Our literature search yielded 6206 publications between 1995 and mid-2018. Of these publications 6042 (97%) were excluded based on title and abstract because they did not fulfil one or more of the inclusion criteria: 3950 (65%) were excluded because not describing an intervention, 1037 (17%) did not meet the first criterion (being conducted in one or more of the European Member States), 1024 (17%) of the European interventions were excluded because they were not focusing on health literacy and 26 (1%) of the studies were excluded because there was no health literacy outcome measure (see also Fig. 1, PRISMA diagram). The remaining 164 publications were retrieved in full text for further assessment, of which 141 failed to meet the inclusion criteria. The main reason for excluding full texts was that they were unrelated to health literacy. Finally, 23 articles were included.
There were not a sufficient number of studies with similar outcome measures or similar interventions to consider quantitative analysis (meta-analysis or statistical pooling) of data; therefore a qualitative analysis was performed. The 23 included intervention studies and their characteristics are summarized in Table 1. The references, the evidence tables, the intervention type and outcome of each of these studies can be found in Table 2. All studies were conducted in North-western Europe, no studies from Eastern and Southern European countries were found. All studies except one  were interventions developed for adults.
Health literacy measure
Studies varied considerably in their measurement of health literacy. Commonly used instruments in the USA to assess health literacy such as the Raped Estimate of Adult Literacy in Medicine (REALM) [18, 19], the Newest Vital Sign (NVS) , Test Of Functional Health Literacy in Adults (TOFHLA) [21,22,23], and the Short Assessment of Health Literacy (SAHL) [23, 24] were used in eight studies. All these measures focus on functional health literacy. Four other studies also focused on functional health literacy skills by assessing reading ability [25, 26] or the level of mild intellectual disability . Three studies measured critical health literacy skills by questionnaire  or interview  or assessing skills towards decision-making . The study by Webb et al.  focussed on functional and interactive health literacy skills by measuring health literacy as the level of verbal and cognitive abilities. Three studies measured health literacy by the level of disease-specific knowledge [20, 32, 33]. One study measured health literacy by the Brief questions to identify inadequate health literacy . In one study the way health literacy was measured was not specified . Numeracy was assessed in four studies [35,36,37,38].
Type of intervention
There was also a huge variation in the type of interventions given: group interventions, individual interventions, web-based interventions, one component interventions (e.g. an information leaflet) and multi-component interventions including chat-groups, lectures, training sessions, a help-desk, computer programs and leaflets among others. Most interventions were web-based interventions (n = 9). The web-based interventions were conducted during the most recent years, most of them (n = 7) in 2015, 2016 and 2017. In only one study it was explicitly mentioned that the patients were involved in the development of the intervention on a module designed for the development of a decision aid about MS-immunotherapy .
Five studies used a Randomised Clinical Trial (RCT) design [19, 23, 26, 27, 34] and five studies a Controlled Clinical Trial (CCT) design [18, 21, 24, 25, 33]. In two studies two groups were compared pre- and post-test (Cohort analytic design) [17, 28], but most studies (n = 9) used the same group that was pretested and post-tested immediately after the intervention (Cohort study). The study by White et al.  used an interrupted time series design and there was one observational study. The type of design in combination with the frequent missing or nor reported use of covariates makes that for most studies (n = 15) the quality was rated as weak (EPHPP , Table 2). The quality of seven studies was rated as moderate on the base of the EPHPP  assessment tool and one study was judged as strong.
Types of intervention studies
The types of interventions in the 23 studies could be categorized as follows;
Interventions aimed at improving (aspects of) the health literacy level of individuals.
Interventions that were specifically tailored to different health literacy levels.
General interventions that aimed at improving health outcomes, which described the specific effects for patients with different health literacy or numeracy levels.
Interventions aimed at improving (aspects of) health literacy
A group training of 2 × 2,5 days in evidence based-medicine for patients, patient counsellors, consumer representatives and healthcare professionals resulted in a significant increase in health related knowledge and in the level of critical health literacy of the participants . In the evaluation of the training they stated that they had broadened their knowledge, were more critical in handling health information and considered themselves more confident on making the right decisions on the basis of the information they found. The content of the training was tailored to the needs of the participants. A second group intervention  specifically focused on so-called ‘hard to reach’ groups (e.g. unemployed women of minority groups and female migrants from Islamic backgrounds). This intervention combined different elements: computer courses, lectures, and language training. Topics related to health and well-being were being discussed. Also this intervention led to an increase in knowledge and comprehensive health literacy. Another group intervention targeted patients with mild intellectual disabilities and was tailored to their verbal and cognitive abilities. In the training, patients were taught how and when to access healthcare . The evaluation showed that the intervention had a significant positive effect on the participants’ ability to recognize disease symptoms, identify illnesses and choose appropriate courses for action.
An intervention that was developed to improve self-care among diabetes patients was evaluated after two years. The patients had received tailored tele-carer education as well as support to change specific lifestyle behaviours . The evaluation showed that these diabetes patients were better able to use knowledge in their day-to-day self-care and expressed a greater control over their self-care decision-making. A UK community study that evaluated the impact of a self-care skills training initially (after 6 months) found a positive effect on decision making skills regarding use of health services (critical health literacy). However, after 12 months the effect was no longer found . In three studies in Denmark, the tele-homecare intervention 'Telekit' was evaluated. The Telekit focuses on the management of COPD in general, how to manage COPD during exacerbations and collect date on the current state of the patient’s health. Both studies did not found a significant difference on functional health literacy [21,22,23]. The Telekit increases the feeling of insecurity, greater freedom, more control and greater awareness of symptoms .
Five interventions specifically focused on the improvement of numeracy skills, i.e. the ability to understand numerical risk information [18, 21, 35,36,37]. The evaluations of these interventions had similar conclusions. In general numerical information is presented in ways too difficult for people with low competencies. Another way of presenting (e.g. by using visual aids and/or lowering the level of detail of information) led to improved understanding in participants with low numeracy competencies.
Interventions tailored to different health literacy levels
Three studies [17, 27, 28] performed an evaluation with an intervention and a control group, comparing the outcome variables. In one study among children with diabetes (age 8–12) the impact of a personalized robot on diabetes knowledge and motivation for self-management was compared to a neutral robot. The reactions of the personalized robot were adjusted to the knowledge level of the child. In the evaluation, children in the intervention group (with the personalized robot) scored higher on diabetes knowledge and motivation for self-management. A tailored training programme on peritoneal dialysis for renal patients with low health literacy resulted in lower incidence of peritonitis and stronger feelings of control and ownership over treatment among the participants in the intervention group, as well as less supervision time needed of nurses . The intervention comprised lowering the amount of written information and using more verbal material, and reducing the use of medical jargon. A computer-tailored intervention for smoking-cessation (booklet and web-based programme) was compared to a general self-help booklet. The tailored approach led to more attempts to quit smoking as well as higher abstinence rates, specifically for participants with lower literacy levels . An intervention that was tailored to the verbal and cognitive abilities of patients with mild intellectual disabilities was evaluated in a one group pre/post-test design . The evaluation showed there was an improvement in symptom recognition, better health-related decision making, improved understanding of medical procedures and a better ability in formulating personal health goals.
Three studies focused on the way of presenting information to persons with different health literacy levels [24, 25, 34]. One study varied in presenting information on spoken versus written text and illustration versus animation. In almost all conditions, the high health literate persons had a better recall on information compared to the low health literate persons, except for the spoken animations. In the spoken animation condition, the low health literate persons recalled the same amount of information as the high literate persons. The other study varied in presenting information on illustrated versus text-only and in not difficult versus difficult texts. Persons with low and high health literacy recalled the not difficult information better than difficult information. Illustrated text improved the recall and attitudes in low health literate persons and had no effect in high health literate persons. Another study stated audio-visual leads to better knowledge. The study also stated that clear, person-based intervention development is more important than interactivity and audio-visual presentation to improve health literacy outcomes.
General interventions that aimed at improving health outcomes, which described the specific effects for patients with different health literacy or numeracy levels
In general, patients with low health literacy benefit less from general interventions compared to patients with higher levels of literacy, e.g. with respect to understanding medication labels [18, 35] and other health messages [19, 20]. In a study on the knowledge level of rheumatoid arthritis patients after being exposed to a pictorial ‘mind map’ together with a arthritis campaign booklet, analysis showed that less literate participants gained fewer knowledge from both the booklet alone and the booklet in combination with the mind map, compared to high literate patients . Similarly, a leaflet was developed to improve gynaecological cancer symptom awareness and to reduce barriers to access medical services . Though in general after reading the leaflet awareness improved and barriers to access medical services were reduced, these effects were less in patients with lower health literacy. In general, patients with low levels of health literacy were found to experience more barriers to access health care services
Four studies reported on outcomes relevant for the daily management of chronic illness or health in general such as knowledge, empowerment, ability to self-manage, decision-making skills, ability to taken an active role in treatment. Increased levels of health literacy were associated with higher levels of empowerment, better decision-making skills, and a more active role in treatment [29,30,31,32]. The evidence were graded as weak due to the fact that results mainly came from uncontrolled studies and results were often based on small groups or a limited number of observations. One study focused on a mobile phone app intervention targeting fruit and vegetable consumption. The information provided via the app where either textual or auditory tailored to the person’s characteristics. The app increased the fruit and vegetable intake, but only in persons with high health literacy .