The Swiss Health Literacy Survey: development and psychometric properties of a multidimensional instrument to assess competencies for health

Abstract

Background  Growing recognition of the role of citizens and patients in health and health care has placed a spotlight on health literacy and patient education.

Objective  To identify specific competencies for health in definitions of health literacy and patient‐centred concepts and empirically test their dimensionality in the general population.

Methods  A thorough review of the literature on health literacy, self‐management, patient empowerment, patient education and shared decision making revealed considerable conceptual overlap as competencies for health and identified a corpus of 30 generic competencies for health. A questionnaire containing 127 items covering the 30 competencies was fielded as a telephone interview in German, French and Italian among 1255 respondents randomly selected from the resident population in Switzerland.

Findings  Analyses with the software MPlus to model items with mixed response categories showed that the items do not load onto a single factor. Multifactorial models with good fit could be erected for each of five dimensions defined a priori and their corresponding competencies: information and knowledge (four competencies, 17 items), general cognitive skills (four competencies, 17 items), social roles (two competencies, seven items), medical management (four competencies, 27 items) and healthy lifestyle (two competencies, six items). Multiple indicators and multiple causes models identified problematic differential item functioning for only six items belonging to two competencies.

Conclusions  The psychometric analyses of this instrument support broader conceptualization of health literacy not as a single competence but rather as a package of competencies for health.

Background

Competencies for the information society

The modern information and knowledge society brings with it many new opportunities but also sets ever‐higher demands for its inhabitants. The ‘active citizen’ is called on to stay informed, make decisions and take action as a voter, a consumer, a patient, etc. Therefore, the extent to which citizens possess the necessary competencies to meet the demands of today’s information and knowledge society is highly relevant in many sectors.

The Organisation for Economic Cooperation and Development (OECD) has helped spearhead efforts to define and measure key competencies for modern societies. ¹ Multidisciplinary expert reviews and discussions evidenced no prior consensus on competencies, yet a basic model summarizing competencies as the interplay between individual capacities (i.e. one’s knowledge and skills, but also motivation and attitudes) and the demands posed in one’s surroundings could be forwarded. In other words, competencies are neither formed nor exercised in a vacuum but depend rather substantially on the social context at a pragmatic level and on values and visions at an ideological level.

Happily, international work around competencies has not remained solely conceptual. The Programme for International Student Assessment (PISA) and the international adult literacy and life skills surveys (IALS/ALL) constitute major endeavours to measure key competencies beginning with the most basic skills (i.e. reading and numeracy) in school‐ and work‐age adult populations, respectively. Switzerland has participated in both multinational assessments, ² , ³ and as in many industrialized countries, one consistent finding has been insufficient reading skills among a large segment of teenagers and work‐age adults.

Link between competencies and health

Although few studies to date have gathered robust data on both competencies and health outcomes, there is a considerable body of evidence that education is one of the most important social determinants of health. ⁴ While schooling and basic competencies of reading and numeracy are strongly correlated, data from the adult literacy surveys demonstrate that they should not be considered one and the same. ⁵ Their findings also suggest that differences in reading and numeracy skills are associated with large differences in health. ⁵ , ⁶ Even in relatively egalitarian societies such as Switzerland, people with higher education still demonstrate healthier behaviours, report less chronic illness, feel healthier and live longer. ⁷ , ⁸ Better employment opportunities and higher income explain such discrepancies only in part, and economists have hypothesized that better cognitive and decision‐making abilities (e.g. better and faster uptake of new information and technologies) may be the key ingredient. ⁹ Given the sheer quantity of new knowledge and innovations on the one hand and high turnover on the other, such competencies may be particularly relevant in the field of health and medicine.

In Europe, several multinational initiatives have underscored the salience of health information. The Informed Patient Study based at Cambridge University ¹⁰ concluded that demand for information and participation are indeed on the rise and that better informed patients are less anxious, more satisfied, more adherent, better at using health resources leading to lower costs and more likely to commence treatment earlier. The European Future Patient Project coordinated by Picker Europe concluded that despite differences in culture and healthcare systems, there were striking similarities across Europe in terms of citizens’ preferences for information, participation and choice, as well as failure of systems to stay abreast with and capitalize on these changes. ¹¹ Finally, two series of multicountry surveys commissioned by the Stockholm Network found that by a large margin, citizens in Europe named ‘giving patients more information about their illness’ the health reform most likely to improve the quality of care. ¹² , ¹³

Health literacy

The field of health literacy took form in the 1990s from the efforts of practitioners and public health researchers to better understand the link between low literacy (as reading skills) and poor health. ¹⁴ In North America, most studies have focused on the impact of low reading skills or poor understanding of health information among patients in the healthcare system. This body of research has provided clear evidence that not only are low reading skills correlated with lower knowledge and skills with respect to the patient’s condition but also poorer prevention, diagnosis and prognosis. ¹⁵ , ¹⁶ Health economists estimate that this kind of low health literacy may cost the American healthcare system up to US$106 billion per year ¹⁷ and the Swiss healthcare system CHF 1.5–2.3 billion per year. ¹⁸ , ¹⁹

Given the lack of dedicated instruments that measure health literacy beyond basic reading skills in clinical settings, the development of instruments that capture health literacy in a more comprehensive manner has been prioritized in both North America ¹⁶ , ²⁰ and Europe. ²¹ Furthermore, given growing interest in health literacy internationally, such instruments will also need to exhibit cross‐cultural validity. The Future Patient Project Switzerland contributes to this priority with a fuller operationalization of the concept of health literacy in the form of a survey instrument which provides empirical data on a large set of core competencies for health in the general population (including a large proportion of people living with a long‐term condition) in a country that has three major cultural‐linguistic regions and a large migrant population.

Methods

This project had several objectives: (i) enumerate core competencies for health and identify indicators that provide some indication (measurement) of those competencies, (ii) gather data on those indicators among a representative general population sample to assess the properties of the instrument and the distribution of competencies in the population and (iii) launch discourse by bringing findings into public and stakeholder fora. This publication describes the conceptual framework, questionnaire development and the psychometric properties of the instrument to address the issues of the dimensionality of health literacy and cross‐cultural performance of the survey items.

Theoretical groundwork

To enumerate core competencies for health, the research team updated their systematic review of the literature in patient‐centred topics of self‐management, patient empowerment, patient education and shared decision making (including informed decision making) and supplemented it with a new search on health literacy (MEDLINE and grey literature since 1990). We identified key publications that elucidate each topic conceptually and/or present empirical measurements for it.

As a first step, we reviewed conceptual definitions of these concepts. While health literacy is a relatively new concept, several working definitions have been forwarded. ²⁰ , ²² The narrowest approaches focus on reading health‐related materials in healthcare settings. ¹⁵ , ²³ Most research on health literacy in North America to date corresponds to this definition. A broader approach places an emphasis on understanding and using information for decision making in health matters. ¹⁶ , ²⁴ While both reading (i.e. working with general or health‐related written materials) and decision making constitute essential components of health literacy, there are many other indispensable competencies for health in modern societies. Indeed, the family of research and policy initiatives on modern literacies – e.g. financial literacy, scientific literacy, computer literacy – is not restricted to reading specific content but covers a broad set of knowledge, skills and attitudes needed to achieve adequate functionality/proficiency or better in everyday life. Although each topic has its own particular history and application, their definitions reveal considerable overlap as evidenced by a selection of representative definitions presented in Table 1.

Table 1.

Selected definitions of health literacy and other patient‐centred concepts

Topic	Definition
Health literacy	Cognitive and social skills that determine the motivation and ability of individuals to gain access to, understand and use information in ways that promote and maintain good health 53
Self‐management	Individual’s ability to manage the symptoms, treatment, physical and psychosocial consequences and lifestyle changes inherent in living with a chronic condition 47
Patient empowerment	An educational process designed to help patients develop the knowledge, skills, attitudes and degree of self‐awareness necessary to effectively assume responsibility for their health‐related decisions 54
Patient education	A systematic learning experience that influences the way the patient experiences his/her illness and/or his/her knowledge and health behaviour and is aimed at improving or maintaining health and learning to cope with a condition 55
Informed decision‐making	A reasoned choice is made by a reasonable individual using relevant information about the advantages and disadvantages of all the possible courses of action, in accord with the individual’s beliefs 48

Essentially, underlying these concepts is a collection of competencies – comprised of knowledge, skill and attitudinal components – geared towards a health‐related outcome. When authors actually specify which competencies a health‐literate person or an informed/educated/empowered patient should possess, we find that not only does each topic cover a broad palate of competencies across various dimensions, but there is considerable overlap between different topics at this level as well. In Table 2, this is evidenced visually, whereby the competencies cited explicitly in selected key publications for each topic are tabulated. If inferred competencies were to be included, even more boxes would be checked off.

Table 2.

Competencies for health as specified in the literature on health literacy and other patient‐centred concepts

Dimension and competence	Concept
	Health literacy	Self‐management	(Patient) empowerment	Patient education	Shared decision‐making	Selected references
A Information and knowledge
AA Access to information about health	X		X	X		16, 55, 56, 57, 58
AB Seeking health information	X	X	X		X	16, 35, 38, 39, 47, 57, 59, 60, 61, 62
AC Ability to understand health information	X			X	X	16, 48, 56, 63
AE Ability to apply health information	X					16, 59
AF Knowledge about condition and treatment	X	X	X	X	X	47, 48, 55, 57, 59, 62, 64
AG Scientific literacy	X					59, 65
B General cognitive skills
BA General life skills	X			X		56, 57, 63, 65, 66
BB Problem solving		X	X	X		54, 61, 62, 63
BC Taking action		X		X		61, 63
BD Ability to understand and give consent	X			X		16, 55, 57
BE Ability to understand risk	X				X	56, 57, 67
BF Critical decision‐making	X	X	X	X	X	35, 38, 39, 48, 56, 58, 59, 63, 64, 67, 68, 69
C Social roles
CA Address social determinants of health	X		X			16, 56, 58, 59, 65
CB Consumer competencies	X	X				59, 68
CC Adapt to work		X	X			60, 62
CD Manage relations with significant others		X	X	X		47, 54, 55, 60, 62
D Medical management
DA First aid	X					59
DB Recognize and act on symptoms		X				47, 60
DC Navigate health care systems	X	X				59, 68
DD Use medications correctly/adherence	X	X		X		25, 47, 55, 60
DE Effective interaction with healthcare providers	X	X	X	X	X	16, 37, 47, 60, 61, 62, 63, 64, 66, 67, 68, 69
DF Healthcare utilization		X				47, 61
DF Informal care	X	X				59, 68
E Healthy lifestyle
Overall	X	X				47, 59, 60
EA Nutrition and diet		X				47, 60
EB Exercise		X				47, 60, 61
EC Non‐smoking		X				47, 60
F/G Motivational skills
FA Manage psychological consequences		X	X		X	47, 48, 60, 61, 62
GA Self‐efficacy	X	X	X			47, 54, 56, 59, 61, 62
GB Emotional intelligence		X		X		70, 71
H Attitudes and values
HA Identify personal preferences (values, needs, goals)	X	X	X	X	X	36, 47, 48, 54, 59, 62, 64, 65, 67

Given the high degree of overlap at both conceptual and content levels, we decided to take a broad, inclusive approach by focusing at the level of shared core competencies – that is, the active ingredients of each concept. Instead of selecting and operationalizing just one concept or definition, this approach would permit assessment of multiple concepts and definitions. The competencies for health chosen for this project are generic – that is, relevant to almost everyone. It is apparent that some competencies resemble discrete tasks and others higher order skills. Just as there is no consensus definition of competencies, there is also no consensus on boundaries along the task–skill–competency continuum. ¹ It has already been demonstrated that skills/tasks in health literacy can often be broken down into more discrete skills/tasks. ²⁵

To present the 30 competencies for health in a succinct manner, the list of core competencies were organized a priori in the following seven dimensions: (i) information and knowledge, (ii) general cognitive skills, (iii) social roles, (iv) medical management, (v) healthy lifestyle, (vi) motivational skills, and (vii) attitudes and values. These dimensions were determined based on the OECD model of competencies ¹ and domains commonly used in patient‐centred concepts.

Operationalizing health literacy

After itemizing the core competencies for health, the next step was to identify indicators and items to measure each competence, drawing on the empirical literature – both health‐ and non‐health‐specific – in English, German and French.

Before 2006, empirical studies on health literacy had been undertaken only in the United States and Canada, yet the instruments used largely measure basic reading skills around health. ²³ , ²⁶ Although useful in identifying patients with low (health‐related) reading skills, these instruments appear to be highly context and language specific. Experts in British Columbia have developed measures of understanding health information based on rich qualitative and quantitative research among vulnerable population groups. ²² As such, some findings may be specific to the groups studied. Outside the health field, the adult literacy surveys have provided the best assessments of basic life skills such as reading and numeracy. Health‐related questions were selected and coded to create index scores for using print material with health content or relevant to health. ²⁷ , ²⁸ Although the assessments and scoring are first rate, there are several drawbacks to this approach. Firstly, the questions were combined a posteriori from an instrument not created to measure health literacy per se. Secondly, while five categories of health activities were identified – health promotion, health protection, disease prevention, health care and maintenance, and systems navigation – only a single index score for health literacy can be calculated. Thirdly, the single index score has proven to be sensitive to the items chosen for its construction, as two different approaches have yielded scores with important discrepancies. ²⁸ , ²⁹

Although research and practice in health literacy are still in their infancy, initiatives in research and practice have been more plentiful for some of the patient‐centred concepts. For example, patient education surveys have been carried out in Finland ³⁰ and Australia ³¹ ; self‐management programmes and assessments have been developed in the United States ³² , UK ³³ and Australia ³⁴ ; shared decision making has been researched extensively in the United States ³⁵ , ³⁶ , UK ³⁷ and Germany ³⁸ , ³⁹ . Although these surveys focus on patients, many of the competencies covered are relevant to the population at large.

Given that many of the competencies enumerated for health literacy and patient‐centred concepts are generic competencies relevant to health, we also reviewed the literature on adult literacy and life skills, including specific areas such as scientific literacy, risk, decision making, responsibility and consumerism.

Survey instrument

We compiled questions for each competence, yielding an original draft questionnaire of over 500 individual items in English, German or French. Through an iterative process involving two senior researchers, we eliminated semantic duplicates opting for the clearest and briefest variant and aimed for high content validity by selecting items for good conceptual coverage of each competence (or indicator thereof), reflecting the current healthcare system and upcoming drivers of change.

A draft source questionnaire of 200 questions was translated into German, French and Italian, with two translations made for each language by different translators. Both versions were used in composing the definitive version in each of the three languages. Based on actual pretests with a small randomly selected sample of the general population, programming errors were corrected, and the questionnaire was furthered shortened to comply with the target time frame. The final version of the questionnaire contained 158 items, corresponding to a 30‐min telephone interview.

As the second objective was to assess as many of the competencies as possible, the actual instrument contained four different families of questions: (i) 127 questions on 30 competencies for health, (ii) 12 questions on socio‐demographic determinants, (iii) 17 questions on health‐related outcomes – i.e. health status and health‐services use – and (iv) two questions on the interview. The questions on socio‐demographics and services use were taken from EUROHIS recommendations, the Swiss Health Survey or Gallup Switzerland standards. Health status may also be considered a possible determinant of health literacy, and health‐services use also a competence.

In the final version of the questionnaire, each competence was measured by 1–4 indicators corresponding to 1–19 questions. No items could be found or devised for three competencies – i.e. ability to apply health information, ability to understand and give consent and navigating healthcare systems – even though navigating healthcare systems may be approximated by using items measuring other competencies. Response formats varied as questions measured knowledge, behaviours/skills, motivation and attitudes.

Data collection and participants

We considered two data collection methods: computer‐assisted telephone interviews (CATI) and face‐to‐face interviews in people’s homes. Owing to funding constraints, CATI was selected which meant a shorter interview time (target length: 30 min), the exclusion of complex questions which could otherwise be aided by visual props and, most importantly, the exclusion of reading skill assessments. As in some Nordic countries, CATI is the standard data collection mode in Switzerland, used in most major surveys such as the Swiss Health Survey.

The target sample size was 1250 respondents sampled randomly from the resident population aged 15+ years speaking one of three official languages. This target would at least yield standard errors associated with the standard sample size of 1000 respondents as well as accommodate over‐sampling of the French‐ and Italian‐speaking regions (actual distribution in 2000, German‐speaking 64%, French‐speaking 20% and Italian‐speaking 6.5%), which permit more meaningful comparisons between and within regions.

The actual data collection was outsourced to ISOPUBLIC, the Swiss member of Gallup International, which operates in accordance with the International Code on Market and Social Research and Swiss data protection laws. In particular, the topic and author of the study were identified upfront to all potential respondents for their consent. Between April and mid‐May 2006, 1255 interviews were conducted among a representative sample of the resident population 15+ years in German (652), French (303) and Italian (300). A total of 12 958 telephone numbers were chosen at random from current telephone registries Twixtel. Three thousand five hundred forty‐six were outright refusals and 156 refused owing to poor health, yielding a response rate of 25.3%. An additional 466 people could not participate owing to language or hearing problems (technically considered ineligible).

The sample was stratified by linguistic region and filled by quota sampling for sex and age group. Separate weights were assigned for both the entire country and separately for each of the three linguistic regions to correspond to the Federal Statistical Office’s distributions for linguistic region, economic region, agglomeration, sex and age group. The statistical errors are ±2.8% for the entire Swiss sample, ±3.9% for the German‐speaking region and ±5.8% for the French‐ and Italian‐speaking regions.

Data analysis/statistical methods

The first objective of the analyses is to determine the dimensionality/factors of health literacy, by assessing whether an overarching single dimension of overall health literacy is supported by the data and, if not, identifying distinct factors. Although the definitions and descriptions of health literacy point to a multidimensional package of knowledge, skills and attitudes, existing measures focus on health‐related reading skills, which is arguably only one competence (in a specific context). In constructing this instrument, we sought to operationalize multiple definitions and descriptions, thereby capturing a wide array of competencies for health. Therefore, we hypothesize that the analyses will confirm the multidimensionality of such an operationalization of health literacy. If health literacy is indeed multidimensional, what are the key dimensions?

Individual item characteristics were examined, focusing on frequencies of responses in each category rather than means and standard deviations, given the wide range of item formats that did not permit meaningful comparisons of means and standard deviations. In addition, patterns of missing data were analysed to determine reasons for missing data. No items were eliminated from analyses owing to missing data, and missing data did not appear to depend on identifiable factors such as order of item administration.

Of the 127 items measuring the competencies for health, 104 were used in these analyses: 73 items pertaining to all survey respondents (citizens) and 31 items pertaining only to respondents who reported having a ‘long‐standing illness or health problem’ or were ‘undergoing a long‐term treatment’ (patients). ‘Patient’ items constitute most or all items for the following five competencies: knowledge about condition and treatment (AF), taking action (BC), recognizing and acting on symptoms (DB), using medications correctly (DD), and effective interaction with healthcare providers (DE). Items excluded from these analyses included follow‐up questions answered only by a subset of respondents and standardized variables measuring health‐services use (DF). All individual item analyses and subsequent factor analyses were conducted using weighted data.

Given the mixed nature of the items (response categories ranged from dichotomous through 10‐point scales), the software Mplus version 3.1 was used to model such complex data. Exploratory factor analyses (EFA) were carried out for all ‘citizen’ and ‘patient’ items and also for each dimension defined a priori for ‘citizen’ items. All items were treated as categorical variables, WLSMV estimation (weighted least square parameter estimates using a diagonal weight matrix with standard errors and mean‐ and variance‐adjusted chi‐square test statistic that use a full weight matrix), a paired present approach for missing data, and promax factor rotation. The number of factors extracted was based on model fit indices, scree plot analyses and the interpretability of factors. In all factor analyses, items with factor loadings >0.30 were retained. Mplus provides a chi‐square fit test and two fit indices: the root mean square error of approximation (RMSEA) and the standardized root mean square residual (SRMR). ⁴⁰ , ⁴¹

Confirmatory factor analysis (CFA) models were run with the model structures generated from the EFA, and the former should still be considered exploratory models despite the name because the basic model structures were generated using the latter. In the CFA models, modification indices were used to identify pairs of items within scales for which model fit would improve if error estimates were freed to covary and for which there appeared to be theoretically justifiable shared method effects. In addition, separate confirmatory factor analyses were run for each individual factor to test the degree of unidimensionality of factors. Mplus maximum likelihood methods for handling missing data were used. In addition to the RMSEA, the Tucker–Lewis index (TLI) and the comparative fit index (CFI) were used to assess fit in CFA models. ⁴⁰ , ⁴²

The second objective of the analyses is to assess the cross‐cultural robustness of the instrument and the factor structure by assessing differential item functioning (DIF) across Swiss linguistic regions. A major challenge in conducting multilingual or cross‐cultural assessment is establishing that the same thing is being measured in each group. If questions are unbiased, respondents from different groups who have similar health literacy characteristics on given dimensions will respond similarly to individual items within each dimension. Methodologically, this means that responses to a given item will be independent of linguistic group membership once respondents are ‘matched’ on their responses across the dimension.

Differential item functioning refers to situations in which an item is not reasonably comparable across groups. It is important to assess DIF to separate true group differences from differences that arise from other factors, such as differences in interpretation of an item across groups. Multiple indicators and multiple causes models (MIMIC) build on the confirmatory factor analysis models developed in the first part of the analysis. The MIMIC model extends the CFA by adding an exogenous variable (e.g. German vs. French or Italian language) to the model. A key model assumption is that once the effect of group membership on the overall factor level is controlled for each factor, there should be no additional effect of group membership on individual items. In instances where group membership predicts item responses after controlling for the effect of group membership on overall factor levels, the assumption is that DIF is present and the item is not functioning equivalently across groups. Multiple indicators and multiple causes models also provide an estimate of the amount of bias caused by items with DIF, as statistical significance may not reflect a substantive effect on overall levels of latent factors at the scale or subscale level.

Multiple indicators and multiple causes models to assess DIF were conducted for ‘citizen’ items for each of the factors identified from the confirmatory factor analyses with sufficient unidimensionality. However, they could not be used to assess DIF for any of the factors from ‘patient’ items given the lack of sufficiently large subgroups. Identification of possible DIF proceeded in stepwise fashion. One factor at a time, each item was initially regressed on French‐ and Italian‐speaking status, controlling for the association between group and the latent factor. The item–language group link with the largest effect on model fit was incorporated into the model, followed by the second largest link after adjustment for the first and so forth until no other item–language group links resulted in significant (P < 0.05) improvements in model fit. The regression of the latent factor on each language group was carried out with and without adjustment after correcting for possible DIF for all items with significant item‐diagnosis links at P < 0.05, not correcting for multiple comparisons. Because multiple significance tests were run, Hochberg’s sequential method ⁴³ was also applied to assess items with potential DIF while controlling for type I error.

To assess the potential effect of items with DIF on the latent factor, regression parameters of the latent factor on each language group were calculated both before and after adjustment for DIF. ⁴⁴ Because the variance of each latent factor is set to 1, these represent, in terms of standard deviations of the latent factors, the degree to which French‐ or Italian‐speaking respondents tend to be higher or lower than German‐speaking respondents. Differential item functioning with an effect on the factor level would result in a substantively different group parameter after adjusting for DIF. Because many of the analyses were conducted on factors with a relatively small number of items, some degree of evidence for DIF is expected, and our attention should be focused on large deviations. Differences in the level of the latent variable may be due to ‘true’ differences or differences in measurement.

Results

The profile of the participants reflected that of the Swiss general population, except that those who completed mandatory schooling or less (13%) and non‐Swiss nationals (11%) were under‐represented. A third reported a long‐term condition or were undergoing long‐term care.

Dimensionality of the Swiss Health Literacy Survey

First, we tested a single‐factor model whereby all 73 ‘citizen’ items were loaded onto one overall factor of health literacy. The model fit poorly (χ²(338) = 5415.8, CFI = 0.60, TLI = 0.63, RMSEA = 0.11).

Second, exploratory factor analysis of all 73 ‘citizen’ items resulted in a four‐factor model with 38 items and reasonable fit ( = 1998.0, RMSEA = 0.07, SRMR = 0.06). As shown in Table 3, the four factors were information and decision‐making (13 items), cognitive and interpersonal skills (eight items), ICT skills (six items) and health activation (11 items), which consists of seeking health information, self‐care and healthy lifestyle. When the model was assessed by confirmatory factor analysis, two items (Q92 and Q93) were removed owing to low loadings, resulting in a 36‐item, four‐factor model with reasonably good fit indicators (χ²(187) = 1302.2, CFI = 0.92, TLI = 0.94, RMSEA = 0.07). The factors information and decision‐making, cognitive and interpersonal skills and ICT skills exhibited moderate unidimensionality, but health activation exhibited low unidimensionality. This finding was consistent with Cronbach’s alphas calculated for each factor: information and decision‐making = 0.72, cognitive and interpersonal skills = 0.81, ICT skills = 0.77 and health activation = 0.60. The cognitive and interpersonal skills and ICT skills factors correlated at 0.50, and the information and decision‐making and ICT skills factors at 0.27. Other factor intercorrelations ranged from −0.05 to 0.16.

Table 3.

Factor structure of the Swiss Health Literacy Survey – all non‐patient items in one model

Item and content			Factor loadings
No.	Original competence	Item	F1	F2	F3	F4
Q3	Access to information about health	Having enough information to choose treatment	0.49	0.03	−0.01	0.03
Q5	Access to information about health services	Having enough information to choose gp	0.54	−0.04	0.12	0.04
Q6	Access to information about health services	Having enough information to choose hospital	0.49	0.05	0.00	0.01
Q7	Access to information about health services	Having enough information to choose sick fund	0.49	−0.01	0.04	−0.01
Q13	Seeking health information	Understanding information on food labels	0.37	−0.01	−0.07	−0.02
Q28	Scientific literacy	Knowledge about science and technology	0.36	0.12	−0.22	−0.09
Q48	Critical decision‐making	Choosing a sick fund	0.52	0.00	0.05	−0.04
Q49	Critical decision‐making	Choosing a gp	0.53	−0.03	0.07	−0.08
Q50	Critical decision‐making	Choosing a medication	0.75	−0.10	−0.09	0.10
Q51	Critical decision‐making	Choosing a medical treatment	0.67	−0.11	−0.09	0.09
Q52	Critical decision‐making	Choosing foods	0.35	0.13	0.07	0.07
Q60	Address social determinants of health	Knowledge of patients’ rights	0.67	0.04	0.04	−0.07
Q61	Consumer competencies	Knowledge of consumer rights	0.60	−0.03	−0.03	−0.14
Q29	General life skills	Reading	−0.10	0.79	0.00	0.16
Q31	General life skills	Numeracy	0.01	0.65	−0.01	−0.04
Q35	General life skills	Communication skills	−0.01	0.70	−0.07	0.03
Q36	General life skills	Problem‐solving	0.07	0.65	−0.07	0.01
Q37	General life skills	Cross‐cultural skills	0.00	0.58	0.06	0.07
Q38	General life skills	Team work	0.01	0.64	0.11	0.00
Q39	General life skills	Learning new things	0.06	0.61	−0.25	−0.02
Q30	General life skills	Writing	−0.07	0.84	−0.01	0.16
Q32	General life skills	ICT skills – computer	−0.09	0.40	−0.81	−0.17
Q33	General life skills	ICT skills – Internet	−0.12	0.39	−0.86	−0.19
Q34	General life skills	Technical skills	0.03	0.47	−0.52	−0.13
Q18	Ability to understand health information	Understanding information on the Internet	0.11	−0.03	−0.77	0.07
Q95	Healthcare utilization	Avoid a gp visit by using the Internet	0.13	−0.09	−0.61	0.30
Q11	Seeking health information	Internet a good source of information about health	0.05	−0.06	−0.88	0.12
Q8	Seeking health information	Seeking health information to avoid gp visit	−0.04	−0.12	−0.25	0.43
Q9	Seeking health information	Seeking health information to prepare for gp visit	0.04	−0.06	−0.33	0.58
Q10	Seeking health information	Seeking additional information after gp visit	−0.04	−0.01	−0.26	0.52
Q12	Seeking health information	information seeking on food labels	0.02	0.14	0.15	0.44
Q14	Seeking health information	Information seeking on PIL	0.01	0.19	0.14	0.43
Q93	Informal care	Avoid a gp visit by treating at home	−0.10	−0.06	−0.21	0.53
Q92	Informal care	Avoid a gp visit by waiting it out	−0.14	−0.03	−0.14	0.54
Q109	Healthy lifestyle	Health consciousness	0.08	0.18	0.14	0.39
Q110	Healthy lifestyle	Activities to maintain health	0.00	0.13	0.13	0.35
Q113	Nutrition and diet	Eating healthy	0.13	0.23	0.29	0.50
Q127	Identify personal preferences	Interest in more self‐care for minor ailments	0.02	0.02	0.06	0.39

Competencies of the Swiss Health Literacy Survey

Given the disappearance of many items/competencies in the global CFA/EFA models, moderate fit and moderate/poor unidimensionality of the four factors identified, we proceeded to analyse each dimension – as defined a priori– separately to discern the extent to which the 30 competencies identified in Table 2 were being measured by the items in the Swiss Health Literacy Survey (HLS·CH) instrument.

Table 4 presents a summary overview of the fit statistics from the confirmatory factor analyses conducted per dimension as defined a priori for ‘citizen’ items and in one model for all ‘patient’ items (shown under the dimension medical management). Multifactorial models with good fit could be erected for five of the seven dimensions information and knowledge, general cognitive skills, social roles, medical management and healthy lifestyle. Exploratory factor analysis suggested no single dimension underlying the three single‐item competencies placed together under motivational skills: managing psychological consequences, self‐efficacy and emotional intelligence. While EFA suggested a good two‐factor fit for half of the items under the dimension attitudes and values, CFA revealed poor fit.

Table 4.

Confirmatory fit statistics for the Swiss Health Literacy Survey in separate dimensional models

Dimension and competence			Measure
Original competence	No. items	Confirmed competence	No. items	χ2	d.f.	P value	RMSEA	CFI
A Information and knowledge	4*		4*	280.3	75	<0.001	0.05	0.92
Access to health information	7	Access to health information	4	2.4	3	0.30	<0.01	1.00
Seeking health information	8	Seeking health information	6	15.5	7	0.03	0.03	0.99
Ability to understand health information Scientific literacy	3 1	Ability to understand information on health and science	5	4.2	5	0.52	<0.01	1.00
		Health information on the Internet	2	na	na	na	na	na
B General cognitive skills	4*		4*	282.6	43	<0.001	0.07	0.98
General life skills	10	Literacy, numeracy and self‐expression ICT skills	4 3	41.5 na	2 na	<0.001 na	0.13 na	0.99 na
Problem‐solving	1	Interpersonal skills and problem‐solving	4	43	2	<0.001	0.12	0.97
Ability to understand risk	4
Critical decision‐making	9	Critical decision‐making	5	172	11	<0.001	0.10	0.94
C Social roles	4*		2*	60.6	10	<0.001	0.06	0.96
Address social determinants of health	6	Address social determinants of health	3	na	na	na	na	na
Consumer competencies	4	Consumer competencies	4	0.9	2	0.63	<0.01	1.00
Adapt to work	1
Manage relations with significant others	1
D Medical management	3*		1*
First aid  Informal care	1 8	First aid and informal care	6	17.2	7	0.02	0.04	0.99
D Medical management [patient items]	3*		4*	87.4	64	0.03	0.04	0.98
Knowledge about condition/treatment Recognize and act on symptoms Adherence	9 2 1	Knowledge about condition/treatment, recognize and act on symptoms and adherence	5	7.1	3	0.07	0.06	0.99
Effective interaction with healthcare providers	13	Effective interaction with healthcare providers – patient communication skills Effective interaction with healthcare providers – communication with doctor	4 7	1 20.9	2 10	0.60 0.02	<0.01 0.06	1.00 0.98
Healthcare utilization	19	Healthcare utilization – assertive use	5	13.3	5	0.02	0.06	0.92
E Healthy lifestyle	3*		2*	9.7	8	0.29	0.01	0.99
Healthy lifestyle overall Nutrition and diet Non‐smoking	2 2 5	Healthy lifestyle overall, nutrition and diet and non‐smoking	4	0.89	2	0.64	<0.01	1.00
Exercise	3	Exercise	2	na	na	na	na	na
F/G Motivational skills	3*		0*
Manage psychological consequences	1
Self‐efficacy	1
Emotional intelligence	1
H Attitudes and values	1*		2*	25.7	7	<0.001	0.05	0.86
Identify personal preferences	11	Identify personal preferences  –Responsibility Identify personal preferences  –Consumerism	4 2	14.5 na	2 na	<0.01 na	0.07 na	0.68 na

d.f., degrees of freedom; RMSEA, root mean square error of approximation; CFI, comparative fit index; *, number of factors; na, not available due to insufficient degrees of freedom.

As an extension of Table 4, Table 5 presents the confirmed items and their factor loadings by confirmed competence and dimensional model. Competencies measured by two‐ or three‐item factors for which fit statistics could not be calculated are also included in Table 5.

Table 5.

Confirmed factors for the Swiss Health Literacy Survey in separate dimensional models

Dimension, competence and item			CFA
No.	Original competence	Item	Factor loading
A Information and knowledge
Access to health information
Q3	Access to information about health	Having enough information to choose treatment	0.64
Q5	Access to information about health services	Having enough information to choose doctor	0.69
Q6	Access to information about health services	Having enough information to choose hospital	0.75
Q7	Access to information about health services	Having enough information to choose sick fund	0.55
Seeking health information
Q1	Access to information about health	Number of sources of health information	0.40
Q8	Seeking health information	Seeking health information to avoid doctor visit	0.63
Q9	Seeking health information	Seeking health info to prepare for doctor visit	0.86
Q10	Seeking health information	Seeking additional health info after doctor visit	0.72
Q12	Seeking health information	Information seeking on food labels	0.24
Q14	Seeking health information	Information seeking on PIL	0.29
Ability to understand information on health and science
Q17	Ability to understand health information	Understanding information in the media	0.39
Q13	Seeking health information	Understanding information on food labels	0.43
Q15	Seeking health information	Understanding information on package insert	0.45
Q27	Knowledge about condition and treatment	Knowledge about lifestyle/prevention	0.34
Q28	Scientific literacy	Knowledge about science and technology	0.53
Health information on the Internet
Q11	Seeking health information	Internet a good source of health information	0.93
Q18	Ability to understand health information	Understanding information on the Internet	0.84
B General cognitive skills
Literacy, numeracy, and communication skills
Q29	General life skills	Reading	0.65
Q30	General life skills	Writing	0.73
Q31	General life skills	Numeracy	0.71
Q35	General life skills	Self‐expression	0.80
ICT skills
Q32	General life skills	Computer skills	0.74
Q33	General life skills	Internet skills	0.74
Q34	General life skills	Technical skills	0.90
Interpersonal skills and problem‐solving
Q36	General life skills	Problem‐solving	0.76
Q37	General life skills	Cross‐cultural skills	0.55
Q38	General life skills	Team work	0.60
Q39	General life skills	Learning new things	0.77
Critical decision‐making
Q48	Critical decision‐making	Choosing a sick fund	0.65
Q49	Critical decision‐making	Choosing a doctor	0.71
Q50	Critical decision‐making	Choosing a medication	0.52
Q51	Critical decision‐making	Choosing a medical treatment	0.50
Q52	Critical decision‐making	Choosing foods	0.53
C Social roles
Address social determinants of health
Q57	Address social determinants of health	Public involvement in health policies	0.30
Q60	Address social determinants of health	Knowledge of patients’ rights	0.78
Q61	Consumer competencies	Knowledge of consumer rights	0.80
Consumer competencies
Q55	Address social determinants of health	Knowledge about politics and current affairs	0.42
Q62	Consumer competencies	Having time to make choices	0.55
Q63	Consumer competencies	Overwhelmed by choices	0.44
Q64	Consumer competencies	Not enough information and support to make choices	0.42
D Medical management
First aid and informal care
Q67	First aid	First aid course	0.32
Q93	Informal care	Avoid a doctor visit by treating at home	0.79
Q96	Informal care	Avoid a doctor visit by visiting a pharmacist	0.47
Q92	Informal care	Avoid a doctor visit by waiting it out	0.78
Q111	Informal care	Self‐care for minor ailments	0.24
Q127	Informal care	Interest in more self‐care for minor ailments	0.32
D Medical management [patient items]
Knowledge about condition and treatment, recognize and act on symptoms and adherence
Q19	Knowledge about condition and treatment	Knowledge about health condition	0.59
Q20	Knowledge about condition and treatment	Knowledge about treatment	0.67
Q68	Recognize and act on symptoms	Monitoring symptoms	0.68
Q69	Recognize and act on symptoms	Acting on symptoms	0.66
Q70	Use medications correctly/adherence	Adherence	0.48
Effective interaction with healthcare providers – patient communication skills
Q58	Address social determinants of health	Membership in patient organization	0.35
Q71	Effective interaction with healthcare providers	Prepare questions for doctor	0.34
Q72	Effective interaction with healthcare providers	Ask doctor questions	0.70
Q73	Effective interaction with healthcare providers	Discuss personal problems with doctor	0.63
Effective interaction with healthcare providers – communication with doctor
Q16	Ability to understand health information	Understanding information from your doctor	0.52
Q23	Knowledge about condition and treatment	Knowledge about how long to take medicine	0.39
Q76	Effective interaction with healthcare providers	Your doctor listens carefully	0.78
Q77	Effective interaction with healthcare providers	Your doctor explains understandably	0.77
Q78	Effective interaction with healthcare providers	Your doctor takes time for questions	0.81
Q79	Effective interaction with healthcare providers	Your doctor offers various treatment options	0.54
Q80	Effective interaction with healthcare providers	Your doctor discusses treatment pros and cons	0.72
Healthcare utilization – assertive use
Q87	Healthcare utilization	Request additional diagnostic examinations	0.56
Q88	Healthcare utilization	Request additional medicines	0.59
Q89	Healthcare utilization	Request referral to another doctor	0.50
Q90	Healthcare utilization	Request referral to an institution	0.58
Q91	Healthcare utilization	Request a medical certificate	0.37
E Healthy lifestyle
Healthy lifestyle overall, nutrition and diet, and non‐smoking
Q109	Healthy lifestyle overall	Health consciousness	0.71
Q110	Healthy lifestyle overall	Activities to maintain health	0.52
Q113	Nutrition and diet	Eating healthy	0.61
Q118	Non‐smoking	Smoking – current	0.44
Exercise
Q117	Exercise	Member of a sports club	0.65
Q115	Exercise	Physical activity	0.32

Using information from both 4, 5, we comment on the competencies per dimension. Within the dimension of information and knowledge, EFA and CFA confirmed three competencies with good unidimensionality. The original competencies ability to understand health information and scientific literacy (measured by a single variable) loaded onto a new factor ability to understand information on health and science. Two questions related to health and the Internet (Q11 and Q18) loaded separately onto a new factor health information on the Internet, but the number of items is insufficient to calculate fit statistics in CFA. No indicators or items were found or created for the original competence ability to apply health information.

Within the dimension of general cognitive skills, the competencies could largely be replicated, albeit with moderate unidimensionality. The original competence general life skills loaded onto three separate factors: (i) literacy, numeracy and self‐expression, (ii) ICT skills, and (iii) interpersonal skills and problem‐solving. The original competence problem‐solving was measured by one question (Q36) and loaded with three interpersonal skills onto one factor. The original competence taking (health‐related) action was measured by two ‘patient’ items (Q40 and Q41) and excluded in EFA owing to poor fit. The items measuring the ability to understand (health‐related) risk loaded neatly onto its own factor in EFA but were excluded in CFA owing to instability. The competence critical decision making (for health) could be confirmed as a factor, but as with the other confirmed factors in this dimension, with CFI in the ideal range but RMSEA not.

Within the dimension of social roles, the competencies addressing social determinants of health and consumer competencies could be confirmed in a good two‐factor solution. Two items (Q55 and Q61) loaded onto a factor rather than its original competence. The competencies adapting to work and managing relations with significant others were each measured with a single question from the MOS SF‐12 and remained separate in EFA.

The dimension medical management included a large proportion of ‘patient’ items, which were assessed together in a separate model. The two original competencies recognizing and acting on symptoms and using medications correctly (adherence) loaded together with a third competence knowledge about condition and treatment (originally placed under the dimension information and knowledge) onto a single factor, which together constitute core self‐management skills. The items for effective interaction with healthcare providers loaded onto two separate factors: patient communication skills and doctor communication skills (as assessed by the patient). An item series on requesting additional services proactively from one’s doctor (placed under the competence healthcare utilization) loaded onto a single factor assertive use. In terms of the ‘citizen’ items, those for the competencies first aid (measured by a single item) and informal care loaded onto a single factor. All five factors demonstrated good unidimensionality.

A good two‐factor solution could be found for the dimension healthy lifestyle. The overall items loaded together with the items for nutrition/diet and non‐smoking onto a single factor with good unidimensionality. The items for exercise loaded onto a second factor.

The three competencies placed a priori under the dimension motivational skills were each measured using a single item. All the inter‐item correlations were below 0.10, suggesting no underlying dimension. Among the 11 items placed under the dimension attitudes and values, only six items loaded onto two factors: responsibility and consumerism. Although the two‐factor solution was good in EFA, CFA revealed poor unidimensionality for the factor responsibility and poor fit overall.

Cross‐cultural validity of the Swiss Health Literacy Survey

Table 6 presents the items whose correlation scores demonstrate statistically significant DIF, after correction for multiple comparisons, according to the MIMIC models. Compared to the Swiss German sample, nine items demonstrated statistically significant DIF for French‐speaking Swiss and 12 items for Italian‐speaking Swiss. Among French‐speaking Swiss, correlation scores were above 0.20 for three items – Q55, Q113 and Q127 – but adjusting for DIF revealed a large effect for only Q55, which exhibited a very large negative effect. This suggests that this item was not reflective of the overall construct for this group and pulled its factor‐level estimate for consumer competencies–coping with choice downward.

Table 6.

Differential item functioning (DIF) of confirmed items in the Swiss Health Literacy Survey by linguistic region in separate dimensional models

Dimension and competence		Significant regression scores suggestive of DIF**
Confirmed competence	No. items	Swiss French vs. German	Swiss Italian vs. German
A Information and knowledge	4*
Access to health information	4	–	–
Seeking health information	6	Q1 = 0.19, Q12 = 0.11	–
Ability to understand information on health and science	5	Q27 = 0.12	Q17 = −0.07
Health information on the Internet	2	na	na
B General cognitive skills	4*
Literacy, numeracy and self‐expression	4	Q31 = 0.14	Q29 = −0.08, Q30 = −0.08
ICT skills	3	na	na
Interpersonal skills and problem‐solving	4	Q37 = 0.06	Q39 = 0.05
Critical decision‐making	5	–	–
C Social roles	2*
Address social determinants of health	3	na	na
Consumer competencies	4	Q55 = −0.36	Q55 = −0.07, Q62 = 0.14, Q63 = −0.08, Q64 = −0.05
D Medical management	1*
First aid and informal care	6	Q93 = 0.13, Q127 = −0.27	Q67 = −0.17, Q127 = −0.11
D Medical management [patient items]	4*
Knowledge about condition and treatment, recognize and act on symptoms and adherence	5	na	na
Effective interaction with healthcare providers – patient communication skills	4	na	na
Effective interaction with healthcare providers – communication with doctor	7	na	na
Healthcare utilization – assertive use	5	na	na
E Healthy lifestyle	2*
Healthy lifestyle overall, nutrition and diet and non‐smoking	4	Q113 = 0.25	Q109 = 0.16, Q110 = 0.14
Exercise	2	na	na

*, number of factors; **, P < 0.05 after adjustment for family‐wise error with Hochberg’s sequential method; na, not available.

Among the Italian‐speaking Swiss, there were no correlation scores above 0.20 for any of the items, but for three competencies – (i) healthy lifestyle overall, nutrition and diet and non‐smoking, (ii) consumer competencies–coping with choice, and (iii) literacy, numeracy and self‐expression – at least half of their constituent items demonstrated DIF. Items Q109 and Q110 exhibited an inflating effect on the competence healthy lifestyle overall, nutrition and diet, and DIF adjustment lowered the factor level estimate considerably. The four items Q55, Q62, Q63 and Q64 exhibited effects in both directions, but the net effect pulled the factor‐level estimate for consumer competencies–coping with choice downward. Because all the items for this competence demonstrate DIF, they probably should not be used in cross‐cultural comparisons. Items Q29 and Q30 exhibited small negative effects, and DIF adjustment pulled the factor‐level score for literacy, numeracy and self‐expression upward.

Validity of the Swiss Health Literacy Survey

Given the limited focus and language‐specific nature of existing instruments measuring health‐related reading and the mode of data collection, we could not include them for use in validation. One important hypothesis, however, is that competencies for health, like life skills, would correlate strongly with educational attainment. Indeed, education is one of the two most important social determinants of competence level in HLS·CH. The other is cultural‐linguistic region. Future publications will explore these issues in detail. But briefly, people with higher education and German‐speaking Switzerland demonstrated higher competencies than their counterparts with clear gradient effects. This finding is supported by known differences in health status, health‐related behaviours, and literacy by education and cultural‐linguistic region. ³ , ⁸ People with higher education were indeed significantly more likely to use new innovations like the Internet and call‐centres for health reasons, lending support to the hypothesis that health literacy may be an important mediator of the relationship between education and health status. ⁴⁵

Discussion

While several definitions for health literacy have been forwarded in the literature, there have been no dedicated instruments that measure citizen‐ and patient‐centred competencies for health beyond basic health‐related reading. The Swiss Health Literacy Survey (HLS·CH) represents an initial effort to operationalize a core selection of generic competencies for health and measure indicators for their component knowledge, skills, motivation and attitudes in the general population, which, of course, includes a high proportion of people living with long‐term conditions in the community.

There is no consensus on the definition and the conceptual delimitation of health literacy. ²⁰ , ²² , ⁴⁶ As many similar concepts have functioned with a multiplicity of definitions and operationalizations (e.g. self‐management ⁴⁷ , informed decision making ⁴⁸ , shared decision making ⁴⁹), this lack of consensus, though lamented by some, is hardly unusual. Rather than forward a new definition, this endeavour has sought to identify and measure the operative ingredients in existing definitions (indeed, for not only health literacy but also related concepts), as identified by leading experts. Indeed, since the first version of the instrument was tested in 2006, the tendency in the literature has not been towards greater unity and focus, but expansion into further areas such as public health knowledge. ⁵⁰ Given the paucity and narrowness of empirical approaches to date, a broad approach to measurement may be particularly informative at this juncture.

Among the two historical strands in health literacy, ⁵⁰ , ⁵¹ this approach is closer to the public health asset model than the clinical risk factor one. Even so, this endeavour has benefitted from strong conceptual and operational work in patient‐centred concepts such as self‐management and shared decision making. As such, we can only concur with the statement that ‘a comprehensive approach to health literacy will include both clinical and public health approaches’. ⁵⁰

Rather than coin a new term, we call this body of competencies for health ‘health literacy’, to anchor this approach to the extensive body of work on competencies and modern literacies at the conceptual, empirical and policy levels. We are cognizant that this usage of ‘health literacy’ is broader than the term’s use in North America in the past decade and a half as the impact of low reading skills – either general (i.e. ‘health and literacy’ in Canada) or health‐related (i.e. ‘health literacy’ in the United States) – on health and use of healthcare services. This broader scope is actually more coherent with other modern literacies such as computer literacy, financial literacy and scientific literacy, which go beyond reading skills and information to include knowledge, skills and attitudes specific to that subject.

The psychometric analyses support the notion that health literacy is a multidimensional construct. The factors in the overall model do not correlate very strongly, suggesting distinct components. Furthermore, the factor information and decision‐making that corresponds best to most definitions of health literacy is itself only moderately unidimensional. Analyses of the two dimensions information and knowledge and general cognitive skills reveal nine distinct factors, four with high unidimensionality and three with moderate unidimensionality. Therefore, even ‘core’ health literacy appears to be made up of distinct components.

In separate dimensional models, many of the competencies could be confirmed by exploratory and confirmatory factor analyses, yielding factors with good unidimensionality. The items also appear to be cross‐culturally robust in the Swiss context. Although DIF was found in several factors, only six items – Q55, Q62, Q63, and Q64 in the competence ‘consumer competencies–coping with choice’ and Q109 and Q110 in the competence ‘healthy lifestyle, nutrition and diet, and non‐smoking’– proved problematic in comparisons between German‐ and Italian‐speaking Swiss. One of these six items Q55 was also highly problematic in comparisons between German‐ and French‐speaking Swiss.

Limitations

While the list of competencies was compiled based on the work of numerous leading experts in the fields of health literacy, self‐management, patient empowerment, patient education, and shared decision making, the choice of indicators and items was limited by several factors.

First, the list of generic competencies for health assessed in this project, though broad, is not exhaustive. There are competencies specific to certain diseases, treatments, settings, lifestyles and life course and could be operationalized in additional modules depending on the focus. As seen in research on quality of life and self‐management, both disease‐specific and generic measures have proven utility. We could not find or create any direct indicators for three of the 30 competencies – that is, ability to apply health information, ability to understand and give consent and navigating healthcare systems. Navigating healthcare systems is a broad competence and may be approximated by using items for other competencies.

Second, the decision to include measurements of all 30 competencies for health within one instrument required brevity. More objective indicators are available for some of the knowledge and skill dimensions. For example, the adult literacy surveys assess one competence such as reading prose text with a battery of graded exercises requiring nearly half an hour of live administration. Such a thorough approach was not possible for this project. Many of the objective test indicators can only be administered face to face, and the mode of data collection CATI also precluded the inclusion of briefer assessments of reading skills (i.e. using print materials) as well as a battery of questions on risk of both medications and lifestyle, which proved difficult to administer via CATI in pretesting.

With a few exceptions, the questions selected for HLS·CH were based largely on self‐reported behaviour, motivation, and attitudes and self‐assessed knowledge and skills. Evidence shows that self‐assessment of health‐specific and general competencies is highly correlated with more objective assessments and can even be very accurate for certain competencies such as computer skills. ³⁴ , ⁵² However, self‐assessed reading strongly underestimates poor or insufficient reading skills. ⁵² The items for the competence ability to understand health information yielded distributions that were closer to findings from the adult literacy surveys. Although the lack of objective assessments of reading and perhaps several other competencies may need to be addressed in future permutations, the advantage of the variables chosen was that they captured the citizens’ experiences and even occasionally solicited citizens’ assessments of resources/barriers in his/her context. Most of these variables are not gathered routinely in either health interview surveys or adult literacy surveys, yet they offer valuable additional information from the citizens’ and patients’ perspective. It must be noted that although the data were collected at the individual level, the intent was not to test individuals or even the population per se. Just as PISA is measured at the individual level, the aggregated results are used to assess the outcome of teaching, schools and education policy.

Third, while this marks a more comprehensive approach to health literacy by augmenting the scope of the competencies measured at the individual level, a truly comprehensive approach must also assess the demands placed in the various contexts relevant to health. There is already a large literature in the United States assessing the match between written materials in the health sector and their consumers, ¹⁴ but indeed, novel approaches to assess the demands posed by larger contexts such as hospitals and supermarkets also need to be carried out from a competencies perspective. Importantly, competencies will change as demands change, so any list of core competencies needs to be revisited and updated over time. The ‘dynamic and holistic’ nature of competencies makes repeated monitoring necessary.

Fourth, low response rate, although typical for such CATI surveys, was coupled with a significant proportion of non‐response owing to health or language issues. A lower percentage of people with little schooling and non‐Swiss nationals participated in the survey than are actually present in the resident population. The weights were not calculated to correct for these variables. While less relevant for the psychometric analyses presented in this study, these sources of potential bias will need to considered when interpreting prevalence estimates.

Subsequent analyses and publications will elucidate the content of the instrument as well as competence profiles for different populations. Perhaps the most important consequence of the multidimensionality of health literacy is that different people may be stronger at different components of health literacy. It means going beyond global scores and looking at finer assessments of component competencies, which may prove more informative for practitioners and researchers alike. However, this data set will not permit analysis of the relationship between objective assessments of reading skills (i.e. using print materials) and other competencies nor address the responsiveness of the items chosen – that is, ability to detect change (after an intervention, for example). These issues need to be addressed in future studies.

Salience of competencies for health

The competencies approach underlines some important shifts for experts in the health field.

• 1Firstly, competencies exist along a continuum. Most often, competencies are not being assessed in terms of presence/absence, but rather degree – that is, how much. The threshold of how much is enough needs to be considered carefully within the context of demands and resources.
• 2Secondly, our complex societies and healthcare systems require multiple competencies, and not just one, even one as essential as reading (i.e. working with print materials). The importance/centrality of reading should not be diminished, but neither should the focus and measurement of health/patient education be limited to reading.
• 3Thirdly, competencies can be taught and learned. While competencies are distributed unevenly within a population, often along socio‐demographic determinants, unlike most of those characteristics that cannot be readily altered, competencies can be taught, facilitated, learned and practised. It is often taken for granted that most citizens and patients possess these competencies for health, but as no one is actually born with any of the competencies listed in Table 3, the questions beg to be asked: ‘Where are these competencies for health actually being taught, and where are they usually being learned?’
• 4Fourthly, competencies are measurable, and as such, health literacy has been forwarded as a measurable outcome of both health promotion and patient empowerment. Analogous to modern educational curricula which have moved from focusing on course content to competencies to be acquired, activities in (patient) empowerment and health/patient education may benefit by being more explicit about the competencies that they seek to target and improve.

For these reasons, a competencies‐based approach can enrich current understanding of health inequalities and social determinants of health by capitalizing on a ‘common currency’ of competencies to span the chasm between patient education and health promotion, between health and education/work sectors, between school‐based learning and adult/lifelong learning, with clear implications for policy and action.