Shared decision-making training in general practice: a rapid review

ABSTRACT

Introduction

Although shared decision-making (SDM) is key to providing patient-centred care, SDM is lacking in primary care. Training programmes seeking to improve GP SDM have yet to be reviewed. Therefore, a rapid review of the literature was conducted to evaluate GP SDM training methods and their outcomes.

Methods

MEDLINE, EMBASE and CENTRAL were systematically searched. Results of the studies included were synthesised narratively. Study quality was appraised using the Medical Education Research Study Quality Instrument (MERSQI).

Results

Seven studies were identified. Study quality was high, with a mean MERSQI score of 17.2/18 (range 16–18). Theory/presentation was the most prevalent training method (n=6). Of the five studies assessing the impact of SDM training on patient outcomes, only one yielded positive results. Contrastingly, both studies assessing clinician behaviour produced positive results.

Conclusions

SDM training improved GP behaviour but the effects on patient outcomes were lacking. SDM training programmes that utilised teaching methods targeting practical SDM skills, such as role play, observed some positive findings. However, because their prevalence was lacking, further research into these methods, and their cost-effectiveness, are needed.

Introduction

Shared decision-making (SDM) is the process of clinicians involving and supporting patients in making informed decisions regarding their healthcare.¹ This is key to providing patient-centred care.² However, evidence suggests that clinicians find implementing SDM challenging.^3,4 Furthermore, evaluation of the Making Good Decisions in Collaboration (MAGIC) SDM improvement programme has highlighted the lack of SDM in clinical practice, particularly in primary care.⁵ This has been attributed to multiple factors, including limited clinician ability to facilitate SDM, the shortage of guidance for GPs on implementing SDM, concerns about time pressures and apprehension about the perceived demands of patients.⁶

A previous systematic review acknowledged that SDM training can improve clinicians' attitudes and knowledge.⁷ Likewise, a rapid review identified that SDM training can enhance doctor–patient communication.⁸ Although this provides insight into SDM training for medical trainees and secondary care doctors, SDM training programmes for GPs have yet to be reviewed.

With the recent introduction of the Personalised Care Institute's SDM training programmes for healthcare professionals, there is a strong need for evidence on delivering SDM training in primary care.⁹ Addressing this research gap can provide clarity on training methods that effectively facilitate SDM in general practice and, hence, contribute to the delivery of patient-centred care.

Therefore, a rapid review was conducted to enable timely synthesis of evidence to answer the following research questions¹⁰:

1. What elements of SDM and teaching methods are used in GP SDM training?

2. What is the educational effectiveness of these interventions according to the components of Kirkpatrick's four levels of evaluating training programmes (Reaction, Knowledge, Behaviour and Patient outcomes)¹¹?

Methods

Search strategy

The search strategy was developed using guidance from the Cochrane Rapid Reviews Methods Group¹² and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.¹³

Ovid MEDLINE, Ovid Embase and The Cochrane Central Register of Controlled Trials (CENTRAL) were systematically searched (3–6 November 2020). Table 1 presents the Ovid MEDLINE search strategy. Minor modifications were made for searches of Ovid Embase and CENTRAL. Backward and forward citation searching identified further studies. Articles were limited to those written in English and published from January 2010 onward, because the MAGIC programme was introduced in 2010¹⁴ and to enable review of more recent studies.

Table 1.

The Ovid MEDLINE search strategy

Group	Terms (combined by the Boolean operator ‘OR’)
1	“shared decision making”.ti,ab. “shared decision-making”.ti,ab. SDM.ti,ab. “shar* decision”.ti,ab. Shared decision making.mp or Decision Making, Shared/
2	education.ti,ab. teach*.ti,ab. train*.ti,ab. intervention*.ti,ab. program*.ti,ab. learn*.ti,ab. course.ti,ab. workshop*.ti,ab.
3	“general practi*”.ti,ab. GP*.ti,ab. “family doctor*”.ti,ab. “primary care”.ti,ab. general practice/ or family practice/
4	competenc*.ti,ab. satisfact*.ti,ab. attitude*.ti,ab. knowledge.ti,ab. outcome*.ti,ab. evaluat*.ti,ab. change.ti,ab. Outcome Assessment, Health Care/
5	#1 AND #2 AND #3 AND #4

Study selection

The search retrieved 1,011 articles. After removal of duplicates, inclusion criteria (Table 2) were used to screen the titles/abstracts, followed by the full texts, of the remaining studies (n=635).

Table 2.

Inclusion criteria

Participants	GPs GP trainees
Intervention	SDM training alone or within context
Outcome measures	Clinician attitude, knowledge or behaviour Patient outcome
Study design	Single group pre and post-test Two-group non-randomised Randomised controlled trial

SDM = shared decision-making.

Both authors screened the first 15% (n=95) of titles/abstracts in duplicate. There were five (8%) disagreements, which were resolved through discussion. TN then screened the remaining titles/abstracts independently, consulting HM where needed.

The full texts of the included titles/abstracts were retrieved and rescreened against the inclusion criteria. Both authors screened the first 28% (n=7) of the full texts in duplicate. No disagreements arose. TN then independently assessed the remaining papers, consulting HM where needed.

TN independently conducted the forward and backward citation searching of the reference lists of the included studies. This yielded an additional 48 records, of which 21 remained after removal of duplicates. These were assessed for eligibility against the inclusion criteria by TN. Only articles meeting the inclusion criteria of the study were included.

The full results of the study selection are presented in Fig 1.

Fig 1.

Study selection. SDM = shared decision-making.

Data extraction

A standardised data-extraction form was developed using items within the Medical Education Research Study Quality Instrument (MERSQI).¹⁵ The data-extraction form comprised the following:

1. Study design, publication year and country of origin

2. Number and type of participants

3. Duration of SDM training programme

4. Description of teaching methods used

5. Elements of SDM included in the intended learning outcomes (ILOs) of the intervention, as described by the Informed Medical Decisions Foundation¹⁶

6. Highest outcome measure according to Kirkpatrick's four levels of evaluating training¹¹

7. Study results and effectiveness using the highest measured outcome according to Kirkpatrick's four levels of evaluating training¹¹

A meeting between both authors was held before data extraction to ensure a robust understanding of the components within the form and to improve inter-rater reliability. The data extraction was conducted in duplicate by both authors. Disputes regarding the elements of SDM covered in the training programmes within the included articles led to refinement of item 5 of the data-extraction form. Subsequently, both authors extracted data on this item together by revisiting the manuscripts. All other disagreements were resolved through discussion.

Study quality appraisal

Study quality was assessed using the MERSQI because of its proven reliability and validity in appraising quantitative medical education research.¹⁵ Both authors calculated the MERSQI scores in duplicate and all disagreements that arose were resolved through discussion. Given that some items were not applicable, the total MERSQI score for each study was calculated as a percentage of the total possible score, and multiplied by 18 to produce a final score out of 18.

In line with previous reviews, studies with a MERSQI score of 14 or more were defined as being high quality.^17–19

Data analysis

Studies were classified according to Kirkpatrick's four levels of evaluation.¹¹ The interventions and their effectiveness were the primary outcomes of interest. Effectiveness was determined using inferential statistics as reported in the studies. Statistical significance was defined as p<0.05 cf. control or using the adjusted relative risk (RR) and 95% confidence interval (CI). Given heterogeneity among the included studies, it was not possible to conduct a meta-analysis. Hence, the results of this review are synthesised narratively.

Results

Seven randomised controlled trials were included in this review.^20–26 The results are summarised in Table 3.

Table 3.

Summary of results

Author	Participants (n)	Outcome measured according to Kirkpatrick's modela	SDM intervention	Elements of SDM coveredb	Teaching method(s)c	Findings	Effectiveness	MERSQI
Wollny et al (2019)20	108	4	SDM in context of type 2 diabetes mellitus; workshop with trained GP (1–1.5 h)	A+B+C+D	3, 4	At 24 months' follow-up: mean decrease in HbA1c of 9 mmol/mol in intervention group compared with 6 mmol/mol in control group	Ineffective: p=0.15	18
Sanders et al (2018)22	68	4	Two sessions on SDM in context of low back-pain (5 h)	A+B+D+E+F	4, 6, 9, 10	At 6 months' follow-up: mean disability scores on Roland-Morris disability questionnaire (0–24) in intervention and control group differed by 0.1	Ineffective: p=0.95	16.5
Tinsel et al (2013)24	Unstated number of GPs (36 GP practices)	4	SDM in context of hypertension (6 h)	C+D+E	2, 3, 8, 10, 11	At 18 months' follow-up, mean systolic blood pressure decreased by 4.5 mmHg in intervention group compared with 4.8 mmHg in control group	Ineffective: p=0.043, (97.5% CI −0.189; 3.69)	17.4
Légaré et al (2012)25	149	4	DECISION+2: SDM in context of antibiotic treatment (4 h)	A+B+C+D+E	3, 10, 11, 12	Absolute difference in percentage of patients deciding to use antibiotics of 25% between intervention and control group	Effective: RR=0.48 (0.34-0.68 95% CI)	18
Légaré et al (2011)26	30	4	DECISION+: SDM in context of antibiotic treatment (6 h)	A+B+C+D+E	3, 10, 11	Absolute difference in percentage of patients deciding to use antibiotics of 28% between intervention and control group	Ineffective: p=0.08	17.4
Tilburgs et al (2019)21	38	3	Two workshops on SDM in context of ACP (6 h)	A+B+D+E	1, 3, 4, 5, 6, 8, 9, 10	At 6 months' follow-up: ACP initiated in 35.4% more patients in intervention group compared with control group	Effective: p=0.002	16.8
Sanders et al (2017)23	42	3	Two training sessions on SDM in context of low back pain (5 h)	A+B+D+E+F	4, 6, 9, 10	OPTION score (0–100), measuring involvement of patients in decision-making, of 38.5 in intervention group versus 23.7 in control group	Effective: p<0.05	16

ACP = advanced care planning; HbA1c = haemoglobin A1c; MERSQI = Medical Education Research Study Quality Instrument; SDM = shared decision-making.

^aKirkpatrick's model¹³: (1) Reaction, (2) Knowledge, (3) Behaviour, (4) Patient outcomes.

^bThe six steps of SDM²²: A=invite patient to participate, B=present options, C=provide information on benefits and risk, D=assist patient in evaluating options, E=facilitate decision-making, F=assist patient with implementation of the decision.

^cTeaching methods: (1) Demonstration of SDM, (2) Group discussion, (3) Instructional guidance, (4) Near-peer teaching, (5) Peer observation, (6) Feedback on performance, (7) Reflection, (8) Roleplay, (9) Small-group learning, (10) Theory/presentation, (11) Videos, (12) Web-based learning.

Publication years ranged from 2011 to 2019. Three studies were conducted in the Netherlands,^21–23 two in Canada^25,26 and two in Germany.^20,24 Participant numbers ranged from 30 to 149. Study participants were GPs in six studies^20–24,26 and GPs and GP trainees in one study.²⁵

Interventions used in SDM training for GPs

SDM training was delivered in the context of clinical processes²¹ or disease management.^20,22–26 The most taught SDM elements, as described by Wexler,¹⁶ were ‘assisting patients in evaluating options’ (n=7), ‘inviting patients to participate’ (n=6), ‘presenting options’ (n=6) and ‘facilitating decision-making’ (n=6). Only two studies described ‘Assisting patients with implementation of the decision’.

All interventions used a combination of teaching methods. The most prevalent teaching methods were theory/presentation (n=6), instructional guidance (n=5) and near-peer teaching (n=4). No study incorporated reflection. Demonstration of SDM, group discussion, peer observation and web-based learning were utilised only once each.

Outcomes of SDM training in general practice

Five studies assessed the effect of GP SDM training on patient outcomes.^{20,22,24–26} Two studies addressed the impact on clinician behaviour.^21,23 The results of the studies addressing patient outcomes are addressed first because this is higher on Kirkpatrick's levels of evaluation.¹¹

Studies assessing patient outcomes

Of the studies assessing patient outcomes,^{20,22,24–26} only one showed statistically significant effects.²⁵

Whereas DECISION+ reduced the percentage of patients choosing to use antibiotics for acute respiratory tract infections, this was not significant compared with the control group (p=0.08).²⁶ Following modifications, DECISION+2 significantly reduced the percentage of patients deciding to use antibiotics to treat acute respiratory infections (RR 0.48, 95% CI 0.34–0.68).²⁵

Furthermore, although a SDM training programme provided in the context of managing hypertension was associated with a reduction in systolic blood pressure at 18 months' post intervention, the decrease observed in the control group was greater (p=0.043).²⁴ However, the authors of this paper found this difference not to be statistically significant at the level of 2.5% (97.5% CI −0.189; 3.69).

Peer-to-peer SDM training delivered in the context of type 2 diabetes mellitus showed a reduction in haemoglobin A1c (HbA1c) at 24 months' post intervention, although this was not significant compared with the control group (p=0.15).²⁰ Additionally, following SDM training based on lower back pain, there was no significant difference in patients' Roland–Morris disability scores (p=0.95).²²

Studies assessing clinician behaviour

Both studies measuring the effect of SDM training on clinician behaviour observed statistically significant changes (p=0.002 and p<0.05).^21,23

Although Sanders et al's²² intervention did not significantly affect patient outcomes (p=0.95), it significantly improved patients' ‘observing patient involvement in decision making’ (OPTION) scores (p<0.05).²³ Moreover, a workshop on SDM in the context of advanced care planning (ACP) resulted in GPs initiating ACP with a significantly greater percentage of patients in the intervention group (p=0.002).²¹

Study quality

The mean overall MERSQI score was 17.2 (range=16–18, SD=0.8). Given that all studies had MERSQI scores >14, they were classed as high quality.

All but two studies were conducted at three or more institutions. The mean (±SD) response rate of participants was 79.4% (±13.9) and there was full reporting of evidence of validity where applicable.

Discussion

This review identified a limited number of studies, none of which were conducted in the UK. Furthermore, the use of decision aids within interventions, differences in study follow-up periods and intervention durations could have confounded the obtained findings. However, high-quality studies were identified. Thus, this review provides an overview of SDM training methods in general practice and their outcomes. This can guide the development of GP SDM training programmes and provide avenues for further research.

SDM training was delivered in context. This is in line with the General Medical Council's recommendation that doctors' continuing professional development learning should be guided by the care they provide.²⁷ Furthermore, most SDM training programmes evaluated in this review covered all the elements of SDM, with the exception of ‘assisting patents with implementation of the decision’, which was included in two studies.^22,23 Although this could indicate that SDM training programmes generally encompass all the necessary components of SDM, it does not provide insight into the extent to which these ILOs were met, and suggests that studies did not fully reported their ILOs. Despite the inclusion of ‘assisting patients with implementation of the decision’ in the intervention ILOs in two studies,^22,23 it is difficult to conclude on the impact of this element within GP SDM training, because the outcomes of these studies are in disagreement.

Additionally, given the positive impact that SDM has been shown to have on adherence to treatment, ‘assisting patients with implementation of the decision’ might be an expected outcome of SDM, without the need for SDM interventions to explicitly state this.²⁸ This could provide a reason for the omission of this SDM element in the interventions studied in this review.

Nonetheless, as illustrated by the challenges encountered by the authors in establishing the ILOs of the programmes, there is a lack of clarity in the objectives of the SDM training programmes. Considering the important role of ILOs in driving learning and facilitating both teachers and learners to focus on observable results, the ILOs of SDM training programmes should be verified before their launch.²⁹ This will ensure that training programmes have clear learning outcomes that involve sufficient components of SDM to enable an effective impact on clinical practice.

GP SDM training programmes comprised a mixture of training methods, including didactic and active training methods. In this review, didactic methods (theory/presentation) were the most common training approach (n=7). This is consistent with Singh et al,⁷ who found that SDM educational programmes mostly comprised didactic components in combination with active learning. However, in this review, didactic teaching methods were most commonly incorporated with instructional guidance on implementing SDM in practice, including desktop tools with prompts (n=4),^{20,21,24–26} whereas in Singh et al's review,⁷ tools for use in clinical practice were uncommon. In the current review, some of the studies that incorporated instructional guidance on implementing SDM in practice obtained positive results on patient outcomes²⁵ (n=1) and clinician behaviour²¹ (n=1), with one study that was found to be statistically ineffective still showing results in the positive direction.²⁶ Thus, these encouraging results indicate a need for further investigation into the use of SDM promoting tools in general practice.

Although collaborative methods of teaching (near-peer teaching, group discussion and small-group learning) were common (n=5),^20–24 their impact on patient outcomes and behaviour were inconclusive. This might also require further investigation.

By contrast, reflection was not explicitly used in any intervention and role play was uncommon (n=2). Consistent with literature on SDM training for secondary care doctors,⁸ this review identified role play as an important training method and attained positive outcomes on clinician behaviour,²¹ although the same cannot be said for patient outcomes.²⁴ Although reflection was not obviously used as a SDM training strategy, medical education theory supports the cyclical use of linking experiences with reflection and further planning in learning.³⁰ Therefore, it is plausible that the implementation of these experiential teaching methods could enhance the effectiveness of SDM training and should be trialled in current SDM training programmes. This can be further enhanced by the provision of feedback, which had a positive effect on GP SDM performance.^21,23

Moreover, exemplar methods of teaching (demonstration of SDM, peer observation and use of videos) were positively associated with improvements in patient outcomes²⁵ and GP behaviour.²¹ This might be because of the illustration of expected SDM performance, as supported by findings that propose behaviour modelling as an effective approach to training.³¹ Thus, the use of exemplars and role play in training could be useful in facilitating SDM in practice.

Given the minimal use of web-based learning in this review (n=1), it is not possible to reach any conclusions regarding their use in GP SDM training. However, the reduction in training time by 2 h, through incorporating an online tutorial within DECISION+,²⁶ might have been responsible for the positive effect on patient outcomes observed by DECISION+2.²⁵ This warrants further investigation into the effects of web-based learning, as well as intervention duration.

Although these findings suggest that GP SDM training leads to improved clinician SDM behaviour, studies assessing the effect of SDM training on patient outcomes largely failed to produce significant improvements.^{20,22,24–26} This could highlight the challenges of using patient outcomes as the outcome measure, as observed in previous systematic reviews.^32,33 This is alluded to by Tinsel et al's findings in which a greater decrease in systolic blood pressure was observed in the control group versus the intervention group.²⁴ However, this might have been because of fluctuations in the systolic blood pressures of participants in the intervention group, rather than the effect of the SDM training programme. Thus, in congruence with Singh et al's systematic review,⁷ outcome assessment for SDM training programmes needs to be improved.

Strengths and limitations

This review is strengthened by the systematic search strategy and rigorous selection criteria used, which ensured that only high-quality studies were included. Additionally, the incorporation of an existing SDM model to describe the individual components of SDM incorporated in each intervention enhanced the construct validity of this review.¹⁶ Although there are many SDM models and there is no set standard on the optimal model, this model was selected because of its simplicity and applicability to clinical contexts.

Despite the inclusion of high-quality literature, this review was limited by the inclusion of only peer-reviewed literature and the absence of grey literature. Thus, potentially useful studies might have been missed. The fact that included studies were limited to those published in the English language further adds to this consequence. Furthermore, items on the data-extraction tool, such as the individual elements of SDM, were subject to interpretation by the authors, which might have reduced the inter-rater reliability. However, this was offset by the authors revisiting the manuscripts together where disputes arose, thus preserving reliability. Finally, intervention cost-effectiveness was not considered in this review. Given the finite provision of resources among institutions, the perceived success of such interventions needs to be balanced with their costs.

Conclusion

Despite the paucity of literature on this topic, this review provides a useful overview of SDM training interventions for GPs and their effectiveness. The SDM model can be applied successfully to a variety of clinical contexts, indicating its applicability to the range of consultations within general practice. However, given the lack of clarity in the ILOs of the SDM interventions, standardisation of SDM training to ensure that set criteria is met by all future programmes is recommended.

Although the SDM training interventions did not appear to significantly improve patient outcomes, they improved clinician behaviour and incorporation of SDM within consultations. Given that this could lead to higher patient satisfaction, incorporation of SDM teaching within GP training is still warranted.

Despite the prominent use of didactics in combination with instructional guidance, followed by collaborative methods, their varying success raises uncertainty over their necessity in GP SDM training. Nonetheless, despite being less frequently used, studies that incorporated teaching methods targeting practical SDM skills, including role play, demonstration of SDM, peer observation and videos (n=4),^21,24–26 showed some positive outcomes.^21,25 Given the minimal findings on these methods in this review, further enquiry into their use is necessary. The cost-effectiveness of such interventions also needs to be considered in future studies.

Key practice implications

• The SDM model can be applied successfully to a variety of clinical contexts, indicating its applicability to the range of consultations within general practice.

• Standardisation of SDM training is recommended to ensure that set criteria are met by all future programmes

• GP SDM training can improve incorporation of SDM within consultations, which could lead to higher patient satisfaction.

• Further research into the effects of GP SDM training interventions on patient outcomes, the use of active teaching methods and the cost-effectiveness of interventions is needed.