Characteristics of GPs responding to an educational intervention to minimise inappropriate prescriptions: subgroup analyses of the Rx-PAD study

Background Interventions aimed at improving GPs’ prescribing practice usually apply a 'one size fits all' when analysing intervention effects. Few studies explore intervention effects by variables related to the GPs’ age, sex, specialist status, practice type (single-handed versus group), practice setting (urban versus rural), and baseline performance regarding the target of an intervention. Aim To explore the characteristics of the GPs responding to a comprehensive educational intervention. Design & setting A secondary analysis of a cluster, randomised educational intervention in Norwegian general practice. Pre-intervention data were captured from January 2005 to December 2005, and post-intervention data from June 2006 to June 2007. The intervention was carried out from January to June 2006. Method Eighty continuing medical education (CME) groups, including 449 GPs aged 27–68 years, were randomly allocated to either an education intervention arm (41 groups, 250 GPs) or a control arm (39 groups, 199 GPs). The primary outcome was GPs' change in potentially inappropriate prescriptions (PIPs) per 100 prescriptions issued to patients aged ≥70 years. The interaction between intervention outcome and variables related to the GPs and their practices were tested. Results Improvements in prescribing were highest among GPs aged 57–68 years (incidence rate ratio [IRR] = 0.77 [95% confidence interval {CI} = 0.73 to 0.81]), those who were specialists (IRR = 0.80 [95% CI = 0.78 to 0.82]), and those who worked in single-handed practices (IRR = 0.75 [95% CI = 0.68 to 0.83]), among GPs with 2.4 to 2.9 PIPs per 100 prescriptions at baseline (IRR = 0.74 [95% CI = 0.70 to 0.78]), and GPs with ≥15 prescriptions per patient per year at baseline (IRR = 0.77 [95% CI = 0.73 to 0.80]). Conclusion The GPs with the lowest adherence to recommended practice at baseline improved their practice most.


Introduction
The incidence of complete Achilles tendon rupture is 18 per 100 000 patient-years 1 and is usually diagnosed clinically by GPs. The extent of clinical misdiagnosis is unknown in Norway, but may be high. 2 This is important as delayed treatment has unfavourable consequences. 1, 3 We report how a GP, with no clinical ultrasound experience, recorded images with a pocket-sized ultrasound device (PSUD) under supervision to confirm a complete Achilles tendon rupture. This could present a new indication for GP ultrasound.

Case report
A 36-year-old man experienced acute pain above the right heel accompanied by an audible snap while sprinting. He immediately had difficulty walking and 3 hours later consulted an on-call GP. Posterior ankle swelling with a tender depression 3 cm proximal to the calcaneum was found. Active plantar flexion against resistance was weak and Simmonds-Thompson test was 'partially positive' on applying a strong calf-squeeze. Based on these findings, calf muscle rupture was diagnosed as the Achilles tendon was thought to be intact. The patient was advised to elevate the foot and wait 2 weeks for improvement. Two days later a second GP, who was aware of a history of an audible snap, considered complete tendon rupture and reexamined the patient. Findings included an absent right heel raise due to weakness, minimal active plantar flexion against gravity and lying prone, significant right ankle swelling without bruising, and an altered angle of declination. Palpation elicited no ankle bony tenderness, yet a painful gap was identified 6 cm proximal from the calcaneal attachment, along the line of the Achilles tendon. Simmonds-Thompson's test was clearly positive. The positive Simmond's triad indicated a clinical diagnosis of complete rupture of the Achilles tendon. A 3.4-8 MHz linear array probe PSUD (VScanÔ dual probe, GE Healthcare), set at a depth of 3.5 cm, was used under the supervision of a rheumatologist experienced in ultrasound. The tendon was enlarged from 1 cm to 6 cm above the calcaneal insertion, where a clear gap was seen ( Figure 1). Two hours later a radiologist-performed ultrasound (LOGIQ E9Ô, GE Healthcare) and reported an enlarged distal tendon and a complete rupture at 5-6 cm from the calcaneal attachment, creating a 2.7 cm blood-filled gap ( Figure 2). Surgical exploration 8 days post-injury found a complete Achilles tendon rupture '5-10 cm above the ankle joint'.

Discussion
Tromsø Hospital serves a large area with a population of approximately 160 000. Between 2010-2014 an average of 21 patients per year were referred by their GP for suspected Achilles rupture.

Introduction
The incidence of complete Achilles tendon rupture is 18 per 100 000 patient-years 1 and is usually diagnosed clinically by GPs. The extent of clinical misdiagnosis is unknown in Norway, but may be high. 2 This is important as delayed treatment has unfavourable consequences. 1, 3 We report how a GP, with no clinical ultrasound experience, recorded images with a pocket-sized ultrasound device (PSUD) under supervision to confirm a complete Achilles tendon rupture. This could present a new indication for GP ultrasound.

Case report
A 36-year-old man experienced acute pain above the right heel accompanied by an audible snap while sprinting. He immediately had difficulty walking and 3 hours later consulted an on-call GP. Posterior ankle swelling with a tender depression 3 cm proximal to the calcaneum was found. Active plantar flexion against resistance was weak and Simmonds-Thompson test was 'partially positive' on applying a strong calf-squeeze. Based on these findings, calf muscle rupture was diagnosed as the Achilles tendon was thought to be intact. The patient was advised to elevate the foot and wait 2 weeks for improvement. Two days later a second GP, who was aware of a history of an audible snap, considered complete tendon rupture and reexamined the patient. Findings included an absent right heel raise due to weakness, minimal active plantar flexion against gravity and lying prone, significant right ankle swelling without bruising, and an altered angle of declination. Palpation elicited no ankle bony tenderness, yet a painful gap was identified 6 cm proximal from the calcaneal attachment, along the line of the Achilles tendon. Simmonds-Thompson's test was clearly positive. The positive Simmond's triad indicated a clinical diagnosis of complete rupture of the Achilles tendon. A 3.4-8 MHz linear array probe PSUD (VScanÔ dual probe, GE Healthcare), set at a depth of 3.5 cm, was used under the supervision of a rheumatologist experienced in ultrasound. The tendon was enlarged from 1 cm to 6 cm above the calcaneal insertion, where a clear gap was seen ( Figure 1). Two hours later a radiologist-performed ultrasound (LOGIQ E9Ô, GE Healthcare) and reported an enlarged distal tendon and a complete rupture at 5-6 cm from the calcaneal attachment, creating a 2.7 cm blood-filled gap ( Figure 2). Surgical exploration 8 days post-injury found a complete Achilles tendon rupture '5-10 cm above the ankle joint'.

Discussion
Tromsø Hospital serves a large area with a population of approximately 160 000. Between 2010-2014 an average of 21 patients per year were referred by their GP for suspected Achilles rupture.

Introduction
The incidence of complete Achilles tendon rupture is 18 per 100 000 patient-years 1 and is usually diagnosed clinically by GPs. The extent of clinical misdiagnosis is unknown in Norway, but may be high. 2 This is important as delayed treatment has unfavourable consequences. 1, 3 We report how a GP, with no clinical ultrasound experience, recorded images with a pocket-sized ultrasound device (PSUD) under supervision to confirm a complete Achilles tendon rupture. This could present a new indication for GP ultrasound.

Case report
A 36-year-old man experienced acute pain above the right heel accompanied by an audible snap while sprinting. He immediately had difficulty walking and 3 hours later consulted an on-call GP. Posterior ankle swelling with a tender depression 3 cm proximal to the calcaneum was found. Active plantar flexion against resistance was weak and Simmonds-Thompson test was 'partially positive' on applying a strong calf-squeeze. Based on these findings, calf muscle rupture was diagnosed as the Achilles tendon was thought to be intact. The patient was advised to elevate the foot and wait 2 weeks for improvement. Two days later a second GP, who was aware of a history of an audible snap, considered complete tendon rupture and reexamined the patient. Findings included an absent right heel raise due to weakness, minimal active plantar flexion against gravity and lying prone, significant right ankle swelling without bruising, and an altered angle of declination. Palpation elicited no ankle bony tenderness, yet a painful gap was identified 6 cm proximal from the calcaneal attachment, along the line of the Achilles tendon. Simmonds-Thompson's test was clearly positive. The positive Simmond's triad indicated a clinical diagnosis of complete rupture of the Achilles tendon. A 3.4-8 MHz linear array probe PSUD (VScanÔ dual probe, GE Healthcare), set at a depth of 3.5 cm, was used under the supervision of a rheumatologist experienced in ultrasound. The tendon was enlarged from 1 cm to 6 cm above the calcaneal insertion, where a clear gap was seen ( Figure 1). Two hours later a radiologist-performed ultrasound (LOGIQ E9Ô, GE Healthcare) and reported an enlarged distal tendon and a complete rupture at 5-6 cm from the calcaneal attachment, creating a 2.7 cm blood-filled gap ( Figure 2). Surgical exploration 8 days post-injury found a complete Achilles tendon rupture '5-10 cm above the ankle joint'.

Discussion
Tromsø Hospital serves a large area with a population of approximately 160 000. Between 2010-2014 an average of 21 patients per year were referred by their GP for suspected Achilles rupture.

Introduction
Last summer our small medical team visited the Calais 'Jungle'. Since that time much has changed and the camp is being demolished and by the time this article is read, it will probably be long gone. Some youngsters are finally being brought to the UK under the 'Dubs' amendment. However, once this camp is cleared it will not solve the ongoing flight of refugees from war torn areas: other camps are already appearing.

July 2016
A young Afghan man caught his finger on a sharp point while trying to cross a barbed wire fence. The finger was partially degloved. He attended the local hospital, where they placed a few sutures, but now, 2 weeks later, the skin is necrotic and the underlying tissue looks infected. He is in danger of losing his finger. A middle-aged Sudanese man has been having rigors and is generally unwell. He says it is similar to when he last had malaria.
A young Ukrainian woman complains of lower back pain and urinary frequency. The paths of these three people may never have crossed; yet here they are, denizens of the Calais Jungle. They turn up to a makeshift primary care 'clinic' that we set up in the heart of the unofficial refugee camp one weekend in July 2016.
With only basic medical supplies, we are immediately challenged by what we see. How can we arrange secondary care for the young Afghan in danger of losing his finger? We try to persuade him to return to the original local hospital, but he is reluctant. It was not a good experience for him the first time round.
With the other two patients, it is easier. They can attend the Salam clinic run by a local association during weekdays. Later, we receive word that malaria has been confirmed in our Sudanese patient.
More people arrive, presenting with scabies, rat bites, tinea, chest infections, and wheezing from inhaling smoke from fires lit to cook and keep warm in their tents at night. We examine a severely malnourished 2-year-old boy. We meet several of the camp's 600 unaccompanied children, at grave risk of sexual exploitation. We learn that there is inadequate safeguarding in place to protect them. A young Eritrean man comes in worried about his eye. He has sustained direct ocular trauma from a rubber bullet, and will never see normally again out of that eye. We see haematomas from police batons, and hear about children being exposed to tear gas again and again ( Figure 1).

The reality
These are no ordinary patients. They have travelled far from home to escape war, poverty, and misery. They have endured personal odysseys to get here, experienced untold hardships, and suffered unimaginable privations. Many have survived the loss of their families, torture, and rape. Their journeys over, for the moment at least, they must make their homes in the Calais Jungle. Their new shelters are in many cases mere tarpaulin covers, and their new beds just rugs on the ground. They own next to nothing. There is little for them to do, besides use their ingenuity to cross the English Channel in search of a better life. They are vulnerable to exploitation, crime, injury, and disease. Potentially violent clashes with local police, with other ethnic groups resident in the Jungle, or local far

Introduction
Older people are generally more vulnerable to adverse drug reactions due to altered pharmacokinetics and pharmacodynamics, comorbidity, and polypharmacy. 1 Inappropriate medication use in frail older people puts them at increased risk for drug-related harms, which may cause impaired quality of life, increased morbidity, and even death. 2,3 In general practice, drugs or combinations of drugs considered potentially inappropriate for older people are reported to make up 5-34% of all prescriptions issued to older patients. [4][5][6][7][8] The Rx-PAD study, a large educational intervention study carried out among 449 GPs in Norway, was carried out betwen 2005-2007. 9 The aim of that intervention was to improve GPs' prescribing practice for older patients (!70 years) in terms of reducing the prevalence of a number of listed explicit criteria for PIPs (Box 1). Average outcome figures 1 year after the intervention revealed that the average proportion of PIPs per 100 prescriptions per GP in the intervention group, as compared with the control group, was reduced by 13% in relative terms. 9 However, it is not known which GPs responded in line with the aims of the intervention. Such information may be relevant for the planning of future comparable educational intervention studies. 10 The aim of this study was to re-analyse the effect of the Rx-PAD intervention study 9 by subgroup, analysing the intervention effect by characteristics linked to the GPs or their practices.

Method
In this subgroup analysis of the Rx-PAD study, 9 outcome measures of seven explanatory variables were explored. In randomised trials, subgroup analyses are commonly applied to examine how various explanatory variables may be associated with the intervention outcome. Subgroup analyses are not based on randomised comparisons and their limitations are well-established: false positives due to multiple comparisons, false negatives due to inadequate power. 10 Although it is almost impossible to fully eliminate such faults in a subgroup analysis, the aim of the present study was to formally test the interaction between the intervention and subgroups among the GPs by using a strict P-value to control for multiple testing.
The Rx-PAD study was designed to assess the effects of an educational intervention given to GPs, which aimed at reducing their prescribing rates of PIPs for older patients. 11 Eighty CME groups comprising 449 GPs were allocated to either intervention (41 groups, 250 GPs) or control (39 groups, 199 GPs). The control group participated in a corresponding educational intervention to improve the prescribing of antibiotics. 12 The two groups were control group for each other. All GPs were part of the Norwegian GP patient-list system and the GPs' prescription data were extracted from the Norwegian Prescription Database (NorPD). 13 PIPs were defined using the 13 explicit items given in Box 1. The clinical relevance of these was validated by the participating GPs in regional meetings 11 and in focus group interviews among participating tutors and GPs. 14 Briefly, the intervention comprised an individual prescription report (sent to all GPs in the intervention group) and two CME-group sessions chaired by a tutor from the research group. 8,9,11 A prescription report disclosed a GP's prescribing practice for the 13 criteria at baseline (NorPD-data from the year prior to the intervention) and as compared with average figures for all participants. During the two CME-group sessions, the tutor presented the evidence for the various criteria (Box 1) and facilitated a structured discussion between the GPs. Based on their individual feedback reports, this discussion aimed to identify individual targets for improvements. 9 Outcomes were based on NorPD-prescription data from the 1-year period after the intervention, as compared with baseline figures and corresponding figures from the control group. 9 This study explored the outcome measure (that is, the number of PIPs per 100 prescriptions) for the seven following subgroups: GPs' age, sex, specialty in general practice or not, setting in a single-handed or group practice, urban or rural practice location, GPs' prevalence of PIPs at baseline relative to prescriptions issued for older patients, and, finally, GPs' baseline prescribing activity for older patients. The latter defined baseline as annual number of prescriptions per older patient who was issued any prescriptions.

Statistical methods
The overall and subgroup prevalence of PIPs at pre-and post-intervention were estimated as proportions of all prescriptions at pre-and post-intervention respectively. The prevalence of PIPs at post-intervention was compared to the prevalence of PIPs at pre-intervention using the relative Box 1. Thirteen explicit criteria for PIPs used for assessing the appropriateness of GPs' prescriptions to older patients (!70 years).

Single drugs
Combination of drugs risk (RR) with 95% CIs. An RR of 1.0 indicates identical prevalence of PIPs at pre-and post-intervention. An RR >1.0 indicates an increase of the prevalence of PIPs at post-intervention relative to preintervention, while a RR <1.0 indicates a decrease of the prevalence of PIPs at post-intervention compared to pre-intervention. The Poisson model is usually considered the basic model for modeling counts and rates. In this cluster randomised trial, each CME group defined a cluster. Therefore, data on rates of PIPs defined as the number of PIPs per 100 prescriptions were analysed using the Poisson regression model extended by including CME group random effects to account for data clustering. Estimates of IRRs with 95% CIs were obtained from the Poisson multilevel models and represent the change in PIPs per 100 prescriptions at post-intervention relative to baseline. If the IRR is <1.0, then a reduction in PIPs was observed, and if it is >1.0, then an increase in PIPs was observed. When the IRR = 1.0, then there was no change in PIPs per 100 prescriptions between pre-and post-intervention. Two steps preceded the modeling of PIPs using the Poisson model. The significance level used was P<0.01.
First, a non-stratified model was fitted to investigate the rate of change of PIPs in the intervention group versus the control group. Using this model, it was possible to investigate the rate of change of PIPs between pre-and post-intervention within the control and intervention groups. The model was adjusted for the following GP characteristics: age, sex, specialisation, type and setting of the practice, baseline values of mean PIPs per 100 prescriptions, and number of prescriptions per older patient. The GPs' age, their annual baseline figures for PIP prevalence, and average number of prescriptions per older patient were split into quintiles by numbers of GPs.
Secondly, the changes in prescription rates of PIPs in intervention and control groups by the seven subgroups of GPs' characteristics were compared. Assuming no difference in intervention effect between the seven independent subgroups, the probability of at least one significant subgroup is given by 1 À (1 À a)7 = 0.30 for a = 0.05. The overall type I error rate was controlled conservatively by using a * = a / 7 = 0.007 in each of the seven subgroup analyses. All analyses were performed in StataSE (version 13).

Results
Summary data showing the prevalence of PIPs pre-and post-intervention are given in Table 1. After the intervention, there were 19 754 PIPs, which accounted for 1.79% of total 1 104 391 prescriptions issued by 449 GPs to patients aged !70 years. The results show a 25% reduction in PIPs after intervention in the intervention group, and a 13% reduction in the control group.

Overall effect of intervening, adjusted for GPs characteristics
IRRs from the Poisson multilevel models represent the change in PIPs per 100 prescriptions at postintervention relative to baseline. In Table 2, the rate of change of PIPs in the intervention group versus the control group is given after adjusting for sub-group characteristics of the GPs. There was no significant difference in PIPs at baseline between the control group and the intervention group, (IRR = 1.  Table 2. Compared to GPs aged 27-42 years, the change in PIP rates among GPs in the age groups 43-48 and 49-52 years was significantly lower, by 15% and 8%, respectively. However, a relatively higher rate of PIPs post-intervention was observed among GPs in the age group 57-68 years, compared to GPs aged 27-42 years. The analysis also showed that GPs who had more PIPs per 100 prescriptions during the 1-year pre-intervention period (baseline), had more PIPs per 100 prescriptions post-intervention. The rate of PIPs per 100 prescriptions was 40% higher among male GPs compared to female GPs, and GPs in urban areas had significantly higher PIPs than peers in rural areas (see Table 2). Table 3 shows that improvements in prescribing were highest among GPs aged 57-68 years, who were specialists, who worked in single-handed practices, who had 2.4 to 2.9 PIPs per 100 prescription at baseline, and among GPs with !15 prescriptions per older patient at baseline.

Sub-group analyses of intervention effect
When the interaction between the intervention and each sub-group was tested, the overall tests were significant (P<0.01); hence, the results of the subgroups with a * set at 0.007 are reported. Reductions of PIPs were observed in all GPs aged !43 years in both the control and the intervention group (P<0.007). However, a 12% significant increase of PIPs was observed among GPs aged 27-42 years in the control group (P = 0.004). PIPs decreased by 20% among specialist GPs in the intervention group and 9% among specialist GPs in the control group. Although a significant reduction (16%; P<0.007) of PIPs was observed among non-specialist GPs in the intervention group, changes of PIPs among non-specialist GPs in the control group were not significant.
In the intervention arm, a 25% reduction of PIPs was observed among GPs in single-handed practices compared to a 20% reduction of PIPs among GPs in group practices (P<0.007). The results for GPs from single-handed practices in the control arm were not significant (P = 0.034), as shown in Table 3. The analysis also showed significant improvements among GPs in the intervention arm, regardless of the mean number of PIPs per 100 prescriptions at baseline. The largest improvement in this group was observed among GPs who had between 2.4 and 2.9 PIPs per 100 prescriptions at Table 2. The rate of change in PIPs in the intervention group versus the control group after adjusting for the GP subgroup characteristics. Estimates of IRRs showing the changes in PIPs per 100 prescriptions at post-intervention relative to 1 year pre-intervention obtained from the Poisson cluster effects regression model. baseline. However, the findings in the control group were not consistent. For example, significant improvements were observed among GPs who had !2.1 PIPs per 100 prescriptions at baseline. Although not statistically significant, an increase of PIPs at post-intervention was observed among GPs with <2.1 PIPs per 100 prescriptions at baseline. Significant reductions of PIPs were observed in the other subgroups in the intervention arm with the exception of GPs who had <8.7 prescriptions per older patient at baseline (see Table 3).

Discussion Summary
This study investigated whether GP characteristics influenced rates of PIPs after an educational intervention. It was found that improvements in prescribing were highest among GPs aged 57-68 years, Table 3. Subgroup analyses comparing changes in prescription rates of PIPs by the seven GP-related characteristics. Unadjusted IRRs and their 95% CIs obtained from the Poisson regression model with cluster effects at CME group level showing the changes in PIPs within the subgroups, a = 0.007. who were specialists, who worked in single-handed practices, who had 2.4 to 2.9 PIPs per 100 prescription at baseline, and among GPs with !15 prescriptions per older patient at baseline. The educational intervention included an exposure of each GP's prescribing profile in a CME group, which may have strengthened their motivation for change. Yet another explanation for a greater reduction in PIPs in patients with a high number of PIPs at baseline is that there were more changes to be made in these patients.

Subgroups
A possible explanation for the high reduction rate of PIPs among the oldest GPs could be that this group more commonly prescribes 'older' drugs, even if newer and possibly safer alternatives are available. Several of the drugs used to measure inappropriateness in the Rx-PAD study had been in the market for decades. The Rx-PAD study demonstrated that being a young doctor was associated with a lower proportion of PIPs at baseline, while their older colleagues were more likely to prescribe inappropriately. 8 That GPs in certain age groups and GPs with a high rate of PIPs at baseline improved most due to the intervention may suggest that future interventions could possibly be more cost effective if they are tailored towards a subgroup of GPs instead of targeting all. In a study analysing factors associated with adolescents non-response to an asthma management programme, Joseph et al 15 argue in favour of subgroup analyses to identify differences in baseline characteristics associated with the primary outcome in a randomised trial. 10 This may be particularly useful if such characteristics are used as basis for creating submodules within the trial through which additional, theory-based strategies could be administrated. 10

Strength and limitations
To the best of the authors' knowledge, this is the first GP-based intervention study to investigate the influence of GPs' characteristics on their change of prescribing practice for older patients across a broad spectrum of therapeutic areas. The Rx-PAD study was a large-scale intervention including about one-tenth of all Norwegian GPs (there were around 4100 GPs in Norway at the time of the study). This strengthens the external validity of the current results. The full model results from the Rx-PAD study are robust, with high power and a cluster randomised design.
Since the GPs recruited in an RCT are not a homogeneous sample, it is not surprising that their responses to an educational intervention vary in many ways. Thus, it may be difficult to apply the results of an educational intervention to individual GPs without considering their individual characteristics. However, results from subgroup analyses are usually unreliable because of reduction in statistical power, which may lead to incorrect conclusions. By formally testing the interactions between the intervention and the subgroups, in addition to using a strict P-value of 0.007, this study managed to control for multiple testing, hence reducing false negatives and positives. That the GPs in the control group also improved their prescribing practice for older patients may partly represent a Hawthorne effect in response to the fact that they knew they were studied. 16 Adding to this is the fact that the control group was allocated to another educational intervention with focus on prescribing quality, although within another therapeutic field (that is, antibiotic use for respiratory tract infections). 12 This may possibly have had effects on their overall prescribing quality.
Despite the age of the data, the results of this study are still relevant; firstly, because the continuous need for improved prescription quality for older people is quite similar today as it was 10 years ago. Secondly, and most important, the focus in this study was to explore how that intervention influenced prescribing patterns among GPs, related to individual and practice setting characteristics.

Comparison with existing literature
To the authors' knowledge, few if any studies with a comparable size and design have explored the intervention effects by variables related to the GPs (age, sex, and specialist status) and their practices (single-handed versus group, and urban versus rural), and baseline performance regarding the target of an intervention. The effects of audit and feedback on prescribing of drugs are reported to be small to moderate, and the relative effectiveness is likely to be greater when baseline adherence to recommended practice is low. 17 This is consistent with this study's findings.

Implications for research
In the control arm of the Rx-PAD study, inappropriate antibiotic prescribing practice strongly correlated with practice activity in terms of annual number of encounters. 12,18 In addition to the explanatory variables used in the present report, it would therefore also have been interesting to correlate GPs' prescribing patterns for older patients to their practice activity in terms of annual patient encounters. However, such data were not recorded for the present arm of the Rx-PAD study. For future research, the inclusion of explanatory variables that also may reflect practice activity more directly are recommended, as well as the amount of time spent per patient in the consultation room.

Funding
The study was carried out with grants from the Norwegian Medical Association, the Norwegian Directorate of Health, and the Research Council of Norway.

Ethical approval
Extraction of data from NorPD was based on written, informed consent from all physicians. The Regional Committee for Research Ethics and the Norwegian Social Science Data Service approved the project in October 2005 (reference 200500838 SM/RH). The Directorate for Health accepted a dispensation from the Health-Professional Secrecy regulations. Data for both patients and GPs were anonymised before analyses.

Provenance
Freely submitted; externally peer reviewed.