Reliability and Validity of the Brazilian Version of the Late-Life Function and Disability Instrument

Design

This quantitative, nonexperimental study was conducted within a more comprehensive study called Maintaining Aging Functions: Older People in the Year 2020 and was approved by the committee on research ethics of the University Hospital of São Paulo University, Brazil. All participants were informed about the research and signed the informed consent statement.

Participants

A convenience sample of 118 older adults was selected. Participants age 60 yr or older who did not require mobility assistance (e.g., help of another person; use of a cane, wheelchair, other mobility aid) and resided in the community were recruited from social, health, and religious institutions located in São Paulo. People younger than age 60 yr or who had communication difficulties, were institutionalized, or did not consent to participate were excluded from the study. All participants were assessed for functional independence in performing activities of daily living using two instruments translated into Brazilian Portuguese: (1) the Barthel Index (Cincura et al., 2009; Mahoney & Barthel, 1965) and (2) the Lawton Instrumental Activities of Daily Living Scale (Lawton & Brody, 1969; Santos & Virtuoso Júnior, 2008).

Data Collection

The study was carried out between March and August 2012. Data were collected through interviews, each conducted in a private setting by a trained examiner who was an occupational therapy practitioner. As part of their training, examiners were instructed in the concepts underlying the instrument and its method of administration. They then conducted pilot interviews with older adults who were not included in the study to give examiners the opportunity to raise any doubts about administering the LLFDI–Br, which the researchers addressed. To evaluate construct validity and internal consistency reliability, the LLFDI–Br was administered to the entire sample. To assess test–retest reliability, the LLFDI–Br was readministered by the same examiner 21 days later to a subsample who agreed to participate.

Instrument

The LLFDI–Br consists of two components: Function and Disability. The Function component measures the ability to perform specific activities that require gross and fine motor skills. This component comprises 32 items that assess three domains: (1) Upper Extremity Function (activities that use the hands and arms; 7 items), (2) Basic Lower Extremity Function (activities that mainly involve standing, bending over, and walking; 14 items), and (3) Advanced Lower Extremity Function (activities with a high level of physical ability and resistance; 11 items; Haley et al., 2002).

The Disability component assesses major life activities. It comprises 16 items and assesses two dimensions: Frequency describes how regularly the respondent participates in each life task, and Limitation addresses the respondent’s ability to perform each life task. Frequency consists of two domains: Social Role (social and community tasks; 9 items) and Personal Role (personal tasks; 7 items). Limitation consists of two domains: Instrumental Role (activities at home and in the community; 12 items) and Management Role (organization or management of social tasks; 4 items).

The questionnaire items are administered through interview, and responses are selected from a 5-point Likert-type scale. In the Function component—for example, “How much difficulty do you have [doing a particular activity]?”—response options include none, a little, some, quite a lot, and cannot do. In the Disability component, response options for Frequency questions—for example, “How often do you do a particular task?”—include very often, often, once in a while, almost never, and never. Response options for Limitation questions—for example, “To what extent do you feel limited in [doing a particular task]?”—include not at all, a little, somewhat, a lot, and completely. Raw scores for all items are transformed to scaled scores from 0 to 100 (Jette, Haley, & Kooyoomjian, 2002, 2003).

Data Analysis

Using the scores for the total sample, descriptive statistical values for the domains of both components of the LLFDI–Br were calculated. Test–retest reliability was calculated using the intraclass correlation coefficient (ICC; Fleiss, 1986) with 95% confidence interval. ICCs ≥.75 indicate excellent, .40–.74 adequate, and <.40 poor reliability (Fleiss, 1986).

Cronbach’s α was calculated to analyze the internal consistency reliability of each domain of the LLFDI–Br. Values ≥.70 indicate excellent; .51–.69 adequate; and <.50 poor internal consistency (Fornell & Larcker, 1981).

To assess the construct validity of the LLFDI–Br, we used factor analysis, a mathematical technique that allows grouping of variables that are highly correlated into a common dimension, or factor (Johnson & Wichern, 2007). Varimax rotation was used to confirm the factors identified.

Sample Characteristics

The study sample comprised 94 (79.7%) women and 24 (20.3%) men ages 60–90 yr (mean [M] age = 71.4 ± 6.9). Three (2.5%) had never attended school, 35 (29.7%) had 1–4 yr of study, 21 (17.8%) had 5–8 yr, 32 (27.1%) had 9–11 yr, and 27 (22.9%) had >12 yr of schooling. The mean Barthel Index score was 99.10 (±3.03), and the mean Lawton Instrumental Activities of Daily Living Scale score was 19.54 (±1.90). Table 1 shows the sample’s mean LLFDI–Br scores by domain.

Test–Retest Reliability

Test–retest reliability was calculated using a subsample of 14 participants. Of this subsample, 12 (85.7%) were female and 2 (14.3%) were male, their age range was 60–90 (M = 68.5 ± 5.8), and 10 (71.4%) had 1–4 yr of schooling and 4 (28.6%) had 5–8 yr of study; the characteristics of the test–retest subsample thus were similar to those of the entire sample. Similar sample sizes were used in calculating the test–retest reliability of the two components of the original English instrument (Haley et al., 2002; Jette, Haley, Coster, et al., 2002).

Test–retest reliability was excellent for the three domains of the Function component of the LLFDI–Br, with ICCs of .95 for Upper Extremity Function, .93 for Basic Lower Extremity Function, and .78 for Advanced Lower Extremity Function. In the Disability component, test–retest reliability was excellent for Social Role (.81), adequate for Personal Role (.53) and Instrumental Role (.59), and poor for Management Role (.27).

Table 2 shows the test–retest reliabilities of the different versions of the LLFDI: the original English (Jette, Haley, & Kooyoomjian, 2002), Hebrew (Melzer et al., 2007), French Canadian (Sakakibara et al., 2013), and Swedish (Roaldsen et al., 2014). The LLFDI–Br showed moderate to excellent test–retest reliability in every domain except Management Role; poor reliability was also observed in the Management Role domain for the original English (Jette, Haley, & Kooyoomjian, 2002) and Hebrew (Melzer et al., 2007) versions.

Internal Consistency Reliability

Internal consistency was excellent for the overall Function component (α = .948) and for its three domains: Upper Extremity Function (α = .819), Basic Lower Extremity Function (α = .889), and Advanced Lower Extremity Function (α = .914). For the Disability component overall, internal consistency was excellent (α = .889). Internal consistency of Social Role (α = .807) and Instrumental Role (α = .856) domains was excellent and of Personal Role (α = .601) and Management Role (α = .611) domains was adequate. As shown in Table 2, the α values for the LLFDI–Br were similar to those of the original English version (Jette, Haley, & Kooyoomjian, 2002), with the exception of Personal Role.

Construct Validity

For the Function component, the three domains were used in the factor analysis: Upper Extremity Function, Basic Lower Extremity Function, and Advanced Lower Extremity Function. These factors explain 53.5% of the total variability. Table 3 shows the factor loadings (with varimax rotation) for the Function component. Factor loadings for the Upper Extremity Function and Advanced Lower Extremity Function domains are similar to those obtained in the original analysis; however, the Basic Lower Extremity Function domain shows differences. Only Items F11 and F26 show high factor loadings in the original domain. Some items had high loadings in the original domain and in other domains (Items F15, F18, and F25 of Basic Lower Extremity Function; Item F3 of Upper Extremity Function; and Item F27 of Advanced Lower Extremity Function). Other items loaded significantly higher in other domains compared with the original domain (Items F2, F10, F12, F14, F21, F22, F23, F28, and F31 of Basic Lower Extremity Function and Item F13 of Upper Extremity Function).

For the Disability component, the four domains were used in the factor analysis: Social Role and Personal Role in the Frequency dimension and Instrumental Role and Management Role in the Limitation dimension. These factors explain 46.6% of the total variance. Table 4 shows the factor loadings for the Disability component. The factor loadings were similar to those obtained in the original analysis, with some exceptions. Some items had high loadings in the original domain and in other domains (Items D1F, D3F, D6F, and D11F of Social Role and Item D15L of Instrumental Role), whereas other items loaded significantly higher in other domains compared with the original domain (Item D10F of Personal Role, Items D13L and D16L of Instrumental Role, and Item D7L of Management Role).