Evaluation of a multicomponent intervention consisting of education and feedback to reduce benzodiazepine prescriptions by general practitioners: The BENZORED hybrid type 1 cluster randomized controlled trial.

BACKGROUND: Current benzodiazepine (BZD) prescription guidelines recommend short-term use to minimize the risk of dependence, cognitive impairment, and falls and fractures. However, many clinicians overprescribe BZDs and chronic use by patients is common. There is limited evidence on the effectiveness of interventions delivered by general practitioners (GPs) on reducing prescriptions and long-term use of BZDs. We aimed to evaluate the effectiveness of a multicomponent intervention for GPs that seeks to reduce BZD prescriptions and the prevalence of long-term users. METHODS AND
FINDINGS: We conducted a multicenter two-arm, cluster randomized controlled trial in 3 health districts in Spain (primary health centers [PHCs] in Balearic Islands, Catalonia, and Valencian Community) from September 2016 to May 2018. The 81 PHCs were randomly allocated to the intervention group (n = 41; 372 GPs) or the control group (n = 40; 377 GPs). GPs were not blinded to the allocation; however, pharmacists, researchers, and trial statisticians were blinded to the allocation arm. The intervention consisted of a workshop about the appropriate prescribing of BZDs and tapering-off long-term BZD use using a tailored stepped dose reduction with monthly BZD prescription feedback and access to a support web page. The primary outcome, based on 700 GPs (351 in the control group and 349 in the intervention group), compared changes in BZD prescriptions in defined daily doses (DDDs) per 1,000 inhabitants per day after 12 months. The 2 secondary outcomes were the proportion of long-term users (≥6 months) and the proportion of long-term users over age 65 years. Intention-to-treat (ITT) analysis was used to assess all clinical outcomes. Forty-nine GPs (21 intervention group and 28 control group) were lost to follow-up. However, all GPs were included in the ITT analysis. After 12 months, there were a statistically significant decline in total BZD prescription in the intervention group compared to the control group (mean difference: -3.24 DDDs per 1,000 inhabitants per day, 95% confidence interval (CI): -4.96, -1.53, p < 0.001). The intervention group also had a smaller number of long-term users. The adjusted absolute difference overall was -0.36 (95% CI: -0.55, -0.16, p > 0.001), and the adjusted absolute difference in long-term users over age 65 years was -0.87 (95% CI: -1.44, -0.30, p = 0.003). A key limitation of this clustered design clinical trial is the imbalance of some baseline characteristics. The control groups have a higher rate of baseline BZD prescription, and more GPs in the intervention group were women, GPs with a doctorate degree, and trainers of GP residents.
CONCLUSIONS: A multicomponent intervention that targeted GPs and included educational meeting, feedback about BZD prescriptions, and a support web page led to a statistically significant reduction of BZD prescriptions and fewer long-term users. Although the effect size was small, the high prevalence of BZD use in the general population suggests that large-scale implementation of this intervention could have positive effects on the health of many patients. TRIAL REGISTRATION: ISRCTN ISRCTN28272199.

Introduction

Clinicians mainly prescribe benzodiazepines (BZDs) to treat anxiety and insomnia, or as adjuvants for treatment of depression [1]. Clinical guidelines advocate short-term use of BZDs [2,3] because long-term use leads to tolerance and dependence and is associated with many adverse effects, including somnolence, daytime drowsiness, memory disruption [4-6], increased risk of falls resulting in hip fracture [7-9], and motor vehicle accidents [10,11]. Other studies expressed concerns about possible links of long-term BZD use with mortality [12,13]. Long-term BZD use is particularly inappropriate for older people [14]. However, many clinicians overprescribe BZDs and chronic use is common [15,16]. Overall number of prescriptions of BZDs and BZD-like drugs (Z-drugs) has modestly decreased over the last 10 years in Europe [17]. The prevalence of BZD use varies greatly among countries and ranges from less than 15 defined daily doses (DDDs) per 1,000 inhabitants per day in the United Kingdom, the Netherlands, and Germany, to more than 85 DDDs per 1,000 inhabitants per day in countries such as Iceland, Portugal, and Spain [18].

The Spanish Medicine Agency (AEMPS) reported an average of 87.6 DDDs per 1,000 inhabitants per day during 2018 [19]. According to the most recent Spanish health survey, an average of 10% of the population in Spain reported consuming a BZD during the 2 weeks prior to the survey. Women and aged 65 years or more were the highest consumers (36%) [20].

Most BZDs are mainly prescribed by general practitioners (GPs). The high variability in prescribing BZDs among practices [21] can be explained local health policies regarding the prescription of BZDs, internal conditions of the healthcare practices, and beliefs and attitudes of the GPs regarding the benefits and risks of BZDs [22-24].

Some authors consider that discussing benefits and risk with the patient before the first prescription could be considered a key component in preventing long-term prescriptions of BZDs [22,25]. In addition, a patient who develops dependence may pose a significant challenge to a GP [22,23,26,27]. GPs can use various strategies to gradually taper BZD use in long-term users [28-34].

To change the BZD prescribing behaviors of health professionals and taper BZD use by long-term users, it is important to reduce the risk of dependence and related adverse events from BZD use. The most common deprescribing interventions include the identifying appropriate patients for deprescribing, providing education and development training to the GPs and patients, and using tailored stepped dose reduction of BZDs [35,36]. The most common implementation strategies are targeting professional behavioral changes by using printed educational materials, educational meetings, educational outreach, local opinion leaders, audit and feedback, computerized reminders, and tailored interventions [37-40]. Audit and feedback and educational meetings are widely used in clinical practice as quality improvement measures. Two systematic reviews that examined 220 randomized controlled trials showed these had a small to moderate effect on changing the behaviors of health professionals. However, there was wide variation in their impact and the extent to which they were implemented [40,41].

Some authors suggest that blending of the effectiveness and implementation stages during development of an intervention could improve the translation of the research findings into clinical practice [42,43].

The BENZORED Phase IV trial is a hybrid type 1 effectiveness and implementation study that evaluates the effectiveness and the implementation of an intervention using a GP training workshop on the appropriate use of BZDs. The intervention encourages GPs to gradually taper BZDs for long-term users and provides monthly feedback to GPs about their BZD prescriptions and access to a support web page. This study also aims to identify barriers and facilitators that affect the implementation of this intervention in primary care settings. In the present manuscript, we report the results of the effectiveness outcomes.

Methods

Study design

This is a multicenter two-arm cluster randomized controlled type 1 hybrid effectiveness–implementation trial (BENZORED trial) conducted between September 2016 and May 2018. Allocation was performed at the level of the primary health centers (PHCs) to reduce bias due to “treatment contamination.” All participant GPs from included PHCs were analyzed, and primary and secondary outcomes were at the GP level. GPs of the PHCs allocated to usual care arm received no intervention.

The details of the protocol were published elsewhere [44] (S1 Study Protocol).

The study protocol was approved by the Primary Care Research Committee, the Balearic Islands Ethical Committee of Clinical Research (IB3065/15), l’IDIAP Jordi Gol Ethical Committee of Clinical Research (PI 15/0148), and Valencia Primary Care Ethical Committee of Clinical Research (P16/024). This study followed the principles outlined in the Declaration of Helsinki (seventh revision).

Only PHCs in which at least two-thirds of the GPs agreed participate were included. Informed consent was waived by the Balearic Islands and the l’IDIAP Jordi Gol Ethical Committee of Clinical Research because the analysis used anonymous administrative data. The requirement for informed consent was not waived by the Valencian Primary Care Ethics Committee, and, therefore, from this region, only GPs who agreed and provided written informed consent were included.

Study population

PHCs from the following regions of Spain were eligible for participation: Balearic Islands (IbSalut), Catalonia (Institut Català de la Salut; Tarragona-Reus district), and Valencian Community (Conselleria de Salut Universal; Arnau de Vilanova-Llíria district). Healthcare in Spain is a public service with universal coverage and free access for the entire population. Spanish primary care system consists of 3 organizational levels: the State’s central administration agency that is responsible of the general coordination and legislation and the pharmaceutical policy; the delivery of health services are responsible of regions (Autonomous communities) and regions are organized by health districts that include PHCs. PHCs are staffed by multidisciplinary teams comprising of GPs, pediatricians, nurses, gynecologists, and physiotherapists. At baseline, we assessed PHC characteristics (total number of patient listed, proportion of patients aged 65 years or more, urban/rural setting, and training practice) and GP characteristics (age, sex, GPs with 3 years specialty training, doctorate degree, resident trainer, and years working as GP).

Randomization and masking

A computer-generated random number table was used to allocate the PHCs to a usual care (control) group or a multicomponent intervention group in a ratio of 1:1. All PHCs were randomized at the same time to maintain allocation concealment and were stratified by health districts, PHCs baseline DDDs per 1,000 inhabitants per day, and proportion of patients older than 65 years to ensure PHCs were balanced to those characteristics. GPs were not blinded to the allocation; however, the pharmacists, researchers, and trial statisticians were blinded to the allocation arm. The primary and secondary outcomes were measured using dispensed prescribing data from the electronic clinical records with predefined indicators.

Intervention

The multicomponent BENZORED intervention consisted of an educational meeting for GPs, audit and feedback, and a support web page.

Educational meetings are commonly used for continuing medical education with the aim of improving professional practice and patient outcomes. Educational meetings include courses, conferences, lectures, workshops, seminars, and symposia [40].

At the start of the trial, the GPs from the intervention group received a 2-hour educational face-to-face workshop in their PHCs. These interventions were delivered by researchers who were GPs, had great expertise in prescribing and deprescribing BZD, and provided training about appropriate procedures for prescribing BZDs. This included educational information about the pharmacological properties of BZDs (biological half-life and equivalent dose of different BZDs), indications for prescription, recommended duration of use, adverse effects, dependence, tolerance, prevalence, and consequences of long-term use. These GPs were asked to discuss with their patients harms and benefits of taking BZD before the first prescription. They received training in a structured BZD discontinuation intervention based on gradual BZD dose reduction and training based on real clinical cases [1,33].

After the initial training workshop, all participating GPs received automated monthly feedback. The audit and feedback, based on the “Feedback Intervention Theory” [45], provided a summary of the clinical performance of healthcare over a specified period of time that was aimed at changing the practices of health professional [41].

Graphical depictions of BZD prescription feedback consisted of a line graph that plotted each GP’s monthly BZD prescription rate in DDDs per 1,000 inhabitants per day for 12 months, and 2 additional line graphs that plotted the BZD prescription rate of the PHC and the health district. The GPs also received a password to access to a support web page that provided additional information to reinforce the messages of the workshop. This website included videos, descriptions of practical cases, and an information leaflet for patients about BZDs, Z-drugs, and sleep hygiene (http://benzored.es).

Outcomes

Primary outcome measure

The primary outcome measure was DDDs of BZDs per 1,000 inhabitants per day after 12 months. Total number of dispensed DDD from the N05BA (diazepam, chlordiazepoxide, potassium clorazepate, lorazepam, bromazepam, clobazam, ketazolam, alprazolam, halazepam, pinazepam, clotiazepam, bentazepam), N05CD (flurazepam, flunitrazepam, triazolam, lormetazepam, midazolam, brotizolam, quazepam, lorprazolam) and N05CF (zopiclone, zolpidem, zaleplon) groups of the Anatomical Therapeutic Chemical (ATC) Classification System code were extracted from the e-prescription databases of each health district. Total number of dispensed DDD was divided by total GP’s patient list / 1,000 × 365 days.

Secondary outcome measures

One secondary outcome measure was the proportion of all patients who were long-term BZD users after 12 months.

The other secondary outcome measure was the proportion of long-term BZD users who were aged 65 years or more after 12 months.

Long-term use was defined as a continuous prescription for any dose of BZD for a minimum of 6 months.

The RELE (IbSalut, Balearic Islands), Rec@p (Institut Català de la Salut; Tarragona-Reus district), and Receta electrónica (Conselleria de Salut Universal; Arnau de Vilanova-Llíria district) are the e-prescription systems and databases used in the 3 health districts included in the study. These databases contain information about all primary care prescription dispensed in community pharmacies, including dispensing date, pharmaceutical product (active ingredient and brand name), dose, and treatment duration.

Data management

Statistical analysis

The sample size calculation was based on the DDDs per 1,000 inhabitants per day. The mean DDDs per 1,000 inhabitants per day was 89.3 DDDs, the standard deviation was 18.7, and the intracluster correlation coefficient was 0.05 [46]. To detect a clinically meaningful difference of 5 DDDs per 1,000 inhabitants per day between the intervention and control groups, with 80% power and a two-sided α value of 0.05, at least 64 PHCs with an average of 10 GPs per PHC were needed. Adjustment for clustering was performed by calculation of 1 + (n − 1) × ρ, where n is the average cluster size and ρ is the intracluster correlation coefficient [46]. For the primary outcome, the effectiveness of the intervention was analyzed using a mixed effects Poisson regression model to account for clustering at the level of the PHC and adjusted for baseline GPs DDDs per 1,000 inhabitants per day. Also, adjusted average marginal effects for group were calculated from the mixed effects Poisson regression models. For secondary outcomes, a random effect Tobit regression with 2 censored values at 0 and 100 was also carried out to analyze the effectiveness of the intervention in the proportion of long-term BZD users after 12 months and proportion of long-term BZD users after 12 months over 65 years old adjusted for GPs baseline proportion of long-term BZD users.

Missing outcomes were accounted for using multiple imputation with chained equation [47] to estimate DDDs per 1,000 inhabitants per day from GPs lost to follow-up. Fifty imputed samples were generated, and estimates were combined using Rubin rules. Baseline values and PHC were included in the imputation model. All statistical analysis were performed using SPSS v.23.0 (IBM, Armonk, NY, USA) using a predefined analysis plan (S1 Analysis plan). Intention-to-treat (ITT) analysis was used to assess all clinical outcomes, and the results are reported in accordance with the Consolidated Standards of Reporting Trials guidelines extension for cluster trials (S1 CONSORT Checklist) [48] and the Template for intervention Description and Replication (S1 TIDieR Checklist).

Subgroup analysis

To assess whether the effect of the intervention on BZD prescription varied according to GP characteristics (sex and working years), characteristic of the GP’s practice list (percentage of patient aged 65 years old and sex) and health districts, a subgroup analysis of effectiveness that was previously planned in the protocol was performed.

Implementation process

Full details of the methods used for evaluation of the implementation of the BENZORED IV intervention were reported elsewhere [49]. To summarize, 40 semistructured interviews and 5 focus group meetings were conducted. GPs were invited to participate in an effort to ensure representativeness for the different health district locations, and they were classified as “low prescribers” or “high prescribers” based on the 12-month final evaluation.

The Consolidated Framework Implementation Research (CFIR) was used to guide development of the focus group meetings and for coding and data analysis [42]. The CFIR is a theoretical framework that provides a list of 41 constructs organized in 5 domains that can negatively or positively influence implementation. Among the 45 GPs contacted, 40 agreed to interviews.

The implementation fidelity was measured as adherence to the intervention, defined as whether “a program service or intervention is being delivered as it was designed” [50]. Adherence was measured using a questionnaire given to all GPS in the intervention group (S1 Questionnaire). This questionnaire asked the GPs to rank adherence to each of the following components of the intervention on a scale from 1 to 10: discussing benefits and risk with the patient before the first prescription; setting limits for the duration from the start of treatment; tailoring stepped dose reduction to discontinue BZD use in long-term users; reviewing BZD prescription feedback; downloading and using the patient information leaflet about BZDs, Z-drugs, and sleep hygiene; and visiting the web page to reinforce information.

Results

Enrollment and disposition of PHCs and GPs

This study was conducted in 3 Spanish health districts, which included 90 PHCs (58 in the Balearic Islands, 20 in the Tarragona-Reus district, and 12 in the Arnau de Vilanova-Llíria district) and 867 eligible GPs (Fig 1). Eighty-one PHCs (90%) agree to participate and were randomly allocated 41 to the intervention arm and 40 to the usual care arm. Overall, there were 749 GPs (86.4%), with 482 from the Balearic Islands, 177 from the Tarragona-Reus district, and 90 from the Arnau de Vilanova-Llíria district. A total of 372 GPs were in the intervention arm and 377 were in the usual care arm. During follow-up, 49 GPs were excluded due to changes in workplace. However, all GPs were included in the ITT analysis.

Fig 1

Flow chart of the study.

GP, general practitioner; ITT, intention-to-treat; PHC, primary health center.

Comparison of the 2 groups at baseline indicated that PHCs in the intervention arm and usual care arm were similar in registered patient population size and percentage of patients aged 65 years or more (Table 1). However, the average DDDs per 1,000 inhabitants per day was slightly higher in the usual care group, 74.6 in the intervention group and 76.9 in the control group, and more GPs in the intervention group were women, GPs with a doctorate degree, and trainers of GP residents.

Table 1

Distribution of PHCs and GPs’ characteristics at baseline.

	Intervention n/N (%)	Control n/N (%)
PHC characteristics	N = 40	N = 41
Baseline DDD per 1,000 inhabitants per day	74.6 ± 31.6	76.9 ± 33.1
Total number of patients listed
<12,500	10/40 (25.0)	11/41 (26.8)
12,500–25,000	21/40 (52.5)	21/41 (51.2)
>25,000	9/40 (22,5)	9/41 (22,0)
Proportion of patients > = 65 years
Mean ± SD	16.3 ± 3.1	16.1 ± 3.6
Urban centers	26/40 (65.0)	28/40 (70.0)
Training PHC	16/40 (40.0)	14/40 (35.0)
GP characteristic	N = 372	N = 377
Age (years)
Mean ± SD	51.7 ± 8.8	51.9 ± 8.6
Women	222/372 (59.7)	206/377 (54.6)
GP with 3 years specialty training	324/372 (87.1)	328/377 (87.0)
PhD (doctorate degree)	36/372 (9.7)	23/377 (6.1)
GP resident trainer	92/372 (24.7)	70/377 (18.6)
Years working as GP
Mean ± SD	21.9 ± 9.6	21.6 ± 9.7
Years working as GP in the actual workplace
Mean ± SD	11.5 ± 8.7	10.7 ± 9.5

DDD, defined daily dose; GP, general practitioner; PHC, primary health center; SD, standard deviation.

Primary outcome

After 12 months, the number of BZD prescriptions (DDDs per 1,000 inhabitants per day) in the intervention group decreased from 74.6 (95% confidence interval [CI]: 71.4, 77.8) to 71.0 (95% CI: 67.8, 74.1) (Table 2). The corresponding decrease in the control group was from 76.9 (95% CI: 73.6, 80.3) to 76.5 (95% CI: 73.2, 79.8). The between groups difference adjusted by baseline BZD prescription in total BZD prescriptions at 12 months (DDDs per 1,000 inhabitants per day) was −3.24 DDDs per 1,000 inhabitants per day (95% CI: −4.96, −1.53, p < 0.001). At the PHC level, the intracluster correlation coefficient of DDDs per 1,000 inhabitants per day of BZDs was 0.30. This was higher than expected from the sample size calculation, indicating substantial within-PHC clustering for BZD prescriptions.

Table 2

Comparison of GPs’ BZD prescriptions, percentage of all BZD long-term users (≥6 months) and percentage of BZD long-term users aged 65 or more.

Unadjusted, adjusted (DDD per 1,000 inhabitants per day and percentage of BZD long-term users at baseline) and estimated ITT results for the control and intervention groups at 12-month follow-up.

			Total
	Intervention Mean ± SD	Control Mean ± SD	Unadjusted mean difference (95% CI)	p-value	Adjusted mean difference (95% CI)	p-value	Estimated ITT mean difference (95% CI)	p-Value
Primary Outcome
DDD per 1,000 inhabitants per day at baseline	74.6 ± 31.6	76.9 ± 33.1
DDD per 1,000 inhabitants per day at 12 months	71.0 ± 29.9	76.5 ± 31.5	−5.38 (−9.94; −0.84)	0.020	−3.45 (−5.09; −1.82)	<0.001	−3.24 (−4.96; −1.53)	<0.001
Secondary outcome
Percentage of BZD long-term users at baseline	9.5 ± 3.9	9.8 ± 3.9
Percentage of BZD long-term users at 12 months	9.1 ± 3.6	9.8 ± 3.8	−0.68 (−1.23; −0.12)	0.016	−0.38 (−0.57; −0.18)	<0.001	−0.36 (−0.55; −0.16)	<0.001
Percentage of BZD long-term users >65 years old at baseline	24.9 ± 8.8	25.3 ± 8.2
Percentage of BZD long-term users >65 years old at 12 months	23.9 ± 7.7	25.2 ± 7.9	−1.28 (−2.45; −0.12)	0.031	−0.90 (−1.46; −0.34)	0.002	−0.87 (−1.44; −0.30)	0.003

BZD, benzodiacepine; CI, confidence interval; DDD, defined daily dose; GP, general practitioner; ITT, intention-to-treat; SD, standard deviation.

We performed subgroup analysis by health districts. The results indicated some differences in the effectiveness of the intervention, the interaction term from DDDs per 1,000 inhabitants per day and group was statistically significant among health districts (p = 0.002). We reported effect sizes by health district in Table 1 of the Supporting information (S1 Table).

The between groups differences adjusted by baseline BZD prescription in the total DDDs per 1,000 inhabitants per day in the Tarragona-Reus was −6.8 (95% CI −12.2 to −1.30, p = 0.023), in the Arnau de Vilanova llíria was −3.3 (95% CI −7.5 to 1.2, p = 0.156) and in the Balearic Islands (IbSalut) was −2.8 (95% CI −4.3 to −1.3, p < 0.001).

Secondary outcomes

The intervention and control groups had statistically significant differences in both secondary outcome measures (Table 2). In particular, the adjusted by baseline BZD prescription, difference of long-term BZD users among all patients was −0.36 percentage points (95% CI: −0.55, −0.16, p < 0.001) and the corresponding adjusted relative reduction was 3.6% (95% CI: 1.65, 8.86, p < .001). In patients aged 65 years or more, the adjusted difference was −0.87 percentage points (95% CI: −1.44, −0.30, p = 0.003) and the corresponding adjusted relative reduction was 3.5% (95% CI: 1.19, 5.71, p = 0.004).

Evaluating the implementation process

A total of 246/372 (66%) GPs from the intervention group answered this questionnaire. These GPs generally valued the multicomponent intervention as flexible and integrable into their practices. However, the 3 components of the intervention were valued differently. They reported the educational workshop was useful for clinical practice and the BZD prescription feedback was easy to interpret. They reported the tailored stepped dose reduction regimen for discontinuing BZDs was complex, and their lack of time and workload were significant barriers to implementation. The material included in the web page was considered helpful for the patients, but they reported some technical problems in accessing the website.

The fidelity of adherence to the intervention differed among the different components. Discussing benefits and risk with the patient before the first prescription had a median adherence score of 8 (interquartile range [IQR]: 7 to 9); setting limits for the duration from the start of the treatment had a median score of 8 (IQR: 7 to 9); tailoring stepped dose reduction to discontinuing BZD use in long-term users had a median score of 7 (IQR: 7 to 8); reviewing BZD prescription feedback had a median score of 7 (IQR: 6 to 8); downloading and using the patient information leaflet about BZDs, Z-drugs, and sleep hygiene had a median score of 7 (IQR: 7 to 8); and visiting the web page to reinforce information had a median score of 5 (IQR: 3.75 to 7).

Discussion

Our major finding is that a multicomponent intervention that targeted GPs was effective in reducing prescriptions of BZDs and in reducing the proportion of long-term BZD users at 12 months.

Comparison with existing literature

Our study indicated a small reduction in the total number of BZD prescriptions. A previous study found greater reductions in the prescriptions of BZDs to elderly patients (measured as DDD per 1,000 inhabitants per day) following continuous educational outreach visits to GPs. This previous study used a more intensive intervention that consisted of visits between GPs and pharmacists every 2 to 8 weeks [51]. However, these more intensive interventions are more difficult to implement in most PHCs. In contrast, other authors found that educational visits with GPs regarding BZD prescriptions were not effective in reducing the number of prescriptions [52,53]. Pimlott and colleagues and Holm also found that an educational intervention and feedback to GPs was not effective [54,55]. However, these studies have some methodological limitations.

Changing inappropriate health interventions is essential for minimizing patient harm, maximizing efficient use of resources, and improving population health. Prescribing BZDs in patients aged 65 years or more is considered potentially inappropriate and should be avoided because the decreased metabolism of long-acting agents and increased sensitivity and risk of cognitive impairment, delirium, falls, fractures, and motor vehicle crashes in older adults [14]. We found that our intervention led to a 3.5% reduction in the proportion of long-term BZDs users who were aged 65 years or more. There is evidence that certain interventions delivered by GPs were effective in reducing the number of long-term BZD users [27]. A previous study of a stepped dose reduction of BZDs with written instructions delivered by GPs to long-term users showed that the intervention was 3 times more effective than usual care [33]. Similar results (25% reduction of BZD use by long-term users) were reported in an intervention consisting of sending an educational booklet and a component for risk self-assessment of BZD use to long-term users [56]. These results differ from our finding that the intervention implemented by GPs provided only a modest reduction in the percentage of long-term BZD users. In more pragmatic clinical trials, GPs receive educational training, and all patients are evaluated, but not all patients receive the intervention. However, a small effect in a pragmatic trial that includes most GPs and all eligible patients is clinically relevant in absolute terms, because overuse of BZDs is widespread in Spain.

The overall reduction in BZD prescriptions that we observed in the intervention group was similar in magnitude to that found in an intervention that included education with audit and feedback [40,41].

Our results indicated that the intervention led to a reduction of 3.2 DDD per 1,000 inhabitants per day. From a health system perspective, this reduction over a 1-year period is clinically important. Some other health systems are steadily decreasing their prescriptions for BZDs. For example, Norway decreased BZD prescriptions from 65.7 to 49.7 DDD per 1,000 inhabitants per day from 2008 to 2018. The decline in Denmark over the same period was 42.4 to 25.9 DDD per 1,000 inhabitants per day. However, such decreases did not occur in Spain, Portugal, or France. In fact, over this same time, Spain had an increased consumption of BZDs from 75.5 to 89.3 DDD per 1,000 inhabitants per day [18].

Strengths and limitations

To our knowledge, this is the largest clinical trial to analyze the effectiveness of an educational intervention for GPs in a primary care setting that focused on reducing prescriptions for BZDs. We included all GPs of the participating PHCs. The study had a high rate of GP participation compared to similar studies [54], and very few GPs were lost to follow-up. We also found differences in effect size among the different health districts, indicating that the characteristics of the health district affected the efficacy of the intervention. For example, the Balearic Islands and Tarragona have local policies for GPs regarding BZD prescribing that include indicators and incentives to motivate GPs to reduce their BZD prescription. Another strength of our study is that it was performed in 3 health districts that have different characteristics. This suggests the results may be applicable in different clinical settings.

Our study had also some limitations. At baseline, the intervention group had more GP training healthcare centers and more GP trainers and physician doctors. This could have led to an overestimate of the effectiveness of the intervention because GP trainers and physician doctors are more likely to follow clinical guidelines and tend to write fewer BZD prescriptions [21]. Although we did not adjust for GP characteristics, we did adjust for baseline BZD prescriptions to minimize the potential of selection bias. GPs in the intervention PHCs may have shared information and strategies with GPs of the usual care PHCs. However, we randomized PHCs to avoid cross-contamination, and the PHCs were mostly in different villages or cities. Moreover, only GPs in the intervention arms had access to the additional material from the website and received individual BZD prescription feedback that was sent to their institutional e-mail addresses.

Implications for clinical practice and future research

A minimal intervention provided to GPs in a primary care setting, which consists of an educational training program and feedback regarding BZD prescriptions, may help to reduce the overall number of BZD prescriptions and reduce the number of long-term BZD users. From a public health perspective, because BZD use is very common in Spain, a small effect on long-term use by individuals could have a large impact in preventing adverse effects related to BZD use at the level of the population, such as falls, fractures, or cognitive impairment, especially in the elderly. In conclusion, our rigorous trial design and theory-based evaluation of the implementation process provided evidence of the effectiveness of a multicomponent strategy that targeted GPs to help reduce BZD prescriptions and the number of long-term users.

Consolidated Standards of Reporting Trials guidelines extension for cluster trials checklist.

(DOC)

Click here for additional data file.

Standards for Reporting Implementation Studies: the StaRI checklist.

(DOCX)

Click here for additional data file.

Template for Intervention Description and Replication checklist.

(DOCX)

Click here for additional data file.

Subgroup analysis of effectiveness of the intervention at 12-month follow-up by health districts.

(DOCX)

Click here for additional data file.

BENZORED protocol.

(DOC)

Click here for additional data file.

Predefined analysis plan and amendments.

(DOCX)

Click here for additional data file.

Implementation Fidelity Questionnaire.

Six items fidelity questionnaire.

(PDF)

Click here for additional data file.

25 Aug 2021

Dear Dr Leiva Rus,

Thank you for submitting your manuscript entitled "Effectiveness of a multicomponent intervention consisting of education and feedback on reducing benzodiazepine prescriptions by general practitioners: BENZORED hybrid type I cluster randomized controlled trial" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Aug 27 2021 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

21 Oct 2021

Dear Dr. Leiva,

Thank you very much for submitting your manuscript "Effectiveness of a multicomponent intervention consisting of education and feedback on reducing benzodiazepine prescriptions by general practitioners: BENZORED hybrid type I cluster randomized controlled trial" (PMEDICINE-D-21-03642R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to three independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Nov 11 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

From the Academic Editor:

1. Please include a description of the intervention, using the TIDIER checklist or equivalent as a guide.

2. Please include the CONSORT checklist for cluster randomized trials.

3. Please describe the intra-cluster correlation coefficient in the Results section and please clarify if this was lower than originally assumed for the sample size calculation.

Other editorial points:

4. Data availability statement: PLOS Medicine requires that the de-identified data underlying the specific results in a published article be made available, without restrictions on access, in a public repository or as Supporting Information at the time of article publication, provided it is legal and ethical to do so. Please see the policy at

http://journals.plos.org/plosmedicine/s/data-availability

and FAQs at

http://journals.plos.org/plosmedicine/s/data-availability#loc-faqs-for-data-policy

Thank you for your willingness to make your data available in a repository. However, the file “repository.sav” appears to be a proprietary format (SPSS dataset) that may reduce the accessibility. We request that authors provide data in file formats that are standard in their field and allow wide dissemination. If there are currently no standards in the field, authors should maximize the accessibility and reusability of the data by selecting a file format from which data can be efficiently extracted.

5. Throughout text: Please include line numbers with the revised version.

6. Abstract: Please report your abstract according to CONSORT for abstracts, following the PLOS Medicine abstract structure (Background, Methods and Findings, Conclusions) http://www.consort-statement.org/extensions?ContentWidgetId=562

7. Abstract: Please specify who was blinded to the intervention and control, and please mention any GPs or PHCs that were lost to follow up in each group.

8. Abstract: Methods and Findings: Please quantify the main results for decline in BZD prescriptions and proportions of long term users over age 65 including both 95% CIs and p values. Please present findings for long term users in general.

9. Abstract: Please report a summary of any adverse events in the Abstract (and please also present these findings in the Results of the manuscript).

10. Abstract: In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

11. Author summary: At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

12. Main text: For citations within the text, please include the reference number within square brackets, placed before the sentence punctuation. Where multiple references are indicated, please do not include spaces within the brackets.

13. Methods: Please include the study protocol/predefined analysis plan, with any amendments, as Supporting Information.

14. Methods: Please describe the nature of GP consent to participate.

15. Methods: Study population: Please mention the three regions are part of Spain.

16. Methods: Please complete the CONSORT checklist and ensure that all components of CONSORT are present in the manuscript.

When completing the checklist, please use section and paragraph numbers, rather than page numbers.

In addition to the CONSORT checklist, we suggest that your implementation research be reported according to Standards for Reporting Implementation Studies statement (STARI). The STARI guidelines can be found here: https://www.equator-network.org/reporting-guidelines/stari-statement/

17. Methods: Please describe the intervention (educational workshop, training on discontinuation/dose reduction intervention, automated monthly feedback, support web page) in greater detail.

18. Methods: Please provide more description of how the DDD per 1000 inhabitants per day was calculated.

19. Methods: Please provide more description of the prescriptions claims databases used to extract prescription records.

20. Methods: Please provide additional description of the subgroup analyses.

21. Methods: The trial registration lists the outcomes of “3. Feasibility, adoption and fidelity of the intervention will be measured by an "ad hoc" questionnaire to measure GP opinion”. (a) Can you please present those results as part of this manuscript, or indicate why that is not possible? (b) Can you please indicate when you plan to publish those results?

22. Methods: Please provide more detail on the implementation aspect of the trial. Please clearly define and differentiate and define your intervention and implementation strategy and outcomes for each. It should be clear how both aspects were tested for this hybrid trial. As described, the multicomponent interventions seem more like implementation strategies.

Please consider using the guidelines published by Proctor et al. to guide your reporting: Proctor EK, Powell BJ, McMillen JC: Implementation strategies: recommendations for specifying and reporting. Implement Sci 2013, 8:139

Please ensure you have included relevant and sensitive implementation outcomes to capture the impact of individual, team-level, and organizational factors that could determine the success/failure of the intervention. Example document on Implementation outcomes. Please use standard terminology for implementation outcomes so that there is consistency in reporting (e.g. adoption, fidelity, penetration, sustainability etc.).

Please clarify whether you used suitable implementation theoretical frameworks to guide the design of the study, analysis, and interpretation of your findings.

23. Results: The sample size listed in the submitted manuscript (749 GPs) and the target from the trial registry (508 GPs) differ. Please explain the discrepancy.

24. Results: Please provide 95% CIs and p values for all primary and secondary outcomes and subgroup analyses described in the text. For the subgroup analyses, please report the interaction between health district and outcome. When reporting adjusted analyses, please also mention the factors adjusted for.

25. Results: Please report any findings pertaining to adverse events for the study including numbers of specific events and whether or not adverse events are thought to be related to the intervention.

26. Discussion: Please present and organize the Discussion as follows: a short, clear summary of the article's findings; what the study adds to existing research and where and why the results may differ from previous research; strengths and limitations of the study; implications and next steps for research, clinical practice, and/or public policy; one-paragraph conclusion.

27. References: Please use the "Vancouver" style for reference formatting, and see our website for other reference guidelines https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

28. Tables 1 and 2: Please include these in the main text of the manuscript.

29. Table 2: Please note in the legend factors adjusted for in the analyses. Please also include results of unadjusted analyses.

30. Supporting Information Table 1: Please include p values for the subgroup analyses by health district, including the analysis for the interaction, for effectiveness of intervention within districts and comparisons between health districts.

Comments from the reviewers:

Reviewer #1: Review

In short: relevant topic, large well-conducted study, conventional intervention, concise presentation.

Abstract

I would recommend to describe the effect size as small, similar to the conclusion in the discussion section.

Introduction

It would be relevant to specify that the prescription rates for BZB in Spain are relatively high compared to other countries (information that I got from the discussion).

The choice of the implementation strategies (education, feedback, and information platform) should be founded in preparatory research and/or conceptual analysis of current practice. Currently it remains unclear why exactly these strategies were chosen, which is not consistent with current implementation science.

Deprescribing (and de-implementation generally) has received considerable interest in recent years. However, this manuscript does not relate to the implementation science literature on the topic.

The chosen approach is conventional for the implementation of recommended practices, although well-conducted at large scale. The authors argue that the prevention of first prescription of BZB has rarely been studied. This may strictly be true, but not prescribing has certainly be a topic in other medication.

Methods

The study is described as Type 1 trial, which implies a primary outcome that relates to clinical or health outcomes. The primary outcome is a medication prescription rate, which seems to me as aspect of physician behaviour, thus Type 3 hybrid effectiveness-implementation trial.

The stratification component of the randomization is not quite clear, because it seems to relate to geographical districts rather than practices.

I would recommend to describe the implementation strategies (education, feedback, info-platform) in more detail, using a reporting guideline such as TIDIER as appendix to the manuscript.

The number of outcomes in this manuscript is 3, which is rather. It seems that information on intervention fidelity is not available or not reported, but would be helpful to interpret the findings.

Results

A participation rate of 90% of invited practices is very high compared to many other studies in primary care. The authors may discuss this (in the discussion section) .

While the study is large and seems well-conducted, the manuscript does not present many results.

Discussion

Similar to the introduction section, the discussion could relate more to the implementation science literature, e.g. Cochrane review on audit and feedback and literature on de-implementation (stopping practices) in healthcare.

Michel Wensing

Reviewer #2: Alex McConnachie, Statistical Review

Vicens et al presents a report of BENZORED, a cluster randomised trial of an educational and feedback intervention for GPs to reduce benzodiazepine prescriptions. This review looks at the use of statistics in the paper.

Overall, I found this to be a nicely written paper. The statistical methods are generally good, with appropriate methods used to account for the cluster randomised design. I do have a few comments, which are generally minor, and should not affect the underlying message of the paper.

Much is made of the "Intention to Treat" analysis. This appears to have been interpreted as an analysis with multiple imputation to account for missing outcome data for some GPs. For me, this is not what ITT means. As the term implies, ITT is about analysing data from a randomised trial according to the intervention that was intended to be delivered. I.e., analysis according to the randomised allocation, regardless of whether the intervention was delivered as planned. In an RCT, all participants should be followed up, even those who withdraw from the intervention.

Loss to follow-up in randomised trials is common, but has nothing to do with the concept of ITT. In my opinion, the primary analysis should be limited to those GPs with follow-up data. This can still be labelled as "ITT", so long as the GPs are analysed according to their original randomised allocation. The amount of missing follow-up data should be reported, and the reader can judge the validity of the primary analysis in the context of this missing data.

As a sensitivity analysis, to assess the impact of missing data on the primary analysis result, multiple imputation is good method to use. However, I always feel that including the randomised group in the imputation process makes this circular; if the outcomes are imputed based on the intervention effect observed in those with complete data, this will tend to reinforce the complete case analysis. I would rather see the multiple imputation carried out without randomised group as a predictor, to see whether the primary analysis result is robust to the assumption of no intervention effect in those with missing outcomes.

That is probably my main point, and is one of emphasis more than anything. Looking at the tables, the complete case analysis appears to give similar results to what is currently described as "ITT". Given the relatively good follow-up achieved in the study, I would not expect multiple imputation without use of randomised group as a predictor to eliminate the intervention effect.

My others points are more minor.

The background reports a figure of 87.6 DDDs per 1000 patients, and a figure of 8.76% of patients taking a DDD each day. These are two different things, so if this is true, then it is quite a coincidence. E.g. 87.6 DDDs per 100 patients could represent about 175 in every 1000 patients (17.5%) all consuming half a DDD each day, or 4.4% of patients taking 2 DDDs every day.

I think the paper would benefit from a clearer statement about the unit of randomisation (PHC) and the unit of analysis (GP). It becomes clear as you go through the paper, but is not immediately obvious. For example, the sample size section talks about data aggregated at the cluster level (PHC), but then makes an assumption about the ICC, which implies analysis of data by GP, clustered within PHCs.

The primary analyses uses Poisson regression, but the intervention effects are reported as mean differences. How were the regression results converted to absolute differences?

There is mention of the trial being reported according to the CONSORT guideline for cluster trials, but I did not see a checklist. One feature that is missing from the results is a report of the ICCs observed. This would be useful for future researchers.

I do not think it is important to report Cohen's d. The outcomes are easily interpretable as they are.

Reviewer #3: Thank you for giving us the opportunity to comment on this interesting submission, and we congratulate the authors on the effort that has been put in.

We have a number of thoughts:

1) While we know how many centres were recruited we were not sure exactly how many GP's in each centre were taking part? Was this a consistently distributed number GPS or was there over-representation with more GPs at some centres? There was also a potentially important imbalance at baseline for DDDs.

2) We struggle to gauge the clinical significance of the findings in this paper. Although the final paragraph of the manuscript suggest that small effects could have high impact in avoiding false fractures or cognitive impairment, we believe that this trial does not provide any evidence on this. This trial only measures change in prescription, and not clinical outcomes.

3) Ultimately, it will be a health economic evaluation that is needed to judge if this sort of intervention is worth pursuing or not. Also, I am guessing that a subsequent manuscript will evaluate whether Primary Care are prepared to take on such an intervention.

Peer review: YK Loke and Navena Navaneetharaja

Any attachments provided with reviews can be seen via the following link:

[LINK]

7 Dec 2021

Submitted filename: Carta al editor con cambios ACEPTADOS CV.docx

Click here for additional data file.

1 Feb 2022

Dear Dr. Leiva,

Thank you very much for submitting your manuscript "Effectiveness of a multicomponent intervention consisting of education and feedback on reducing benzodiazepine prescriptions by general practitioners: BENZORED hybrid type I cluster randomized controlled trial" (PMEDICINE-D-21-03642R2) for consideration at PLOS Medicine.

Your revised paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to two of the original reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of the remaining points of Reviewer 2, we cannot accept the manuscript for publication in the journal in its current form, but we would like to consider another revised version that addresses the reviewer's and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we may seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Feb 22 2022 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1. Data availability statement: Thank you for providing the link to access the dataset. Please revise the statement to “All data are available from the Zenodo repository (https://doi.org/10.5281/zenodo.5607352).” or similar. There may be a typo in the DOI provided, please check if this should be: https://doi.org/10.5281/zenodo.5607352

Please also check that no identifying information (GP, PHC name) are included in the file.

2. Abstract: Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions). Please incorporate the sentence describing the study objective as the final sentence Background.

3. Abstract: Methods and Findings: Please mention that analyses were intention to treat.

4. Abstract: Methods and Findings: Line 121-122: Please mention the characteristics for which the two groups were not apparently balanced.

5. Abstract: Methods and Findings: In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

6. Throughout: Please place the in-text citations within square brackets before the sentence punctuation, for example [1].

7. Throughout: Please carefully check the text for typos/grammatical errors.

8. Methods: Line 239: Please include a copy of the original study protocol as a supporting information file, and please refer to it here (for example, S1 Protocol).

9. Methods: Line 244: Please include the information regarding participant (PHC) consent and include the information that the ethical review boards waived the requirement for GP consent.

10. Methods: Line 342: Please reference the analysis plan in the Supporting Information files here.

11. Methods: Line 344-345: Please revise the CONSORT statement to: "These results are reported as per the Consolidated Standards of Reporting Trials (CONSORT) guideline extension for cluster randomized trials (S1 CONSORT Checklist)."

12. Methods: Line 362: Please provide a copy of the questionnaire as a supporting information file.

13. Results: Line 401: Please summarize the key findings from the additional analyses of effects by health district.

14. Discussion: Line 455-456: Please provide references and clarify this statement: “Prescribing BZDs in patients aged 65 years or more is considered inappropriate and should be avoided.”

15. Checklists: Thank you for including the 3 reporting checklists. Please revise the checklists, using section and paragraph numbers to refer to locations within the text (e.g. Methods, paragraph 1). Please do not refer to page numbers.

16. Supporting information Table 1: Please provide a legend for this table.

17. Analysis Plan: Please note at what point during the study the amendment to the analysis plan was made.

Comments from the reviewers:

Reviewer #2: Alex McConnachie, Statistical Review

I thank Vicens and colleagues for their consideration of my original points.

On the matter of the meaning of ITT, I think we can agree to disagree. I accept that I may be in the minority in my opinion, even if I think I am right!

However, this got me thinking about the analysis. By imputing the missing outcomes for GPs who moved out of the PHC area, the analysis is estimating the effect of the intervention under the scenario that all GPs remained within their original PHC. I am not sure this is realistic.

Also, in the analysis, it appears that only those who expressed an interest in the study prior to randomisation are included. Therefore, the analysis is estimating the impact of the intervention in a subgroup of GPs, rather than the overall impact in the population of all GPs. If implemented, some GPs might decline the intervention, and these GPs are not being included in the analysis.

Since randomisation took place at the PHC level, the fairest analysis, under ITT, might be to include all GPs within each PHC, regardless of whether they received the intervention, since the randomisation resulted in the intervention being available in some PHCs but not others. Since outcomes were collected from electronic records, would it have been possible to collect complete outcome data for all GPs in each PHC at baseline and at follow-up, removing the need for imputation of missing data?

I suppose one problem with this approach would be that for those GPs who move into an area during the study, there would be no baseline value for the outcome variables, making the analysis at GP level more difficult.

All this being said, I think that what the authors have done is very good. Both the complete case and multiple imputation results are presented, and are similar. Given that GPs who were unwilling to take part were excluded prior to randomisation, what are the implications in terms of the population-level effect of the intervention? Did these GPs, in intervention PHCs, have access to the intervention? If so, did they receive it, or did they tend to decline?

The method of estimating the marginal mean difference in DDDs from a Poisson model is fine, but should be mentioned in the statistical methods section.

Finally, I still have a problem with the text "The Spanish Medicine Agency (AEMPS) reported an average of 87.6 DDDs per 1000 inhabitants-per-day during 2018. Thus, approximately 9% of the population of Spain is receiving on average daily equivalent of 10mg of diazepam." I do not believe that it is possible to deduce the percentage of the population who are receiving a medication, based on the average DDDs per 1000 participants. Approximately 90 DDDs per 1000 population could represent 9% of the population receiving an average of 1 DDD each, or it could represent 18% of the population receiving an average of 0.5 DDDs each. There is no way to tell.

Reviewer #3: Thank you for revising the manuscript. I don't have any further points to raise.

Any attachments provided with reviews can be seen via the following link:

[LINK]

17 Feb 2022

Submitted filename: respuesta plos medicine final 14 feb.docx

Click here for additional data file.

16 Mar 2022

Dear Dr. Leiva,

Thank you very much for re-submitting your manuscript "Effectiveness of a multicomponent intervention consisting of education and feedback on reducing benzodiazepine prescriptions by general practitioners: BENZORED hybrid type I cluster randomized controlled trial" (PMEDICINE-D-21-03642R3) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by one of the reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Mar 23 2022 11:59PM.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

From the academic editor:

-Abstract: Methods and Findings Line 125: As the primary outcome is reported as absolute difference, please similarly report the secondary outcome as absolute rather than relative difference: “The intervention group also had a smaller number of adjusted relative long-term users overall (-3,6% (95%CI: 1.85,8.88, p>0,001) and a smaller number of adjusted relative long-term users over age 65 years (.- 3.5% (95%CI: 1.19, 5.71 p=0,004)”

-Methods: Study population: Line 248: I think it might be helpful to insert a brief few sentences on how PHC is organised in Spain (regions, districts, PHCs, GPs).

-Methods: Line 305: The second secondary outcome measure is perhaps better considered a sub-group analysis (people over 65 years who are long term BZD users). Was there any heterogeneity of effect between those >65 years and those <=65 years? It looks like there was also a pre-specified sub-group analysis by district (e.g. described at line 339-341 and reported at line 406). It would be good to have a sub-heading on pre-specified sub-group analyses and ensure all are reported in the results.

-Results, Table 2: Please include a footnote to describe what covariates were included in the adjusted model. It is difficult to work out from the main manuscript text.

-Results: I am a little surprised that with the large actual ICC (0.3) relative to the sample size calculations (0.05) that the study was adequately powered to show a difference – the effect size is smaller and the number of general practices is only a little higher compared to the original calculations. But it is hard to calculate the actual study power based on the data providers (e.g the patient cluster sizes at the PHC level are not given). Could this be checked?

-Results: Also related to the point above it would be helpful to understand a little more what is happening at the PHC level to cause such a high level of clustering? This is related to my comment above on PHC structure in Spain. Most of the discussion focuses on the level of GPs and while I can understand within general practice clustering I am a little unclear what is driving PHC level clustering.

Other editorial points:

1. Title: Please revise to: “Evaluation of a multicomponent intervention consisting of education and feedback to reduce benzodiazepine prescriptions by general practitioners: The BENZORED hybrid type I cluster randomized controlled trial” and please update this in the manuscript submission system as well as the text of the manuscript.

2. Data availability statement: We suggest revising to: “The de-identified data underlying the study’s findings are publicly available in the Zenodo repository: https://doi.org/10.5281/zenodo.5970482”

3. Throughout: Please edit for minor typos and grammatical errors throughout.

4. Abstract: Line 112-113: “GPs were not blinded to the allocation.” Please also mention who was blinded to the allocation.

5. Abstract: Line 114: Please clarify if “tapering” rather than “withdrawal” would be preferable here.

6. Abstract: Line 115: We suggest revising to: “...with monthly BZD prescription feedback and access to a support webpage.”

7. Abstract: Line 116-117: Please clarify the number of GPs (and PHCs) lost to follow up in each group.

8. Abstract: Line 119: Please define “long term users” in this sentence.

9. Abstract: Line 131: Please provide a definition of “physician doctor” as this may not be a universally understood term.

10. Author Summary: Line 165-167: We suggest revising to: “... successfully reduced the proportion of long-term users, which could potentially reduce adverse outcomes related to long-term use.” or similar.

11. Line 172: Please change “Background” to “Introduction” as the heading for this section.

12. In-text citations: Throughout, please include a space between the reference bracket and the preceding word, for example [1]. Please also check that all reference brackets are placed before the sentence punctuation, where applicable.

13. Methods: Please clarify the nature of the informed consent provided by GPs at PHCs included in the study.“Only PHCs in which at least two-thirds of the GPs agreed and provided written informed consent to participate were included. Informed consent was waived by the Balearic Islands and the l'IDIAP Jordi Gol Ethical Committee of Clinical Research because the analysis used anonymous administrative data. The requirement for informed consent was not waived by the Valencian Primary Care Ethics Committee, and therefore, from this region, only PHCs from which all GPs agreed and provided written informed consent were included.” or similar, if this is accurate.

14. Methods: Line 294-295: “This web site included videos, descriptions of practical cases, and an information leaflet for patients about BZDs, Z-drugs, and sleep hygiene (http://benzored.es).” Please note that the link provided does not seem to be functional, please include an accessible link.

15. Methods: Line 300-301: “Total number of dispensed DDD from the N05BA, N05CD, and N05CF groups of the Anatomical Therapeutic Chemical (ATC)Classification System code…” please clarify what prescription drugs are included in these codes.

16. Results: Line 378-380: We suggest clarifying to: “This study was conducted in three Spanish health districts, which included 90 PHCs (58 in the Balearic Islands, 20 in the Tarragona-Reus district, and 12 in the Arnau de Vilanova-Llíria district) and 867 eligible GPs (Figure 1).”

17. Results: Line 391-392: Please report the baseline DDD per 1000 inhabitants per day for the intervention and control groups in the text.

18. Results: Line 418-425: Please mention in the text what factors were adjusted for in these analyses.

19. Results: Line 417: For secondary outcomes, it would be helpful to report the absolute differences, as mentioned by the Academic Editor (please see the comment on the Abstract).

20. Results: Line 427-436: Please present data on these results, e.g. percentage of respondents who reported on each item of the questionnaire. Please consider adding a table with complete results for these outcomes.

21. Discussion: Line 459-462: “In contrast, other authors found that educational visits with GPs regarding BZD prescriptions were not effective in reducing the number of prescriptions[52,53]. Pimlott et al. and Holm also found that an educational intervention and feedback to GPs was not effective[54,55].” Some additional discussion on why these educational/feedback interventions may have been less effective would be helpful to provide context.

22. Discussion: Line 468-470: Please revise if it might be more accurate to interpret this as: “We found that our intervention led to a 3.5% reduction in the proportion of long-term BZDs users who were aged 65 years or more.”

23. Discussion: Line 472-473: We suggest revising to “A previous study of a stepped-dose reduction of BZDs, with written instructions delivered by GPs to long-term users…” or similar.

24. Discussion: Line 477-479: Please clarify here and throughout, as the secondary analysis revealed a reduction in proportion of long-term BZD users, rather than a reduction in BZD use by long term BZD users. We suggest changing to: “These results differ from our finding that the intervention implemented by GPs provided only a modest reduction in the percentage of long-term BZD users.”

25. Discussion: Line 503-505: Please include slightly more context here, briefly discussing how these local policy differences could have influenced your results. Please also include relevant citations: “For example, the Balearic Islands and Tarragona have local policies regarding BZD prescribing that include indicators and incentives.”

26. Discussion: Line 506-507: We suggest revising to: “This suggests the results may be applicable in different clinical settings.”

27. Line 538: Please remove the “Author contributions” section from the main manuscript text, and ensure all information is completely entered into the relevant sections of the manuscript submission system.

28. Line 553: Please remove the “Funding statement” section from the main manuscript text, and ensure all information is completely entered into the relevant sections of the manuscript submission system.

29. Line 564: Please remove the “Competing interests statement” section from the main manuscript text, and ensure all information is completely entered into the relevant sections of the manuscript submission system.

30. Line 566: Please remove the Ethics and Dissemination section from the end of the manuscript, and ensure the information is included in the appropriate location of the Methods section (this information could be added to the “Study Design” sub-section, or as a separate sub-section). The sentence “All data are available upon reasonable request.” can be removed as the information on accessing the publicly available data is included in the Data Availability statement of the manuscript submission system.

31. References: Please check the formatting of each reference in the list. Please use the "Vancouver" style for reference formatting, and see our website for other reference guidelines https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

For example:

Reference 4: The journal title should be “BMJ”

Reference 6: The journal title is missing.

Reference 9: This reference is incomplete.

Reference 27: This reference is incomplete.

Reference 30: Please check the journal title.

32. Table 1: Please define “GP Specialty Training” as this was not mentioned in the Methods (please describe assessment of GP characteristics in the Methods).

33. Table 2: In the legend, it would be helpful to define that the table reports on the percentage of all patients who are long term BZD users (and how this is defined), and the percentage of long-term BZD users who are older than 65 years. Please mention the factors adjusted for in the analyses in the legend (i.e. please briefly describe the unadjusted, adjusted, and estimated ITT analyses).

34. Supporting Information Table 1: Please remove the duplicate file (the file without p values).

35. TIDieR, StaRI, and CONSORT Checklists: Please remove the older versions of these files.

36. S1 Implementation Fidelity Questionnaire: Please include a copy of the questionnaire as a supporting information file.

37. Study protocol: Thank you for including the original copy of the study protocol (“renamed_11888”). Please include a version in English, and we suggest renaming the file (e.g. S1 Protocol, or similar).

Comments from Reviewers:

Reviewer #2: Alex McConnachie, Statistical review

I thank the authors once again for their responses. The section about DDDs and percentage of users has been removed, which was my only real objection. Whilst I feel there are other ways that the analysis could have been done, I am happy with the analyses as presented, and I have no further comments to make.

Any attachments provided with reviews can be seen via the following link:

[LINK]

1 Apr 2022

Submitted filename: Carta para Plos ultima revision 22 Marzo sin comentarios.docx

Click here for additional data file.

7 Apr 2022

Dear Dr Leiva,

On behalf of my colleagues and the Academic Editor, Dr Peiris, I am pleased to inform you that we have agreed to publish your manuscript "Evaluation of a multicomponent intervention consisting of education and feedback to reduce benzodiazepine prescriptions by general practitioners: The BENZORED hybrid type 1 cluster randomized controlled trial." (PMEDICINE-D-21-03642R4) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

Prior to final acceptance, please address some minor points:

At line 123, please adapt the text to "forty"; and use "follow-up" here and at any other instances;

At line 162, please spell out "DDD" at first use in the Author Summary;

Please remove the academic editor's name from reference 5;

Please remove the duplicated text from reference 9; and

Please correct reference 34, removing the Orcid and the author name following the title.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Richard Turner PhD, for Caitlin Moyer, Ph.D.

Senior Editor, PLOS Medicine

rturner@plos.org