Downloaded from http://jnnp.bmj.com/ on March 10, 2015 - Published by group.bmj.com
Epilepsy
RESEARCH PAPER
Comprehensive educational plan for patients with epilepsy and comorbidity (EDU-COM): a pragmatic randomised trial Simone Beretta,1 Ettore Beghi,2 Paolo Messina,2 Francesca Gerardi,2 Francesca Pescini,3 Andrea La Licata,3 Luigi Specchio,4 Mariangela Ferrara,4 Maria Paola Canevini,5 Katherine Turner,5 Francesca La Briola,5 Silvana Franceschetti,6 Simona Binelli,6 Isabella Giglioli,6 Carlo Andrea Galimberti,7 Cinzia Fattore,8 Gaetano Zaccara,9 Luciana Tramacere,9 Francesco Sasanelli,10 Marta Pirovano,10 Carlo Ferrarese1 ▸ Additional material is published online only. To view please visit the journal online (http://dx.doi.org/10.1136/ jnnp-2013-306553). For numbered affiliations see end of article. Correspondence to Dr Simone Beretta, Department of Neurology, San Gerardo Hospital, Via Pergolesi 33, Monza 20900, Italy; [email protected] Received 14 August 2013 Revised 28 October 2013 Accepted 25 November 2013 Published Online First 8 January 2014
To cite: Beretta S, Beghi E, Messina P, et al. J Neurol Neurosurg Psychiatry 2014;85:887–892.
ABSTRACT Background The impact of educational strategies in the management of adverse treatment effects and drug interactions in adult patients with epilepsy with comorbidities remains undetermined. Objective The EDU-COM study is a randomised, pragmatic trial investigating the effect of a patienttailored educational plan in patients with epilepsy with comorbidity. Methods 174 adult patients with epilepsy with chronic comorbidities, multiple-drug therapy and reporting at least one adverse treatment effect and/or drug interaction at study entry were randomly assigned to the educational plan or usual care. The primary endpoint was the number of patients becoming free from adverse treatment events and/or drug interactions after a 6month follow-up. The number of adverse treatment events and drug interactions, health-related quality of life (HRQOL) summary score changes and the monetary costs of medical contacts and drugs were assessed as secondary outcomes. Results The primary endpoint was met by 44.0% of patients receiving the educational plan versus 28.9% of those on usual care ( p=0.0399). The control group reported a significantly higher risk not to meet successfully the primary endpoint at the end of the study: OR (95% CI) of 2.29 (1.03 to 5.09). A separate analysis on drug adverse effects and drug interactions showed that the latter were more sensitive to the effect of educational treatment. Quality of life and costs were not significantly different in the two groups. Conclusions A patient-tailored educational strategy is effective in reducing drug-related problems ( particularly drug interactions) in epilepsy patients with chronic comorbidities, without adding significant monetary costs. Registered at ClinicalTrials.gov, identifier NCT01804322, (http://www.clinicaltrials.gov).
drug interactions in variable proportions.3–5 The incidence of AEs tends to increase with the number of drugs and drug combinations.6–8 Complex therapeutic schedules theoretically increase the risk of drug-related problems, and may affect patients’ quality of life and increase medical contacts, hospital admissions and related costs.9 10 In this context, personalising drug information may reduce the number and severity of drug-related problems and their complications.11 Patient– physician interactions and educational strategies may improve clinical outcomes and quality of life in other chronic disorders like asthma and diabetes.12–15 However, patient information was not consistently found to be associated with a better health outcome and to cost saving.16–18 The management of the patient with epilepsy and comorbidity is particularly complex, because the burden of possible AEs and drug interactions may be challenging for the patient and the physician. Combinations of AEDs and other compounds are present in several patients, tend to increase with age and are also expected to increase the risk of AEs, the number of medical contacts and health-related costs. In this regard, an educational plan may help the patient and the caring physician by alerting them on the possible adverse effects of drug combinations. The objective of the EDU-COM study was to verify whether a standardised educational plan administered as a patient-tailored counselling session, is followed by an increased proportion of adult patients with epilepsy and comorbidity becoming free from clinically relevant adverse treatment effects and/or drug interactions, compared to usual clinical practice. Secondary analysis included the effects of the two approaches on health-related quality of life (HRQOL) and direct medical costs.
INTRODUCTION
MATERIALS AND METHODS Study design
Epilepsy is a chronic condition requiring long-term pharmacological treatment. In adult patients, epilepsy is associated with other clinical conditions in a significant proportion of cases.1 2 Antiepileptic drugs (AED) are accompanied by adverse events (AE) and
This is a randomised, controlled, single-blind, pragmatic non-pharmacological trial. Consecutive patients were randomised from December 2009 to December 2011 to receive a standardised educational plan (experimental arm) or to usual care, that is, the
Beretta S, et al. J Neurol Neurosurg Psychiatry 2014;85:887–892. doi:10.1136/jnnp-2013-306553
887
Downloaded from http://jnnp.bmj.com/ on March 10, 2015 - Published by group.bmj.com
Epilepsy management of the AEs/drug interactions as usually done in clinical practice according to each unit’s modalities (control arm). The study protocol was approved by the ethics committee of San Gerardo University Hospital, and by the ethics committees of each of the eight other participating epilepsy centres. Written informed consent was obtained from each participant. Patients’ follow-up ended at June 2012. The standardised educational plan consisted of the discussion (1 h counselling) with the patient about each of the following points in this order: the cause and nature of any AEs and/or drug interaction; the tolerability profile of each drug present in the schedule, illustrated as a simple list including the commonest AEs presented in decreasing order of frequency; the clinical manifestations (if any) associated with the current drug interaction(s); any contraindication to the use of over-the-counter drugs potentially interfering with the current treatment schedule; the reasons for and the potential benefits of the suggested treatment change; an encouragement to withdraw any potentially interfering or contraindicated drug. The educational plan was administered at admission and in the same form after 1 month, as a reminder. The educational plan used in the EDU-COM study was conceived to be at the same time standardised (following a predefined sequence, as shown in online supplementary figure S1) and patient-tailored (adaptable to the individual patient profile). A centralised computer-generated allocation sequence was used for the randomisation; no blocking or stratification procedures were adopted. Inclusion criteria were as follows: age 18 years or older; established diagnosis of epilepsy; presence of one or more concurrent clinical conditions requiring chronic treatment (ie, daily treatment lasting 1 month or longer); clinically relevant AE attributable to the present treatment(s) and/or clinically relevant drug interaction; possible modification of the treatment schedule to eliminate AE(s) and/or risky drug interactions. For the purposes of this study, a clinically relevant AE was any symptom and/or clinical and/or laboratory sign related to the current drug regimen requiring active monitoring (according to the caring physician’s judgment); a clinically relevant drug interaction is an interaction requiring active monitoring (with clinical visits and/ or biochemical analyses according to the caring physician’s judgment) and/or treatment change. Exclusion criteria were as follows: non-modifiable treatment schedule; patient unable to understand or comply with an educational plan (the application of this exclusion criterion was left to the caring physician’s judgment); patient unable or unwilling to release a written informed consent. Drug-related AEs and drug interaction could be checked using a comprehensive internet-based drug database ((http://www. codifa.it), Italian language, related to the Istituto Superiore di Sanità).
Outcome assessment Outcome assessment was performed in person at baseline (M0), at 1 month (M1) and 6 months (M6) after admission. For the purposes of the study, treating and evaluating physicians were represented by different persons. Neither the patients nor the caring physicians involved in administering interventions were blind to the assigned arm. On the contrary, physicians assessing outcomes were blind to the assigned arm. At admission (M0), all patients were interviewed for the collection of: demographic data; clinical factors (date of diagnosis, seizure frequency in the preceding 6 months and seizure type(s), epilepsy syndrome, comorbidities); treatment regimen; number, type and severity of AEs. At each follow-up visit (M1 and M6 888
visit), the patients were inquired about the outcome of the AE, the occurrence of any new symptom, seizure frequency, any change in the treatment schedule, any medical contact or hospital admission. HRQOL was assessed at admission and at each visit (M0, M1, M6) using the Italian validated translation of the QOLIE-31 Inventory.19 The data for adverse drug events were collected during each visit as follows: generic description, followed by categorisation using the MedDRA SOC (2008) terminology; severity defined as: mild, moderate, severe, potentially lethal; relation to a particular drug present in the treatment schedule (defined by the evaluating physician using as a reference http://www.codifa.it). The number of drug interactions was defined by a centralised independent epilepsy specialist (SB) at the end of study, using the treatment forms included in the Case Report Form and taking as a reference a comprehensive internet-based drug database (http://www.codifa.it). The primary outcome was defined as the number of patients becoming free from AEs and/or drug interactions at end of study (M6). More specifically, to meet the primary outcome, patients admitted with only AEs or drug interactions had to become free of any event (whether AE or interactions) at M6, while those with AEs and drug interactions had to become free to at least one of the two components. Thus, the term ‘success’ was used for patients meeting the primary outcome during the study. The secondary outcomes were defined as follows: the number of patients free from drug interactions/adverse drug events at the intermediate visit (M1); the number of adverse drug events and drug interactions during the study period; the QOLIE-31 score changes from M0 to M6; the cost of drugs, medical contacts and hospital admissions during the study period.
Data quality assurance and control of bias Monitoring was carried out by a dedicated epilepsy specialist (SB), with regular on-site monitoring visits and verification of: protocol adherence, screening failures, accurate capture of clinical data, protection of safety of participants. To control for the principal sources of bias, the following strategies were adopted: consecutive patients were screened to minimise selection bias; the caring physicians (neurologists, epilepsy specialists) were involved as treating physicians to ensure correct application of the inclusion criteria, and choose the best treatment change; the physicians assessing outcome (neurologists or neurology residents) were different from the treating physicians.
Determination of sample size and study power According to the results reported from previous studies,3 we estimated that a change in treatment schedules under usual care was followed by the disappearance of AEs in 40% of patients. To be considered effective, an educational plan was expected to increase this proportion to 60% (20% absolute difference). For this difference to become statistically significant with an 80% power (1-β) and a 5% level of significance ( p=0.05), a total of 100 patients per treatment arm were planned to be enrolled. Recruitment was terminated after the enrolment of 187 patients (174 with confirmed eligibility criteria), leading to a reduction of the study power to 76%.
Statistical analysis The data were analysed using the Statistical Analysis System (SAS Institute, Cary, North Carolina, USA.) package for PC (V.9.2). The statistical analysis plan included descriptive statistics to compare the baseline characteristics of the two populations
Beretta S, et al. J Neurol Neurosurg Psychiatry 2014;85:887–892. doi:10.1136/jnnp-2013-306553
Downloaded from http://jnnp.bmj.com/ on March 10, 2015 - Published by group.bmj.com
Epilepsy and the distribution of the treatment schedules, the number and type of AEs, along with the cost items (medical contacts, hospitalisations, drugs) and the HRQOL scores. Univariate analysis was carried out using the χ2 test, the Wilcoxon–Mann Whitney test and the Friedmann test as appropriate. All analyses were performed in the intent-to-treat population (unless otherwise specified). Missing data on the primary endpoint were considered as failures. Missing data on the HRQOL scores were imputed using the Last Observation Carried Forward (LOCF) approach. Multivariable logistic regression was assessed on the primary endpoint using a ‘forward’ selection approach. Results are reported as ORs with 95% CIs. Comparisons between the two arms in term of number of AEs and drug interactions were assessed using Poisson regression analysis. HRQOL score was also analysed using repeated measures ANOVA with ‘treatment’ as the between-groups level (experimental and control), ‘time’ as the within-group level (baseline and M6) and the interaction term ‘treatment×time’. The ANalysis Of VAriance (ANOVA) repeated measures model was carried out using the PROC MIXED SAS procedure and choosing an ‘unstructured’ correlation matrix. All tests are two-tailed with significance set to 5%. We decided not to adjust for multiple comparisons as we used a single primary endpoint and all other secondary endpoints are considered subsidiary and, as such, results concerning these outcomes can only have an exploratory rather than a confirmatory interpretation.20
RESULTS Study population A total of 187 patients were screened, and 174 (93%) were confirmed eligible and randomly assigned to usual care (n=83) or standardised educational plan (n=91). Of these, 172 (99%) completed the protocol (see figure 1). Table 1 describes the intention-to-treat (ITT) population in terms of demographic characteristics, education, epilepsy (syndrome, disease duration and seizure frequency), number of comorbidities and drugs. The two treatment arms were well balanced for all these variables. Most patients had a focal
epilepsy syndrome less than 1 seizure/month, a normal neurological examination and a complex therapeutic regimen (3 or more drugs).
Efficacy of education plan: primary end point Table 2 reports the results obtained on the primary endpoint. The analysis on the ITT dataset was the primary analysis. The proportion of patients becoming free from AEs and/or drug interactions at the end of the study was significantly higher in the experimental arm ( p=0.0399). The results were confirmed by the adjusted multivariable logistic model that showed that the control group reported a significantly higher risk not to meet successfully the primary endpoint at the end of the study: OR (95% CI) 2.29 (1.03 to 5.09) ( p=0.0420).
Efficacy of education plan: secondary end points As shown in table 3, the number of patients becoming free from AEs and/or interactions at 1-month visit showed no difference between the two arms (p=0.1897). The number of AEs showed a non-significant difference and a substantial reduction over time (Friedmann test: p
Epilepsy
RESEARCH PAPER
Comprehensive educational plan for patients with epilepsy and comorbidity (EDU-COM): a pragmatic randomised trial Simone Beretta,1 Ettore Beghi,2 Paolo Messina,2 Francesca Gerardi,2 Francesca Pescini,3 Andrea La Licata,3 Luigi Specchio,4 Mariangela Ferrara,4 Maria Paola Canevini,5 Katherine Turner,5 Francesca La Briola,5 Silvana Franceschetti,6 Simona Binelli,6 Isabella Giglioli,6 Carlo Andrea Galimberti,7 Cinzia Fattore,8 Gaetano Zaccara,9 Luciana Tramacere,9 Francesco Sasanelli,10 Marta Pirovano,10 Carlo Ferrarese1 ▸ Additional material is published online only. To view please visit the journal online (http://dx.doi.org/10.1136/ jnnp-2013-306553). For numbered affiliations see end of article. Correspondence to Dr Simone Beretta, Department of Neurology, San Gerardo Hospital, Via Pergolesi 33, Monza 20900, Italy; [email protected] Received 14 August 2013 Revised 28 October 2013 Accepted 25 November 2013 Published Online First 8 January 2014
To cite: Beretta S, Beghi E, Messina P, et al. J Neurol Neurosurg Psychiatry 2014;85:887–892.
ABSTRACT Background The impact of educational strategies in the management of adverse treatment effects and drug interactions in adult patients with epilepsy with comorbidities remains undetermined. Objective The EDU-COM study is a randomised, pragmatic trial investigating the effect of a patienttailored educational plan in patients with epilepsy with comorbidity. Methods 174 adult patients with epilepsy with chronic comorbidities, multiple-drug therapy and reporting at least one adverse treatment effect and/or drug interaction at study entry were randomly assigned to the educational plan or usual care. The primary endpoint was the number of patients becoming free from adverse treatment events and/or drug interactions after a 6month follow-up. The number of adverse treatment events and drug interactions, health-related quality of life (HRQOL) summary score changes and the monetary costs of medical contacts and drugs were assessed as secondary outcomes. Results The primary endpoint was met by 44.0% of patients receiving the educational plan versus 28.9% of those on usual care ( p=0.0399). The control group reported a significantly higher risk not to meet successfully the primary endpoint at the end of the study: OR (95% CI) of 2.29 (1.03 to 5.09). A separate analysis on drug adverse effects and drug interactions showed that the latter were more sensitive to the effect of educational treatment. Quality of life and costs were not significantly different in the two groups. Conclusions A patient-tailored educational strategy is effective in reducing drug-related problems ( particularly drug interactions) in epilepsy patients with chronic comorbidities, without adding significant monetary costs. Registered at ClinicalTrials.gov, identifier NCT01804322, (http://www.clinicaltrials.gov).
drug interactions in variable proportions.3–5 The incidence of AEs tends to increase with the number of drugs and drug combinations.6–8 Complex therapeutic schedules theoretically increase the risk of drug-related problems, and may affect patients’ quality of life and increase medical contacts, hospital admissions and related costs.9 10 In this context, personalising drug information may reduce the number and severity of drug-related problems and their complications.11 Patient– physician interactions and educational strategies may improve clinical outcomes and quality of life in other chronic disorders like asthma and diabetes.12–15 However, patient information was not consistently found to be associated with a better health outcome and to cost saving.16–18 The management of the patient with epilepsy and comorbidity is particularly complex, because the burden of possible AEs and drug interactions may be challenging for the patient and the physician. Combinations of AEDs and other compounds are present in several patients, tend to increase with age and are also expected to increase the risk of AEs, the number of medical contacts and health-related costs. In this regard, an educational plan may help the patient and the caring physician by alerting them on the possible adverse effects of drug combinations. The objective of the EDU-COM study was to verify whether a standardised educational plan administered as a patient-tailored counselling session, is followed by an increased proportion of adult patients with epilepsy and comorbidity becoming free from clinically relevant adverse treatment effects and/or drug interactions, compared to usual clinical practice. Secondary analysis included the effects of the two approaches on health-related quality of life (HRQOL) and direct medical costs.
INTRODUCTION
MATERIALS AND METHODS Study design
Epilepsy is a chronic condition requiring long-term pharmacological treatment. In adult patients, epilepsy is associated with other clinical conditions in a significant proportion of cases.1 2 Antiepileptic drugs (AED) are accompanied by adverse events (AE) and
This is a randomised, controlled, single-blind, pragmatic non-pharmacological trial. Consecutive patients were randomised from December 2009 to December 2011 to receive a standardised educational plan (experimental arm) or to usual care, that is, the
Beretta S, et al. J Neurol Neurosurg Psychiatry 2014;85:887–892. doi:10.1136/jnnp-2013-306553
887
Downloaded from http://jnnp.bmj.com/ on March 10, 2015 - Published by group.bmj.com
Epilepsy management of the AEs/drug interactions as usually done in clinical practice according to each unit’s modalities (control arm). The study protocol was approved by the ethics committee of San Gerardo University Hospital, and by the ethics committees of each of the eight other participating epilepsy centres. Written informed consent was obtained from each participant. Patients’ follow-up ended at June 2012. The standardised educational plan consisted of the discussion (1 h counselling) with the patient about each of the following points in this order: the cause and nature of any AEs and/or drug interaction; the tolerability profile of each drug present in the schedule, illustrated as a simple list including the commonest AEs presented in decreasing order of frequency; the clinical manifestations (if any) associated with the current drug interaction(s); any contraindication to the use of over-the-counter drugs potentially interfering with the current treatment schedule; the reasons for and the potential benefits of the suggested treatment change; an encouragement to withdraw any potentially interfering or contraindicated drug. The educational plan was administered at admission and in the same form after 1 month, as a reminder. The educational plan used in the EDU-COM study was conceived to be at the same time standardised (following a predefined sequence, as shown in online supplementary figure S1) and patient-tailored (adaptable to the individual patient profile). A centralised computer-generated allocation sequence was used for the randomisation; no blocking or stratification procedures were adopted. Inclusion criteria were as follows: age 18 years or older; established diagnosis of epilepsy; presence of one or more concurrent clinical conditions requiring chronic treatment (ie, daily treatment lasting 1 month or longer); clinically relevant AE attributable to the present treatment(s) and/or clinically relevant drug interaction; possible modification of the treatment schedule to eliminate AE(s) and/or risky drug interactions. For the purposes of this study, a clinically relevant AE was any symptom and/or clinical and/or laboratory sign related to the current drug regimen requiring active monitoring (according to the caring physician’s judgment); a clinically relevant drug interaction is an interaction requiring active monitoring (with clinical visits and/ or biochemical analyses according to the caring physician’s judgment) and/or treatment change. Exclusion criteria were as follows: non-modifiable treatment schedule; patient unable to understand or comply with an educational plan (the application of this exclusion criterion was left to the caring physician’s judgment); patient unable or unwilling to release a written informed consent. Drug-related AEs and drug interaction could be checked using a comprehensive internet-based drug database ((http://www. codifa.it), Italian language, related to the Istituto Superiore di Sanità).
Outcome assessment Outcome assessment was performed in person at baseline (M0), at 1 month (M1) and 6 months (M6) after admission. For the purposes of the study, treating and evaluating physicians were represented by different persons. Neither the patients nor the caring physicians involved in administering interventions were blind to the assigned arm. On the contrary, physicians assessing outcomes were blind to the assigned arm. At admission (M0), all patients were interviewed for the collection of: demographic data; clinical factors (date of diagnosis, seizure frequency in the preceding 6 months and seizure type(s), epilepsy syndrome, comorbidities); treatment regimen; number, type and severity of AEs. At each follow-up visit (M1 and M6 888
visit), the patients were inquired about the outcome of the AE, the occurrence of any new symptom, seizure frequency, any change in the treatment schedule, any medical contact or hospital admission. HRQOL was assessed at admission and at each visit (M0, M1, M6) using the Italian validated translation of the QOLIE-31 Inventory.19 The data for adverse drug events were collected during each visit as follows: generic description, followed by categorisation using the MedDRA SOC (2008) terminology; severity defined as: mild, moderate, severe, potentially lethal; relation to a particular drug present in the treatment schedule (defined by the evaluating physician using as a reference http://www.codifa.it). The number of drug interactions was defined by a centralised independent epilepsy specialist (SB) at the end of study, using the treatment forms included in the Case Report Form and taking as a reference a comprehensive internet-based drug database (http://www.codifa.it). The primary outcome was defined as the number of patients becoming free from AEs and/or drug interactions at end of study (M6). More specifically, to meet the primary outcome, patients admitted with only AEs or drug interactions had to become free of any event (whether AE or interactions) at M6, while those with AEs and drug interactions had to become free to at least one of the two components. Thus, the term ‘success’ was used for patients meeting the primary outcome during the study. The secondary outcomes were defined as follows: the number of patients free from drug interactions/adverse drug events at the intermediate visit (M1); the number of adverse drug events and drug interactions during the study period; the QOLIE-31 score changes from M0 to M6; the cost of drugs, medical contacts and hospital admissions during the study period.
Data quality assurance and control of bias Monitoring was carried out by a dedicated epilepsy specialist (SB), with regular on-site monitoring visits and verification of: protocol adherence, screening failures, accurate capture of clinical data, protection of safety of participants. To control for the principal sources of bias, the following strategies were adopted: consecutive patients were screened to minimise selection bias; the caring physicians (neurologists, epilepsy specialists) were involved as treating physicians to ensure correct application of the inclusion criteria, and choose the best treatment change; the physicians assessing outcome (neurologists or neurology residents) were different from the treating physicians.
Determination of sample size and study power According to the results reported from previous studies,3 we estimated that a change in treatment schedules under usual care was followed by the disappearance of AEs in 40% of patients. To be considered effective, an educational plan was expected to increase this proportion to 60% (20% absolute difference). For this difference to become statistically significant with an 80% power (1-β) and a 5% level of significance ( p=0.05), a total of 100 patients per treatment arm were planned to be enrolled. Recruitment was terminated after the enrolment of 187 patients (174 with confirmed eligibility criteria), leading to a reduction of the study power to 76%.
Statistical analysis The data were analysed using the Statistical Analysis System (SAS Institute, Cary, North Carolina, USA.) package for PC (V.9.2). The statistical analysis plan included descriptive statistics to compare the baseline characteristics of the two populations
Beretta S, et al. J Neurol Neurosurg Psychiatry 2014;85:887–892. doi:10.1136/jnnp-2013-306553
Downloaded from http://jnnp.bmj.com/ on March 10, 2015 - Published by group.bmj.com
Epilepsy and the distribution of the treatment schedules, the number and type of AEs, along with the cost items (medical contacts, hospitalisations, drugs) and the HRQOL scores. Univariate analysis was carried out using the χ2 test, the Wilcoxon–Mann Whitney test and the Friedmann test as appropriate. All analyses were performed in the intent-to-treat population (unless otherwise specified). Missing data on the primary endpoint were considered as failures. Missing data on the HRQOL scores were imputed using the Last Observation Carried Forward (LOCF) approach. Multivariable logistic regression was assessed on the primary endpoint using a ‘forward’ selection approach. Results are reported as ORs with 95% CIs. Comparisons between the two arms in term of number of AEs and drug interactions were assessed using Poisson regression analysis. HRQOL score was also analysed using repeated measures ANOVA with ‘treatment’ as the between-groups level (experimental and control), ‘time’ as the within-group level (baseline and M6) and the interaction term ‘treatment×time’. The ANalysis Of VAriance (ANOVA) repeated measures model was carried out using the PROC MIXED SAS procedure and choosing an ‘unstructured’ correlation matrix. All tests are two-tailed with significance set to 5%. We decided not to adjust for multiple comparisons as we used a single primary endpoint and all other secondary endpoints are considered subsidiary and, as such, results concerning these outcomes can only have an exploratory rather than a confirmatory interpretation.20
RESULTS Study population A total of 187 patients were screened, and 174 (93%) were confirmed eligible and randomly assigned to usual care (n=83) or standardised educational plan (n=91). Of these, 172 (99%) completed the protocol (see figure 1). Table 1 describes the intention-to-treat (ITT) population in terms of demographic characteristics, education, epilepsy (syndrome, disease duration and seizure frequency), number of comorbidities and drugs. The two treatment arms were well balanced for all these variables. Most patients had a focal
epilepsy syndrome less than 1 seizure/month, a normal neurological examination and a complex therapeutic regimen (3 or more drugs).
Efficacy of education plan: primary end point Table 2 reports the results obtained on the primary endpoint. The analysis on the ITT dataset was the primary analysis. The proportion of patients becoming free from AEs and/or drug interactions at the end of the study was significantly higher in the experimental arm ( p=0.0399). The results were confirmed by the adjusted multivariable logistic model that showed that the control group reported a significantly higher risk not to meet successfully the primary endpoint at the end of the study: OR (95% CI) 2.29 (1.03 to 5.09) ( p=0.0420).
Efficacy of education plan: secondary end points As shown in table 3, the number of patients becoming free from AEs and/or interactions at 1-month visit showed no difference between the two arms (p=0.1897). The number of AEs showed a non-significant difference and a substantial reduction over time (Friedmann test: p
