Social Science & Medicine 49 (1999) 459±467
Prescribing benzodiazepinesÐa critical incident study of a physician dilemma Preben Bendtsen*, Gunnel Hensing, Lotta McKenzie, Anna-Karin Stridsman Department of Health and Environment, Division of Preventive and Social Medicine, Faculty of Health Sciences, S-581 85 LinkoÈping, Sweden
Abstract Use of benzodiazepines has been discussed extensively both among the public and within the medical society. The aim of this study was to explore the quality of dilemmas experienced by physicians when prescribing benzodiazepines. A questionnaire was sent to 213 Swedish General Practitioners. The critical incident technique was chosen as an appropriate method for surveying professional experiences. Concern for the patient and threats to the integrity of the physician were common dilemmas. The physicians did not believe that the patients were telling the truth or did not trust the patients' ability to handle the medicine. The most frequent consequences of the dilemmas were worry about a disturbed relationship with patients indicating an uncertainty as to how to create a good relationship with them. The participants in the study were aware of the national guidelines for prescribing benzodiazepines, but due to insucient time a prescription was often chosen as a way to handle the dilemmas. Improvement in the rational use of benzodiazepines is not achieved by the medical board making new rules but rather by oering physicians education in communication and negotiating skills as well as more time with the individual patient who is requesting benzodiazepines. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Benzodiazepines; Dilemma; Physician; Prescription; Primary health care; Sweden
Introduction The ®rst benzodiazepine, chlordiazepoxide, was introduced in 1961. After this a number of similar substances have been released. Benzodiazepines are now among the most widely prescribed anxiolytic and hypnotic medications due to their eectiveness in relieving anxiety and insomnia (Perez and Tudor, 1993; Uhlenhuth et al., 1995; Simon et al., 1996; MoralesSuarez-Varela et al., 1997). When ®rst introduced,
* Corresponding author. Tel.: +46-1322-5487, fax: +461322-5490. E-mail address:
[email protected] (P. Bendtsen)
benzodiazepines appeared to be a safer alternative than previous drugs such as barbiturates with regard to the consequences of excessive use. However, during the last decade or two there has been growing concern both among the public and among physicians about the side eects of benzodiazepines and their potential for dependence in long-term use (Cormack and Howells, 1992; Isacson, 1997). The most frequently reported side-eects have focused upon impaired psychomotor performance and cognitive function, especially memory, withdrawal symptoms, dependence and abuse (Mattila et al., 1986, 1988; Livingston, 1994; Ranstam et al., 1997). Furthermore the use of benzodiazepines has been questioned as the conditions for which they are most commonly used often have their
0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 9 9 ) 0 0 1 3 3 - 1
460
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
origin in personal and social circumstances and not truly physiological processes (Morales-Suarez-Varela et al., 1997). In addition, there is no evidence of benzodiazepines being eective for more than 3±4 months (Ranstam et al., 1997). The prevalence of benzodiazepine use in populationbased studies varies from 2±10% of the adult population. The use of benzodiazepines increases with age and is more common among women than men. Of the total users of benzodiazepines 15±30% are reported to be continuous users and have, in most cases, developed a physical dependence (Isacson et al., 1992; Livingston, 1994; Morales-Suarez-Varela et al., 1997; Woods and Winder, 1995; Simon et al., 1996; Isacson, 1997). Both epidemiological and clinical studies focus either on the occurrence or consequences of benzodiazepine use. However, the physician's perspective is often missing (Isacson, 1997). In a previous study on uncomfortable prescribing decisions by Bradley, tranquillisers and hypnotics were the second most common reason for discomfort among the prescribing physician (Bradley, 1992a). Although written information about irrational prescribing is available, physicians need more knowledge about how to communicate this to the patients (Bradley, 1992b; Boixet et al., 1996). BjoÈrndal and Fugelli (1989) found good theoretical knowledge about when and how to prescribe benzodiazepines among physicians but found, despite this, a great variation in prescription practices with respect to benzodiazepines (BjoÈrndal and Fugelli, 1989). In a literature review, Isacson (1997) found that female and younger physicians prescribed psychotropics to female patients more often than to male patients, and did so more frequently than male and older physicians did. Dierences among various specialisations of physicians have not shown any consistent pattern (Heiman and Wood, 1981; Isacson, 1997). Hence, the great variation among physicians with regard to prescribing benzodiazepines is not clearly understood. In Sweden and several other countries various steps have been taken to control the prescription of benzodiazepines. The Swedish Medical Board has published recommendations about how to prescribe these e.g. never prescribe benzodiazepine at the ®rst visit or to an unknown patient, oer nonpharmacological support or psychotherapy as the ®rst choice and restrict the use of benzodiazepine to more severe cases of anxiety or insomnia (Socialstyrelsen, 1990; Cormack and Howells, 1992). In addition, many countries have introduced special numbered prescriptions, sometimes in duplicate or triplicate, for registration and follow-up (Schwartz and Blank, 1991). Pharmacoepidemiological reports have pointed out that many physicians do not follow the expert recommendations (Simon et al., 1996). On the basis of this it is plausible to believe that physicians may ex-
perience dilemmas in prescribing benzodiazepines due to the con¯icting advice in the public and professional debate about whether or not to prescribe (Livingston, 1994; Peturson, 1994). The present study was undertaken in order to explore the qualities of dilemmas general practitioners in Sweden experience when prescribing benzodiazepines. Furthermore, the decisions to prescribe or not were analysed in order to obtain more knowledge of factors in¯uencing the prescription of benzodiazepines. Methods and study population Methods The critical incident technique was chosen as an appropriate method for this study. This technique was ®rst described by Flanagan (1954), who de®ned a critical incident as ``any observable human activity that is suciently complete in itself to permit inferences and predictions to be made about the person performing the act'' (Flanagan, 1954). The critical incident technique can be described as part of a phenomenological approach in the sense of describing, analysing and interpreting individuals' understanding of phenomena in their surroundings. The purpose of a ``critical incident study'' is to capture problematic situations experienced in relation, for example, to performing professional tasks. The subjects of the study are asked to describe de®ned incidents, the consequences of those and what they did to cope with them (Bradley, 1992b; Norman et al., 1992). The term ``critical incident'' emanates from the early studies performed by Flanagan among air force pilots (Flanagan, 1954). Semantically ``dilemma'' is a better term for studies performed within the health care setting and was therefore used in this study. ``Dilemma'' can be de®ned as ``a perplexing or awkward situation perceived by a professional to cause disturbance in the performance of a work task'' (Timpka et al., 1995; Hensing et al., 1997). The term ``dilemma'' was chosen as a synonym for ``critical incident'' throughout this paper. A questionnaire developed earlier and used for studies among physicians (Timpka et al., 1995) and social insurance ocers (Hensing et al., 1997) was revised for the purpose of this study. The introductory part of the questionnaire contained demographic questions on sex, age and number of years in occupation. The central question runs: ``Can you describe the last occasion when you experienced a dilemma in relation to the prescription of benzodiazepines?'' The follow-up questions let the subjects describe:
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
461
Table 1 Reasons for the patient's request for benzodiazepines as experienced by physicians (n=90) Consultation situation
Number (%) of dilemmasa
(1) Continuation of previous prescriptions (2) Anxiety, initiation of treatment for depression (3) Person with known or former drug abuse (4) ``Diuse'' request for benzodiazepine (5) Unspeci®ed psychiatric problems (6) Insomnia (7) Somatic disease (8) Prescription lost (9) Abuse in relative (10) ``Mourning''
33 (37) 29 (32) 28 (31) 23 (26) 17 (19) 16 (18) 14 (16) 6 (7) 5 (6) 5 (6)
a
. . . . .
A dilemma could be classi®ed in one or more categories.
whether benzodiazepines were prescribed or not; reasons for that decision; why a dilemma was experienced; measures taken to handle the situation and suggestions for future changes in order to prevent similar situations.
The questionnaire was tested on seven physicians, and some minor changes were made regarding linguistic clarity. Study population Together with an introductory letter the questionnaires were mailed personally in the autumn of 1996 to all physicians (n=213) working as general practitioners (GP) in the county of OÈstergoÈtland, which has more than 400,000 inhabitants and is situated in southern Sweden. General practitioners were chosen since benzodiazepines are mainly prescribed by this group of physicians (Morales-Suarez-Varela et al., 1997). The study population includes both specialised GPs and physicians in training. A reminder letter was sent out and the ®nal study group consisted of 113 physicians, 44 female and 69 male, giving a response rate of 53%. As the total population of physicians consisted of an equal amount of males and females, we had a slightly higher percentage of males compared with those not participating in the study. The years of experience in the group of participants were similar to the remaining group of physicians. There was no signi®cant dierence in the age distribution between male (n=69) and female (n=44) physicians. An internal dropout consisted of 14 individuals who did not answer the entire questionnaire, usually by omitting the open-ended questions, or by giving information on the dilemmas which was too scanty for further analysis. Finally, nine physicians indicated that they had never experienced a dilemma in relation to
prescribing benzodiazepines. Thus the qualitative analysis was based on answers from 90 physicians.
Analysis of data Theoretically the analysis of data was based on a semiotic tradition (Silverman, 1993), which means that an internal analysis of the written answers were made. The open-ended questions were analysed through several separate readings of the material by the third and fourth author. Each questionnaire was initially read as a whole (``horizontal reading''). Then the questions were read as a group i.e. the question on dilemmas experienced was read from all questionnaires (``vertical reading''). The horizontal reading gave a contextual picture of the cases that the physicians referred to, whereas the vertical reading was made in order to capture common features of the answers. A preliminary set of empirically de®ned categories was created on a broad basis, which lead to a large number of categories. The underlying strategy was to avoid interpretations at this stage but rather to create more categories. The same two authors then compared and discussed each category until satisfactory agreement was found concerning the type of categories. The discussion was complemented by reading and reanalysing of the questionnaires when necessary. During this process the categories were systematically sorted, leading to a reduction of the number of categories. All questionnaires were used for the analysis, but after reading and analysing the great majority of the questionnaires no new categories were formed, which indicated that satisfactory saturation had been found concerning the type of categories. Based on this categorisation an independent analysis was made by the ®rst author, who read all the answers. The material was then discussed with the third and fourth author, and after consensus was reached,
462
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
Table 2 Reasons given by physicians (n=90) for the dilemma experienced Reason for dilemma
(A) Concern for the patient (1) Fear of making a wrong decision (2) History of abuse (3) Fear of initiating dependence (4) Fear of harming the patient due to toxicity (B) Integrity of the physician (5) Feeling of manipulation (6) Suspicion that others take the medicine (7) Physician feels threatened due to pressure from the patients (8) Disturbed patient±physician relation (C) Miscellaneous (9) Unknown patient (10) Physician generally restrictive (11) Time limitations a
Number (%) dilemmas where physician prescribed (n=67)a 26 24 9 7
Number (%) dilemmas with no prescription (n=23)a
(39) (36) (13) (10)
4 (17) ± ± ±
10 (15) 8 (12) 4 (6) 3 (4)
3 (13) ± 9 (39) 7 (35)
7 (10) 5 (7) 3 (4)
± ± ±
A dilemma could be placed in more than one category.
necessary revisions were made. At this stage the main changes involved cutting down the number of categories and labelling some. Finally, the second author, with extensive experience of the critical incident technique from earlier studies, examined a random sample of the categorisations, making additional revisions. Results Reasons for requesting benzodiazepines In the ®rst part of the analysis the dierent reasons for wanting benzodiazepines were scrutinised, and each consultation situation was categorised into 10 dierent categories as displayed in Table 1. The most common consultation situation where a dilemma was experienced occurred when patients just wanted an iteration of an earlier prescription and the doctor for various reasons reacted to this request (category 1). In many cases the patients were considered abusers or former drug abusers or gave a diuse reason for wanting benzodiazepine (category 3 and 4). Nevertheless, various psychiatric symptoms such as anxiety, unspeci®ed symptoms, insomnia and mourning were described by the patients in more than 75% of the situations where the doctor experienced a dilemma (category 2, 5, 6, 10). Reasons for the experienced dilemma In Table 2 the results of the analysis of the reasons for the dilemma are displayed. The physicians are divided into two groups Ð those who prescribed
(n=67) and those who did not (n=23) prescribe benzodiazepines in the situation. On the basis of the physicians' report we categorised the reasons for the dilemma into 11 categories. From these, two major groups of categories were found. One group, categories 1±4, was related to the physicians ``concern for the patients''. These categories indicate that the physician is aware of the risks of abuse and dependency when prescribing benzodiazepines and that this is the most frequent reason for a dilemma. With regard to concern about harming the patient, the dilemma for the physician was usually whether benzodiazepines were the right choice or not. One physician wrote: (quotations selected from questionnaires) ``A common dilemma is when I am expected to continue a prescription initiated by a specialist in psychiatry'' (category 1). The other main group of reasons for the dilemma was concerned with the ``integrity of the physician'' as displayed in Table 2, categories 5±8. The patient's attitude and behaviour when requesting the benzodiazepines was apprehended by the physicians as a threat to their professionalism or the interpersonal relationship due to a feeling of being manipulated. A disturbed patient±physician relationship was explicitly stated in 3 dilemmas (category 8). One physician wrote about an elderly woman on multiple drugs. He had tried to control the prescriptions but the patient made sure that she met dierent physicians at the clinic: She is known to be ``shopping'' around at dierent doctors, asking for prescriptions. She seems to
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
463
Table 3 Reasons for declining a prescription of benzodiazepines (n=23) Reason (1) (2) (3) (4) (5) a
Number (%) of physicians ``con®rming'' the speci®c reasona
Abuse or suspicion about abuse Doctor oering alternative No indication for treatment with benzodiazepine Never introducing benzodiazepine to younger patients Unknown patients
21 (91) 4 (17) 3 (13) 2 (9) 2 (9)
A dilemma could be classi®ed in one or more categories.
have an excellent ability to persuade the physicians. She often excuses her behaviour with a story about her purse being stolen. Furthermore, it is impossible to prove that she has not been robbed of her drugs when she claims it. I am not sure whether she really uses the tablets herself (category 5 and 6). An additional example of a dilemma due to a feeling of being manipulated was given by another physician: A man with several years of benzodiazepine and alcohol abuse who had just moved to my part of town wanted me to take over the prescription of benzodiazepines. The patient admits that he had been attending two physicians where he previously lived in order to get the medicine. Will I be the only prescriber now? (category 5). The last three categories were concerned with prescribing benzodiazepines to an unknown patient, the physician being restrictive in general, and time limitation.
The dilemmas reported by the 23 physician who did not prescribe benzodiazepines could only be classi®ed into four of the 11 categories displayed in Table 2. Three of these categories were concerned with the integrity of the physician i.e. a feeling of manipulation, the physician feels threatened by the patient and a disturbance of the patient±physician relation. Reasons for declining a prescription In most cases (91%) suspicion about abuse was given as an explanation for declining prescription (Table 3). Only four out of the 23 patients to whom a benzodiazepine prescription was refused were oered an alternative, as could be read from the questionnaires. One physician pointed out that benzodiazepines were normally oered as a routine in order to overcome side eects when initiating SSRI therapy: Decided in agreement with the patient not to treat with benzodiazepines due to previous abuse (category 1).
Table 4 Reasons given by the physicians (n=67) for prescribing benzodiazepines Category ``Reasons'' (1) Planning to follow up the prescription or temporary prescription (2) Indication for treatment (3) Continuation of previous treatment (4) Low risk of abuse according to the physician (5) Patient expectation (6) Elderly patient (7) Avoiding other kind of abuse (8) Avoiding extra workload for the physician (9) Maintenance of physician±patient relation (10) Avoiding abstinence (11) Treatment for ``mourning'' (12) Time limitation (13) Superior's recommendation a
A dilemma could be classi®ed in one or more categories.
Number (%) of physicians ``con®rming'' the speci®c reasona 24 (36) 22 (33) 17 (25) 14 (21) 10 (15) 5 (8) 5 (8) 3 (4) 3 (4) 3(4) 2 (3) 2 (3) 1(2)
464
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
Reasons for prescribing benzodiazepines The reasons for prescribing the medicine were categorised in thirteen categories, as displayed in Table 4. The ®rst four categories contain reasons indicating that a prescription of benzodiazepines was given under controlled conditions or that the medication was an adequate choice. The remaining nine categories indicated more dubious reasons for prescribing benzodiazepines. For example ®ve physicians referred to the patients' age as a reason for prescribing: The patient, an elderly woman, could not wait until her usual doctor came back. The risks of abuse were less due to the patient's age (category 6). Another ®ve physicians indicated that benzodiazepines were prescribed in order to avoid the abuse of alcohol or other drugs: The patient should use the benzodiazepines when necessary in order to avoid narcotics (category 7). Another physician described a heroin addict who was about to lose his family due to his abuse: The patient asked for benzodiazepines since he cannot aord to buy heroin due to unemployment. I could not ®nd any alternative but to prescribe (category 7).
Suggestion by the physician in order to overcome future dilemmas Finally, the physicians were asked to indicate what action should be taken in order to prevent similar incidents in the future. Around 20% (n=19) had no suggestions, but the majority 46% (n=41) referred to the medical board recommendations. Other suggestions were to give more time to each patient 10% (n=9), improved consultation with specialists 11% (n=10), while ®ve physicians (6%) suggested more government control. As the government had recently increased the control by introducing a new serial-numbered prescriptions form, we asked whether this had changed the participating physicians' routines. The majority, 75%, indicated no change in their daily practice, whereas the remainder reported that they had become more restrictive. Discussion The study underscores the diculties physicians' ex-
perience when deciding whether or not to prescribe benzodiazepines. Overall the decision whether or not to prescribe seems to be rather complex and results in ``cognitive dissonance'', as Bradley puts it (Bradley, 1992a). Any decision involving the choice between two or more alternatives can result in cognitive dissonance, leading to discomfort, or a dilemma as described in this study. The physicians experienced dilemmas in clinical situations that appear quite justi®ed with respect to proper indications for prescribing benzodiazepines. Hence, the majority of the problems presented were psychiatric (Table 1). A majority (39%) mentioned ``fear of making a wrong decision'' as a reason for the dilemma. If the dilemma was related to uncertainty about the diagnosis or correct treatment, the physicians ought to be able to consult a psychiatrist. However, the participants in the study do not mention the possibility to refer to a psychiatrist for consultation concerning the diagnosis or suggestions about alternative pharmacological or psychosocial treatment (Table 3). Such failure to consult a psychiatric specialist on the part of primary health care physicians when uncertain about diagnosis and treatment is in line with earlier studies. GPs are known from several earlier studies not to detect psychiatric disorders among their patients (Paykel and Priest, 1992). In this study most patients presented with psychiatric symptoms, and the problems seemed instead to be uncertainty as how to handle psychiatric needs. An educational program presented to GPs on the Swedish island of Gotland resulted in better management of patients suering from depression and even lowered the suicide rate (Rutz, 1992). Considering that psychiatric disorders have been appointed by the WHO as one of the most important diagnostic groups in causing ``disability adjusted life years'' (DALYs), it is urgent to increase knowledge and skill in the management of these patients (WHO, 1996). It is beyond both the purpose and methods of this study to discuss whether the dilemmas experienced are relevant or not in relation to the Medical Board recommendations or the best clinical management. However, such questions would be of interest for further study. Lack of con®dence in the patients was a common reason for experiencing dilemmas. The physicians did not believe that the patients were telling the truth or did not trust the patients' ability to handle the medication. Reasons given for this distrust were fear of abuse, fear of initiating dependence, suspecting others of taking the medicine or feeling threatened by the patients. Irrespective of whether this disturbance is an eect of a genuine concern for the patients or a fear of being manipulated, the consequence is a dilemma and possibly a threat to a good patient±physician relationship. Such disturbance counteracts the recommen-
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
dations of the Medical Board that nonpharmacological support should be oered as the ®rst choice. Worries related to a disturbance in the relationship to the patient were a frequently mentioned reason for a dilemma both in those incidents where the physician prescribed or declined to prescribe benzodiazepines (categories 5±8, Table 2). This was mentioned more explicitly in 35% of those dilemmas where the physician declined to prescribe benzodiazepines (category 8, Table 2). A general concern about future disturbances in the relationship to the patient due to a refusal to prescribe benzodiazepines might be a result of uncertainty about how to create a good relationship with the patients. A ``no'' to a request for benzodiazepines could mean a genuine concern for the patient Ð that the physician cares about the patient. This ought to strengthen rather than disturb the relationship between the physician and the patient. Earlier studies have emphasised the need for additional training of physicians in communication skills (Bradley, 1992a,c; Barter and Cormack, 1996). There seems to be a need for education in how to avoid prescribing when the physicians ®nd no clinical indications for benzodiazepines. Communication skills and how to pursue nondrug alternatives have been suggested as the most urgent educational needs for physicians, which our study emphasises. Two previous studies found a lack of knowledge among the prescribing physicians with respect to what the patients themselves think about their medicine. By asking the patients about their experiences and beliefs physicians can open a dialogue and inform the patient about what they think about the medicine (Britten et al., 1995; Barter and Cormack, 1996). Only a few participants have mentioned insucient time in dealing with the patients. However, a considerable number of the participants (36%) indicate that they are planning to ``follow up the patient at a later stage'' (Table 4). This could also indicate a lack of time. This ®nding con®rms an earlier study, where high prescribers of benzodiazepines indicated that they had prescribed because of a heavy workload, stating that writing prescriptions saved time (Cormack and Howells, 1992). If the physicians are to be able to communicate their reasons for declining to prescribe benzodiazepines they obviously need both communication skills and time. Still, we cannot be certain that additional time with each patient requesting benzodiazepines will facilitate a better prescription practice of benzodiazepines. This would be an interesting topic for future research. The participants in the study were aware of the guidelines and the new routine concerning benzodiazepine prescriptions. They were also very likely to be aware of the public and professional debate about
465
the risk of dependence in long-term users and the fact that benzodiazepines have a strong abuse potential. Such factors are related to the community level, i.e. the physician as part of a public authority rather than being the patient's therapist (Lipsky, 1980). As a civil servant, the physician is obliged to ful®l certain duties and sanctions can be taken to correct those who do not follow their obligations. It might well be that recommendations from the Medical Board contribute to their experience of dilemma. To some extent, physicians have to realise the ideal presented in recommendations in situations when time pressure, lack of knowledge, needy patients and inadequate psychiatric health services make it a ``mission impossible'' (Lipsky, 1980). In addition non drug alternative treatment is not easy available (Boixet et al., 1996). Seen from this perspective, the dilemmas are rather an eect of an unrealistic health and prescribing policy than a lack of knowledge among the physicians. The ®nding that most of the participants referred to the guidelines when asked to suggest action to prevent future dilemmas re¯ects this. Despite the present guidelines, the physicians still experience a number of dierent dilemmas. New guidelines would probably not make a considerable change in the amount or the quality of dilemmas. Methodological considerations This study has a semi-qualitative approach, and the aim was to explore the qualities of the dilemmas that physicians experience (Patton, 1980). In an interview study, further interviews can be made until no new aspects or qualities of the phenomenon under study are found. A similar procedure can be used in a questionnaire study, i.e. starting the study with a smaller sample, which is increased if needed. We chose another approach and distributed the questionnaires to the whole population of general practitioners ful®lling the inclusion criteria. We have no reason to believe that the type or quality of the dilemmas were dierent in the participating group compared to the nonrespondents. There might, however, be a bias regarding how often those dilemmas are experienced or what the consequences of the dilemmas were. It could be expected that those physicians who were most interested in, or most disturbed by, benzodiazepine prescription would participate to a larger degree in a study of this character. Therefore the frequency or consequences of dilemmas was not analysed in this study. All methods have their own limitations and advantages (Patton, 1980; SaÈljoÈ, 1988; Silverman, 1993). The original critical incident technique was based on interviews, which makes it possible to ask follow-up questions in order to clarify the descriptions of dilemmas and their consequences. This is not possible if ques-
466
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467
tionnaires are used. An advantage of using questionnaires is that it is possible to collect data from a larger sample of individuals. In our study we approached included the total population of physicians working as GPs Ð a methodological advantage. However, a larger sample size could only to a certain extent compensate for using an open-ended questionnaire, which limits the amount of material compared to an interview. Written questionnaires might generate shorter and less developed answers and the respondent probably writes less about complicated and time-consuming matters. However, earlier studies with the method used here showed that the respondents made thorough descriptions of both the dilemmas and their consequences (Timpka et al., 1995; Hensing et al., 1997). The reason for this could be that it is easier to admit mistakes and that the respondent more willingly admits failures in a written questionnaire rather than in an interview. Conclusion On the basis of our ®ndings, improvement in the rational use of benzodiazepines is not achieved by the medical board making new rules but rather by oering physicians education in communication and negotiating skills. The logistic diculties experienced by many physicians also need to be addressed. In particular there seems to be a need for more time to be spent with patients requesting benzodiazepines, and nonpharmacological alternatives should be easily available for the physicians. References Barter, G., Cormack, M., 1996. The long-term use of benzodiazepines: patients' views accounts and experiences. Family Practice 13, 491±497. BjoÈrndal, A., Fugelli, P., 1989. Can regional dierences in consumption of tranquillisers and hypnotics be explained by variations of general practitioners' threshold of prescribing? A methodological study. Scandinavian Journal of Primary Care 7, 67±71. Boixet, M., Batlle, E., Bolibar, I., 1996. Benzodiazepines in primary health care: a survey of general practitioners prescribing patterns. Addiction 91, 549±556. Bradley, C.P., 1992a. Factors which in¯uence the decision whether or not to prescribe: the dilemma facing general practitioners. British Journal of General Practitioners 42, 454±458. Bradley, C.P., 1992b. Turning anecdotes into data: the critical incident technique. Family Practice 9, 98±103. Bradley, C.P., 1992c. Uncomfortable prescribing decisions: a critical incident study. British Medical Journal 304, 294± 296. Britten, N., Brant, S., Cairns, A., Hall, W.W., Jones, I., Salisbury, C., Virji, A., Herxheimer, A., 1995. Continued
prescribing of inappropriate drugs in general practice. Journal of Clinical Pharmacy and Therapeutics 20, 199± 205. Cormack, M., Howells, E., 1992. Factors linked to the prescribing of benzodiazepines by general practice principals and trainees. Family Practice 9, 466±471. Flanagan, J.C., 1954. The critical incident technique. Psychological Bulletin 51, 327±358. Heiman, E.M., Wood, G., 1981. Patient characteristics of clinical attitudes in¯uencing the prescribing of benzodiazepines. Journal of clinical psychiatry 42, 71±73. Hensing, G., Timpka, T., Alexanderson, K., 1997. Dilemmas in the daily work of social insurance ocers. Scandinavian Journal of Social Wellfare 6, 301±309. Isacson, D., 1997. Long-term benzodiazepine use: factors of importance and the development of individual use patterns over time: a 13-year follow-up in a Swedish community. Social Science & Medicine 44, 1871±1880. Isacson, D., CarsjoÈ, K., Bergman, U., Blackburn, J.L., 1992. Long-term use of benzodiazepines in a Swedish community; an eight-year follow-up. Journal of clinical Epidemiology 45, 429±436. Lipsky, M., 1980. Street-level bureaucracy. Dilemmas of the individual in public services. Russel Sage Foundation, New York. Livingston, M.G., 1994. Benzodiazepine dependence. British Journal of Hospital Medicine 51, 281±286. Mattila, M., SeppaÈlaÈ, T., Mattila, K.J., 1986. Combined eects of buspirone and diazepam on objective and subjective tests of performance in healthy volunteers. Clinical Pharmacological Therapy 40, 620±626. Mattila, M.J., Mattila, M., Aranko, K., 1988. Objective and subjective assessments of the eects of ¯upentioxol and benzodiazepines on human psychomotor performance. Psychopharmacology 95, 323±328. Morales-Suarez-Varela, M., Jaen-Martinez, F., LlopisGonzaoÈez, A., Silla Sobrecases, B., 1997. Sociodemographic characteristics of female habitual benzodiazepine consumers in the catchment area of a health care centre. Scandinavian Journal of Social Medicine 25, 176±179. Norman, I.J., Redfern, S.J., Tomalin, D.A., Oliver, S., 1992. Developing Flanagan's critical incident technique to elicit indicators of high and low quality nursing care from patients and their nurses. Journal of Advanced Nursing 17, 590±600. Patton, M.Q., 1980. Qualitative evaluation methods. Sage Publications Inc, London. Paykel, E.S., Priest, R.G., 1992. Recognition and management of depression in general practice: consensus statement. British Medical Journal 305, 1198±1202. Perez, G.G., Tudor, R., 1993. A post-marketing surveillance survey on the use of dipotassium clorazepate in clinical practice in the Philipines. Current Medical Research and Opinion 13, 175±186. Peturson, H., 1994. The benzodiazepine withdrawal syndrome. Addiction 89, 1455±1459. Ranstam, J., Merlo, J., Blennow, G., Hanson, B.S., OÈstergren, P.O., Melander, A., 1997. Impaired cognitive function in elderly men exposed to benzodiazepines or
P. Bendtsen et al. / Social Science & Medicine 49 (1999) 459±467 other anxiolytics. European Journal of Public Health 7, 149±152. Rutz, W., 1992 Evaluation of an educational program on depressive disorders given to general practitioners on Gotland: short and long-term eects. Ph.D. thesis, Faculty of Health, department of Psychiatry, University of LinkoÈping, Sweden. Schwartz, H.I., Blank, K., 1991. Regulation of benzodiazepine prescribing practices: Clinical Implications. General Hospital Psychiatry 13, 219±224. Silverman, D., 1993. Interpreting qualitative data. SAGE Publications, London. Simon, G.E., VonKor, M., Bsarlow, W., Pabiniak, C., Wagner, E., 1996. Predictors of chronic benzodiazepine use in a health maintenance organisation sample. Journal of Clinical Epidemiologi 49, 1067±1073. Socialstyrelsen, 1990. Beroendeframkallande psykofarmaka. Att behandla och foÈrebygga beroende och missbruk
467
(Addictive psychopharmacological drugs. How to treat and prevent dependence and abuse). AllmaÈnna RaÊd, Socialstyrelsen (National Medical Board), Stockholm. SaÈljoÈ, R. (Ed.), 1988. The Written World. Springer-Verlag, Berlin. Timpka, T., Hensing, G., Alexanderson, K., 1995. Dilemmas in sickness certi®cation among Swedish physicians. European Journal of Public Health 5, 215±219. Uhlenhuth, E.H., Balter, M.B., Ban, T.A., Yang, K., 1995. International study expert judgement on therapeutic use of benzodiazepines and other psychotherapeutic medications. II. Pharmacotherapy of anxiety disorders. Journal of Aectice Disorders 35, 153±162. WHO, 1996. Investing in health research and development. Report of the ad hoc committee on health research relating to the future intervention options. Geneva. Woods, J.H., Winder, G., 1995. Current benzodiazepine issues. Psychopharmacology 118, 107±115.
