Background

Rapid molecular diagnostic tests for pneumonia pathogens may improve antimicrobial stewardship (AMS). Results become available within 1 to 6 h, with accurate detection of multiple respiratory bacteria, viruses, and antimicrobial resistance genes directly from respiratory secretions without culture [1]. Commercially available rapid tests for pneumonia presently comprise the BioFire FilmArray Pneumonia panel (bioMérieux), the Unyvero Hospitalised Pneumonia panel (Curetis) and the FTD Respiratory Pathogens 33 (Fast Track Diagnostics) [1]. Other tests are in development.

Such tests may be particularly useful in intensive care units (ICUs), where patients with pneumonia are frequent [2, 3], there is an increase the risk of rapid deterioration and death [3, 4], and thus there is a demand for urgent antimicrobial treatment [5, 6]. Recommended practice for a suspected pneumonia include the prescription of empiric broad-spectrum antibiotics, with refinement once results of laboratory cultures become available (typically after 48–72 h) [5]. This approach is improvable because: (a) unnecessary antibiotics increase the risk of adverse consequences including direct toxicity, drug interactions, and Clostridium difficile infection [7], (b) empirical cover may prove ineffective for patients with drug resistant organisms [8], and (c) liberal broad-spectrum use drives antimicrobial resistance (AMR).

Molecular diagnostics can identify pathogens and their resistance genes within hours rather than days, potentially directing early tailored antimicrobial therapy [9]. Randomised-control trials (RCTs) are comparing outcomes and AMS in molecular diagnostic and conventional microbiology-guided antimicrobial treatment for pneumonias [1]. For example, INHALE (ISRCTN16483855) has implemented the evaluation of point-of-care molecular tests for pneumonias in 12 UK ICUs [10].

Clinicians’ endorsement and uptake of molecular tests is crucial to maximise engagement in the context of such a RCT and for any subsequent deployment. Yet, adoption may be impeded if users harbour unaddressed concerns or if device usage is incompatible with local practice [11]. For instance, a recent RCT found that a highly sensitive rule-out test did not improve AMS for ICU patients with suspected VAP, likely due to incongruity with local prescribing culture [12].

Clinicians’ beliefs about molecular diagnostics remain largely unknown, with the UK Department of Health and Social Care identifying a lack of understanding frontline needs as a potential delayer to the clinical adoption of molecular tests [13]. Accordingly, we explored intensivists’ beliefs of molecular diagnostics before the tests’ implementation for the INHALE RCT.

Methods

This study is reported following Standards for Reporting Qualitative Research guidelines [14]. It employed vignette-based interviews (VBIs) using an interpretivist approach to understand clinicians’ beliefs of molecular diagnostics as a decision aid. The data are derived from the same interview transcripts as Pandolfo and colleagues [15], but were analysed separately and meet Fine and Kurdek’s criteria for publishing multiple reports from one dataset [16].

Setting

Interviews occurred in four UK ICUs that varied in patient population and prevalence of multi-drug resistant organisms. One was a large district general hospital in Norfolk; the remainder were London-based and comprised a tertiary referral hospital, a paediatric hospital, and a private hospital with an international patient population. None had molecular diagnostic tests for pneumonia available at the time of this research.

Inclusion criteria

All intensivists practicing at the four participating ICUs were eligible to participate. Clinicians who could spare time from clinical duties were recruited via local promotion after ward rounds.

This research received Health Research Authority approval before data collection; all participants gave written informed consent.

Vignette-based interview methodology

Semi-structured VBIs explored clinicians’ antibiotic decision-making processes. Participants read a vignette depicting a hypothetical patient exhibiting signs of pneumonia (see Additional file 1 for vignette and interview guide). They then applied their expertise to determine whether to wait or to start antimicrobial treatment. We explored the perceived utility of molecular test results for this patient, followed by further detailed discussions of issues raised. Since participants may not have been familiar with specific devices, interviews asked generally about molecular tests for pneumonia. We posited a notional six-hour result turnaround time because this is achievable with two devices considered for the INHALE RCT. The vignette was pilot-tested with five non-participating ICU consultants and had no ‘correct’ answers because its purpose was to encourage reflection on decision-making processes.

Interviews were audio-recorded and conducted face-to-face in each site’s ICU by NB and YJ. Both have clinical pharmacy backgrounds, with qualitative research and interviewing experience.

Analysis

Interviews were anonymised, professionally transcribed, and entered into NVivo V.12. AMP (a research psychologist), NB, and YJ verified transcription accuracy. Data were analysed by AMP and NB using thematic analysis, following Braun and Clarke’s recommendations [17].

We adopted an inductive approach whereby data were coded to capture themes that represented a pattern of responses across transcripts. Themes were then deductively mapped onto the Necessity-Concerns Framework (NCF). NCF proposes that patients’ adherence to recommended treatment plans is influenced by their beliefs about the importance of their treatment and their treatment concerns [18]. Consistent with Pandolfo and colleagues [15], we adapted the principles of NCF to explore intensivists’ beliefs regarding molecular diagnostics as a decision aid for their patients’ treatments. ‘Molecular Diagnostic Necessity’ refers to clinicians’ perceived importance of molecular diagnostic results in practice while ‘Molecular Diagnostic Concerns’ are their beliefs about the consequences associated with test adoption [19].

Results

Interviews occurred between August and December 2018, were between seven and 20 min in length, and enrolment continued until data saturation [20]. Total audio-recording duration was approximately 4.5 h.

Participants comprised seven early-career trainees, sixteen middle-grade trainees, and eleven consultants. Eleven participants were employed at Hospital 1, ten at Hospital 2, seven at Hospital 3, and six at Hospital 4. All intensivists practicing at Hospital 2 treated children and neonates; remaining participants treated adult patients. Interviews were individual except in one case where one early-career trainee and one middle-grade trainee were interviewed together.

We describe clinicians’ perceived importance of molecular diagnostics followed by their concerns. Tables 1 and 2 show sub-themes and supporting quotations for Molecular Diagnostic Necessity and Molecular Diagnostic Concerns themes, respectively. There were no notable differences between paediatric- and adult-treating doctors’ beliefs.

Table 1 Supporting quotations for Molecular Diagnostic Necessity theme
Full size table
Table 2 Supporting quotations for Molecular Diagnostic Concerns theme
Full size table

Molecular Diagnostic Necessity: molecular diagnostic results improve antibiotic prescription practices

Positive results facilitate choosing targeted antibiotics

Positive molecular diagnostic results (i.e., detection of bacterial pathogen by the test) were generally believed to facilitate antibiotic choice by rapidly identifying the organism(s) and predicting their antibiotic susceptibilities (Quotes 1–3). Currently, this information would be available approximately 72 h after the initial prescribing decision, leading to lengthy courses of potentially sub-optimal therapy. Many participants felt that molecular test results would encourage either starting appropriate antibiotics at the outset or swiftly de-escalating empirical therapy to narrow-spectrum antibiotics.

Positive results lower threshold for starting antibiotics

In addition to improving antibiotic choice, ten clinicians asserted that positive results would lower their threshold for starting antibiotics. For example, one trainee likened molecular diagnostics to other infection indicators (e.g., white cell count in blood) which give “proof” of an infection but should be combined with the clinical context before making an antimicrobial decision (Quote 4). This decision was frequently described as reflecting a combination of clinical factors, including prospectively, the molecular test result. None of our participants said that they would start antibiotics solely based on positive results (Quote 5).

Some doctors stated that molecular diagnostics would not influence their threshold to start antimicrobial therapy (Quote 6). These clinicians believed that they would exclusively use this test to choose appropriate antibiotics because their decision to start antibiotics would only be based on the clinical context (Quote 7).

Negative results indicate absence of respiratory infection

Negative results (i.e., no detection of bacteria or resistance genes) would generally be interpreted to indicate that a respiratory infection was unlikely. Some clinicians believed that a negative result in a clinically stable patient would encourage withholding or stopping antibiotics (Quotes 8–9), whereas a negative result in a deteriorating patient would be interpreted to indicate a non-respiratory source of infection (Quote 9).

Molecular diagnostic results increase confidence in prescribing decisions

When currently making antibiotic decisions, some participants reported experiencing negative emotions like “angst” and “panic” (Quotes 10–11). They described uncertainty regarding whether their prescribing was appropriate and worried that it contributed to AMR (Quotes 10–11). This reflects the present difficulty of verifying the appropriateness of antibiotic decisions until laboratory culture results become available.

Such participants believed that the information provided by the molecular test would make them “happier” and more “confident” in their prescribing (Quotes 10–11). They believed that their prescribing decisions would be beneficial to their patient and society and felt that access to these tests would encourage AMS practices.

Molecular Diagnostic Concerns: integrating molecular diagnostics into practice

Unfamiliarity with molecular diagnostic test capabilities

Many clinicians stated that they needed to familiarise themselves with the molecular test before using its results in practice. First, they wanted more information about the test, including its sensitivity, specificity, and its place in the diagnostic process (Quote 12). Several raised misconceptions about the test’s capabilities (e.g., it would prove the absence of all possible resistance genes and mutations; Quote 10). Second, they wanted first-hand experience to verify the test’s suitability (Quote 13). These participants felt a sense of caveat emptor—as future users of the test’s results, these clinicians believed that familiarisation was essential to understand the test’s capabilities and to make an informed decision about its usefulness in practice (Quotes 13–14). These participants would not adopt the test if, after familiarisation, they concluded that it did not meet their standards or would negatively impact patient care.

Molecular diagnostics detecting non-pathogenic bacteria may lead to over-treatment

Concerns were common that the molecular test would detect organisms present, but of no consequence to the patient (Quote 15). Many clinicians worried that the test would detect non-pathogenic bacteria, and that, paradoxically, this might encourage unnecessary antibiotic prescriptions (Quote 16). To mitigate against antibiotic over-treatment based on detection of colonisation rather than infection, some participants stressed that they would not automatically prescribe based on the test result. They would combine the results with the clinical context to determine the likelihood that the bacteria detected were causing current disease before making a decision (Quote 17).

Molecular diagnostics failing to detect pathogens may lead to under-treatment

Some clinicians reported that negative results would not be reassuring enough to withhold or stop antibiotics. These doctors were uncertain whether the test was able to detect all possible respiratory pathogens (Quote 18). They consequently believed that negative results should be overridden by other evidence of respiratory infection (e.g., chest X-ray consolidation), which would lead them to prescribe or continue antibiotics (Quote 19).

Concern of patient deterioration while awaiting molecular diagnostic results

Many early-career and middle-grade trainees assumed that they would need to withhold antibiotics until molecular test results become available; no consultants took this view. Consequently, some junior doctors expressed discomfort with the prospect of withholding antibiotics for up to 6 h. They worried that both the vignette and their actual patients would significantly deteriorate within this period (Quotes 20–21), believing that withholding antibiotics would be indefensible if the patient was exhibiting signs of sepsis (Quote 22).

Other trainees felt more comfortable with withholding antibiotics until receiving molecular test results. These participants believed that the patient was stable enough to ‘wait and watch’ and valued using the results to aid their ultimate prescription decision (Quotes 23–24).

Discussion

This is the first study exploring UK intensivists’ salient beliefs influencing molecular diagnostic test uptake. Consistent with the NCF, facilitators related to clinicians’ beliefs that molecular tests would improve AMS whereas barriers related to their concerns about integrating these tests into clinical practice [19].

Many participants were open to using the molecular test because they believed that its results would improve antibiotic prescription practices. They saw positive test results, with early specific identification of a pathogen, as facilitating targeted antimicrobial therapy. Similarly, they viewed negative results to signal the likely absence of respiratory infection and thought that these results would provide reassurance that withholding an antibiotic prescription was appropriate. Our findings are consistent with arguments that molecular diagnostics would aid the optimisation of antimicrobial therapy [1, 9], and that it could balance the competing priorities of individual patients with AMR [21, 22].

Despite clinicians recognising the potential of molecular tests, concerns about tests’ integration into clinical practice may impede uptake. Consistent with previous assertions [1, 21], participants worried that a molecular test’s high sensitivity would lead to reporting of colonising bacteria, which, paradoxically, could prompt unnecessary antimicrobial therapy. They also raised concerns about unfamiliarity with the test, misapprehensions about its capabilities, uncertainty whether it would fail to detect pathogens, and worry about withholding antibiotics until receiving test results. Clinicians may also hold these beliefs regarding the ability of conventional microbiology to find organisms, especially when antibiotics have been deployed [23].

Individual and environmental factors may also influence prescribers’ acceptance of molecular diagnostics. Individual clinicians’ trust in molecular diagnostics may affect its usage; for example, prescribers with low levels of trust in the machine likely would harbour concerns while having low beliefs about the device’s utility. Similarly, the prescribing environment may affect machine uptake. For instance, unit-specific norms may encourage continuing empiric antibiotics for a minimum course; as such, receiving rapid test results may not impact antibiotic de-escalation or stopping. Thought should also be given to machine placement, whether at the point-of-care or in the laboratory.

Our findings show the importance of understanding clinicians’ beliefs about molecular diagnostic tests to ensure that they understand their rationale, and to sufficiently address their concerns. To encourage molecular diagnostic uptake in ICU—whether for RCTs or for permanent use—implementation strategies should bolster users’ necessity beliefs while reducing concerns; this may be achieved with relatively simple solutions (summarised in Table 3).

Table 3 Recommendations to facilitate molecular test uptake
Full size table

This study has limitations. Firstly, our interviews were conducted before the molecular diagnostic test was deployed for the INHALE trial. Different beliefs may emerge after implementation and should be explored. Secondly, we interviewed ICU intensivists as the INHALE RCT runs with the molecular test located in ICUs; however some clinicians have recommended collaborating with microbiologists during test implementation [21, 24]. Future research will investigate ICU microbiologists’ beliefs of molecular diagnostics to understand their barriers and facilitators to adoption. The INHALE team has also collaborated with unit microbiologists to create site-specific prescribing algorithms that aid translation of machine output to prescribing advice. Thirdly, clinicians’ beliefs may differ for non-respiratory molecular tests and should be examined in future research. Lastly, while we have endeavoured to sample from diverse hospitals, we recognise that clinicians’ beliefs may vary elsewhere in the UK and in other countries.

Conclusions

This study is the first to characterise prescribers’ beliefs about molecular diagnostics as facilitators and barriers to applying it to antibiotic prescribing for patients with suspected pneumonia. Our findings demonstrate the importance of prescribers’ beliefs as determinants of molecular diagnostic uptake. If this technology proves successful in improving antibiotic prescribing and stewardship, interventions should understand and address these beliefs to ensure the optimal application of molecular diagnostics in clinical practice.