Telephone-based guided self-help for adolescents with CFS: A non-randomised cohort study-Lloyd et al

[Dolphin]

This abstract doesn't list lots of areas where there were no improvement/no differences, or indeed how some of the differences mentioned in the abstract aren't clear cut (some other people might have left them out). I'll try to list them below.

Telephone-based guided self-help for adolescents with chronic fatigue syndrome: A non-randomised cohort study.

Behav Res Ther. 2012 May;50(5):304-12. doi: 10.1016/j.brat.2012.02.014. Epub 2012 Mar 10.

Lloyd S, Chalder T, Sallis HM, Rimes KA.

Source

Department of Psychological Medicine, Institute of Psychiatry, King's College London, Weston Education Centre, Cutcombe Road, London, UK. Samantha.lloyd@kcl.ac.uk

Erratum in
• Behav Res Ther. 2013 Aug;51(8):518.

Abstract


The aim of this study was to gain preliminary evidence about the efficacy of a new telephone-based guided self-help intervention, based on cognitive-behavioural principles, which aimed to reduce fatigue and improve school attendance in adolescents with chronic fatigue syndrome (CFS).

A non-randomised cohort design was used, with a two-month baseline period.

Sixty-three 11-18 year-old participants recruited from a specialist CFS unit received the intervention.

Participants received six half-hour fortnightly telephone sessions and two follow-up sessions.

Fatigue and school attendance were the main outcomes and the main time point for assessing outcome was 6 months post-treatment.

Using multi-level modelling, a significant decrease in fatigue was found between pre-treatment and 6 month follow-up, treatment effect estimate = - 5.68 (-7.63, -3.72), a large effect size (Cohen's d = 0.79).

The decrease in fatigue between pre and post-treatment was significantly larger than between baseline and pre-treatment.

A significant increase in school attendance was found between pre-treatment and 6 month follow-up, effect estimate = 1.38 (0.76, 2.00), a medium effect size (d = -0.48).

Univariate logistic regression found baseline perfectionism to be associated with better [corrected] school attendance at six-month follow-up.

In conclusion, telephone-based guided self-help is an acceptable minimal intervention which is efficacious in reducing fatigue in adolescents with CFS.

 

[Dolphin]

This study used the Oxford criteria.

---------

Measurements were generally taken at 5 timepoints:
Baseline
At: 2 months later (just before treatment/therapy started)
At: 5 months (i.e. 3 months later, after therapy had just finished)
At: 8 months (i.e. 3 months after therapy had finished)
At: 11 months (i.e. 6 months after therapy had finished)

 

[Dolphin]

The very first sentence is:

Chronic fatigue syndrome (CFS) is characterised by severe disabling fatigue, present for more than 50% of the time and affecting both physical and mental functioning, which is not accounted for by organic illness.

The usual way of putting it is something like not explained by other physical or mental illness.

In the introduction, all the examples given are from psychological conditions/disorders e.g. anxiety, depression, OCD and eating disorders.

 

[Dolphin]

Although a telephone intervention, it wasn't aimed at the most ill/disabled:

An additional inclusion criterion was that participants were able to attend a face-to-face assessment in order to confirm diagnosis of CFS; this intervention was not aimed at the most severely affected, housebound adolescents, who were likely to require (and were offered) more intensive face-to-face intervention at home.

-------

Intervention

The main aims of the intervention were to address cognitive and behavioural factors that may be acting to perpetuate fatigue and other symptoms. This approach was based on the assumption of the possibility of recovery, rather than merely symptom management.

The young person was assisted in developing a regular sleep routine and encouraged not to sleep in the daytime.

The intervention involved a self-help manual written for this purpose, based partly on a cognitive behavioural model described in an existing book for adolescents (Chalder, 2002). The manual included an explanation of CFS and associated symptoms, advice on managing the condition and how to set targets.

Chalder, T. (2002). Self help for chronic fatigue syndrome: A guide for young people. Witney, Oxon: Blue Stallion Publications.

 

[biophile]

Univariate logistic regression found baseline perfectionism to be associated with better [corrected] school attendance at six-month follow-up.

So perfectionism is now good in CFS?

 

[Dolphin]

Wald tests were used to check for the equality of the treatment effect estimates during the baseline to pre-treatment and pretreatment to end of treatment phases. A significant Wald test result indicates that the change during these two periods is significantly different. Although the main time point for assessing outcome was six months follow-up, the Wald tests compared the baseline to pre-treatment phase with the pre-treatment to end of treatment phase. This was because these two phases were more comparable in terms of the length of time of each phase. In other words, a significant Wald test result for a comparison between the baseline to pre-treatment phase and the pre-treatment to 6 month follow-up phase could potentially have reflected simply the increased length of time elapsed since pre-treatment rather than the effects of the intervention.

(Actually, the period from baseline to pre-treatment was only two months, versus three months between the pre-treatment end of treatment).

Anyway, for five of the six measures there wasn't a statistical difference on this!

Measure where there was a difference:
Fatigue

Measures where there wasn't a difference:
-School attendance
- Impairment (Social Adjustment Scale) (despite the title, this isn't measuring a psychological construct but physical functioning/similar)
-Depression
-Adjustment
-Anxiety.

You might ask how does this fit in with what is in the abstract:

A significant increase in school attendance was found between pre-treatment and 6 month follow-up, effect estimate = 1.38 (0.76, 2.00), a medium effect size (d = -0.48).

Here's an example:
School attence:
36.28%: Baseline
42.99%: At: 2 months later (just before treatment/therapy started)
48.80%: At: 5 months (i.e. 3 months later, after therapy had just finished)
53.40%: At: 8 months (i.e. 3 months after therapy had finished)
59.27%: At: 11 months (i.e. 6 months after therapy had finished)

So there is a difference between the pre-treatment and 6-months follow-up (42.99% vs 59.27%) but one can't be sure this wasn't just due to the passage of time as in the two months at the start before treatment, there was an improvement (from 36.28% to 42.99%).
[Actually the Wald test only compares 36.28% and 42.99% vs 42.99% and 48.80%, but one appears a similar trend might also explain the 59.27% attendance figure.

 

[Dolphin]

Global improvement and satisfaction
Both adolescents and parents were asked to rate how the adolescent was compared to before treatment on a 7-point scale from ‘very much better’ to ‘very much worse’. Adolescents and parents were also asked how satisfied they were overall with the treatment received on a 7-point scale from ‘very satisfied’ to ‘very dissatisfied.’

So a 7-point CGI scale (as in the PACE Trial)

Global improvement and satisfaction
The majority (71.4%) of participants felt they had experienced some degree of improvement at the end of treatment. Parent ratings of improvement were higher with 79.6% of parents feeling there had been some degree of improvement. Sixty-six percent of adolescents and 70.4% of parents were either very or moderately satisfied at the end of treatment.

It is unclear to me whether "some degree of improvement" would include a CGI of 3 or not i.e. "a little better".

 

[Dolphin]

Predictors of main outcomes at 6 month follow-up

In order to investigate predictors of fatigue at 6 month follow-up the following hypothesized risk factors were entered into a linear regression model: length of illness, baseline scores for depression, anxiety, impairment and adjustment, maternal mental wellbeing, self-oriented perfectionism, socially prescribed perfectionism, amount of handbook content read, age and gender. In univariate analyses, none of the above variables were found to be significantly associated with fatigue at 6 months follow-up (see Table 4).

Logistic regression was carried out with the same predictor variables but omitting impairment, in order to predict poor school attendance (less than 70% attendance) at 6 month follow-up. In univariate analyses, baseline self-oriented perfectionism and socially-prescribed perfectionism were found to be significantly associated with increased odds of having a poor outcome for attendance (see Table 4). When entered into a multivariate model neither variable remained significant.

(For some reason Adjustment is blank for "school attendance")
This means that they looked at 11 possible predictors for fatigue and none were predictors on their own.

For school attendance, they looked at 10 possible predictors for fatigue and none were predictors on their own and two were statistically significant. This is at the level of p<=0.05. However given the number of predictors checked, this finding could be due to chance

Also, the confidence intervals for the two that were statistically significant only barely missed zero:

Poor outcome for school attendance
at 6 month follow-up (univariate or and 95% CIs):
Self-oriented perfectionism: 1.073 (1.002, 1.149)*
Socially-prescribed perfectionism: 1.111 (1.009, 1.214)*

If one number had been minus, it wouldn't have been statistically significant.

Perfectionism wasn't set as a categorical measurements so these may be points per unit which makes them a little stronger.

 

[Dolphin]

Univariate logistic regression found baseline perfectionism to be associated with better [corrected] school attendance at six-month follow-up.

So perfectionism is now good in CFS?

Interesting. This must be what this erratum refers to:

http://www.ncbi.nlm.nih.gov/pubmed/22459729

Erratum in

• Behav Res Ther. 2013 Aug;51(8):518

The original paper has:

Univariate logistic regression found baseline perfectionism to be associated with poorer school attendance at six-month follow-up.

in abstract

and they then go on to talk about this (when the corrected version says the opposite happened)

This is the first study to show that higher baseline levels of perfectionism are associated with lower school attendance following an intervention for CFS in adolescents. This finding is consistent with cognitive behavioural models of CFS that highlight the important role of perfectionism in this condition (e.g. Surawy et al., 1995). It is possible that the more perfectionist adolescents were more avoidant of school attendance due to concerns about negative evaluation of their performance. It is not clear why baseline perfectionism did not also predict fatigue severity after treatment. It is possible that if the more perfectionist individuals were more likely to be avoiding school, their perfectionistic beliefs were not currently causing sufficient distress or unhelpful behavioural responses as to be contributing to fatigue. This areawarrants further research as the minimal nature of the intervention meant that there was less time to fully address perfectionism. Adolescents with high pre-treatment levels of perfectionism may require more comprehensive interventions. This is in line with the findings of previous research which suggested that depressed adults with high levels of perfectionism may respond better to longer, more intensive treatments in comparison to minimal interventions (Blatt & Ford, 1994). However, as multiple associations were tested within this

So it looks like the wording in the abstract is not good.

An erratum has been posted for this paper:

so perhaps it deals with this point.