There is a substantial body of literature addressing each of these questions (why do people quit therapy and what predicts positive outcomes); unfortunately there are no easy answers. In part, this is because the literature has looked at these questions from a range of angles, including client characteristics (age, race, gender, motivation, education level, diagnosis), therapist characteristics (demographics, years of experience, therapeutic orientation [cognitive vs psychodynamic]), therapeutic dynamic characteristics (client-therapist demographic match, therapeutic alliance), and treatment characteristics (whether treatment is manualized, group vs individual therapy, type of therapy). I’ve provided brief answers to both below, but you’ll likely need to do additional literature searches if you have a question about a specific group of patients or therapy.
Why do people quit therapy?
Swift & Greenberg (2012), conducted a meta-analysis of 699 studies and found a nearly 20% dropout rate for therapy. They found that younger clients, clients who were attending university clinics, those with personality or eating disorders, and those seen by providers-in-training had higher dropout rates. Some of these factors may be correlated; younger clients may be more likely to attend university clinics, and those clinics often use graduate student therapists-in-training to provide services. The study also found dropout rates were higher in studies where the therapist rated/decided if a client had dropped out, which may reflect the therapists’ perspective. Additionally, therapy that was not time-limited and not manualized had higher dropout rates. Whether it was group vs individual therapy, the orientation of therapy, provider race or gender, and client race or employment status had no effect. Results were mixed for other client demographic variables such as gender, education and marital status.
Another meta-analysis of 11 studies found that dropout was higher when therapeutic alliance, the relationship between the client and therapist, was weaker (Sharf, Primavera, & Diener, 2010). However, the authors noted that this was moderated by the client’s level of education and the length of treatment. Clients with more education were less likely to dropout. The longer the treatment the higher the dropout rate, though, as in the Swift & Greenberg (2012), it may also reflect the perception of the person reporting dropout. Clients may have felt the therapy was completed, while therapists may have felt the client “dropped out.”
Finally, a recent meta-analyses examined the impact of accommodating client preferences on treatment outcomes and dropout rates (Swift, Callahan, Ivanovic, & Kominiak, 2013). There is extensive literature exploring the importance of client preferences related to type of therapist (sex, race), type of treatment (therapy and meds, therapeutic orientation), or style (collaborative, prescriptive) on therapeutic outcomes. Swift and colleagues (2013) found that accommodating client preferences resulted in better outcomes and decreased dropout; these effects did not vary based on client demographic characteristics (age, sex, marital status, etc.). They also noted that clients who were not accommodated were especially likely to dropout of short term therapy.
What predicts therapeutic outcome?
Several recent studies examined the role that therapeutic alliance had on outcomes. A Swedish study of 646 patients found that alliance impacted symptom level in the next therapy session in a feedback loop, such that stronger alliance predicted stronger change in symptoms, which in turn strengthened alliance (Falkenstrom, Granstrom, & Holmqvist, 2013). However, the authors noted that the effect was stronger for some clients (personality disorder patients), than others, indicating that client factors also impact outcomes. Therapy type (supportive, CBT, psychodynamic) did not impact symptom reduction.
Another study found only a weak indication that therapeutic relationship impacts outcomes in complex treatment for psychosis (Priebe, Richardson, Cooney, Adedeji, & McCabe, 2011); these results may reflect a similar finding to Falkenstrom and colleagues (2013), in that diagnosis impacts the role of alliance in outcomes. McEvoy, Burgess, & Nathan (2014) found that another client factor, level of interpersonal problems, predicted dropout and poorer outcomes, regardless of alliance. They also found that type of treatment impacted this relationship, as those patients in CBT who had better alliance were more likely to stay in treatment, but that neither alliance nor personal problems predicted outcomes in those clients. A 2012 meta analysis questioned previous findings that found alliance predicted outcomes (Fluckiger, Del Re, Wampold, Symonds, & Horvath); they found small effects of alliance and outcomes early in therapy, but not for longer-term therapy and did not find sustained relationships with any other commonly noted predictors, such as type of therapy. In contrast, a recent study (Gullo, Lo Coco, & Gelso, 2012) found that a genuine and real (i.e. accurate and not distorted by past conflict or bias) therapeutic relationship predicted successful outcomes for those clients who stayed with treatment, but not those clients who terminated early.
In terms of other client characteristics that can impact outcomes, one study of 60 Caucasian and Asian American clients found that those with more somatic symptoms and who used avoidant coping had poorer psychosocial functioning and psychological discomfort at end of treatment (Kim, Zane, & Blozis, 2012). They also found that those who preferred Non-English language therapy had worse outcomes and that this was mediated by avoidant coping. The authors proposed that Asian American patients who preferred non-English language therapy may have used coping styles that were culturally relevant to them, but which fit poorly with the style of therapy they received. The expression of psychological symptoms and illness is culturally defined, and should be kept in mind when reading meta-analyses that may reflect the dominant culture.
Another meta-analysis of 23 studies found a small to medium effect of homework compliance (completing therapeutic assignments given by the therapist) on client outcomes, which was present regardless of the client’s primary symptom presentation (Mausbach, Moore, Roesch, Cardenas, & Patterson, 2010). This may reflect the importance of client buy-in and effort in therapy, though it should be noted the effect size varied according to certain methodological factors.
Other analyses have examined the importance of therapist characteristics or client-therapist match on outcomes. A meta-analysis of 64 studies found therapist sex did not impact client outcomes or dropout (Bowman, Scogin, Floyd, & Mckendree-Smith, 2001). As noted in my answer to the dropout question above, Swift and colleagues (2013) noted a range of domains in which clients can prefer to be matched in regards to therapy. A 2011 (Cabral & Smith) meta-analyses of 52 studies found almost no effect for race-ethnicity matching and outcomes when looking across all studies. However, the authors did note that African Americans had mildly better outcomes when matched with African American therapists, a trend which did not hold for the other racial and ethnic groups in the study (Caucasians, Hispanics, and Asian Americans). A second meta-analyses of 7 studies found similar results, with a negligible effect size for ethnic match across all samples, though matching had a greater effect for ethnic minority samples than Caucasians in terms of preventing dropout and increasing the number of sessions attended (Maramba & Hall, 2002). Again, these findings call attention to the importance of multicultural conceptualization when identifying treatments and approaches for clients; this diversity is also an excellent example as to why your questions have no simple answers.
Note: None of the studies I found, focusing on recent, meta-analytic literature, addressed whether the clients found the therapy “helpful” (as indicated in your question) as a primary research question.
Bowman, D., Scogin, F., Floyd, M., & Mckendree-Smith, N. (2001). Psychotherapy length of stay and outcome: A meta-analysis of the effect of therapist sex. Psychotherapy, 38(2), 142-148
Cabral, R. R., & Smith, T. B. (2011). Racial/Ethnic matching of clients and therapists in mental health services: A meta-analytic review of preferences, perceptions, and outcomes. Journal of Counseling Psychology, 58(4), 537-554. DOI: 10.1037/a0025266
Falkenstrom, F., Granstorm, F., & Holmqvist, R. (2013). Therapeutic alliance predicts symptomatic improvement session by session. Journal of Counseling Psychology, 60(3), 317-328. DOI: 10.1037/a0032258
Fluckiger, C., Del Re, A. C., Wampold, B. E., Symonds, D., Horvath, A. O. (2012). How central is the alliance in psychotherapy? A multilevel longitudinal meta-analysis. Journal of Counseling Psychology, 59(1), 10-17. DOI: 10.1037/a0025749
Gullo, S., Lo Coco, G., & Gelso, C. (2012). Early and later predictors of outcome in brief therapy: The role of real relationship. Journal of Clinical Psychology, 68, 614-619. DOI: 10.1002/jclp.21860
Kim, J. E., Zane, N W., & Blozis, S. A. (2012). Client predictors of short-term psychotherapy outcomes among Asian and White American outpatients. Journal of Clinical Psychology, 68, 1287-1302. DOI: 10.1002/jclp.21905
Maramba, G. G., & Hall, G. C. N. (2002). Meta-analyses of ethnic match as a predictor of dropout, utilization, and level of functioning. Cultural Diversity and Ethnic Minority Psychology, 8(3), 290-297. DOI: 10.1037//1099-9809.8.3.290.
Mausbach, B. T., Moore, R., Roesch, S., Cardenas, V., & Patterson, T. L. (2010). The relationship between homework compliance and therapy outcomes: An updated meta-analysis. Cognitive Therapy Research, 34, 429-438. DOI: 10.1007/s10608-010-9297-z
McEvoy, P. M., Burgess, M. M., & Nathan, P. (2014). The relationship between interpersonal problems, therapeutic alliance, and outcomes following group and individual cognitive behavior therapy. Journal of Affective Disorders, 157(20), 25-32. DOI: 10.1016/j.jad.2013.12.038.
Priebe, S., Richardson, M., Cooney, M., Adedeji, O., & McCabe, R. (2011). Does the therapeutic relationship predict outcomes of psychiatric treatment in patients with psychosis? A systematic review. Psychosomatics, 80(2), 70-77. DOI: 10.1159/000320976
Sharf, J., Primavera, L. H., & Diener, M. J. (2010). Dropout and therapeutic alliance: A meta-analysis of adult individual psychotherapy. Psychotherapy Theory, Research, Practice, Training, 47(4), 637-645. Doi: 10.1037/a0021175
Swift, J. K., Callahan, J. L., Ivanovic, M., Kominiak, N. (2013). Further examination of the psychotherapy preference effect: A meta-regression analysis. Journal of Psychotherapy Integration, 23(2), 134-145. DOI: 10.1037/a0031423
Swift, J. K., & Greenberg, R. P. (2012). Premature discontinuation in adult psychotherapy: A meta-analysis. Journal of Consulting and Clinical Psychology, 80(4), 547-559. doi: 10.1037/a0028226