Reliability of the Clinical Outcome Variables Scale When Administered Via Telephone to Assess Mobility in People With Spinal Cord Injury

Article Outline

• Abstract
• Methods
  • Phase 1: Equivalence Reliability of COVS and TCOVS
    • Participants
    • Procedure
  • Phase 2: Test-Retest Reliability of the TCOVS
    • Participants
    • Procedure
  • Analysis
• Results
  • Equivalence Reliability of TCOVS and COVS
  • TCOVS Test-Retest Reliability
• Discussion
  • Study Limitations
• Conclusions
• Acknowledgments
• References
• Copyright

Abstract 

Barker RN, Amsters DI, Kendall MD, Pershouse KJ, Haines TP. Reliability of the Clinical Outcome Variables Scale when administered via telephone to assess mobility in people with spinal cord injury.

Objective

To examine the equivalence reliability and test-retest reliability of the Clinical Outcome Variables Scale (COVS) when administered via telephone (TCOVS) to people with spinal cord injury (SCI).

Design

Equivalence (telephone administration vs in-person) and test-retest reliability study.

Setting

Assessments conducted in participants’ home environment.

Participants

Equivalence reliability was examined in a convenience sample of 37 people with a diagnosis of traumatic SCI who had been discharged from the Queensland Spinal Injuries Unit to the community. In a separate group of participants, test-retest reliability of COVS when administered via telephone was examined in 43 people with SCI who were randomly selected from the Queensland Spinal Cord Injuries Service records.

Interventions

Not applicable.

Main Outcome Measures

Reliability was assessed at the subscale and composite score level using intraclass correlation coefficients (ICC_2,1) and Bland-Altman limits of agreement.

Results

Reliability was good for TCOVS and COVS for the composite score (ICC=.98), mobility subscale (ICC=.97), and ambulation subscale (ICC=.99). Reliability was also good for TCOVS test and retest assessments for the composite score (ICC=1), mobility subscale (ICC=1), and ambulation subscale (ICC=1). For all comparisons, most data points were within the 95% limits of agreement and the width of limits of agreement were considered to be clinically acceptable.

Conclusions

The study findings confirm the equivalence and test-retest reliability of the TCOVS in an SCI population when administered by trained raters.

Key Words: Outcome assessment (health care), Rehabilitation, Reproducibility of results, Spinal cord injuries, Telephone

PEOPLE WITH TRAUMATIC spinal cord injury (SCI) can now expect to live close to a normal life span; however¹ many may experience age-related health issues at an earlier age than the general population.², ³ To investigate these health issues, the Queensland Spinal Cord Injuries Service (QSCIS) is undertaking a longitudinal study to track changes in people with SCI across the lifespan. Assessment occurs via telephone interview, a method of data collection that is easily administered, cost effective, and inclusive of a geographically dispersed group of people. Many outcome measures that are included in the assessment are traditionally administered in-person, particularly functional measures. It is necessary, therefore, to confirm that information obtained by telephone interview is similar to information obtained by in-person observation (equivalence reliability) and that information obtained by telephone interview is consistent over time (test-retest reliability). Only then can any variability in assessment findings be attributed to real change, and not simply to variability in reporting.

One of the key functional outcome measures to be used in the longitudinal study is the Clinical Outcomes Variable Scale (COVS), a clinician rated, composite measure of mobility that is used routinely across the continuum of care provided by QSCIS.⁴ It has also been used for general rehabilitation populations and specific diagnostic groups such as stroke, traumatic brain injury, amputations, and musculoskeletal injuries, and has been applied in acute, inpatient, and outpatient rehabilitation settings, and community settings.⁴, ⁵, ⁶ In its current form, it consists of 13 items, which comprise rolling (2 items), lying to sitting (1 item), sitting balance (1 item), transfers (2 items), ambulation (4 items), wheelchair mobility (1 item), and arm function (2 items). All 13 items are rated by a clinician through observation and assessment of task performance according to detailed guidelines. Each COVS item is scored on a 7-point scale ranging from 1 (fully dependent mobility) to 7 (normal independent mobility). COVS scores are generally reported as a single composite score ranging from 13 to 91. In an SCI population, 2 subscales have also been reported, of which the first is a general mobility score ranging from 7 to 49 and the second is an ambulation score ranging from 5 to 35.⁷ As such, COVS offers a more relevant and more complete profile of mobility after SCI when compared with other mobility tools. A list of individual items and those that are included in the composite score and subscale scores is outlined in table 1. The findings of previous studies have shown that COVS has acceptable construct validity when used in an SCI population, particularly when the 2 subscales are used.⁷ It has been found to discriminate between respondents across lesion levels, completeness of injury level, American Spinal Injury Association impairment grade and walking status at the individual item, composite score, and subscale score levels. In addition, COVS has been found to have acceptable internal consistency and interrater reliability between hospital and community settings and greater sensitivity than the FIM instrument in measuring changes over time for assessment of mobility in people with SCI.⁸

Table 1. COVS Individual Items: Inclusion in the Composite Score and Subscale Scores

The purpose of the present study was (1) to examine the equivalence reliability of COVS scores obtained through in-person functional assessment and through telephone interview and (2) to examine test-retest reliability of the COVS when administered via telephone (TCOVS).

Back to Article Outline

Methods 

We examined equivalence reliability and test-retest reliability in 2 study phases, using 2 different study samples and study procedures. Informed consent was obtained from participants prior to data collection. Ethics approval for this project was provided by the Princess Alexandra Hospital Human Research Ethics Committee.

Phase 1: Equivalence Reliability of COVS and TCOVS 

We examined equivalence reliability of COVS in a convenience sample of 41 people recently discharged from the Queensland Spinal Injuries Unit after either their primary rehabilitation or their readmission for management of secondary conditions. All people who attended the Transitional Rehabilitation Program between September 2004 and November 2005 were invited to participate. This program assists people with SCI in the transition from hospital rehabilitation to community living.⁹ Because it is part of the QSCIS continuum of care, all people undergoing rehabilitation with QSCIS were eligible. The criteria for inclusion consisted of (1) a diagnosis of traumatic SCI, (2) aged over 15 years at time of injury, (3) discharged from hospital to a community setting more than 4 weeks previously, (4) living within the same home environment on each testing occasion, and (5) adequate cognitive and verbal skills to complete a telephone interview. The criteria for exclusion were (1) ventilator dependent, (2) difficult psychosocial situation (eg, substance abuse), (3) poor knowledge of English, or (4) limited access to a telephone. A total of 53 people were eligible to participate, of whom 41 agreed to participate and 37 completed both assessments within the 1-week period. The final study sample was considered representative of people with SCI on the QSCIS records in terms of age, sex, and level and completeness of injury. The average age ± standard deviation (SD) of participants was 41±18 years, of which 31 (84%) were men and 6 (14%) were women, 18 (49%) had tetraplegia and 19 (52%) paraplegia, and 19 (51%) had a complete injury and 18 (49%) an incomplete injury. The average time since injury was 8±3.7 months with the exception of 1 participant whose injury had occurred 41 years previously.

Three to 4 weeks after discharge from hospital, we approached potential participants in person, provided with a verbal and written explanation of the study purpose, procedure, and consent process, and asked if they were willing to participate. To prevent temporal bias and order effects, each person who agreed to participate was randomly allocated (with the throw of a die) to 1 of 2 groups: group A (TCOVS assessment first, COVS assessment second) or group B (COVS assessment first, TCOVS assessment second). Each participant was then assessed within the home environment on 2 occasions by trained raters according to group allocation. Both assessments were completed within a 1-week period to ensure that the functional ability of participants would not change substantially. To prevent recall on the part of the trained raters, 1 rater conducted the COVS in-person assessments and the second rater conducted the telephone assessment. For the in-person assessment, participants were asked to perform each task and were then rated on their performance. The TCOVS assessments were performed via a guided telephone interview format in which written questions were provided to participants in advance. The rater encouraged participants to describe their current abilities and prompted them to do so with the statement “If I was to visit you today and ask you to perform these tasks, describe how you would do them.” Both raters were blind to previous assessments. All data were collected according to standard procedures⁴ by the 2 raters, both of whom were physiotherapists trained in the administration of both the COVS and the TCOVS and experienced in administration of the COVS within an SCI population. Time taken to conduct both in-person and telephone assessments was recorded.

Phase 2: Test-Retest Reliability of the TCOVS 

We examined test-retest reliability of the TCOVS in a random sample of 43 people with SCI from the QSCIS records who met the same inclusion criteria as for phase 1. Prior to randomization, stratification into 6 groups according to year of injury occurred (<5y, 5–10y, 10–15y, 15–20y, 20–25y, >25y), with 7 participants then selected from each group. This selection process was used to ensure equivalence across strata and representation of the SCI population over the last 50 years. Contact details were obtained by searching QSCIS records and telephone directories. Potential participants were contacted by telephone, provided with a verbal explanation of the study purpose, procedure, and consent process, and invited to participate. The selection process was repeated until 7 participants were recruited to each group. Contact was made with 56 people, of whom 45 agreed to participate and 43 completed both assessments within an average time period of 15±2 days. Two people failed to complete the second assessment due to a major change in their personal situation (n=1) and because they were unavailable (n=1). The final study sample was considered representative of people with SCI on the QSCIS records in terms of age, sex, and level and completeness of injury. The average age of participants was 46±10 of which 35 (81%) were men and 8 (19%) women, 18 (42%) had tetraplegia and 25 (58%) paraplegia, and 21 (49%) had a complete injury and 22 (51%) an incomplete injury.

Test-retest reliability of TCOVS was examined together with the core of established measures included in the QSCIS longitudinal study. Each participant completed assessments twice. To prevent recall, each occasion was separated by a 2-week period, as recommended by psychometric texts.¹⁰ It was not anticipated that the functional ability of participants would change substantially over this period, given the rate of progression in the SCI population observed clinically. Once verbal consent had been provided, an appointment for the first interview was made and information sheets, consent forms, and the interview questionnaire were sent to the participant. Immediately after the first interview, the participant was asked to destroy or dispose of the interview questions. A week later the participant was contacted a second time to make an appointment for the second telephone interview, scheduled to occur 2 weeks after the initial interview. A copy of the interview questionnaire was then mailed to the participant a second time. At the beginning of the second interview, the participant was asked if they had experienced any major change in their health or life, physically or emotionally, to rule out the possibility that a functional change had occurred between the 2 testing sessions.

Analysis 

We analyzed the data using SPSS software.^a Analysis was performed for the total composite score and 2 subscales consistent with the use and interpretation of COVS in clinical practice and research. Analysis of individual items was also performed in order to identify sources of error.

Equivalence reliability and test-retest reliability were assessed using 2 methods: intraclass correlation coefficients (ICCs)¹¹ and the Bland-Altman limits of agreement.¹² An ICC_2,1 was selected on the basis that each participant was assessed by the same raters and that the raters were expected to represent the population of raters who may be using this procedure in practice (as we assume people to use this in practice will also be trained appropriately). Although the ICC is designed primarily for use with interval data, the ICC can be applied without distortion to data on the ordinal scale when intervals between such measurements are assumed to be equivalent.¹³ Therefore, the ICC was used on the assumption that the intervals between categories were equivalent, given that ordinal subscales were added together to get the overall score. The ICC values higher than .75 were considered to represent good reliability and values below .75 to represent poor to moderate reliability.¹⁴ Limits of agreement were examined by plotting the mean score and the difference between the 2 scores. The 95% limits of agreement were computed as the mean difference ±1.96 SD of the difference scores. Plots were generated using Microsoft Excel.^b

Back to Article Outline

Results 

Equivalence Reliability of TCOVS and COVS 

The mean time period between assessments ± SD was 4±4 days. The average time taken to administer the TCOVS assessment (5min) was substantially less than the average time taken to administer the COVS in-person assessments (35min).

Correlation between the TCOVS and COVS was good for the composite score (ICC=.98), the general mobility subscale (ICC=.97), and for the ambulation subscale (ICC=.99). ICCs, mean scores, mean difference between paired scores, and limits of agreement for the composite scores and individual items are displayed in table 2. Correlation between TCOVS and COVS individual items was also good (ICC range, .89–.99) with the exception of sitting balance (ICC=.52) and wheelchair mobility (ICC=.42), for which there was only moderate correlation.

Table 2. Equivalence Reliability of TCOVS and COVS for the Composite Score, Mobility Subscale, Ambulation Subscale, and Individual Item Scores

Item Aggregate	ICC2,1	TCOVS	COVS	TCOVS & COVS	TCOVS − COVS	95% Limits of Agreement
						Lower	Upper
Composite score	.98(.96–.99)	54±18	55±20	55±19	−2±4	−9.8(−7.5to−12.1)	6.5(4.2–8.9)
Mobility subscale	.97(.94–.98)	37±12	39±13	38±13	−2±3	−8.3(−6.5to−10.2)	4.5(2.7–6.3)
Ambulation subscale	.99(.97–.99)	11±9	11±9	11±9	0±2	−3.1(−2.3to−3.9)	4.1(2.0–3.7)
Individual items
Roll right	.95(.91–.97)	5.3±2.3	5.5±2.2	5.4±2.2	−0.2±0.7	−1.6(−1.2to−2.0)	1.2(0.8−1.6)
Roll left	.95(.91–.97)	5.4±2.3	5.5±2.2	5.5±2.2	−0.2±0.7	−1.6(−1.2to−2.0)	1.2(0.8–1.6)
Supinetosit	.91(.83–.95)	5.1±2.2	5.4±2.2	5.3±2.2	0.3±1.0	−2.2(−1.6to−2.7)	1.6(1.1–2.2)
Sitting balance	.52(.24–.72)	3.7±2.1	4.9±2.3	4.3±1.9	−1.1±2.1	−5.3(−4.1to−6.5)	3.1(1.9–4.2)
Horizontal transfer	.91(.84–.95)	4.9±2.4	5.1±2.3	5.0±2.3	−0.2±1.0	−2.1(−1.6to−2.7)	1.8(1.2–2.3)
Vertical transfer	.94(.88–.97)	3.2±2.3	3.3±2.4	3.3±2.3	−0.1±1.0	−1.7(−1.2to−2.2)	1.6(1.1–2.1)
Walking	.97(.94–.98)	2.0±1.8	1.9±1.9	2.0±1.8	0.1±0.5	−0.9(−.06to−1.1)	1.0(0.7–1.2)
Walking aids	.99(.98–.99)	1.8±1.6	1.8±1.6	1.8±1.6	0.0±0.2	−0.5(−0.3to−0.6)	0.5(0.3–0.6)
Walking distance	.96(.93–.98)	1.9±1.8	2.0±2.0	2.0±1.9	−0.1±05	−1.1(−0.8to−1.4)	1.0(0.6–1.3)
Walking speed	.89(.80–.94)	2.0±1.7	2.1±2.0	2.0±1.8	0.1±0.9	−1.8(−1.3to−2.3)	1.7(1.2–2.2)
Wheelchair mobility	.42(.11–.65)	6.1±.68	5.7±1.0	5.9±0.7	0.4±0.9	−1.3(−0.8to−1.8)	2.2(1.7–2.7)
Right arm function	.89(.79–.94)	6.2±1.2	6.0±1.4	6.1±1.3	0.1±0.6	−1.1(−0.8to−1.5)	1.4(1.0–1.7)
Left arm function	.90(.81–.95)	6.0±1.6	6.1±1.4	6.0±1.5	−0.1±0.7	−1.4(−1.0to−1.8)	1.3(0.9–1.6)

NOTE. For each of the paired scores, the columns represent the ICC (95% confidence interval [CI]), the mean ± SD of the TCOVS, COVS, and TCOVS and COVS, the difference between TCOVS and COVS, and the 95% limits of agreement (CI).

Bland-Altman plots (fig 1) showed good agreement between TCOVS and COVS. The difference and 95% limits of agreement between the paired means for the composite scale were −2 (−9.8 to 6.5). The difference between paired means for the general mobility subscale were −2 (−8.3 to 4.5) and for the ambulation scale were 0 (−3.1 to 4.1).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  Bland-Altman plots that examine the agreement between TCOVS and COVS for the (A) composite score; (B) general mobility score; and (C) ambulation score. The y axis represents the difference between TCOVS and COVS paired scores. The x axis represents the mean of TCOVS and COVS paired scores. Dashed lines represent 95% limits of agreement.

TCOVS Test-Retest Reliability 

Correlation between TCOVS test and TCOVS retest was good for the composite score (ICC=1), the general mobility subscale (ICC=1), the ambulation subscale (ICC=1), and for all individual items (ICC>.95 for all items) (table 3). Bland-Altman plots (fig 2) showed good agreement between TCOVS test and TCOVS retest. The difference and 95% limits of agreement between the paired means were 0 (−4.1 to 3.7) for the composite score, 0 (−2.5 to 2.8) for the mobility subscale, and 0 (−3.1 to 2.6) for the ambulation subscale.

Table 3. Test-Retest Reliability of TCOVS 1 and TCOVS 2 for the Composite Score, Mobility Subscale, Ambulation Subscale, and the Individual Item Scores

Item Aggregate	ICC2,1	TCOVS 1	TCOVS 2	TCOVS 1 & 2	TCOVS 1 − 2	95% Limits of Agreement
						Lower	Upper
Composite score	1.00	58±21	58±21	58±21	0.0±1.5	−4.1(−3.1to−5.2)	3.7(2.6–4.7)
Mobility subscale	1(0.99–1.00)	37±12	37±12	37±12	0.0±1.1	−2.5(−1.8to−3.2)	2.8(2.1–3.5)
Ambulation subscale	1(0.99–1.00)	15±11	15±11	15±11	0.0±1.5	−3.1(−2.4to−3.9)	2.6(1.8–3.3)
Individual items
Roll right	0.99(0.98–0.99)	5.5±2.0	5.4±2.1	5.5±2.0	0.1±0.3	−0.6(−0.4to−0.8)	0.7(0.6–0.9)
Roll left	0.99(0.98–0.99)	5.3±2.2	5.3±2.2	5.3±2.2	0.0±0.3	−0.7(−0.5to−0.8)	0.7(0.5–0.9)
Supinetosit	0.99(0.98–1.00)	5.1±2.2	5.2±2.2	5.2±2.2	−0.1±0.3	−0.6(−0.5to−0.8)	0.6(0.4–0.7)
Sitting balance	0.99(0.98–0.99)	3.6±2.4	3.7±2.4	3.6±2.4	−0.1±0.4	−0.8(−0.6to−1.0)	0.7(0.5–0.9)
Horizontal transfer	1(0.99–1.00)	5.2±2.6	5.2±2.6	5.2±2.6	0.0±0.2	−0.4(−0.3to−0.5)	0.4(0.3–0.5)
Vertical transfer	1.00	4.0±2.4	4.0±2.4	4.0±2.4	0.0±0.2	−0.5(−0.4to−0.6)	0.4(0.2–0.5)
Walking	0.95(0.91–0.97)	2.8±2.4	2.9±2.4	2.9±2.4	−0.1±0.8	−1.6(−1.2to−2.0)	1.4(1.0–1.8)
Walking aids	1.00	2.7±2.4	2.7±2.4	2.7±2.4	0.0±0.0	0.0	0.0
Walking distance	1.00	2.5±2.2	2.5±2.2	2.5±2.2	0.0±0.0	0.0	0.0
Walking speed	0.96(0.93–0.98)	2.6±2.3	2.6±2.3	2.6±2.3	0.0±0.9	−1.7(−1.2to−2.1)	1.7(1.3–2.2)
Wheelchair mobility	0.98(0.96–0.99)	6.4±0.7	6.3±0.7	6.3±0.7	0.0±0.2	−1.2(−0.9to−1.6)	1.2(1.0–1.6)
Right arm function	0.99(0.98–0.99)	6.2±1.3	6.3±1.3	6.2±1.3	0.0±0.2	−0.5(−0.4to−0.6)	0.4(0.5–0.3)
Left arm function	1.00	6.2±1.1	6.2±1.1	6.2±1.1	0.0±0.0	0.0	0.0

NOTE. For each of the paired scores, the columns represent the ICC (95% CI), the mean ± SD of the TCOVS 1, TCOVS 2, and TCOVS 1 and TCOVS 2, the difference between TCOVS and COVS and the 95% limits of agreement (CI).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 2. 

  Bland-Altman plots that examine the agreement between TCOVS and COVS for (A) composite score; (B) general mobility score; and (C) ambulation score. The y axis represents the difference between TCOVS 1 and TCOVS 2 paired scores. The x axis represents the mean of TCOVS 1 and TCOVS 2 paired scores. Dashed lines represent 95% limits of agreement.

Back to Article Outline

Discussion 

The results of this study indicate that COVS assessment of people with SCI by trained raters is similar when COVS is administered via telephone interview and by in-person observation. Assessment via telephone interview was also found to be consistent over time. These findings support the use of TCOVS as a reliable measure of mobility when administered by trained raters. Coupled with the fact that it can be administered quickly and to a geographically dispersed group of people, TCOVS provides a convenient, time efficient, and geographically inclusive method for the collection of data across the lifespan for people with SCI. The reliability of the TCOVS also compares more than favorably with the reliability of the telephone version of the FIM instrument motor subscale, which appears to be the only other functional measure with demonstrated reliability when administered by telephone to people with SCI.¹⁵

The equivalence reliability of TCOVS composite score and 2 subscales was good, illustrated by both high correlation coefficients and more than 90% of data points within the 95% limits of agreement. Although we would argue that the width of the limits of agreement is clinically acceptable, there may be some scope for reducing these limits, however. Inspection of the individual items revealed that 1 item, the sitting balance item, had only moderate correlation and wide limits of agreement. Although this may reflect real differences in the SCI person’s experience, recall, and perception of sitting balance compared with the physiotherapist’s direct observation, it may also reflect the shortcomings of the operational definitions for the sitting balance item. According to the standard COVS guidelines, sitting balance is assessed with legs over the side of the bed with no hands for support, a position that would not be used functionally by the majority of people with SCI. Surprisingly, no assessment is made of balance in long sitting or of sitting with 1 hand for support, both of which represent functional sitting balance strategies used by people with SCI and incorporated into training during rehabilitation.¹⁶ It is probable, therefore, that the reliability of this item could be improved by modifying the operational definitions to include balance in long sitting and with support of 1 hand, thereby making the item more relevant to everyday life for people with SCI. Given that the COVS is used widely in other diagnostic groups, however, further investigation of this item in diagnostic groups other than SCI would be necessary prior to item modification.

A second individual item that also appeared to show a sizeable amount of error between in-person and telephone assessment was the wheelchair mobility item. Interestingly, there was only moderate correlation between the TCOVS and COVS scores, but acceptable limits of agreement. A small mean difference between TCOVS and COVS scores ± SD (0.4±0.9) for the wheelchair mobility item would suggest that the level of error was low. High mean scores for TCOVS ± SD (6.1±0.68) and COVS (5.7±1.0) on a scale of 7 suggests that scores were aggregated in the top part of the scale, however, which led to a lower correlation coefficient. Therefore, rather than a high level of error, the low correlation coefficient reflects a skewed distribution likely to have poor sensitivity to further improvements, as previously reported.⁸

Study Limitations 

It is important to highlight a number of methodologic limitations that may have influenced the results of this study. First, the findings may represent an underestimate of equivalence reliability because 2 different raters were used to conduct the assessments. On that basis, differences between paired scores may reflect not only a different method of administration, but also different raters or a combination of both. Even so, any difference is likely to be small, because interviewers were highly trained and experienced in the use of the TCOVS, familiar with functional abilities after SCI, and comfortable with administering this tool with people with SCI. Second, participants in this study had all been discharged from hospital in the previous 2 months and would therefore have undergone in-person COVS assessments immediately prior to discharge from hospital. As a result, it is possible that reliability estimates could have been biased. On the one hand, participants would have had a recent opportunity to attempt each task in a manner consistent with the operational definitions for COVS assessment. If, instead, the sample had consisted of people with SCI of longer duration, the reliability might have been lower for items that measured tasks not performed on a regular basis, such as vertical transfers. On the other hand, people with longer duration SCI may have been more aware of their exact functional limitations. In contrast, participants who had been recently discharged from hospital may still have been exploring their functional limits while taking on new challenges and opportunities within their home and community environment. Additionally, if this was the case, the period between assessments may not have been sufficiently brief to ensure that functional status had not changed. Ideally, future research should be conducted to confirm the equivalence reliability of TCOVS in a population of people with SCI of longer duration.

Test-retest reliability was good to almost perfect for the composite score, the subscales and individual item scores. This suggests that people with SCI are consistent in the way in which they score their mobility when both tests are performed by the same rater. Because the findings arose from people who have sustained SCI over the past 50 years, it is reasonable to suggest that they may be generalized to the population of people with SCI of traumatic origin.

The strength of the findings may be due to attributes of the COVS that are known to enhance the reliability of self-report measures.¹⁷ With the exception of the sitting balance item, the COVS includes functional items that are relevant to day-to-day life of people with SCI, with each item defined using nontechnical jargon. The strength of the findings may also be attributed to the standardized method of administration by a single trained rater who was a physiotherapist with experience in the use of the TCOVS and familiar with function after SCI. Further research would be required to determine the reliability of TCOVS when performed by physiotherapists unfamiliar with function after SCI or physiotherapists inexperienced in the use of COVS.

Alternatively, it is possible that the test-retest reliability is inflated due to a number of methodologic limitations. Participants may not have destroyed the original survey as requested, allowing them to repeat their responses from the first interview. In addition, the 2-week period between each test may have been brief enough to help ensure no change in mobility status, but may not have been sufficiently long to prevent the participant from recalling their responses on the first test. This is unlikely, however, given that TCOVS was administered as part of a large survey questionnaire that included many functional measures.¹⁸

For both components of this study, like other similar studies, analysis was based on persons who agreed to participate, who had access to the telephone, and whose status could be considered uncomplicated. Although this is likely to have led to nonresponse bias, it is not clear how this bias would have influenced reliability. This highlights the need to ensure the reliability of TCOVS when used to assess the functional status of a more representative sample of people with SCI, including people who experience difficulties with the telephone interview format, such as people with hearing, visual, literacy, or communication problems or psychologic barriers. In addition, further research for this purpose could include administration of the TCOVS with or without a proxy.

Back to Article Outline

Conclusions 

The findings of this study show that TCOVS, when used by trained raters, offers a feasible and highly reliable method for assessing mobility in the home environment. The TCOVS can be considered, therefore, as a suitable tool to track outcomes for people with SCI across the lifespan.

Suppliers

Back to Article Outline

Acknowledgments 

We thank Jennifer Campbell, BPhty, for expert advice regarding the COVS and to Sarita Schuurs, BPhty, and Brooke Wadsworth, BPhty, for assistance with data collection.

Back to Article Outline

References 

1. Frankel HL, Coll JR, Charlifue SW, et al. Long-term survival in spinal cord injury: a fifty year investigation. Spinal Cord. 1998;36:266–274
   • View In Article
   • MEDLINE
2. Krause JS, Broderick L. A 25-year longitudinal study of the natural course of aging after spinal cord injury. Spinal Cord. 2005;43:349–356
   • View In Article
   • MEDLINE
   • CrossRef
3. Charlifue S, Lammertse DP, Adkins RH. Aging with spinal cord injury: changes in selected health indices and life satisfaction. Arch Phys Med Rehabil. 2004;85:1848–1853
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (89 KB)
   • CrossRef
4. Seaby L, Torrance G. Reliability of a physiotherapy functional assessment used in a rehabilitation setting. Physiother Can. 1989;41:264–271
   • View In Article
5. Low Choy N, Kuys S, Richards M, Isles R. Measurement of functional ability following traumatic brain injury using the Clinical Outcomes Variable Scale: a reliability study. Aust J Physiother. 2002;48:35–39
   • View In Article
   • MEDLINE
6. Przybylski BR, Dumont ED, Watkins ME, Warren SA, Beaulne AP, Lier DA. Outcomes of enhanced physical and occupational therapy service in a nursing home setting. Arch Phys Med Rehabil. 1996;77:554–561
   • View In Article
   • Abstract
   • Full-Text PDF (957 KB)
   • CrossRef
7. Campbell J, Kendall M. Investigating the suitability of the clinical outcome variables scale (COVS) as a mobility outcome measure in spinal cord injury rehabilitation. Physiother Can. 2003;55:135–144
   • View In Article
   • CrossRef
8. Campbell J, Kendall M. In: A move in the right direction: the COVS (Clinical Outcome Variables Scale) as an outcome measure of mobility in spinal cord injury rehabilitation. 2003;Proceedings of the International Spinal Cord Society Australasian Branch Conference; Nov 20-22; Fremantle (Aust).
   • View In Article
9. Kendall MB, Ungerer G, Dorsett P. Bridging the gap: transitional rehabilitation services for people with spinal cord injury. Disabil Rehabil. 2003;25:1008–1015
   • View In Article
   • MEDLINE
   • CrossRef
10. Nunnally JC. Psychometric theory. 2nd ed.. New York: McGraw-Hill; 1978;
    • View In Article
11. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;2:420–428
    • View In Article
    • CrossRef
12. Bland JM, Altman DG. Applying the right statistics: analyses of measurement studies. Ultrasound Obstet Gynecol. 2003;22:85–93
    • View In Article
    • MEDLINE
    • CrossRef
13. Tinsley H, Weiss D. Interrater reliability and agreement of subjective judgements. J Couns Psychol. 1975;22:358
    • View In Article
14. Portney L, Watkins M. Foundations of clinical research. 2nd ed.. Upper Saddle River: Prentice Hall; 2000;
    • View In Article
15. Masedo AI, Hanley M, Jensen MP, Ehde D, Cardenas DD. Reliability and validity of a self-report FIM (FIM-SR) in persons with amputation or spinal cord injury and chronic pain. Am J Phys Med Rehabil. 2005;84:167–176quiz 77-9, 98
    • View In Article
    • MEDLINE
    • CrossRef
16. Shirado O, Kawase M, Minami A, Strax TE. Quantitative evaluation of long sitting in paraplegic patients with spinal cord injury. Arch Phys Med Rehabil. 2004;85:1251–1256
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (164 KB)
    • CrossRef
17. Abramson JH, Abramson ZH. Survey methods in community medicine. 5th ed.. Edinburgh: Churchill Livingstone; 1999;
    • View In Article
18. Shea JB, Morgan RL. Contextual interference effects on the acquisition, retention, and transfer of a motor skill. J Exp Psychol. 1979;5:179–187
    • View In Article

• a Version 13; SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.
• b 2002; Microsoft Corp, One Microsoft Way, Redmond, WA 98052.