Patient Education and Counseling 52 (2004) 716

Review

Improving patients communication with doctors: a systematic review of intervention studies

Jane Harrington, Lorraine M. Noble*, Stanton P. Newman
Department of Psychiatry and Behavioural Sciences, Unit of Health Psychology, University College London, Wolfson Building, 48 Riding House Street, London W1N 8AA, UK Received 12 April 2002; received in revised form 7 December 2002; accepted 27 December 2002

Abstract A systematic review of the literature examined intervention studies designed to increase patients participation in medical consultations. Twenty-ve papers describing twenty studies met the inclusion criteria. About half of the intervention studies were randomised placebo controlled experimental designs. The studies were predominantly conducted in outpatient or primary care settings, with 50% of the interventions presented in a written form. Overall, half of the interventions resulted in increased patient participation, with slightly more signicant results found for bids for clarication than question-asking. However, of the 10 written interventions only two reported a signicant increase in question-asking. Patient satisfaction was the most commonly measured outcome, but few signicant improvements were found. However, there were signicant improvements in other outcomes, including perceptions of control over health, preferences for an active role in health care, recall of information, adherence to recommendations, attendance, and clinical outcomes. Few studies examined the links between patient characteristics and the success of the interventions. Future research needs to establish which forms of intervention are most effective and practical, for which groups of patients. # 2003 Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Doctorpatient communication; Intervention studies; Patient participation

1. Introduction The importance of doctorpatient communication in the process of health care has been established, with studies demonstrating clear links between the quality of communication and patient satisfaction, adherence and clinical outcomes [13]. Recognition of the importance of doctorpatient communication has led to the formal teaching of communication skills being considered as an essential aspect of medical education, and this is now an integral component of most medical curricula [4]. Whilst concerted efforts at behavioural change have been directed at health care professionals, considering only the doctors input into the consultation has the consequence of neglecting half of the relationship. The patients contribution to the consultation has also been found to be related to outcomes. Patients who are more active during consultations (in terms of asking questions, proffering information, and expressing opinions) are more likely to understand the
Corresponding author. Tel.: 44-20-7679-9465; fax: 44-20-7679-9426. E-mail address: lorraine.noble@ucl.ac.uk (L.M. Noble).
*

treatment rationale and recommendations and there is a growing body of evidence to suggest that more verbally active patients may experience improved medical outcomes [57]. Patients who are educated, have higher incomes, and are older, have been found to receive more information [8], which may indicate that different styles of communication are more effective in eliciting information from doctors. Alternatively, these patients may elicit more information because of doctors responses to certain subgroups of patients (e.g. positive stereotyping or subjective judgements about the patients informational needs or requirements). However, despite the evidence suggesting that active participation is benecial, in practice patients often contribute little to the consultation apart from answering direct questions. For example, patient question-asking has been found to be the least frequent category of verbal behaviour in medical visits [9]. A number of researchers have set out to redress the balance, by intervening to improve the patients contribution to their consultations. Six separate interventions to train patients to communicate more effectively with their doctors were reviewed by Anderson and Sharpe [10]. The evidence suggested that interventions were benecial, although there

0738-3991/$ see front matter # 2003 Elsevier Science Ireland Ltd. All rights reserved. doi:10.1016/S0738-3991(03)00017-X

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

were too few studies to make reliable comparisons among different forms of interventions. This systematic review specically examines intervention studies directed at patients. It incorporates the six studies that were subject of the earlier review and considers how the evidence on this topic has accumulated to date. This review examines intervention studies designed to improve patients communication with their doctors. The studies are considered in terms of: (a) the populations investigated; (b) design; (c) the types of interventions used; (d) the impact of the interventions on patients participation during the consultation and on outcomes; and (e) the links between patient characteristics and the success of the interventions.

tion, communication intervention, communication-skillstraining, patient involvement, patient empowerment and patient education. The search terms patient-participation and patient education were combined with the term communication to focus the search. Papers were included if they described an intervention designed to improve patients communication with doctors in any setting, reported data about the impact of the intervention on patients communication, and were published in English. The reference sections of papers that met the inclusion criteria were also searched, although this failed to yield any additional studies. 2.2. Overview of studies identified Twenty-ve papers describing twenty separate studies met the inclusion criteria [1135] (Table 1). The discrepancy was due to investigators reporting ndings from the same samples in different papers. To avoid confusion, papers using the same data set have been considered as a single study. Two-thirds of the studies were published within the last decade, indicating an increasing recognition of the importance of patient participation in the consultation. The studies were conducted in four countries, with the majority from the USA.

2. Methods 2.1. Identification of studies Literature searches were conducted to identify published reports of studies using Medline (19662001), PsycLIT (19662001), and BIDS Social Science Citation Index (19812001). The key search terms were patient-participa-

Table 1 Studies of interventions to improve patients communication with doctors References Country Population N Sample characteristics Agea Gender (female, %) Mainly female 100 Not reported 58 0 81 100 43 84 Not reported 59 100 61 54 70 52 76 71 83 19 RCT RCT Quasi RCT and Quasi RCT RCT RCT RCT RCT UC RCT RCT RCT RCT Quasi RCT RCT RCT RCT RCT Placebo Placebo Placebo Placebo Placebo Placebo Placebo Placebo Placebo UC True None Placebo Placebo True True True True True Placebo Design Control groups

[11], [12] [13] [14] [15], [16], [17] [18] [19] [20] [21] [22] [23] [24] [25] [26], [27] [28] [29] [30] [31] [32], [33] [34] [35]

USA USA UK USA USA USA USA USA Australia UK UK USA UK USA USA Australia UK USA USA The Netherlands

Outpatientmedical clinic 200 Simulated patientshypertension 36 Primary care 131, 127 Outpatientpeptic ulcer, diabetes, 252 hypertension, breast cancer Outpatienthypertension 150 Family medicine clinic 67 Outpatientobstetrics and gynaecology 53, 49 General paediatric practice 141 Outpatient and 142 inpatientmedical oncology Primary care 80 Outpatientmedical oncology 95 Outpatientbreast cancer 60 Primary care 120 Primary care 201 Primary care 20 Outpatientmedical oncology 60 Outpatientdermatology, 1077 gynaecology, orthopaedics Family practice clinic/private offices 150 Community sites 314 Outpatientscardiology 103

Median 50 Median 28 Not reported 4755 (all groups) 58 35 39 (study 2) 8 51 Not reported 44 59 43 46 37 53 Not reported 45 77 64

Papers reporting findings from the same patient samples are grouped together; N: the number of patients providing data for the study; RCT: randomised controlled trial; Quasi: quasi-experimental; UC: unclear; Placebo: placebo control group; True: non-intervention group; None: no control group. a Mean age reported unless otherwise specified.

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

3. Results 3.1. Populations studied The settings were primarily outpatient and primary care clinics (Table 1), with one study conducted in the community [34], and one study considering inpatients [22]. No studies considered mental health services, despite evidence that patients attending these services often have unspoken requests [36]. One study used healthy volunteers as simulated patients [13], which is a procedure of questionable validity. The sample sizes varied considerably, with a range from 20 to 1077 (Table 1). The majority of studies were conducted with adults, with a mean or median age of 2877 years. Just one of the interventions was directed at children [21]. In most of the studies the sample was mainly or all female (Table 1). Although most of the studies described where the intervention took place, many failed to explain the process of care, for example, the sequence of events after the patients arrival at the clinic, or the range of health professionals providing care. Additionally, few reported whether patients attended for more than one consultation, or described patients previous experience with the service, despite the possibility that familiarity and pre-established interaction patterns may inuence patients contribution to the consultation. Only half of the studies reported the timing of the intervention in relation to patients previous experience with the service [15,19,22,24,25,28,3032,35]. In six further studies, previous experience was implied [11,14,23,26, 29,34]. 3.2. Design Over half of the studies (11/20) used a randomised controlled experimental design, comparing one or more experimental conditions with an attention placebo control (Table 1). In three of these [11,21,28], the doctors or clinic sessions were randomised, rather than individual patients. Six studies using a randomised controlled trial design failed to include an attention placebo control condition, and three used a quasi-experimental design, both of which reduce the condence with which improvement in patients communication can be attributed to the intervention. All but two of the studies [23,24] assessed the impact of the intervention on the process of communication and outcomes of the consultation. Three studies examined the inuence of patient characteristics on responses to the interventions [13,18,26], considering a small range of variables. Fifteen of the twenty studies used only post-intervention measurement. Only four studies examined baseline and post-intervention measures to enable changes due to the intervention to be examined [15,24,25,32].

3.3. Overview of the interventions 3.3.1. Aims of the interventions All of the interventions were designed to increase patient participation during the consultation. The specic behaviours most frequently encouraged were question-asking, raising concerns, and requesting clarication or checking understanding (Table 2). In two studies, no specic behaviours were targeted. Most of the interventions were delivered immediately before the patients appointments (Table 2). In ve studies, interventions were delivered prior to the day of the appointment to allow patients more time to familiarise themselves with the content of the intervention. 3.3.2. Modes of presentation Written interventions were most common, followed by face-to-face coaching and videotape (Table 2). A few studies used a combination of modes. In two studies, the intervention was simultaneously directed at other participants in the consultation: either doctors [28] or doctors and parents [21]. Four studies compared the impact of different interventions, either using the same mode with different content [13,14,18] or comparing different modes of presentation [25]. 3.3.3. Duration of interventions The duration of the interventions varied considerably, from 10 to 25 min for the video interventions to a 2 h combined face-to-face and written intervention (Table 2). In several studies, patients were allowed as much time as they wanted with the intervention. However, nearly half of the studies failed to report the duration of the interventions. Both the different types of interventions and their duration have practical implications, if the intention is that they will be used in routine practice. 3.3.4. Content of the interventions The written interventions tended to be presented in booklet or checklist form, sometimes with instructions to write down any questions that the patient may have had. The faceto-face interventions were generally based on coaching patients in question-asking techniques and the video interventions were based on modelling techniques. Modelling was also employed in one of the face-to face interventions [34] (Table 2). 3.4. Effects on the process of communication Pendleton [37] distinguished between three types of variables in doctorpatient communication research: antecedents, process and outcome. The twenty studies considered in this review examined the effects of the interventions on the process and outcome of communication. Process variables include measures of the verbal and non-verbal interactions between the doctor and the patient. Outcome variables can be immediate, intermediate, and long-term

10

Table 2 Overview of intervention studies References Behaviours encouraged Question-asking Question-asking, clarification Mode of presentation Face-to-face Video, face-to-face Time of intervention delivery Immediately pre-appointment Immediately pre-appointment Duration (min) 10 2025, Not reported Brief description of experimental intervention To work through a question-asking protocol, identifying questions patients may have concerning five main aspects of illness (aetiology, duration, severity, prevention) and treatment. Model expresses confusion and asks for clarification during a partially confusing presentation on symptoms, causes and consequences of hypertension. All patients participated in a live teaching session (on dietary advice); 50% were told to ask questions if confused during the session. One group received permission to ask questions and the other received guidance on how to check understanding. Using a treatment algorithm as a guide, patients read their medical records and were coached to ask questions and negotiate medical decisions with their doctors. Video 1. Model expresses confusion and asks for clarification during a partially confusing presentation on definitions, control, complications and medication for hypertension. Video 2. Model discloses problems and concerns about hypertension, including difficulties with the regimen. Booklet educating patients to recognise information needs and verbalise requests for information. Based on 33 questions covering five areas, state of body, reasons for state of body, treatment, reasons for treatment, and impact on life. Patients provided with a list of possible concerns, with instructions to write down at least three questions to ask the doctor and to take the list to the appointment. Note from the doctor to patient giving express permission to ask questions. Children: featured a child modelling effective communication. Parent: emphasised the importance of doctorpatient communication and fostering the childs active involvement. Doctor: presented research evidence on health consequences of doctorpatient communication and demonstrated building rapport, checking childrens understanding of medical information. Prompt sheet with 11 items designed to encourage patient active participation in the consultation. Patients encouraged to write down specific questions and to take sheet into consultation. Instructions encouraging patients to think about reason for attending, problems experienced, worries, etc. Patient instructed to tell the doctor all these things clearly, early on, and to ask for more information on points of interest. Patients provided with an audiotape of their previous consultation (first consultation, primary bad news, i.e. diagnosis, or secondary bad news i.e. unsuccessful initial treatment), to take home and listen to as many times as they wished. Interactive programme consisting of text, graphics, audio narration, music, and video clips. Programme divided into four sections, introduction, understanding the problem, treatment options and the experiences of other women. Brochure, based on the above, conveying the same message. Leaflet asking patients to identify their problems, think about the causes, treatment, etc. and provided space for the patient to write down their ideas and then voice these concerns during the consultation and to ask questions about diagnosis. Patients completed a 25 item questionnaire containing five categories of concerns, desire for medical information, psychosocial assistance, therapeutic listening, general health advice and biomedical treatments.

[11], [12] [13]

[14] [15], [16], [17]

[18]

Question-asking, checking, understanding Question-asking, decision-making, overcoming barriers Question-asking, clarification, raising concerns Question-asking

Written Face-to-face

Immediately pre-appointment Immediately pre-appointment Immediately pre-appointment

Not reported 20

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

Video

14

[19]

Written

Immediately pre-appointment Immediately pre-appointment Immediately pre-appointment

Self-determined

[20]

Question-asking, obtaining information Question-asking, raising concerns, decision-making, active role Question-asking

Written

Not reported

[21]

Video

10 (child), 10 (parent), 15 (doctor)

[22]

Written

Immediately pre-appointment Immediately pre-appointment Up to one month pre-appointment Immediately pre-appointment

Not reported

[23]

Raising concerns, seeking information No specific behaviours

Written

Not reported

[24]

Audiotape

Length of audiotape

[25]

Question-asking, raising concerns, expressing opinions Raising concerns

Multimedia, written

3540, 1520

[26], [27]

Written

Immediately pre-appointment Immediately pre-appointment

Not reported

[28]

Raising concerns

Written

Not reported

[29]

Question-asking, information-seeking and verifying Question-asking

Face-to-face

Immediately pre-appointment

Not reported

[30]

Written, face-to-face

Immediately pre-appointment

Not reported, not reported

[31]

Question-asking

Written

2 weeks, pre-appointment 23 days, pre-appointment

Self-determined

[32], [33]

[34]

Information seeking and verifying, information-provision Active role

Written

Self-determined

Patients presented with information aimed at improving their understanding of the typical communication goals relevant to the three parts of the examination (history-taking, examination, conclusion). It was stressed that patients should become the information seeker and doctor information provider. Prompt sheet, based on 17 frequently asked questions by patients. Coaching based on discussion about the importance of question-asking, generating questions, comparison of questions with the prompt sheet and adding new questions. Exploration of benefits and barriers to question asking, and question-asking rehearsed. A help card and an information letter to help patients prepare and prioritise questions to ask in consultation covering patients condition (7 items), tests and treatment (12 items) other (4 items) 14 page training booklet formatted like a workbook with examples and space for notes. Prompted patients to write questions on a variety of topics and to list the details of their symptoms and related medical history. Written summary of training booklet. Group programme including modelling of both undesirable and desirable behaviours, roleplaying exercises, and discussion of potentially negative effects of passive involvement. Provision of cue cards with a list of desirable active behaviours and a booklet for recording reasons for visit, medications, and questions for the doctor, with room for recording the answers. Frequently asked question checklist. 49 items covering 10 topics: nature of disease, physical state, medication, risk factors and lifestyle, treatment, examination, disease related problems in the family, psychosocial problems, practical matters, other problems.

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

Face-to-face, written

Up to 3 months pre-appointment

2h

[35]

No specific behaviours

Written

1 week, pre-appointment

Self-determined

11

12

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

[37]. Immediate outcomes include patient satisfaction and recall, intermediate outcome variables include adherence to recommendations, and long-term outcomes include changes in health status or lifestyle. 3.4.1. Measures of the process of communication In the majority of studies, the process of communication was measured using interaction analysis, usually from audiotaped recordings [11,1315,18,19,2426,28,30,32], with two studies using video [21,29]. Six of these used previously reported interaction analysis systems [11,15, 18,21,22,24], with the remainder using newly devised systems, but since 1995 none of the studies have employed reliable and valid methods. The use of different systems made comparisons of results across studies difcult, particularly when the reliability of the measures had not been established. Furthermore, studies using new measures were less likely to examine inter-rater reliability than studies using previously reported methods, which is a serious methodological aw. Researchers coding the interactions between doctors and patients were reported to be blinded to group allocation in only six of the studies [15,18,19, 21,26,32], leaving the remaining studies open to the possibility of bias. The interaction analysis systems were used to examine a wide range of communicative behaviours, measuring frequency, duration and ratios (e.g. patient to doctor behaviours). The most commonly measured variable was patient question-asking. However, there was otherwise little consistency in the measures used. In several of the studies [20,23,31,34,35], the process of communication was assessed by obtaining patients or
Table 3 Impact of interventions on the process of communication References [11], [13] [14] [15], [18] [19] [20] [21] [22] [23] [24] [25] [26], [28] [29] [30] [31] [32], [34] [35] [12] Overall patient participation Sig " Sig " Sig " Sig " Sig " NS " NS " Sig " Sig " [27] NS " Sig " Sig " [33] Sig " NS " NS " Patient question asking Sig " Sig " NS " Sig " NS " NS " NS " NS " Sig " Sig " Sig " Sig " Sig "

doctors perceptions of the consultation, and this was often the only measure of the process of communication. None of the studies using perceptions as a measure of process attempted to establish the reliability or validity of the criteria used. The sole use of subjective perceptions as an index of the process of communication is problematic, as subjective perceptions and behavioural measures have been found to correlate poorly [25,38]. Nearly half of the studies measured only one or two categories of communicative behaviour, thus failing to take the opportunity to make use of the richness of the data available. Five studies measured 10 or more categories of communicative behaviour [11,15,21,22,29]. 3.4.2. The impact of the interventions on the process of communication Although the studies used different measures of patient participation, overall it was clear that the interventions had the effect of encouraging patients to be more active in their consultations (Table 3). Out of 16 studies that examined variables related to patient participation, 10 reported a signicant increase and ve reported a non-signicant increase. Broken down by mode of presentation, all but one of the six face-to-face interventions reported a signicant increase in overall participation [11,13,15,29,30], and all of the video interventions reported signicant increases [13,18,21]. However, of the 10 written interventions, only two reported a signicant increase [14,32]. Question-asking was the most frequently targeted specic behaviour, with equal numbers of studies reporting signicant increases and non-signicant trends (Table 3). The range of question-asking was low in many studies, with a

Patient clarification Sig "

Consultation length

Expressed affect Sig "

Doctor encourages patient participation Sig "

[16], [17]

Sig "

Sig "

Sig " Sig "

NS "

(Sig "): significant increase; (Sig #): significant decrease; (): no change; (NS): non-significant change; blank: not assessed.

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

13

range of 07% questions per consultation reported for control groups and 09% for intervention groups. Question-asking appeared to represent very little of the patients verbal contribution to the consultation, three studies reporting the proportion gave a range of 0.078.7% [11,22,24]. Differences among the studies may have been related to the criteria used for question-asking. For example, one study [29] that included direct and indirect questions, found that patients asked more questions than those reported in other studies. Signicant increases in requests for clarication were more consistently reported (Table 3), raising the possibility that patients perceived it as less threatening to increase their contribution based on the information that the doctor was providing, rather than to ask spontaneous questions. Six studies compared the effects of different interventions. One found that instructing patients to ask questions most effective, followed by a modelling video [13], and another found that guidance in checking understanding was more effective than giving permission to ask questions [14]. Other ndings indicated that modelling of requesting clarication was more effective than modelling of raising concerns [18]. A further study [30] found that while a question prompt sheet resulted in increased question-asking behaviour, adding face-to-face coaching did not encourage further question-asking. Two studies found no differences between their interventions, one comparing two similar written interventions [20] and one comparing written and multi-media presentations of the same information [25]. Anderson and Sharpe [10] suggested that future intervention studies should be expected to extend assessments beyond question-asking behaviour, but there is little evidence that investigators have broadened the range of communicative behaviours under consideration and question-asking in its various guises still predominates. Increasing patient participation might be expected to signicantly lengthen the consultation, an outcome that would not generally be welcomed by doctors. However, the evidence suggested that this concern may be unwarranted. In ve of the seven studies that considered this, increasing patient participation did not result in an increase in consultation length. Where differences in consultation length between the groups were observed, in one study [26] the difference was relatively small (1.2 min), but in the other [28] it was considerably longer (6 min) (Table 3). Although the aim of the interventions was to increase patient participation, there was also evidence that the consultation could be modied in other ways. In one study, where the intervention consisted of giving patients an audiotape of their consultation [24], a signicant reduction was found in patients subsequent requests for information that had been given during that consultation. Two studies considered the impact of the intervention on expressions of affect during the consultation. In one study [11], patients who had received the intervention were rated by independent observers as more anxious and angry, and

their doctors as more matter-of-fact, angry, and less sympathetic. This suggested that increased patient participation was uncomfortable for both parties, raising a concern that although the intervention may have been effective, the consequences were detrimental. In a further study [15,17] the intervention resulted in signicantly increased displays of affect from doctors and patients, although the only detail provided about this was that this included both positive and negative emotions. Despite these interesting ndings from these earlier studies, expressed emotion has not been considered in any of the more recent investigations. The impact of the interventions on doctors behaviour was examined in three further studies, two of which found signicant increases in the encouragement doctors gave to patients to ask questions (Table 3), and the third nding that experimental group consultations were more likely to be rated by the doctor as a good consultation [23]. 3.5. Effects on outcomes of the consultation Immediate outcomes were the most commonly considered, with patient satisfaction the single most frequently examined outcome. Most of the studies used newly-devised measures of immediate outcomes or modied versions of existing scales, which raises questions about validity and reliability, and makes comparisons among studies problematic. Intermediate and long-term outcomes were considered in only four studies [11,15,26,32], presumably due to the resource implications of following up patients over time. Attendance was considered in three studies and one study measured adherence to both medications and behavioural treatments (Table 4). Perhaps the most important long-term outcome variable, disease control, was considered in only one study. The lack of emphasis on outcomes that have a clear practical importance in health care is a serious shortcoming of work in this eld. 3.5.1. The impact of the interventions on outcomes of the consultation The studies generally found no signicant increase in patient satisfaction due to the interventions, with only two exceptions (Table 4). Overall, high levels of satisfaction were reported, with ceiling effects specically reported in three studies [20,28,34]. The high frequency of ceiling effects and the insensitivity of measures of satisfaction has been previously noted as an issue limiting ndings on this dimension [39,40]. Two studies considering doctor satisfaction found no signicant change [20,28]. Signicant improvements were found in the accuracy of patients recall of the consultation in three of the ve studies which examined this, although none of the ve studies considering patients knowledge of their illness found signicant changes (Table 4). This discrepancy might be due to the fact that measures of recall emphasise treatment recommendations, which may be more salient to patients, but also

14

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

Table 4 Impact of the interventions on patient outcomes References [11], [12] [13] [14] [15], [16], [17] Satisfaction Sig # Sig " or Sig # Sig " preference for active role, Sig " perceived health Significant improvement hypertension and diabetes Knowledge Recall Response to illness Sig " locus of control Mood Adherence Attendance Sig " short term Disease control

[18] [19] [20] [21] [22] [23] [24] [25] [26], [27] [28] [29] [30] [31] [32], [33] [34] [35]

NS " NS " Sig "

Sig " Sig " locus of control Sig " preference for active role NS adjustment NS # anxiety anxiety

Sig " Sig " NS " NS adjustment perceived health perceived health

NS optimism long-term Sig # anxiety, Sig # depression anxiety Sig " Sig "

NS "

Sig # anxiety

(Sig "): significant increase; (Sig #): significant decrease; (): no change; (NS): non-significant change; blank: not assessed.

may reect differences in the way information about diagnosis and treatment are given in the consultation. The aim of many of the interventions was to encourage patients to be involved in their health care. Signicant increases in patients perceptions of control over their health and preferences for an active role in their health care were found in the four studies that considered this. Only one of the ve studies considering patients perceptions of their own health or adjustment to illness found a signicant improvement (Table 4). However, these perceptions were assessed soon after the consultation, and it could be argued that these constructs would be more appropriately measured as long-term outcomes. Despite the evidence about increases in observable affect from the two studies mentioned above, patient and doctor self-reports about the consultation were more positive. Patient anxiety was the most commonly measured mood state, and decreases in anxiety due to the intervention were noted in three out of ve studies (Table 4). Furthermore, no effects on doctors satisfaction with, or perceptions of, the consultation were found in any of the studies that considered this [20,26,28]. Although longer-term outcomes were rarely examined, results were encouraging. Signicant improvements in attendance, adherence and disease control were found for three separate samples of patients with chronic conditions [11,1517,27,33].

3.6. The influence of patient variables on the impact of the interventions If interventions are to be used routinely to improve patients communication with their doctors, it is important to determine not only which interventions are successful and how they affect outcomes, but also whether certain groups of patients may benet. Only three studies considered this issue. Increases in question-asking were found to be higher for patients from social classes I and II, and increases in the length of the consultation were particularly evident in patients who were younger, male, and from social classes I and II [26]. Two studies considered assertiveness, but their generalizability was limited by their sampling techniques: in one study, healthy volunteers were used to represent simulated patients [13] and in another, the sample consisted only of male patients [18]. Little can therefore be concluded, except that it remains a possibility that patient characteristics can affect responses to an intervention. 3.7. Quality of reporting and methodological issues The studies reviewed clearly demonstrated that non-signicant ndings are common in this area of research. In a few studies, however, it was difcult to determine the results of signicance testing due to ambiguous reporting. Four particular problems were evident: describing non-signicant

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716

15

trends as marginally signicant or approaching signicance, implying in the summary of the study that the ndings were signicant, misusing the convention of asterisking signicant ndings in a table of results, and failing to report the results of signicance testing for all relevant comparisons [11,18,20,26,28]. Several studies provided information about the relationships between input, process and outcome variables in the consultation, but failed to clearly report on the impact of the intervention [12,13,1619,22,2426]. Few of the studies examined changes in measures from baseline to post-intervention and post-intervention, with most relying on a post-test control group design. This design is not ideal as it fails to provide evidence that any differences between experimental and control groups are a direct consequence of the intervention, because the assumption that both groups have equal baseline levels of performance is not tested. In the context of a communication intervention the collection of baseline data for certain measures, for example, anxiety, knowledge and health status as such measures would provide supporting information regarding the efcacy of the intervention. In terms of statistical analysis, several studies performed multiple correlations on the same data set [12,17,2426], thus increasing the risk of Type I errors. Only two studies reported power analyses [26,30]. The absence of a power calculation is problematic, as the reader cannot determine whether non-signicant results are a consequence of an inadequate sample size, or whether signicant results (for example, correlations) are related to a very large sample size. In several studies it was difcult to ascertain the numbers of patients participating, as some investigators emphasised the number recruited rather than the number providing data [11,13,14,24] or failed to report exactly how many patients comprised the experimental and control groups [15,22]. It would appear that since the publication of the Anderson and Sharpe review [10], some of the criticisms the authors made regarding the quality of reporting and sample size still have not been adequately addressed.

Encouraging patient participation appeared to lead to a greater sense of control and preference for a more active role in consultations in a number of studies, although the results were less clear when considering satisfaction, knowledge, and recall. 4.1. Further research Several areas warrant further investigation. Firstly, it needs to be determined whether responses to interventions are moderated by patient characteristics, to identify which patients would derive greatest benet from a communication intervention. Secondly, although both difcult and costly, the impact of interventions on longer-term clinical outcomes should be assessed. Studies should also take advantage of the richness of the data available from audio or videotaped recordings, given that interventions may improve patient participation or affect the consultation in more ways than simply increasing patient questionasking. Future research also needs to address the basic aws evident in some of the studies reviewed here. In particular, the results of the interventions should be presented clearly, the numbers of participants in each group should be reported, and patients prior experience with the service should be described. The key methodological issues that need to be addressed are the inclusion of the appropriate control groups, appropriate sample size, proper use of randomisation and blinding, and the use of valid and reliable measures. 4.2. Practice implications At present, partially due to the diversity of the interventions, there is insufcient evidence to draw denitive conclusions about which forms of intervention are most successful in which contexts, although there was some evidence that the more intensive face-to-face interventions were more effective than written interventions. This raises the issue about what is practical compared to what is effective, which is important if the ultimate aim is for interventions to be used routinely. Both the nancial and other resource implications of developing and implementing interventions need to be considered, and more emphasis could be placed on comparing the effectiveness of different modes in achieving a particular outcome.

4. Conclusions The studies considered in this review generally demonstrated that interventions directed at patients can be successful in increasing patient participation, and that, in some circumstances, this can be achieved without an increase in consultation length. Where the ndings about patient participation were statistically non-signicant, there were trends in the expected direction. The overall increase in patient participation was most frequently assessed by examining patient question-asking or attempts to clarify information, although it has been noted that question-asking often represents only a small proportion of the amount of overall talk. It is possible that patient question-asking may be regarded by patients and doctors as inappropriately assertive behaviour.

Acknowledgements This work was undertaken by Jane Harrington, Lorraine Noble and Stanton Newman, who received funding from the NHS London Regional Ofce, Research and Development Programme. The views expressed in the publication are those of the authors and not necessarily those of the NHS or the Department of Health.

16

J. Harrington et al. / Patient Education and Counseling 52 (2004) 716 [21] Lewis CC, Pantell RH, Sharp L. Increasing patient knowledge, satisfaction, and involvement: randomized trial of a communication intervention. Pediatrics 1991;88:3518. [22] Butow PN, Dunn SM, Tattershall MHN, Jones QJ. Patient participation in the cancer consultation: evaluation of a question prompt sheet. Ann Oncol 1994;5:199204. [23] Fredrikson LG, Bull PE. Evaluation of a patient education leaet designed to improve communication in medical consultations. Patient Educ Counsel 1995;25:517. [24] Ford S, Falloweld L, Hall A, Lewis S. The inuence of audiotapes on patient participation in the cancer consultation. Eur J Cancer 1995;31A:22649. [25] Street RL, Voight B, Geyer C, Manning T, Swanson GP. Increasing patient involvement in choosing treatment for early breast cancer. Cancer 1995;76:227585. [26] McCann S, Weinman J. Empowering the patient in the consultation: a pilot study. Patient Educ Counsel 1996;27:22734. [27] McCann S, Weinman J. Encouraging patient participation in general practice consultations: effect on consultation length and content. Psychol Health 1996;11:85769. [28] Hornberger J, Thom D, MaCurdy T. Effects of a self-administered previsit questionnaire to enhance awareness of patients concerns in primary care. J Gen Int Med 1997;12:597606. [29] McGee DS, Cegala DJ. Patient communication skills training for improved communication competence in the primary care medical consultation. J Appl Commun Res 1998;26:41230. [30] Brown R, Butow PN, Boyer MJ, Tattersall MHN. Promoting patient participation in the cancer consultation: evaluation of a prompt sheet and coaching in question-asking. Br J Cancer 1999; 80: 2428. [31] Fleissig A, Glasser B, Lloyd M. Encouraging out-patients to make the most of their rst hospital appointment to what extent can a written prompt help patients get the information they want? Patient Educ Counsel 1999;38:6979. [32] Cegala DJ, McClure L, Marinelli TM, Post DM. The effects of communication skills training on patients participation during medical interviews. Patient Educ Counsel 2000;41:20922. [33] Cegala DJ, Marinelli T, Post D. The effects of patient communication skills training on compliance. Arch Fam Med 2000;9:5764. [34] Tennstedt SL. Empowering older patients to communicate more effectively in the medical encounter. Clin Geriatric Med 2000;16:61 70. [35] Martinali J, Bolman C, Brug J, van den Borne B, Bar F. A checklist to improve patient education in a cardiology outpatient setting. Patient Educ Counsel 2000;42:2318. [36] Noble LM, Douglas BC, Newman SP. What do patients want and do we want to know? a review of patients requests of psychiatric services. Acta Psychiatr Scand 1999;100:3217. [37] Pendleton D. Doctorpatient communication: a review. In: Pendleton D, Hasler J. (Eds.), Doctorpatient Communication. London: Academic Press; 1983. [38] Street RL. Analyzing communication in medical consultations. Med Care 1992;30:97688. [39] Balogh R, Simpson A, Bond S. Involving clients in clinical audits of mental health services. Int J Qual Health Care 1995;7:343 53. [40] Calnan M. Towards a conceptual framework of lay evaluation of health care. Soc Sci Med 1988;27:92733.

References
[1] Hall JA, Roter DL, Katz BA. Meta-analysis of correlates of provider behavior in medical encounters. Med Care 1988;26:65775. [2] Roter DL, Hall JA, Katz NR. Patientphysician communication: a descriptive summary of the literature. Patient Educ Counsel 1988; 12:99119. [3] Stewart MA. Effective physicianpatient communication and health outcomes: a review. Can Med Assoc J 1995;152:142333. [4] Simpson M, Buckman R, Stewart M, Maguire P, Lipkin M, Novak D, et al. Doctorpatient communication: the Toronto consensus statement. Br Med J 1991;303:13857. [5] Rost K, Carter W, Inui T. Introduction of information during the initial medical visit: consequences for patient follow-through with physician recommendations for medication. Soc Sci Med 1989; 28: 31521. [6] Roter DL, Hall JA. Doctors talking to patients/patients talking to doctors. Westport (CT): Auburn; 1993. [7] Street RL. Active patients as powerful communicators: the communicative foundation of participation in care. In: Robinson WP, Giles H, editors. The new handbook of language and social psychology. Chichester (UK): Wiley. p. 54160. [8] Street RJ. Information-giving in medical consultations: the inuence of patients communicative styles and personal characteristics. Soc Sci Med 1991;32:5418. [9] Roter DL, Hall JA, Katz NR. Relations between physicians behaviors and analogue patients satisfaction, recall, and impressions. Med Care 1987;25:43751. [10] Anderson LA, Sharpe PA. Improving patient and provider communication: a synthesis and review of communication interventions. Patient Educ Counsel 1991;17:91134. [11] Roter DL. Patient participation in the patient-provider interaction: the effects of patient question asking on the quality of interaction satisfaction and compliance. Health Educ Monogr 1977;281315. [12] Roter DL. Patient question asking and physicianpatient interaction. Health Psychol 1984;3:395409. [13] Wallston BS, Wallston KA, DeVellis BM, McLendon E, Percy J. Modication of question asking behaviour in high and low assertive women through modeling and specic instructions. Soc Behav Pers 1979;6:195204. [14] Robinson EJ, Whiteld MJ. Improving the efciency of patients comprehension monitoring: a way of increasing patients participation in general practice consultations. Soc Sci Med 1985;21:9159. [15] Greeneld S, Kaplan S, Ware JE. Expanding patient involvement in care: effects on patient outcomes. Ann Int Med 1985;102:5208. [16] Greeneld S, Kaplan S, Ware JE, Yano EM, Frank HJL. Patients participation in medical care: effects on blood sugar and quality of life in diabetes. J Gen Int Med 1988;3:44857. [17] Kaplan SH, Greeneld S, Ware JE. Assessing the effects of physician patient interactions on the outcomes of chronic disease. Med Care 1989;27:11027. [18] Anderson LA, DeVellis B, DeVellis RF. Effects of modeling on patient communication, satisfaction, and knowledge. Med Care 1987;25:104456. [19] Tabac ER. Encouraging patient question-asking: a clinical trial. Patient Educ Counsel 1988;12:3749. [20] Thompson SC, Nanni C. Patient-oriented interventions to improve communication in a medical ofce visit. Health Psychol 1990;9: 390404.