Summary of main findings
This study shows that Swedish GPs find it easier to start than to stop prescribing psychotropic drugs and that there is some reluctance to alter other physicians’ prescriptions. A considerable share of GPs believe between-practice variation in psychotropic drug prescribing is high, given equal symptoms. Similar proportions of GPs believe patients take psychotropic drugs for reasons that are mostly medical, or mostly socioeconomic, respectively. Although few GPs report actually prescribing new drugs, there is a tendency to believe, or to believe that patients think, that new and expensive drugs are more effective than older and cheaper ones. GPs are on average satisfied with their level of prescribing of antidepressants and sedatives. The prescribing of antipsychotics, on the other hand, is more often rated as too low than too high, and only about half the GPs report prescribing them.
Attitudes, beliefs and behaviours
Among the GPs in this study, six out of ten found it easier to start than to stop prescribing psychotropic drugs, whereas the opposite was true for only one in ten. The ease of starting and stopping a drug may be linked to the tendency of that drug to cause dependency, making sedatives of the benzodiazepine type particularly difficult to stop [12, 17]. One Dutch-Swedish study found, for example, that two thirds of patients who started taking benzodiazepines were still taking them one year after initial prescription, and one third were still taking them after eight years [22].
Of course, many factors other than the drug’s pharmacological properties influence prescribing decisions. Commonly, GPs are faced with the request to repeat a prescription originally issued by another physician, a situation that poses a particular set of problems. In the present study, GPs more frequently found it difficult rather than easy to change a colleague’s prescription. The assessment of another physician’s reason for prescribing may be difficult because of limited information in the medical records [23]. Moreover, when a prescription is “inherited” from another physician, the receiving GP is likely to consider some of the responsibility remains with the initiator [11]. This tendency may be more pronounced for problematic drugs such as benzodiazepines and opioids, where assuming full responsibility may imply that the GP should take on the demanding and sometimes thankless task of trying to reduce dosages [11].
Psychiatric drug therapy is often seen as being more arbitrary than its somatic counterpart, a finding we highlighted in a paper entitled “Psychiatry is not a science like others” [10]. In the present study, about half of the respondents believed there was considerable variation in psychotropic prescribing (for equal symptoms) between GP practices; the other half was undecided and a few considered variation to be negligible. These findings are in agreement with variations in psychotropic prescribing practices nationally, with higher use of antidepressant and sedative drugs in western Sweden and higher use of antipsychotic drugs in the north [7]. Internationally, there are also significant differences in the choice and volume of psychotropic drugs prescribed between regions and nations, with more frequent use among women, the elderly, and the socially deprived [3,4,5]. Even accounting for these factors, however, much variation remains unexplained and is sometimes put down to the “diverse prescription habits of physicians” [4].
In the present study, the GPs were overwhelmingly in favour of using psychotherapy rather than psychotropic drugs for mild psychiatric disease, a finding that may indicate that Swedish GPs define “mild” disease in terms of “not needing psychotropic drugs”. The high availability of psychotherapy in Swedish primary care, illustrated by our finding that virtually all respondents had access to psychotherapy at their practice, makes referral feasible. Internationally, on the other hand, a scarcity of psychotherapists is often cited as a factor that contributes to high levels of prescribing of psychotropic drugs in primary care [15,16,17].
The question concerning the nature of problems treated with psychotropic drugs divided the GPs into two equally sized camps. Eight out of ten respondents saw these problems as either predominantly socioeconomic or predominantly medical, and only two in ten chose the in-between option. The framing of a patient’s condition is, indeed, often contentious in general practice: does the patient actually suffer from a psychiatric disease, or are the symptoms more properly viewed as manifestations of his/her difficult life situation [13, 24]? Making a distinction between medical and socioeconomic factors as the cause of ill health may be difficult in any health care setting. It is, however, arguably more difficult in primary care, as GPs treat patients with milder symptoms and may also be more aware of patients’ circumstances through long-established contact and knowledge of other family members [15, 25].
Most respondents, seven out of ten, believed health care staff behaved similarly towards users and non-users of psychotropic drugs. Nevertheless, almost all of the remainder believed that behaviour towards users of these drugs was worse than that towards other patients. In terms of stigmatisation of the mentally ill, the secular trend is probably one of decreasing stigma [10, 13].
Six out of ten GPs were neutral about the association between a psychotropic drug’s newness and its perceived efficacy. Three out of ten believed new drugs were more effective than old ones. These proportions were very similar to the GPs’ estimates of patients’ opinions about efficacy in relation to expensiveness. Whereas some GPs do believe new and expensive drugs are more effective [10], most research indicates that GPs place little emphasis on cost in treatment decisions [26]. Internationally, physicians have been found to make largely inaccurate estimates of medicine prices [27, 28]. Very few GPs in our study claimed to actually prescribe new psychotropic drugs, a finding in line with previous research indicating that Swedish GPs tend to follow the therapy recommendations issued by Drugs and Therapeutic Committees [29].
Decisions about drug therapy may be influenced by prescribing technology. Multi-dose dispensing, where a patient’s drugs are automatically dispensed in plastic bags corresponding to each instance of administration, is common among the elderly in Sweden [30]. At the time of the study, the electronic interface for prescribing multi-dose dispensed drugs had a function for repeating all the patient’s drugs by a single click of the mouse [10]. In our study, the question about dose-dispensed drugs and en masse renewal, i.e. without individual consideration of each medication, yielded mainly negative answers. However, a quarter of the GPs claimed to do this more or less often. Consistent with this finding, other studies have shown that the drug regimens of patients with multi-dose dispensing are reassessed less frequently [31], and that patients with normal prescriptions have more appropriate drug regimens than those with multi-dose dispensing do [23, 32]. Concerns about the safety of the repeat-all function finally led to its removal.
The GPs in the present study were, overall, content with their prescribing levels in relation to medical needs: seven out of ten were satisfied with their prescribing of antidepressants, and six out of ten with their prescribing of sedatives and antipsychotics. The latter category stood out, however, in two respects. First, whereas almost all GPs had prescribed from the first two drug classes, only about half claimed to have prescribed antipsychotics recently. Second, whereas the distribution of dissatisfied prescribers was symmetrical for antidepressants and sedatives, it was highly skewed towards the “too low” side for antipsychotics. It thus appears that GPs view antipsychotics as a drug group that is rarely prescribed and somewhat underused. Prescription rates of antipsychotics are, as noted in the introduction, indeed much lower than those of the other main drug categories. A potential explanation for the unexpected finding of perceived underuse may be that antipsychotics may be interpreted semantically as “drugs for psychosis”, thereby evoking a context wherein patients are often reluctant to comply with therapy and prone to use too little medication.
Drivers of increased prescribing
As seen in the introduction, the prescribing of psychotropic drugs is rising in Sweden. The increase has, however, largely been confined to antidepressants, mirroring the expanding use of drugs from this category in several European countries [3]. Meanwhile, the international trend for the prescribing of sedatives is mixed [4], whereas that of antipsychotics seems to be rising [5]. Keeping this in mind, what are the implications of our findings for the wider aim of the project – understanding the determinants of psychotropic drug prescribing in primary care?
First, the finding that GPs are satisfied with their level of prescribing in spite of high (and rising) levels of antidepressant use, may indicate that this drug category is seen as safe and useful. Granting that many GPs feel uneasy about increasing levels of antidepressant prescribing [13], the dominant attitude may still be that antidepressants are comparatively innocuous [10, 15]. It may be hypothesised that high levels of antidepressant prescribing are tolerated, as long as the level of sedatives prescribed is under control.
Second, we show that GPs find it easier to start than to stop prescribing psychotropic drugs. Importantly, we did not ask if this led to GPs actually starting prescribing drugs more often. Nevertheless, if acted upon, this inclination would tend to inflate prescribing rates over time. Moreover, the GPs indicated some reluctance to change a colleague’s prescription, which would similarly favour increased prescribing, if acted upon. Requests for repeating other physicians’ prescriptions may be problematic for reasons already highlighted; additionally, it is often expected that a drug regimen initiated by a psychiatrist will be continued after referral back to primary care [10]. More generally, changing or refusing to repeat another physician’s prescription may be construed as a sign of disrespect towards that colleague, and may therefore be avoided.
Third, touching on the previous point, there are a number of “not strictly medical” considerations that are clearly of importance in understanding the mechanisms of psychotropic prescribing in primary care. Patients’ expectations are a major factor, and GPs often report that patients seek simple solutions to complex psychosocial problems [10, 13]. In line with this, we found that many GPs regard socioeconomic factors as relevant for psychotropic prescribing. Whatever the GP holds to be true about a symptom being “medical” or not may, however, be of less immediate significance than the practical need of getting through a day’s work. Thus, a high work load, lack of alternative therapies and a general sense of being overwhelmed by patients’ predicaments may lead to the prescription of psychotropics as a coping mechanism [16]. Awareness of this from personal experience may be one reason why GPs consider psychiatric drug treatment to be somewhat arbitrary.
Strengths and limitations
The high number of respondents is a strength of the present study. Nevertheless, the response rate is a limitation; we do not know if non-responders’ replies would have differed from those obtained in the study. However, the response rate is similar to that in many other studies based on questionnaires sent to physicians (e.g. [12, 17]), and the distribution of specialists and interns among the responders was similar to that in the overall population. Characteristics of responding and non-responding GP practices were also similar in terms of location and ownership, after controlling for the number of registered patients. Practices with more registered patients were more likely to respond; an expected finding as larger practices usually have more physicians. Further, the returned questionnaires were largely complete, with little missing data.
A limitation of the study is that a 20-item questionnaire is unable to explain more than a small part of a complex phenomenon such as psychotropic drug prescribing. A further limitation of the study is that the questionnaire used was not extensively validated. In addition, we mainly used the umbrella term “psychotropic drugs” in the questionnaire, after presenting the three main psychotropic drug categories. This choice reflects our previous use of this term [10] and is based on the view that psychotropic medications have many common features. In addition, the labels attached to psychiatric drug categories are poorly matched to the drugs’ actual effect and clinical use. Antidepressants, for example, are arguably more useful in anxiety disorders than in depression [33]. Further, in contrast to much other research, we did not restrict the questions to patients in a certain age range or with a specific psychiatric disorder. This approach generated knowledge with a broad scope, which we believe is relevant to primary care.
Finally, we consider it a strength that the input of all physicians working in primary care was sought, including temporary staff, and that the questionnaire we used was based on our previous research using qualitative methodology [10]. Future research could focus on integrating qualitative and quantitative components, using a mixed methods design, with the aim of further elucidating physicians’ rationales for decision making about psychotropic drugs.