Appropriate prescribing in asthma and its related cost in east LondonBMJ 1995; 310 doi: https://doi.org/10.1136/bmj.310.6972.97 (Published 14 January 1995) Cite this as: BMJ 1995;310:97
- Jeannette Naish, general practitionera,
- Patricia Sturdy, research officera,
- Peter Toon, general practitionera
- a City and East London General Practice Database Project, Department of General Practice and Primary Care, Medical College of St Bartholomew's and the London Hospitals, Basic Medical Sciences, Queen Mary and Westfield College, London E1 4NS
- Correspondence to: Dr Naish.
- Accepted 25 November 1994
Objectives: To determine the patterns of preventive to reactive prescribing for asthma among general practices in the City and East London Family Health Services Authority area and their relation to prescribing cost.
Design: Descriptive study of asthma prescribing during April 1992 to March 1993. Prescribing data were linked with general practice and population data on one database.
Setting: City and East London Family Health Services Authority area, including all general practices in contract with the authority, which covers the inner city London Boroughs of Hackney, Tower Hamlets, and Newham and the Corporation of the City of London.
Subjects: All 163 general practices as at 1 June 1993.
Main outcome measures: Ratios of prescribed inhaled corticosteroids plus cromoglycates (prophylactic treatment) to bronchodilators; distribution of the cost of asthma prescribing; distribution of overall generic prescribing; proportion of asthma generic prescribing; distribution of cost of overall drugs prescribed per prescribing unit.
Results: Practices approved for band 3 health promotion or asthma surveillance and those with a general practitioner trainer had on average higher ratios of prophylactic to bronchodilator treatment and significantly higher asthma drug costs than other practices. Those practices with high levels of overall generic prescribing had significantly higher prophylactic to bronchodilator ratios than thosewith lower levels of generic prescribing. Practices with higher levels of asthma drug generic prescribing also had significantly higher prophylactic prescribing. However, the proportion of generically prescribed asthma drugs was lower than overall generic prescribing. There was no correlation between the ratio of prophylactic to bronchodilator asthma prescribing and the proportion of overall drugs expenditure, but high spending practices spent significantly more on asthma drugs.
Conclusions: Pressure to reduce the cost of asthma prescribing may lead to a lowering of the ratio of prophylactic to bronchodilator treatments. However, reducing prophylactic prescribing would run contrary to the British Thoracic Society guidelines and might worsen the quality of asthma care.
Practices approved for vocational training, asthma surveillance, or band 3 health promotion show a higher level of prophylactic prescribing
A higher level of prophylactic prescribing suggests that these practices are fo
In asthma prescribing good clinical practice and cost consciousness do not seem to run together
Pressure to reduce the costs of asthma prescribing may lead to a reduction in prophylactic treatment and lower the quality of asthma care
General practice prescribing costs are high and rising. The indicative prescribing scheme seeks to control them and increase cost effectiveness.1 There is, however, little consensus on what constitutes good prescribing and how to measure it. Though overall cost, even for specific drugs or categories, shows wide variation, it is too crude a measure. Low costs may reflect commendable parsimony or failure to diagnose. They are affected by morbidity in the population served. Prescribing may save money in other aspects, such as consultation time, while prescribing may reduce other health care costs—for example, when a therapeutic trial of treatment is used instead of investigation or an expensive drug avoids hospital treatment. Generic prescribing is often promoted as good practice1 2 but it is not clear how it relates to cost effectiveness.
Baker and Klein studied the effect of population characteristics and practice organisation on general practice activities between family health services authorities,3 but this masks substantial variations between practices. We studied factors related to variation in prescribing costs between individual practices within one family health services authority.
We chose to look firstly at asthma. Asthma has a high prevalence, particularly in our area,4 and is one of the few conditions in which there is wide consensus on what constitutes good practice.5 Several studies have suggested that prescribing could be improved by increasing the ratio of prophylactic (inhaled corticosteroids and cromoglycates) to bronchodilator drugs prescribed.6 7 8 9 10 This could decrease morbidity and hospital admission rates. In our area admission rates for asthma are 80% above the national rates.11
Jones investigated 59 practices with an interest in asthma and found that though their average total prescribing costs were lower than their family health services authority average, their respiratory drug costs were higher.12 McGavock found that high spending urban practices prescribed proportionately more systematic (prophylactic) drugs for asthma than low spending practices.13 There are no general practice studies of the generic prescribing of asthma drugs.
We decided to look at the relation between overall drug costs, overall asthma drug costs, generic prescribing rates, and prophylactic to bronchodilator ratio. We also related these to other factors which have face validity as likely to be markers for good prescribing in this area—namely, approval for asthma surveillance, band 3 health promotion, and vocational training.
Subjects and methods
Applying a method analogous to that of Baker and Klein3 to individual practices, the City and East London General Practice Database allows study of factors which influence general practice activity. It uses routinely collected practice based information from the family health services authority, Prescription Pricing Authority, Office of Population Censuses and Surveys, and the district health authority. Issues of confidentiality and data security were negotiated with the local medical committee and City and East London Family Health Services Authority.
Data on general practitioner characteristics, practice resources, andpractice populations for all 163 practices in east London (as at 1 June 1993) were collated on one database. Included in these data was health promotion banding, approval for asthma surveillance (after July 1993), and the presence of a general practitioner trainer (as at 1 October 1993). The source of these data was family health services authority files. Prescribing variables were obtained from general practitioners' prescribing analyses, and cost was then linked to the dataset and analysed to investigate the relations, if any, through statistical analysis. Practice data were compiled on a Foxpro 2 database and analysed with the SPSS-PC.
The number of practice prescribing units is the denominator. This figure is the number of patients aged under 65 plus three times those aged over 65.14 Other variables were the number of items prescribed, their net ingredient cost, and the percentage of items prescribed generically. These variables were included for all drugs and for asthma drugs15 for the period April 1992 to March 1993. Data were collected for the 163 practices in the City and East London Family Health Services Authority area but those with incomplete datasets were excluded from some calculations; the number analysed ranged from 155 to 158 practices. There are no dispensing practices in the city and east London area.
Analyses were based on the ratio of prophylactic (inhaled corticosteroids (BNF, ch 3.2)15 plus cromoglycates (BNF, ch 3.3))15 to bronchodilator drugs (BNF, ch 3.1)15 for asthma treatment and the net ingredient cost of these asthma drugs per practice prescribing unit.
The distribution of the prophylactic to bronchodilator ratios was found to be sufficiently normal for use of an independent samples t test on the raw data. However, the net ingredient cost of asthma drugs per prescribing unit was highly positively skewed, and in order to satisfy the requirements for a t test logs were taken before the test was carried out. The relation between continuous variables was investigatedby means of Pearson's correlation coefficient and Spearman's rank correlation coefficient. Because there were several outliers and because of the problems associated with analysing percentage data Spearman's rank correlation coefficient was calculated to confirm Pearson's correlation coefficient. The median was used as the measure of central tendency to display the results in the tables.
The three measurement variables used are listed in table I. The prophylactic to bronchodilator ratios had a near normal distribution but the unit cost for asthma drugs was highly positively skewed. Table II shows that the unit cost of asthma prescribing was significantly higher in practices with a general practitioner trainer (P<0.001), practices approved for band 3 health promotion (P<0.001), and those approved for asthma surveillance (P=0.007). These practices also had significantly higher prophylactic to bronchodilator item ratios than their counterparts. Band 3 and training practices had significantly higher cost ratios (P<0.001).
Practices with a higher level of overall generic prescribing showed a higher unit cost for asthma drugs (table III). The percentage of overall generic prescribing was positively correlated with prophylactic bronchodilator ratios for items and costs (P<0.001). There was also a positive correlation between the percentage of overall generic prescribing and the generic prescribing of asthma drugs, giving a Pearson coefficient of 0.52 (P<0.001).
Correlation coefficients were calculated from individual practice data.
The mean percentage of generic prescribing for asthma drugs was 19.3%, as compared with 44.5% for overall drugs. The mean percentage of generic prescribing for each group of asthma drugs was as follows: cromoglycates 8.6%; corticosteroids 7.3%; bronchodilators 23.7%. Table IV shows a positive relation between the percentage of generic prescribing for asthma drugs and the prophylactic to bronchodilator ratios for items and costs (P<0.001). The net ingredient cost of asthma drugs per prescribing unit increased with the percentage of generic prescribing of asthma drugs (P=0.001).
There were no correlations between the percentages of overall drug expenditure and the asthma drug prescribing ratios for costs and items (P=0.12 and P=0.12 respectively). Table V shows that high spending practices spent more on asthma drugs per prescribing unit than those with lower levels of expenditure.
Our findings must be interpreted with caution. Though prescribing analyses and cost data derived from prescriptions dispensed and paid for are likely to be the most reliable available, neither items dispensed nor net ingredient cost is an ideal measure.16 An “item” is a variable quantity of medication, and the choice of drug and whether it is prescribed generically will influence net ingredient cost. As in asthma prescribed dosage may vary widely, and compliance could be a problem,10 even if we had an accurate measure of the quantity of drug prescribed—for example, by defined daily dosage—this would reflect patient morbidity and attitudes to treatment as well as prescribing behaviour.
Nor is the practice prescribing unit an ideal denominator. It has been criticised as too crude an attempt to reflect demographic factors affecting morbidity.17 The practice list size information from which it is derived is often inaccurate, particularly in inner city areas with high population morbidity. Higher prescribing rates may reflect “cleaner” lists owing to better practice organisation, which reduce the denominator, as much more prescribing increases the numerator. Our use of a prescribing ratio cancels out some but not all of these sources of inaccuracy.
Though the individual doctor would be the most logical unit for studying prescribing behaviour, the practice is the most realistic. Few practices can separate the prescribing of individual doctors within the group. Despite these difficulties these data indicate interesting features both of asthma prescribing in particular and of the use of prescribing data in general.
Practices approved for vocational training, asthma surveillance, or band 3 health promotion show a higher ratio of prophylactic to bronchodilator prescribing, suggesting that they are following British Thoracic Society guidelines more closely (table II).5 The congruence of these factors suggests that they are indeed related to “good prescribing” in this area, though clearly they are not synonymous with it. However, the unit cost for asthma drugs is significantly higher in these practices, supporting Jones's view that higher expenditure in this area reflected good rather than bad practice.12
The higher the level of overall generic prescribing and generic prescribing of asthma drugs the higher were the unit cost of asthma drugs and the prophylactic to bronchodilator ratios (tables III and IV). This suggests that the more cost conscious practices prescribe more appropriately. However, the percentage of generic prescribing of all asthma drugs (19.3%) was far lower than the percentage of overall generic prescribing (44.5%). The Audit Commission has highlighted the low level of generic prescribing of bronchodilators nationally,1 but we found that generic prescribing of corticosteroids and cromoglycates—both under 10%—was even lower than that ofbronchodilators (23.7%). This may reflect differences in the ease of writing proprietary rather than generic names, perception of lack of generic equivalents, or merely habit. Clearly generic prescribing is not a valid measure of good quality prescribing, though it may be desirable to avoid waste.
Good clinical practice and cost consciousness do not seem necessarily to go together. Pressure to reduce drug costs therefore may not increase the cost effectiveness of prescribing; indeed, it may do the opposite. Our study discloses a need for a more sophisticated approach to improving the quality of prescribing, taking into account the many factors which influence prescribing decisions in general practice. We intend to continue to use the general practice database—now expanded to include population variables from census data—to develop our understanding of these factors.
We thank the staff at the City and East London Family Health Services Authority for providing data. We also thank the Prescribing Pricing Authority for the data that was obtained by Jeannette Naish in her capacity as medical adviser to the family health services authority. Dr Chris Griffiths was most helpful with his constructive criticism. Enid Hennessy, of the department of epidemiology and medical statistics, gave valuable statistical advice. The City and East London General Practice Database project is funded by the City and East London Family Health Services Authority and the former North East Thames Regional Health Authority.