Environmental Barriers Experienced by Amputees: The Craig Hospital Inventory of Environmental Factors–Short Form

Study Design and Sample 

The study was a cross-sectional survey design that we conducted as part of a larger project, the Limb Loss Research and Statistics Program (LLRSP). The LLRSP is a partnership between researchers at the Johns Hopkins University Bloomberg School of Public Health and the Amputee Coalition of America (ACA), with a mission to advance the understanding of the epidemiology and consequences of limb loss. Using an existing database containing information on persons who contacted the ACA between 1998 and 2000, we identified 6500 amputees. Information contained in the database included etiology and level of amputation, demographic information (date of birth, and sex), and mailing address and telephone number. A sample of 1538 amputees was identified. The sample was comprised of 608 persons with amputation secondary to peripheral vascular disease, including those with diabetes mellitus; 579 persons with traumatic amputation; and 351 persons with amputation due to malignancy of a limb or joint.

Eligibility criteria included age 18 to 84 years, amputation of either an upper- or lower-limb or bilateral amputation of upper or lower limbs, and persons who spoke English. We conducted a proxy interview if the interviewer determined that the subject was unable to complete an interview due to a mental or physical impairment. The study was reviewed and approved by the Committee on Human Research at the Johns Hopkins Bloomberg School of Public Health.

Survey Procedure 

Subjects were mailed a letter from the ACA that outlined the purpose and objectives of the survey and contained a contact sheet for persons interested in participating in the study to return to the study center. Two weeks after the letter was mailed, if the subject had not returned a contact sheet, trained interviewers began attempts to contact the subject by telephone. Half of all subjects returned a contact sheet. Tracing procedures (eg, telephone directories, internet directory search engines) were employed in the case of returned letters with no forwarding address and nonworking telephone numbers. Prior to the interview, oral informed consent was obtained from the subjects by the interviewer. Interviews were conducted between January and August 2001.

Measures 

The survey was comprised of measures of general health and well-being, function, use and satisfaction of prosthesis, environmental barriers, and unmet medical needs. While previous analysis has focused on use of prosthesis, depressed mood, and pain,8, 9, 10 our primary outcome measure for this analysis was perceived environmental barriers. This variable was measured using the CHIEF-SF, a survey instrument designed to measure subjects’ perceptions of the frequency and magnitude of barriers in their environment. The CHIEF-SF has been found to be both valid and reliable in differentiating the perceived barriers of disabled and nondisabled persons and among persons with varying impairments.11 The instrument was developed to operationalize 5 conceptual characteristics of the environment—accessibility, accommodation, resource availability, social support, and equality—and to be applicable across disability groups without regard to diagnostic categories. The CHIEF-SF has a 12-item scale. The items are scored on their frequency from 0 to 4 (0, never; 1, less than monthly; 2, monthly; 3, weekly; 4, daily) and magnitude from 0 to 2 whether the barrier is perceived as “no problem,” “a little problem,” or “a big problem” (1, little problem, 2, big problem). A frequency by magnitude product score ranging from 0 to 8 is derived to measure the overall impact of the barrier. Higher scores indicate a greater impact of environmental barriers. The 12 items are grouped into 5 subscales or domains, (1) policies, (2) physical/structural, (3) work/school, (4) attitudes/support, and (5) services/assistance. The subscale score is comprised of an average of the mean frequency-magnitude score for each item in the subscale. For the work/school subscale, only subjects currently working or in school were eligible to answer these questions. Subscale scores are not computed for respondents who did not answer all questions contained within each subscale domain.

Additional measures included in this analysis were demographic information (sex, current age, race, ethnicity), etiology of amputation, years since amputation, number of hours of prosthesis wear in an average day, amputation level, and education level (less than high school, high school graduate or general equivalency degree [GED], some college or college graduate). Community type (rural vs urban) was computed using subject zip code and area defined by the Multiple Statistical Area Urban Influence Urban Continuum codes. Residential region of the United States (Northeast, Midwest, South, West) was computed using state of residence in accordance with those defined by the U.S. Census. Information collected on household size and family income was used to generate poverty status (poor, near poor, not poor) according to the U.S. Census definition.12

Data Analysis 

Our primary analysis used mean frequency-magnitude subscale scores of the CHIEF-SF. However, because our data were skewed rather than normally distributed, we calculated a binary outcome for each subscale by establishing a cutoff point equal to a mean subscale score of 3 or higher, indicating barriers that occur at least monthly and are perceived to be a big problem or barriers that occur at least weekly and are perceived to be a little or big problem.

As part of the analysis we compared amputee responses to that of a general population sample by using previously published normative CHIEF-SF data from the 1999 Colorado Behavior Risk Factor Surveillance System (BRFSS) that we obtained from researchers at Craig Hospital of Denver, CO.11 The BRFSS is a national telephone survey designed to measure state-level prevalence of major behavioral risk factors that are associated with premature morbidity and mortality.13 The 1999 Colorado BRFSS data set is comprised of 2259 telephone interviews conducted among a random sample of Colorado residents aged 18 years and older. In addition to a core set of questions on health behaviors, the 1999 version of the survey included the CHIEF-SF. To identify persons with a disability within the sample, a series of questions were asked that assessed whether the individual experienced limitations in work, limitations in activities due to health or impairment, difficulties with memory or learning, and/or use of special equipment to ambulate. Finally, comparisons were made with a sample of 124 persons with SCI and 120 persons with TBI obtained from Craig Hospital. These subjects were initially recruited to test the CHIEF-SF psychometric properties among specific groups with disability. A binary outcome was computed using a cutoff of frequency-magnitude product score of greater than or equal to 3 for all 5 subscales as well as the overall total score for the data from the BRFSS and Craig Hospital sample.

Chi-square analysis was performed to compare persons with and without significant subscale scores (frequency-magnitude score ≥3). Bivariate analysis was conducted using simple logistic regression. We used multivariable logistic regression to control simultaneously for potential confounding factors to include sex, current age (in years), etiology, time since (in years) and level of amputation, prosthetic use (in hours/day), number of comorbid conditions, education level, household poverty status, and the presence of stump pain, phantom pain, and back pain. Odds ratios (ORs) with 95% confidence intervals (CIs) are reported. All analysis was performed using Stata.a Statistical significance was determined at the α equal to .05.

Mean Frequency-Magnitude Scores 

Mean frequency-magnitude scores ± standard deviation ranged from a high of 1.67±1.84 for the physical/structural subscale to a low of 0.68±1.29 for the work/school subscale (table 2). Analysis of frequency-magnitude subscale scores by etiology demonstrated that, across all subscales except the services/assistance subscale, the mean subscale scores were significantly lower for cancer amputees than traumatic amputees. Although mean subscale scores reported by cancer amputees were consistently lower than those of dysvascular amputees across the board, only mean attitudes/support and services/assistance scores proved to be significantly lower.

Table 2. Mean Product Scores for Survey Participants and Percentage With Mean Subscale Score of 3 or Higher by Etiology for the CHIEF-SF

	Reported Barriers	Total		Dysvascular⁎		Trauma		Cancer		F†
		Mean ± SD	Percentage ≥3	Mean ± SD	Percentage ≥3	Mean ± SD	Percentage ≥3	Mean ± SD	Percentage ≥3
	Policies	1.11±1.84	15.0	1.16±1.77	15.9	1.24±2.05	17.8	0.82±1.55	8.9	.03
	Policies: businesses	1.04±2.01		1.21±2.06		1.09±2.12		0.70±1.72
	Policies: government	1.21±2.29		1.15±2.17		1.42±2.57		0.95±1.89
	Physical/structural	1.67±1.81	23.1	1.65±1.80	21.0	1.82±1.90	29.4	1.44±1.67	15.8	.04
	Surroundings	1.21±1.88		1.26±1.87		1.36±2.03		0.88±1.62
	Natural environment	2.15±2.36		2.03±2.28		2.32±2.53		2.00±2.16
	Work/school	0.68±1.29	12.7	0.63±1.22	7.7	0.88±1.47	10.5	0.43±0.97	3.9	.01
	Attitudes	0.76±1.70		0.70±1.63		0.96±1.90		0.50±1.41
	Help	0.62±1.48		0.54±1.36		0.82±1.75		0.36±0.99
	Attitudes/support	0.91±1.48	0.1	0.95±1.47	14.7	1.06±1.64	15.1	0.56±1.11	5.3	.00
	Attitudes home	0.67±1.67		0.79±1.81		0.73±1.75		0.38±1.18
	Discrimination	1.15±2.02		1.13±2.00		1.39±2.26		0.74±1.42
	Services/assistance	0.99±1.39	10.0	1.12±1.44	13.2	0.99±1.39	9.8	0.79±1.30	7.8	.02
	Transportation	0.85±1.92		1.23±2.21		0.71±1.78		0.50±1.48
	Medical care	1.15±2.13		1.22±2.03		1.23±2.23		1.10±2.16
	Help home	1.05±1.91		1.17±2.02		1.10±1.95		0.82±1.68
	Information	0.85±1.76		0.93±1.83		0.87±1.81		0.74±1.58
	CHIEF-SF total	1.12±1.27	11.2	1.19±1.27		1.22±1.33		0.85±1.13		.00

⁎ Includes persons with diabetes mellitus. † Analysis of variance across etiologies to compare mean subscale scores.

Comparison to General Population: Colorado BRFSS and TBI and SCI Sample 

Overall, 87% of amputees surveyed reported perceived barriers in 1 or more areas with the majority (54%) reporting barriers in 4 or more areas. Persistent barriers (frequency-magnitude product score ≥3) were reported by 62% of amputees in 1 or more environment domains.

When compared with the nondisabled population from the 1999 Colorado BRFSS, amputees were significantly more likely to perceive persistent barriers (mean frequency-magnitude score ≥3) in all areas except work/school barriers (fig 1). Similarly, amputee perception of persistent barriers was greater than the general disabled population from the BRFSS with the exception of work/school and services/assistance barriers. When compared with the sample of persons with SCI from Craig Hospital, a different pattern emerged. A greater proportion of persons with SCI were significantly more likely than persons with amputation to perceive persistent barriers (frequency-magnitude score 3) overall and in the areas of policies and physical/structural. When we compared the percentage of amputees with mean frequency-magnitude scores 3 or more with the group with TBI, however, there were no significant differences between the groups.

View Large Image Download to PowerPoint

Fig 1. Percentage with mean CHIEF-SF subscale score of 3 or greater. A comparison of persons with amputation and nondisabled and disabled respondents from the Colorado 1999 BRFSS, and sample of persons with TBI and SCI from Craig Hospital, Denver, CO. Data source: Whiteneck et al.11

Predictors of Significant Barriers 

After adjusting for covariates in the extended model, women were less likely to perceive barriers in the physical/structural environment (OR=0.7; 95% CI, 0.5–0.9) than men. Age was a significant factor, with amputees aged 55 to 64 years less likely to perceive barriers than those aged 18 to 44 years in the areas of physical/structural environment (OR=0.5; 95% CI, 0.3–0.9), attitudes/support (OR=0.4; 95% CI, 0.2–0.8), and services/attitudes (OR=0.4; 95% CI, 0.2–0.8). The same was true for amputees aged 65 years and older in 2 of those areas, attitudes/support (OR=0.3; 95% CI, 0.1–0.8) and services/assistance (OR=0.4; 95% CI, 0.1–0.9). The number of hours of prosthesis wear each day was also a significant predictor, with amputees who wear a prosthesis 9 or more hours a day being 50% less likely to perceive barriers in the area of attitudes/support (95% CI, 0.3–0.8) and 70% less likely to perceive barriers in the area of services/assistance (95% CI, 0.1–0.8) than those amputees who did not have or wear a prosthesis. With the exception of the work/school environment, amputees who reported having 2 or more comorbid conditions were 2 to 3 times more likely to perceive barriers than those who reported no comorbid conditions (policy: OR=2.1; 95% CI, 1.0–4.5; physical/structural: OR=2.3; 95% CI, 1.3–4.2; attitudes/support: OR=3.6; 95% CI, 1.6–8.2; services/assistance: OR=3.0; 95% CI, 1.1–8.0). Likewise, poverty was a significant predictor of increased perception of barriers in the environment, with the exception of work/school, with amputees with a household income at or near the poverty line being 2 to 3.5 times more likely to perceive barriers than those that were not poor (policies: OR=2.6; 95% CI, 1.4–4.8; physical/structural: OR=2.1; 95% CI, 1.2–3.6; attitudes/support: OR=2.1; 95% CI, 1.3–3.4; services/assistance: OR=3.5; 95% CI, 1.9–6.7). Amputees with stump pain were nearly 2 times more likely to perceive barriers in the physical/structural environment than those who reported no pain (95% CI, 1.2–2.8). The same was true of amputees with back pain (OR=1.7; 95% CI, 1.2–2.6).