Due to growing international concern about the quality of prescribing in primary care, researchers and policy makers have made interventional strategies to improve prescribing. Drug expenditures are big burden and threaten of health care budgets. It is challenging task to improve prescribing pattern in medical practice.
The prescribing habit by general practitioner is a complex activity and depends on the interplay of many factors. In recent decades, these factors have been shown to influence family physician prescribing pattern (1).( 1 effect of advertising) Studying factors associated with family physicians' prescribing is considered to be of high value since high percentage of drugs are prescribed by primary care physicians. These factors interact in non linear and unpredictable ways (2). (2 factors influencing GP Allan).
As result of various influences, prescribing habit changes of the individual physician usually occur slowly. (3)( drug prescription patterns Bjerrum).
For the scope of this review we classify factors associated with family physicians' prescribing into four categories. The first category includes factors related to physicians; age, sex, years of experience, and continuous medical education. The second category includes factors related practice settings; size of practice, number of patients, guidelines and use of drug formulary. The third category includes factors related to drugs: advertisement and cost. The fourth category includes factors related to patients; age, sex, comorbidity and multiple healthcare providers. Therefore, the aim of this review is to identify and to assess factors affecting family physician prescribing behaviour. The research question formulated to state, what are the factors affecting family physician prescribing behaviour.
In this literature review we chose broadly inclusive search strategy with two stages. In stage one, a search has been conducted using the following key words: "prescription", "prescribing", "prescribing patterns" prescribing attitudes", "prescribing factors", "prescribing indicators", primary care prescribing" and "GP prescribing". In second stage , after figuring out the factors associated with physicians prescribing patterns based on first stage, a second search has been conducted using the following terms "guidelines and prescribing", "drug cost", "drug advertising", " drug formulary", " polypharmacy", repeat prescriptions" and "new drug". We searched midline, Pub med and Eric from 1990 to 2009. From articles satisfying preliminary inclusion criteria, the reference lists were reviewed. Based on the initial review our final inclusion and exclusion criteria were determined.
The inclusion criteria are: a) articles assessed prescribing in primary care setting, b) article retrieving information from prescription database. We excluded articles written in language other than English, studies that assessed specific drug group or drugs for specific disease. There were no geographic limitations.
We found 31 studies that met our criteria. All studies used a database for data collection, 12 and 13 prospective and retrospective studies respectively (table 1).
Factors related to physician:
It has been found that there is a significant relation between certain primary care physician characteristics and their prescribing behavior. Younger primary care physicians have higher rates of new drug utilization. Female sex and recent graduation i.e. less years in practice are associated with high drug utilization rates (4,5,6,7,8,9) (1, 4,18, 19,31,32).One study showed that no influence of physician age or number of years in practice on polypharmacy in particular (6) (18). High prescribers did not differ significantly from low prescribers in age, number of years in practice, mean practice size or patient age. (9)(32)
Gill et al has found no effects of physician's ethnicity and place of graduation on prescribing patterns (10).(2), However, two studies demonstrated that physicians who were foreign trained tend to have high prescribing rates and cost (9)(28,32) . Also medical school found to be a factor associated with higher new drug utilization (7)(19). Continuous medical education (CME) has an effect acceleration of new drug adoption (5)(4). An educational intervention programs improve prescribing patterns and may result in significant clinical benefits (11,12) (3, 25). It is also noticed that prescribing habits are influenced by scientific papers, specialist recommendations and meetings (3) (14). Financial incentive found to have a temporary effect on changing prescription behavior (13)(26). One study showed that guidelines had a little effect on antihypertensive drug use. (14)(13) Adoption of new drug is important prescribing factor. Among five drugs studied Steffensen et al found poor agreement between early, intermediate and late prescribers. Late prescribing was associated with female physicians.(5) (4)
Factors related to practice
There is a linear correlation between the number of prescribed drugs and number of general practitioners in the practice (15, 16) (12, 15). Physicians with large practice prescribed more drugs than those with small practice (4, 16)(1, 15). In terms of polypharmacy, one study showed 56% of prescribing variation between general practitioners could be explained by predictors related to practice structure, workload, clinical work profile and prescribing profile (17)(6).
It has been noticed that high workload practices tend to have a high prescribing rates.(4,18) (1,27), However in practices with large number of listed patients ,physicians prescribed fewer drugs per patient compared to practices with low number of listed patents.(6)(18) McCarthy et al found a significant correlation between the number of drugs prescribed and the number of physician working in the practice.(15) (12) The diffusion time of new drug after its release is longer in partnership practice compared to single handed practice, the mean diffusion times are 41 and 119 days for partnership and single handed practices respectively.(5) (4) Fee-for-service type of practice was considered to be associated with higher rates of new drug utilization. (7)(19).
Physicians practicing in rural areas and having high proportion of elderly have lower new drugs utilization rates than those practicing in urban areas. (7)(19).
Use of drug formulary and agreed verbal prescribing policy had no significant association with the number of drug prescribed. (15)(12). Computerized reminders have some effect on physician prescribing behavior.(19) (24)
Factors related to drugs
Physicians' interaction with drug industry found to start as early as medical school.(20) (373.26).It has been found that as many as 80% of GP's in both partnership and single handed practices had prescribed new drug 6 weeks and 21 weeks after its release respectively.(5) (4)
Tamblyn et al addressed that the new drugs have 8 to 17 fold differences in utilization rate, and were prescribed by 1.3%-22.3% of physicians. (7)(19). There is a linear relationship between polypharmacy and underprescribing. The higher the number of the drugs, the higher the estimated probability of underprescribing is.(21) (23). Provision of drug cost information in a computer based patient record system was found to have no effect on overall prescription drug cost to patients, however there was differences in individual drug classes. It also has been found that physicians are unfamiliar with the costs ofÂ medications they commonly prescribe . (22)(33) One study has indicated that a significant proportion in volume and costs is directly affected by hospital-initiated prescriptions. (9)(32) Repeat prescribing accounted for the vast majority of all items as well as prescribing costs. It accounted for 75% and 81% of all items and prescribing costs respectively. (23)(7)Among elderly patients, the average prescription was 99.4% per 100 general practitioner contacts; 72.1% were repeat prescriptions.(24) (17).
Ashly et al has found that prescribing and professional behavior appear to be affected by the present extent of physician-industry interactions.(25)(29) Reducing interactions between physicians and pharmaceutical sales representatives has resulted in improved prescribing.(8) (31)
Factors related to patients
The drug utilization rate increases with patient's age. Patient's age has more significant effect on drug utilization rate compared to patient's sex (26, 27) (11, 16). Repeat prescriptions significantly increase with patient's age. It has been found as high as 72 - 90% for patients aged 85 and over (23,24)(7,17). In terms of patient's sex, female patients were found to be given more drug items but less repeat prescription than male patients (27)(16). In elderly population, more than 60% of repeated prescribing was for female patients. (24) (17). Among patients aged 79 and under, female patients were prescribed to significantly more times than males (28)(9). Buck et al found that female sex was associated with potentially inappropriate medications (29)(22).
Patients with greater number of chronic conditions, multiple healthcare providers and multiple clinic visits have higher risk of developing polypharmacy and persistent polypharmacy (30) (8). Prescribing rates as well as costs increase with morbidity.(31) (10).
In this literature review we observed that physicians prescribing behaviors are affected linearly or inversely by many factors.
A significant relation has been found between certain physician characteristics and prescribing behavior. The findings that younger male physicians had higher prescribing rate may be related to a causal link between some physician characteristics, prescribing behavior and patient outcomes. It is not clearly known why sicker patients would seek young or male physicians, but these physicians may favor more aggressive treatment than female physicians and older colleagues.(4)(1) Higher rates of drug utilization among younger physicians may be related the propensity for aggressive intervention, more established prescribing behavior in older physicians or targeted marketing practices. (32,33)(19---47,48) The finding that male physician had higher rates of new drug utilization was supported by other studies.(34) (19---8) Female sex, small list size, lower diagnostic activity per patient and restrictive attitude toward pharmacotherapy fit into topology of conservative physicians. Supported by some studies, conservative physicians described as being light users of drugs. (35, 36) (4---9,19)
It is surprising that for those physicians who qualified from different countries, their ethnicity had no effect on their prescribing behavior.(10)(2) This could be related to secondary socialization which occurs in about 5 to 6 years. (37)(2---2) Socialization through postgraduate training and practicing in group practice changes prescribing behavior.(38) (2-----3)
There was no direct link between postgraduate training level and prescribing behavior. The level of postgraduate training can be a factor in determining how readily doctors accept commercial sources of prescribing information. Handouts from pharmaceutical companies were rated as very important or important sources of CME by significantly fewer certified members than non-certified members of the College of Family Physicians of Canada. (39)(presc by can----59) Interventional CME for treatment of chronic diseases for example bronchial asthma resulted in some improvement in prescribing behavior. (11)(3) CME and other social facts have been found accelerators for new drug adoption. (40, 41)(4---9,19) It has been suggested that there is a link between increasing age, non-attendance at CME courses, and inappropriate prescribing.(42) (presc by can--24). But there was no enough data, however, to explore this hypothesis further. Other studies do not support this explanation.(43,44, 45)(presc by can 44-47) The purpose of the implementation of clinical guidelines is to improve quality of care. However, studies have showed that the US National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC) guidelines seemingly had little effect on the pattern of antihypertensive drug prescribing.(14, 46, 47) (13, 13----21) Two possible reasons, first is that physicians may be reluctant to change drug therapy because of already well controlled blood pressure. Second reason may be that prescribing behavior was influenced by pharmaceutical manufacturer promotional activities. (14)(13)
Practice setting characteristics have been shown to influence prescribing behavior. The data clearly demonstrated a relation between polypharmacy and practice setting characteristics. Practices with large number of patients have fewer drugs prescribed per patient compared with practices with low number of patients. (48, 49)(18-22,23) This finding was consistent with other studies.(6) (18) Busy working physician were more inclined to prescribe multiple drugs than physicians with low workload. (50, 51)(18----24,25) It has been noticed that new drugs have been adopted by partnership practices faster than single-handed practices. (5)(4) The type of practice also influences physicians' use of drugs. Salaried physicians practicing in government-funded community health centers had better prescribing patterns than physicians in fee-for-service group practices.(52) (pres bycan 5) Free-for-service practices were associated with higher rates of new drug utilization.(7)(19) the magnitude of this association was not large enough to expect major cost saving related to new drug use.(53)(19---52). It has been noticed that fee for service patients were more likely to follow JNC guidelines than the patients with health maintenance organization insurance. Thus, the patients with health maintenance organization insurance had no preference for encouraging their physicians to select more cost-effective drugs. In contrast, fee for service patients appeared to have more preference for selecting lower cost drugs.(14) (13) Although, financial incentives represent a non-voluntary strategy to implement change in medical practice, it had a limited, temporary effect on the prescribing behavior.(13, 54) (26, 26-----18)
Working in rural areas influenced prescribing behavior. Lower rates of new drug utilization among physicians working in rural areas may be due relative isolation of rural physicians from colleagues who may have influence in the decision to prescribe new treatment.(55) (19----22)
Only one study concluded that there was no significant difference in number of different drugs prescribed by practices operating a formulary from that found among practices with no formulary. This could be due to, that all physicians in those practices may not comply with the formulary, or formularies contained a narrow range of drugs.(15) (12) However, this finding was not supported by other studies. The use of formulary has been found to influence prescribing behaviors and reduce costs. (56, 57)(12---1,3)
There are many factors related to drugs that may influence physician prescribing behavior. Early use of new drugs may not be compatible with appropriate prescribing. Newness should not be seen as a virtue in a pharmaceutical product and that it is crucial that physicians think more carefully before prescribing a new drug.(58)(presc by can) Small proportion of physicians prescribed new drugs even for drugs that were known as providing substantial improvement over existing treatment.(7,59) (19, 19---8) Costs of health care are escalated by increased spending and use of prescription medications. There was no enough evidence that physicians prescribing behaviour affected by awareness of drug cost (60)(20). It may be that physician prescribing behaviour isÂ insensitive to cost information. other factors such as drugÂ efficacy, , patient compliance, side effects and peer recommendationsÂ may be more important.(61) (20---5) Particularly for chronic care medicationsÂ that have proven to be effective for an individual patient,Â cost may be a minor factor. However, several studiesÂ have shownÂ that education of physicians about drug prices can change prescribingÂ behaviour and reduce cost by improving selectionÂ of cost-effective drug treatment.(62,63,64) (33-18,19,20) In qualitative studies drug price was a recurrent theme and was mentioned as the main reason for choosing first line treatment. Price was also mentioned as the reason for drug switch.(2) (factors Allan)
It has been noted that when physicians were aware that patients would have to pay out of their own pockets for prescriptions, or they learned from patients' complaints to them, they modified their prescribing behavior accordingly.(58)( pres by can)
At primary care level, as high as two thirds of all prescriptions were repeated. One possible reason may be the impact of hospital prescribing in volume was most obvious with repeat prescriptions for patients with chronic disorders.(9) (32) Second possible reason is that large proportion of repeat prescription issued during indirect contact.(24) (17) In UK study, it has been found that 23% of the patients had been receiving repeat prescriptions for more than a year without seeing their family physicians.(65)(17---2) Practices with high number of patients on repeat prescriptions were found to have an increased risk of polypharmacy. It has been noticed that practices using a wide range of different drugs had a high prevalence of polypharmacy.(66) (18---31)
Physicians' sources of information about pharmaceutical agents are likely to be a major factor in prescribing behavior. The drug representatives visited physicians on frequent bases using a wide variety of promotional techniques including drug samples, gifts, and educational materials.
Accepting drug samples was associated with preference and prescription of new drug (25,67)(29, 29-40) In one study, it has been found that 85% of medical students believe it is improper for politicians to accept a gift, whereas 46% found it improper for themselves to accept gift of similar value from a pharmaceutical company.(68) (29---9)
All educational materials sponsored by pharmaceutical industries including funding for travel or lodging to attend educational symposia, industry-paid Meals, pharmaceutical representative speakers ,CME sponsorship and honoraria, research funding influenced prescribing.(25)(29) The finding that advertizing on clinical software had little effect on prescribing behaviour was similar to other studies results when examining the relationship between prescribing and advertizing in journal. One study found no relation between the advertizing and for a drug and the amount and prescribing by physicians. (69)(5---11)
It is reasonable that practices with high proportion of elderly patient had high rates of drug utilization. The observation that some of practices had sicker patients than others; this observation may be due to that sicker patients chose specific practices or physician prescribing behaviour may have made their patients sicker. (4)(1) It is not clear why sicker patients chose particular practices. One study has found that practices with high proportion of elderly patients were associated with greater likelihood of prescribing of new drugs, but lower new drug utilization. It has been suggested that physicians faced patients with coexisting disease.(70, 71) (19-53,54) The finding that patients in the remote regions had low prevalence of drug prescribing may have been because of limited access to medical services. (72)(Quesinable presc) Females were prescribed more medications than males. When gender-specific medications are excluded the differences are less pronounced. (73)(9-16)When female-specific therapeutic groupings and treatments are removed, differences still exist between male and female prescribing.
Physicians prescribing behavior appears to be influenced by multiple factors. Majority of studies in this review retrieve their data from health database. However, these comprehensive health databases have no information on the indications for drug treatment or ascertainment of comorbidity that may have affect prescribing behavior. Thus, attributes of the practice population need to be considered as potential biases. Data is lacking on combination of each factor to patient outcomes, this gap in the literature needs to be addressed. Therefore, it is difficult to measure the appropriateness of physicians prescribing. Physicians prescribing behavior can be improved by implementation of slowly progressing changes. Finally, prescribing is a clinical decision; studies; of clinical decision making are about people, behavior and contexts. They need both quantitative and qualitative approaches.
Davidson et al., 1995
336 general practices
Gill et al., 1997
310 general practitioners
Denig et al.,1998
181 general practitioners
Steffenson et al., 1999
95 general practitioners
Handerson et al., 2008
1336 general practitioners
Bjerrum et ak., 2000
173 general practitioners
Harris et al., 1996
115 general practices
Chon et al., 2009
Mortin et al., 2002
31 general practices
McGavock et al., 1988
23 general practices
Fernandez et al., 2008
McCarthy et al 1992
362 general practioners
Guo et al., 2003
7.3 million prescriptions
Bjerrum et al., 2002
Bjerrum et al., 2001
173 general practioners
Rberts et al., 1993
90 general practices
Straand et al., 1999
Bjerrum et al., 1999
173 general practices
Tamblyn et ak., 2003
1661 general practitioners
Omstein etal., 1999
Grimmsmann et al., 2009
730 general practices
Buck et al., 2009
Kuijpers et al., 2008
Martens et al., 2007
53 general practitioners
Straand et al., 2006
600 general practitioners