Variations in general practitioners’ prescribing patterns ... · prescribing were observed for...
1
20 60 100 140 180 Variations in general practitioners’ prescribing patterns in Ireland: preliminary results Motterlini N 1 , Dimitrov BD 1 , Bradley M 1 , Bennett K 1,2 , Fahey T 1 . 1 HRB Centre for Primary Care Research, Department of General Practice, Royal College of Surgeons in Ireland, Republic of Ireland. 2 Department of Pharmacology and Therapeutics, Trinity College Dublin, Republic of Ireland. Background Evidence concerning variation in clinical practice has been accruing in the US with the advent of the Dartmouth Atlas Health Project (www.dartmouthatlas.org). Information on practice variation raises important issues concerning efficacy, effectiveness and quality of healthcare delivered in national and international contexts. 1 Objectives This study was aimed at quantifying the variations in prescribing patterns of common drugs amongst general practitioners (GPs) in Ireland using the Health Service Executive Primary Care Reimbursement (HSE-PCRS) prescribing database. Methods Study population Data was obtained from the Irish General Medical Services (GMS) scheme pharmacy claims database from the HSE-PCRS. Thirty percent of the Irish population is covered under the GMS scheme. The scheme is means tested and provides free health services to those who are unable to afford them, thus it over-represents socially- deprived populations. The GMS database contains routinely collected data from pharmacy claims for dispensed medications. All prescription items are coded using WHO’s Anatomical Therapeutic Chemical (ATC) classification. Basic demographic information (age and sex) are recorded. No information on diagnosis or disease condition is available. We analyzed all patients aged ≥15 years in the HSE-PCRS database between January and December 2010. First, we selected the ATC broad classes (e.g. J01, N02) that were most frequently prescribed, and second, we extracted the drug category with the highest prevalence of prescriptions within each class. Data analysis Prescribing prevalence is composed of the number of individuals receiving at least one prescription for a particular drug class (the numerator) and the number of GMS eligible population data for that year (the denominator). For the calculation of annual prescription prevalence, we divide the numerator by the denominator and report per 1,000 population. Age-sex standardised prescribing ratios (SPRs) were determined for each GP. The SPR is a measure of the extent to which the number of patients on a particular drug is above average, average or below the group norm, taking account of age and sex. An SPR of 100 is average for the group of GPs being studied, an SPR of 200 is twice the average and an SPR of 50 is half the average. 2 The variation in SPRs between GPs was measured and expressed by extremal quotient (EQ), coefficient of variation (CV%) and systematic component of variation (SCV) which represents the variation considered systematic, thus beyond chance. 3 The larger the values of these measures, the higher the variability in prescribing among GPs. Figure 1: Prescribing prevalence per 1000 GMS population for drug category (2010) Results Descriptive statistics During the study period, 2,261 GPs prescribed at least one medication for over 80% of the GMS eligible population (n=1,359,561). Second line antibiotics (384/1000 GMS population) demonstrated the highest prescribing prevalence, while inhaled bronchodilators showed the lowest prevalence (112/1000 GMS population) among the drug categories analysed (Figure 1). Practice variation A consistency between the reported measures according to their relative magnitude of the variations was observed (Table 1). In terms of SCV, the greatest variation in prescribing were observed for paracetamol (6.82) and first line antibiotics (5.11). Proton pump inhibitors and platelet aggregation inhibitors had the lowest variations in prescribing (1.39 and 0.70, respectively) among the drug categories selected. Table 1: Variation statistics for drug category (2010) The distribution graphs in Figure 2 show the variation in SPRs across GPs for each drug category. Each blue dot represents one GP. The chart summarises two features of the data. The first is a measure of dispersion: the SPR for the highest GP is up to four times higher than the lowest GP, suggesting substantial variation. Secondly, the distribution graph shows that the variation is caused by several outliers, indicating that the variation is persistent and widespread across the country. Figure 2: Age-sex standardised prescribing ratios (SPRs) for drug category (2010) Conclusions Preliminary analysis in relation to variation in GP prescription patterns in Ireland can provide very important and valuable information on practice variation and help prioritise future research studies to improve the quality of prescribing. The observed variation for paracetamol and 1 st line antibiotics prescribing is most likely due to variability in clinical indication (pain and infection). Lower variability for statins and platelet inhibitors reflects firmer clinical indication, for instance in secondary prevention of cardiovascular disease. Acknowledgements: We thank the HSE-PCRS and the Health Intelligence Unit for supplying the data on which the study is based. Funding: This research was supported by the Health Research Board of Ireland through the HRB Centre for Primary Care Research. Conflict of Interest Statement: The authors declare no conflict of interest. Corresponding author: Nicola Motterlini, HRB Centre for Primary Care Research, RCSI Ireland, [email protected] References 1. Wennberg JE. Time to tackle unwarranted variations in practice. BMJ 2011; 342:687-690. 2. Johnson Z, Hayes C, Dack P. The standardized prescribing ratio - a new method for comparing prescribing between GPs, controlling for patient age and sex. Pharmacoepidemiology and Drug Safety 1997; 6 (5): 337-345. 3. McPherson K, Wennberg JE, Hovind OB, Clifford P. Small-area variations in the use of common surgical procedures: an international comparison of New England, England, and Norway. NEJM 1982; 307:1310-4. ATC class Drug Category EQ CV SCV N02 PARACETAMOL 4.40 30.9% 6.82 J01 1 st LINE ANTIBIOTICS 3.76 26.8% 5.11 R03 INHALED BRONCHODILATORS (BETA-2 AGONISTS) 4.05 29.3% 4.96 N05 BENZODIAZEPINES 3.65 26.4% 4.84 J01 2 nd LINE ANTIBIOTICS 3.19 23.4% 3.82 M01 NSAIDs 3.22 22.0% 2.88 C10 STATINS 2.55 18.8% 1.54 A02 PROTON PUMP INHIBITORS 2.61 17.8% 1.39 B01 PLATELET INHIBITORS 2.47 17.9% 0.70 0 50 100 150 200 250 300 350 400 Prevalence per 1000 individuals Age-sex SPR 2 nd line Antibiotics NSAIDs Proton Pump Inhibitors 1 st line Antibiotics Paracetamol Benzodiazepines Statins Platelet Inhibitors Inhaled Bronchodilators (Beta-2 Agonist) Paracetamol 1 st line Antibiotics Inhaled Bronchodilators (Beta-2 Agonist) Benzodiazepines 2 nd line Antibiotics NSAIDs Statins Proton Pump Inhibitors Platelet Inhibitors 100 140 180 20 60
Transcript of Variations in general practitioners’ prescribing patterns ... · prescribing were observed for...
2060
100
140
180
Variations in general practitioners’ prescribing patterns in Ireland: preliminary resultsMotterlini N1, Dimitrov BD1, Bradley M1, Bennett K1,2, Fahey T1.
1HRB Centre for Primary Care Research, Department of General Practice, Royal College of Surgeons in Ireland, Republic of Ireland.2Department of Pharmacology and Therapeutics, Trinity College Dublin, Republic of Ireland.
BackgroundEvidence concerning variation in clinical practice has been accruing in the US withthe advent of the Dartmouth Atlas Health Project (www.dartmouthatlas.org).Information on practice variation raises important issues concerning efficacy,effectiveness and quality of healthcare delivered in national and internationalcontexts.1
ObjectivesThis study was aimed at quantifying the variations in prescribing patterns ofcommon drugs amongst general practitioners (GPs) in Ireland using the HealthService Executive Primary Care Reimbursement (HSE-PCRS) prescribing database.
MethodsStudy populationData was obtained from the Irish General Medical Services (GMS) scheme pharmacyclaims database from the HSE-PCRS. Thirty percent of the Irish population is coveredunder the GMS scheme. The scheme is means tested and provides free healthservices to those who are unable to afford them, thus it over-represents socially-deprived populations. The GMS database contains routinely collected data frompharmacy claims for dispensed medications. All prescription items are coded usingWHO’s Anatomical Therapeutic Chemical (ATC) classification. Basic demographicinformation (age and sex) are recorded. No information on diagnosis or diseasecondition is available.
We analyzed all patients aged ≥15 years in the HSE-PCRS database between Januaryand December 2010. First, we selected the ATC broad classes (e.g. J01, N02) thatwere most frequently prescribed, and second, we extracted the drug category withthe highest prevalence of prescriptions within each class.
Data analysisPrescribing prevalence is composed of the number of individuals receiving at leastone prescription for a particular drug class (the numerator) and the number of GMSeligible population data for that year (the denominator). For the calculation ofannual prescription prevalence, we divide the numerator by the denominator andreport per 1,000 population.
Age-sex standardised prescribing ratios (SPRs) were determined for each GP. TheSPR is a measure of the extent to which the number of patients on a particular drugis above average, average or below the group norm, taking account of age and sex.An SPR of 100 is average for the group of GPs being studied, an SPR of 200 is twicethe average and an SPR of 50 is half the average.2
The variation in SPRs between GPs was measured and expressed by extremalquotient (EQ), coefficient of variation (CV%) and systematic component of variation(SCV) which represents the variation considered systematic, thus beyond chance.3
The larger the values of these measures, the higher the variability in prescribingamong GPs.
Figure 1: Prescribing prevalence per 1000 GMS population for drug category (2010)
ResultsDescriptive statisticsDuring the study period, 2,261 GPs prescribed at least one medication for over 80%of the GMS eligible population (n=1,359,561).Second line antibiotics (384/1000 GMS population) demonstrated the highestprescribing prevalence, while inhaled bronchodilators showed the lowest prevalence(112/1000 GMS population) among the drug categories analysed (Figure 1).
Practice variationA consistency between the reported measures according to their relative magnitudeof the variations was observed (Table 1). In terms of SCV, the greatest variation inprescribing were observed for paracetamol (6.82) and first line antibiotics (5.11).Proton pump inhibitors and platelet aggregation inhibitors had the lowest variationsin prescribing (1.39 and 0.70, respectively) among the drug categories selected.
Table 1: Variation statistics for drug category (2010)
The distribution graphs in Figure 2 show the variation in SPRs across GPs for eachdrug category. Each blue dot represents one GP. The chart summarises two featuresof the data. The first is a measure of dispersion: the SPR for the highest GP is up tofour times higher than the lowest GP, suggesting substantial variation. Secondly, thedistribution graph shows that the variation is caused by several outliers, indicatingthat the variation is persistent and widespread across the country.
Figure 2: Age-sex standardised prescribing ratios (SPRs) for drug category (2010)
ConclusionsPreliminary analysis in relation to variation in GP prescription patterns in Ireland canprovide very important and valuable information on practice variation and helpprioritise future research studies to improve the quality of prescribing.The observed variation for paracetamol and 1st line antibiotics prescribing is mostlikely due to variability in clinical indication (pain and infection). Lower variability forstatins and platelet inhibitors reflects firmer clinical indication, for instance insecondary prevention of cardiovascular disease.
Acknowledgements: We thank the HSE-PCRS and the Health Intelligence Unit forsupplying the data on which the study is based.
Funding: This research was supported by the Health Research Board of Irelandthrough the HRB Centre for Primary Care Research.
Conflict of Interest Statement: The authors declare no conflict of interest.
Corresponding author: Nicola Motterlini, HRB Centre for Primary Care Research,RCSI Ireland, [email protected]
References1. Wennberg JE. Time to tackle unwarranted variations in practice. BMJ 2011; 342:687-690.2. Johnson Z, Hayes C, Dack P. The standardized prescribing ratio - a new method for comparing
prescribing between GPs, controlling for patient age and sex. Pharmacoepidemiology and DrugSafety 1997; 6 (5): 337-345.
3. McPherson K, Wennberg JE, Hovind OB, Clifford P. Small-area variations in the use of commonsurgical procedures: an international comparison of New England, England, and Norway. NEJM1982; 307:1310-4.
ATC class Drug Category EQ CV SCV
N02 PARACETAMOL 4.40 30.9% 6.82
J01 1st LINE ANTIBIOTICS 3.76 26.8% 5.11
R03 INHALED BRONCHODILATORS (BETA-2 AGONISTS) 4.05 29.3% 4.96
N05 BENZODIAZEPINES 3.65 26.4% 4.84
J01 2nd LINE ANTIBIOTICS 3.19 23.4% 3.82
M01 NSAIDs 3.22 22.0% 2.88
C10 STATINS 2.55 18.8% 1.54
A02 PROTON PUMP INHIBITORS 2.61 17.8% 1.39
B01 PLATELET INHIBITORS 2.47 17.9% 0.70
0
50
100
150
200
250
300
350
400
Pre
vale
nce
per
10
00
ind
ivid
ual
s
Age
-sex
SP
R
2nd line Antibiotics
NSAIDs ProtonPump
Inhibitors
1st line Antibiotics
Paracetamol
Benzodiazepines
Statins PlateletInhibitors
InhaledBronchodilators(Beta-2 Agonist)
Paracetamol
1st line Antibiotics
InhaledBronchodilators(Beta-2 Agonist)
Benzodiazepines
2nd line Antibiotics
NSAIDs
Statins
ProtonPump
Inhibitors
PlateletInhibitors
10
01
40
18
02
06
0
