How best to use and evaluate Patient Information Leaflets given during a consultation: a systematic review of literature reviews

Abstract Background In the past, several authors have attempted to review randomized clinical trials (RCT) evaluating the impact of Patient Information Leaflets (PILs) used during a consultation and draw some general conclusions. However, this proved difficult because the clinical situations, size and quality of RCTs were too heterogeneous to pool relevant data. Objective To overcome this 30‐year stalemate, we performed a review of reviews and propose general recommendations and suggestions for improving the quality of PILs, how to use them and methods for evaluating them. Methodology We searched five databases for reviews, systematic reviews and meta‐analyses describing PILs. We drew general and condition‐linked conclusions concerning the impact of PILs. Checklists summarize criteria for quality PILs, and ways of using and evaluating them. Results Of 986 articles found, 24 reviews were pertinent; the five oldest considered the impact of PILs irrespective of the condition the patient consulted for; the 19 more recent ones mostly addressed precise clinical situations. Discussion Whatever the clinical situation, PILs improve patients' knowledge and satisfaction. For acute conditions, in the short‐term PILs also improve adherence to treatment. For chronic diseases, invasive procedures or screening situations, their impact on adherence varies depending on the context, how the PILs are given and the invasiveness of the intervention. Conclusion PILs are considered to be very useful, especially for acute conditions where the patient is the first to suffer from lack of information. We propose checklists for writing, designing, using and evaluating PILs in RCTs to enable comparisons between different studies.


| INTRODUCTION
Since the 1970s, various authors have investigated the use of Patient Information Leaflets (PILs) [1][2][3][4] and have suggested that they are helpful for patients, particularly as they improve recall of what was said during the consultation. 2,5,6 Although more and more information is available through the Internet, patients continue to ask for more written information. 7,8 However, the availability of PILs does not necessarily guarantee access to quality information tailored to the needs of each patient. [9][10][11][12] In the 1990s, Dixon and Park underlined the importance of developing recommendations to improve the quality of PILs. 13 Although health-care institutions and the research community have developed guidelines to help create PILs, 6,14-16 the use of these is rarely reported by the medical and research community. Other authors have looked at how PILs are used in everyday practice. 4,16,17 There is general agreement that PILs should be handed out by the physician at an opportune moment during the consultation, 6,18 should target patient expectations 10,16,[18][19][20][21] and that the form should take into account the patient's preferences. 18,22 The PILs should back up what the physician says 6, [16][17][18] but should in no way be a substitute for oral information, preferred by the majority of patients. 4,[16][17][18] However, research protocols generally do not take these considerations into account: the PILs is sometimes distributed by the nurse, 23,24 sometimes by a pharmacist, 25,26 by a clinical research assistant or by another person. Sometimes it is sent by email 27 or by post 28 in spite of the fact that informing the patient is now considered by medical institutions as the physician's responsibility. 14,29 In view of problems such as these, the most recent review of "generalist" literature on PILs (i.e. not specific to a given clinical situation) conducted in 1998 16 stressed the need for further research in general, and in particular using randomized clinical trials (RCTs).
Since then, many groups have used RCTs to assess the impact of PILs in specific clinical situations, such as in chronic illness, 22 contraception 30 screening, 31 chest pain in the emergency room, 23 preparing for surgical interventions 9 or in consultations with a primary care physician. 1,2,32 However, problems of heterogeneous research protocols remain, both in the choice of primary outcome and in the main measurement technique, resulting in conclusions that are sometimes contradictory. 32 Studies concerning a given condition have been reviewed within the appropriate field.
We have attempted to summarize the diverse reviews, both general and specific to given conditions (literature reviews, systematic reviews and meta-analyses), made to date. We clarify the impact of PILs by evaluating their effect on main outcomes, and specify their prescription according to condition and terms of use. In addition, we propose a checklist for writing, designing and using PILs with recommendations for the standardization of research protocols that assess PILs.
Filters used were "meta-analyses", "literature reviews", "systematic reviews" without temporal or language restrictions. By chance, all the reviews found were in English. After merging the results from the different search engines, duplicate publications were removed. The relevant free access articles and those available through our university or national research organisations were recovered. Otherwise, if the title or abstract were relevant, the authors were contacted by email and if there was no response, their articles were ordered from the publisher. Review articles cited in extracted articles were also used. For our purposes, the PILs should contain information on the disease for which the patients consulted. Hence, we excluded reviews of leaflets aimed at multifaceted studies in which no leaflet-specific effect could be extracted, reviews concerning decision aids, or on patient consent documents and reviews of patient empowerment tools.
To extract all relevant articles, two primary care physicians (MT and JT) separately assessed all articles found by the search engines using the titles and abstracts. Only articles selected by both were retained, and when they disagreed, the abstracts were reassessed by one and checked by the other. Selected abstracts were discussed with an expert (MS) in PILs and disagreements resolved by consensus. We checked the reference lists of the selected reviews for additional relevant publications.

| Data extraction
Data were extracted from the full texts of the selected reviews by the two primary care doctors working independently. A standardized form was used to record the relevant characteristics of the included reviews: methodology, condition studied, population, intervention, outcome measures, study quality, the number of articles included in the review, the total number of patients (if available) and main conclusions. We (MS and AG) checked the concordance rate between the doctors and resolved any disagreement by consensus (MT, JT, MS and AG).
To anticipate and resolve disagreements regarding the terminology for the different outcomes (for example, for some authors "adherence" relates only to drug adherence, 33 for others, it includes respect of the drug regime, lifestyle changes and changes in diet 34 ), we built a framework using a multidisciplinary phenomenological patient-centred approach by grouping outcomes according to the type of impact

| Methodological quality assessment
We evaluated the methodological quality of included reviews using the criteria of the Cochrane Handbook for Interventional Systematic Reviews and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Each review article was rated as "low quality", "good quality" or "very good".

| Article selection
Our search identified 986 unique records of which 950 did not meet our inclusion criteria following the screening of titles and abstracts. Of the remaining 36 articles, after evaluation of the full text, only 24 met the inclusion/exclusion criteria (Fig. 2). The rate of concordance for data extraction between the two doctors was 95%.

| Review characteristics
Between 1990 and 2012, five literature reviews evaluated the "general" impact of PILs on patients, whatever the medical conditions: the oldest three focused exclusively on PILs 5,6,16 while the two more recent looked at the impact of any form of information on patients. 9,20 Since 2012 (up to August 2015), all the reviews we found were specific to a given situation or condition. Seven reviews looked at information on drugs (including one on contraceptives and one on drugs for psychiatric disorders), three reviews concerned PILs for cancer patients, three were on PILs intended to be given before a screening examination or surgery, three were about common acute conditions and three on chronic diseases. The main characteristics of all these reviews (outcome measures, etc.) are summarized in Table 1.
According to the PRISMA checklist for study quality, 10 were of very good quality, nine were good and five were of poor methodological quality.    19,42 However, the extent of this impact also depends on the clinical context and/or invasiveness of the intervention (improvement in knowledge from 18% to 57% depending on the case) 9,31 and the timing of receiving the information. 4,6,9,42 For example, in the case of cancer, written information in the form of new patient information booklets or packages improved patient knowledge and reduced confusion especially if it was provided to the patient before their first clinical appointment rather than at the first appointment.

| The impact of PILs on the psyche, cognitive and emotional state of the patient
Before a surgical procedure, 95% of patients want to be informed about the associated risks 9. However, the information is misunderstood T A B L E 1 ( C o n t i n u e d ) 25% of the time, and patients deem the information inadequate (15% of leaflets) or perceive the leaflet as simply a way to protect the judicial rights of the surgeon. 9 Conversely, immediately after surgery may be an inappropriate moment to give written information because at this point patients seem to retain very little of the information provided. 9 Few studies have evaluated the impact of PILs on knowledge beyond 15 days after reception. 9 However, it is agreed overall that PILs improve patient satisfaction. 5,10,17,20 Many other psychological dimensions (sense of personal selfefficacy, decisional conflict, doctor-patient communication, empathy, trust, listening, etc.) have also been explored over the past 20 years, but as definitions of outcomes and the tools used to assess them vary between teams and disciplines, we were unable to draw clear conclu-

| The impact of PILs on patient behaviour
The dimensions studied were adherence (nowadays the term adherence is preferred to compliance) to treatment or lifestyle, return to work, rates of reconsultation, absenteeism and participation in screening. When used prior to surgery, the PILs can sometimes lead patients to refuse surgery (3.2% to 14.6% depending on the procedure). 9 Similarly, in the context of screening, PILs can lead to acceptance or potentially to refusal to undergo an examination. For example, one study showed that PILs could improve consent to screening with a prostate-specific antigen test but did not improve the acceptance of screening involving a digital rectal examination. 31 Another concluded that before a colposcopic procedure they were useful for obtaining patient consent. 40 Nevertheless, in a review of studies on screening programmes for several conditions, Fox 31 found that PILs had no clear effect on screening uptake. Due to their potential influence in decisions, PILs must be thoughtfully provided and used with care.
For common conditions, PILs improve adherence to medication and advice. 32 PILs also decrease the number of repeat visits to the primary care physician. 32 For example, in the case of lower back pain, 39 the precise instructions given by the leaflet boosted patient confidence, and improved beliefs in the effectiveness and adherence to short-term exercises but did not improve the rate of attendance at appointments.
In the context of chronic conditions, the concept of clinical inertia has recently appeared in the literature. This refers to inappropriate behaviour documented in evidence-based studies in given clinical situations (diabetes, hypertension, dyslipidaemia) and focuses on the determinants from both the patient's perspective and the physician's perspective. 42   1. For cancers, PILs diminish levels of anxiety. 7 It has been noted that for cancer in particular, information must be tailored to the needs of the individual patient to achieve better outcomes. 10,20 2. Prior to an invasive procedure, some authors find that PILs can increase anxiety, 9 while others find they have no effect. The active information reduced levels of anxiety and depression, but without reducing the number of cases of depression. In contrast, passive information increased anxiety.

| Impact on physician's behaviour
Patient Information Leaflets should be nuanced, distributed thoughtfully and personalized by the physician during the consultation. They must be hand delivered and must be considered in the same way as a medical prescription. Three of four studies presenting data on the prescription of antibiotics in primary care showed significant reductions in the number of prescriptions for groups receiving leaflets. 16,32 Among the reviews we studied, only De Bont et al. 32 looked at the physician's prescribing behaviour, despite the fact that their behaviour has a direct impact on the patient in the quality of care (reduction in the number of redundant prescriptions) and on public health (through costs incurred). There is a lack of research into this aspect that merits greater study. For long-term treatments, physicians should consider using other educational tools to supplement PILs.

| Towards the standardization of PILs
In  (Table 2).
T A B L E 2 Checklist for quality Patient Information Leaflets (PILs) according to the current literature

| Towards the standardization of research protocols evaluating PILs
When examining data from the literature, it is difficult to know whether the lack of impact of a patient information leaflet is related to the quality of the leaflet itself, the way it was used, the precise clinical situation in which it was tested or the quality of the research protocol. To overcome this impasse, it seems essential that researchers use consensually accepted standardized tools, including procedures for drafting PILs (such as those recommended here) ( to produce a synthesis of the data or to define a threshold of effectiveness of PILs by dimension and clinical situation. These endpoints are nevertheless essential in the evaluation of any PILs. There is a need to develop generic scoring systems independent of the pathological context so as to allow comparison between studies.

| CONCLUSION
When well written and used at the appropriate time, PILs can improve patients' knowledge and/or patients' satisfaction whatever the clinical situation and induce better adherence to treatment, to diet and to

T A B L E 3 Points to consider when evaluating Patient Information Leaflets (PILs)
Type of PILs PILs written and designed according to a defined methodology and/or complying with the guidelines (see Table 2) 12,14,15 Way of using PILs Hand delivered at the same time as verbal information (or, if sent prior to a consultation, by post or email, at least read together with the physician during the consultation) 15,16,46 Tailored/customized according to the patient's profile by the physician during the consultation (e.g. by underscoring certain items) [10][11][12]17,18 Given at an opportune moment during the consultation 9,18 Given only if the patient wants PILs 18

Study design
Randomized allocation of patients (or cluster randomization) to PILs or a control group 14,15,29 Single blind because the physician has to go through the PILs with the patient Control group without PILs (oral information alone)

Outcomes
Primary outcome using one previously validated score or measure Acute conditions Impact on patient Outcomes using one or more of the main outcome measures commonly used in RCT Psychic and cognitive impact Test of comprehension/knowledge of condition Satisfaction Behavioural impact Behaviour/adherence to treatment and to advice according to the objectives of the PILs (writer's intention) Reconsultation rates Therapeutic outcomes Pain Depression Anxiety Impact on physician Number of drugs prescribed Number of examinations or laboratory tests prescribed Impact on both patient and physician Doctor Patient Communication effectiveness It remains for research teams to work on the elaboration of generic scores independent of the clinical situations so as to allow better comparison between PILs and to set clear guidelines for their usage.