Development and validation of a questionnaire assessing the perceived control in health care among older adults with care needs in the Netherlands

To understand how older adults view control in relation to health care and to obtain a conceptual model for the development of a measurement instrument, a qualitative study was conducted [19]. Thirty-two older adults, in the age of 65–96 and mostly living independently, participated in either an in-depth interview (n = 20) or a focus group discussion (n = 12) to reflect upon their experiences with care and what factors caused them to feel (a lack of) control. A conceptual model was developed, summarizing the key factors that constitute perceived control in health care among older people. Five constituting factors were identified, as presented in Table 1.

The target population consisted of older adults with health or functioning problems and care needs. Professional care may concern the general practitioner (GP), medical specialist, formal home care (e.g., domestic help, personal care, or nursing care), and non-medical types of care, such as help or care from physiotherapists, dietitians, dentists, but also help in the form of practical aids (e.g., walking devices) which are provided by agencies. By informal care, we refer to recurrent help or care from the social network, for example from partner, family members, friends, or neighbors. The types of care/help that are given by the several caregivers were not defined, but may include medical treatment, consult or advice; psychosocial care or advice; personal or nursing care; practical support; and emotional support.

Using our conceptual framework (Table 1), the topics from our qualitative interview study were converted into a tentative list of 63 items. This list underwent several stages of recurrent testing and adaptation in collaboration with multiple parties.

(1) We performed a pilot test with three older people, to test a strongly abbreviated version of the original list; for this purpose, we used a cognitive debriefing method [20, 21]. (2) Consultation was sought from three scientists in the field of aging by e-mail. (3) For further improvement, we sought the opinions from members (older adults) of the advisory panel that is associated with our research project. (4) A total of 198 older adults filled out a newly revised version of the item list. These were participants from the Longitudinal Aging Study Amsterdam (LASA), which is a cohort study started in 1992, aimed at investigating the trajectories and interrelationships of several aspects of functional change with aging, among older adults in the ages of 55 and over [22]. The aim of this fourth stage was to explore basic statistics, such as the response rates, and to collect written evaluations from the participants concerning the items. (5) Lastly, data from the member check in the qualitative study—in which 11 participants responded either on their interview report or on a summary document of the conceptual model (Table 1)—were used for a final adaptation of the questionnaire contents.

To avoid possible overlap with concepts such as quality of care, we focused on older people’s perceived ‘personal’ control and no longer included items about perceived support from care professionals/organizations and perceived support from services/infrastructure (see topics between square brackets, Table 1). Consequently, the questionnaire will be limited to the confidence in people’s own efforts, on (1) organizing care and (2) management in the home setting; and also (3) perceived support from one’s social network was believed to be greatly interwoven to people’s overall sense of personal care-related control, and was kept in the instrument. These three elements represent people’s self-reliance, i.e., their perceived own control resulting from efforts by themselves possibly in combination with efforts of people in their informal network. Furthermore, the availability of an emergency plan was included as we believed this subtopic also reflects self-reliance to some degree.

Two of the subcategories within the constituting factor organizing professional care (Table 1) were identified and incorporated as separate parts in the questionnaire. First, ‘communication with care professionals’ was elaborated with three more items, because on micro-level our interview data showed that communicating with care professionals includes multiple aspects, such as providing information to the physician, asking questions, and participating in decisions. As these three aspects are, additionally, reflected in existing viewpoints about doctor–patient interaction [23], we felt the need to distinguish between these levels of communication in the questionnaire. Second, ‘planning (more) complex care in the future’ was considered as an independent topic, because its importance was emphasized by respondents in the member check: In response to these respondents’ comments, two items were added, of which ‘perceived sufficiency of financial resources’ represented a new subtopic. Eventually, each subtopic within Table 1 is converted into a minimum of one and a maximum of three items.

The final instrument is a self-report questionnaire, counting 29 items and existing of two main parts (Table 2, first column). Part A includes four items of which three are rated on an 11-point Likert scale (0 = not at all to 10 = completely). These cover the instrument’s overall topics, i.e., to what extent one feels to be in control over one’s health care, to what extent one feels to be supported by people in their social network, and the extent to which one feels that personal control in care is important. The remaining item of part A has a nominal response scale and focuses on who is the main person responsible for the received care, according to the respondent. Part B consists of 25 items that are divided in various parts to ease questionnaire administration, i.e., structured according to separate types of effort in relation to care: (B1) organizing professional care (eight items), (B2) communication with healthcare professionals (four items), (B3) health management in the home situation (four items), (B4) planning (more) complex care in the future (four items), and (B5) perceived support from the social network (five items). These address people’s perceived personal control in care with or without structural help from significant others in their network (B1–B4), or explicitly address the extent to which people feel supported by the informal network surrounding them (B5). All these items have a five-point Likert scale (1 = not able or with great difficulty to 5 = with great ease). Because we regard perceived control in health care to be an overall feeling that originates from multiple experiences in health care, it is required that people may look back upon a substantial time span. Thus, people are instructed to rate their situations by giving an average evaluation of the experiences regarding a certain aspect over the past year.

To determine the construct validity of the health care-related perceived control questionnaire, we investigated the factor structure (structural validity) and the relationship with associated constructs (hypotheses testing).

Second, for the hypotheses testing, we selected measures that assess related concepts, and were either included in the same interview, i.e., a sense of mastery [5], or derived from the previous LASA main cycle in 2011/12, i.e., self-efficacy [24, 25], self-esteem [26], and social loneliness, a subscale from the loneliness instrument by De Jong Gierveld and Kamphuis [27]. The control-related instruments reflect: control over events and situations in life in general (mastery, seven items); the belief in one’s ability to organize or execute certain behaviors to produce given attainments in general (self-efficacy, 12 items); and the overall evaluation of one’s own worth (self-esteem, with a four-item adapted version of the Rosenberg scale). We expected these to moderately positively correlate with the items that fall within the ‘perceived personal control’ component (.30 ≤ r ≤ .50). Social loneliness (five items) refers to the perceived ‘absence of a broader engaging network’ (p. 122) and includes issues of having enough people in one’s network to rely on in case of difficulty [28]. This subscale was expected to negatively correlate with the ‘perceived support’ component in our questionnaire (r ≥ −.30). We applied one-tailed Spearman’s rho analysis for nonparametric data, as our data were not normally distributed and because directions of the correlations were hypothesized a priori.

The reliability was investigated by determining the internal consistency (Cronbach’s alpha) of the final scales that are formed, based on the results of the EFA and our own decisions following interpretation of these EFA results. A Cronbach’s α value between .70 and .90 was considered adequate [29].

The EFA included all items except for items 1–4 (Part A of the questionnaire), because these items either attempt to capture an overall score of the other specific items (items 1 and 2), or because these were not intended to contribute to the perceived control score at all (items 3 and 4). Based on our hypothesis that the items were to be divided into a ‘perceived personal control’ component and a ‘perceived support’ component, we tested a two-factor model. This hypothesis was supported by the factor structure obtained from EFA, with the first factor (perceived personal control) showing an eigenvalue of 8.5 and accounting for 34.0 % of the variance, and with 2.0 and 8.1 %, respectively, for the second factor (perceived support). However, we found three items with loadings under the minimum acceptable level. These items seemed to be coherent on a conceptual level, all relating to expectations regarding future (more) complex care. Therefore, we conducted a final analysis, presetting the number of factors at three. Consequently, a model emerged (see Table 2) that appeared to be the most adequate. The factors that we established were identified as: (I.) ‘perceived personal control in health care’ (13 items); (II.) ‘anticipated personal control regarding future health care’ (three items); and (III.) ‘perceived support from the social network’ (three items). The total variance explained by these factors was 48.5 %. Six items were excluded from the factor structure, either because these did not show clear contrast (items 19, 20, 24, and 29), or because they showed loadings beneath the minimum acceptable level for all factors (items 20 and 24), or lastly because we considered the items to be conceptually deviant from the main contents of the factor that they were statistically assigned to (items 16 and 29) or to belong to more than just one factor on conceptual grounds (item 27, i.e., this item fitted on factor III ‘perceived support’; however, having an emergency plan is not purely a matter of support, but may also reflect own efforts and therefore may belong to the ‘perceived personal control’ factor as well).

The internal consistency of the three scales—with a total of 19 items—was: .90 for the ‘perceived personal control’ scale (13 items); .71 for the ‘anticipated personal control’ scale (three items); and .77 for the ‘perceived support’ scale (three items). For each subscale, average scores can be calculated with higher scores reflecting higher levels of perceived (anticipated) personal control or perceived support, respectively. To reflect an overall level of perceived control in health care, we suggest to average the scores of all Part B items—including the single items not integrated in the factor structure. In addition, an overall level of people’s perceived ‘own control’ can be represented by the score on item 1.

We found a moderate (.30–.50) association between the domain of ‘perceived personal control’ and the sense of mastery, self-efficacy, and self-esteem instruments in positive direction; and between the ‘perceived support’ domain and social loneliness in negative direction (Table 4). This confirmed our expectation that higher perceived personal control in care was accompanied by greater perceptions of mastery, self-efficacy, and self-esteem; and that higher perceived support was related to lower experienced social loneliness. ‘Anticipated personal control regarding future care’ was less strongly associated with the control-related constructs than expected (.19–.25, i.e., under the minimum of r = .30). The subscales within our questionnaire correlated moderately positively with one another, varying from .30 to .46 (Table 4).