Search Results
The search results yielded 1105 titles and abstracts through Medline, the Web of Science, and additional resources. There were 145 full-text studies checked for eligibility and a total of 21 RCTs met inclusion criteria (see Figure 1) [24].
Study Characteristics
Table 1 reports the characteristics of included studies for both interventions (health coaching), and control groups (usual care), including sample size, mean age of participants, intervention duration, personnel, and mode of delivery (e.g., face-to-face, telephone-based, web-based). The included studies comprised 21 RCTs published between 1950 and 2022. A total of 3848 participants were included in the 21 studies, of whom 1988 were randomised to receive coaching interventions and 1860 were allocated to control groups. The majority of studies (n=10) were conducted in the US [31, 32, 36, 37, 39-41, 43, 45, 46], two were conducted in Canada [44, 48], and Taiwan [29, 47], and the rest were conducted once in different countries including Turkey [28], South Korea [30], Norway [33], Finland [34], Germany [35], Belgium [38], and Australia[42]. In the 18 studies that reported gender of participants, 54% of participants were female. The mean age of the recruited participants was 58.9 (SD = 6.3). Due to the inconsistent reporting of other demographic and socioeconomic characteristics, such as education, ethnicity and income status, across the 21 papers we were unable to report them here. The recruitment of participants was varied and drawn from different communities including ethnic community centres [36], community health centres [30, 31, 33, 44], community advertisement [33, 35, 46, 48], primary care or hospital clinics [28, 39, 42, 43, 47] and databases [29, 34, 38, 41]. For clinical factors, including HbA1c, there were no discernible changes between the intervention and control groups at baseline. The mean HbA1c level across all studies at baseline was 8.5% (SD = 0.7). All the RCT interventions reported improvements in favour of coaching groups over control groups in reductions of HbA1C. The effect size (d) of interventions was varied across the included studies and ranged from 0.01 to 1.3. Two interventions had a large effect size (d=1.3 and d=0.8) [36, 40], and six interventions reported a medium effect size (d=0.5 to d=0.7) [29, 30, 35, 41, 44, 45]. The remaining thirteen interventions had small (d ≥ 0.20) [28, 31, 32, 38, 42, 43, 46-48] or trivial (d < 0.20) [33, 34, 37, 39] effect sizes. The median effect of health coaching interventions on HbA1c was d = 0.30, a small effect size.
Mode of delivery and intervention duration
Health coaching was delivered through various methods including exclusive telephone-based [31, 38, 40, 42, 46, 48], exclusive web or mobile-based remote patient monitoring/electronic assistance (ERPM/EA) systems [37] or in combinations of face-to-face and telephone-based [28, 29, 36, 39, 41, 44, 45, 47]; face-to-face and ERPM/EA [30] telephone-based and ERPM/EA [33-35] or face-to-face, telephone-based and ERPM/EA [32, 43]. Interventions provided both in person and over the phone were shown to be more successful, with median d=0.5 (range = 0.1-1.3) [28, 29, 36, 39, 41, 44, 45, 47], compared to interventions delivered exclusively over the phone, with median d=0.3, range of 0.6 (0.2-0.8) [31, 38, 40, 42, 46, 48]. Only six studies reported follow-ups after interventions, which ranged from 6-12 months [28, 33, 35, 38, 42, 48]. One study recommended that longer exposure to health coaching sessions could lead to more effective outcomes [48]. However, because it was the only intervention that claimed this and had a medium impact (d=0.3), this conclusion should be regarded with caution. In addition, it was indicated that interventions lasting six months or less were found to have greater impacts on the reduction of HbA1c than any other programmes' durations with median d=0.8 (range=0.7-1.3) [35, 36, 40].
Delivery personnel
Different people delivered the health coaching interventions. In four studies, the health coaching intervention was delivered by untrained personnel [28, 29, 31, 42, 43], while the remaining 17 interventions reported training of the interventionist on health coaching. Seven studies relied on nurses to deliver coaching sessions [30, 31, 33, 36, 38, 44, 48], four studies provided interventions by trained health coaches [32, 34, 35, 37], and only one study was delivered by health coaches certified by the International Coach Federation (ICF) [49]. The remaining interventions were delivered by different professionals, including dental care providers [28], community health workers [36], dieticians [42], medical staff [39,45], pharmacists [29], psychologists [46], college students [43], peer patients, [41], and physicians [30]. The median effect size of interventions delivered by trained health coaches was d=0.3 (range = 0.01-1.3) versus a median of 0.3 with a range of d=0.3 (0.2-0.4) of those delivered by untrained health coaches.
Behavioural framework and theory use
The heterogeneity of interventions was evident in relation to the employed approaches and underpinning theories. Out of the 21 papers, five studies did not report the use of theories [29, 30, 31, 35, 37, 42]. The remaining 15 were grounded in different theories or frameworks. Most studies employed motivational interviewing [28, 32, 33, 36, 38, 41, 44, 45, 47, 48], two studies used cognitive-behavioural therapy [40, 44], two studies used the transtheoretical model [33, 39], and self-efficacy theory and social-cognitive theory were each used once [28]. The median effect size of MI-based interventions was d=0.4 with a range of 1.2 (1.3-0.1). In contrast, the remaining interventions that did not use MI had a median effect size of d=0.2 and a range of 0.8 (0.8-0.01). Intervention details, such as the details of the coaching curriculum adopted were absent from many papers [29, 30, 33, 38, 44, 45]. Few studies provided a health coaching protocol with details of session activities [28-31, 33, 35, 37-39, 44]. Only one study was identified as a good example of an intervention supported by clear descriptions of the health coaching programme [43].
Identified BCTs
A total of 23 BCTs were identified across the 21 studies reviewed (see Table 2). Interventions were varied in terms of the number of BCTs that were utilized in each intervention, ranging from 0 to 9 BCTs. The median of BCTs used across all interventions was 5. Table 2 lists the BCTs used in each intervention.
The most frequently coded BCT was 1.1 goal setting (behaviour), which has identified in 14 interventions (60.8%) [28, 31-36, 39-44, 47]. 1.2 problem solving was the second most commonly identified BCT, reported in 10 interventions (43.5%) [32, 33, 36-40, 42, 43, 46]. 3.1 social support (unspecified) was reported in four studies [32, 34, 37, 39]. 1.7 review outcome goals, 1.8 behavioural contract, 2.2 feedback on behaviour, 4.1 instruction on how to perform a behaviour, 8.7 graded tasks, 12.5 adding objects to the environment, and 2.5 monitoring outcome(s) of behaviour by others without feedback were each used once in six interventions (4%) [28, 30, 37, 38, 40, 42]. No BCTs were identified in one study [45]. There was no evidence of an association between the number of BCTs used in an intervention and its effect size (see Figure 2). Table 2 provides more details about the frequencies of the BCTs used in each intervention.
BCTs and intervention effectiveness
Due to the heterogeneity of the studies, it was not possible to perform a meta-analysis, and instead, an exploratory analysis was conducted to assess the effectiveness of each intervention. An overview of the use of different BCTs and effect sizes found in each study is presented in Table 2. The most effective intervention based on the effect size (d=1.3), used four BCTs: 1.1 goal setting (behaviour), 1.2 problem Solving, 1.3 goal setting (outcome), and 2.4 self-monitoring of outcome of behaviour [36]. The other intervention with a large effect size (d=0.8) reported nine BCTs: 2.7 feedback on outcomes of behaviour, 1.1 goal setting (behaviour), 3.1 social support (unspecified), 1.7 review outcome goals, 1.2 problem Solving, 1.4 action planning, 8.7 graded tasks, 1.6 discrepancy between current behaviour and goal, and 10.4 social reward [40].
Two BCTs that were used in the two most effective interventions [36, 40], i.e., 1.1 goal setting (behaviour) and 1.2 problem solving, were also the most used BCTs in three out of six interventions with a medium effect [35, 41, 44]. The use of 2.3 self-monitoring of behaviour and 2.4 self-monitoring of outcome(s) of behaviour in interventions with medium effect sizes was reported twice [35, 44]. The most commonly used BCT across those with small or trivial effect was 1.2 problem solving, which was reported in seven of interventions with small or trivial effect [32, 33, 37, 39, 42, 43, 46]. 1.1 goal setting (behaviour) was the second frequently used BCT which was reported in six [32-34, 39, 42, 43], and 3.1 social support (unspecified) reported in four studies [32, 34, 37, 39].
Quality of the included studies
Although some studies showed good methodological quality due to their low bias [29, 34, 35, 38, 47], the majority were weak because of either high or unclear risk of bias [28, 30-33, 37, 39-46, 48]. Twelve of the 21 studies 21 [29, 31, 33-35, 38, 40, 42, 44, 45, 47, 48] described the method of randomization generation and 11 studies [29,31,34,35,38,41,42,44,45,47,48] used a concealed allocation schedule. The methodological quality of blinding participants and personnel on the assignment of participants to study groups were generally low due to either high or unclear bias in procedures across most studies and insufficient detail. Across all the included studies, attrition bias and selective outcome reporting bias were low and not detected. Table 3 and Figure 3 provide further details about the quality of the included studies.