This is a repository copy of Early response to psychological therapy as a predictor of

depression and anxiety treatment outcomes: A systematic review and meta-analysis.

White Rose Research Online URL for this paper:

https://eprints.whiterose.ac.uk/146850/

Version: Accepted Version

Article:
Beard, J. and Delgadillo, J. orcid.org/0000-0001-5349-230X (2019) Early response to
psychological therapy as a predictor of depression and anxiety treatment outcomes: A
systematic review and meta-analysis. Depression and Anxiety, 36 (9). pp. 866-878. ISSN
1091-4269

https://doi.org/10.1002/da.22931

This is the peer reviewed version of the following article: Beard, JI, Delgadillo, J. Early
response to psychological therapy as a predictor of depression and anxiety treatment
outcomes: A systematic review and meta‐analysis. Depress Anxiety. 2019, which has been
published in final form at https://doi.org/10.1002/da.22931. This article may be used for
non-commercial purposes in accordance with Wiley Terms and Conditions for Use of Self-
Archived Versions.

Reuse
Items deposited in White Rose Research Online are protected by copyright, with all rights reserved unless
indicated otherwise. They may be downloaded and/or printed for private study, or other acts as permitted by
national copyright laws. The publisher or other rights holders may allow further reproduction and re-use of
the full text version. This is indicated by the licence information on the White Rose Research Online record
for the item.

Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by
emailing eprints@whiterose.ac.uk including the URL of the record and the reason for the withdrawal request.

eprints@whiterose.ac.uk
https://eprints.whiterose.ac.uk/
 Author’s Manuscript

Note: This is a pre-print peer reviewed article, accepted for publication on 25.05.2019. Please
do not copy or share without the author’s permission.
Citation: Beard, J.I.L., & Delgadillo, J. (in press). Early response to psychological therapy as
a predictor of depression and anxiety treatment outcomes: A systematic review and meta-
analysis. Depression and Anxiety. doi: 10.1002/da.22931

Early response to psychological therapy as a predictor of depression and anxiety

treatment outcomes: A systematic review and meta-analysis

Jessica I.L. Beard1 and Jaime Delgadillo2*

1
Department of Psychology, University of Sheffield, United Kingdom

2
Clinical Psychology Unit, Department of Psychology, University of Sheffield, United

Kingdom

* Correspondence: jaime.delgadillo@nhs.net

1
ABSTRACT

Background: Previous studies indicate that early symptomatic improvement, typically

observed during the first 4 weeks of psychological therapy, is associated with positive

treatment outcomes for a range of mental health problems. However, the replicability,

statistical significance and magnitude of this association remains unclear.

Aim: The present study reviewed the literature on early response to psychological

interventions for adults with depression and anxiety symptoms.

Methods: A systematic review and random effects meta-analysis was conducted, including

studies found in Medline, PsychINFO, SCOPUS, Web of Science, and through reference lists

and reverse citations.

Results: Twenty-five eligible studies including 11091 patients measured early response and

examined associations with post-treatment outcomes. It was possible to extract and/or

calculate effect size data from 15 studies to conduct a meta-analysis. A large pooled effect

size (g = 0.87 [95 % CI: 0.63, 1.10] p < .0001) indicated that early responders had

significantly better post-treatment outcomes compared to cases without early response, and

this effect was larger in anxiety (g = 1.37) compared to depression (g = 0.76) measures. Most

studies were of good quality and there was no evidence of publication bias. The main

limitations concerned insufficient statistical reporting in some studies, which meant that the

meta-analysis only included 60% of reviewed studies, and it was not possible to examine

effect sizes according to different outcome questionnaires.

Conclusions: There is robust and replicated evidence that early response to therapy is a

reliable prognostic indicator for depression and anxiety treatment outcomes.

Key words: early response; psychological therapy; depression; anxiety

2
1. INTRODUCTION
Symptomatic improvements that occur during the initial sessions of psychological therapy,

and which are of a statistically and/or clinically significant magnitude, have been termed

early response. As early as the 1980s, psychotherapy researchers have observed that the bulk

of symptomatic improvement occurs within the first month of therapy, with 60% of total

improvement in cognitive behavioural therapy for depression occurring by week four (Rush,

Kovacs, Beck, Weissenburger, & Hollon, 1981). Since then, numerous other studies have

reported early response patterns across different mental health problems and psychological

therapies (Bell, Waller, Shafran, & Delgadillo, 2017; Bradford et al., 2011; Delgadillo et al.,

2014; Doyle, Grange, Loeb, Doyle & Crosby, 2009; Gois et al., 2014; Hunnicutt-Ferguson,

Hoxha & Gollan, 2012; Rubel et al., 2015).

Some studies have found that early symptomatic response is associated with better

post-treatment outcomes (e.g., Crits-Christoph et al., 2001; Grilo, Masheb & Wilson, 2006;

Lutz et al., 2017; Stauffer et al., 2011; van Calker et al., 2009). However, the magnitude of

this association varies considerably across studies and clinical samples. For example, in

samples of patients accessing guided self-help for depression, Tadić et al. (2010) reported an

odds ratio of 1.33 whereas Delgadillo et al. (2014) reported an odds ratio of 12.60.

Furthermore, others have found mixed results. For example, Arnow et al. (2007) found no

significant associations between early response and treatment dropout. Başoğlu et al. (1994)

found that early response predicted some post-treatment outcomes (panic attacks) but not

others (anticipatory avoidance). In addition, there is still no agreed-upon time scale of when

early response occurs by in psychological therapy (Haas, Hill, Lambert & Morrell, 2002;

Rubel et al., 2015). For example, some studies have shown that this can happen by week two

or three, whilst others have examined this by week eight or later (Arnow et al., 2007; Bell et

al. 2017; Rubel et al., 2015; Tadić et al., 2010). Also, different psychological therapies have

3
varying numbers of sessions within the first month of treatment. For example, Tang and

DeRubeis (1999a) state that there are not always four sessions within the first month of

cognitive behavioural therapy (CBT); instead, in treatment lasting around 12 to 20 weeks, 40

to 60% of sessions occur within the first month. Studies assessing early response are also

heterogeneous, since they use different symptom measures and baseline levels of distress

differ among study samples.

Therefore, whilst it appears to be that people with early symptomatic improvements

tend to have a better prognosis, the replicability, statistical significance and magnitude of the

association between early response and post-treatment outcomes remains unclear. The current

systematic review aimed to synthesise what is known about the association between early

response and psychological treatment outcomes in adult patients seeking treatment for

depression and anxiety.

2. METHODS

2.1. Protocol and Registration

The review protocol was prospectively registered in the PROSPERO database

(http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42018089123).

2.2. Search Strategy and Study Selection

Table 1 specifies the inclusion and exclusion criteria that guided this review. Four databases

were searched in order to search for titles, abstracts and key words (Appendix A) on February

13, 2018: Medline, PsychINFO, SCOPUS and Web of Science. The search was limited to

peer reviewed articles or book chapters published in English. No date restrictions were

applied. A full list of excluded studies and reasons for exclusion are listed in Appendix B.

4
 The search also included the terms “sudden gains”, which can be defined as sudden

and statistically significant symptomatic improvements that occur between two consecutive

sessions during therapy (Tang & DeRubeis, 1999b). Although sudden gains are conceptually

different to early response (which can be gradual and over more than two sessions), sudden

gains that occur early in treatment can overlap with and contribute to an early response

pattern. Therefore, studies where the median sudden gain session occurred within the first

four weeks of therapy were also included; consistent with evidence that most sudden gains

occur within the first month (e.g., Tang & DeRubeis, 1999a). Titles and abstracts were

screened by the first author, followed by a full-text review. Of the studies eligible for the

review, reverse-citations and reference list searches were conducted by hand to identify

further eligible studies. A PRISMA diagram summarising the study selection process is

presented in Figure 1.

[Table 1]

[Figure 1]

2.3. Quality and Risk of Bias Assessment

Two reviewers independently assessed the quality and risk of bias for each study using the

Quality Assessment Tool for Observational Cohort and Cross-sectional Studies and the

Cochrane Handbook for Systematic Reviews of Interventions for Controlled Trials (Higgins,

Altman & Sterne, 2017). Quality ratings were highly consistent and did not require mediation

by a third reviewer. See Appendix C for quality assessment summary tables.

5
2.4. Data Analysis

In addition to a narrative synthesis of all reviewed studies, a random effects meta-analysis

was conducted using the statistical package Meta-Analysis via Shiny (MAVIS; Hamilton,

2011). Studies that examined the predictive value of early response and which reported

sufficient statistical information to calculate standardised effect sizes were included in the

meta-analysis. Hedges g was calculated to adjust for unequal sample size across groups,

comparing post-treatment outcomes between early responders versus cases without early

response. Heterogeneity was examined using the Q and I2 statistics (Higgins et al., 2003).

Potential publication bias was examined using a weight-function model (Vevea & Hedges,

1995) as well as the fail-safe N calculation using Rosenthal’s method (Rosenthal, 1991).

There were several sources of heterogeneity in the design of the studies, including in

the symptom measures used, the time-point at which early response was defined, and the

time-point at which outcomes were assessed (e.g. post-treatment, at 6-months follow-up). We

therefore followed a principled and consistent approach to enable meta-analysis, where we

selected the outcome measure with the largest effect size (in studies with multiple outcomes),

at the measurement point closest to session 4 (in studies with multiple measurements), with

the outcome defined at the end of treatment (or most proximal follow-up assessment). In a

subgroup analysis, we separately meta-analysed samples using measures for depression or

anxiety symptoms. Other sources of heterogeneity such as differences in the types of

treatments and study design were examined by removing systematically different studies

from the meta-analysis and examining the influence on pooled effect sizes and indices of

heterogeneity. The small number of studies (<20) that were included in meta-analysis and the

imbalance in the number of studies per each of the above categories of heterogeneity

precluded the use of categorical moderator analysis, according to conventional guidelines

(Rubio‐Aparicio, Sánchez‐Meca, López‐López, Botella, & Marín‐Martínez, 2017).

6
[Table 2]

3. RESULTS

3.1. Study characteristics

Design and sample characteristics. Twenty-five studies met the inclusion criteria and are

described in Tables 2 and 3, including analyses of controlled trials (n= 16) and observational

studies (n= 9). Many studies focused exclusively on depression (n= 16), whilst only three

focused exclusively on anxiety disorders. Sample sizes ranged from 23 to 5484 participants,

with nearly all of the studies including more females than males.

Measures. A wide range of measures were used to assess early response, as shown in

Table 2. The Beck Depression Inventory (BDI) and the Hamilton rating scale for depression

(HAMD) were most commonly used to assess depression symptoms. A variety of different

measures were used to assess anxiety, including the Beck Anxiety Inventory (BAI), Penn

State Worry Questionnaire, and Panic Disorder Severity Scale-self report. Measures of

psychological distress included the Outcomes Questionnaire (OQ-45), and the Clinical

Outcomes in Routine Evaluation (CORE). Only two studies used general psychological

distress measures rather than disorder-specific measures to assess symptom change (Lutz et

al., 2017; Rubel et al., 2015), whilst two used a combination of both (Lutz, Stulz & Köck,

2009; Schibbye et al., 2014).

Defining early response. Many studies conceptualised early response by quantifying

changes between intake assessments and week four of treatment, as reported by 15 of the 25

studies, while others conceptualised this as far as the 8th week. Sixteen studies used standard

7
and similar criteria to define early response, including reliable change (RC; Jacobson &

Traux, 1991) a specific percentage reduction on the measures used (e.g., >25% reduction of

symptom severity from baseline to the week of early response), or early remission of

symptoms based on established diagnostic cut-off scores. Other studies defined early

response using methods such as hierarchical agglomerative cluster analysis (HAC) or Growth

Mixture Modelling (GMM) to identify different trajectories of change among participants.

The three studies measuring sudden gains used Tang and DeRubeis’ (1999b) definition, and

each study modified the criteria to fit the measures used and the diagnosis of patients. With

regards to the relationship between early response and outcomes, 18 of the 25 studies clearly

stated the methods used to predict outcomes among early responders. These methods

included receiver operating characteristic curves (ROC), logistic regression, general linear

models (GLM), GMM, trajectory analysis, linear modelling of growth curves, cluster

analysis, piecewise growth mixture modelling (PGMM) and HAC.

Interventions. Most of the studies focused solely on psychological interventions (n=

15), where cognitive behavioural therapy (CBT) was the most commonly studied

intervention. Ten studies included samples that accessed either pharmacotherapy or

psychological therapy (both treatments were combined in 3 of these studies).

Of the studies using exclusively psychological therapies, two used internet-based

cognitive-behavioural therapy, one used group telephone therapy, and another pooled data for

patients who accessed individual, group and internet-delivered therapies. Other psychological

therapies included cognitive behavioural analysis system of psychotherapy (CBASP),

supportive-expressive dynamic psychotherapy, interpersonal psychotherapy (IPT),

behavioural activation, metacognitive therapy, short-term psychodynamic supportive

psychotherapy (SPSP), behavioural therapy and self-control therapy. Thirteen studies used a

variety of psychological or pharmacological treatments but not all of them provided

8
information about each type of therapy, and the different therapies were not all analysed

separately.

3.2. Quality Assessment

Overall, most reviewed studies had low risk of bias. More than half (15 = 60%) were rated as

“good” quality studies and a further 8 (32%) were rated as “fair”. Only two were rated as

“poor” quality studies with high risk of bias. The most common sources of bias included the

failure to report dropouts / excluded cases, lack of recruitment information, little or no

description of treatments, and lack of information about randomization procedures and

blinding in clinical trials. A detailed risk of bias assessment for each study can be found in

supplementary Appendix C.

3.3. Early response and treatment outcomes

Pooling available data across studies indicated that approximately half (47.9%) of study

participants were classified as early responders. The majority of these studies found support

for a positive association between early response and post-treatment outcomes as summarised

in Table 3. Some studies additionally investigated the influence of pre-treatment changes that

may have occurred prior to exposure to therapy (e.g., between intake assessments and the

first therapy session). These studies found that early response to therapy remained a

significant predictor after controlling for pre-treatment changes that might denote regression

to the mean (Delgadillo et al., 2014; Lutz et al., 2017). Furthermore, early response was

associated with clinical outcomes across different severities of psychological distress (e.g.,

mild, moderate, severe depression), different forms of psychotherapy / pharmacotherapy and

even in a placebo control group (see: Tadić et al., 2010).

Only one study did not support an association between early response and outcomes,

although they specifically focused on dropouts (Arnow et al., 2007). Two studies found

9
equivocal relationships between early response and positive outcomes. Başoğlu et al. (1994)

found no association between early remission in panic symptoms and improvement in

anticipatory anxiety and avoidance at post-treatment. However, they did find an association

between early improvement in phobia symptoms and a reduction in panic attacks. Van et al.

(2008) compared psychological therapy with a combination of pharmacotherapy and

psychotherapy. In the psychotherapy condition, early non-response significantly predicted

final non-response. However, this was not the case for remission rates. In the combined

therapy condition, both final non-response and remission rates differed significantly for those

who showed early non-response and early response. This may suggest that the medication in

the combined therapy influenced the predictive value of early response. Therefore, the

potential influence of combined treatments was examined using a subgroup meta-analysis

described in the next section.

Some of the reviewed studies did not specifically test early response – outcome

associations, but instead focused on when the largest changes occurred, how many cases

showed early response, or which variables predicted early response (Gildengers et al., 2005;

Heckman et al., 2017; Jordan et al., 2014; Rabin, Kaslow & Rehm, 1984). The key findings

from these studies are summarised in Table 3.

[Figure 2]

3.4. Meta-analysis

A primary meta-analysis was conducted using effect size data from 15 studies (including 18

unique samples and 3956 participants) to examine associations between early response and

post-treatment outcomes. The weighted mean effect size was g = 0.87 [95 % CI: 0.63, 1.10] p

10
< .0001, indicating that early responders had significantly better treatment outcomes

compared to cases without early response (see Figure 2). Cochrane’s Q-test revealed

evidence of significant heterogeneity in the sample (Q(17) = 246.12, p < .0001) and the I2

statistic was 87.73%, indicating a large degree of heterogeneity. The weight-function model

likelihood ratio test did not indicate evidence of publication bias; x2(1) = 3.08, p = .08.

Calculation of the fail-safe N indicated that 3200 studies with non-significant results would

be needed to conclude that the effect of early response is not statistically significant.

A series of subgroup analyses were conducted to examine the potential influence of

methodological features. Removing samples (k = 6) with combined (psychotherapy and

pharmacological) treatments, samples (k = 4) from studies that applied latent clustering of

symptom trajectories, samples (k = 4) from studies that investigated sudden gains, and

samples (k = 6) from studies with “fair” (instead of “good”) quality ratings made little

difference to the pooled effect sizes (g = 0.82 to 0.92) and indices of heterogeneity (I2 =

87.46 to 91.83%) obtained in subsamples. However, removing samples (k = 4) from

observational studies considerably reduced the pooled effect size (g = .74) and indices of

heterogeneity (I2 = 79.69%; Q(13) = 66.72, p < .0001) observed in the subgroup of RCT

samples. Removing two studies (Başoğlu et al., 1994; Steinman et al., 2013) that specifically

examined panic disorder interventions also considerably reduced the pooled effect size (g =

.77) and indices of heterogeneity (I2 = 79.70%; Q(15) = 230.87, p < .0001) observed in the

subgroup of RCT samples. Furthermore, separate meta-analyses in the samples analysing

depression (k = 15, n = 3690) and anxiety (k = 5, n = 2214) measures revealed markedly

different effect sizes (depression g = 0.76 [0.58, 0.94]; anxiety g = 1.37 [0.86, 1.88]) and

indices of heterogeneity (depression I2 = 76.27%; anxiety I2 = 90.00%). Forest plots for these

separate meta-analyses are shown in supplementary Appendix D.

11
4. DISCUSSION

4.1. Summary of evidence

This systematic review provides robust evidence of a positive association between early

response to psychological therapy and post-treatment outcomes. A large and statistically

significant overall effect size (g = 0.87) indicated that early responders tended to have better

treatment outcomes compared to other patients. Translating this effect size into an odds ratio

scale (4.84), would indicate that early responders were at least four times more likely to attain

positive treatment outcomes compared to other cases. This association was replicated across

various forms of therapy including CBT, psychodynamic and interpersonal psychotherapies,

and guided self-help interventions delivered in person, in groups, via telephone and via

internet. Furthermore, there was low risk of bias in most of the reviewed studies, which were

generally of good methodological quality, and we found no evidence of publication bias.

The effect of early response on anxiety outcomes was considerably larger (g = 1.37)

compared to the effect on depression outcomes (g = 0.76). However, the index of

heterogeneity for anxiety studies was much higher (I2 = 90% vs. 76%), suggesting that the

influence of early response may vary across anxiety disorders. In particular, large effect sizes

(g = 1.80 to 1.87) were observed in the studies that examined early response to panic disorder

treatments (Başoğlu et al., 1994; Steinman et al., 2013), and the exclusion of these two

studies reduced the overall heterogeneity of meta-analytic findings. The design of studies was

another relevant source of heterogeneity, as we found that the index of heterogeneity was

somewhat attenuated in the subsample of randomised controlled trials (after excluding

observational studies). This is plausibly explained by the stringent inclusion/exclusion criteria

applied in clinical trials, which typically yields more diagnostically homogeneous samples

compared to naturalistic observational studies.

12
4.2. Methodological Considerations

The studies included in this review varied considerably in the methods used to assess early

response. For example, Rubel et al. (2015) compared three different approaches: growth

mixture modelling (GMM), reliable change (RC) and clinical improvement. They found that

at session three approximately 35% of the participants met criteria for RC, and 16.3% had

achieved clinically significant improvement. On the other hand, when using GMM, only

7.2% showed early positive response by session three. This shows that GMM is a more

conservative approach, identifying much fewer patients as showing early response than the

RC and clinical improvement methods. This was also found when comparing the studies by

Gilboa-Shechtman and Shahar (2006) and Lutz et al. (2009) which re-analysed the same

dataset (Elkin et al., 1989). Although both studies included the same number of participants

from the same dataset, they used different methods to identify early response and to predict

outcomes. Consistent with Rubel et al. (2015), Lutz et al. (2009) identified fewer patients as

showing early response (n=99) when using GMM, compared to Gilboa-Shechtman and

Shahar (2006) who used linear modelling of growth curves (n=122).

Early response was most often defined by week 4, although other cut-offs were used

in different studies. For example, Steidtmann et al. (2013) used weeks four, six and eight as

they are considered to be clinically useful decision points. Similarly, Schibbye et al. (2014)

used week five as it represented the first half of the treatment. The studies by Arnow et al.

(2007), Başoğlu et al (1994) and Tadić et al. (2010) defined early response by week two of

treatment. Alternatively, some studies retrospectively identified a week of early response via

the use of statistical methods that modelled latent trajectories of change (e.g., Heckman et al.,

2017; Lutz et al., 2009; Steinman et al., 2012). In spite of these methodological differences,

subgroup analyses removing latent trajectory profiling studies from meta-analysis did not

considerably change the pooled effect size or indices of heterogeneity. However, Rubel et al.

13
(2015) argue that a minimum of three sessions are required to conduct GMM, suggesting that

there may not have been enough information within the first two weeks of treatment to assess

the predictive value of early response. Therefore, it may be that week two is too early to find

significant differences between those who do and do not show early response, which may

possibly explain the mixed results found by Arnow et al. (2007), Başoğlu et al (1994) and

Tadić et al. (2010). Overall, studies that defined early response by week four or later

consistently found associations with post-treatment outcomes.

In total, only 10 of the 25 (40%) studies clearly stated how they handled missing data

in their analyses. Not appropriately managing dropouts can lead to biased interpretations of

results (Dumville, Torgerson & Hewitt, 2006; Mallinckrodt et al., 2001). Therefore, it may be

that including and examining all participants who dropped out of treatment could potentially

influence the results concerning early response and treatment outcomes. However, reviewed

studies that used an intention-to-treat approach still found large and statistically significant

effects of early response over treatment outcomes.

4.3. Strengths and limitations

To our knowledge, this is the first meta-analytic review to assess the association between

early response and psychological therapy outcomes. Strengths of this review include the

prospective registration of the review protocol and the study selection across multiple

databases, which had no date restriction in order to gather all eligible studies. Reverse and

forward citation searches were also conducted. The review also included a quality assessment

by two independent assessors. However, the review excluded grey literature, studies

published in languages other than English, and authors of eligible studies were not contacted

to request data.

14
 Some limitations should be taken into account when interpreting the findings of this

review. Due to such a wide variety of measures and methods used to assess early response,

directly comparing findings was difficult. Although a series of subgroup analyses were

carried out, there was still a large amount of heterogeneity left unexplained. This is likely to

have been due to the large number of measures used across the studies, with only two studies

in the meta-analysis using the same depression measure. This meant that a subgroup analysis

to assess the effect of specific measures was not possible. We also note that only 60% of

reviewed studies provided sufficient information to compare treatment outcomes between

cases with and without early response. In spite of the limited number of studies that

contributed to meta-analysis, there was no evidence of publication bias and the fail-safe N

(3200) indicated that the meta-analytic results were robust.

Sudden gains that occur early in treatment arguably overlap with the early response

phenomenon. On this basis, we imposed inclusion criteria to ensure that we only analysed

data from studies where the median sudden gain session occurred by week 4. This a priori

exclusion criterion resulted in a stricter exclusion of sudden gains studies and a more liberal

inclusion of early response studies. We argue that this is appropriate, given the focus of the

current review. A more liberal inclusion of sudden gains studies would risk confounding

these constructs, making it difficult to ascertain whether or not early response predicts better

treatment outcomes (over and above the already known effect of sudden gains). We note that

a subgroup analysis excluding sudden gains studies made no difference to the pooled meta-

analysis results, so the inclusion of these studies is unlikely to bias results in any way.

Whilst the inclusion of RCTs and observational studies allowed for the inclusion of a

greater number of studies, there are limitations in both designs. Observational studies have

the advantage of including patients who are encountered in routine practice, thus

strengthening the external validity of results. However, internal validity can be compromised

15
in observational studies, since they often lack rigorous features that are present in RCTs, such

as the use of control groups, randomisation and blinding, adherence to treatment manuals,

and stringent inclusion and exclusion criteria. RCTs included in this review also varied

considerably in their patient selection criteria and how well they handled missing data.

Although the study design accounted for a proportion of heterogeneity (as described above),

it is nevertheless clear that early response is a phenomenon that occurs across a wide variety

of study designs and settings. Another point to consider is that a large number of secondary

analyses were included in the review. These studies should be interpreted with caution as

secondary analyses can lack statistical power and multiple testing can inflate the chances of

type 1 error.

4.4. Implications for research, theory and practice

It is important to consider the findings of this review in the context of previous research.

Although sudden gains are conceptually different from early response, unless they occur by

week four of therapy, they are both important for predicting post-treatment outcomes. A

meta-analysis conducted by Aderka, Nickerson, Bøe and Hofmann (2012) found results

consistent with the current review. They found robust evidence for a positive association

between sudden gains and primary outcomes in depression and anxiety symptoms. However,

whilst the current review focused on primary symptoms of depression and anxiety, Aderka et

al. (2012) also examined secondary symptoms. Compared to the large effect size in primary

symptoms (g=0.62, [0.43, 0.80]), the effect size for secondary symptoms was small and non-

significant (g=0.37, [-0.34, 1.07]). This suggests that sudden gains in primary symptoms may

not necessarily lead to improvements in secondary symptoms. As the current review did not

address this, future research or reviews into early response should also focus on secondary

symptoms. Secondly, Aderka et al. (2012) found effect sizes of sudden gains to be much

larger in CBT interventions compared to non-CBT interventions. Only eight of the 25 studies

16
in the current review used therapies other than some form of cognitive therapy. This

highlights the need for more research into early response in treatments other than cognitive

therapies. Additionally, Aderka et al. (2012) found no differences in effect sizes across

depression and anxiety, while the present review indicates a stronger influence of early

response on anxiety outcomes compared to depression.

Previous authors have suggested that the early response effect could be explained by a

trend of improvement that began prior to the start of treatment, which might be indicative of

regression to the mean (Beckham, 1989). Two of the reviewed studies did in fact observe that

early pre-treatment gains are associated with trajectories of improvement, however both

studies still found significant effects of early response during treatment after controlling for

pre-treatment gains (Delgadillo et al., 2014; Lutz et al., 2017). On this basis, the effect of

early response cannot be entirely explained by regression to the mean. Furthermore, the

pooled rate of early response across studies (47.9%) was considerably higher than the rate of

short-term remission observed in waitlist control groups (~20%) of depression treatment

trials (Posternak & Miller, 2001; Whiteford et al., 2013); hence the early response effect

cannot simply be explained by spontaneous remission.

An alternative explanation could be derived from the findings by Tadić et al. (2010),

which indicated that individuals in a placebo control group also experienced early response at

an equal or higher rate than the CBT group and with similar effects on the eventual remission

of symptoms. This could potentially be explained by the nonspecific factors discussed by

Ilardi and Craighead (1994), who suggest that early response may simply be due to a

“readiness to change”, and is therefore a quasi-placebo response to factors such as increased

hope and expectations of improvement. Howard et al. (1993) have previously referred to this

as a remoralisation effect which they attributed to common factors that are likely to be

present in most forms of psychotherapy. This explanation fits within the wider literature on

17
the placebo effect in healthcare. For example, prior studies have shown experimentally that

perceiving clinicians as warm, friendly, empathic and competent enhances the placebo effect

(Howe, Goyer, & Crum, 2017; Kaptchuk et al., 2008). However, given that the four other

studies in this review using a control condition either only reported early response in the

treatment condition or pooled the data, it was not possible to determine whether early

response occurs consistently across the placebo conditions. Future individual patient data

meta-analyses of trials comparing psychotherapies with placebo could help to elucidate the

extent to which the early response phenomenon may be partly or fully explained by a quasi-

placebo effect.

An important clinical implication of this literature is that response to treatment during

the first month is a reliable prognostic indicator across multiple forms of psychological care

for common mental health problems. The fourth session could be seen as a timely

opportunity to review treatment outcomes and to identify potential obstacles to improvement,

which is consistent with the principles of routine outcome monitoring and feedback-informed

treatment (Carlier et al., 2012; Lambert et al. 2003).

4.5. Conclusions

In conclusion, a robust evidence-base supports the notion that patients showing reliable

symptomatic improvements within the initial therapy sessions tend to have a much better

prognosis after treatment. Therefore, routinely monitoring early response is of paramount

importance, as it has the potential to identify those who may have a protracted or limited

response to treatment. A clinical review using validated outcome measures at the fourth

session of therapy is recommended, in order to assess progress and to identify potential

obstacles to improvement in cases that have not yet shown reliable symptomatic

improvements.

18
DATA AVAILABILITY STATEMENT

All relevant data sources will be made available upon written request to the corresponding

author.

CONFLICT OF INTERESTS

The authors declare no conflict of interests.

REFERENCES

Aderka, I. M., Nickerson, A., Bøe, H. J., & Hofmann, S. G. (2012). Sudden gains during

psychological treatments of anxiety and depression: A meta-analysis. Journal of

Consulting and Clinical Psychology, 80(1), 93-101.

Arnow, B. A., Blasey, C., Manber, R., Constantino, M. J., Markowitz, J. C., Klein, D. N., ...

& Rush, A. J. (2007). Dropouts versus completers among chronically depressed

outpatients. Journal of Affective Disorders, 97(1), 197-202.

Başoğlu, M., Marks, I. M., Kiliç, C., Swinson, R. P., Noshirvani, H., Kuch, K., & O'Sullivan,

G. (1994). Relationship of panic, anticipatory anxiety, agoraphobia and global

improvement in panic disorder with agoraphobia treated with alprazolam and

exposure. The British Journal of Psychiatry, 164(5), 647-652.

Beckham, E. E. (1989). Improvement after evaluation in psychotherapy of depression:

Evidence of a placebo effect? Journal of Clinical Psychology, 45(6), 945-950.

19
Bell, C., Waller, G., Shafran, R., & Delgadillo, J. (2017). Is there an optimal length of

psychological treatment for eating disorder pathology? International Journal of

Eating Disorders, 50(6), 687-692.

Bradford, A., Cully, J., Rhoades, H., Kunik, M., Kraus-Schuman, C., Wilson, N., & Stanley,

M. (2011). Early response to psychotherapy and long-term change in worry symptoms

in older adults with generalized anxiety disorder. The American Journal of Geriatric

Psychiatry, 19(4), 347-356.

Braun, J. D., Strunk, D. R., Sasso, K. E., & Cooper, A. A. (2015). Therapist use of Socratic

questioning predicts session-to-session symptom change in cognitive therapy for

depression. Behaviour Research and Therapy, 70, 32-37.

Carlier, I. V., Meuldijk, D., Van Vliet, I. M., Van Fenema, E., Van der Wee, N. J., & Zitman,

F. G. (2012). Routine outcome monitoring and feedback on physical or mental health

status: Evidence and theory. Journal of Evaluation in Clinical Practice, 18, 104 –110.

Crits-Christoph, P., Connolly, M. B., Gallop, R., Barber, J. P., Tu, X., Gladis, M., &

Siqueland, L. (2001). Early improvement during manual-guided cognitive and

dynamic psychotherapies predicts 16-week remission status. The Journal of

Psychotherapy Practice and Research, 10(3), 145-154.

Delgadillo, J., McMillan, D., Lucock, M., Leach, C., Ali, S., & Gilbody, S. (2014). Early

changes, attrition, and dose–response in low intensity psychological

interventions. British Journal of Clinical Psychology, 53(1), 114-130.

Dew, M. A., Reynolds, C. F., Houck, P. R., Hall, M., Buysse, D. J., Frank, E., & Kupfer, D.

J. (1997). Temporal profiles of the course of depression during treatment: predictors

of pathways toward recovery in the elderly. Archives of General Psychiatry, 54(11),

1016-1024.

20
Doyle, P. M., Le Grange, D., Loeb, K., Doyle, A. C., & Crosby, R. D. (2010). Early response

to family‐based treatment for adolescent anorexia nervosa. International Journal of

Eating Disorders, 43(7), 659-662.

Dumville, J. C., Torgerson, D. J., & Hewitt, C. E. (2006). Research methods: reporting

attrition in randomised controlled trials. BMJ: British Medical Journal, 332(7547),

969-971.

Elkin, I., Shea, M. T., Watkins, J. T., Imber, S. D., Sotsky, S. M., Collins, J. F., ... & Fiester,

S. J. (1989). National Institute of Mental Health treatment of depression collaborative

research program: General effectiveness of treatments. Archives of General

Psychiatry, 46(11), 971-982.

Gilboa-Schechtman, E., & Shahar, G. (2006). The sooner, the better: Temporal patterns in

brief treatment of depression and their role in long-term outcome. Psychotherapy

Research, 16(3), 374-384.

Gildengers, A. G., Houck, P. R., Mulsant, B. H., Dew, M. A., Aizenstein, H. J., Jones, B. L.,

... & Reynolds III, C. F. (2005). Trajectories of treatment response in late-life

depression: psychosocial and clinical correlates. Journal of Clinical

Psychopharmacology, 25(4), S8-S13.

Gois, C., Dias, V. V., Carmo, I., Duarte, R., Ferro, A., Santos, A. L., ... & Barbosa, A. (2014).

Treatment response in type 2 diabetes patients with major depression. Clinical

Psychology & Psychotherapy, 21(1), 39-48.

Grilo, C. M., Masheb, R. M., & Wilson, G. T. (2006). Rapid response to treatment for binge

eating disorder. Journal of Consulting and Clinical Psychology, 74(3), 602-613.

21
Haas, E., Hill, R. D., Lambert, M. J., & Morrell, B. (2002). Do early responders to

psychotherapy maintain treatment gains? Journal of Clinical Psychology, 58(9),

1157-1172.

Hamilton, W. (2011). Package ‘MAVIS’ (1st ed.). Retrieved from https://

www.researchgate.net/publication/280711501_MAVIS_Meta_Analysis _via_Shiny

Heckman, T. G., Heckman, B. D., Anderson, T., Bianco, J. A., Sutton, M., & Lovejoy, T. I.

(2017). Common factors and depressive symptom relief trajectories in group

teletherapy for persons ageing with HIV. Clinical Psychology &

Psychotherapy, 24(1), 139-148.

Higgins, J. T., Altman, D. G., & Sterne, J. A. C. (2011). Assessing risk of bias in included

studies. In J. T. Higgins & S. Green (Eds.), Cochrane Handbook for Systematic

Reviews of Interventions version 5.1.0 (updated March 2011; Chap.8). London,

England: The Cochrane Collaboration. Retrieved from

https://www.handbook.cochrane.org

Howard, K. I., Lueger, R. J., Maling, M. S., and Martinovich, Z. (1993). A phase model of

psychotherapy outcome: causal mediation of change. Journal of Consulting and

Clinical Psychology, 61(4), 678-685.

Howe, L. C., Goyer, J. P., & Crum, A. J. (2017). Harnessing the placebo effect: Exploring the

influence of physician characteristics on placebo response. Health Psychology,

36(11), 1074.

Hunnicutt-Ferguson, K., Hoxha, D., & Gollan, J. (2012). Exploring sudden gains in

behavioral activation therapy for major depressive disorder. Behaviour Research and

Therapy, 50(3), 223-230.

22
Ilardi, S. S., & Craighead, W. E. (1994). The role of nonspecific factors in cognitive‐behavior

therapy for depression. Clinical Psychology: Science and Practice, 1(2), 138-155.

Jacobson, N. S., & Truax, P. (1991). Clinical significance: a statistical approach to defining

meaningful change in psychotherapy research. Journal of Consulting and Clinical

Psychology, 59(1), 12-19.

Jordan, J., Carter, J. D., McIntosh, V. V., Fernando, K., Frampton, C. M., Porter, R. J., ... &

Joyce, P. R. (2014). Metacognitive therapy versus cognitive behavioural therapy for

depression: a randomized pilot study. Australian & New Zealand Journal of

Psychiatry, 48(10), 932-943.

Kaptchuk, T.J., Kelley, J.M., Conboy, L.A., Davis, R.B., Kerr, C.E., Jacobson, E.E., Kirsch,

I., Schyner, R.N., Nam, B.H., Nguyen, L.T. and Park, M. (2008). Components of

placebo effect: randomised controlled trial in patients with irritable bowel syndrome.

BMJ, 336(7651), 999-1003.

Lambert, M. J., Whipple, J. L., Hawkins, E. J., Vermeersch, D. A., Nielsen, S. L., & Smart,

D. W. (2003). Is it time for clinicians to routinely track patient outcome? A meta-

analysis. Clinical Psychology: Science and Practice, 10, 288 –301.

Lewis, C. C., Simons, A. D., & Kim, H. K. (2012). The role of early symptom trajectories

and pretreatment variables in predicting treatment response to cognitive behavioral

therapy. Journal of Consulting and Clinical Psychology, 80(4), 525-534.

Lutz, W., Arndt, A., Rubel, J., Berger, T., Schröder, J., Späth, C., ... & Fuhr, K. (2017).

Defining and Predicting Patterns of Early Response in a Web-Based Intervention for

Depression. Journal of Medical Internet Research, 19(6), e206.

23
Lutz, W., Stulz, N., & Köck, K. (2009). Patterns of early change and their relationship to

outcome and follow-up among patients with major depressive disorders. Journal of

Affective Disorders, 118(1), 60-68.

Mallinckrodt, C. H., Clark, W. S., & David, S. R. (2001). Accounting for dropout bias using

mixed-effects models. Journal of Biopharmaceutical Statistics, 11(1-2), 9-21.

Masterson, C., Ekers, D., Gilbody, S., Richards, D., Toner-Clewes, B., & McMillan, D.

(2014). Sudden gains in behavioural activation for depression. Behaviour Research

and Therapy, 60, 34-38.

Posternak, M. A., & Miller, I. (2001). Untreated short-term course of major depression: a

meta-analysis of outcomes from studies using wait-list control groups. Journal of

Affective Disorders, 66(2), 139-146.

Rabin, A. S., Kaslow, N. J., & Rehm, L. P. (1984). Changes in symptoms of depression

during the course of therapy. Cognitive Therapy and Research, 8(5), 479-487.

Rosenthal, R. (1991). Meta-analytic Procedures for Social Research (Vol. 6). Sage.

Rubel, J., Lutz, W., Kopta, S. M., Köck, K., Minami, T., Zimmermann, D., & Saunders, S.

M. (2015). Defining early positive response to psychotherapy: An empirical

comparison between clinically significant change criteria and growth mixture

modeling. Psychological Assessment, 27(2), 478-488.

Rubio‐Aparicio, M., Sánchez‐Meca, J., López‐López, J. A., Botella, J., & Marín‐Martínez, F.

(2017). Analysis of categorical moderators in mixed‐effects meta‐analysis:

Consequences of using pooled versus separate estimates of the residual between‐

studies variances. British Journal of Mathematical and Statistical Psychology, 70(3),

439-456.

24
Rush, A. J., Kovacs, M., Beck, A. T., Weissenburger, J., & Hollon, S. D. (1981). Differential

effects of cognitive therapy and pharmacotherapy on depressive symptoms. Journal of

Affective Disorders, 3(3), 221-229.

Schibbye, P., Ghaderi, A., Ljótsson, B., Hedman, E., Lindefors, N., Rück, C., & Kaldo, V.

(2014). Using early change to predict outcome in cognitive behaviour therapy:

exploring timeframe, calculation method, and differences of disorder-specific versus

general measures. PloS one, 9(6), e100614.

Stauffer, V. L., Case, M., Kinon, B. J., Conley, R., Ascher-Svanum, H., Kollack-Walker, S.,

... & Lieberman, J. (2011). Early response to antipsychotic therapy as a clinical

marker of subsequent response in the treatment of patients with first-episode

psychosis. Psychiatry Research, 187(1), 42-48.

Steidtmann, D., Manber, R., Blasey, C., Markowitz, J. C., Klein, D. N., Rothbaum, B. O., ...

& Arnow, B. A. (2013). Detecting critical decision points in psychotherapy and

psychotherapy+ medication for chronic depression. Journal of Consulting and

Clinical Psychology, 81(5), 783-792.

Steinman, S. A., Hunter, M. D., & Teachman, B. A. (2013). Do patterns of change during

treatment for panic disorder predict future panic symptoms? Journal of Behavior

Therapy and Experimental Psychiatry, 44(2), 150-157.

Tadić, A., Helmreich, I., Mergl, R., Hautzinger, M., Kohnen, R., Henkel, V., & Hegerl, U.

(2010). Early improvement is a predictor of treatment outcome in patients with mild

major, minor or subsyndromal depression. Journal of Affective Disorders, 120(1), 86-

93.

25
Tang, T. Z., & DeRubeis, R. J. (1999a). Reconsidering rapid early response in cognitive

behavioral therapy for depression. Clinical Psychology: Science and Practice, 6(3),

283-288.

Tang, T. Z., & DeRubeis, R. J. (1999b). Sudden gains and critical sessions in cognitive-

behavioral therapy for depression. Journal of Consulting and Clinical Psychology,

67(6), 894-904.

Van, H. L., Schoevers, R. A., Kool, S., Hendriksen, M., Peen, J., & Dekker, J. (2008). Does

early response predict outcome in psychotherapy and combined therapy for major

depression? Journal of Affective Disorders, 105(1), 261-265.

van Calker, D., Zobel, I., Dykierek, P., Deimel, C. M., Kech, S., Lieb, K., ... & Schramm, E.

(2009). Time course of response to antidepressants: predictive value of early

improvement and effect of additional psychotherapy. Journal of Affective

Disorders, 114(1), 243-253.

Vevea, J. L., & Hedges, L. V. (1995). A general linear model for estimating effect size in the

presence of publication bias. Psychometrika, 60(3), 419-435.

Whiteford, H. A., Harris, M. G., McKeon, G., Baxter, A., Pennell, C., Barendregt, J. J., &

Wang, J. (2013). Estimating remission from untreated major depression: a systematic

review and meta-analysis. Psychological Medicine, 43(8), 1569-1585.

26
Table 1. Inclusion and exclusion criteria

Review question
Does early response to psychological therapy predict post-treatment depression or anxiety outcomes?
Inclusion criteria Exclusion criteria
Population Adult patients 18 years and over accessing psychological Studies exclusively with children and/or adolescents under 18.
interventions for anxiety (including Obsessive-Compulsive Disorder,
Studies exclusively with non-clinical samples.
Post-Traumatic Stress Disorder, other anxiety disorders) and/or
depression.
Intervention Any form of psychological therapy or intervention with the aim of Studies that do not include psychological interventions for depression
treating depression or anxiety and delivered in any modality and/or anxiety.
(individual, group, internet-based etc.).
Comparator A between-groups comparison of those who do and do not show Studies where no comparisons are made between cases with and
early response to psychological therapy. without early response to psychological treatment.
Studies comparing treatment outcomes between cases that had Studies where sudden gains tended to occur after the fourth week of
sudden gains early in treatment (e.g. sample median sudden gain therapy.
occurred within the first 4 weeks) and cases without sudden gains.
Outcomes The statistical significance and magnitude of the association between Studies where early response is not measured.
early response and post-treatment depression and anxiety outcomes.
Studies where post-treatment depression / anxiety outcomes are not
reported.
Setting Any settings where psychological interventions are usually delivered,
including online or telephone therapy, in any country.
Study design Randomised controlled trials and observational studies. Grey literature, such as dissertation theses, which are not published in
peer reviewed scientific journals.
Studies published in peer reviewed scientific journals, in the English
language. Editorials, newspaper or magazine articles and other forms of media.
Literature sources not in published in English.

27
Table 2. Characteristics of studies included in the review

First Author and Year Study Design Study Setting Primary Disorders Analysed Intervention Outcome Intervention
N Condition Measuresj Duration
Arnow et al. (2007) RCTb NA MDD 681 CBASP, HAMD-24 12 weeks
Nefazadone or
COMBh
Başoğlu et al. (1994) RCTb UK and US, 2 PD, agoraphobia 154 AE, PE, AR or PR Main phobia 8 to 16 weeksl
outpatient sites targets: avoidance
and feark
Beckham. (1989) Observational US, university-based Depression 23 CBT BDI NA
research site
Bradford et al. (2011) RCTb US, Primary care GAD 76 CBT PSWQ 12 weeks
clinic
Braun et al. (2015) Observationalc NA Depression 55 CT BDI-II, HRSD 16 weeks
b e g
Crits-Christoph et al. RCT NA Mixed 98 CT, SE BDI, BAI 16 to 52
(2001) weeksm

Delgadillo et al. (2014) Observationalc UK, 1 primary care Mixede 1850 LiCBT PHQ-9, GAD-7 Up to 10
mental health service weeks

Dew et al. (1997)a Observationalc US, psychiatric Depression 95 Combined NT and HAMD-17 18 weeksn
institute and clinic IPT
Gilboa-Schechtman et al. RCTb US, 3 university MDD 162 CBT, IPT, IMI- BDI, HRSD 16 weeks
(2006) research sites CM or PLA-CM

Gildengers et al. (2005)a RCTb US, university based MDD 395 IPT, NT or PX HRSD-17 12 weeks
research site
Gois et al. (2014) RCT Portugal, 3 outpatient MDD 30 IPT or Sertraline MADRS 24 weeks
diabetes clinics
Heckman et al. (2017) RCTb US, telephone based Depression 103 Group GDS 12 weeks
Teletherapy
Hunnicutt-Ferguson et Observational NA MDDf 42 BA QIDS-SR 16 weeks
al. (2012)a

Jordan et al. (2014) RCT pilot study NZ, university clinical MDD, Bipolar II 48 MCT vs CBT QIDS 16-C 12 weeks
research unit

28
Lewis et al. (2012) Observational US, university training Mixede 173 CBT BDI NA
clinic
Lutz et al. (2009) RCTb US, 3 university MDD 162 CBT, IPT, IMI- BDI, HSCL-90 16 weeks
research sites CM or PLA-CM
Lutz et al. (2017) RCT Internet-basedd Depression 409 ICBT PHQ-9 12 weeks
Masterson et al. (2014) RCTb UK, mental health Depression 40 BA PHQ-9 12 weeks
practices
Rabin et al. (1984) RCT NA Depression 98 BT, CT or CBT BDI 10 weeks
and Self-Control
Group Therapy
Rubel et al. (2015) Observational Germany, 26 Mixede 5484 Psychotherapyi GMH Not fixedo
counselling, medical
or mental health sites
Schibbye et al. (2014) Observational Sweden, intenet-based Mixede 112 ICBT OQ-45, CORE- 10 weeks or
10, MADRS-S, 15 weeksp
PDSS-SR, LSAS-
SR
Steidtmann et al. (2013) RCTb Outpatient sitesd MDD 352 CBASP or IDS-SR 12 weeks
COMBh
Steinman et al. (2013) Observationalc US, university PD 36 CBGT PDSS 12 weeks
research clinic
Tadić et al. (2010) RCTb Germany, outpatient Depression 223 Sertraline or CBT HAMD-17 10 weeks
facility
Van et al. (2008) RCTb Netherlands, 2 Depression 190 SPSP or SPSP HAMD-17 16 weeks
outpatient facilities with FX or VX
RCT, randomised control trial; NA, not available; UK, United Kingdom; US, United States; NZ, New Zealand; MDD, major depressive disorder; PD, panic disorder; GAD, generalised anxiety disorder;
CBASP, cognitive behavioural analysis system of psychotherapy; AE, alprazolam plus exposure; PE, placebo plus exposure; AR, alprazolam plus relaxation; PR, placebo plus relaxation; CBT, cognitive
behavioural therapy; LiCBT, low intensity CBT (guided self-help); CT, cognitive therapy; SE, supportive-expressive dynamic psychotherapy; NT, Nortriptyline; IPT, interpersonal psychotherapy; IMI-CM,
imipramine plus clinical management; PLA-CM, placebo plus clinical management; PX, Paroxetine; BA, behavioural activation; MCT, meta-cognitive therapy; ICBT, internet cognitive behavioural therapy; BT,
behavioural therapy; CBGT, cognitive behavioural group therapy; SPSP, short-term psychodynamic supportive psychotherapy; FX, Fluoxetine; VX, venlafaxine; HAMD, Hamilton rating scale for depression;
BDI, Beck depression inventory; PSWQ, Penn state worry questionnaire; HRSD, Hamilton rating scale for depression; BAI, Beck anxiety inventory; MADRS, Montgomery-Åsberg depression rating scale; GDS,
geriatric depression scale; QIDS-SR, quick inventory of depressive symptomatology-self report; QIDS-C, quick inventory of depressive symptomatology-Clinician assessed; HSCL, Hopkins symptom checklist;
PHQ, patient health questionnaire; GMH, general mental health; OQ, outcomes questionnaire; CORE, clinical outcomes in routine evaluation; PDSS-SR, panic disorder severity scale-self report; LSAS-SR
Liebowitz social anxiety scale-self report; IDS-SR; inventory of depressive symptomatology-self report; SG, sudden gain.
a
Open trial studies, therefore the participants and therapists were not blind to treatment conditions.
b
RCT refers to a secondary analysis of RCT.
c
Observational refers to a secondary analysis of observational studies.
d
Country of study not reported
e
Mixed refers to a variety of depression and anxiety disorders.
f
Healthy controls were also enrolled to take part in the study
g
Here we report the largest sample reported by the authors, which was used to investigate the predictive value of early response.

29
h
COMB refers to the psychotherapy used and the medication used combined
i
Does not state what psychotherapy methods were used
j
Only measures used to assess early response or early sudden gains
k
Main phobia targets were modified from Marks and Mathews (1979). Includes avoidance, self and assessor rated, and fear, self and assessor rated.
l
Psychological treatment lasted 8 weeks, medication could continue to week 16, 20 weeks for patients in the cross validation sample
m
16 weeks for depression or anxiety with 3 monthly booster sessions, 52 weeks for cases with avoidant or obsessive-compulsive personality disorder
n
The authors focused only on the 18 weeks of therapy before randomisation (secondary analysis of RCT)
o
Treatment was not fixed to a strict time limit, sessions varied from 4 to 109 sessions (m=9.76, SD = 8.25)
p
10 weeks for depression and panic disorder, 15 weeks for social anxiety disorder

30
Table 3. Findings reported by studies included in the review

First Author Early N (% of total) Reported Statistics/Findings

and Year Response Early Others
Timingf Responders
Arnow et al. Week 2 NA NA No difference between dropouts and completers in terms of early response (main effect F=.33,
(2007) p=.57; dropout x treatment F=.11, p=.89).

Başoğlu et al, Week 2-4 NA NA Early improvement (by week 4) in avoidance of main phobia targets explained 45% of variance in
(1994) clinical global impression (CGI) ratings by week 8. Week-4 improvement in fear explained 24%
of variance in patients’ global impression (PGI) ratings by week 8.

Beckham. Session 4-6 NA NA The BDI at session 6 correlated with the final session BDI (r=.80, p<.001).
(1989)

Bradford et Week 4 23 (30.3%) 53 Responders at 3 months had a significantly larger magnitude of change at 4 weeks compared to
al. (2011) non-responders (t(74)=3.423, p<.001). This was the same for 15 month responders (t(74)=3.069,
p=.003).

Braun et al. Week 4 NA NA Symptom improvements from session 1 to 4 were significant and substantial (d=1.59, t=5.58,
(2015) p<.001).

Crits- Week 4 NA NA Change on the BDI and BAI measures from baseline to weeks 2, 3 and 4 significantly predicted
Christoph et remission at week 16 (all p<.005). At week 4, the Area under the curve (AUC) values for early
al. (2001) change in each outcome measure were .73 for BDI and .77 for BAI.

Delgadillo et Week 4 NA NA For patients whose treatment lasted ≥5 sessions, reliable improvement by session 4 predicted
al. (2014) post-treatment reliable and clinically significant improvement; with significant odds ratios in
depression (pooled OR = 12.60, p < .001) and anxiety (pooled OR = 21.10, p < .001) measures.

Dew et al. Week 4-6 29 (30.5%) 66 62% of the early response group already had below threshold HAMD scores by week 6. All
(1997) subjects had these low scores by weeks 8-10. By week 4, early responders had a higher
percentage of cases with remission of symptoms compared to non-early responders (x2 = 13.1, p <
.01).

31
Gilboa- Week 4 122 (75.3%) 40 The end of treatment BDI (F = 18.17, p < .01) and HRSD (F = 22.59, p < .01) scores were
Schechtman significantly lower for the rapid responders group compared to others.
et al. (2006)

Gildengers et Week 2-5 181 (47.9%) 197 Early responders in NT+IPT reached HRSD score of 10 by week 5, PX+IPT by week 3 and
al. (2005) NT/PX by week 2.

Gois et al. Week 6 22 (73.3%) 8 Early responders showed significant improvement in depression over time in both IPT and
(2014) Sertraline groups (F(2.4)=27.73, p<.001).

Heckman et Week 4 32 (30.5%) 73 Post-treatment scores for early responders M=7.43, SD=4.2, delayed responders M=6.41,
al. (2017) SD=4.64, non-responders M=17.3, SD=6.17.

Hunnicutt- SG median pre- 15 (35.7%) 27 Sudden gain patients had significantly different rates of treatment response to non-sudden gain
Ferguson et gain session 1a participants (X2=5.09, p<.05).
al. (2012)b

Jordan et al. Week 4 NA NA Participants in both therapies showed clinically significant improvements on the QIDS at week 4
(2014) (MCT d=.74, 95% CI [.30, 1.17]; CBT d=.73, 95% CI [.31, 1.14]).

Lewis et al. Session 5 NA NA Around 76% of total symptom change was observed after intake, reaching 85.47% of total
(2012) symptom change at session 5.

Lutz et al. Week 8 99 (61.1%) 63 Class membership (early response and non-early response) was significantly associated with
(2009) reliable improvement post-treatment (X2(9)=74.8, p<.001).

Lutz et al. Week4 343 (83.9%) 66 Rapid early response effect size d=2.9; moderate early improvement d=1.19.
(2017)

Masterson et SG median pre- 17 (42.5%) 23 At post-treatment sudden gain groups had a mean PHQ score of M=5.3, SD=3.63, whereas non-
al. (2014) gain session 2a sudden gain participants had M=10.2, SD=6.8; t(35)=2.92, p=.006.

Rabin et al. Week 3-4 NA NA For the majority of depressive symptoms, the largest changes occurred during the first 2 to 4
(1984) sessions.

Rubel et al. Session 3 396 (7.2%) 5088 Early responders showed the highest pre-post effect sizes (ds=1.88-2.16) and the highest share of
(2015) reliably improved patients at post-treatment (90-93%).

32
Schibbye et Week 4 NA NA The disorder specific measures at week 4 explained variance in outcomes at post-treatment (panic
al. (2014) disorder R2=.34, p<.01; depression R2=.41, p>.01; social anxiety disorder R2=.43, p<.01).

Steidtmann Week 6-8 NA NA Percentage of symptom reduction at weeks 4, 6 and 8 predicted post-treatment HRSD remission
et al. (2013) status in combined treatment (week 4: x2 = 6.26) and psychotherapy (week 4: x2 = 4.05).

Steinman et SG session 2 19 (52.8%) 17 Cluster membership (sudden gains versus others) was a significant predictor of post-treatment
al. (2013) outcomes, accounting for 25% of the variability in six-month follow-up scores.

Tadić et al. Week 2 95 (42.6%) 128 In both CBT and Sertraline, early response was a highly sensitive predictor of later stable
(2010) response (76-82%) and stable remission (71-72%).

Van et al. Week 8 109 (57.4%) 81 Early responders and non-responders in the psychotherapy condition were significantly different
(2008) from each other in terms of final non-response (x2=5.069, p<.01), but this was not the case for
remission rates (x2=.994, p=.319). In combined therapy, there were significant differences in
terms of non-response (x2=16.019, p<.001) and remission (x2=12.435, p<.001).
SG, sudden gain; NA, not available. Overall mean proportion of cases with early response across studies = 47.9%
a
Median pre-gain session refers to the therapy session immediately preceding the sudden gain (Tang & DeRubeis, 1999b)

33
Figure 1. PRISMA flow diagram of the systematic study selection

34
Figure 2. Random effects meta-analysis: post-treatment effect sizes comparing early responders vs. others

35