Uncertainty in clinical practice – an interview study with Swedish GPs on patients with sore throat

Uncertainty is inevitable in clinical practice and perhaps most prominent in work in primary care [1], where patients consult with a wide range of problems, with symptoms in early phases of disease and with problems of life which blur with medicinal problems. Three types of uncertainty have been identified: technical, due to inadequate scientific information; personal, from not knowing the patients’ wish; and conceptual, from problems in applying abstract data to concrete situations [2]. The last decades’ focus on EBM and guidelines sought to remedy one of the identified factors, the insufficient knowledge. Uncertainty is one factor explaining variation in practice and unnecessary use of resources, i.e. laboratory tests in primary care [3]. One of the few studies characterizing individual GPs’ diagnostic strategy in relation to stress from uncertainty showed a positive correlation between anxiety due to uncertainty and diagnostic activity [4, 5].

The physicians’ emotional reactions to uncertainty are described as stress and concern about bad outcomes [1]. Tolerance for uncertainty has been shown to vary between physicians. Experienced physicians tolerate uncertainty better than younger ones and males better than females [1, 6]. A strategy for tolerance is avoidance of or reluctance to disclose uncertainty, learnt during the education of the young doctors [7].

Uncertainty is also discussed as a factor for unnecessary use of antibiotics, which is a threat to modern healthcare. In one recent systematic review, diagnostic uncertainty was found to be the principal indirect fact for inappropriate antibiotic prescribing [8], while fear of possible complications was identified in another [9]. GPs who felt greater discomfort in the face of uncertainty seemed to prescribe antibiotics more often [10]. Recent years’ achievements in research concerning infections have resulted in expanded evidence relevant for the primary care population and evidence-based guidelines are published in many Western countries for common infections in primary care. As another suggested tool to decrease uncertainty and increase appropriate antibiotic prescribing, near-patient tests have been introduced in primary care in several Western countries [8]. Thus important means to reduce uncertainty are now used for common infections in primary care.

In Sweden guidelines for patients with sore throat (pharyngotonsillitis) from 2001 were updated in 2012 [11, 12]. The guidelines aim at a rational use of antibiotics by identifying patients likely to benefit from antibiotic treatment. It is based on reliable evidence that sore throat is a self-limiting disease and the benefit of antibiotics primarily is to relieve symptoms in patients with an infection with Streptococcus group A (GAS). The guidelines recommend the use of the clinical Centor criteria (absence of cough, fever >38.5 °C, tender lymphadenitis and tonsillar coating) to increase the prevalence of patients with GAS infection and thereafter to verify infection with GAS using the near-patient test Rapid Antigen Detection Test for GAS (RADT). There is ample evidence that it is impossible to distinguish between virus and GAS with the naked eye [13]. At the time of the study the guidelines recommended using RADT to verify the presence of GAS when ≥2 Centor criteria were present and antibiotic treatment when the test was positive. The use of C-reactive protein (CRP) test is not recommended in either version of the guidelines as there is limited evidence of benefit in patients with sore throat [11, 12]. Near-patient RADT and CRP in primary care have been used in the Nordic countries for about 25 years. Relevant differential diagnoses (DD) and complications are listed in the guidelines [11, 12].

Studies of the actual behaviour of physicians as well as descriptions of what coping with uncertainty entails have been called for [1]. More complex problems create more uncertainty for the GP than straightforward problems. Managing patients with a sore throat has been used as an example of an uncomplicated clinical problem [1]. However, even this clinical situation contains elements of uncertainty and concern for bad outcomes. In a recent interview study with GPs concerning the management of patients with a sore throat, bad outcomes (potentially threatening DD and complications) were identified as a constant theme for all respondents [14, 15]. Therefore the aim of this paper was to describe strategies for coping with uncertainty in patients with pharyngotonsillitis in relation to guidelines,

The 25 participating GPs were between 33–64 years of age, half were women and 22 GPs had a working experience ≥5 years [14].

Four strategies for coping with uncertainty were identified, one of which was compliant with guidelines, “Adherence to guidelines”, and three were idiosyncratic: “Clinical picture and CRP”, “Expanded control”, and “Unstructured”. The residual uncertainty differed for the different strategies: in the strategy “Adherence to guidelines” and “Clinical picture and CRP” uncertainty was avoided, based either on adherence to guidelines or on clinical picture and near-patient CRP; in the strategy “Expanded control” uncertainty was balanced based on expanded control; and in the strategy “Unstructured” uncertainty prevailed in spite of redundant examination and anamnesis (Table 1).

In this interview study with 25 strategically selected GPs from different parts of Sweden we explored the management of patients with sore throat and identified the residual uncertainty. Four strategies for coping with uncertainty were identified, one of which was “Adherence to guidelines”, and three were idiosyncratic: “Expanded control”, “Clinical picture and CRP” and “Unstructured”. The residual uncertainty differed for the different strategies: in the strategy “Adherence to guidelines” and “Clinical picture and CRP” uncertainty was avoided, in “Expanded control” fear of bad outcomes was balanced, but in the strategy “Unstructured” uncertainty prevailed and the GP was out of control. To our knowledge no earlier study has explored strategies for coping with uncertainty in relation to a clinical problem in primary care. This study gives new insights into how the inherent uncertainty in clinical practice is managed in relation to guidelines in patients with sore throat in primary care.

A maximum variety sampling of participants from different parts of Sweden was achieved. Also, none of the invited GPs declined to be interviewed. The analysis with both inductive and deductive elements required iterative close readings of the interviews, which increases the validity of the results [16].

One weakness of the study is that four different interviewers may have decreased the reliability of the interviews. Still, all used the interview guide and the interviews were read and discussed continuously in order to reach consistency. Four of the interviewers had been involved in implementing the sore throat guidelines and thus could have been perceived as experts on the informants, which may have biased the interviews.

In this study we used expressions of fear in connection with DD or complications as a measure of concern for bad outcomes. In the interview manual questions of difficulties and concerns in patients with sore throat were asked but did not include direct questions about how GPs tolerated uncertainty. Earlier studies on physicians’ reactions to uncertainty used the “Physician reaction to uncertainty” scale questions, which besides concerns for bad outcomes also measures anxiety due to uncertainty and reluctance to disclose uncertainty to patients and physicians [1]. Thus the result of our study is not directly comparable with studies where the scale was used. However, this qualitative study deepens the understanding of the clinical practice with regard to managing uncertainty in primary care.

Near-patient tests (RADT and CRP) have been used for more than two decades in the Nordic countries, and the generalizability of this study might seem low to countries where the test is not in use. However, the test is being introduced in more countries, and therefore the result of our study might be relevant also for other countries. Finally, no observation of actual behaviour was included in the study. A link between what you say and what you do may not always be straightforward.

Scientific medicine is described as a symbolic system for coping with fears and uncertainties in medicine [1]. For the absolute majority of GPs uncertainty was avoided and our study thus confirms earlier findings [1, 5, 7]. However, the result may have differed if a more complex patient problem been studied. If the diagnosis seems unclear, more complications remain possible, thus diagnostic uncertainty and fear of possible complications seem to be related [8]. One consistent theme in studies of uncertainty is fear of personal inadequacy and failure. Young physicians must learn to control uncertainty or to be paralysed by it [1]. Denial and avoidance of uncertainty is a strategy to tolerate uncertainty together with physicians’ propensity to resolve uncertainty by action rather than inaction [1, 5].

For patients as common as those with a sore throat, GPs have to develop a strategy that decreases uncertainty to be able to last in their work. A greater stress of uncertainty was indicated by the GPs using both the strategy “Expanded control”, where uncertainty was managed by an excessive workload, and the strategy “Unstructured”, where uncertainty was not controlled. Personality traits seem related to diagnostic reasoning, and in one study of GPs the personality trait Neuroticism was associated with more anxiety due to uncertainty [17]. Stress caused by uncertainty correlates with work satisfaction and a higher risk of burnout [18]. Thus the GPs using the strategies “Expanded control” and “Unstructured” seem at risk of not coping with work in the long run.

All GPs knew of the guidelines but only one third of the interviewed GPs used the strategy “Adherence to guidelines” and felt no uncertainty. In the light of the systematic work carried out in Sweden with distributed guidelines in printed short version and regular outreach visits to all health centres this looks like a small number [19]. However, as the guideline did not include a dominating conception among GPs using the strategy “Clinical picture and CRP”, namely unspecified bacterial infection, it was not perceived relevant for their clinical practice. A similar mismatch between the content of the guideline and the conception of GPs might explain the results from the German study of GPs by Schneider et al. The use of guidelines did not contribute to understand the GPs strategies to manage uncertainty, which the authors suggest might be due to the gap between guidelines and the actual clinical practice [4].

In this study an excess of laboratory tests compared to the guidelines was described in all strategies except “Adhering to guidelines”. In the strategy “Clinical picture and CRP” the result of the CRP tests induced certainty in the GPs. However, it could be said that this was a false sense of certainty. There are only two small studies describing CRP in patients with sore throat, and therefore the evidence to interpret the result lacks validity [20, 21]. Perhaps this delight in CRP is an expression of “Our stubborn quest for security” as Kassier put it [22]. He calls for us to weed out tests that are ineffective or poorly predictive.

Tolerance of uncertainty seems to affect test-ordering behaviour [3]. Increase of diagnostic activity was found in the strategy “Expanded control” as in an earlier study of German GPs [4]. Risk-adverse GPs in France more often used RADT and more often prescribed antibiotics when RADT was not used compared to risk-tolerant GPs [23]. In our study no measures of tolerance of uncertainty were included.

GPs have to cope with uncertainty, especially in common patient problems like sore throat. A few GPs did not manage to do so or had to put in excessive work to control uncertainty with risk of lower work satisfaction and burnout. For these GPs reflective writing or case-based reasoning might be a remedy [12, 24, 25]. The majority of the interviewed GPs, however, had no feeling of uncertainty when dealing with patients with a sore throat. Although the greater part of these GPs used a strategy, which did not adhere to guidelines, they deemed that they did not have any problem. To change habits and preconceptions in situations like that is not easy. Addressing current conceptions and fears of bad concerns in new or revised guidelines might be one way forward. Small group discussions together with feedback and benchmarking of performance indicators might be another way to go.

The majority of the GPs avoided uncertainty and deemed they had no problems. Their strategies either adhered to guidelines or comprised excessive use of tests. Thus use of guidelines as well as use of more near-patient tests seemed associated to reduced uncertainty, although the later strategy at the expense of compliance to guidelines. A few GPs did not manage to cope with uncertainty or had to put in excessive work to control uncertainty.