Ilardi and Craighead (1994) made the observation that 60–70% of the symptom improvements during CBT occurred in the first four weeks. The authors concluded that cognitive modification techniques are not extensively applied in these weeks and thus that Beck`s cognitive hypothesis of CBT cannot explain this early improvement. Tang and DeRubeis (1999b) criticized Ilardi and Craighead`s method (Ilardi and Craighead, 1994; 1999) assuming homogeneity among patient`s time course which is inconsistent with the empirical data. They concluded that symptom change time course data do not contradict Beck’s cognitive hypothesis.

Furthermore, Rachmann (1999) commented on the occurrence of early changes as significant findings in psychotherapy research from a theoretical point of view and raised the attention to the need for recognizing and explaining delayed-onset therapeutic changes in particular. He referred to the commonly stated argument, that CBT sessions commonly initiate a process that then might unfold after and between the sessions, implying that the evidence of change emerges after a certain delay. Further, he claimed for studying early changes in CBT for anxiety disorders rather than depression or other disorders as this could lead to better and more enlightening results according to his insights. Following the key study by Ilardi and Craighead (1994) he commented that early changes are as surprising as they are significant. He critically commented that the rapid early responses raised the uncomfortable possibility that a large part of the response to treatment is nonspecific, and the responses bear some resemblance to the rapid early response that is characteristically found in placebo reactions. The main difference is that placebo reactions tend to be transient whereas the changes observed after CBT are reasonably enduring which may lead to assumption that rapid early changes reflect non-specific effects other than placebo. The author stated a very important and so far rather underrepresented issue: early responses to treatment do occur, their significance is undeniable, but far more interesting are the delays in response to treatment and its relation to outcome to explain this phenomenon. To conclude, Rachmann (1999) raised suspicion about the well accepted assumption of a vast amount of therapists that all or virtually all therapeutic changes in CBT occur during the sessions, stating that they do not (e.g. consolidation of information).

Haas and colleagues (2002) investigated the question of whether early responders to psychotherapy maintain their treatment gains on the data of 147 college student clients. They defined early response based on a score representing the difference between client-obtained symptom distress session scores and expected scores derived from a large study on typical treatment responses. The authors basically concluded from the results that early positive response was associated with fewer psychological symptoms at therapy termination and follow-up and maintenance of therapy gains. A comparison to pharmacological research has been drawn within this study as well: a drug is deemed therapeutic when it induces an effect that differs from a placebo, whilst time here often distinguishes between an active drug agent from a placebo effect, implying that the long-term effects and the course of drug action are important indicators of its efficacy. The authors raised attention to the fact that an early response could mean a premature change in the symptomatology due to client characteristics in contrast to drug effects, and this “placebo” response would often be linked to poorer long-term outcomes and particularly relapse during follow-up. Several other studies investigated this possibility and support empirically the contention that a premature drug response foretells poor long-term therapeutic outcomes.

The authors concluded that (especially in CBT for patients with depression) there is a considerable variability with respect to the impact of time of response to treatment. Mainly, early responders showed differences in process variables and showed better long-term outcomes (e.g. early responders appeared to proceed from one problem to another while delayed responders continued to revisit the same therapy topic repeatedly across sessions; which suggested a better negotiation and completion of therapeutic tasks that were prerequisite for therapy-induced change for early responders). The authors pointed to the problematic issue of identifying the active mechanisms linking early response to long-term outcome: it may be indicative of the ability to negotiate the steps in therapy which are essential for maintaining long-term gains. On the one hand side, there is common agreement, that (in CBT) the most critical change tasks occur in the first few sessions and a favorable early response towards this may simply indicate the accomplishment to this task. On the other hand, early response may simply indicate a client’s readiness to change (or resourcefulness) and their response is maybe more towards common factors rather than specific CBT interventions. This idea would be supported by the statement that most of the i.e. substantial improvement occurs before, not after, most of the specific therapeutic operations have been initiated (Rachmann, 1999). Additionally there is a lack of agreements how to define and measure the concept of early response. The authors mainly addressed two questions in this study: a clarification of the predictive role of early and positive response to therapy for outcome at termination and follow-up and if those clients being identified as showing early response maintain their gains until follow-up. Based on the concept of clinical significance by Jacobson and Truax (1991), patients’ symptom course has been categorized as: recovered, improved, deteriorated, casualties and unchanged. The authors conclude from the results that the response rates were predictive of long-term outcomes while controlling for initial severity. Moreover, in the groups of recovered and improved patients, more rapid responses had been found, thus rapid responses seem to be associated with clinically meaningful change at termination and follow-up and the participants who showed rapid response rates maintained their treatment gains following the termination. The chosen method in this study and given the fact that there is no agreed-upon definition in psychotherapy outcome literature for early response, the method of combining sessionby-session scores in relation to a set of expected responses based on normative data in the present study, may provide a unique and potentially useful definition of early response that is able to take into account typical response patterns of clients with similar initial disturbances. Moreover this definition also allows controlling for regression towards the mean, which is highly relevant in this context: clients with very high initial symptom scores are expected to show greater reductions compared to those with lower levels.

The authors draw the conclusion that clients who were categorized as early responders in contrast to all other study participants responded better to therapy and maintained these gains up to two years. This implies that whatever form of “placebo effect” can be excluded from possible explanations, as then one would expect much higher relapse rates once the treatment ended. Instead of this the authors found early response to be related to remission of symptoms at termination but unrelated to relapse. The authors critically interpret the results in line with theoretical implications and raise doubt about the therapeutic techniques as key actor: they summarize that the results suggest that many clients showed a substantial response to treatment so early that it is hard to attribute their change in symptomatology to the active ingredients of most treatments as posited by most therapeutical systems. In that respect the findings are consistent with Ilardi and Craighead’s (1999) assertions that change in symptomatology may have less to do with the presumed active components of therapy and more to do with nonspecific factors that are present across therapies (e.g. client-therapist-relationship). If those common factors account for a portion of early response to treatment, then this might strengthen the raised argument that common factors are central in a patient’s healing process. The authors further considered the question about the necessity (for many clients) of specific therapeutic techniques that are often introduced later in the therapeutic encounter. Haas and colleagues (2002) suggested further research on early response to treatment and therapeutic alliance, while doubting as well that therapeutic relationship needs time to develop and would probably not be able to account fully for the phenomenon of early response. A list of possible other factors contributing to an explanation of early treatment response was given by the authors: patient characteristics (especially traits like being more receptive to psychotherapeutic influence, organization, coherence, ability to maintain focus, higher levels of motivation or being more “ready” for therapy), better working alliance between therapist and client and therapist experience.