Relationship-undermining statements by psychotherapists with clients who present with marital or couple problems.

This study examined the prevalence of relationship undermining statements by psychotherapists as reported by clients in individual therapy who presented with relationship problems, and whether these statements were associated with worse outcomes for client relationships. Participants (n = 101) reported on recollections of whether their therapist had suggested that their partner would never change, had a diagnosable personality/mental health disorder, had negative motives, that the relationship was doomed from the start or beyond repair now, or that divorce/breakup was their best option. Findings showed high prevalence of these undermining statements and associations with poorer relationship outcomes and shorter duration of therapy. We discuss potential explanations for this phenomenon and offer implications for the training of therapists who treat individual clients with relationship problems.

INTRODUCTION

Concerns about marital and other intimate relationship issues are a common presenting problem in individual psychotherapy (Farrell et al., 1987; Greenspan & Kulish, 1985; Nakash et al., 2018; Pajer, 2018; Rosen et al., 2013). However, there is a lack of training for individual therapists on how to address these relationship issues (Gurman & Burton, 2014). Over the years, some couple therapists have expressed concerns about the potential damage to couple relationships when individual therapists accept their (individual) client's perspective against the partner or suggest a diagnosis for a spouse they have never met (Doherty, 2002; Fox, 1968; Hurvitz, 1967). Statements like these could potentially undermine the client's relationship. However, little is known about the actual practices of individual therapists when clients ask for help with relationship problems. We present the findings from a pilot study of individuals who presented with relationship problems in individual therapy. We were specifically interested in the prevalence of what we consider to be relationship‐undermining statements by therapists, as recalled by their clients, and whether these statements are associated with therapy outcomes.

LITERATURE REVIEW

Mental health treatment is widely accessed in the United States. According to a Harris poll, 27% of American adults engaged in talk therapy or medical treatment for mental health issues during a two‐year period (Relias Media, 2004). Although precise figures on the number of people presenting to individual therapy with couple relationship problems are not available, some studies indirectly bear on the question. Greenspan and Kulish (1985) reported that “family or marital discord” was the most frequently reported reason for seeking treatment in their sample of clients in long‐term psychotherapy. Gill (1985), in a study that compared clinical psychotherapy records from 1938 to a sample taken from 1978, reported that in 1938, 28% of the records indicated the presenting problem of “interpersonal problems” while in 1978 the number had risen to 55%.

More recent studies indicate that “relationship” problems account for 20.4% and 23.6% of the presenting problems reported by outpatient and inpatient veterans (respectively) with Post Traumatic Stress Disorder (Rosen et al., 2013). Cheng (2000) investigated how clients (n = 269) listed the most troubling problems for which they were seeking help in individual psychotherapy. She determined that “relationships” was the third most troubling problem reported. In a survey of both therapists and clients on common presenting problems, Nakash et al. (2018) found that marital problems were mentioned specifically by 19.6% of clients and 21.4% of therapists as the main problem. In an Australian study of 273 clients who presented alone for relationship counseling, Lohan et al. (2021) reported that the most common presenting problems were personal issues related to their relationship (53%), relationship improvement (40%), and the relationship's future (29%). Despite ambiguity in some of these studies about client's marital status or whether the presenting problems were in the couple relationship, it is reasonable to surmise that couple problems are an important issue that many clients bring to individual psychotherapy.

Concerns about individual therapy for couple problems

This study can be viewed as part of a relatively recent tradition of research on the negative effects of therapy, in this case effects on family members and significant others (Lilienfeld, 2007). In an article on individual therapy for relationship problems, Gurman and Burton (2014) argued that individual therapists sometimes accept their client's view or presentation of the relationship problems and can ultimately “side” against the non‐attending partner and may even offer a diagnosis for the partner. This can prevent arriving at a fuller, more systemically oriented view of the presenting problem and may become a source of conflict in the relationship (“My therapist agrees with how I see things.”).

After noting that some clients may misperceive their therapist's empathy for agreement that the spouse is almost entirely responsible for the marital problems, Gurman and Burton go on to write:

Unfortunately, it also commonplace to hear of therapists, insufficiently trained to think systemically, who actually express such a view. Sometimes this takes the form of diagnosing, either confidently or speculatively, the partner whom the therapist has never met. Sometimes the therapist asks “leading questions” about why the client stays in the relationship despite his suffering, etc., implying that this must reflect pathology on his (sic) part, as anyone else would see that the relationship is not worth saving. Such untoward therapist positions may implicitly encourage divorce or at least fail to foster change within the couple relationship and possibly worsen the couple’s problems by reinforcing the attending partner’s polarized position. (p. 476–477)

A backdrop for Gurman and Burton was their observation that the overwhelming proportion of therapists in the United States receive their training in individual psychotherapy, with relatively little attention to the insights and techniques that have emerged in the field of couples therapy. Even among marriage and family therapists, the most common form of practice is individual therapy (Northey, 2002).

In addition to an ongoing concern expressed by couple therapists about harm to the relationships of clients who present individually, there is a small literature on the topic by psychotherapists who mainly treat individuals. In the early 1980s, Phillips addressed ethical issues involved when marital relationships deteriorate following the kind of individual psychotherapy that does not engage the perspective of the spouse (Phillips, 1983). Following that paper, a handful of studies examined whether marital and other romantic relationships deteriorate following individual therapy. The results have been mixed, with indications that individual therapy helps many relationships while leading to deterioration in others (Brody & Farber, 1989; Hunsley & Lee, 1995; Pomerantz & Seely, 2000; Roberts, 1996). The largest study of outcomes of individual therapy for relationship functioning (Lohan et al., 2021) found no improvement in relationship satisfaction following individual therapy conducted by therapists who were trained in couples therapy, even though there was improvement in individual psychological distress.

A clue to understanding this differential impact of individual therapy on relationships comes from a small case study by Lefebvre and Hunsley (1994), where the outcomes were different depending on the client's and spouse's understanding of the problems being treated. When both partners agreed that the client's distress was from biologically based mental illness, the relationship improved after individual treatment. When the client perceived the problems as stemming largely from the spouse's behavior, not biology, the relationship deteriorated. Thus, it makes sense to focus on the potential risks of individual therapy when relationship problems are a key focus, as opposed to general psychotherapy for a range of mental health issues, only some of which the client may associate with a relationship problem.

One thoughtful article on the dangers to be avoided when providing individual therapy for relationship problems was offered by psychiatrist Schwartz (2005) in an essay in the Harvard Review of Psychiatry. He proposed that psychotherapists are pulled toward negative assessments of their clients’ relationships for two main reasons. First, therapists tend to focus on the current relationship crisis or stalemate, which may lead them to not account for the prior vitality or the future possibilities for the relationship. Our assessments, he asserts, are influenced by the first stories we hear from our clients. Second, the attentive and empathic skills offered by the therapist can be used by clients to cast their relationship partner in an unfairly negative light. Therapists in turn, because of our empathic engagement, are “powerfully drawn to our patient's point of view in their assessment of others” (Schwartz, 2005, p. 276).

Nowhere in the relevant literature is there data on client perceptions of how their therapist talked during sessions about the client's relationship or about their relationship partner. There is a clear need to document the extent to which therapists may undermine clients’ relationships by what they say during the everyday practice of individual psychotherapy. We hope that this pilot study will spur further research in this area.

THE PRESENT STUDY

We attempt to address two questions with this study:

What is the prevalence of relationship‐undermining statements by individual therapists when treating married clients who present with relationship problems? We had no specific predictions about prevalence beyond the expectation that the frequency of relationship‐undermining actions by a therapist would be sufficient enough to be of concern. As this is a first attempt to study this phenomenon, we refrained from speculating in advance on the degree of the prevalence beyond the notion that would not be rare or infrequent.

How does the extent of relationship‐undermining statements by individual therapists relate to the outcomes of therapy and the number of sessions attended? We anticipated that more areas of undermining would be associated with poorer relational outcomes. That is, the more domains in which the relationship the therapist expressed undermining statements about, the worse the outcomes. We also included number of sessions as a proxy for early termination of therapy if the client feels less hope for the relationship.

We addressed these questions through a survey of a national, online, convenience sample of 101 individuals who brought couple/relationship problems to individual therapy. Therapist undermining was assessed through asking clients to report what their therapists said about the relationship or the relationship partner. Two important caveats: the participants reported on therapy that happened at any point in the past, and we cannot assess the degree to which clients may distort what they hear from their therapists. Our aim here is not offer definitive evidence but to open the research door to an important but neglected topic: What happens in individual psychotherapy that may undermine the relationships of clients?

METHODS

Sample and recruitment method

Participants volunteered for the study by responding to a Human Intelligence Task (HIT) on the Amazon Mechanical Turk (MTurk) platform. They were initially screened through self‐report questions. To be included in the study, participants must have attended or are currently receiving individual therapy for problems associated with their marriage or long‐term committed relationship. After responding to screening questions, potential participants were excluded if they: (1) had seen a counselor with their spouse or partner, or (2) had seen a religious leader and not a mental health professional. (We were interested in information from those who had accessed a licensed mental health professional rather than a member of the clergy.) Although not representative, MTurk provided access to a convenience sample that was diverse in terms of age, race, gender, and sexual orientation (see sample characteristics below).

MTurk, which has been widely used in social science studies since its creation (Petri et al., 2020), has been shown to provide diverse samples with valid responses from survey research when compared to in‐person sampling methods (Casler et al., 2013). We followed recommendations in the literature on recruiting optimal samples and on obtaining high quality responses (Paolacci & Chandler, 2014). Specifically, we used screening questions to establish a focused sample and a set of novel questions developed for the current study. We minimized the time required to complete the survey, and prohibited participants from taking the screener questions more than once. Institutional Review Board (IRB) approval was obtained before the survey was posted on the MTurk platform. Data gathering took place during 2018–2019.

The resulting sample of 101 participants had an average age of 34 (SD = 8.36), with a range from 23 to 72. They were 60% male and 40% female. Most were currently married (64% or in a long‐term committed relationship, 21%). The primary racial groups were White (50%) and Asian (43%). As for educational levels, 71% had some college or an associate's degree, 20% had high school or less, 7% had a college degree, and 3% had a post‐graduate degree. In terms of sexual orientation, 75% described themselves as heterosexual and 25% as gay, lesbian, or bisexual.

In order to describe the sample in terms of the problems they brought to their therapy, the survey included a checklist of possible problems they may have discussed with their counselor. The checklist of 20 couple problems has been validated in prior research on couples in the divorce process (de Graaf & Kalmijn, 2006; Doherty et al., 2016). The most frequently problems reported (in order of frequency) were: growing apart, not able to talk together, sexual problems, the spouse's personal habits/problems, how the spouse handles money, and infidelity. In general, the problems these individuals brought to individual therapy appear similar to those documented by Doss et al. (2004) as problems couples report bringing to couples therapy.

Measures

Therapist relationship‐undermining statements

We developed six questions to identify areas of therapist relationship‐undermining statements. Because this is the first study on the topic, we created new measures based on the literature cited above as well as our own clinical experience with clients reporting on what previous therapists had said, as well as anecdotal reports by therapists in case consultations and supervision sessions in a variety of clinical settings. We asked participants to rate “how true” the presence of the therapist comments were in their experience of the therapy. The specific instructions were: Thinking about your most recent counseling experience for relationship problems, please indicate how true the following statements are (if you are in counseling now, answer based on your current experience):

Response options were not at all true, somewhat true, moderately true, mostly true, and completely true. (There was also a don't know/can't remember category.) Specifically, we asked how true it was that their counselor had:

Suggested that your spouse could not change without having met that person.

Suggested a personality or mental health diagnosis of your spouse without having done an individual assessment.

Suggested that the marriage is probably beyond repair.

Indicated that divorce is your best or most realistic option.

Suggested negative motives (like being selfish or deliberately trying to be hurtful) behind your spouse's actions.

Suggested that your relationship was a bad match from the beginning.

In order to avoid a potential demand characteristic in which the participant would assume that we are only interested in negative therapist behavior, we preceded these items about undermining statements with a series of questions about how true it was that the therapist showed empathy, was active with questions and observations, showed hope for the relationship, and demonstrated other positive skilled behaviors.

Because our research question was not about the frequency of therapist‐undermining statements but instead the number of areas in which the therapists made undermining statements, we dichotomized each of the six items so that an item score represented the absence of the respective undermining statement in each of the six types of statement (0, not at all true) or the presence of the comment (1, somewhat true to completely true). We then summed the six dichotomized items to produce a total therapist‐undermining score that could range from 0 to 6, with higher scores representing more domains of undermining statements. The summed scale demonstrated good internal consistency (α = 0.91), indicating that therapists making one type of statement were likely to also engage in others, as reported by the client.

Client outcomes

Because the participants could have done therapy at any time in the past, standard outcome measures asking about current relationship functioning post therapy were not appropriate for this study. Instead, we developed three basic self‐report items asking about their perception of how the therapy may have influenced their relationship at the end of their therapy work.

How was your marriage/relationship after the counseling? Responses could range from 1 (much worse) to 5 (much better).

What happened with the relationship between you and your partner after you stopped attending counseling sessions (if you are currently in counseling, answer based on the state of your relationship now)?” Response categories were: 1 = Divorced/broke up, 2 = Separated without a plan to get back together, 3 = Stayed together.

How helpful was the counseling in resolving the relationship problems that you discussed with your counselor? Responses could range from 1 (Not at all Helpful) to 5 (Very Helpful).

For the number of therapy sessions with the counselor, participants were asked “About how many counseling sessions have you attended?” Respondents were prompted to enter a numerical answer based on their most recent therapy experience. The average was 7.7 sessions (SD = 9.06), with a range from 1 to 55 sessions.

Analysis plan

For the first set of analyses, we tested the continuous variables for violations of linearity, normality, and homoscedasticity. We used the descriptive statistics to determine the frequency of therapist relationship‐undermining statements. To test whether therapist relationship‐undermining statements predicted therapy outcomes, we used multivariate regression for the three continuous variables and multinomial regression (UCLA, 2020) for the categorical variable of relationship status after therapy (stayed together versus separated/divorced). We used SPSS v27 for all analyses.

Assumptions were met for the relationship quality post‐treatment and self‐reported helpfulness of therapy variables but not for the session frequency variable, which demonstrated skewness and kurtosis outside of the desired range. Although the highest reported session frequency (55 sessions) was much higher than the rest of the responses and resulted in skewness and kurtosis outside of the desired range, it was deemed to be a reasonable response to the question. Since multivariate regression can be robust to undesirable skewness and kurtosis, the outlier case was included in the final models.

RESULTS

We present the findings here in three sections: the individual undermining items, the scale score of relationship‐undermining items, and outcomes. Table 1 contains a correlation matrix of the study variables, and Table 2 contains descriptive statistics on the outcome variables.

TABLE 1

Correlation matrix of study variables

Variables	1	2	3	4	5
1 Helpful therapist behaviors	−
2 Therapist‐undermining statements	−.12	−
3 Session frequency	−.21*	−.32**	−
4 Relationship quality	.28**	−.28**	.11	−
5 Overall helpfulness	.46**	−.04	−.01	.52**	−

* <.05; ** <.01

TABLE 2

Descriptive statistics for outcome variables (N = 101)

	Range	Mean	SD
Overall helpfulness	1–5	3.93	1.11
Relationship quality post‐treatment	1–5	3.74	1.14
Number of sessions	1–55	7.70	9.06

Prevalence of the relationship‐undermining variables

All six of the undermining areas were reported by at least half of the clients as somewhat true to completely true. Specifically, the therapist suggesting that the spouse was unlikely to change was reported by 64% of participants; suggesting that the spouse had a personality disorder or other mental health diagnosis (without an individual assessment) was reported by 55% of the participants; suggesting that the marriage was beyond repair by 57%; indicating that divorce/separation was the best option by 55%; suggesting negative motives in the spouse by 62%; and suggesting that the marriage was a bad match from the beginning by 59% of clients.

We then decided to relax the cut‐off for the presence of the undermining behaviors by including “somewhat true” responses as indicating the absence of that kind of statement, leaving moderately to completely true responses as indicating the presence of that kind of statement. We did this to determine whether prevalence remained high under a less stringent measurement condition that allows for some memory of a therapist statement to count as zero presence of that domain. With that assessment rubric, the prevalence rate of relationship‐undermining actions was lower but remained substantial: for the same sequence of items as above, the figures were 46%, 43%, 43%, 39%, 39%, and 40%.

Undermining scale descriptive findings

Table 3 shows descriptive findings for the therapist relationship‐undermining scale, which was the summed scores for the six items (based on the more stringent cut off scores). The mean was 3.27 (SD = 3.29), indicating that the average client reported about 3 of the 6 areas of undermining statements from their therapist. Seventy‐eight percent (78%) of clients reported undermining statements in at least one area, with 44% reporting statements in five or six of the areas.

TABLE 3

Reported number of different relationship‐undermining statements (N = 101)

Number of reported undermining statements	Number of clients (%)
0	22 (21.8%)
1	10 (9.9%)
2	13 (12.9%)
3	2 (2%)
4	10 (9.9%)
5	16 (15.8%)
6	28 (27.7%)
Mean = 3.27; SD = 2.39; Range = 0–6

Outcome findings

Results of the multivariate linear regressions showed that the frequency of reported relationship‐undermining statements was significantly associated with relationship quality after the therapy (B = −0.14, SE = 0.04, p < 0.01). As predicted, the more areas of therapist‐undermining statements, the lower the relationship quality.

The results for the multinomial regression for relationship status after therapy indicated that therapist‐undermining statements significantly predicted a greater likelihood of separating or divorcing (B = 0.177, SE = 0.057, Wald = 9.774, p = 0.002). Each one‐unit increase in undermining therapist scores (specifically, the addition of one area of undermining) was associated with a 18% increase in the odds of separating/divorcing versus staying together.

Relationship undermining was not associated with self‐reported treatment helpfulness (B = −0.02, SE = 0.05, p = .71).

The prediction for number of sessions of treatment was also supported by the regression findings (B = −1.20, SE = 0.34, p < .01). That is, the more areas of therapist‐initiated undermining, the fewer therapy sessions. The average number of sessions for the lower range of undermining scores (0–2) was 12 sessions, while the average for the higher range (4–6) was 5 sessions.

DISCUSSION

This study explored the prevalence and impact of practices by individual therapists that have the potential to harm the marriage and couple relationships of clients who present with problems in those relationships. The findings provide evidence that, from the perspective of clients, therapist relationship‐undermining statements are highly prevalent. That is, many individual therapists, when presented with a client's marital or relationship problem, tend to portray an absent spouse in highly unfavorable ways: as not likely to change, as having a personality disorder or mental health diagnosis, and as having negative motives. In terms of the relationship, they are also likely to suggest that it was fatally flawed from the beginning and is beyond repair now, and say that divorce is the most reasonable option in front of the client. In terms of client outcomes, our findings indicate that these statements are associated with poorer outcomes for the relationship, greater likelihood of separation and divorce, and fewer sessions. There was no association found for perception of the overall helpfulness of the therapy.

Realizing that these are provocative findings, we want to emphasize some important caveats before interpreting them. First, all measures were client reports, not based on therapist reports or observations of sessions; as mentioned, clients may have differing interpretations of what their therapists said or meant. Second, we did not ask how often the therapist was expressing agreement with the client versus offering an uninvited opinion; there might be different effects accordingly. Third, the sample consisted of clients who had done therapy over a period of years, which means that their memories might be less reliable than a survey of clients recently completing therapy. (We did not ask about which years participants were in therapy.)

Fourth, the findings are correlational associations as opposed to causal ones; we cannot conclude that the therapy outcomes were due to therapist practices. For example, clients whose therapists made more negative statements may have been already more likely to have poorer outcomes. Fifth, separation is not always an undesirable outcome of couple therapy; however, it is concerning when separation is associated with skeptical therapist statements about the spouse or the relationship (with the possible exception of abuse situations). Sixth, fewer sessions are not necessarily an indicator of inadequate therapy; however, it is concerning when shorter duration is associated with skeptical therapist statements. Seventh, the sample was not obtained in a way to represent the population of individuals seeking psychotherapy for relationship problems. Eighth, we did not ask whether the therapist had seen the client's partner in any of the sessions. Finally, the measures were developed for this study and lack prior evidence for their reliability and validity.

With these caveats and limitations in mind, we believe that this study suggests that a number of practices by therapists are overdue for questioning because they are widespread and have the potential to undermine marriages and other committed couple relationships. This conclusion is consistent with prior literature that raised concerns about how individual therapists sometimes harm their clients’ relationships (Gurman & Burton, 2014).

On a more speculative level, we also suspect that many experienced couples therapists, after years of hearing couple therapy clients’ reports about what their individual therapists have said, will not be surprised by these findings. In our work with couples on the brink of divorce (Doherty & Harris, 2016), we routinely ask clients about prior therapy experiences, and the empirical findings of this study are consistent with what we hear from clients, even for those who were satisfied with the services of their prior therapists. In that regard, some clients are pleased that their therapist indicated that their spouse probably has a personality disorder, which might make these clients inclined to report that the therapy was “helpful” even if the relationship deteriorated. This could account for the lack of association between therapist‐undermining statements and the helpfulness outcome variable.

An important clarification of the findings is that we are not suggesting that the therapist statements were necessarily inaccurate, just that (a) the therapist cannot know for sure that their comments are accurate (e.g., diagnosing at a distance, interpreting negative motives, knowing that the relationship was fatally flawed from the outset), and (b) the statements create a risk for harming the relationship. Nor are we proposing a rule that a therapist must never, in any circumstances, utter a statement like the ones on our list. Raising the possibility of an alcohol problem after a spouse's third DWI would be an example of an appropriate comment. Our main point is that the clustering of this kind of statement—for example, combining the suggestion of a personality disorder and a fatally flawed origin of the relationship—may lead to negative outcomes for the client's relationship. In other words, we suspect that the clients pick up on the therapist's overall tone and attitude toward the relationship partner and the relationship itself—how negative, neutral, or hopeful the therapist is.

A key question is why therapists would engage in practices that could undermine their clients’ relationships? Certainly, there is no reason to believe that therapists are anti‐relationship or wish to do harm. Knowing that the findings of this study do not address this question, we nevertheless propose several possible explanations:

There is little explicit training of therapists in how to work with individual clients on relationship issues (Gurman & Burton, 2014). There are many established techniques for helping individual clients with depressive thoughts or anxious feelings, but little guidance for helping them with a troubled marriage. To be clear, we are not saying that the training of individual therapists does not involve supervision on their work with relationship problems, but that there is little in the way of explicit skills and published literature, with the exception of Interpersonal Psychotherapy (IPT; Stuart & Robertson, 2012). This may result in therapists making up what they do on their own, possibly reflecting their own biases about relationships.

Therapists may feel obligated to show understanding and empathy toward their individual client, which can lead them to fully accept the client's portrayal of their partner and of their own behavior in the relationship (Schwartz, 2005).

When therapists become frustrated with the lack of client progress on relationship problems, they may resort to suggesting that the therapy is stalling because the spouse has a personality disorder or that the relationship itself is likely to fail. In supervision and case consultations, we have observed therapist frustration with a case being paired with these and other undermining statements reported in this study.

When a client is considering divorce and seems stuck in a distressed marriage, it may be tempting for the therapist to recommend divorce (directly or indirectly) in order to help the client get beyond the impasse of indecision. This could be a reflection of therapists’ challenges in sitting with a client's ambivalence and not responding out of their own anxiety to offer a solution.

Some therapists may lack the systemic perspective necessary for imagining how their individual client is also contributing to negative interaction patterns with their partner, and how the partner may feel and respond in these situations. The spouse becomes a toxic figure or at best a clueless character in the client's life drama.

Ethical and clinical considerations

The universal ethical mandate for clinicians is to, “first, do no harm.” This research suggests that therapists not infrequently say things to their clients that have the potential to harm their relationships. As mentioned, we assume positive intentions, but many therapists may take an individualistic focus that does not sensitize them to the effects of their interventions on clients’ families, intimate partners, social networks, and communities. But because the human self is inherently relational, non‐attending partners and family members are always ethical stakeholders in individual psychotherapy (Doherty, 2022). This is one reason that suggesting a diagnosis for a relationship partner without having met and assessed that person crosses an ethical line in terms of the potential for harm and for the misuse of professional opinion.

A possible implication for ethical preparation of therapists is for educators to teach about the potential for relationship undermining in therapy. We recommend including case studies and clinical vignettes that illustrate common undermining statements and help students learn how to ask questions and deliver therapeutic statements in relationally oriented and systemically attuned ways. Above all, the mental health community can do a better job of understanding the relational context of clients’ lives and the ways our interventions have relational impact.

Finally, because clients may misconstrue what their therapists are saying about their relationship partner or the relationship itself, therapists might adopt the practice of checking in with individual clients about what they “heard” us say during the session. For example, the questions “What did you hear me say about your relationship?” might be a good way to wrap up a therapeutic conversation and offer the opportunity to clarify any client misinterpretations. It could also be a good check for the therapist who may be unwittingly engaging in disparaging the client's partner or suggesting that the relationship is somehow doomed.

Implications for future research

We still know very little about the phenomenon explored in this study. An important next step would be to obtain therapists’ perspectives on their routine practices with individual clients who come with relationship problems. What do therapists report saying when they discuss relationship or partner‐specific problems with an individual client, and what is their rationale for what they say? Do they see how what they say about the spouse or the relationship could be interpreted differently than intended? In other words, we need a better understanding of what therapists say, and what they intend, in addition to what clients hear. Ideally this would be accompanied by transcripts or recordings of sessions so that we would assess the interplay between what the therapist says and the client says, which could reveal how much the therapist initiates undermining observations versus passively more affirming clients’ observations about their partner and the relationship.

Another area to explore would be positive comments by therapists about the client's partner and the relationship. There is much to learn about the proportions of positive and negative comments and about the possibility that some therapies may encourage a preponderance of negative comments that might offset otherwise effective individual therapy for relationship problems. Finally, studies in this area should enroll clients as the embark on therapy and examine the nature and seriousness of the relational problems as well as their goals for treatment. This kind of study would also allow the use of standardized outcome measures.

Just as psychotherapy has the power to heal, it has the power to harm (Lilienfeld, 2007). We hope that this study, notwithstanding its pilot status, opens the door to further research on iatrogenic effects of individual therapy on marital and couple relationships, and to some soul searching among psychotherapists about what may be a common blind spot in our work.