CIQ Properties

Community (re)integration refers to being in (returning to) the mainstream of family and community life, by persons with impairments and disabilities due to injury, chronic illness, or old age. Other terms for the same or similar phenomena include: independent living; normalization; deinstitutionalization; mainstreaming;lack of handicap; social participation. Community integration may be measured by collecting information on:

• activities, including instrumental activities of daily living (IADLs), hobbies and productive activity in household, work, school, and community;
• family, kin, and work relationships;
• social roles in the marital dyad, family and society at large;
• and sometimes physical independence, community mobility, and economic independence.

The Community Integration Questionnaire (CIQ) was intended to be a brief, reliable measure of an individual's level of integration into the home and community. It is important to note that this questionnaire represents a finite set of indicators of community integration, and as such does not encompass all possible indicators of integration.

A detailed analysis of the clinimetric properties of the CIQ is contained in the review paper by Dijkers in the Journal of Head Trauma Rehabilitation, 1997. (See references). The following is a short summary, supplemented by information from research published after that date:

Reliability. Results of reliability studies have been mixed. Based on the (Pearson) correlations reported in the earliest study, the interrater reliability of the CIQ appears in the "acceptable" range (Willer et al., 1993). However, the intraclass correlation coefficient (ICC), a more appropriate measure, resulted in much lower numbers, according to a later investigation (Tepper et al., 1996), especially for the Home (H) dimension. The most recent research (Sander et al., 1997) also suggests that in home integration there is the greatest discrepancy between reports by subjects with TBI and those by their proxies. In the latter study, the person with TBI tends to report higher values than the proxy for all three components.

Sander et al. also calculated kappa for the correspondence between person with TBI and proxy reports on each of the 15 individual CIQ questions Sander et al, 1997). Kappas ranged from .42 (shopping) to .94 (school). While these figures are in the moderate or better range for kappa, there was a tendency for patients to rate themselves as more integrated on the Home and Productivity (P) items, and on three of the Social Integration (S) items.

The traditional cut-off point for acceptable internal consistency as measured by coefficient alpha is .80. For the CIQ total, three out of four studies (Willer, Linn & Allen, 1994, Willer, Ottenbacher and Coad, 1994; Corrigan and Deming, 1995; Heinemann and Whiteneck, 1995) have reported levels that exceed this criterion. However, the corresponding values for the S and P dimensions are much lower, especially for the latter - quite likely due to the fact that alpha is based on only two variables, which are very dissimilar from one another: work/study/volunteering, and travel.

Distribution issues. Corrigan and Deming (1995) noted negatively skewed distributions for the premorbid data for CIQ total, positive skews for H for all four samples, negative skews for all samples for the S dimension, and negative skew for P for the TBI-premorbid sample, but positive skew for the three others. Various kurtosis (peakedness/flatness of the distribution) problems were also noted. They recommended that the P subscale not be used independently from its contribution to the CIQ total score. Distribution problems that could not be resolved completely by transformations of the CIQ scores were also noted by Willer, Ottenbacher & Coad (1994). Non-parametric statistics may need to be used in order to deal with potentially non-normal distributions.

Hall et al. (1996) noted that about half of their subjects reached a score equal to or greater than the mean of a control sample on the H and S dimensions. (equivalent to the average value for the Willer, Linn & Allen (1994) control sample) at one or two years post-injury. On the P dimension, however, less than one quarter reached the controls' average. The authors state that this limits the CIQ's usefulness. However, they do not report what percentages reached the theoretical maximum for each scale - the true measure of a ceiling effect.

Validity. No formal content or face validity studies of the CIQ have been done, but it was developed utilizing a panel consisting of both consumers and professionals with expertise in TBI outcome studies

CIQ subscales and total score have been found to have correlations with subscales and total score on CHART (another measure of community integration), impairment and disability, time since injury, and subjective quality of life (Heinemann and Whiteneck, 1995). However, two small studies reported results that contradicted these findings (Burleigh et al., 1997; Smith et al., 1998). Most studies find negative correlations between CIQ and subtotals (except sometimes H) and age. Females tend to have higher CIQ scores for total, H and S than males, but lower P scores.

Sensitivity. The available research shows that the CIQ can validly distinguish between persons with TBI and non-disabled controls (Corrigan and Deming, 1995; Willer, Ottenbacher & Coad, 1994; Gordon et al., 1999). With one exception, persons with TBI are less integrated along all dimensions than controls. In a comparison between individuals with mild TBI and normal controls, individuals with mild TBI scored significantly lower on the productivity subscale (Paniak et al, 1999). There were no differences between these two groups in the home and social integration subscales. In one study (Willer, Ottenbacher & Coad, 1994), CIQ scores distinguished between three groups of persons with TBI living in settings differentiated by supervision/support level: independent in the community, in the community with some (natural) support, and in an institution such as a nursing home, rehabilitation facility, etc. Willer et al. (1999) reported gain in CIQ (sub)scores for people with TBI receiving residential rehabilitation services between the second and third anniversary of injury, while a control group who received no or minimal home-based services did not show a gain. In another study, at most a Evidence of changes in CIQ (sub)scores over time are absent from the literature, with one exception where a very small increase from the first to the second anniversary of injury was reported for a group that in large majority did not any more receive rehabilitation services (Hall et al, 1996). Corrigan et al. (1998) on the other hand, found trends toward improvement in all three subscales and total score in a cross-sectional study covering anniversaries one through four.