Adaptation and validation of the Instrumental Expressive Social Support Scale in Portuguese older individuals*

Objective to adapt and validate the Instrumental Expressive Social Support Scale (IESS) in a sample of older people. Method methodological study. The sample of 964 community-dwelling older people was randomly divided into two groups. The first group was used as a calibration sample to study the number of factors underlying social support through Principal Axis Factoring, and the second group as a validation sample to test the “best fit” model through Confirmatory Factor Analysis. Results exploratory Factor Analysis suggested a three-factor solution, which was confirmed by Confirmatory Factor Analysis. The factors were similar to those in the pre-existing dimensions of the original instrument and were named as Sense of control (α = 0.900), Financial support (α = 0.802), Familiar and socio-affective support (α = 0.778). Confirmatory Factor Analysis showed acceptable fit. The model’s goodness-of-fit indexes were satisfactory (χ2/df = 5.418; CFI = 0.903; NFI = 0.884; RMSEA = 0.098). The convergent validity was supported by associations between social support and medication adherence and positive affect. The discriminant validity was evidenced by association with negative affect. The reliability analysis showed high values of internal consistency. Conclusion the instrument proved to be a valid measure for the assessment of social support in older people.

economic status, and the influence of each variable is often complex and they usually interact with other factors. A higher perceived support is associated with being a women or living with a partner (3,10) . A higher level of education was also found to be associated with positive social support (10) . Age is also determinant and the old-old were also found to report lower levels of social support from friends when compared to the young-old (11) .
The association between social support and health outcomes is well documented. Social support is important not only for promoting better mental health, but also for a good physical health, reducing mortality by 50%, independently of age, gender and other health conditions (12) . Low perceived social support was found to be associated with poor self-rated health in older women (13) .
In its turn, the high support from family, friends and social groups are important predictors of disease outcomes, both in Hispanic and Caucasian samples (14) .
The association between social support and subjective well-being in older age has also been established (15)(16)(17)(18) . Positive and negative affects are two of the three components of subjective well-being (SWB), which also includes life satisfaction. Existing evidence shows that social support is positively correlated with positive affect and inversely correlated with negative affect (16) . For example, in a study with older persons (mean age of 73) found that social support was associated with positive affect (19) . This same result was found in a study conducted in Australia, showing a positive association between social support and positive affect (20) .
A strong positive association was also reported between life satisfaction and social support in a study involving a sample of community-dwelling older adults (19) .
A strong association was found between lack of social support and psychological distress in homedwelling older adults (21) . Depression is relatively common in the elderly, and social support can act as a buffer, protecting them from negative affect (22)(23) .
When comparing the association between age and social support in different age groups, stronger associations with well-being were evidenced in older adults (23) .
A previous research has already established the association between social support and patient medication adherence, namely in old and chronically ill persons (24) . A former study has demonstrated that social support influences diabetes medication adherence and non-pharmacological treatment (25) . The relationship Lima L, Santos C, Bastos C, Guerra M, Martins MM, Costa P.
between social support and medication adherence is particularly significant in older people. In this age group, most people suffer from multiple chronic illnesses (e.g. hypertension, cholesterol, diabetes) and need to take several medications. Older adults are also the largest users of prescribed medication (26) .
Lack of social support of home-dwelling elderly persons was also suggested as contributing to medication nonadherence, and the prevalence of nonadherence was shown to be higher in individuals who lived in their own houses (27) . Additionally, other studies stress that the problem of medication nonadherence is increasingly high in those persons living alone in their own houses, with little support from family or friends (12) .
The assessment of social support needs to be carefully considered, depending on the type of research, as well as on the characteristics of the population under study. Most importantly, when addressing older adults, the instruments in use must focus on distinctive aspects of this age group, namely on their social roles, relationships and psychological development. More specifically, and considering that the autonomy of the elderly persons is usually replaced by increased dependency on their close relatives and friends, it is fundamental that the instrument clearly captures the affective dimension in perceived social support. Moreover, due to the reduced functionality and independence, it is also important to assess the way old adults perceive social support. This perceived social support is characterized either by attitudes of respect towards the autonomy of the dependent person or, in contrast, social support is perceived as a form of excessive control and lack of sense of empowerment, because "the perception of personal control plays a critical role in the health and well-being of an older person" (28) . Finally, the economic dimension is also important, since Portugal is a country in which older people are an economically deprived/ vulnerable group, The Instrumental Expressive Social Support Scale (IESS) meets all these demands since it includes items that measure all these aspects of the perceived social support (29) .
The IESS scale was previously adapted to the Portuguese population and the results evidenced good psychometric properties (29) . Reliability was assessed through internal consistency and the Cronbach's alpha was 0.83 for the total scale. Exploratory factor analysis indicated six factors accounting for 62.1% of the variance.
The three factors that explained most of the variance observed were: Factor 1 -socio-affective support; Factor 2 -sense of control and Factor 3 -financial support.
The IESS has also been used in a study with cardiac patients, in which a moderate negative correlation was found between social support and perceived stress (30) . The instrument was also used in a sample of patients with vertebra-medullar lesion and a negative association was found between social support and depression (31) .
The aim of this study was to validate the Portuguese version of the Instrumental Expressive Social Support Scale (29) in older adults.

Method
In this cross-sectional and observational study, a non-probabilistic and convenience sampling technique was used, whose subjects were recruited as part Primary school (4 years) educational level was found in 70% of the sample (n=696).
For validation purposes, the total sample was randomly divided into two different samples (EFA and CFA). An overview of the characteristics of the study participants is presented in Table 1.
Several instruments were used. The Instrumental Expressive Social Support Scale has been previously adapted to Portuguese (29) . The IESS scale is a multidimensional measure of social support that includes 20 items grouped into three dimensions. A 5-point Likert scale was used to determine the frequency by which participants were bothered with the described issues in the last 6 months (1 -"always or almost always"; 2 -"many times"; 3 -"sometimes"; 4 -"rarely"; and 5 -"never". The total score is calculated by the sum of the items scores and may vary between 20 and 100, with a higher total score reflecting a better perception of social support and absence of presented problems. The Reported Adherence to Medication (RAM) Scale Portuguese version (32) is used to assess the levels of medication adherence, which includes the frequency by which patients adjust or change the prescribed dosages.
It measures the levels of agreement as "sometimes forgetting to take, or sometimes altering the medication

Results
The Principal Axis Factoring (PAF) method was used for a first exploratory data analysis (with oblimin rotation and without forcing the previous number of factors), aiming to understand how data were naturally grouped.
From this analysis, items 4, 8, 9 and 12 were excluded due to their low communalities (lower than 0.30). Later, a second exploratory factor analysis was performed and the results showed that items were grouped into three factors and all items (excluding items 18, 19 and 20) were loaded into a single factor, with values above 0.30, as indicated in Table 2. ). An analysis of the modification indices was conducted and the model was re-specified through correlation between errors from items 5 and 6, 3 and 5, and this modified model (Model 2) showed a better fit for the data (34) .
Considering that, theoretically, social support is a multidimensional construct and that, empirically, the factors showed strong correlations with each other, a second-order factor was extracted, which allowed to calculate a total score for the social support scale, thus producing a third model (Table 3).   socio-affective support, Sense of control and Financial support. Some items loaded in more than one factor, but all were grouped into the factor where their loading was higher.
The first factor, named "Familiar and socio-affective support", groups the items that measure what the expressive dimension of social support usually describes.
This dimension evaluates whether respondents feel or believe that their family and friends are close and affectionate, and that they are available for sharing their problems. It has been argued that close relatives and friends have different roles in providing social support in old age, but they both represent important sources of love and affection, and contribute to subjective well-being (18) . The second factor, "Financial support", represents what is usually described as instrumental support, since it assesses if older people feel that they have sufficient financial support for their needs and if they feel able to manage their finances. As previously stated, this dimension is particularly relevant for Portuguese old adults, as they are a significant part of an economically deprived group (3) . Finally, the third factor, "Sense of control", includes items that evaluate how respondents feel that their close relationships are capable of respecting their autonomy and independence by providing support that is not over controlling. A review of the literature showed that older people have a strong inner drive towards autonomous decision-making, despite the dependency (28) .
Subsample B was used for the scale validation and to confirm the 3-factor structure of the IESS scale, in order to show its usefulness in assessing social support in older adults. To the best of our knowledge, this is the first study conducted in order to examine the factor structure of the IESS scale and previous research papers only reported the exploratory analysis in the study of the psychometric properties of the instrument (29) . The inclusion of 2 correlations between errors in the model was necessary, but the results obtained by CFA suggest that the 3-factor model structure performed the best on Lima L, Santos C, Bastos C, Guerra M, Martins MM, Costa P.
15. Pinquart M, Sorensen S. Influences of socioeconomic status, social network, and competence on subjective well-being in later life: a meta-analysis. Psychol