Patient Education and Counseling 75 (2009) 340351
Contents lists available at ScienceDirect
journa l homepage: www.e lse1. Introduction
Health literacy has been a focus for attention in North Americasince the 1990s and is now listed as one of the four key priorities inpublic health in the US . More recently it has become recognizedas a health issue in Europe and elsewhere. For example, work hasbegun to appear in the British medical literature highlighting howhealth literacy is central to involving patients in their care and toreducing health inequalities . Health literacy is a broad andcomplex concept, which has been dened in a number of waysincluding the capacity to obtain, process and understand basichealth information and services needed to make appropriatehealth decisions , the cognitive and social skills whichdetermine the motivation and ability of individuals to gain accessto, understand and use information in ways which promote andmaintain good health , and as a critical empowerment
strategy to increase peoples control over their health, their abilityto seek out information and their ability to take responsibility .
Literacy skills and numeracy skills are two key components ofthis wider concept of health literacy , and are the focus of thepresent review. One study has dened the reading element ofliteracy as the ability to understand texts and to obtain informationfrom different sources including signs and symbols , but a widerperspective would also include writing. Numeracy skills refer tothe ability to understand and use numerical, graphical andmathematical information . A national survey of the literacyand numeracy skills of working age adults in England found that16% had limited literacy and 47% had limited numeracy . Thereis no one agreed denition of what constitutes limited literacy, andour use of this term should simply be taken to mean having feweror lower literacy or numeracy skills than other people. Althoughliteracy levels are associatedwith education, ethnicity and age  anumber of studies have shown that having limited literacy ornumeracy skills acts as an independent risk factor for poor health, through medication errors [12,13] and poorer under-standing of disease and treatments . A recent systematicreview of the relationship between literacy and health outcomes
Received 2 October 2008
Received in revised form 23 December 2008
Accepted 8 January 2009
Objective: To evaluate the published literature on the effects of complex (multi-faceted) interventions
intended to improve the health-related outcomes of individuals with limited literacy or numeracy.
Methods: Weundertook a systematic review of randomized and quasi-randomized controlled trials with
a narrative synthesis. The search strategy included searching eight databases from start date to 2007,
reference checking and contacting expert informants. After the initial screen, two reviewers
independently assessed eligibility, extracted data and evaluated study quality.
Results: The searches yielded 2734 non-duplicate items, which were reduced to 15 trials. Two
interventions were directed at health professionals, one intervention was literacy education, and 12
were health education/management interventions. The quality of the trials wasmixed, 13/15 trials were
conducted in North America, and all focused on literacy rather than numeracy. 13/15 trials reported at
least one signicant difference in primary outcome, all favoring the intervention group. Only 8/15 trials
measured direct clinical outcomes. Knowledge and self-efcacy were the class of outcomemost likely to
Conclusion: A wide variety of complex interventions for adults with limited literacy are able to improve
some health-related outcomes.
Practice implications: This review supports the wider introduction of interventions for people with
limited literacy, particularly within an evaluation context.
2009 Elsevier Ireland Ltd. All rights reserved.
* Corresponding author at: Health Service and Population Research Department
(PO29), David Goldberg Centre, Institute of Psychiatry, Kings College London, De
Crespigny Park, London SE5 8AF, UK. Tel.: +44 20 7848 0739; fax: +44 20 7277 1462.
E-mail address: firstname.lastname@example.org (S. Clement).
0738-3991/$ see front matter 2009 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.pec.2009.01.008Complex interventions to improve the hA systematic review
Sarah Clement a,*, Saima Ibrahimb, Nicola CrichtonaHealth Service and Population Research Department (PO29), David Goldberg Centre, Inb Institute of Primary Care and Public Health, Faculty of Health and Social Care, Londonc Institute for Healthcare Studies, Division of General Internal Medicine, Feinberg School
A R T I C L E I N F O A B S T R A C Talth of people with limited literacy:
Michael Wolf c, Gillian Rowlands b
tute of Psychiatry, London, UK
uth Bank University, London, UK
Medicine, Northwestern University, Chicago, USA
vier .com/ locate /pateducou
S. Clement et al. / Patient Education and Counseling 75 (2009) 340351 341concluded that limited literacy is related to several adverse health-related variables, including knowledge about health and healthcare, hospitalization, global measures of health and some chronicdiseases . Qualitative research has demonstrated the shameand practical difculties that patients with limited literacy canexperience when interacting with the health care system, as wellas the coping strategies they employ to circumvent these [15,16].These ndings have intensied the growing international recogni-tion of this highly prevalent problem and the need for action.
To date, researchers and practitioners have used a number ofapproaches to attempt to lessen the impact of limited literacy ornumeracy on health outcomes. Some of these strategies can betermed simple interventions, such as the use of simplied writtenlanguage, pictorial material, and audio/audiovisual resources.Other strategies are classied as complex interventions. A complexintervention is dened as one that comprises of a number ofseparate elementswhich seemessential to theproper functioningofthe intervention although the active ingredient of the interventionthat is effective is difcult to specify . . . The componentswill usuallyinclude behaviors, parameters of behaviors (e.g. frequency, timing),andmethods of organizing and delivering those behaviors (e.g. typeof practitioner, setting) . . . [and] may be at the level of individualpatient care, . . . organizational or service modication . Acomplex interventions to improve the health of people with limitedliteracy or numeracy might, for example, provide health educationthrough a combination of verbal presentation, pictorial materialsand checking for understanding, ormight haveoneprimaryelementthat is in itself multi-component or involves human interaction,such as literacy education. By contrast, examples of simpleinterventions in this eld include the sole use of a pamphlet writtenin simplied languageor the soleuseof aDVD. Existingevidencehasdemonstrated that simple interventions responding to limitedliteracy, such as the use of audiotapes, have been met with variablesuccess with no consistent ndings regarding whether thisintervention increases knowledge, produces behavior change orreduces distress . We postulate that complex interventions arelikely to be needed to effectively tackle the complex health-relatedneeds of peoplewith limited literacy or numeracy, a viewwhich canalso be inferred from the work of others [2,19].
To our knowledge, no existing systematic review has beenconducted specically focusing on complex interventions toimprove the health of people with limited literacy or numeracy.Coulter and Ellins [2,20] conducted a policy overview rather than asystematic review, and only a small part of the work examinedliteracy. One systematic review focused on one type of simpleintervention, namely audiotapes . Pignone et al.  havesystematically reviewed a mix of interventions, the majority(n = 15) being simple and the minority (n = 5) complex. Theirreview was restricted to studies published in English, fromdeveloped countries, and those where a formal assessment ofliteracy had been made. Furthermore, only studies publishedbetween 1980 and 2003 were considered, and as health literacyresearch is a vastly expanding eld, a signicant number of studieswill have been published since 2003.
The aim of our review was to evaluate the published literatureon the effects of complex interventions intended to improve thehealth-related outcomes of people with limited literacy ornumeracy.
2.1. Search strategy
Eight electronic databases were searched: Medline (1966);Cumulative Index to Nursing & Allied Health Literature (CINAHL,1982); Cochrane Central Register of Controlled Trials (CENTRAL,1800); PsycINFO (1887); SCOPUS database (1966); BritishEducation Index (1975); Educational Resources InformationCenter (ERIC, 1966), and Australian Education Index (1979).Searches were undertaken in March and April 2007 and eachdatabase was searched from its earliest date. No languagerestrictions were used. The search strategy for Medline usedsubject heading and textword searching, combining terms asfollows ($ indicates truncation): ((literacy-related terms, e.g.reading/, literac$) OR (numeracy-related terms, e.g. mathe-matics/, numera$) OR (educational terms, e.g. educationalstatus/)) AND (the set of terms for identifying clinical trials fromthe Cochrane Handbook ). The searches for the other databaseswere based on the Medline search strategy, but for the non-healthdatabases a set of health terms (e.g. health/, disorder$, patient$,nurs$) were added to limit the studies to those with relevance tohealth. Full details of the electronic searches are available as anelectronic supplementary le to this paper.
The reference lists of all studies included in our review, and ofother relevant reviews [20,21] were inspected to identify furtherpossible studies. Three groups of expert informants werecontacted: the authors of all of the included studies; researchersknown to be working in the eld of health literacy; and the leads ofongoing studies in the area identied through searches on theNational Research Register, the Health Services Research Projectsin Progress database, and the Cochrane Central Register ofControlled Trials. The 78 expert informants were asked to identifyany published studies likely to meet the inclusion criteria that hadnot yet been included in the review.
2.2. Selection of studies
The titles and abstracts of the studies yielded by the searchwere screened to exclude those that clearly did not meet theinclusion criteria. Initially, 150 studies were independentlyscreened by two researchers (SI and SC) then discussed to aidreliability, with the remaining studies being screened by oneresearcher (SI). Full papers were obtained for papers not screenedout. Two researchers (SI and SC) independently assessed thesepapers against the inclusion criteria, with any differences ininclusion decisions being resolved via discussion, with a thirdresearcher (GR) acting as arbiter.
The inclusion criteria are shown in Table 1, and aimed to selectpublished papers reporting randomized controlled trials (RCTs) orquasi-randomized controlled trials of complex interventionsintended to improve outcomes for people with limited literacyor numeracy, which included at least one health-related outcome.
2.3. Data extraction, quality assessment and synthesis
Data on study characteristics, study quality and ndings wereextracted independently by two researchers (SI and SC) fromunmasked publications. Any discrepancies were resolved viadiscussion and arbitration with GR and MW, and statistical issuesresolved with statistician NC.
The quality of the studies was assessed using the Delphi List, which we modied by deleting the two criteria relating toblinding of patients and professionals (as these are rarely possiblein the types of intervention reviewed here), and replacing themwith two criteria from the CONSORT statement , namelyinclusion of an a priori sample size calculation and of a participantow diagram.
Statistical aggregation of ndings was deemed inappropriategiven the variety of different measures of outcome used and therange of time periods to follow-up. Consequently a narrativeanalysis was undertaken with ndings presented in tabular form,with supplementary data in the text. The synthesis covered both
S. Clement et al. / Patient Education and Counseling 75 (2009) 340351342overall study populations and subgroup analyses performed ondata from participants with limited literacy/numeracy where thelatter has been reported by the study authors. We have restrictedour reporting of the effectiveness of the interventions at the nal
Table 1Inclusion criteria.
Data type Primary research, quantitative data
Participants Adults (including adults consulting on behalf of dependents,
and professionals who may be the target of an intervention,
all participants on whom outcomes are reported must
Health condition Any or none
Intervention Complex intervention (more than one element) ANDintended to improve outcomes for people with limited
literacy/numeracy, evidenced by either:
mention of literacy or numeracy in the description ofthe population or
mention of literacy or numeracy in the description ofthe intervention
Comparator Any (active or inactive control)
Outcomes At least one of the following health-related outcomes
Clinical outcomes (physical or psychological) Health knowledge Health behaviors Self-reported health status/quality of life Self-efcacy/condence in relation to health/health
Utilization of health care Health professional behavior
Study types Randomized controlled trials Quasi-randomized controlled trials
Published journal paperfollow-up point and focused on outcomes dened by the trialists asprimary, although summary data are also presented on secondaryoutcomes. We were unable to report test-values to accompany pvalues where these were missing from the original papers.
3.1. Search results
The searches yielded 2734 non-duplicate items, which werereduced to 17 papers  reporting 15 trials, as described bythe ow diagram in Fig. 1. Eleven expert informants replied to therequest for additional studies, suggesting 46 studies, of which 10had already been screened out and 36 had not been previouslyidentied but were screened out or excluded as they did not meetinclusion criteria. References for all excluded studies are availablefrom the corresponding author.
3.2. Trial characteristics and study designs
The trial characteristics, primary outcome(s) measured andresults for these primary outcomes are shown in Table 2.