Professional Documents
Culture Documents
Mejora de Validez en Investig Del Tratamiento de Comorbilidad
Mejora de Validez en Investig Del Tratamiento de Comorbilidad
655665, 2006
doi:10.1093/schbul/sbl009
Advance Access publication on July 18, 2006
Gregory J. McHugo1,2, Robert E. Drake2, Mary F. holism), drugs (National Institute on Drug Abuse), and
Brunette2, Haiyi Xie2, Susan M. Essock3, and mental illness (National Institute of Mental Health) has
Alan I. Green4 led to an unfortunate lack of responsibility and cooper-
2
Dartmouth Psychiatric Research Center, 2 Whipple Place, Suite ation concerning the study of co-occurring disorders.
202, Lebanon, NH 03766; 3Division of Health Services Research, Treatment research of co-occurring disorders has been
Mt Sinai School of Medicine; 4Department of Psychiatry, summarized recently in terms of specific psychosocial in-
Dartmouth Medical School terventions,8 residential interventions,9 treatment princi-
pertains to the generalizability of the findings across the ple, residential treatment programs must follow counter-
operations and measures of a study, and external validity acting and restrictive regulations of the Department of
pertains to the generalizability of the findings across peo- housing and urban development (HUD), and clinicians
ple, settings, and time. Co-occurring disorders research are often reluctant to assign unstable patients to cloza-
has been weak in all 4 domains of inferential validity. As pine by randomization. Despite several quasi-experimen-
the evidence-based medicine movement grows within tal studies that support residential dual-diagnosis
behavioral health, the lack of valid research will hinder treatment9 and clozapine,11 there have been no success-
the development of evidence-based practices, treatment fully completed rigorous RCTs of either intervention.
guidelines and algorithms, and decision supports for Despite the power and prestige of the RCT, there are
co-occurring disorders. many situations in which true experiments are impossi-
The accrual of evidence in support of an intervention ble, unnecessary, premature, or uninformative.20 Even
comes from many directions and with various claims to if randomization is possible in a given situation, experi-
scientific rigor and inferential validity. The challenge be- mental control may not be, so each threat to inferential
fore the co-occurring disorders field now is to design in- validity has to be evaluated carefully. For example, ran-
tervention studies that address key clinical questions in domization to a community treatment team vs standard
the efficacy of an intervention is necessary but seldom One such reality is the decision-making process in
sufficient to ensure its widespread adoption in routine which clinicians and patients engage regularly. If, on
practice settings. We believe that the widespread adop- the one hand, a study contains a rigid protocol for mak-
tion of interventions to treat substance abuse among peo- ing decisions about treatment dose and duration without
ple with co-occurring disorders will occur only when regard to co-occurring conditions, engagement in serv-
studies lead to inferences with high external validity ices, adherence to treatment, and patient preferences, it
and include procedures with high ecological validity. will have low ecological validity because its procedures
may not simulate the complex decision process that is
External Validity used in routine practice. If, on the other hand, a study
protocol reflects the usual decision making that deter-
A study has high external validity if its findings apply to
mines which intervention to try first, how long a given
the intended population of patients, providers, and set-
trial lasts, what outcomes are tracked, how to sequence
tings. Despite the importance of external validity, the de-
interventions, and how to handle nonadherence and
liberate focus on internal validity among investigators
dropout, it will more closely mimic the decision process
has led to a great deal of irrelevant and nongeneralizable
in routine care and have higher ecological validity.
adaptive designs have been used to evaluate the treatment treatment options as essentially equivalent in terms of
of depression in primary care settings.32,33 In these stud- the likelihood of success. Equipoise-stratified randomiza-
ies, rule-based but flexible strategies for care manage- tion is a method for retaining the maximum number of
ment, which included education, medication, specialty participants possible, while still allowing contrasts be-
care, and adherence, side effect, and outcome monitor- tween conditions (interventions) to which participants
ing, were contrasted with usual care and found to be have been assigned randomly. At the point of consent,
more effective. participants are informed of the treatment options of-
Because treatment for co-occurring disorders does not fered in the experiment, and they decide which options
have well-established guidelines currently, randomized are acceptable to them. As participants enroll in the ex-
adaptive designs are an appealing option. By adding ran- periment, a number of strata will emerge depending on
dom assignment at each decision point, studies can con- the subsets of the treatment options that participants
trast multiple treatment strategies, each of which involves are willing (and eligible) to accept. The only people
a sequence of options that depend on factors such as re- who are excluded are those who are willing to accept
sponse to prior treatment, adherence, and intervention only one or none of the available options. Otherwise,
modality. These studies are efficient in developing and the study can accommodate all volunteers willing to be
A typical efficacy trial of one of the medications might its own right. Too often, treatment nonadherence and
last for 36 months, and the outcome might be the change dropout are considered nuisances rather than key out-
in the average daily quantity of alcohol consumed. comes, although they are among the preconditions
Surely, the long-term perspective of the former study that determine the magnitude and scope of the treatment
leads to outcomes that are more meaningful and clinically effect.
significant for patients and clinicians, although short- Deciding what outcomes to measure is a central issue
term studies to establish treatment efficacy and tolerabil- when studying co-occurring disorders. Mental health and
ity are needed first. One implication of this is that efficacy substance-abuse outcomes are relatively independent
studies should include short-term (proximal) outcomes dimensions, and they are largely independent of out-
that predict long-term (distal) outcomes rather than comes in other domains, such as employment, quality
solely using short-term outcomes that are supposed sur- of life, general health, and residential status.41,42 Having
rogates of the long-term outcomes but may not actually used a multidimensional assessment, the challenge then is
predict them. to decide whether to treat the outcome within each do-
Recovery from both mental illness and substance main separately and hierarchically (ie, primary vs second-
abuse is a longitudinal process, not a short-term out- ary outcomes) or to combine outcomes into a single
Mediators are links in a causal chain of outcomes, some form of dual-diagnosis treatment; the challenge
whereby levels of an intervening (proximal) outcome is to bring it under algorithmic control via explicit pro-
causally influence more distal outcomes. Mediators are tocols so that interventions are tested against meaningful
often process variables, such as services received or standards.
knowledge gained, which occur after enrollment in an in- Even if usual care can be standardized prior to the start
tervention but before outcomes are assessed. For exam- of a research project, it may be difficult to keep it stan-
ple, the effect of naltrexone on alcohol abuse may depend dardized over time. For example, randomization within
on adherence with the medication regimen. In this case, clinics to either usual care or a new intervention can lead
treatment adherence mediates the effect of naltrexone on to numerous threats to inferential validity. It is difficult to
alcohol use. A range of statistical methods has been pro- control drift, diffusion, and compensatory efforts when
posed to detect mediator effects, and they differ in the control group (ie, usual care) clinicians become aware
trade-off between Type I errors and statistical power.50 of experimental interventions. In addition, various efforts
to rescue clients, such as transferring them to supervised
Measuring Substance Use housing or changing medications, may also, of necessity,
undermine experimental control. Quasi-experimental re-
of acknowledgment of a mental illness or denial and min- ple contacts from the patients social world such as
imization regarding substance-use disorder, either of landlords, friends, coworkers, and family) work well for
which can lead to disinterest in treatment, in general, longitudinal studies,57 but they do not ensure compliance
and medications, in particular, in the early stages of treat- with regular and frequent research appointments to mon-
ment. As patients learn more about their disorders and itor medication treatment. Instead, frequent reminders,
become interested in managing them, they are often transportation, child-care support, and community out-
more willing to try medications as a management tool. reach may be needed in order to enhance patients ability
Consequently, studies of medications for people with to participate regularly in this type of study. A further
dual disorders are more feasible if they recruit people complication is that patients in later stages of treatment
who are already in either the active treatment or the re- who are more stable are often reluctant to participate in
lapse prevention stages of recovery. Restricting the trea- medication studies. Clinicians may view their stability as
ted population in this manner limits the generalizability tenuous, and there may be reluctance on their part to
of the study, but paradoxically, it may enhance its exter- stress the patient by involvement in research or by chang-
nal validity, by allocating treatment in the research as it ing medication. Studies of add-on medications, such as
would be allocated in routine practice. naltrexone for alcohol-use disorders, may be easier to im-
stages of recovery and are more representative of the recovery suggests that there are distinct subgroups within
population. the larger population of patients with co-occurring dis-
One problem with conducting studies in routine set- orders, which are characterized in part by their rela-
tings is that usual care varies widely from setting to set- tionships and responses to treatment over time.60 One
ting. As discussed above, there is a pressing need to group rejects community-based treatments and has uni-
standardize usual care in order to clarify the effects of formly poor outcomes over at least 10 years. A second
new medication and psychosocial interventions. In addi- group engages in treatment rapidly, enters substance-
tion to clarifying usual care, critical features of routine abuse recovery rapidly, and has a stable course over 10
settings can be expected, based on empirical studies, to years. The majority of patients occupy 2 groups that
influence outcomes. For example, the local criminal jus- have intermediate courses, one with a fluctuating course
tice system,58 the types and amounts of housing programs over 10 years, and another with a course of slow but
that exist,11 and the amounts of family contact and sup- steady improvements over 10 years. If these subgroups
port that patients have59 may moderate treatment effects. can be validated across samples and characterized by
In addition, medications for mental and physical health common individual differences and clinical features, it
may interact with treatment for substance abuse and would make sense to classify patients early in treatment
pharmacological, cognitive behavioral, social network- An earlier version of this article was presented at the
based, self-help, and other strategies could be tested. National Institute on Alcohol Abuse and Alcoholism/
When patients are securely in remission (eg, at least National Institute on Drug Abuse/National Institute
6 months without a relapse of either disorder), they are of Mental Health workshop on Methodology of
in the relapse prevention stage. A variety of relapse pre- Conducting Pharmacologic Clinical Trials in Patients
vention interventions have been developed for mental ill- with Alcohol/Drug Dependence and Psychiatric Comor-
nesses and for substance-use disorders, but none has been bidity, Bethesda, MD, in February 2006.
tested specifically with dual-diagnosis patients.63 Because
patients at this stage of recovery are participating in treat-
ment reliably, are relatively stable, and yet relapse is com-
mon, this would be an ideal group for RCTs to examine References
interventions for successful relapse prevention. 1. New Freedom Commission on Mental Health. Achieving the
Promise: Transforming Mental Health Care in America. Rock-
Summary ville, Md. DHHS Publication No. SMA-033831; 2003.
2. Drake RE, Brunette MF. Complications of severe mental ill-
663
G. J. McHugo et al.
15. Wells KB. Treatment research at the crossroads: the scientific depression (STAR*D) study. Psychiatr Clin North Am.
interface of clinical trials and effectiveness research. Am J 2003;26:457494.
Psychiatry. 1999;156:510. 35. Schneider LS, Tariot PN, Lydetosos CG, et al. National In-
16. Carroll KM, Rounsaville BJ. Bridging the gap: a hybrid stitute of Mental Health Clinical Antipsychotic Trails of In-
model to link efficacy and effectiveness research in substance tervention Effectiveness (CATIE): Alzheimer Disease trial
abuse treatment. Psychiatr Serv. 2003;54:333339. methodology. Am J Geriatric Psychiatry. 2001;9:346360.
17. Rothwell PM. External validity of randomized controlled tri- 36. Breslin FC, Sobell MB, Sobell LC, Cunningham JA, Sdao-
als: to whom do the results of this trial apply? Lancet. Jarvie K, Borsoi D. Problem drinkers: evaluation of
2005;365:8293. a stepped-care approach. J Subst Abuse. 1999;10:217232.
18. Zimmerman M, Chelminski I, Posternak MA. Generalizabil- 37. Brooner RK, Kidorf M. Using behavioral reinforcement
ity of antidepressant efficacy trials: differences between de- to improve methadone treatment participation. Sci Pract
pressed psychiatric outpatients who would and would not Perspect. 2002;1:3848.
qualify for an efficacy trial. Am J Psychiatry. 2005;162: 38. Lavorie PW, Rush AJ, Wisniewski SR, et al. Strengthening
13701372. clinical effectiveness trials: equipoise-stratified randomiza-
19. Rounsaville BJ, Carroll KM, Onken LS. A stage model of be- tion. Biol Psychiatry. 2001;50:792801.
havioral therapies research: getting started and moving on 39. Fenton WS, Blyler CR, Heinssen RK. Determinants of med-
from stage I. Clin Psychol Sci Pract. 2001;8:133142. ication compliance in schizophrenia: empirical and clinical
664
Enhancing Validity
53. Sigmon SC, Stitzer ML. Use of a low-cost incentive interven- 59. Clark RE. Family support and substance use outcomes for
tion to improve counseling attendance among methadone persons with mental illness and substance use disorders.
maintained patients. J Subst Abuse Treat. 2005;29:253258. Schizophr Bull. 2001;27:93101.
54. Kranzler HR, Van Kirk J. Efficacy of naltrexone and acam- 60. Xie H, Drake RE, McHugo GJ. Are there distinctive trajec-
prosate for alcoholism treatment: a meta-analysis. Alcohol tory groups in substance abuse remission over 10 years? An
Clin Exp Res. 2001;25:13351341. application of the group-based modeling approach. Adm
55. Mann K, Lehert P, Morgan MY. The efficacy of acamprosate Policy Ment Health Ment Health Serv Res. 2006;33:423432.
in the maintenance of abstinence in alcohol-dependent indi- 61. Osher FC, Kofoed LL. Treatment of patients with psychiatric
viduals: results of a meta-analysis. Alcohol Clin Exp Res. and psychoactive substance use disorders. Hosp Community
2004;28:5163. Psychiatry. 1989;40:10251030.
56. Drake RE, Xie H, McHugo GJ, Green AI. The effects of clo- 62. Corrigan JD, Bogner J, Lamb-Hart G, Heinemann AW,
zapine on alcohol and drug use disorders among patients with Moore D. Increasing substance abuse treatment compliance
schizophrenia. Schizophr Bull. 2000;26:441449. for persons with traumatic brain injury. Psychol Addict
57. Scott CK. A replicable model for achieving over 90% follow- Behav. 2005;19:131139.
up rates in longitudinal studies of substance abusers. Drug 63. Drake RE, Wallach MA, McGovern MP. Preventing relapses
Alcohol Depend. 2004;74:2136. to substance use disorder among clients with severe mental ill-
58. Swartz MS, Swanson JW, Wagner HR, Burns BJ, Hiday VA. nesses: what do we know and what do we do next? Psychiatr
665