Development of the Taiwanese version of the Health Enhancement Lifestyle Profile (HELP-T)

Objectives To develop and validate a Taiwanese version of the Health Enhancement Lifestyle Profile (HELP-T) for community-dwelling older Taiwanese adults (≥ 55 years). Methods The original Health Enhancement Lifestyle Profile (HELP) is a 56-item self-report questionnaire measuring various aspects of health-related lifestyles in older adults. The standard cultural-adaptation procedure was used for questionnaire translation and modification. A field test was conducted for culturally specific item selection, rating-scale analysis, and psychometric validation of the HELP-T in a sample of 274 community-dwelling older adults via classical test theory. Results The 59-item HELP-T is culturally adapted from the original 56-item HELP. The original 6-point rating scale was modified to a 3-point scale for easy use by Taiwanese older adults. The HELP-T had good internal consistency (Cronbach’s alpha = 0.82). The test-retest reliability for the total score was high (0.92), and moderate to high (range: 0.57–0.92) for subscales. The construct validity was supported by the significant correlations between each subscale and the total score (Spearman’s rho = 0.41–0.67, p < 0.0001) and by the ability of the scores to significantly discriminate between participants with different levels of self-rated health (p = 0.0001). Conclusions The HELP-T is a suitable clinical tool for assessing and monitoring lifestyle risk factors, establishing client-centered lifestyle intervention goals, and determining the outcomes of lifestyle interventions.


Introduction
This study aimed to develop a Taiwanese version of the HELP (HELP-T). For proper crosscultural use, standard procedures were adopted to linguistically and culturally adapt the HELP, and CTT was used to determine the appropriateness of the rating scales and to confirm the reliability and validity of the HELP-T.

Instrument
The HELP has two major sections: (1) personal and health information: demographics, diagnoses and self-rated health (i.e., excellent, good, fair, or poor), and (2) seven subscales aforementioned that measure different aspects of a health-related lifestyle. Each subscale contains eight items that ask how often a person engaged s in various health-promoting or risky behaviors during the previous 3 months; a 6-point rating scale is used: never (score 0), 1-3 days/ month (score 1), 1-2 days/week (score 2), 3-4 days/week (score 3), 5-6 days/week (score 4), or 7 days/week (score 5). For each HELP subscale, a total score between 0 and 40 can be computed: a higher score means a higher frequency of health-promoting behavior [15].

Study phases
The study was conducted in two phases: (1) generating the preliminary HELP-T, and (2) evaluating the appropriateness of the rating scale and determining the intrument's reliability and validity. National Taiwan University Hospital's Institutional Review Board approved the study (201203041RIC). Written informed consent was obtained from all participants.
Phase one: Generating the preliminary HELP-T. A series of procedures were adopted to generate a preliminary version of the HELP-T.
Translating and culturally adapting the HELP The HELP was translated using a forward and backward translation procedure [18]. A review committee consisting of 6 experts including the research team (authors), 4 occupational therapists, one nurse, and one physical therapist, all of whom are experienced with geriatric care resolved wording discrepancies and determined conceptual and semantic equivalence between the two versions. Finally, the author of the original HELP, who was proficient in both languages, approved the two versions.
Several items from the original HELP were modified for cultural appropriateness. First, items with activities or objects with which older Taiwanese adults are usually not familiar were modified. For example, we replaced canned soup, hot dogs, bacon, sausage with local foods such as pickled cucumber, fermented bean curd, and kimchi in a Diet item. Second, some items were added with more activity examples that are culturally relevant; for example, "mahjong" was added to a Leisure item and Asian martial arts to an Exercise item. Moreover, because negatively worded questions are not commonly used in Mandarin, we rephrased them accordingly; for example, "How often during a week do you tend to ignore the routine for grooming and personal hygiene?" was revised to "How often during a week do you perform grooming and personal hygiene?" Creating additional culturally specific items We conducted three focus groups to gather information about health-related lifestyles from different perspectives. This procedure aimed to explore addtional culturally specific items that were not included in the original HELP. The first focus group included eight healthcare professionals (two occupational therapists, a physical therapist, a physician, a nurse, a dietician, a social worker, and a public health policy maker) specialized in geriatric care. The other two focus groups were separately conducted in southern and northern Taiwan, each with eight community-dwelling older adults of different ages and sex. Members of the focus groups discussed their definitions, experiences and perceptions regarding healthy and unhealthy activities of daily living. The same review committee analyzed the minutes of the focus groups and suggested eight culturally specific items (one for the Exercise subscale, four for the Social and Productive subscale, one for the Leisure subscale, and one original item [separated into two items] for the Leisure subscale) (see Results).
Examining culturally specific items We used various criteria to determine the psychometric properties of the new items: (a) items with a mean between 1 and 4 (to prevent a floor or ceiling effect), (b) items with a median between 1 and 4, (c) correlation to the domain: r > 0.4, (d) a within-the-domain itemdeleted reliability < 0.7, and (e) a significant difference in the mean scores between highest and lowest 1/3 groups [19]. The preliminary version of the HELP-T included the 56 original items and those culturally specific items that met the 5 criteria above (see Results).
Phase 2: Evaluating the appropriateness of the HELP-T's rating scale, and of its reliability and validity. Participants We enrolled 274 community-dwelling older adults (age > 55) who were cognitively intact and able to communicate in Mandarin or Taiwanese. Convenience and snowball sampling methods were used to recruit participants from a variety of diverse community sites across different regions of Taiwan.

Data collection procedures
The preliminary version of the HELP-T was administered through on-site paper-and-pen questionnaires in groups or face-to-face interviews by the first author. About 20 to 40 minutes were needed to complete the HELP-T.

Data analysis
Negatively conceptualized items were reverse coded for scoring. Kolmogorov-Smirnov tests and distribution plots were used to examine the normally hypothesis of the HELP-T total score and subscales. Therefore, parametric and non-parametric statistics were used in the following analyses, respectively.
Examining and modifying the rating scale Many participants who completed the HELP-T through a face-to-face interview commented that the 6-point scale was too detailed and that they had difficulty in choosing their answers. Some response categories were rarely selected (< 10% of the participants). For more than half the items, participants used only 3 or fewer response categories. Therefore, we collapsed the 6-point scale into a 3-point scale by combining adjacent categories. As a result, a new 3-point rating scale was proposed: never or 1-3 days/month (score = 0), 1-4 days/week (score = 1), and 5-7 days/week (score = 2). Because the correlation coefficients between the scores of two rating scales was 0.985, we used the new 3-point rating scale for the subsequent analyses.

Examining reliability and validity
Cronbach's alpha (α) was used to determine the internal consistency for the total score of the HELP-T. An α of at least 0.80 was considered good [20]. For test-retest reliability, 28 participants completed the HELP-T twice within an interval of 11-14 days. The intraclass correlation coefficient (ICC) with a one-way random effects model (1,1) [19,21] was used to determine test-retest reliability of the HELP-T total score and subscale scores. An ICC of at least 0.75 was considered high, between 0.75 and 0.40 was considered moderate, and less than 0.40 was considered low [19].
The construct validity of the HELP-T was examined using hypothesis testing and discriminant validity. The hypothesis testing method evaluated the correlations of scores from the seven subscales and the total HELP-T score. We hypothesized that there would be significantly moderate-to-high correlations across the subscale scores and the total score, and that there would be small-to-moderate correlations across the seven subscales. Because the normality of the subscales was not assumed, Spearman's correlation coefficient (rho [ρ]) was used for this hypothesis testing. Discriminant validity was used to test the HELP-T scores from participants who rated their health as excellent or good, and those who rated their health as fair or poor. An independent t test or Mann-Whitney U test (if normality was not assumed) was used to compare scores between the two groups. SPSS 20 (IBM Corp., Somers, NY, USA) was used for all statistical analyses. Significance was set at p < 0.05. For multiple testing, significance was adjusted using the Bonferroni correction.

Selecting culturally specific items
Among the eight culturally specific items suggested, four met all criteria and were selected: "gather with family members you don't live with", and "contact family members you don't live with" in the Social and Productive Activities subscale, and "do art and music activities, play musical instruments, or sing (karaoke)" and "do gardening, planting, or crafts" (split from one original item) in the Leisure subscale (Table 2). Therefore, the final version of the HELP-T includes 59 items: 10 in Social and Productive Activities, 9 in Leisure, and 8 in the other 5 subscales.

Descriptive data
The mean ± SD total score of the 59-item HELP-T (3-point rating scale) was 53.59 ± 11.41 (Table 3). We divided the mean of each subtotal score by the number of items in each subscale. The mean total scores of the Activities of Daily Living (1.45 ± 0.63), Diet (1.44 ± 0.67), and Other Health Promotion and Risk Behaviors (1.29 ± 0.67) subscales were right-skewed; thus, participants did them more than "1-3 days/month" but less than "5-7 days/week". The Exercise (0.59 ± 0.67), Leisure (0.58 ± 0.64), and Social (0.41 ± 0.59) subscales had the lowest mean total scores. They were left-skewed; thus, participants did them less than "1-4 days/week".

Reliability
The Cronbach's α of the HELP-T total score was 0.82, which indicates good internal consistency. The test-retest reliability (ICC [95% CI]) of the HELP-T total scores was 0.92 (0.83-0.96), which indicated excellent score agreement.  (Table 3). Most items in the HELP-T showed moderate-to-good individual item agreement.

Validity
Most items reached an acceptable level of validity and were significantly correlated (p < 0.05) ( Table 3). The construct validity of the HELP-T was supported by the significant correlation  rated their health as excellent or good and of those who rated their health as fair or poor (Table 5). These findings supported the discriminant validity of the HELP-T: the HELP-T distinguished between older adults with different self-rated levels of health.

Discussion
The 59-item HELP-T is the first health-related lifestyle assessment designed specifically for older Taiwanese adults. The HELP-T was culturally adapted from the original 56-item English version by adding two culturally specific items about family activities and by splitting one Leisure item into two. The original 6-point rating scale was modified to a 3-point scale to make it easier for older Taiwanese adults to use. The HELP-T total score had good internal consistency, and most HELP-T domains showed acceptable-to-good test-retest reliability and good construct validitythe HELP-T is suitable for measuring various aspects of lifestyle factors and behaviors in older Taiwanese adults. Health is generally conceptualized in some dimensions that are universal across cultures, but other dimensions vary by culture [22][23][24]. We added two items about family activities  [25] reported that 82.8% of older Taiwanese adults get together with their children at least once a week. Our study confirmed that health-related lifestyle activity profiles completed by older Taiwanese adults reflect their social and cultural values. Although the original 6-point rating scale might make the HELP more sensitive by reflecting small incremental lifestyle changes, we used a 3-point scale in the HELP-T for three reasons. First, our statistical analysis indicated that the scores from the two rating scales were highly correlated. Second, too many choices can compromise a person's decisiveness, especially for the elderly and those with a low level of formal education [26]. Compared with the original HELP study [15], in which 91.2% of the participants in the U.S. had completed secondary or higher education, only 66.1% of our participants had. Third, because of potential variations in the lifestyle context within the 3-month survey time-frame (e.g., holidays and other special events), some participants commented that they could recall only the approximate frequency for each of the activities and behaviors included in the HELP-T and that 6 choices made the questions difficult to answer. The internal consistency of the HELP-T was acceptable. The test-retest reliability was good overall, and it was fair-to-good for all subscales, but "Diet" and "Other Health Promotion and Risk Behaviors", for which it was unsatisfactory. We found that many items in these subscales depended upon changes in social and temporal contexts. For example, within a week, one may have several social events involving eating too much or eating unhealthy food and drinking too much alcohol (e.g., banquets, parties, rituals). Similarly, it is common for older Taiwanese adults to take over-the-counter medications for mild and brief symptoms of illness (e.g., pain, cold). These behaviors might have jeopardized our test-retest results.
The construct validity of the HELP-T was first supported by the interrelationships between the different lifestyle behaviors subscales. The significant low-to-moderate correlations between the 7 subscales indicate that each subscale contributes a somewhat related but distinctive aspect to the measure of a healthy lifestyle. Clinically, the scores from different HELP-T subscales can help identify areas of strength and weakness in a person's lifestyle. Thus, service planning can be individualized to meet each older adult's personal needs.
The construct (discriminant) validity of the HELP-T was also supported by the ability of the scores to distinguish between participants who perceived themselves to be in good health and those who did not. Lifestyle behaviors are reported [27] to be responsible for at least 50% of how healthy one is. Others have reported that older adults who perceive their health as poor are less likely to exercise [28] and perform self-care [29], and that they are more likely to engage in risky behaviors, such as smoking, heavy alcohol drinking, and poor eating habits [30]. Our results echo these previous findings [16].
It is noteworthy that, although the original HELP and the HELP-T consist of 7 subscales that yield subtotal scores, a healthy lifestyle does not entail a high score for every subscale. Individuals must prioritize their own needs to develop personal plans that allow them to achieve a balanced, healthy lifestyle. We recently also developed a HELP-T Intervention Plan Form along with a Clinician Guide [31], in which an individual client and the clinician are instructed to establish their goals for change and to identify their targeted HELP-T activities to achieve the goals.
The generalizability of our findings is limited, however, because we enrolled only a small nonrandom sample from Taiwan. Future studies should include larger and more representative random samples of older adults in Taiwan.

Conclusions
This study adapted the HELP for cross-cultural use with older Taiwanese adults. We modified both the content and the rating scale to make HELP-T suitable for older Taiwanese adults. The HELP-T is a valid and useful tool that enables clinicians to understand the health-promoting habits and routines of older Taiwanese adults, helps them establish goals for lifestyle change, and yields client-centered lifestyle monitoring and recommendations.