Reliability and validity of Short Form 36 Version 2 to measure health perceptions in [CFS]

[Dolphin]

Reliability and validity of Short Form 36 Version 2 to measure health perceptions in a sub-group of individuals with fatigue.

Disabil Rehabil. 2011;33(25-26):2596-604. doi: 10.3109/09638288.2011.582925. Epub 2011 Jun 20.

Davenport TE, Stevens SR, Baroni K, Van Ness JM, Snell CR.


Source
Department of Physical Therapy, Thomas J. Long School of Pharmacy and Health Sciences, University of the Pacific, Stockton, California, USA. tdavenport@pacific.edu

Abstract*

PURPOSE:

To determine the validity and reliability of Short Form 36 Version 2 (SF36v2) in sub-groups of individuals with fatigue.

METHOD:

Thirty subjects participated in this study, including n = 16 subjects who met case definition criteria for chronic fatigue syndrome (CFS) and n = 14 non-disabled sedentary matched control subjects.

SF36v2 and Multidimensional Fatigue Inventory (MFI-20) were administered before two maximal cardiopulmonary exercise tests (CPETs) administered 24 h apart and an open-ended recovery questionnaire was administered 7 days after CPET challenge.

The main outcome measures were self-reported time to recover to pre-challenge functional and symptom status, frequency of post-exertional symptoms and SF36v2 sub-scale scores.

RESULTS:

Individuals with CFS demonstrated significantly lower SF36v2 and MFI-20 sub-scale scores prior to CPET.

Between-group differences remained significant post-CPET, however, there were no significant group by test interaction effects.

Subjects with CFS reported significantly more total symptoms (p < 0.001), as well as reports of fatigue (p < 0.001), neuroendocrine (p < 0.001), immune (p < 0.01), pain (p < 0.01) and sleep disturbance (p < 0.01) symptoms than control subjects as a result of CPET.

Many symptom counts demonstrated significant relationships with SF36v2 sub-scale scores (p < 0.05). SF36v2 and MFI-20 sub-scale scores demonstrated significant correlations (p < 0.05).

Various SF36v2 sub-scale scores demonstrated significant predictive validity to identify subjects who recovered from CPET challenge within 1 day and 7 days (p < 0.05).

Potential floor effects were observed for both questionnaires for individuals with CFS.

CONCLUSION:

Various sub-scales of SF36v2 demonstrated adequate reliability and validity for clinical and research applications.

Adequacy of sensitivity to change of SF36v2 as a result of a fatiguing stressor should be the subject of additional study.


PMID: 21682669 [PubMed - indexed for MEDLINE]

*I've given each sentence its own paragraph

 

[Dolphin]

At 1 day follow-up, 93% of control subjects reported full recovery within 24 h of the second CPET, compared to 19% of subjects with CFS (p<0.001).

Thirty-one percent of subjects with CFS still rated themselves as incompletely recovered at 1-week follow-up.

Twenty-five percent of control subjects reported no symptoms during the follow-up period, compared to no subjects with CFS.

SF36v2 may demonstrate floor effects in some subjects with CFS. Considering SF36v2 sub-scale scores, a ‘0’ rating was recorded by two subjects (13%) for physical function, three subjects for role physical (19%), two subjects for role emotional (13%), five subjects for vitality (31%), one subject for mental health (6%), four subjects for social function (25%), one subject for general health (6%) and one subject for health transition (6%).

------
Last paragraph, before the limitations section which is at the very end:

Measures that predict the duration of PEM following CPET challenge can provide researchers and clinicians with a potentially valuable marker of prognosis and indicator of improvement.

In this study, various SF36v2 sub-scales demonstrated significant predictive validity.

All physical component sub-scales of the SF36v2 demonstrated significant predictive value to determine recovery within 1 day, as well as the social function and vitality sub-scale scores from the mental component.**

To predict recovery within 1 week, the bodily pain, vitality, and role emotional subscales became significant predictors with moderate to robust diagnostic accuracy.***

Future work should aimto replicate the cut scores identified in this study, as well as prospectively validate their prognostic importance to predict clinical status in individuals with CFS who are receiving treatment.

Floor effects in some individuals with CFS may limit the value of both SF36v2 sub-scale scores to properly capture limited health perceptions.

This may partially explain the absence of significant group by test interaction effects on SF36v2 sub-scale scores observed in this study.

The presence of floor effects in SF36v2 raises the possibility that either a fatigue-specific or health condition-specific questionnaire, such as MFI-20, may be necessary for patients with CFS.

The relative clinimetric performance of fatigue-specific and general health questionnaires should be an area for further research.

*I've given each sentence its own paragraph

**The cut-offs:
1 day later:

Physical function (48%) 0.880 (0.697–1.00) 0.82 0.82 0.57 0.75 4.5 0.21 0.001*{
Role physical (57%) 0.865 (0.706–1.00) 0.79 0.88 0.90 0.77 6.9 0.23 0.001*{
Bodily pain (76%) 0.911 (0.764–1.00) 0.85 0.81 0.85 0.81 4.4 0.18 50.001*{
General health (28%) 0.898 (0.000–1.00) 0.85 0.81 0.85 0.81 4.4 0.18 50.001*
Role emotional 0.659 (0.449–0.869) – – – – – – 0.157
Vitality (41%) 0.836 (0.672–1.00) 0.85 0.81 0.85 0.81 4.4 0.18 0.003*
Social function (69%) 0.854 (0.695–1.00) 0.79 0.90 0.90 0.79 7.9 0.23 0.002*
Mental health 0.672 (0.467–0.876) – – – – – – 0.227
Health transition 0.424 (0.180–0.669) – – – – – – 0.501

This meant that individuals who scored below the bolded marks on that particular SF-36 v2 subscale before doing any testing were likely not to have recovered 24 hours after the first exercise test.

***The cut-offs:
7 days later:
7-day recovery

Physical function 0.771 (0.594–0.947) – – – – – – 0.061
Role physical 0.717 (0.531–0.903) – – – – – – 0.133
Bodily pain (39%) 0.829 (0.681–0.977) 0.90 0.58 0.90 0.58 2.2 0.17 0.009*
Role emotional (71%) 0.875 (0.699–1.00) 0.90 0.58 0.90 0.58 2.2 0.17 0.009*
Vitality (22%) 0.792 (0.630–0.953) 0.88 0.58 0.88 0.58 2.1 0.20 0.043{
Social function 0.683 (0.438–1.00) – – – – – – 0.204
Mental health 0.742 (0.483–1.00) – – – – – – 0.094
General health 0.758 (0.550–0.967) – – – – – – 0.073

Health transition 0.242 (0.00–1.00) – – – – – – 0.073

This meant that individuals who scored below the bolded marks on that particular SF-36 v2 subscale before doing any testing were likely not to have recovered 6 days after doing the second of two exercise tests.

 

[Dolphin]

The SF-36 v2 questionnaire can be seen here:
http://www.healthmeasurement.org/pub_pdfs/_QUESTIONNAIRE_SF-36, version 2.pdf

A lot of the questions are ask about the last four weeks, so it's probably not that surprising that in this study, with the small sample sizes there were, no group x test interactions that were statistically significant: for a lot of the scores for people with CFS, there was virtually no change. However, other sorts of tests at seven days might have found differences.

 

[Guido den Broeder]

A rather weird study. This is by no means a validation. As you say, the questions are about the last four weeks. But no participant will have answered with that in mind.

Nor can the conclusions be generalized. The study says nothing about the validity and reliability of the SF36 for measuring the effects of some therapy, which is how it is most often used (and where it fails hopelessly).

 

[Dolphin]

I suppose I should say that the SF-36 v2 scores at 7 days were only a part of the study.
They did compare the baseline SF-36 v2 scores with MFI-20 baseline scores and baseline SF-36 v2 scores with those who then went on to report symptoms.

There isn't really a problem with "last four weeks" aspect for the baseline SF-36 v2 scores; the "four week problem" only arises when one brings in the scores at seven days, so only part of the study.

 

[biophile]

"Thirty-one percent of subjects with CFS still rated themselves as incompletely recovered at 1-week follow-up." When I overdid it once in a GET program, I thought I was virtually "recovered" by the next session one week later, but then I did less and suffered more later, suggesting I was not completely recovered. There may be an extra period of vulnerability once the obvious post-exertional symptoms subside, something which I have observed many other times as well.

 

[alex3619]

Interesting. I wonder though if there are any studies showing how changes in attitude can alter SF-36 scores without altering real functioning? That is the confounding factor we appear to be facing.

 

[Valentijn]

This is something I wondered about for PACE and deterioration - I think I scored 0 for one of the SF-36 scales while just mostly housebound and able to get to clinics once in a while. So how do they determine deterioration for someone that starts at 0? And if deterioration of greater than 20 points is required (was that mentioned somewhere?), that would mean that everyone at 20 and under technically could not deteriorate, even if their condition became much worse.

 

[Dolphin]

This is something I wondered about for PACE and deterioration - I think I scored 0 for one of the SF-36 scales while just mostly housebound and able to get to clinics once in a while. So how do they determine deterioration for someone that starts at 0? And if deterioration of greater than 20 points is required (was that mentioned somewhere?), that would mean that everyone at 20 and under technically could not deteriorate, even if their condition became much worse.

I imagine you scored 0 on Role Physical. A lot of people with ME score 0 on that.
However, that wasn't used. What they sued was SF-36 physical function. It is the 10 questions at number 3 here: http://www.anapsid.org/cnd/files/sf36.pdf . I very much doubt people able to take part in the trial would have scored 0 on this: it's 5 points for each "yes, limited a little" and 10 points for each "no, not limited at all".

However, requiring a 20 point drop at two consecutive points is quite a big drop: somebody would really have to be very ill to do that if they started at, say, 30 points.