A widely accepted credo among medical and mental health professionals, attributed to the Greek physician and “father of medicine” Hippocrates, is Primum non nocere (“First, do no harm”). Yet, despite the signal importance of this credo, the field of psychotherapy has displayed relatively little interest in the question of potentially harmful psychological treatments. For example, in 2006, the American Psychological Association (APA) Presidential Task Force on Evidence-Based Practice barely mentioned the problem of harmful treatments.
There is no question that psychotherapy is helpful on balance. Meta-analyses consistently demonstrate that a broad spectrum of psychotherapies, including behavior, cognitive-behavioral, interpersonal, and insight-oriented therapies, exert positive effects on a variety of psychological problems. These problems include mood disorders, anxiety disorders, insomnia, and bulimia nervosa. Nevertheless, this positive assessment of the state of the psychotherapy literature must be balanced against one sobering fact: A nontrivial number of clients become worse following psychotherapy.
Estimates of psychotherapy deterioration—that is, worsening of symptoms following psychotherapy— vary across controlled studies, but typically range from 3% to 10%. In the substance abuse literature, estimates of deterioration have often been higher, averaging 10% to 15%. These percentages have led some researchers, such as Hans Strupp, Richard Stuart, and Scott Lilienfeld, to suggest that psychotherapy can in some cases be iatrogenic, that is, capable of causing harm.
Moreover, comprehensive meta-analyses (quantitative reviews) of the psychotherapy outcome literature have been reasonably consistent in revealing that control groups—groups that do not receive treatment— show higher end-state functioning than groups that receive treatment in a nontrivial minority of studies. A classic meta-analysis of psychotherapy outcome studies by Mary Lee Smith and Gene Glass found this to be true in 9% of the scientific investigations of the efficacy of psychotherapy. Later meta-analyses have reported similar or even higher percentages of negative outcomes. Even more troubling, meta-analyses of treatments for adolescent behavioral problems have suggested that as many as 29% of studies yield negative outcomes.
Troubling as they are, reported percentages of psychotherapy deterioration and negative effect sizes from meta-analyses are difficult to interpret. As the logical error of post hoc, ergo propter hoc (“after this, therefore because of this”) reminds us, deterioration following psychotherapy is not the same as deterioration resulting from psychotherapy. At least some of the deterioration reported in previous studies could reflect worsening due to extratherapy variables, such as stressful life events. Moreover, certain individuals who deteriorate following psychotherapy might have been even worse off had they not received psychotherapy. Therefore, estimates of psychotherapy deterioration, although suggestive of therapy-induced harm in some cases, cannot be interpreted directly as evidence for the negative effects of therapy.
The same holds for negative effect sizes from meta-analyses. Such effect sizes, especially if close to zero, could reflect random sampling error around a mean effect of zero. As a consequence, treatments that yield small negative effects could be merely inert rather than harmful per se. It should be noted, however, that inert treatments can cause harm indirectly by leading clients to forgo more efficacious therapies. Economists refer to this side effect as an opportunity cost, meaning that ineffective treatments may lead consumers to forfeit opportunities for successful treatment.
Randomized controlled trials are the best means of ascertaining whether deterioration following psychotherapy is attributable to the effects of psychotherapy. In randomized controlled trials, investigators randomly assign some clients to receive the treatment, and others to either receive no treatment or an alternative treatment. If treated groups show worse functioning than untreated control groups following psychotherapy in replicated randomized controlled trials, one can be reasonably confident that a psychological treatment is harmful.
Yet even here, the absence of mean differences between treatments can mask iatrogenic effects. For example, certain treatments may markedly increase the variance in outcomes but leave the average outcome largely or entirely unchanged. In these instances, psychotherapies have produced improvement in some clients but worsening in others. As a consequence, consumers of psychotherapy outcome research must be certain to attend not merely to differences in mean scores following treatment but to potential differences in variance.
Harm as a Multifaceted Phenomenon
Making matters more complex, harm following psychotherapy is a multifaceted phenomenon. First, harm can be manifested in myriad forms, including a worsening of the signs and symptoms of a disorder, emergence of new signs and symptoms, excessive dependency on therapists, and reluctance to seek future treatment. Traditionally, psychotherapy outcome researchers would regard only the first two of these forms of harm as “deterioration.” In some cases, as in certain coercive restraint therapies (described below), psychological treatments may even produce serious physical harm.
Second, harm can occur in certain symptom domains (e.g., heightened anxiety) but not in others (e.g., depressed mood). This point underscores the need to assess harmful effects using multiple indicators of psychological functioning.
Third, certain treatments may be harmful primarily for the relatives of clients rather than clients themselves. As this entry will discuss below, facilitated communication for autism and recovered memory techniques appear to have led to numerous uncorroborated allegations of sexual abuse against the family members of clients.
Fourth, certain treatments that are efficacious in the long term may engender symptom worsening in the short term. For example, some effective marital therapies appear to result in time-limited increases in marital distress early in treatment, probably because they force partners to confront long-avoided emotional conflicts.
Potentially Harmful Therapies
Attention has only recently been accorded to the development of criteria for ascertaining whether a form of therapy is potentially harmful. In recent work, Scott Lilienfeld and his colleagues have generated a provisional list of potentially harmful therapies. Readers are advised to bear in mind that this list is preliminary. It will almost certainly be revised in coming years as additional data become available.
To qualify as a potentially harmful therapy, a treatment should be demonstrated to (a) be harmful across replicated studies (ideally randomized controlled trials), or (b) result in marked increases in adverse events shortly after the introduction of treatment, events that otherwise rarely occur (e.g., recovered memories of trauma), as reported across multiple case studies. Inclusion of a treatment on the list of potentially harmful therapies does not mean that it is harmful for all or even most clients. Instead, a treatment’s being designated as potentially harmful should alert therapists to the distinct possibility that the treatment may pose harm in certain cases.
Scott Lilienfeld perused the scientific literature and compiled a list of potentially harmful therapies in 2007. This literature provides reasonably clear-cut evidence that the following six treatments—taken from Lilienfeld’s more comprehensive compilation— qualify as potentially harmful therapies.
Critical incident Stress Debriefing
Critical Incident Stress Debriefing, commonly called crisis debriefing, is a widely used treatment designed to ward off symptoms of posttraumatic stress disorder in individuals exposed to trauma. Several thousand counselors administered crisis debriefing or variants of this treatment to the witnesses of the September 11, 2001, terrorist attacks in New York City. Crisis debriefing is typically conducted in a single group session and lasts several hours, although it is occasionally spread out across several sessions. In traditional crisis debriefing, therapists strongly encourage group members to discuss and “process” their negative emotions associated with the trauma. They outline the symptoms of posttraumatic stress disorder that group members are likely to experience following trauma, and discourage group members from discontinuing participation once the session is underway. With respect to the lattermost point, some crisis debriefing counselors even ask group members to retrieve members who have left in mid-session.
A meta-analysis of randomized controlled trials for crisis debriefing yielded a small negative effect size (d = -.11) for posttraumatic stress disorder symptoms. That figure indicates that individuals exposed to trauma who receive crisis debriefing actually end up with slightly more symptoms of posttraumatic stress disorder than individuals who receive no treatment. Moreover, the results of three randomized controlled trials indicate that crisis debriefing can be markedly harmful for at least some trauma-exposed individuals, perhaps because this procedure impedes natural recovery processes. Incidentally, at least some evidence from randomized controlled trials suggests that grief therapies, which also encourage the expression of intense negative affect following painful experiences, can be iatrogenic for individuals undergoing relatively normal grief reactions.
Scared Straight Programs
Scared Straight programs became popular in the 1970s and were featured in an influential 1978 documentary on Rahway State Prison in New Jersey. The programs aim to frighten at-risk teenagers away from a life of crime by showing them the terrifying realities of prison life. Many of these programs introduce troubled adolescents to convicts in prison.
A meta-analysis of seven randomized controlled trials and quasi-experimental studies of Scared Straight programs revealed that they increased the chances of reoffending by 60% to 70%. The mechanisms underlying these negative effects are unknown, although some authors have suggested that Scared Straight programs may contribute to further alienation among a subgroup of already alienated teenagers. The research evidence for other “get tough” interventions with troubled adolescents has been similarly inconclusive or negative. For example, the research support for popular “boot camp” programs for adolescent criminals is mixed. Some studies show significant positive effects, but others show significant negative effects. The factors that account for these varied outcomes are unknown. As a consequence, there is at present no way for practitioners to know whether boot camp interventions will be beneficial or harmful.
Facilitated communication is based on a radical theory that infantile autism and related developmental disabilities are primarily movement disorders, not mental disorders. According to facilitated communication proponents, individuals with autism suffer from developmental apraxia (i.e., a disorder that interferes with the child’s ability to perform well-coordinated movements, including speaking). This condition supposedly accounts for their poor or even (in severe cases) absent verbal abilities, as well as other movement difficulties. Therefore, advocates of facilitated communication argue that individuals with autism and similar disorders can purportedly generate communications using a computer keyboard or letter pad with the assistance of a trained facilitator who offers a combination of guidance and resistance to the individuals’ hands.
Controlled studies in which individuals with autism and their facilitator were shown different stimuli and had to type out which stimulus they saw provide overwhelming evidence for an ideomotor effect—a phenomenon in which the facilitator unconsciously makes subtle motions that influence the communications of autistic individuals. Specifically, in these studies the word typed was in essentially all cases the word seen by the facilitator, not the word seen by the autistic individual. Indeed, there is no compelling evidence that any facilitated communications are generated by autistic individuals themselves. Moreover, facilitated communication has been associated with at least five dozen allegations of sexual abuse against the parents and relatives of autistic children. Most of these allegations have never been corroborated by objective evidence.
Coercive Restraint Therapies
Coercive restraint therapies are a subset of attachment therapies, which are based on the unsupported notion that certain psychological difficulties, including aggression and oppositionality, stem from aberrant early attachment experiences. These experiences include a premature or abnormally difficult birth. Coercive restraint therapies, like other attachment therapies, are intended to ameliorate these adverse experiences. They include holding therapy, in which therapists physically restrain children or adolescents in an effort to release suppressed rage, and rebirthing therapy, in which therapists attempt to engage the client in reenacting the trauma of birth. In some variants of rebirthing therapy, practitioners wrap the child in blankets to create an analogue birth canal, and they simulate the birth process by squeezing on the child repeatedly.
There are no controlled studies supporting the efficacy of coercive restraint therapies or other attachment therapies. Moreover, several children have been injured or suffocated to death during coercive restraint therapies—including 10-year-old Candace Newmaker, who died in Colorado in 2000 during a rebirthing session.
Recovered Memory Techniques
Several surveys from the 1990s suggest that a sizable proportion of therapists (perhaps 25%) make regular use of suggestive techniques to unearth purportedly recovered memories of early trauma, particularly child sexual abuse. These techniques include repeated therapist prompting of memories, hypnosis (including hypnotic age regression), guided imagery, and “body work,” which encourages clients to access “bodily memories” of early abuse.
The question of whether suggestive techniques can ever uncover accurate memories of abuse remains scientifically controversial. Nevertheless, laboratory research leaves little doubt that these techniques can lead to false memories in a substantial percentage of participants. Moreover, findings that recovered memory techniques can engender memories of alien abductions and even past-life child abuse offer existence proofs that at least some of the memories produced by these techniques are false.
There is also strong reason to suspect that false memories can lead to significant harmful effects in both clients and their family members. For example, data from recovered memory legal claims in Washington State indicate that suicidal ideation increased nearly sevenfold and that psychiatric hospitalizations increased nearly fivefold over the course of recovered memory treatment. Although these findings do not provide definitive evidence of causality, they raise serious concerns about the potential negative effects of such treatment. Moreover, many families have been torn apart by uncorroborated accusations of sexual abuse by children against parents.
Dissociative Identity Disorder-Oriented Treatment
Dissociative identity disorder-oriented (DID-oriented) treatment is designed to elicit alter personalities (i.e., alternate personalities or alters) in individuals suspected of having DID (formerly known as multiple personality disorder). The core premise of this treatment is that patients with DID harbor hidden identities that must be brought forth for improvement to occur. DID-oriented treatment methods include a variety of suggestive treatment techniques, including contacting supposed alters through hypnosis, introducing alters to each other, and mapping out the relations among alters. One prominent DID-oriented therapist advocates the use of a “bulletin board” to allow alters to post messages to each other. Another advocates “inner board meetings” as a method of permitting alters to communicate.
Multiple lines of evidence suggest that DID-oriented techniques more often create alters than uncover them, probably by treating poorly integrated aspects of patients’ personalities as though they were independent identities. For example, only about 20% of patients with dissociative identity disorder display clear-cut alters prior to treatment. Alters emerge in the remaining 80% of patients only following DID-oriented therapy. Moreover, the number of alters tends to increase over the course of DID-oriented therapy. This finding is worrisome given that the number of alters in dissociative identity disorder patients is associated with a significantly longer time to “fusion,” that is, the integration of alters into a single personality.
Although research evidence indicates that psychotherapy is generally helpful, a growing body of data suggests that certain psychological treatments can be harmful for certain clients and their families. The fields of counseling and clinical psychology are awakening to the need to devote additional resources to the identification of potentially harmful therapies. Efforts to (a) understand the psychological mechanisms underlying the iatrogenic effects of potentially harmful therapies, (b) find better means of detecting and preventing these effects, and (c) educate psychotherapists and mental health consumers regarding the hazards of harmful psychological treatments should therefore be accorded greater attention.
- Boisvert, C. (2003). Negative effects in psychotherapy: Research findings and clinical implications. Directions in Clinical and Counseling Psychology, 15(4), 37—48.
- Bootzin, R. R., & Bailey, E. T. (2005). Understanding placebo, nocebo, and iatrogenic treatment effects. Journal of Clinical Psychology, 61, 871-880.
- Hunsley, J., & Lee, C. M. (1995). The marital effects of individually oriented psychotherapy: Is there evidence for the deterioration hypothesis? Clinical Psychology Review, 15, 1-22.
- Lilienfeld, S. O. (2007). Psychological treatments that cause harm. Perspectives on Psychological Science, 2, 53-70.
- Lilienfeld, S. O., Lynn, S. J., & Lohr, J. M. (Eds.). (2003). Science and pseudoscience in clinical psychology. New York: Guilford Press.
- Lynn, S. J., Lock, T., Loftus, E. F., Krackow, E., & Lilienfeld, S. O. (2003). The remembrance of things past: Problematic memory recovery techniques in psychotherapy. In S. O. Lilienfeld, S. J. Lynn, & J. M. Lohr (Eds.), Science and pseudoscience in clinical psychology (pp. 205-239). New York: Guilford Press.
- Mash, E. J., & Hunsley, J. (1993). Assessment considerations in the identification of failing psychotherapy: Bringing the negatives out of the darkroom. Psychological Assessment, 5, 292-301.
- McCord, J. (2003). Cures that harm: Unanticipated outcomes of crime prevention programs. Annals, AAPSS, 587, 16-30.
- Mohr, D. C. (1995). Negative outcome in psychotherapy: A critical review. Clinical Psychology: Science and Practice, 2, 1-27.
- Moos, R. (2005). Iatrogenic effects of psychosocial interventions for substance use disorders: Prevalence, predictors, prevention. Addiction, 100, 595-604.
- Rhule, D. M. (2005). Take care to do no harm: Harmful interventions for youth problem behavior. Professional Psychology: Research and Practice, 36, 618-625.
- Smith, M. L., & Glass, G. V. (1977). Meta-analysis of psychotherapy outcome studies. American Psychologist, 32, 752-760.
- Strupp, H. H., Hadley, S. W., & Gomez-Schwartz, B. (1978). Psychotherapy for better or worse: The problem of negative effects. New York: Wiley.
- Stuart, R. B. (1970). Trick or treatment: How and when psychotherapy fails. Champaign, IL: Research Press.
- Werch, C. E., & Owen, D. (2002). Iatrogenic effects of alcohol and drug prevention programs. Journal of Studies on Alcohol, 63, 581-590.