INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To improve the quality of asthma care provided by doctors, international and national evidence-based guidelines for the management of asthma have been developed (1). Guidelines, however, are not self-implementing (7, 8). Additional strategies that encourage doctors to accept and actually use guidelines are crucial for achieving high quality of care. International and national guidelines are more difficult to implement than local or internally developed guidelines (8). Just publishing national or international guidelines in a journal or mailing them to the target group is not sufficient. Dissemination in an educational intervention or as part of continuing education increases the doctors' attention to the contents and makes them more aware of the possible consequences for their personal practice. However, even if doctors are well informed about what to do, they often do not perform according to this knowledge. Specific educational strategies are needed to implement evidence-based recommendations in medical practice (8). When developing such strategies, attention should be given to the specific barriers to successful application of those recommendations in certain groups of doctors and practices (7, 9, 10). Barriers within doctors relate to competence, motivation and attitudes, and personal characteristics such as learning style, whereas barriers within practices exist as doctors do not work entirely independently (7, 11). Theories of adult learning stress the importance of motivation; the doctors must see the need and be willing to change their behavior to increase their professional competence (11, 12). Because many different factors and barriers may play a part in the change process, intensive efforts seem to be necessary to alter practice (7, 13). Learning methods that have proven effective include interactive and problem-solving exercises combined with feedback on performance (9, 14). Combined strategies that deal with different types of barriers seem to be more effective than single separate strategies (17, 18).

Prescribing of drugs is a common form of treatment. Irrational prescribing is a global problem and continuing medical education has become an international concern (19, 20). As a consequence of this internationalization, attention should be paid to the influence of context variables on the effectiveness of educational strategies. There are potential benefits of international collaboration and cooperation (13, 21). Educational interventions conducted so far are often limited to one specific setting. The generalizability of these findings to other settings is uncertain (13).

The aim of this study was to explore whether a specific educational approach has a similar effect when used in different health care settings. For this purpose, within the framework of the international Drug Education Project (DEP) an educational program to implement asthma guidelines, consisting of a combination of auditing and peer group discussions, was tested in four countries: The Netherlands (NL), Norway (N), Sweden (S), Slovakia (SK). The effect of the program was evaluated on changes in knowledge, attitudes, and prescribing behavior regarding the treatment of asthma in a randomized controlled study. Both the overall effect of the program, and a comparison of its effects in each of the four countries are presented.

Health Care Settings

In The Netherlands, Norway, and Sweden, the vast majority of asthma patients are treated by a general practitioner (GP). GPs may consider a consultative referral to a specialist without the specialist taking over the treatment. Referral is practiced for severe cases that do not respond to standard treatment, or in cases with difficulties in diagnosing. Thereafter, the patients may be treated for a period of time by a specialist and if the condition stabilizes, the patient is usually advised to seek maintenance therapy and management with his or her GP. Visiting a specialist without referral by a GP is not possible in The Netherlands, whereas in Sweden it is not compulsory to have a GP referral note to see a specialist. Especially in emergency situations, patients may be treated directly by a hospital-based specialist. In Norway, GPs are usually the gatekeepers to the specialists, but it is possible for a patient to consult one of the few private specialists in pulmonary medicine directly. However, this service is rather expensive as the fee is decided by the specialist. In Slovakia, most asthma patients are treated by lung specialist and/or allergologists. GPs in Slovakia have a gatekeeper role, but asthma patients are referred to lung specialists or allergologists working in outpatient clinics, because maintenance treatment with antiasthma drugs may only be prescribed by specialists. These prescriptions may then be repeated by the GP.

In The Netherlands, all prescriptions are of limited duration with a maximum of 3 mo, after which the patient has to consult the GP for a renewal. In Norway, prescriptions may also be issued for a maximum duration of 3 mo, but prescriptions may be repeated up to three times without consulting a GP. In Sweden, a GP can prescribe drugs for maintenance treatment for a longer period, up to 1 yr. However, such drugs are dispensed in 3-mo segments because reimbursement works on this basis. In Slovakia, there are no strict limitations on the amount of drugs prescribed, but the amount prescribed for maintenance treatment is usually for a 3-mo period.

In The Netherlands, national guidelines for the diagnosis and treatment of chronic nonspecific lung diseases in general practice were published in the national journal for GPs by the Dutch College of General Practitioners in 1992 (2, 3). These guidelines include recommendations on treatment of asthma. The Swedish guidelines for the treatment of asthma, issued in 1993 by the Medical Products Agency, were intended both for lung specialists and GPs (5). They were published in a national journal, and sent to all physicians and pharmacies in Sweden. The Slovak Medical Society issued national guidelines for diagnosis and treatment of asthma in 1996 (6). Those guidelines were especially developed for lung specialists and allergologists, and sent to them free of cost. In Norway, no official guideline concerning the treatment of asthma had been published at the time of the study. However, the Norwegian Drug and Therapeutics Formulary, an independent publication with contributions from experts in various fields and financial support from health authorities, includes recommendations regarding asthma treatment. This formulary is distributed once every 2 yr free of cost to Norwegian physicians and pharmacists (4).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

The educational program was intended for small groups of doctors and was tested and evaluated in a randomized controlled trial in each participating country. The study design, the educational approach, and the outcome measures were similar in all countries. Each country was allowed to organize and plan the recruitment, data collection, and meetings with doctors in a way that fitted their national situation best.

Randomization was conducted on group level (22). In The Netherlands and Norway groups of doctors were stratified on geographical region and in Sweden on group size. A parallel design was used; half of the groups in each country received the asthma education and formed the so-called asthma intervention arm. The other groups functioned as control groups and received an educational program on another disease (uncomplicated urinary tract infections). In Slovakia, because of the involvement of lung specialists and allergologists, the control groups did not receive an educational program.

Data on knowledge, attitudes, and prescribing behavior regarding the treatment of asthma were collected for all groups before and after the intervention. The effect of the intervention was assessed on each level, i.e., knowledge, attitudes, and prescribing, providing insight into which steps of the dissemination and implementation process were successfully taken. To determine whether the educational program was effective in increasing the actual implementation of the existing guidelines, the impact on prescribing behavior was considered most relevant.

Study Population

The number of groups involved in the study varied from 20 in Slovakia to 36 in Sweden (Table 1), whereas the number of doctors who participated ranged from 81 in Slovakia to 204 in Sweden. In The Netherlands, GPs have regular meetings in so-called pharmacotherapy counseling groups. In Sweden, most GPs work in group practices at health care centers. These existing groups were used when recruiting participants for the study. In The Netherlands doctors from 24 of the 98 pharmacotherapy counseling groups invited to participate agreed to do so (24%). The average number of GPs in the participating groups did not differ significantly from that in the nonparticipating groups in this country (7.5 versus 8.8 GPs per group). Also, the average age of the GPs in the groups did not differ (47.1 versus 46.2 yr). However, the participating groups did meet more frequently; 83% of the participating groups held more than 5 group meetings each year, whereas in the other Dutch groups this percentage was 65. Five percent of the nonparticipating groups was actually not active, i.e., having no meetings at all in 1 yr. In Sweden, 35% of the health care centers invited participated (39 of 112, merged to 36 groups for practical reasons). Doctors in Norway and Slovakia were recruited on an individual basis and grouped for the purpose of the educational intervention. The participation rate was 31% in Norway (199 of 635), and 20% in Slovakia (81 of 405). In Norway, the participating doctors did not differ from the nonparticipating doctors regarding age, gender, and practice organization (mean age 44.1 versus 45.8 yr; 23% versus 22% female doctors; 29% versus 28% doctors working in a single-handed practice). No data were available for the nonparticipating doctors in Sweden and Slovakia.

Characteristics of participants, including number of years in practice and gender, were collected on the baseline knowledge and attitudes questionnaire (Table 1).

Educational Strategy

The educational strategy can best be described as a self-learning method based on individual auditing and feedback of performance for small peer groups. No authoritative teaching was involved. Two educational meetings were planned for each peer group. These meetings took place in the vicinity of the doctors' practices, either in a conference room of some public building, or at a local pharmacy, health care center, or clinic. During these meetings the groups were asked to discuss their management of asthma patients in the light of existing guidelines. To enable this discussion the research teams prepared feedback material for each group.

During the first intervention meeting, each doctor received individual feedback on his decision policies regarding the treatment of specific asthma cases. Before this first meeting, the doctors received these cases on paper, and had been asked to answer several questions for each case. The feedback material offered introspection in optimal and suboptimal prescribing routines (as defined in existing guidelines), and in specific patient characteristics that appeared to trigger suboptimal prescribing decisions. In addition, a short overview of the contents of the guidelines was distributed during the first meeting. The groups were encouraged to reach a consensus regarding the optimal prescribing policies for both the maintenance and exacerbation treatment of asthma patients. The purpose was that the doctors would internalize the externally developed guidelines. During the second intervention meeting, the doctors received feedback on their actual prescribing. The feedback was developed to offer introspection into possible discrepancies between the doctor's current practice and the optimal prescribing policies according to the guidelines. It thus showed the need for change, and stimulated the doctors to take action.

All feedback was provided on an individual doctor level, but presented and discussed in the group settings. A stepwise procedure was followed during the meetings, including: familiarizing with the feedback material, comparing and evaluating the feedback on prescribing decisions in the light of the guidelines, and trying to reach a consensus on particular management decisions. Doctors were identified in the feedback material by code numbers known only to the doctor. In most cases, however, the doctors decided to disclose their identity during the meetings. Doctors not present during a meeting received the feedback material after the meeting by mail or directly from a colleague. The organization of the meetings was adjusted to the needs and wishes of groups of doctors in each country. In Sweden, Norway, and Slovakia, members from national research teams (so-called facilitators) were present at all meetings to guide and help the group when necessary. In The Netherlands, where the groups were used to conduct their meetings without outside interference, members of the research team were only present at the first meeting. In Sweden and Norway, the facilitator offered the feedback material. In some groups the facilitator acted as technical chairperson, encouraging the discussion among the participating doctors. Other groups decided to choose one of their own members as chairperson. In The Netherlands and Slovakia, the groups selected one of their members as moderator, who prepared the meetings and presented the feedback material. Before the first intervention meeting, these moderators received a short explanation of the feedback material and how to present it during the meetings.

The feedback material focused on matters related to the key messages from the existing guidelines and recommendations (1). The recommendations on asthma treatment in the participating countries were similar. All emphasized the inflammatory nature of asthma. The first key message is to use inhaled corticosteroids promptly. A stepwise approach is recommended, i.e., to initiate anti-inflammatory therapy if a patient uses bronchodilators continuously, and to increase the dosage of inhaled corticosteroids if adequate control, indicated by frequent use of bronchodilators, is not achieved. The second key message is to treat severe exacerbations with oral corticosteroid courses, and to refrain from the routine use of antibiotics.

For preparing the feedback on decision policies to be used at the first meeting, the doctors received two series of written simulated cases (see APPENDIX). In The Netherlands, Norway, and Slovakia the case material was sent by mail to the doctors participating in the asthma intervention arm. In Sweden, all doctors received and completed the case material at a preintervention visit of one of the researchers at each health care center. The first series, focusing on the first key message, consisted of asthma patients who visited the doctor's office to get a repeat prescription. The cases systematically varied in their age, use of -agonists, use of inhaled corticosteroids, current symptoms, and self-measured peak expiratory flow (PEF) values. The doctors were asked for each case whether they considered the patient adequately treated, and which changes they would make regarding the treatment. The second series, focusing on the second key message, consisted of asthma patients who came to the doctor's office on a semi-emergency visit. They varied in their age, current symptoms (breathing problems, cough/sputum, fever), and current PEF values. Their daily intake of inhaled corticosteroids was already 1,600 µg, and the doctors were asked whether they would prescribe an oral corticosteroid course and/or an antibiotic. All answers of the doctors were analyzed to determine how the various patient characteristics influenced the treatment decisions (so-called clinical judgment analysis) (23, 24). This resulted for each doctor in a personal decision policy for maintenance treatment (based on the first series of cases), and one for the treatment of exacerbations (based on the second series of cases). Both the doctors' decisions and decision policies were presented as feedback material during the first meeting.

For preparing the feedback on prescribing behavior needed for the second meeting, prescribing data of the baseline period were used. This feedback on actual prescribing showed: graphs of patients who were prescribed high levels of bronchodilators without any inhaled corticosteroids or with (too) low levels of inhaled corticosteroids (first key message), and graphs of the number of oral corticosteroid and specific antibiotic courses prescribed to asthma patients (second key message).

In all countries except Slovakia, one or more local pharmacists were sometimes participating in the educational group meetings. More characteristics of the meetings in the different countries are presented in Table 2. The most important difference was that the Slovakian groups conducted only one instead of two intervention meetings, and they did not receive any feedback on their actual prescribing. This was because the prescribing data were not available in time in Slovakia.

Outcome Measurements

Prescribing behavior. Prescription data of a 6-mo (NL, S, SK) or 12-mo period (N) were collected before and after the intervention through pharmacies, insurance companies, or directly from computerized databases of doctors dispensing drugs in their practice. Baseline data were collected in the different countries between May 1994 and August 1996. To control for seasonal influences on asthma treatment, outcome data of a comparable period were collected after the intervention (between September 1995 and August 1997).

Patients age 18 to 49 yr who received at least one antiasthma drug (as defined in the Anatomical Therapeutic Chemical classification index; ATC code: R03 [25]) during the selected 6-mo or 12-mo period were defined as asthma patients. By only including 18- to 49-yr-olds, an attempt was made to exclude patients with chronic obstructive pulmonary disease (COPD).

The data material consisted of all antiasthma drugs (ATC code R03), oral glucocorticosteroids (ATC code H02AB), and oral antibiotics commonly used for respiratory tract infections (Table 3) dispensed to the defined asthma patients. In Norway, only those antibiotic prescriptions could be included that contained information on the birth date of the patient.

For all prescriptions the number of defined daily dosages (DDD [25]) was available. Antibiotics prescribed for maintenance treatment were excluded (i.e., prescriptions for > 14 DDDs). Oral glucocorticosteroid prescriptions were included if prescribed as a short course (i.e., prescriptions for  14 d or  30 DDD).

To evaluate the effect of the educational intervention on the prescribing behavior, four prescribing indicators were developed and calculated per group for each country. The general quality of maintenance treatment (first key message) was assessed by the following indicators:

1. Proportion of patients receiving inhaled corticosteroids of all defined asthma patients.

Two additional indicators giving insight in the degree to which a stepwise approach is followed were estimated:

a. Proportion of patients receiving on average per day > 0.25 DDD (on average 1 inhalation) inhaled bronchodilators (₂-agonists or anticholinergics) without anti-inflammatory treatment (inhaled corticosteroids or cromoglycates) of all defined asthma patients receiving inhaled bronchodilators.

b. Proportion of patients receiving on average per day < 0.5 DDD inhaled corticosteroids (< 400 µg budesonide/beclomethasone or < 300 µg fluticasone) and > 0.5 DDD inhaled bronchodilators (on average more than 2 inhalation/day; ₂-agonists or anticholinergics) of all defined asthma patients receiving a combination of inhaled corticosteroids and bronchodilators.

The quality of exacerbation treatment (second key message) was assessed by:

1. Proportion of oral corticosteroid courses of all antibiotic and oral corticosteroid courses dispensed to defined asthma patients.

Knowledge and Attitudes

The extent to which knowledge and attitudes were in agreement with the national guidelines was assessed 3 to 6 mo before and 3 to 6 mo after the educational intervention by use of a questionnaire. The original English questionnaire has been translated in all countries to their own language and back again to English to check the translation, using experienced translators. The questionnaire was sent by mail to all participants except in Sweden where the questionnaire before the intervention was distributed during a preintervention visit of one of the researchers to each health care center.

Knowledge was assessed about maintenance treatment, treatment of asthma exacerbations, PEF measurement, and inhalation technique (in total 10 questions). Attitude statements concerned maintenance treatment, treatment of asthma exacerbations, PEF measurement, patient self-management, and patient follow-up (in total 12 statements). Besides these overall scores, separate subscores were calculated for those knowledge questions and attitudes statements explicitly related to the two key messages. Subscore A1 included 3 questions about maintenance, and A2 included 4 questions about exacerbations. Subscore B1 and B2 included each 2 statements regarding maintenance and exacerbations, respectively. For each country, knowledge and attitudes scores before and after the intervention were calculated per group, expressing the proportion of responses in agreement with the national guidelines of the total number of questions answered (missing data were excluded). No negative markings were given for incorrect replies. Only answers of doctors who completed more than two-thirds of the questions needed for a specific score were included in the analysis.

Analysis

The effect of the educational intervention on knowledge and attitudes was evaluated using a classic linear regression model, whereas the effect on prescribing behavior was analyzed by use of a logistic regression model (26). The effect of the intervention, the trend effect, and the baseline difference between control and intervention arm were estimated simultaneously as suggested by Huck and McLean (27). A multilevel model was used taking into account the hierarchical structure of the data (prescriptions were nested within patients, patients were nested within doctors, and doctors were nested within groups of doctors) by including random effects (28). The significance of the regression coefficients was tested by evaluating their estimates divided by their standard errors against standard distribution. A 2-tailed significance level of 0.05 was used. Effect size estimates (Hedges's g) were calculated as the difference between the mean change of the intervention group and the mean change of the control group divided by the pooled standard deviation (29, 30). A meta-analysis was conducted to assess the combined effect of the intervention in the four countries. Overall effect sizes were calculated by weighting the individual effect size estimates for sample size as outlined by Hedges and Olkin, i.e.,_j · g/w_j, where w_j = n^c_jnⁱ_j/n^c_j + nⁱ_j (30). Significance testing was conducted by calculating the overall statistics using the inverse normal method, i.e.,_j/k where k equals the number of studies included (29, 30).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Characteristics of Participating Groups and Intervention Meetings

Within each country, groups participating in the asthma intervention arm and control arm were comparable (Table 1). Some differences existed between the countries, including a higher proportion of female doctors in Sweden and Slovakia, and, on average, less years of practice experience in Sweden.

The percentage of doctors attending the intervention meeting was 100% in Slovakia (first and only meeting) whereas in Norway on average 68% of the doctors per group were present during both the first and the second meeting (Table 1). The average duration of the meetings in Norway was about twice as long compared with the meetings in the other countries. The time between both meetings varied between the countries from 1 wk in Norway to about 10 wk in Sweden.

Overall Effect of Educational Intervention

The educational program significantly changed the doctors' overall attitudes (effect size 0.75) and part of their prescribing behavior in the desired direction (effect sizes 0.66, 0.20, 0.24, and 0.71), but there was no positive effect on their overall knowledge (Tables 4 and 5). General trends were observed regarding overall knowledge, and regarding two of the four prescribing indicators (Tables 4 and 5). Restriction of the analysis to doctors who were actually present during at least one of the meetings, somewhat increased most effect sizes (effect sizes for changes in prescribing behavior 0.67, 0.25, 0.27, and 0.66), but did not show more significant changes (data not shown).

Effect of Educational Intervention in Each Country

The program was most effective in The Netherlands, while limited effect was observed in Norway and Slovakia, and no significant effect was seen in Sweden. In these latter countries, the baseline levels were often less in agreement with the guidelines, and relatively strong general trends were observed toward guideline recommendations. In none of the countries the results significantly changed when restricting the analysis to doctors who were actually present during at least one of the meetings (data not shown).

Looking at prescribing behavior, the largest effects (effect size > 1) can be seen in The Netherlands in the proportion of asthma patients receiving inhaled corticosteroids, and in the prescribing of oral corticosteroid courses relative to antibiotic courses for the treatment of asthma exacerbations (Table 4). The prescribing of oral corticosteroid courses increased also significantly in Norway (effect size 0.87), whereas in Slovakia there was a considerable but not significant improvement (effect size 0.79). Moderate effects (effect sizes between 0.4 and 0.8) were seen regarding prescribing more inhaled corticosteroids to asthma patients in Slovakia and Norway, and having less patients on continuous bronchodilator monotherapy in Slovakia after the intervention. These changes were not statistically significant when compared with changes observed in the control groups. Looking at the effect on knowledge of the doctors, the only significant change was observed in Slovakia regarding the knowledge about the treatment of asthma exacerbations (Table 5). Regarding the doctors' overall attitude scores, large effects (effect sizes > 0.8) were observed in The Netherlands, Norway, and Slovakia, but in the latter country this effect was not significant. These improvements were in part the result of changes in attitudes concerning the treatment of asthma exacerbations. Attitudes concerning maintenance treatment were already to a large extent in agreement with the guidelines in all countries before the educational intervention, leaving little room for improvement.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study evaluated and compared the effect of an educational program to implement guidelines on asthma treatment in four European countries with different health care settings. An important result of this study is that a program that has some effect in one health care setting does not necessarily give the same effect elsewhere. In general, the educational program improved both the doctors' attitudes as well as some of their prescribing behavior. No positive effects were seen regarding the doctors' knowledge. These results are in line with what could be expected from a program that concentrates on the implementation of guidelines and not on the transfer of knowledge (31). When looking at the results in each country, the program improved prescribing of inhaled corticosteroids in The Netherlands, treatment of exacerbations in The Netherlands and Norway, attitudes in The Netherlands, Norway, and Slovakia, whereas knowledge improved slightly in Slovakia. No significant effect was observed in Sweden. These varied results warrant a careful analysis.

The educational program focused on the maintenance treatment of asthma as well as the treatment of asthma exacerbations, but improvement in maintenance treatment was only seen in The Netherlands. Although attitudes about maintenance treatment were already to a large extent in agreement with the guidelines in all countries before the educational meetings, prescribing behavior was not optimal in any of the countries. However, a change in chronic treatment may be more difficult to implement and to detect than a change in an acute treatment. To implement messages regarding maintenance treatment, doctors must alter a routine treatment that the patient may feel comfortable with. The doctor is less likely to be faced with a patient habituated to a particular drug regimen when changing treatment of exacerbations, a more acute condition. In addition, not all asthma patients may have visited their doctor for their maintenance treatment during the period for which outcome data were collected. In all countries except The Netherlands, patients can get prescriptions for their maintenance treatment valid for a year. Only in Norway, outcome data were collected for a whole year.

A number of factors beyond the scope of the research design may have influenced the findings. A first consideration in the interpretation of the findings is the degree to which doctors were accustomed to the mode of learning: the small group interventions. The educational program was developed for and used by groups of doctors. The Netherlands boasts a well- established nationwide system of pharmacotherapy counseling groups which has been in place since 1991; GPs meet regularly to discuss pharmacotherapy and during the meetings individualized prescription feedback is often provided. This experience of participating doctors with small group continuing medical education programs could have contributed to the effect of the intervention in The Netherlands. This is in line with earlier findings in this country. Doctors who were accustomed to discussing their prescribing in peer groups changed their behavior more as a result of such peer group meetings than doctors who were not used to this approach (14). In Sweden the vast majority of GPs have been working in group practices for more than two decades, but were, at the time of the study, not accustomed to discussing pharmacotherapy and prescribing feedback within these groups on a regular basis. In Norway and Slovakia the groups were newly formed for the intervention, and were not used to discussing pharmacotherapeutic subjects with each other.

Second, the countries differed substantially with respect to the level of recognition of guidelines as "gold standards" for medical practice. Although identical key messages were delivered in all four countries, based on international and national guidelines, there are differences in the nature and distribution of the existing national guidelines, and this is likely to influence the possible effect of a guideline. Only in The Netherlands, national guidelines were developed by GPs and intended primarily for their use. This guideline initiative has been quite successful and highly accepted, because it is initiated and "owned" by the GPs themselves (32). This could have contributed to a larger impact of an educational program to implement such a guideline. In countries other than The Netherlands, guidelines were not specifically written for general practice. In The Netherlands, Sweden, and Norway, national recommendations for asthma treatment were published 2 to 4 yr before this study started. In Slovakia, national guidelines on the treatment of asthma became available during the study. The difference in availability of recommendations between Slovakia and the other countries probably has influenced the extent to which knowledge, attitudes, and prescribing behavior were already in agreement with the guidelines in the different countries before the intervention. A comparison of the baseline situation showed that the gap between recommendations and practice, representing the potential for improvement, varied considerably between the countries, and especially in Slovakia there was much room for improvement.

One might expect that the room for improvement is related to the effect of the intervention. When looking at knowledge and attitudes, the largest improvements were indeed seen when the baseline performance was low (i.e., knowledge and attitudes regarding the treatment of exacerbations in Slovakia). Looking at changes in prescribing, however, the most significant changes were seen when the room for improvement was relatively small (i.e., use of inhaled corticosteroids in The Netherlands, use of oral corticosteroids in both Slovakia and The Netherlands). When there was much room for improvement, there were changes observed in the intervention groups, but these were often matched by similar changes in the control groups. This could be explained by the notion that general trends toward guidelines recommendations are more likely to occur when there are large discrepancies between these guidelines and actual practice. The impact of one specific intervention may then shrink into insignificance in comparison to the general effect of other activities in a country.

Surprisingly, attendance of the doctors was not clearly related to the impact of the intervention. Although in The Netherlands the number of doctors attending two meetings was higher than in countries where less effect was seen, limiting the analysis to doctors attending the meetings did not significantly increase the effects in those countries. One reason for this might be that nonattenders often received feedback material afterwards and may have discussed the material with one or more colleagues. Such interaction has been observed previously regarding doctors working in health care centers (33). In Slovakia, where the doctors did not receive feedback on their actual prescribing and only one meeting was held with each group, smaller effects were to be expected regarding changes in prescribing. However, for three of the four prescribing indicators the effect sizes were quite large in Slovakia.

The average size of participating groups varied from four doctors in Slovakia to more than seven in The Netherlands. However, it can not be concluded from the results that a smaller or larger group size is associated with increased effectiveness of the intervention. The duration of the meetings or the time between both meetings varied considerably between the countries, but these variables were not related to the magnitude of effect.

Some limitations regarding the detection of possible changes in this study need to be addressed. First of all, the data used to detect changes were dispensing data. These reflected only prescriptions for drugs the patient chose to use, not necessarily all prescriptions issued by doctors, therefore representing an amalgam of doctor and patient behavior (34). There are however no indications to expect a difference neither between the countries nor before and after the intervention in this respect.

Second, the power to detect differences in prescribing was smaller than expected, because the variance between and within groups was larger than estimated. Especially in Slovakia, where only 20 groups of four doctors were included this explains why some large effects were not significant at the 0.05 level (29).

Third, general trends in the treatment of asthma in agreement with the recommendations as seen for two prescribing indicators also reduced the power to detect possible improvements due to the educational intervention. This is similar to findings of Feder and coworkers in the United Kingdom (35). External factors, such as increased publicity and other educational activities regarding asthma management, may be the cause of these trends. Additionally, in Sweden the outcome assessment may have been biased by the fact that the doctors in the control group also responded to the written simulated cases of asthma patients during the preintervention visit, and might have been influenced by this experience.

Finally, our prescribing indicators may not detect all changes in prescribing behavior. The general indicator assessing the quality of maintenance treatment (indicator 1) does not take into account the amount of drugs prescribed to each individual patient. Including such information could provide an indicator that is more sensitive to specific dosage changes in prescribing behavior. In the two prescribing indicators giving insight in the degree to which the doctors apply the recommended stepwise approach (indicators 1a and 1b), the amount of drugs prescribed was included. However, this inclusion is subject to some error. Estimates have to be made of the average daily amount of inhaled bronchodilators a patient receives. Such estimates can be made most accurately if at least two prescriptions have been delivered during the period of data collection. This was not always the case, resulting in too high estimates of the inhaled bronchodilator use. Especially in Slovakia and Sweden, where the habit is to prescribe large packages of bronchodilators, these overestimations occur more often, and therefore some possible improvements in the maintenance treatment after the intervention may have gone undetected.

This study shows that improvements in drug treatment are possible with a program based on self-learning in small peer groups. Different attitudes and experiences with guidelines as well as with continuing medical education programs, and different opportunities for change combined with prevailing trends of prescribing behavior, are possible factors that may explain why a certain message or a particular educational strategy works in one country, but not in another. Tailoring of the educational strategy in each country seems to be necessary to enhance the implementation of guidelines in practice.