CBT for chronic fatigue syndrome in a rehabilitation setting:effectiveness and predictors of outcome

[Dolphin]

Cognitive behavioural treatment for chronic fatigue syndrome in a rehabilitation setting: effectiveness and predictors of outcome.

Behav Res Ther. 2011 Dec;49(12):908-13. doi: 10.1016/j.brat.2011.09.004. Epub 2011 Sep 28.

Schreurs KM, Veehof MM, Passade L, Vollenbroek-Hutten MM.

Source
University of Twente, Enschede, The Netherlands. k.m.g.schreurs@utwente.nl

Abstract


Cognitive behavioural therapy (CBT) was combined with graded exercise therapy (GET) for patients with chronic fatigue syndrome (CFS) in an uncontrolled implementation study of an inpatient multidisciplinary group therapy.

During the intake procedure, 160 CFS patients completed a questionnaire on fatigue related measurements, physical impairment, depression, somatic and psychological attributions, somatic focus, and sense of control over symptoms.

Pre-treatment physical activity level was measured with an actometer.

At baseline, post-treatment and 6-month follow-up individual strength, subjective fatigue and physical impairment, were reassessed.

Large effect sizes were found on subjective fatigue (1.2 post-treatment; 1.2 follow-up) and physical impairment (-.9 post-treatment; -.9 follow-up), Clinically significant improvement was found in 33.8% of the participants at post-treatment and 30.6% at follow-up.

Individual strength at post-treatment was predicted by level of physical activity before treatment, and by sense of control over symptoms and physical activity at follow-up.

Clinically significant improvement in subjective fatigue was predicted by not receiving a disablement insurance benefit, shorter duration of fatigue, higher sense of control over symptoms and, at follow-up by more pre-treatment physical activity.

In conclusion, the intervention was effective for CFS patients.

Cognitive behavioural factors that perpetuate fatigue symptoms are also predictors of treatment outcome.

 

[Dolphin]

This was an inpatient programme. It presumably is expensive.

Patients were referred to the rehabilitation program by a general practitioner or a medical specialist. They were referred for a CFS program and the diagnosis of CFS had to be confirmed. In an intake procedure prior to the treatment, inclusion and exclusion criteria were assessed. One to three months after this procedure, patients started treatment. Treatment consisted of 12 weeks in which patients were alternately five days in the institute and nine days at home. Subsequently, they were three months at home and finally they were treated for a last five days. The total treatment took thus about 25 weeks in which patients received a total of 75 therapy sessions. The program was offered to groups of six patients by a team consisting of a psychologist, a rehabilitation physician, a physiotherapist, an occupational therapist, a sports instructor and a social worker. The daily activities in the group were supervised by social workers educated in group supervision. The program consisted of CBT interventions aimed at diminishing the influence of perpetuating factors. Cognitive restructuring, problem solving, and individual goal setting, were part of the interventions of psychologists and social workers. GET was given by physiotherapists and sport instructors. Interventions of occupational therapists were aimed at optimizing the level of daily activities.

Follow-up was six months later.

Fatigue was assessed with the CIS-20, an instrument that has been shown to have high validity and reliability (Vercoulen et al., 1994, 1998). The total scale is a measure of individual strength and consists of 20 items assessing four dimensions of fatigue: subjective experience of fatigue, concentration, motivation and physical activity level. Items are scored on a 7-point Likert scale. The score, ranging from 20 to 140, was computed by summing all items. Higher scores indicate lower levels of individual strength. In addition to the total CIS, the subjective fatigue subscale can be used separately. This score (range 8e56) was computed by summing the eight items of the subscale. Internal consistency of the total scale and the subjective fatigue subscale in this study was satisfactory, with Cronbach’s alphas of .8 and .7, respectively.

Physical impairment was measured with the, frequently used and psychometrically sound, physical subscale of the Medical Outcome Study-Short Form-36 Health Survey (SF-36) (Stewart & Ware, 1992). This scale, consisting of 10 items, ranges from 0 to 100 with higher scores representing less physical impairment. A score 65 indicates the absence of severe disability. In our sample, Cronbach’s alpha was .9.

Here are the basic results:

Mean (sd) (n = 160).

Subjective fatigue (fatigue subscale of CIS-20, possible scores 8-56, higher is worse):
Baseline: 50.5 (6.0)
Post-treatment: 38.3 (12.0)a
Follow-up: 38.4 (12.2)a

Individual strength (CIS-20 - summation of all 20 items, possible scores 20-140)
Baseline: 108.4 (14.5)
Post-treatment: 85.7 (24.1)a
Follow-up: 86.0 (22.9)a

Physical impairment (SF-36 Physical functioning):
Baseline 51.9 (19.0)
Post-treatment: 67.4 (20.6)b
Follow-up: 69.3 (18.2)a

a Significant (p < 0.001) difference with T1, tested with a paired t-test.
b Significant (p < 0.001) difference with T1, tested with Wilcoxon signed ranks
test.

 

[Dolphin]

They use actometers, but only at baseline (or at least that's all they report anyway):

The level of physical activity was measured with the Actilogscale V3.0 (Van der Werf, Prins, Vercoulen, Van der Meer, & Bleijenberg, 2000; Vercoulen et al., 1994). This actometer was attached to the ankle and worn for 14 consecutive days before the start of treatment. A higher score indicates a higher activity level.

They make a bit of a fuss about how their study is better than another study as they use actometers (rather than a subjective measure) to look at predictors, without highlighting that their own study doesn't use actometers as an outcome measure:

CFS patients who were less active before treatment were more fatigued afterwards and showed less change in fatigue. This result corroborates earlier findings in which activity was also measured by means of an objective accelerometer (Prins et al., 2001). A recent study with activity level measured by a self-reported questionnaire did not find a relationship between activity level and outcome (Kemke et al., 2010). These contradictory results are probably caused by differences in measurement methods. In a recent study it was concluded that self-reported measures of activity level all suffer from the inherent problem that people are not able to judge this behaviour correctly (Scheeres, Knoop, Meer, & Bleijenberg, 2009). Self reported activity level may be particularly problematic in patients with CFS since they worry a lot about the activities they feel not able to perform.

They again praise themselves for using actometers in the concluding paragraph:

A large sample of patients was involved in this study, and we were able to use standardized measurement and, an objective measurement of activity patterns. The study was limited by the selection bias of patients admitted to a tertiary rehabilitation centre and the absence of a control group. However, the study also shows that CBT combined with GET may be effective in the less controlled circumstances of daily practice.

 

[Dolphin]

Inclusion criteria were: a confirmed diagnosis of CFS according to the CDC criteria; being older than 18 years; severe fatigue (Checklist Individual Strength- 20 (CIS-20) subscale subjective fatigue 35); being motivated to participate in a self-management program; being able to follow therapies for 5 h a day. Exclusion criteria were: severe psychopathology (Symptom Checklist-90 (SCL-90) >295); extensive cognitive deficits; severe psychosocial problems of an acute nature that needed solution before treatment; enforced motivation by a legal procedure concerning financial benefit. Only a small proportion of patients (N = 6) who were strongly motivated to participate in the program despite being involved in a legal procedure, were admitted to the program.

Out of 160 patients,149 completed the program (7% dropped out of treatment) and, 116 (73%) finished the follow-up measurement six months after the end of the program. There were no differences in demographic variables between completers and participants with missing measurements. Patients who dropped out of treatment did not differ from those who stayed in treatment in functioning, and cognitive and behavioural variables at intake. Patients who did not finish the follow-up measurements had less symptoms of depression at intake, but did not differ from completers in all other variables at intake, nor at the measurements at the start or end of program.

Patients were defined as clinically significantly improved if they had a reliable change index >1.96 on the CIS-20 subjective fatigue subscale, a CIS-20 subjective fatigue score 35 and a SF-36 physical score 65 (Scheeres et al., 2008).

 

[Sean]

They make a bit of a fuss about how their study is better than another study as they use actometers (rather than a subjective measure) to look at predictors, without highlighting that their own study doesn't use actometers as an outcome measure:

If they had used them at outcome, and they showed a favourable result, they would surely have highlighted it. Demonstrating a clear objective improvement at outcome, strongly correlated to self-report measures and CBT/GET, would be a serious feather in their cap.

Individual strength at post-treatment was predicted by level of physical activity before treatment, and by sense of control over symptoms and physical activity at follow-up.

Clinically significant improvement in subjective fatigue was predicted by not receiving a disablement insurance benefit, shorter duration of fatigue, higher sense of control over symptoms and, at follow-up by more pre-treatment physical activity.

Translation: Patients who are less sick at the start of the trial tend to have better outcomes on subjective self-report measures.

 

[WillowJ]

Translation: Patients who are less sick at the start of the trial tend to have better outcomes on subjective self-report measures.

yes, I was thinking the reasoning looked a little circular.

Is it customary to use two different tests for significance on different data bits? just curious...

 

[Dolphin]

When these results are combined with some other results, for some factors the results are not consistent:

Some of the variables of the theoretical model of perpetuating factors that has been developed by Vercoulen et al. (1998), have also prognostic value for treatment success. Higher level of physical activity before the start of treatment, a shorter duration of fatigue complaints, not receiving a financial benefit and, a high sense of control over complaints at admission, were related to treatment success. The results corroborate with the studies of Prins et al. (2001) who also found that sense of control over complaints was a predictive factor of treatment outcome and, that causal attributions were not a predicting outcome. Some other studies have found that causal attributions predict outcome (Butler et al., 1991; Chalder et al., 2003; Darbishire et al., 2005; Deale et al., 1997).

Depression at intake was not a significant predictor of fatigue at the end of treatment or six months after treatment. It may be that the measurement of depression in the current study is confounded by expectations of patients. Depression was measured right before the intake procedure. Possibly, a number of patients may have been reluctant to admit psychological problems to prevent that professionals would attribute their fatigue to psychological causes. Others may have been desperately in need of help and therefore admitting many psychological problems. Our result is in contrast to a recent study that used the same model of potential prognostic factors and found that depression was the only predictor of post-treatment fatigue (Kemke et al., 2010). The prognostic role of psychological and psychiatric problems remains unclear with some studies finding a negative effect of distress and depression on treatment outcome (Bentall et al., 2002; Darbishire et al., 2005) and others finding that psychiatric comorbidity is not related to treatment outcome (Deale et al., 1997; Prins et al., 2005). We recommend that future studies use the same measurement instrument, by preference a structured psychiatric interview, to assess depression and other comorbid psychiatric diagnoses.

In contrast to other studies, age (Quarmby et al., 2007) was not an independent predictors of treatment outcome. Receiving a disablement insurance benefit (Bentall et al., 2002; Prins et al., 2002) has been found to negatively influence treatment outcome, and it is also one of the predictors of recovery six months after treatment. Finally, in this study a longer duration of complaints was related to less favourable outcomes, which is in contrast with other studies that did not find such an effect (Butler et al., 1991; Quarmby et al., 2007). Duration of complaints in this sample is longer than in the most studies (e.g., Scheeres et al., 2008). Patients are only treated in a tertiary institution when all other treatment facilities are exhausted. Our results show that it may be advisable to start treatment at an earlier moment.

A better understanding of the pathophysiology of the condition, I think would give more interesting answers than more and more studies like this.

 

[Dolphin]

Is it customary to use two different tests for significance on different data bits? just curious...

Not that common, I think.

Here is why they did it:

To study treatment effectiveness, changes in subjective fatigue, individual strength and physical impairment were analysed with paired t-tests (normally distributed variables) and Wilcoxon signed ranks tests (not normally distributed variables).

 

[Dolphin]

When these results are combined with some other results, for some factors the results are not consistent:

A better understanding of the pathophysiology of the condition, I think would give more interesting answers than more and more studies like this.

Another frustrating thing about all this is all the signs are that they're not particularly interested in the predictors in terms of deciding who should or shouldn't get a therapy.

Also, the things measured are only done at one moment in time, while there could be some variation e.g. in terms of self-efficacy/degree to which one feels in control. Also, this could also change during therapy.

 

[biophile]

So, after decades of CBT/GET for CFS research, weakish contradictory findings for almost all the pillars of their model, except for the circular reasoning factor of disability benefits, the effects of therapy-infused self-efficacy response bias on reported symptoms, and the actometers which have previously debunked presumed increases in activity but now they did not report on in this study? Notice the self-praising for using actometers despite not reporting the actual results at followup.

The Dutch researchers still use the Vercoulen et al model of CFS. Did they ever acknowledge or respond to the study by Jason et al which long ago debunked that model for CFS but found that it fit chronic fatigue for psychiatric diagnoses?

Another frustrating thing about all this is all the signs are that they're not particularly interested in the predictors in terms of deciding who should or shouldn't get a therapy.

CBT/GET proponents might get more interested in a hurry when biomedical science starts offering patients real treatments and then newly resourced ex-patients start suing for the quackery pushed onto them over the years. Guessing from this study, it looks like the more severely affected are less likely to respond to CBT/GET? Re predictors and mediators, limited information on the PACE results suggests nothing which stands out except a small effect on altering patients' perceptions.

 

[Snow Leopard]

One issue that does not get mentioned much is not all GET or CBT is the same. The model that Jason has studied has quite different expectations (and seems much more acceptable to patients), but seems to have similar outcomes. Yet this is quite different to the model that is used in Europe.

 

[Simon]

...a high sense of control over complaints at admission, were related to treatment success. The results corroborate with the studies of Prins et al. (2001) who also found that sense of control over complaints was a predictive factor of treatment outcome

This finding strikes me as a little odd.

If you were just looking at the natural course of the illness, you might expect those who reported a greater sense of control over their symptoms to do better than those who reported less control. (I don't think the paper considers this, but I haven't read the full text).

However, this study uses a combination of CBT and GET, both of which explicitly aim to give patients a greater sense of control over symptoms. So you might expect those struggling most to control their symptoms to improve the most if the therapies do what they claim: through the wonders of CBT and GET these patients gain greater control over theri symptoms and so improve. Instead the reverse seems to be true - those reporting least control over their symptoms did less well than those who already had a better sense of control over their symptoms before the study started. That suggests that the 'control' aspect of CBT/GET doesn't have much impact.

Or have I missed something important here?

 

[Dolphin]

This finding strikes me as a little odd.

the study said:

...a high sense of control over complaints at admission, were related to treatment success. The results corroborate with the studies of Prins et al. (2001) who also found that sense of control over complaints was a predictive factor of treatment outcome​

If you were just looking at the natural course of the illness, you might expect those who reported a greater sense of control over their symptoms to do better than those who reported less control. (I don't think the paper considers this, but I haven't read the full text).

However, this study uses a combination of CBT and GET, both of which explicitly aim to give patients a greater sense of control over symptoms. So you might expect those struggling most to control their symptoms to improve the most if the therapies do what they claim: through the wonders of CBT and GET these patients gain greater control over theri symptoms and so improve. Instead the reverse seems to be true - those reporting least control over their symptoms did less well than those who already had a better sense of control over their symptoms before the study started. That suggests that the 'control' aspect of CBT/GET doesn't have much impact.

Or have I missed something important here?

Yes, there could be something to what you are saying.

I think what Prins et al (2001) may be referring to is not baseline sense-of-control (although it's not fully clear to me), which is different to what was found in this current study.

If one looks at Table 1 in

How Valid Is The Model Behind Cognitive Behavior Therapy For Chronic Fatigue Syndrome? An Evaluation Of The Additional Data From The Trial By Prins Et Al.
http://www.iacfsme.org/BULLETINSUMM...tenPrinsEvaluation8289/tabid/436/Default.aspx

one sees how sense of control improved with CBT. That's what one would expect to happen. Indeed it was the only one of the three theoretical factors that was significant:

Guided Support and treatment as usual revealed that CBT had no significant impact on somatic attributions and focusing on bodily symptoms, and that in line with established guidelines, these two variables were not mediating factors. The only variable in the model showing an effect of CBT was sense of control.

If a treatment is supposed to work by improving sense of control, baseline measurements should matter too much because CBT should be able to improve low scores. And as you say, it might be likely that one could improve lower scores more than higher ones (although it might depend if there are ceiling effects - if no ceiling effects, and plenty of scope for everyone to improve, one might not be sure how things would go).

 

[xchocoholic]

Criteria - severely fatigued but could handle 5 hour of therapy daily. How is
that severe ?

I couldn't have tolerated this even in the beginning of my cfs much less now.