Impact of a cognitive behavioural therapy training program on family physician practices

Abstract

Objective

To investigate changes in FPs’ self-reported clinical practices after participation in a comprehensive 1-year cognitive behavioural therapy (CBT) training course.

Design

Cross-sectional study.

Setting

Norway.

Participants

Family physicians.

Main outcome measures

Impact of the CBT training course on FPs’ delivery of CBT to patients and their use of referral to specialized mental health care providers.

Results

Of the 217 FPs who had participated in the training course between 2009 and 2016, 124 completed the survey (response rate=57.1%); 99.2% of participating FPs reported using CBT tools daily in patient consultations, more than three-quarters reported changing the way they organized their workdays to accommodate CBT, and 75.0% reported using structured CBT consultations at least monthly after completing the course. The most common patient groups receiving structured CBT were those experiencing mild or moderate depression (22.8%), anxiety disorders (30.4%), or a combination of an anxiety disorder and depression (43.5%). The odds of making fewer referrals to specialized mental health care providers were 5.4 times higher among FPs who used Socratic questioning (P=.02), 4.7 times higher among those who provided consultation summaries (P=.01), and 3.3 times higher among those who had participated in a refresher course (P=.05).

Conclusion

Comprehensive training in CBT promotes the use of CBT tools and strategies in family practice. Further longitudinal research (ideally randomized controlled studies) on patient outcomes related to CBT provided in family practices is required.

There were approximately 5400 family physicians in Norway in 2021,¹ and the average panel size of Norwegian FPs in 2016 was 1120 patients.² A study published in 2015 found 12.2% of appointments with FPs in Norway had mental health problems cited as the main reason for the encounters, and 26.4% of appointments included a presentation of a mental health problem, mostly mild or moderate depression or anxiety disorders.³ This is consistent with data from other countries.⁴

Primary care physicians are well positioned to be able to intervene early in the course of mental health disorders and are gatekeepers to specialist services.⁵ Yet only a small fraction of patients with mental health disorders are offered adequate treatment,^5,6 and FPs may lack the skills needed to treat patients with mental health problems.⁷ National Institute for Health and Care Excellence guidelines recommend cognitive behavioural therapy (CBT) for the most common mental health disorders.⁸ Cognitive behavioural therapy strategies and skills^9-20 have been shown to be helpful in treating and managing patients with mental health disorders in primary care,^8,21-23 and seems to be effective in treating patients with medically unexplained symptoms such as non-chronic low back pain and some somatic disorders (eg, irritable bowel syndrome).^24-27 By employing CBT strategies and skills, FPs can help patients cultivate healthier and more realistic thinking and behaviour.²⁸

While some Norwegian FPs have participated in brief (usually 2-day) CBT courses, these programs may not provide them with sufficient knowledge to deliver effective treatment.^6,29,30 A Canadian study demonstrated that a 10-session course delivered as 2 hours per week showed promise in teaching complex CBT skills to mental health care providers.³¹ To provide more comprehensive training, in 2009 the Norwegian Medical Association and the Norwegian Association for Cognitive Behavioral Therapy introduced a 1-year CBT course, which more than 200 FPs have now completed.²⁶

The course is structured around two 2-day and three 1-day workshops, totalling 7 full days over a year. The workshops include lectures, group work, and role play and cover an introduction to CBT, treatment of mild or moderate depression (as well as the most common anxiety disorders and medically unexplained physical symptoms), and special challenges encountered when using CBT. The content was designed to be of maximum relevance to FPs. Between workshops, physicians receive a reading list of relevant articles and participate in smaller evening counselling groups led by a trained peer supervisor using videos of consultations conducted by participating FPs. The course ends with a 2-hour written examination.

Participating FPs have stated 2 main reasons for participating in the course. The first reason is to treat their patients more effectively themselves, as supportive talk and medication alone do not appear to be sufficient, and specialized mental health care is not easily accessed.²⁶ The second reason is to do something more active for patients than just “containing“ their problems, or serving in an unsatisfying “container” function, where FPs merely listen to their patients’ complaints without being able to develop an active treatment plan.⁹ Participants want to use CBT techniques to take a more active role in their patients’ treatment and create shared responsibilities for their recovery.

All FPs who complete the 1-year course are offered an annual 2-day follow-up course that addresses specific, more advanced CBT skills.

The purpose of this study was to understand family physicians’ experiences with a comprehensive 1-year CBT training course and how it affected their use of CBT, the kinds of consultations in which they used CBT, and their referral patterns to psychiatric specialist services. In addition, we wanted to learn about the types of patients being treated, the CBT strategies and skills used, the impact of CBT on FPs’ satisfaction, whether it changed their containing role, and their perspectives on whether this benefited their patients.

METHODS

Study design

The target group for the study was all FPs who had attended 1 of the 8 annual 1-year CBT courses between 2009 and 2016. In 2017 we sent them an introductory email inviting them to participate in our study. Those who agreed to participate received a second email with a link to the questionnaire, which was sent using the Questback program. Two additional email invitations were sent to those who had not responded to the first.

The questionnaire was based on a previous qualitative study²⁶ and had been evaluated in a pilot study with 5 FPs with CBT training and adapted based on their feedback. It included questions about practice characteristics: sex, number of patients on their panel, number of FPs in the practice, type of practice funding, and whether the physician was a specialist in family medicine.

The CBT-related questions included the number of monthly structured CBT consultations FPs provided and their use of CBT strategies and skills, such as setting an agenda, using problem lists (ie, the patient makes a prioritized list of problems), using Socratic questioning (open neutral questioning), using the ABC (Activating event, Belief, Consequence) model,³² assigning a daily activity log, discussing the panic cycle (which describes the connection between triggers, emotions, bodily reactions, and catastrophic thinking), discussing the anxiety curve (the effect of trigger on anxiety), using case formulation (exploring the relationship between life experiences and anxiety development), providing psychoeducation, providing a summary at the end of the consultation, and assigning and reviewing homework.

Participants

A total of 217 FPs were contacted through the introductory email and follow-ups, yielding 124 participants (response rate=57.1%).

Outcome and statistical analysis

We used descriptive statistics to calculate frequencies and percentages. Chi-square tests were used to investigate associations among dichotomous variables. Binary responses were analyzed using logistic regression. Two steps preceded the modelling of these data. Univariate (unadjusted) logistic regression models were first fitted to the data. All covariates with P≤.20 and P≤.05 in the univariate analysis were then used to fit 2 separate adjusted logistic regression models. The models were compared using the Bayesian information criterion, which states that among nested models, the model with the smallest Bayesian information criterion is the better fit. To assess the accuracy of the final fitted logistic regression model, we plotted the sensitivity of the test against specificity (the false-positive rate) to produce the receiver operating characteristic curve (Figure 1), which had an area under the curve of 0.80. All analyses were performed using Stata SE 15, and the statistical significance level was set at P≤.05.

Figure 1.

Receiver operating characteristic curve for CBT referrals: The area under the curve suggests an 80% chance of correctly distinguishing FPs who made few referrals from those who made many (N=124).

RESULTS

Of the 124 participating FPs, 63.7% were female, compared with 42.3% in Norway’s overall FP population (Statistics Norway data, 2017), and 78.2% were family medicine specialists. Among the respondents, 22.6% had finished the course within the previous year, 32.3% within the previous 2 years, and 45.2% within 3 or more years prior to the survey.

Practice size and organization

Among participants, 55.6% had between 700 and 1199 patients, 42.7% had between 1200 and 1699 patients, and 1.6% had between 1700 and 2199 patients (Table 1). In terms of funding models, 92.7% of participants were self-employed and publicly funded (paid by per capita grant), 5.6% had fixed salaries from their municipalities, and 1.6% obtained their income from patient payments only (Table 1). This is similar to data for all Norwegian FPs (Statistics Norway data, 2017).

Table 1.

Characteristics of FPs who participated in the CBT training program and of their practices

CHARACTERISTIC	FEMALE FPs (n=79)	MALE FPs (n=45)	TOTAL (N=124)
Percentage of respondents, %	63.7	36.3	100.0
Panel size, n (%)
• 700-1199	42 (53.2)	27 (60.0)	69 (55.6)
• 1200-1699	35 (44.3)	18 (40.0)	53 (42.7)
• 1700-2199	2 (2.5)	0 (0.0)	2 (1.6)
FPs in practice, n (%)
• 1	3 (3.8)	1 (2.2)	4 (3.2)
• 2-3	27 (34.2)	6 (13.3)	33 (26.6)
• 4-5	35 (44.3)	21 (46.7)	56 (45.2)
• 6-7	11 (13.9)	11 (24.4)	22 (17.7)
• >7	3 (3.8)	6 (13.3)	9 (7.3)
Funding of practice, n (%)
• Fixed salary	4 (5.1)	3 (6.7)	7 (5.6)
• Per capita grant	74 (93.7)	41 (91.1)	115 (92.7)
• Patient payment	1 (1.3)	1 (2.2)	2 (1.6)
Family medicine specialization, n (%)
• Specialist	61 (77.2)	36 (80.0)	97 (78.2)
• Non-specialist	18 (22.8)	9 (20.0)	27 (21.8)
Recertifications, n (%)
• 0	10 (12.7)	3 (6.7)	13 (10.5)
• 1-3	34 (43.0)	15 (33.3)	49 (39.5)
• >3	35 (44.3)	27 (60.0)	62 (50.0)
Years since CBT course, n (%)
• 0-1	22 (27.8)	6 (13.3)	28 (22.6)
• 2-3	25 (31.6)	15 (33.3)	40 (32.3)
• >3	32 (40.5)	24 (53.3)	56 (45.2)

CBT—cognitive behavioural therapy.

Patients

Participating FPs reported that the most common patient groups receiving structured CBT were those experiencing mild or moderate depression (22.8%), anxiety disorders (30.4%), or a combination of an anxiety disorder and depression (43.5%).

Structured and unstructured CBT consultations

Among FP participants, 75.0% reported using structured CBT consultations at least monthly and 99.2% reported using CBT skills during regular patient visits daily. While the numbers of CBT consultations provided per month and per day did not show consistent differences based on FP sex, female FPs were more likely to provide only 1 to 4 structured CBT consultations per month (P=.03), male FPs were more likely to provide 13 to 16 structured CBT consultations per month (P=.04), and female FPs were more likely to provide 9 to 12 unstructured CBT consultations per day (P=.05). Male FPs were generally more likely than female FPs to provide higher numbers of structured CBT consultations (Figure 2).

Figure 2.

Proportions of female and male FPs who A) provide structured CBT consultations per month and B) provide unstructured CBT consultations per day

All the elements of CBT taught in the course were used in structured CBT consultations and in other types of visits (Figure 3).³² More than half of the FPs used problem lists, Socratic questioning, psychoeducation, and consultation summaries in all types of clinical consultations.

Figure 3.

Proportions of FPs using CBT elements in structured consultations versus other consultations

Changes in family physicians’ practices

Respondents described 3 important changes in the way they practised after having completed the course. Two-thirds of respondents said they tended to make fewer referrals to specialized mental health care, two-thirds said they were much less likely to take a passive “containing” role with patients, and three-quarters said they had changed their workweeks to include more dedicated time for CBT.

Factors associated with fewer referrals. Data from model 2 (Table 2)³² show factors significantly associated with fewer referrals to mental health professionals. The odds of making fewer referrals were 5.4 times higher among FPs who used Socratic questioning (P=.02), 4.7 times higher among those who made consultation summaries (P=.01), and 3.3 times higher among those who had participated in a CBT refresher course (P=.05).

Table 2.

Factors associated with fewer patient referrals to mental health professionals

FACTOR	UNIVARIATE ANALYSIS OR (95% CI), P VALUE	MODEL 1* (BIC=133.09) OR (95% CI), P VALUE	MODEL 2* (BIC=124.32) OR (95% CI), P VALUE
Asking Socratic questions	5.31 (1.59-17.72), P=.01	5.89 (1.32-26.25), P=.02	5.39 (1.27-22.84), P=.02
Providing summaries at consultation end	4.94 (1.74-14.07), P<.01	5.30 (1.40-20.08), P=.01	4.72 (1.37-16.32), P=.01
Participating in a supervised education group	2.60 (1.22-5.56), P=.01	1.09 (0.35-3.35), P=.88	1.06 (0.34-3.25), P=.92
Having fixed time for CBT	2.78 (1.21-6.37), P=.02	1.44 (0.45-4.58), P=.54	1.44 (0.46-4.47), P=.53
Participating in a refresher course	2.90 (1.34-6.26), P=.01	3.52 (1.04-11.86), P=.04	3.32 (1.01-10.95), P=.05
Having an office colleague who uses CBT	4.02 (1.42-11.35), P=.01	1.99 (0.50-7.89), P=.33	2.09 (0.56-7.89), P=.28
Having economic incentives	2.60 (1.10-6.13), P=.03	1.89 (0.54-6.67), P=.32	1.82 (0.53-6.30), P=.34
Providing psychoeducation	1.88 (0.71-4.95), P=.20	0.76 (0.21-2.81), P=.68	NA
Assigning homework	2.59 (0.81-8.24), P=.11	0.86 (0.16-4.49), P=.86	NA
Setting an agenda	0.75 (0.28-2.06), P=.58	NA	NA
Using problem lists	1.12 (0.31-4.07), P=.86	NA	NA
Reviewing homework	1.36 (0.30-6.13), P=.69	NA	NA
Using the ABC model	1.30 (0.35-4.85), P=.69	NA	NA
Assigning a daily activity log	1.33 (0.55-3.20), P=.52	NA	NA
Discussing the panic cycle	0.97 (0.40-2.36), P=.94	NA	NA
Using the anxiety curve	1.69 (0.66-4.29), P=.27	NA	NA
Using a case formulation approach	0.60 (0.25-1.46), P=.26	NA	NA
Evaluation	1.11 (0.46-2.70), P=.81	NA	NA
Cooperating with local specialists	1.32 (0.54-3.21), P=.54	NA	NA
Having patients experience positive effects	2.03 (0.61-6.72), P=.25	NA	NA

ABC—Activating event, Belief, Consequence³²; BIC—Bayesian information criteria; CBT—cognitive behavioural therapy; NA—not applicable; OR—odds ratio.

^*Models 1 and 2 are based on variables with P≤.20 and P≤.05, respectively, from the univariate analysis. Model 2 was chosen for the analysis because of its smaller BIC estimate.

Factors associated with avoidance of “containing” role. We found no factors significantly associated with reducing the container function feeling among FPs in adjusted model 2 (Table 3).³² In the unadjusted univariate analysis, the use of Socratic questioning, psychoeducation, assignment of homework, participation in supervised education groups, having patients experience a positive effect, participation in a refresher course, and having an office colleague using CBT were statistically significantly associated with a higher probability of avoiding the containing role with patients.

Table 3.

Factors associated with FPs being less likely to occupy a container role (ie, listening to patients’ complaints without developing an active treatment plan)

FACTOR	UNIVARIATE ANALYSIS OR (95% CI), P VALUE	MODEL 1* (BIC=138.35) OR (95% CI), P VALUE	MODEL 2* (BIC=125.49) OR (95% CI), P VALUE
Asking Socratic questions	4.21 (1.30-13.67), P=.02	3.28 (0.74-14.47), P=.12	3.09 (0.79-12.10), P=.11
Providing psychoeducation	3.60 (1.35-9.63), P=.01	1.89 (0.57-6.29), P=.30	2.10 (0.66-6.75), P=.21
Assigning homework	4.21 (1.30-13.67), P=.02	2.16 (0.44-10.76), P=.35	2.58 (0.56-11.81), P=.22
Participating in a supervised education group	2.25 (1.05-4.82), P=.04	1.03 (0.32-3.28), P=.96	1.00 (0.32-3.09), P=.99
Having patients experience positive effects	7.50 (1.91-29.50), P<.01	2.23 (0.12-40.73), P=.59	2.44 (0.12-48.02), P=.56
Participating in a refresher course	2.44 (1.13-5.27), P=.02	2.85 (0.84-9.66), P=.09	2.62 (0.83-8.23), P=.10
Having an office colleague who uses CBT	2.81 (1.06-7.50), P=.04	3.64 (0.79-16.72), P=.10	4.11 (0.91-18.47), P=.07
Assigning a daily activity log	2.03 (0.82-5.02), P=.13	0.87 (0.27-2.76), P=.81	NA
Using the anxiety curve	1.98 (0.77-5.10), P=.16	1.29 (0.40-4.24), P=.67	NA
Evaluation	2.05 (0.83-5.08), P=.12	1.62 (0.51-5.16), P=.41	NA
Setting an agenda	1.44 (0.55-3.81), P=.46	NA	NA
Using problem lists	0.79 (0.20-3.18), P=.74	NA	NA
Reviewing homework	1.53 (0.34-6.88), P=.58	NA	NA
Using the ABC model	0.51 (0.10-2.52), P=.41	NA	NA
Discussing the panic cycle	1.43 (0.58-3.52), P=.44	NA	NA
Using a case formulation approach	1.18 (0.47-2.96), P=.73	NA	NA
Providing summaries at the end of consultations	1.06 (0.36-3.13), P=.91	NA	NA
Cooperating with local specialists	1.49 (0.60-3.71), P=.39	NA	NA
Having fixed time for CBT	1.49 (0.68-3.29), P=.32	NA	NA
Having economic incentives	1.63 (0.71-3.70), P=.25	NA	NA

ABC—Activating event, Belief, Consequence³²; BIC—Bayesian information criteria; CBT—cognitive behavioural therapy; NA—not applicable; OR—odds ratio.

^*Models 1 and 2 are based on variables with P≤.20 and P≤.05, respectively, from the univariate analysis. Model 2 was chosen for the analysis because of its smaller BIC estimate.

Changes in workday organization. Most participants (75.5%) reported changing the way they organized their workdays in order to accommodate CBT. These changes included 12.9% reserving 1 day a week for CBT, 37.1% setting aside specific time for CBT consultations, and 58.4% doubling the time they allotted for CBT sessions.

DISCUSSION

Our findings indicate that following participation in a 1-year CBT course, FPs changed how they managed patients with mental and somatic health problems, and at least some participants were able to maintain this for more than 3 years after completion of the course. Family physicians in this study reported making fewer referrals to psychiatric specialist services and were less likely to serve in a containing role. Almost all FPs reported using CBT strategies and skills on a daily basis.

Although many FPs complete CBT courses of varying duration and teaching methods, the usefulness of these courses is seldom evaluated. A study by King et al from London, England, is one of the few that has addressed FPs’ experiences with CBT training in shorter courses.²⁹ The researchers conducted a parallel group, cluster randomized controlled trial of a package of 4 half-days of brief CBT training and concluded that FPs need more training to acquire sufficient skills to help patients with depression. Our training is provided over 12 months and is more comprehensive than the course for FPs in London,²⁹ which may explain the differences in our findings.

More than three-quarters of our respondents reported changing the organization of their workdays to accommodate CBT, suggesting they perceived CBT to be useful.

Although mild or moderate depression and anxiety disorders were the most common disorders treated with CBT, all respondents reported using CBT skills when treating other diagnoses such as pain, psychosomatic disorders, and medically unexplained symptoms.

Participants used most of the CBT skills taught in the courses, which they found easy to integrate within a primary care visit. The 4 most popular, used in more than 50% of all consultations, were problem lists, Socratic questioning, psychoeducation, and consultation summaries. These were also associated with reductions in the number of patient referrals and in family physicians’ reported feelings of “helplessness” when treating these patients.

It is also worth noting that FPs who participated in small follow-up group meetings with colleagues after the course to maintain their skills anecdotally reported making fewer referrals to specialist services, and they reported feeling more effective in managing patients’ problems.

Strengths and limitations

A 57.1% response rate after 2 reminders is considered acceptable for a Questback survey.³³ The respondents’ population profile largely corresponded to the profile of FPs in Norway. New survey responses after each reminder round did not change the results and supported the notion that the selection of respondents was reasonably representative. It is possible, however, that respondents with more positive experiences of the CBT courses were overrepresented in our study.

Implications for clinical practice and for future research

Comprehensive CBT training promotes the use of CBT skills and strategies in primary care, and appears to facilitate change in the way physicians practice. The use of CBT may contribute to earlier interventions with people experiencing mental distress, reducing the burden of a person’s illness and fewer pressures on the second-line services. Our findings may also be helpful in planning family medicine residency training and continuing professional development offerings.

Conclusion

As a result of CBT training, referrals to specialized mental health services decreased and FPs were more satisfied and less frustrated in being able to provide more active and planned care for patients.

Editor’s key points

• ▸ Cognitive behavioural therapy (CBT) strategies have been shown to be effective in treating and managing patients with mental health disorders in primary care, but brief training programs, such as 2-day courses, may not provide FPs with sufficient knowledge to deliver effective treatment.

• ▸ Following a year-long CBT training program in Norway that also provides annual refresher courses, 75.0% of participants reported providing structured CBT consultations at least monthly and 99.2% reported using CBT skills daily for regular patient visits. Mild or moderate depression and anxiety disorders were the most common disorders treated with CBT.

• ▸ The 4 most popular CBT tools, used in more than half of all consultations, were problem lists, Socratic questioning, psychoeducation, and consultation summaries. The CBT tools and strategies associated with FPs referring fewer patients to specialized mental health care providers were the use of Socratic questioning, the provision of consultation summaries, and participation in a refresher course on CBT.

Points de repère du rédacteur

• ▸ Il a été démontré que les stratégies de thérapie cognitivo-comportementale (TCC) sont efficaces dans le traitement et la prise en charge en soins primaires des patients atteints de troubles de santé mentale, mais des programmes de formation courts, comme d’une durée de 2 jours, pourraient ne pas fournir aux MF assez de connaissances pour offrir un traitement efficace.

• ▸ En Norvège, après un programme de formation en TCC d’une durée d’un an, qui offre aussi des cours de rafraîchissement annuels, 75,0 % des participants ont signalé offrir des consultations en TCC structurées au moins chaque mois, et 99,2 % ont dit se servir de leurs compétences en TCC chaque jour au cours des visites régulières de leurs patients. La dépression de légère à modérée et les troubles d’anxiété étaient les problèmes les plus souvent traités par la TCC.

• ▸ Les 4 outils de la TCC les plus populaires, utilisés dans plus de la moitié de toutes les consultations, étaient les listes de problèmes, le questionnement socratique, la psychoéducation et les résumés de consultation. Parmi les outils et les stratégies de la TCC qui sont associés à la réduction du nombre de demandes de consultations spécialisées en santé mentale figuraient le questionnement socratique, la fourniture de résumés de consultation et la participation à des cours de rafraîchissement sur la TCC.