Understanding the assessment of psychotropic drug harms in clinical trials to improve social workers' role in medication monitoring.
For the past generation, drug therapy has become a primary form of helping mental health clients, many of whom derive substantial benefits from it. The responsible use of modern medications for all of the helping professions, however, requires ongoing education that includes not only the intended effects of psychotropic medications, but also any unintended, harmful effects. Based on a largely unmet need in the social work literature, the purpose of this integrative review is to facilitate social work practitioners' understanding of current practices in clinical trial research for assessing psychotropic drugs' harm. This could lead to more significant involvement of practitioners in detecting psychotropic drug harms in their clients, thereby contributing to improved practices in mental health. We performed a comprehensive review of interdisciplinary sources on the policies, procedures, and practices of assessing harms in clinical drug testing, including U.S. Food and Drug Administration (FDA) guidelines for safety analyses, biomedical and statistical reviews of trial design limitations and possibilities for assessing harms, and descriptions of actual practices in assessing and reporting harms of psychotropic drugs. This integrative review is presented in the spirit of advancing evidence-based practice, in which highlighting what we do not know is as important in informing clients as is sharing what we do know, both desirable and undesirable (Gambrill, 2006). The review concludes with recommendations for social work practice, research, and policy.
Extensive pharmaceutical industry involvement in producing prescription drug knowledge and disseminating it to professionals, policymakers, and the public has resulted in well-documented publication biases and considerable uncertainty about drug efficacy and safety (Medawar, Hardon, & Herxheimer, 2004; Melander, Ahlqvist-Rastad, Meijer, & Beermann, 2003; Perlis et al., 2005; Turner, Matthews, Linardatos, Tell, & Rosenthal, 2008; Wazana, 2000). There have been publicized concerns about the association of suicidal thinking and behavior with antidepressant, stimulant, and anticonvulsant drugs (Alonso-Zaldivar, 2006; Healy & Whitaker, 2003; Mundy, 2008), and discoveries during litigation against drug companies about hidden or minimized adverse events (AEs) (Avorn, 2006; Kesselheim & Avorn, 2007; Kondro & Sibbald, 2004; State of Connecticut v. Eli Lilly and Company, 2008). As a result, professionals and the public are alert to the fact that some drug effects, especially of a behavioral or psychological nature, remain unrecognized at the time of drug approval and are too tardily recognized after a drug has reached the market. Such uncertainty as to the nature and intensity of drug effects poses a serious problem for making informed decisions about the drug treatment of distress and disorder.
As a preliminary to this review, we sought to assess social work's involvement in evaluating drug safety issues by searching Social Work Abstracts, Social Services Abstracts, and Sociological Abstracts (for the previous 15 years) for the key words side effects, adverse effects, harm, safety, and psychopharmacology along with medication or psychotropic. We retrieved 38, 175, and 120 citations, respectively. Reviewing this literature, we concluded that current textbooks (Bentley & Walsh, 2006; Dziegelewski & Leon, 2001) and review articles (Bentley, 2003; Farmer, Bentley, & Walsh, 2006; Walsh, 1999) on psychopharmacology and social work largely aim to educate social workers on types of medication, indications, known side effects, concerns for special populations, and ethical issues with regard to professional values. Few reports, however, scrutinize clinical trials. Otis and King (2006) advised counselors on various unanticipated effects and discontinuation reactions that remain unaccounted for in clinical trials. Cohen (2002) and Cohen and Jacobs (2007) outlined methodological and conceptual limitations of randomized controlled drug trials, raising doubts about the scientific evidence supporting widespread drug use. It appears that social workers are continually encouraged to understand and monitor drugs' side effects, but little discussion occurs among social workers about how such effects are identified and communicated in drug research.
OVERVIEW OF CLINICAL TESTING OF PSYCHOTROPIC DRUGS
Psychotropic or psychoactive drugs discussed in the present review are those available by prescription, consumed as medicines, and intended--by definition--to alter the behavior, mood, thinking, or feeling of the individual ingesting the drug. Psychotropic drugs commonly produce a variety of anticipated and unexpected, gross and subtle, acute and long-lasting, physical and psychological effects in an individual.
Most of the information available to physicians, nonmedical professionals, and the public about a drug's effects typically comes from clinical studies or tests required by the FDA of the drug's manufacturer (usually called the sponsor) for a new drug approval. The studies progress as follows: preclinical testing on animals; phase I testing on small numbers of healthy adults to ensure no gross toxicities; phase II open-label testing to find preliminary efficacy data and appropriate dose ranges in people with the targeted illness or condition; and phase III placebo-controlled testing in several sites, to provide more definitive efficacy and additional safety information from larger samples. Phase III trials are usually short-term (three to eight weeks), double-blind, randomized, controlled trials (RCTs), which are expected--assuming homogeneity of patients, their proper randomization, and strict "blindability"--to show causal links between treatment and outcome. The RCT is considered the "gold standard" method for ascertaining a drug's efficacy, which means that in at least two phase III trials, the drug has outperformed placebo (an inert pill) on symptom rating scales scores--or, given the null hypothesis-testing nature of trials, that the drug probably has a smaller than 5 percent chance of being less efficacious than a placebo (Avorn, 2004). The assessment of adverse effects or events (both defined as "any unfavorable and unintended sign [including an abnormal lab finding, for example], symptom, or disease temporarily associated with the use of a medicinal product, whether or not considered related to the medicinal product" [FDA, 1995, pp. 2-3]) is typically considered a secondary outcome and, as we document later, receives much less investment of energy and creativity than the assessment of intended effects. Nonetheless, most of the safety information found on the FDA-approved drug label derives from this phase of testing. Finally, phase III trials frequently occur in the context of multisite, often multicountry trials conducted by private firms called contract research organizations (CROs), hired and paid by the pharmaceutical company that intends to manufacture and exclusively market the tested drug.
GUIDELINES ON AE ASSESSMENT IN CLINICAL TRIALS
FDA guidelines stipulate the format of the clinical safety review prepared by the sponsor for FDA approval of a new drug. The review is intended to allow the FDA reviewer to decide "whether safety has been assessed by all methods reasonably applicable to evaluate safety and whether the drug is safe for its intended use" (Murphy & Roberts, 2006, p. 36). The term "safety" is ubiquitous in published clinical trial reports but may have conflicting uses. The FDA describes safety to mean that "the demonstrated benefits of a drug outweigh its known and potential risks for the intended population and use" (Galson, 2005, p. 6). The term also describes or implies an absence of harm, or an absence of evidence of harm from a drug (Enas & Goldstein, 1995; Ioannidis et al., 2004). In a 31-page guidance document on conducting and preparing a clinical safety review, only one paragraph outlines methods for eliciting AEs in clinical trials (FDA, 2005). It explains that open-ended questions or checklists may be used to elicit AEs and states that, depending on the method, "results can differ greatly and may lead to marked differences in reported adverse event rates across studies (it would not usually be appropriate to pool results obtained using both methods)" (FDA, 2005, p. 20).
The FDA stipulates that AEs recorded by the study clinicians should be coded by the sponsor according to one of two formal dictionaries: COSTART (Coding Symbols for Thesaurus of Adverse Reaction Terms) or MeDRA (Medical Dictionary for Regulatory Activities). MeDRA, for example, contains five hierarchical levels of terms, with the "preferred term" as the basic unit for reporting. For a physical effect, a verbatim report by the patient or clinician, such as "always tired," would likely be mapped to the preferred term "fatigue" (Nilsson & Koke, 2001). Verbatim terms for different conditions, such as "sunburn" and "rash," may both be coded to the preferred term "rash." For psychological and behavioral effects, the connection between terms is less straightforward. Preferred terms for a reported event such as an angry outburst include "affect lability," "aggression," "behavior abnormal," "disturbance in social behavior," "irritability," and "mood swings."
Despite the standardized list of terms, there is "considerable discretion to the classifier to choose the term that best reflects the verbatim term reported by the investigator" (FDA, 2005, p. 20). Terms should not be too narrow, too broad, or exaggerate or minimize a finding, and they should be consistent across treatment groups and sites. AEs may also be graded for intensity as mild, moderate, or severe, but no standardized definitions or criteria exist for assigning such levels (Vitiello et al., 2003). Consistency among investigators in interpreting and coding AEs is complicated by the geographic dispersion, cultural and language differences, and oversight difficulties inherent in multisite, multicountry trials (Getz, 2007). Moreover, marketing imperatives to "get a drug through" the FDA hurdle and receive market approval and possibly generate huge financial returns to its sponsor may impede the collection of comprehensive and reliable information (Busfield, 2006).
Finally, the FDA (1995) has required the sponsor to evaluate all AE information received from site investigators and to report serious unexpected events within 15 days and life-threatening events within seven days to the FDA. The FDA further requests a sponsor to provide detailed information on each case of death, serious adverse event (defined as "any untoward medical occurrence that at any dose results in death, is life-threatening, requires inpatient hospitalization or prolongation of existing hospitalization, results in persistent or significant disability/incapacity, or is a congenital anomaly/birth defect" [FDA, 1995, p. 5]), or discontinuation from a trial due to an AE (FDA, 2005). Tables for common AEs (as defined by the sponsor) and a review of laboratory findings should also be included. This review of source AE data, as well as private negotiations between FDA and the sponsor, determines the nature and presentation of the "official" safety information printed in the drug's label.
Thus, the FDA-approved label is supposed to be "the closest one can get to the truth regarding the scientific information known about a drug" (Murphy & Roberts, 2006, p. 36). Until the Supreme Court ruling in Wyeth v. Levine (2009), the FDA during the past decade supported the position that a sponsor should not be expected to provide AE information other than that required to be printed in the drug label (Kesselheim & Avorn, 2007). At the same time, other litigation has revealed negotiations between the FDA and a pharmaceutical company spanning five years over changing the drug's label to include AE data that had not been available to the public. Except for a few cases in which evidence is revealed during litigation, it remains largely unknown what information in the drug label remains absent, minimized, or reconfigured as a result of the negotiation process. Increasingly, the FDA is making its own safety and clinical reviews of drugs available directly to the public through its Web site, and these provide additional details of effects sparsely covered in the drug label (Cohen, Hughes, & Jacobs, 2009).
In addition to FDA guidelines, the National Institutes of Health (NIH) has endorsed Data and Safety Monitoring Plans (DSMPs) for every clinical trial (FDA, 1996; NIH, 1998). A DSMP must specify parameters for collecting AE information, such as the instrument, the frequency of observations, and plans for event reporting. However, industry observers agree that the DSMP is "frequently neglected" and "often less well defined than other components of a research protocol," resulting in a number of FDA warning letters to investigators for failing to follow or implement safety monitoring (White, Field, & Wolf, 2007, p. 52). Investigative reports have further revealed that some clinical trials are not in compliance with either medical ethics or federal regulations for safety, implementing few or no safeguards to protect subjects' lives (U.S. Department of Health and Human Services, 2007).
The globalization of clinical trial research contributes to these deficiencies. In 2006, 41 percent of clinical trial principal investigators under FDA regulation operated outside the United States (Getz, 2007), where cost savings and easier access to patients attract CROs. However, the incidence of poor-quality trial data and investigator noncompliance and fraud in informed-consent AE reporting and other domains is "substantially higher" for these investigators. The FDA does not possess the resources to closely monitor sites globally, making these problems difficult to contain.
SCOPE OF AE ASSESSMENT IN RCTS
RCTs in psychopharmacology are designed to test hypotheses about the efficacy of the study drug (FDA, 1995). A clinical trial is
not designed to test specified hypotheses about safety nor to measure or identify adverse reactions with any pre-specified level of sensitivity.... The approach typically followed is to screen broadly for adverse events and to expect that this will reveal the common adverse reaction profile of a new drug and will detect some of the less common and more serious adverse reactions. (FDA, 1995, p. 5)
Thus, this experimental design is not designed or powered to detect the multiple, diverse, and sometimes unanticipated AEs inherent in any form of drug use, especially psychotropic drug use (Cohen & Jacobs, 2007; Ioannidis & Lau, 2002).
The short duration and limited sample size of the RCT is best suited to identify acute and proximal AEs rather than those that emerge after long-term drug exposure or distal to drug exposure (Vitiello et al., 2003). A nine-week trial enrolling approximately 300 participants, for example, could detect short-term events occurring in 1/100 subjects, the rate defined as "frequent" by the FDA. A trial with a sample size of 100 and a hypothetical 3 percent AE rate is underpowered to detect events significantly because the confidence interval (CI) (the estimated interval at which the effect would occur in the population) would include zero (CI: -0.3 percent to 6.3 percent) (Rief, Avorn, & Barsky, 2006). Infrequent events occurring at a rate between 1/100 and 1/1,000 and rare events occurring less often than 1/1,000 are not likely to be detected in most clinical trials. However, infrequent events, such as liver toxicity, may be detected early by closely following more common, less serious events, such as transaminase elevation. Hepatoxicity is the most frequent reason why drugs are pulled from the market, but its discovery could be expedited in some cases with reliable and systematic premarketing assessment (McPherson & Hemminki, 2004; Temple & Himmel, 2002). Similarly, rare life-threatening events are not expected to show in a clinical trial database because the appearance of such events in preclinical animal studies or uncontrolled trials would obviate future testing in large RCTs. The occurrence of a single rare life-threatening event in a phase III trial, then, "would represent a serious safety problem" for a drug (FDA, 2005, p. 9). It is important to also note that a finding of zero events in a finite sample (0/n) still carries a CI with an upper limit of 3/n, signifying that the true rate of the event in the population could exceed the zero estimate (Hanley & Lippman-Hand, 1983). Findings of no serious AEs in a clinical trial or nonstatistically significant differences in rates of AEs between treatment groups, then, do not exclude the possibility that major AEs remain undetected (Ioannidis, Mulrow, & Goodman, 2006). As we discuss briefly in the next section, this may be especially the case regarding events of a psychological or behavioral nature. These may be more subtle in presentation or insidious in onset and easily confused with subjects' co-occurring psychological difficulties, given that individuals must meet criteria for one or more mental disorders to participate in psychopharmacology phase II or III clinical trials.
AE ELICITATION METHODS
The identification of frequent or reasonably expected AEs, and perhaps less frequent ones, depends on the consistent use of a range of assessment methods during the trial (Rief et al., 2006). To capture both anticipated and unexpected AEs, those conducting the RCT would need to use a "flexible vigilance method," with both drug-specific AE scales and systematic assessments of a broader range of physical and psychological signs and symptoms (Greenhill et al., 2003). Flexible vigilance is particularly important for capturing psychological and behavioral effects. In a classic article describing "behavioral toxicity" caused by various substances, including psychopharmacological agents, Summerfield (1978) explained that higher level mental faculties are typically affected before lower level physical functions:
A very serious consequence [of intoxication] is a loss of self-critical monitoring of whatever one may be doing. The loss of concentration can easily be such that the carrying out of any kind of skilled activity is ... impaired in the particularly dangerous way that the person concerned is unaware of the process of behavioral deterioration to which he or she is being subjected. ... Only therefore by looking for impairments of functions immediately dependent on the highest levels for their control and coordination might any adverse effect be detectable at all. It is a profound conceptual issue that has spent more time in oblivion than in recognition. (pp. 336-337)
Currently, excluding basic laboratory tests of normal physical function, three primary methods exist for eliciting information about AEs. Spontaneous, open-ended questioning is the most common method used--or reported--in psychopharmacology trials (Greenhill et al., 2004; Wernicke, Faries, Milton, & Weyrauch, 2005). The investigator may ask the participant a general question such as, "Have you had any health problems since your last visit?" In most trials, visits are held weekly or every two weeks, and they usually last 15 to 20 minutes (during which other assessments, of therapeutic effects and vital signs, are also conducted). A second method is a drug-specific checklist. A trial of a stimulant medication may include a list of anticipated AEs--for example, appetite suppression, abdominal pain, or insomnia--that the subject or investigator is expected to identify easily. This inquiry approach provides an opportunity to characterize events by timing, frequency, duration, and severity (Enas & Goldstein, 1995). Finally, a body system review is a detailed inquiry covering any number of potential AEs associated with each body area or bodily system.
Each approach has proponents and critics. Open-ended inquiry is most frequently defended by industry sponsors, whereas researchers and academics with no industry links urge that comprehensive assessments be implemented. Critics regard open-ended questioning as unreliable and vulnerable to the variability with which participants to observe, recall, have associations with the drug, and report AEs (Vida & Looper, 1999). In particular, events affecting higher level cognition are unlikely to be identified and spontaneously reported by the participant within the standard time frame of a RCT. Evidence also exists that some trial participants make a living from this activity and have learned not to report any problem to ensure a smooth trial and earn their full compensation at trial's end (Smith & Evans, 2005). In sum, AE rates may be underestimated using this method and often are highly variable across trials. The advantages of the method remain its speed and ease of administration.
A structured assessment results in higher AE rates than open-ended questioning and provides more reliable and comparable data across trials (Bent, Padula, & Avins, 2006; Wallin & Sjovall, 1981). AEs that are difficult to collect because of participant embarrassment, such as sexual dysfunctions, or the subtlety of the event, such as declining or dulled cognitive function, are more likely to be elicited with a specific inquiry (Vida & Looper, 1999). However, critics of specific inquiries argue that these amass irrelevant complaints commonly occurring in the general population and obscure the ability to discriminate true drug-related events (Bent et al., 2006). These problems might be avoided by taking structured baseline assessments to increase the rate of reported baseline events and by conducting pre-and posttreatment comparisons. A second limitation to specific assessment is that the range of potential events is restricted to expected or probable ones from a given drug.
Finally, the body system review, which systematically covers a wider range of potential AEs, still largely focuses, in practice, on physical events and requires more time to complete. Cost considerations weigh heavily here, as even small changes in procedures can translate into considerable expenses, an unattractive option for sponsors who spend very large sums to conduct clinical trials and, in a fiercely competitive climate, attempt to maximize returns on their uncertain initial investment (Gassman, Reepmeyer, & von Zedtwitz, 2008; Goozner, 2004).
A few examples illustrate discrepancies that may result from different elicitation methods. Depending on the method, the rate of sexual dysfunction in selective serotonin reuptake inhibitor (SSRI) antidepressants trials ranges from 2 percent to 73 percent (Safer, 2002). According to Safer, "in one instance, a drug company-sponsored review covering over 3,000 subjects treated with SSRIs simply did not list any sexual side effects on its 23-item side effect table" (p. 585). In two pediatric trials and one adult trial in which both open-ended and specific inquiry were used, the open-ended method more often produced differences in AE rates between drug and placebo groups (Wernicke et al., 2005). The specific inquiry produced fewer overall differences between groups but a higher number of statistically significant differences. A pediatric study in which 59 youths receiving psychotropic medication were interviewed using each of the three methods in succession resulted in 195 elicited AEs (Greenhill et al., 2004). Fifty (26 percent) AEs were identified by the open-ended inquiry, an additional 16 (8 percent) by the drug-specific inquiry, and an additional 129 (66 percent) by the body system review. Of 17 events rated as severe, six (37 percent) were detected only by the body system review. Of 37 events that led to some medical or clinical action, 19 (53 percent) were identified by the body system review. This method took an average of 19 minutes to administer, compared with four minutes for the other methods. The authors of the study concluded that the longer duration made it unlikely that a body system review would be incorporated into standard assessment procedures.
CURRENT PRACTICES IN AE ASSESSMENT IN CLINICAL TRIAL RESEARCH
With no guidelines or broad consensus to hold sponsors and investigators accountable, how consistent and adequate are current practices in drug research? Cohen and Hughes (2003) analyzed 11 publications reporting on 1,584 subjects who had taken atomoxetine (Strattera) as part of phase II and III trials. Atomoxetine was approved by the FDA in 2002, in what its sponsor claimed was an unusually stringent regulatory environment, and heralded as the first drug indicated for both adults and children diagnosed with attention-deficit/hyperactivity disorder. On average, four rating scales per study were used to assess efficacy (range: 3 to 7), compared with 0.1 scales to assess AEs (range: 0 to 1). The average number of words in the publication describing methods for evaluating efficacy was 211 (range: 61 to 394), compared with 21.2 for safety (range: 0 to 35). Nine of the 11 publications (81.8 percent) used unsolicited, spontaneous reports or open-ended questioning to assess AlEs. In an updated analysis of premarketing and postmarketing studies of atomoxetine, the authors found that little had changed in assessment methods despite a number of safety warnings about that drug being issued by the FDA in the intervening years (Cohen et al., 2009).
A 2003 systematic review of AE elicitation methods in pediatric psychopharmacology retrieved 196 citations, including 65 (33 percent) that used open-ended inquiry methods, 45 (23 percent) that used drug-specific checklists, and 86 (44 percent) that focused on basic laboratory tests to screen for AEs (Greenhill et al., 2003). No study used a comprehensive body system review. The use of laboratory tests as the principal means of monitoring for AEs is especially inadequate because it disregards psychological and behavioral events. To assess a psychoactive substance as if its primary undesirable effects can be determined by measurable bodily changes--such as an increase or decrease in heart rate, blood pressure, or weight--is a gross simplification of psychotropic drugs' actions. Greenhill et al. (2003) also found that of 27 trials of stimulant medications, all but two questioned only a parent on the behavior of the youth. The type of informant accounted for the largest source of variation in AE assessment using a drug-specific checklist, with direct elicitation of the youth, resulting in a wider range of AEs than asking the parent or the teacher.
AE information from other medical- and health-related areas appears similarly insufficient, and even neglected, in published clinical trial research (Bernal-Delgado & Fisher, 2008; Ioannidis & Lau, 2001). Of 192 drug trials from seven medical areas, only 39 percent had adequate reporting of AE information, including providing reasons for study discontinuations and defining and reporting severity of AEs (Ioannidis & Lau, 2001). Reporting of AE information required a third of a page in most reports, taking up on average 12 percent of the results section. In 59 percent of reports, this space was equal to or less than that devoted to the authors' names and affiliations. A review of 185 pharmacology RCTs published in seven high-ranked medical journals found that approximately 75 percent of trials "merely said that patients were routinely assessed for the presence of ADRs [adverse drug reactions]" and provided no further elaboration (Loke & Derry, 2001, p. 3). Nearly 14 percent of trials made no mention of AEs in any part of the publication. Similar results are observed in reviews of published reports of mental health interventions, where over 80 percent of 142 trials were judged to have inadequate reporting of AEs (Papanikolaou, Churchill, Wahlbeck, & Ioannidis, 2004). Across 103 drug trials, less than one-tenth of a page, or 7 percent, of the entire Results section was devoted to reporting AEs. The researchers of one review remarked that "most high-quality trials amass an enormous amount of information about safety and adverse effects during their conduct as part of regulatory requirements. Yet the selective filtering of all these data into a quarter of a page can hardly be adequate" (Ioannidis & Lau, 2001, p. 443).
Additional evidence of unreliable assessment methods is reflected in widely varying AE rates in placebo-treated groups of drug trials (Rief et al., 2006, Vida & Looper, 1999). An analysis of trials for nine antidepressants found rates of headache and agitation in these groups ranging from 10 percent to 35 percent, and 2 percent to 22 percent, respectively (Vida & Looper, 1999).There was also high correlation (Pearson r [greater than or equal to] .7) of drug event rates with placebo event rates, with agitation, nervousness, and headache showing correlations of at least .95. Because the variation in placebo event rates is not likely caused by pharmacologic properties of the inert placebo, the high correlation suggests that "a significant proportion of active drug AER [adverse event rate] heterogeneity is associated with nonpharmacologic heterogeneity," such as miscellaneous methodological factors (Vida & Looper, p. 419). The occurrence of AEs in inert placebo groups is poorly understood and subject to multiple interpretations (Weihrauch & Gauler, 1999), but an obvious inference is that "if the great variability of adverse effect reports in placebo groups is due to unreliable and invalid assessment methods, the same problem applies for adverse effect reports in active treatment groups" (Rief et al., 2006, p. 158).
Related to this observed deficiency in assessment are inconsistent or plainly misleading methods for recording and coding AEs. In 2003, the FDA requested several pharmaceutical companies to reexamine all AE data pertaining to suicidality from clinical trials of antidepressants in pediatric populations (Leslie, Newman, Chesney, & Perrin, 2005). Each company responded to the request using different methods and thresholds for coding suicidality. For example, some companies only recorded events that investigators considered to have resulted from active treatment, whereas others recorded all events regardless of a presumed association with treatment. AE terms were coded inappropriately--for instance, using the ambiguous term "emotional lability" in place of "suicidal ideation"--resulting in the true frequency of suicide-related acts being obscured in original FDA reviews of these trials. In a second, independent analysis of antidepressants and suicidality, FDA clinical safety reviews for sertraline (Zoloft) and paroxetine (Paxil) were revealed to inaccurately assign suicidal events occurring during the placebo lead-in phase of a trial (which occurs before randomization of the participants to a treatment group) to the placebo treatment group (Healy & Whitaker, 2003). Across studies, removing these events from the placebo treatment groups would have reduced the number of suicidal events in these groups by half and considerably increased differences in suicidality between active treatment and placebo groups.
IMPLICATIONS OF INADEQUATE AE ASSESSMENT METHODS
Widely varying and noncomparable AE rates across clinical trials make it difficult for physicians and patients to make informed decisions based on a benefit-to-harm relationship (Bent et al., 2006). The low statistical power of a single RCT to comprehensively evaluate multiple AE endpoints can be partially addressed by meta-analytic techniques that analyze results across many trials, potentially providing more robust estimates for frequencies of AE, identifying subgroups of patients who may be at greater risk of harm, and detecting less common AEs (Brewer & Colditz, 1999). Unfortunately, inconsistent assessment methods prevent the comparability of results and merging of data from different trials (Rief et al., 2006). Meta-analysis of AE information remains difficult and inconclusive (Ioannidis & Lau, 2002).
With such limited and unreliable information from RCTs about a drug's AE profile, the introduction of a drug to the market "should essentially be considered a therapeutic experiment" (Wood, 2000, p. 1825). In some cases, millions of individuals may be exposed to a new drug within months of its FDA approval. It is not surprising then that half of drug withdrawals occur within two years of the drug's introduction (Lasser et al., 2002). Between 1975 and 1999, of 548 new chemical entities that were approved for marketing, 45 (8.2 percent) eventually received a "black box" on the official drug label, the FDA's strongest warning for serious or life-threatening AEs, and 16 (2.9 percent) were withdrawn from the market (Lasser et al., 2002). More recently, between 1998 and 2005, serious AEs reported to the FDA through MedWatch, its passive postmarketing surveillance system, increased 2.6-fold, and reported deaths increased 2.7-fold (Moore, Cohen, & Furberg, 2007). This spontaneous, voluntary surveillance system mostly used by health care professionals, however, is known to capture as little as only 1 percent of actual adverse drug reactions (Kessler, 1993). Postmarketing surveillance studies, known as phase IV trials, are expected to provide a more realistic portrait of drugs' AEs, as these are assessed during ordinary conditions of use, with more ordinary patients than highly selected clinical trial participants. However, as of 2006, more than 70 percent of phase IV trials that sponsors had agreed to conduct as a condition of FDA approval of their drug had not yet started (Avorn, 2007).
Overall, a substantial portion of the safety information about psychotropic drugs--especially newer ones--derives from clinical trials, yet for many reasons explored in this review, these trials do not allow clinicians to construct a realistic portrait of potential and expected drug harms. We suggest possible actions of social workers at research, practice, and policy levels that might remedy some of this deficiency and increase the knowledge base of the profession as it continues its major involvement with medicated clients (see Table 1).
It has been recommended that psychiatric drugs should be studied first and foremost as psychoactive drugs (Jacobs & Cohen, 1999). This means that prior to measuring a narrow effect of interest in RCTs, a variety of other methods should have already revealed the drug's full range of psychoactive effects (Charlton, 2001). Phase I trials monitoring healthy adults for gross toxicity and phase II trials seeking an optimum dose range are designed for FDA approval and barely contribute to drug knowledge in the aforementioned sense. Unfortunately, there is no preestablished protocol for human psychoactive drug investigations designed to reveal important and subtle psychological alterations produced by drugs, some of which may be experienced as adverse, unpleasant, and undesirable. More in-depth, exploratory methods suited for this purpose are needed prior to phase III clinical trials and within which phase III results can be interpreted. Social workers' contribution to the development of these methods could include collecting ethnographic accounts of clients' personal difficulties and ongoing treatment effects to supplement the diagnoses and numerical scores by which difficulties and effects are summarized today (Cohen & Jacobs, 2007). Social workers could elicit and document their client's version of treatment effects and assess the client's story and behavior in the context of what is known as a result of standard drug research and what remains unknown because of limitations in this research. Also, in clinical trial research, the use of informants who know trial participants well and can supplement evaluations conducted by (stranger) investigators is in its infancy and needs more development. Until such methods are implemented with creativity, care, and rigor, commercially funded claims about both the efficacy and tolerability of newly introduced drugs should be scientifically scrutinized and interpreted with great caution.
Related to this, social work practitioners should move beyond medication management roles that emphasize medication compliance to more substantive roles in minimizing drug harms and ensuring client safety. A comprehensive, one-page treatment emergent effects checklist assembled by Cohen and Sengelman (2008) for clinical practice is freely available for download at the independent, publicly funded CriticalThinkRx project Web site (http://www.criticalthinkrx.org/pdf/m7/ Module7-Medication-Monitoring-Checklist.pdt). It covers a wide range of physiological, psychological, and behavioral effects commonly associated with prescribed psychotropic agents. Practitioners should complete it with clients before, during, and after medication use to track new and ongoing effects and make sound judgments of effects' relation to drug treatment, and they should share the results of their assessments with clients, families, and other professionals. Because they have been downplayed in conventional research studies, numerous effects of psychotropic drugs (including impaired cognition and judgment, agitation, aggression, psychological indifference, and others, such as drug discontinuation reactions) are not widely discussed. Although they might cause substantial disability, the effects are unlikely to be attributed by clients to drugs' actions and are least likely to be so reported by practitioners to MedWatch, for example. Increased sensitivity and vigilance by social workers to such effects would contribute to the formation of a more accurate portrait of the benefits and harms of psychotropic drugs prescribed as medicines and could prevent unnecessary suffering.
Finally, at the policy level, the production of knowledge about a drug's effects must not be monopolized by the one stakeholder that has immense financial stakes in the drug's commercial success. The slow recognition of many AEs is likely explained in significant part by conflicts of interest throughout the drug testing and reporting process due to extensive pharmaceutical industry involvement (Angell, 2008). To improve this situation, control over the conduct and dissemination of scientific research must be removed from pharmaceutical companies or greatly diversified. Some scholars suggest increased public funding of drug research and the involvement of various private for-profit and nonprofit stakeholder groups in funding (Baker, 2008; Vitiello, Heiligenstein, Riddle, Greenhill, & Fegert, 2004). Former New England Journal of Medicine editors recommended that clinical trials still be primarily funded, but not controlled, by pharmaceutical companies (Relman & Angell, 2002). They suggested setting up a centralized nonprofit institution to mediate between research proposals and the distribution of funds, which would be awarded to academic centers and teaching hospitals to conduct research. Alternatively, a department within NIH could be established to administer clinical trials of prescription drugs (Angell, 2008).
Deficiencies of our drug safety system require involvement from those positioned at all levels in the system of testing, administering, and monitoring prescription drugs. We think that social workers should support policy initiatives aimed at displacing pharmaceutical companies as the primary sponsors of research for the drugs they market. In addition, as more social workers are involved in the delivery of mental health services than any other professional group, social workers are in a position to play key roles in supplementing present limited understanding of drugs' effects by systematically eliciting and publishing comprehensive accounts of clients' treatment experiences and finding creative ways to include these accounts in "what is known" about desirable and undesirable drug effects.
The number of adults and youths in distress who are treated with psychopharmacological agents continues to rise, and the uncertainties surrounding the effects of drug treatments are proving to be enormously consequential. The identification, description, and prevention of harm to clients from psychotropic drugs will surely rank as a top feature of mental health work in immediate years to come.
Original manuscript received October 4, 2008
Final revision received June 4, 2009
Accepted June 15, 2009
Alonso-Zaldivar, R. (2006, March 23). FDA panel urges strong warning on ADHD drugs: Cases of hallucinations cited in advisers' recommendation. Retrieved October 1, 2007, from http://www.sfgate.com/cgi-bin/article. cgi?f=c/a/2006/03/23/MNGSDHSMI71.DTL& hw=fda+panel+urges+strong+warning+on+ ADHD+drugs&sn=001&sc=1000
Angell, M. (2008). Industry-sponsored clinical research: A broken system. JAMA, 300, 1069-1071.
Avorn, J. (2004). Powerful medicines: The benefits, risks, and costs of prescription drugs. New York: Alfred A. Knopf.
Avorn, J. (2006). Dangerous deception: Hiding the evidence of adverse drug effects. New England Journal of Medicine, 355, 2169-2171.
Avorn, J. (2007). Paying for drug approvals--Who's using whom? New England Journal of Medicine, 356, 1697-1700.
Baker, D. (2008).The benefits and savings from publicly funded clinical trials of prescription drugs. International Journal of Health Services, 38, 731-750.
Bent, S., Padula, A., & Avins, A. (2006). Better ways to question patients about adverse medical events. Annals of Internal Medicine, 144, 257-261.
Bentley, K. (2003). Introduction to the special issue: Psychiatric medication issues for social workers, counselors, and psychologists. Social Work in Mental Health, 1(4), 1-3.
Bentley, K., & Walsh, J. (2006). The social worker and psychotropic medication: Toward effective collaboration with mental health clients, families, and providers (3rd ed.). Pacific Grove, CA: Brooks/Cole.
Bernal-Delgado, E., & Fisher, E. (2008).Abstracts in high profile medical journals often fail to report harm. BMC Medical Research Methodology, 8(14). Retrieved from http://www.biomedcentral.com/14712288/8/14
Brewer, T., & Colditz, G. (1999). Postmarketing surveillance and adverse drug reactions. JAMA, 281, 824-829.
Busfield, J. (2006). Pills, power, people: Sociological understandings of the pharmaceutical industry. Sociology, 40, 297-314.
Charlton, B. (2001). Fundamental deficiencies in the megatrial methodology. Current Controlled Trials in Cardiovascular Medicine, 2(1), 2-7.
Cohen, D. (2002). Research on the drug treatment of schizophrenia: A critical appraisal and implications for social work education. Journal of Social Work Education, 38, 1-24.
Cohen, D., & Hughes, S. (2003, October 4). What's in a non-stimulant? A preliminary review of published studies of Strattera (atomoxetine). Paper presented at the 7th Annual Conference of the International Center for the Study of Psychiatry and Psychology, Chicago.
Cohen, D., Hughes, S., & Jacobs, D. (2009). The deficiencies of drug treatment: The case of Strattera. In S. Timimi & J. Leo (Eds.), Rethinking ADHD: From brain to culture (pp. 313-333). New York: Palgrave Macmillan.
Cohen, D., & Jacobs, D. (2007). Randomized controlled trials of antidepressants: Clinically and scientifically irrelevant. Debates in Neuroscience, 1, 44-54.
Cohen, D., & Sengelman, I. (2008). CriticalThinkRx: A critical curriculum on psychotropic medications. Retrieved September 25, 2008, from www.criticalthinkrx.org
Dziegelewski, S., & Leon, A. (2001). Social work practice and psychopharmacology. New York: Springer.
Enas, G., & Goldstein, D. (1995). Defining, monitoring, and combining safety information in clinical trials. Statistical Medicine, 14, 1099-1111.
Farmer, R., Bentley, K., & Walsh, J. (2006). Advancing social work curriculum in psychopharmacology and medication management. Journal of Social Work Education, 42, 211-229.
Galson, S. (2005). Statement on initiatives regarding drug safety. Retrieved September 12, 2007, from http://www. fda.gov/ola/2005/drugsafety0505.html
Gambrill, E. (2006). Evidence-based practice and policy: Choices ahead. Research on Social Work Practice, 16, 338 357.
Gassman, O., R.eepmeyer, G., & von Zedtwitz, M. (2008). Leading pharmaceutical innovation: Trends and drivers for growth in the pharmaceutical industry (2nd ed.). Berlin: Springer-Verlag.
Getz, K. (2007). Global clinical trials: Activity in the details. Applied Clinical Trials, 16(9), 42-44.
Goozner, M. (2004). The $800 million pill: The truth about the cost of new drugs. Berkeley: University of California Press.
Greenhill, L.,Vitiello, B., Fisher, P., Levine, J., Davies, M., Abikoff, H., et al. (2004). Comparison of increasingly detailed elicitation methods for the assessment of adverse events in pediatric psychopharmacology. Journal of the American Academy of Child and Adolescent Psychiatry, 43, 1488-1496.
Greenhill, L., Vitiello, B., Riddle, M., Fisher, R, Shockey, E., March, J, et al. (2003). Review of safety assessment methods used in pediatric psychopharmacology. Journal of the American Academy of Child and Adolescent Psychiatry, 42, 627-633.
Hanley, J., & Lippman-Hand, A. (1983). If nothing goes wrong, is everything all right? JAMA, 249, 1743-1745.
Healy, D., & Whitaker, C. (2003). Antidepressants and suicide: Risk-benefit conundrums.Journal of Psychiatry and Neuroscience, 28, 331-337.
Ioannidis, I., Evans, S., Gotzsche, R, O'Neill, R., Altman, D., Schulz, K., et al. (2004). Better reporting of harms in randomized trials: An extension of the CONSORT statement. Annals of Internal Medicine, 141,781-788.
Ioannidis, I., & Lau, J. (2001). Completeness of safety reporting in randomized trials: An evaluation of 7 medical areas. JAMA, 285, 437-443.
Ioannidis, I., & Lau, J. (2002). Improving safety reporting from randomized trials. Drug Safety, 25(2), 77-84.
Ioannidis, I., Mulrow, C., & Goodman, S. (2006). Adverse events: The more you search, the more you find. Annals of Internal Medicine, 144, 298-300.
Jacobs, D., & Cohen, D. (1999).What is really known about psychological alterations produced by psychiatric drugs? International Journal of Risk & Safety in Medicine, 12, 37-47.
Kesselheim, A., & Avorn, J. (2007).The role of litigation in defining drug risks. JAMA, 297, 308-311.
Kessler, D. (1993). Introducing MEDWatch: A new approach to reporting medication and device adverse effects and product problems. JAMA, 269, 2765-2768.
Kondro, W, & Sibbald, B. (2004). Drug company experts advised staff to withhold data about SSRI use in children. Canadian Medical Association Journal, 170, 783.
Lasser, K., Allen, E, Woolhandler, S., Himmelstein, D., Wolfe, S., & Bor, D. (2002).Timing of new black box warnings and withdrawals for prescription medications. JAMA, 287, 2215-2220.
Leslie, L., Newman, T., Chesney, J., & Perrin, J. (2005). The Food and Drug Administration's deliberations in antidepressant use in pediatric patients. Pediatrics, 116, 195-204.
Loke,Y., & Derry, S. (2001). Reporting of adverse drug reactions in randonfised controlled trials--A systematic survey. BMC Clinical Pharmacology, 1(3). Retrieved September 15, 2007, from http://www. biomedcentral.com/1472-6904/I/3
McPherson, K., & Hemminki, E. (2004). Synthesising licensing data to assess drug safety. British Medical Journal, 328, 518-521.
Medawar, C., Hardon, A., & Herxheirner, A. (2004). Depressing research [Letter to the editor]. Lancet, 363, 2087.
Melander, H., Ahlqvist-Rastad,J., Meijer, G., & Beermann, B. (2003). Evidence-b(i)ased medicine--Selective reporting from studies sponsored by pharmaceutical industry: Review of studies in new drug applications. British Medical Journal, 326, 1171-1173.
Moore, T, Cohen, M., & Furberg, C. (2007). Serious adverse drug events reported to the Food and Drug Administration, 1998-2005. Archives of Internal Medicine, 167, 1752-1759.
Mundy, A. (2008). Drugs' link to suicide risk cause concern. Retrieved September 25, 2008, from http://online. wsj.com/public/article/SB121556144610237551. html?mod=2_1566_leftbox
Murphy, S., & Roberts, R. (2006)."Black box" 101: How the Food and Drug Administration evaluates, communicates, and manages drug benefit/risk.Journal of Allergy and Clinical Immunology, 117(1), 34-39.
National Institutes of Health. (1998). National Institutes of Health policy for data and safety monitoring. Retrieved December 6, 2007, from http://grants.nih.gov/ grants/guide/notice-files/not98-084.html
Nilsson, M., & Koke, S. (2001). Defining treatment-emergent adverse events with the medical dictionary for regulatory activities (MeDRA). Drug Information Journal, 35, 1289-1299.
Papanikolaou, P., Churchill, R.,Wahlbeck, K., & Ioannidis, I. (2004). Safety reporting in randomized trials of mental health interventions. American Journal of Psychiatry, 161, 1692-1697.
Perlis, R., Perlis, C., Wu,Y., Hwang, C., Joseph, M., & Nierenberg, A. (2005). Industry sponsorship and financial conflict of interest in the reporting of clinical trials in psychiatry. American Journal of Psychiatry, 162, 1957-1960.
Otis, H., & King, J. (2006). Unanticipated psychotropic medication reactions. Journal of Mental Health Counseling, 28, 218-240.
Relman, A., & Angell, M. (2002). America's other drug problem. New Republic, 227(25), 27-41.
Rief, W., Avorn, J., & Barsky, A. (2006). Medication-attributed adverse effects in placebo groups. Archives of Internal Medicine, 166, 155-160.
Safer, D. (2002). Design and reporting modifications in industry-sponsored comparative psychopharmacology trials.Journal of Nervous and Mental Disease, 190, 583-592.
Smith, M., & Evans, D. (2005). Miami test center lures poor immigrants as human guinea pigs. Retrieved September 25, 2008, from http://www.bloomberg. com/apps/news?pid= 10000006&sid=aspHJ_ sFen1s&refer=home
State of Connecticut v. Eli Lilly and Company. (2008). Retrieved September 25, 2008, from http://www. ct.gov/ag/lib/ag/health/elililly-zyprexa.pdf
Summerfield, A. (1978). Behavioral toxicity--The psychology of pollution. Journal of Biosocial Science, 10, 335-345.
Temple, R., & Himmel, M. (2002). Safety of newly approved drugs. JAMA, 187, 2273-2275.
Turner, E., Matthews, A., Linardatos, E., Tell, R., & Rosenthal, R. (2008). Selective publication of antidepressant trials and its influence on apparent efficacy. New England Journal of Medicine, 358, 252-260.
U.S. Department of Health and Human Services. (2007). The Food and Drug Administration's oversight of clinical trials (OEI-01-06-00160). Retrieved May 20, 2009, from http://www.oig.hhs.gov/oei/reports/oei-01-06-00160.pdf
U.S. Food and Drug Administration. (1995). Clinical safety data management: Definitions and standards for expedited reporting. Retrieved September 17, 2007, from http:// www.fda.gov/cder/guidance/iche2a.pdf
U.S. Food and Drug Administration. (1996). Good clinical practice: Consolidated guidance. Retrieved September 17, 2007, from http://www.fda.gov/cder/ guidance/959fnl.pdf
U.S. Food and Drug Administration. (2005). Reviewer guidance: Conducting a clinical safety review of a new product application and preparing a report on the review. Retrieved September 17, 2007, from http://www. fda.gov/cder/guidance/3580fnl.pdf
Vida, S., & Looper, K. (1999). Precision and comparability of adverse event rates of newer antidepressants. Journal of Clinical Psychopharmacology, 19, 416-426.
Vitiello, B., Heiligenstein, J., Riddle, M., Greenhill, L., & Fegert, J. (2004). The interface between publicly funded and industry-funded research in pediatric psychopharmacology: Opportunities for integration and collaboration. Biological Psychiatry, 56, 3-9.
Vitiello, B., Riddle, M., Greenhill, L., March, J., Levine, J., Schachar, R., et al. (2003). How can we improve the assessment of safety in child and adolescent psychopharmacology? Journal of the American Academy of Child and Adolescent Psychiatry, 42, 634--641.
Wallin,J., & Sjovall, J. (1981). Detection of adverse drug reactions in a clinical trial using two types of questioning. Clinical Therapeutics, 3, 450-452.
Walsh, J. (1999). Anti-anxiety medications: A review for social workers. Social Work in Health Care, 30(1), 31-49.
Wazana, A. (2000). Physicians and the pharmaceutical industry: Is a gift ever just a gift? JAMA, 283, 373-380.
Weihrauch, T., & Gauler, T. (1999). Placebo--Efficacy and adverse effects in controlled clinical trials. Drug Research, 49, 385-393.
Wernicke, J., Faries, D., Milton, D., & Weyrauch, K. (2005). Detecting treatment emergent adverse events in clinical trials: A comparison of spontaneously reported and solicited collection methods. Drug Safety, 28, 1057-1063.
White, S., Field, L., & Wolf, D. (2007). Monitoring the monitors. Applied Clinical Trials, 16(9), 52-60.
Wood, A. (2000). Thrombotic thrombocytopenic pupura and clopidogrel--A need for new approaches to drug safety. New England Journal of Medicine, 342, 1824-1826.
Wyeth v. Levine, 555 U.S. (2009).
Shannon Hughes, MSW, is a doctoral candidate, and David Cohen, PhD, LCSW, is professor, Robert Stempel College of Public Health and Social Work, Florida International University. Address correspondence to Shannon Hughes, School of Social Work, Florida International University, 11200 SW 8th Street, Miami, FL 33199; e-mail: firstname.lastname@example.org.
Table 1: Recommendations for Social Workers' Roles Regarding Psychotropic Drugs Possible Roles for Level Issued to Be Addressed Social Workers Research Lack of understanding of * Collect and publish psychotropic drugs' full ethnographic accounts of psychoactive effects clients' treatment experiences * Include data from informants who know the client well Practice Minimizing drug harms to * Use Medication Treatment clients and maximizing Emergent Effects clients' well-being Checklist with clients before, during, and after medication use * Share results of assessments with treatment team Policy Conflicts of interest * Support policy resulting from initiatives aiming to extensive pharmaceutical displace pharmaceutical industry involvement in companies as the primary drug testing and sponsor of research for reporting the drugs they market