive episodes report having panic attacks. Some data suggest that when panic attacks accompany depression, treatment response may be less favorable (22). This observation, however, may be confounded by the duration of untreated panic disorder prior to the initiation of effective therapy or by the type of pharmacotherapy chosen (61).

Generalized anxiety disorder (GAD) is another prominent anxiety disorder. It appears to be about three to four times more common than panic disorder. GAD is a somewhat confusing disorder; its diagnostic criteria have shifted considerably over the last 10–15 years (79). When codified in 1980, the time-course criterion required 1 month of continuous or persistent symptoms. In DSM-IIIR this was lengthened to 6 months, and a greater emphasis was placed on the presence of unrealistic worry as a key symptom. In DSM-IV, the 6-month time frame has been maintained, and excessive worry is now the cardinal feature. Because of these changes, studies done in the 1980s may not be comparable to those of agents evaluated in the early 1990s. Patients seen in primary care settings often do not meet full GAD diagnostic criteria. This is particularly important because marketing data suggest that about 75–80% of prescribing of anxiolytic agents is by primary care physicians; benzodiazepines account for the vast majority of these prescriptions.

Additional aspects of panic disorder and other forms of anxiety disorders are considered in more detail elsewhere in this volume (see Maintenane Drug Treatment for Schizophrenia and Atypical Antipsychotic Drugs). In this chapter, we discuss selected aspects of the understanding and pharmacotherapy of acute anxiety and briefly explore future directions for treatment. Because the benzodiazepines are still the major class of compounds used in anxiety, considerable attention will be devoted to their molecular pharmacology, current patent status including new agents, regulatory issues, drug interactions, and receptor partial agonists (benzodiazepines and 5-HT1A). The remaining sections of the chapter deal with other drugs, namely, SSRI and tricyclic antidepressants, cholecystokinin antagonists, and 5-HT3 receptor antagonists.

Benzodiazepines remain the mainstay of our treatment armamentarium for acute anxiety. They have been extensively utilized and studied, and their efficacy and relative safety compared to other antianxiety agents currently marketed in the United States are well established; the interested reader is urged to consult recent comprehensive reviews that have appeared elsewhere (47, 78, 94). Although thousands of benzodiazepine-like drugs have been synthesized, no new benzodiazepines have been introduced into clinical practice in the United States as antianxiety agents since the last volume in this series.

Central nervous system (CNS)-active benzodiazepines bind at complex, membrane-spanning, heteromeric protein structures that comprise the GABA-A receptors in gray matter. The current assumption is that these structures are pentameric. The highest receptor densities are found in synaptosomal fractions from cortex, followed by hypothalamus, cerebellum, hippocampus, and striatum. Receptor populations are also found in midbrain, medulla oblongata-pons, and spinal cord. Benzodiazepines allosterically modify GABA-mediated opening of a chloride ion channel in the center of the GABA-A receptor. Other modifiers acting at different binding sites include barbiturates, ethanol, and neurosteroids. Benzodiazepine-induced conformational changes increase GABA's receptor affinity, and thus increase the frequency of ion channel openings. Chloride ion influx hyperpolarizes GABA neurons and produces GABA's inhibitory interneuronal effects, including actions on dopaminergic, serotonergic, and noradrenergic neurons. For example, inhibition by GABA-ergic neurons decreases locus coeruleus noradrenergic neuron firing.

The noncovalently associated protein subunits surrounding the chloride channel by convention are labeled alpha (a), beta (b), and gamma (g). Other polypeptide subunits have been identified (e.g., d, r), but their functional significance is not established. Various GABA-A receptor isoforms are assembled from gene family products that yield six, three, and three different subtypes of a, b, and g subunits, respectively. In vitro transfection data and other studies suggest that there is considerable regional heterogeneity in the subunit composition of GABA-A receptors, but most are comprised of an a1b2g2 pattern. In this type of receptor complex, sometimes called the type I benzodiazepine subtype, GABA per se binds to b subunits, while photoaffinity labeling studies suggest that benzodiazepines bind to a site on the a subunit (a1) (85). So-called type II imidazopyridine-sensitive (e.g., alpidem, zolpidem) subunits contain a2/a3 subunits, while an imidazopyridine-insensitive type II receptor subtype contains the a5 subunit. Receptors containing a4 or a6 subunits have low affinity for benzodiazepines and are sometimes called diazepam-insensitive receptors. Receptors containing g subunits have increased affinity for benzodiazepines (89). Unfortunately, in vivo strategies are not available to establish the accuracy or clinical significance of these in vitro observations. Also, it is not known what subunits or combinations are expressed in vivo in single neurons; it is surely possible that a single cell could carry multiple receptor subtypes. GABA-A receptors have a clear homology with glycine and nicotinic acetylcholine receptors.

All eight benzodiazepines currently indicated for GAD have exceeded their patent protection; nevertheless, alprazolam, diazepam, and lorazepam remain among the top 100 drugs in sales in the United States. Alprazolam recently received approval (but without a patent extension) as the only benzodiazepine specifically for use in panic disorder (50, 68, 86). Other benzodiazepines, however, have also been used successfully in panic disorder (32, 47, 77, 78). A slow-release formulation of alprazolam has been developed and is being reviewed by the FDA. Clonazepam's sole approved indication is as an anticonvulsant even though it is frequently prescribed for its anti-panic (47, 78, 87) and anti-manic effects. Adinazolam, with its active metabolite [mono-N-desmethyladinazolam (NDMAD)], is a potent triazolobenzodiazepine which, in a slow-release formulation, appears to have efficacy in patients with panic disorder (35, 37, 70, 82). There is some suggestion that use of the slow-release preparation may be associated with less interdose (therapeutic tolerance) and discontinuation-associated (rebound) intensification of symptoms than occurs with some rapidly eliminated, shorter half-life benzodiazepines (e.g., alprazolam). As with other triazolobenzodiazepines, the biotransformation of adinazolam to NDMAD appears to involve the cytochrome P-450 3A subfamily pathway. NDMAD, which is more potent than its parent compound, is renally eliminated (34). It seems likely that caution will be required in the use of adinazolam in patients with renal impairment. (N.B., A recent review by an FDA advisory panel did not recommend approval of adinazolam for use in panic disorder.)

Perhaps in part because of their extensive use and cost impact, benzodiazepine use has been subjected to increasingly close scrutiny and monitoring. In New York State, for example, the promulgation of triplicate prescription procedures since 1989 has significantly reduced their use. Ostensibly, such programs are initiated to reduce alleged inappropriate prescribing practices and illicit diversion. Unfortunately, however, what has mainly been accomplished is a decrement in state (e.g., Medicaid) and third-party (e.g., Blue Cross and Blue Shield) outlays for benzodiazepines. Other anxiolytic agents and alcohol-containing products are routinely substituted by some physicians or patients; these substituted therapies generally are less safe or less effective. There have been negative impacts not only in patients with anxiety disorders, but also in elderly patients in nursing homes and patients with epilepsy. An increase in deaths due to overdosages with meprobamate has been reported. Adequate impact data have not been collected and analyzed, and the larger community of concerned clinicians and scientists has not become adequately involved with this increasingly problematic issue (4, 10, 72, 74, 75, 76, 80, 91, 96). Other regulatory actions have been under consideration, including class labeling for benzodiazepines used as hypnotics and unit-of-use packaging (57).

Underlying many of these actions is an unsubstantiated belief that benzodiazepine misuse and abuse are widespread and a significant public health problem. In this regard, a survey conducted by the late Mitchell B. Balter and colleagues is of interest (8). A peer selection process using primary nominators from 44 countries generated a panel of international experts on the pharmacotherapy of anxiety. Sixty-six members of this clinician–researcher cohort overwhelmingly concluded that strict and differential restrictions on the use of benzodiazepines are not warranted. This conclusion reflected the opinion of the expert panel that the relative abuse liability of benzodiazepines is low and that qualitative differences in abuse liability among the benzodiazepines are minimal.

Pharmacodynamic drug interactions (e.g., CNS depression and its consequences) between benzodiazepines and other CNS agents, including ethanol, are well-recognized. However, during their initial years of use, one of the "virtues" of benzodiazepines was thought to be their relative lack of clinically significant pharmacokinetic interactions. This property was a clear advantage, for example, over the barbiturates and meprobamate which are known inducers of hepatic oxidative metabolism. This conclusion reflected the fact that a few benzodiazepines are metabolized mainly by conjugation (e.g., oxazepam, lorazepam), whereas others such as diazepam, with its active metabolite desmethyldiazepam, may have alternative pathways via different cytochrome P-450 isozymes (e.g., 2C3, 2D6, 3A4) that could mitigate the clinical significance of any pharmacokinetic interactions. Triazolobenzodiazepines (e.g., alprazolam, midazolam, triazolam), however, present different concerns: They are metabolized preferentially by 3A subfamily isozymes, and both in vitro and in vivo data suggest that clinically significant increments in sedation and its consequences may occur when triazolobenzodiazepines are coadministered with other agents that inhibit this pathway either competitively (e.g., fluoxetine) or through complex formation (e.g., erythromycin and other macrolides, cimetidine) (36, 40, 55, 58, 88, 90, 95). Benzodiazepines do not appear to have significant effects on the metabolism of other agents (i.e., when they are viewed not as substrates but as potential competitive inhibitors).

A partial agonist by commonly accepted classical definition has less intrinsic activity than a full agonist for the same receptor. Potency is not the differentiating variable, because potency comparisons are made from the dose at which a given drug attains 50% of a target pharmacodynamic effect (i.e., its ED50). Two drugs (e.g., a full agonist and a partial agonist) could differ according to their intrinsic activities and yet have the same ED50 or potency. Another view of partial agonists presumes that they have reduced intrinsic activity and require a greater degree of receptor occupancy as compared to full agonists to achieve the same effect. However, it is also possible for the same drug to be a partial agonist at one receptor site and a full agonist at another. For ligand-gated ion channels, full and partial agonists modulate their allosteric transition from their inactive (closed) to active (open) states to varying degrees, perhaps because of different affinities for specific protein subunits in the receptor complex. This understanding of full and partial agonists fits well with drug actions at ligand-gated chloride ion channels (e.g., GABA-A receptors) where no second messengers are involved. It is possible that more frequent contact with the benzodiazepine receptor is required for partial agonists than for full agonists to maintain the GABA receptor in the active (open) conformation.

When G proteins and second messengers are involved (e.g., 5-HT1A receptors), the same drug acting at the same receptor type can produce effects ranging from antagonism to almost full agonism (48). Drug action may depend on the state of the coupling in the G-protein–receptor complex which may vary from precoupling to dissociation. Receptor coupling to different combinations of G-protein subunits can also influence second messenger effects. Some of this variation may also be a function of the degree of receptor reserve (i.e., spare receptors). For the medications in the following two subsections, there is still limited understanding about the specifics of their actions as partial agonists.

The search for benzodiazepine receptor partial agonists derives from the desire to find agents that will have comparable or better efficacy, tolerability, and safety when compared to currently available benzodiazepines, and yet cause fewer of the unwanted properties attributed to benzodiazepines (e.g., sedation, physical dependence, tolerance) (41). It should be kept in mind that GABA-A receptors in contrast to GABA-B receptors do not couple to second messengers.

A variety of structurally distinct compounds have been studied that have partial agonist properties at GABA-A receptors. Some act as full agonists at certain GABA-A receptor subtypes while acting as partial agonists at others; some appear to act as partial agonists at all GABAA receptor subtypes. Among the agents that have received the most attention are abecarnil (84), alpidem (66), bretazenil (54, 62), and suriclone (53). All have been investigated for their effectiveness in animal models and in patients with anxiety. Bretazenil, for example, has reasonable effectiveness in animal models and yet in comparison to benzodiazepines (e.g., diazepam) produces less sedation, anticonvulsant tolerance, muscle incoordination, self-administration, and discontinuation-related behaviors (62). Alpidem (an imidazopyridine) (66, 67), abecarnil (a b-carboline) (7), and suriclone (a cyclopyrrolone) (3) appear to be the most extensively studied of these agents in humans.

Alpidem is a high-affinity full agonist comparable to benzodiazepines at a2b1g2S recombinant receptors, but it is a low-potency partial agonist at a2b1g1 recombinant receptors at which benzodiazepines act as higher-potency partial agonists (89). Abecarnil, by contrast, is a partial agonist at both of these recombinant receptors (89). Although not studied in this same way, suriclone binds to membranes that have been fully and irreversibly photolabeled with flunitrazepam, which suggests some binding at a locus that is not identical to the a1b2g2 subtype (97). In standard models, however, suriclone has high affinity for both type I and type II benzodiazepine receptors. Curiously, some cyclopyrrolones do not potentiate the sedative effects of ethanol (31).

Despite their presumed advantages, no benzodiazepine partial agonists have been approved for anxiolytic use in the United States. Abecarnil's development program is currently quiescent, although anxiolytic research in Europe is quite active. Suriclone is no longer under study, and alpidem was recently withdrawn from clinical use in Europe because of hepatotoxicity (note that hepatotoxicity does not appear to be a problem with zolpidem, a structurally and pharmacologically similar agent recently marketed in the United States as a hypnotic). Although it seems likely that other benzodiazepine partial agonists will be developed, it remains unclear whether their putative selectivity based on in vitro binding data will translate into any demonstrable clinical advantages (e.g., reduced sedation, tolerance development, or dependence liability).

Understanding the role of serotonergic agents in anxiety disorders is not straightforward. The CNS actions of serotonin are mediated by multiple receptor subtypes and subfamilies that can be present on the same neuron and involve different intracellular signaling systems (42). Azapirones such as buspirone and ipsapirone (13) are generally classified as 5-HT1A receptor partial agonists. The role of the 5-HT1A receptor in the pathogenesis of anxiety disorders is discussed in detail elsewhere in this volume (see chapter by Coplan et al.). In vitro studies suggest that these compounds are full agonists at presynaptic autoreceptors and partial agonists at postsynaptic receptors. Presynaptic 5-HT1A receptors also appear to have a large receptor reserve, whereas postsynaptic receptors do not. A given agonist acting at the same receptor in a specific cell line can show differences in intrinsic activity depending on receptor density. Receptor–effector coupling also can vary as a function of the second messenger system involved (e.g., adenylyl cyclase versus phospholipase C) (48). It is unclear how such putative differences translate into the effectiveness of these agents in GAD. Agonist effects at presynaptic autoreceptors could temporarily reduce serotonin concentrations and have anxiolytic effects, but desensitization should occur with chronic administration. Postsynaptic partial agonist effects could lead to reduced postsynaptic agonist effects from serotonin per se. This overly simplified explanation based on in vitro findings does not adequately take into account the increasingly complex information accumulating about serotonin receptors and their varied nature and second messenger systems. Some non-azapirone 5-HT1A receptor full agonists are in the early stages of development. Their availability may help to clarify the role of this receptor in anxiety. The slow onset of action of the azapirones is consistent with the idea that second messenger systems set into motion a cascade of downstream events that takes more time to effect change than drugs that work directly on ligand-gated ion channels.

Azapirones are not as consistently effective as benzodiazepines in patients with GAD. Some evidence suggests that these agents have a better chance of helping patients not previously benefited by or exposed to benzodiazepines. In such a population, 60–80% of patients could be expected to benefit. (45). Compared to benzodiazepines, they cause less sedation, motor impairment, and memory loss. They also do not appear to cause clinically significant discontinuation syndromes. Unfortunately, they do not seem to stop panic attacks (50, 81). However, when they are effective at reducing overall levels of anxiety they may decrease the frequency of panic attacks in some patients.

Tricyclic antidepressants (TCAs) and specific serotonin reuptake inhibitors (SSRIs) appear to be beneficial in some patients with GAD, mixed anxiety and depression, or panic disorder. The numbers of such patients in placebo-controlled trials are growing, but none of these drugs has received FDA approval for any of these indications. Among the TCAs, imipramine (6, 20, 25, 45, 49, 63, 93) and clomipramine (21, 30, 51) have been the most extensively studied. Among the SSRIs, fluoxetine and fluvoxamine have been the most extensively studied, especially for their role in panic disorder (6, 11, 20, 25, 28, 29, 30, 38, 39, 45, 46, 59, 63, 73, 83, 92, 93). Relative to benzodiazepines, onset of action in any of these anxiety disorders for any of these agents is slow, ranging from 2 to 6 weeks. Patients with panic disorder have received the most attention in controlled trials with SSRIs, and their ability to reduce the frequency of panic attacks appears to be independent of concomitant depressive symptomatology. Because of their relative lack of unwanted anticholinergic and cardiovascular effects and weight gain, SSRIs are favored over TCAs by many clinicians. In panic-disorder patients treated with TCAs and SSRIs, there may be an initial increase in jitteriness that decreases treatment acceptance (tolerability) (65, 69). Prior warning through patient education or the temporary use of b-adrenergic receptor antagonists or a benzodiazepine may be beneficial. Sexual dysfunction (e.g., anorgasmia, retarded ejaculation) associated with SSRI use can also affect patient acceptance and compliance. In clinical practice, it is increasingly common for patients with panic disorder to be started on a benzodiazepine (e.g., alprazolam, clonazepam) while an antidepressant (e.g., imipramine, fluoxetine) is being phased in. Such combination regimens need to take into account potential drug interactions (see above).

Cholecystokinin (CCK), in addition to its peripheral role as a regulator of gastrointestinal functions (e.g., inhibition of gastric emptying), is also a centrally acting neuropeptide with high-affinity binding sites in cortex, hippocampus, and amygdala (9, 71). Its predominant form in brain is a sulfated octapeptide (CCK-8). Data suggest that some forms of CCK may be mediators of anxiety symptomatology; benzodiazepine receptor agonists (both full and partial) have been shown to antagonize CCK-induced excitation of hippocampal neurons in the rat (15, 16), and long-term benzodiazepine use in the rat has also been shown to lower CCK-induced CNS activation (14). Animal models also support the anxiogenic effects of a naturally occurring metabolite of CCK-8 [i.e., the COOH-terminal tetrapeptide (CCK-4)] (26). This anxiogenic effect is likely mediated by both CCK-A and CCK-B receptors in brain, and it may explain the anxiolytic properties of devazepide, a non-benzodiazepine CCK-A receptor antagonist (43). Devazepide also causes a dose-dependent inhibition of CCK-8-stimulated increases in levels of ACTH and b-endorphin in rat brain (64). In small-cohort studies, the anxiogenic properties of CCK-4 have been observed in normal volunteers (an effect that could be blocked by lorazepam) (27) and in patients with panic disorder (17, 18, 19). In two patients studied by the latter group, CCK-4-induced panic attacks were blocked by pretreatment with imipramine (17, 18), an observation that curiously is not reported in Ref. 19. Others have noted the anxiogenic effects (in a small series of panic disorder patients and controls) of the homologous CCK pentapeptide fragment, pentagastrin (1). In one study, patients with panic disorder had lower CCK-8 CSF levels than did normal controls (60). Possible explanations for this finding include (a) altered metabolism of CCK-8 or conversion into CCK-4, (b) reduced CCK-A or CCK-B receptor density, or (c) increased CCK-A or CCK-B receptor sensitivity in panic disorder patients as compared to controls. Taken together, these studies support a possible role for CCK in the pathogenesis of anxiety, the use of CCK-4 as a diagnostic challenge test for panic disorder, and a potential role for specific CCK-A or CCK-B receptor antagonists in the acute treatment of anxiety.

The 5-HT3 receptor antagonist and antiemetic agent, ondansetron, is currently being investigated as a potential treatment for GAD, panic disorder, and social phobia. Because the 5-HT3 receptor is a ligand-gated ion channel for cations, it is possible that drug actions at this receptor subtype may be more rapid than for drugs acting at other serotonin receptor subtypes. Some data also suggest that 5-HT3 receptors may mediate CCK release. From microinjection studies in rats, the amygdala has been postulated to be a key site for the anxiolytic activity of 5-HT3 antagonists (44). Some rodent models that quantify social interaction or time spent in dark versus light test boxes and primate (marmoset) responses to a confrontational human threat test have supported the further testing of ondansetron as an antianxiety agent (23, 24, 52, 56). Data available from human trials are limited but supportive at the present time. Four hundred and two patients with GAD were treated in the United Kingdom in a multicenter (58 general practitioners), 4-week, fixed-dose, double-blind trial comparing ondansetron at 1.0 mg t.i.d. and 4.0 mg t.i.d. to diazepam 2.0 mg t.i.d. and placebo (16, 17). Using HAM-A total scores, both the higher-dose ondansetron patients and the diazepam patients were significantly more improved by 3 weeks than those on placebo. At trial completion (i.e., 4 weeks), the lower-dose ondansetron and diazepam patients were more improved than those on placebo. This study, which slightly favored the 1.0 mg t.i.d. ondansetron group, was complicated by a high placebo response (45% met improvement criteria on the HAM-A). On the Global Improvement Scale, the improvement rates were 50%, 43%, 48%, and 38%, respectively, for patients on ondansetron 1.0 mg t.i.d., ondansetron 4.0 mg t.i.d., diazepam 2.0 mg t.i.d., and placebo. The study author also claims an absence of abrupt discontinuation-related rebound anxiety for both ondansetron groups yet "significant rebound" for the diazepam patients (i.e., a 1-week later HAM-A increment of 2.3 points) (56).

Further discussion and understanding of ondansetron is not possible at this time because the trial just noted is published only in summary form, and no other data, including trials in the United States, appear to be published. Limited data have also been published for other 5-HT3 receptor antagonists (e.g., zacopride [its enantiomers may have differential activity] (2, 12), tropisetron, zatosetron) but not for granisetron or bemesetron.

Hepatic metabolism data have recently been published for both ondansetron and tropisetron (33). In studies using human liver microsomes, the oxidative metabolism of tropisetron was reduced to a greater degree by 1.0 mM quinidine than was the case for ondansetron (67% and 18%, respectively), suggesting that both drugs are substrates for cytochrome P-450 2D6. By contrast, ondansetron hydroxylation was modestly reduced (27%) by the 3A subfamily inhibitors, cyclosporin and triacetyloleandomycin, and this was even less the case for tropisetron (<10%). Coadministration of drugs that are known inhibitors of these isozymes (e.g., specific serotonin reuptake inhibitors) could have both pharmacokinetic and pharmacodynamic consequences. In addition to being substrates to varying degrees for these isozymes, both 5-HT3 antagonists appear to be weak competitive inhibitors of the 2D6 pathway; however, the inhibitory constants for both drugs are well above their likely therapeutic concentrations. Ondansetron (Ki = 31 mM) appears to have comparable effects as a weak competitive inhibitor for the 3A subfamily pathway (e.g., cyclosporin metabolism), while tropisetron's inhibiting effects would seem to be inconsequential (Ki = 2.1 mM).

COMMENT

Benzodiazepines continue to be consistently effective agents for acute anxiety in patients with GAD and panic disorder. Although not totally free from unwanted problems, their benefit-to-risk ratio is extremely favorable. Unfortunately, no benzodiazepine receptor partial agonist has emerged as a viable alternative. Azapirone 5-HT1A receptor partial agonists are somewhat less predictably effective than benzodiazepines in patients with GAD and are relatively ineffective in the acute treatment of panic attacks; nevertheless, they offer some advantages over benzodiazepines in patients for whom they prove to be effective (e.g., less sedation or memory impairment). A major limitation in the use of azapirones is their slow onset of action. Tricyclic antidepressants and SSRIs also have a slow onset of action. For patients who tolerate the side effects that are characteristic of either class, agents from both classes may reduce symptomatology in GAD or significantly lower the frequency of panic attacks in patients with panic disorder. Cholecystokinin antagonists and 5-HT3 receptor antagonists have the potential for a more rapid onset of anxiety reduction; unfortunately, further research will be essential to clarify their place in the treatment of acute anxiety.

This work was supported, in part, by a grant (MH-34223) from the Department of Health and Human Services.

published 2000