Evaluation of an artificial intelligence-based decision support for the detection of cutaneous melanoma in primary care: a prospective real-life clinical trial.

BACKGROUND: Use of artificial intelligence (AI), or machine learning, to assess dermoscopic images of skin lesions to detect melanoma has, in several retrospective studies, shown high levels of diagnostic accuracy on par with - or even outperforming - experienced dermatologists. However, the enthusiasm around these algorithms has not yet been matched by prospective clinical trials performed in authentic clinical settings. In several European countries, including Sweden, the initial clinical assessment of suspected skin cancer is principally conducted in the primary healthcare setting by primary care physicians, with or without access to teledermoscopic support from dermatology clinics. OBJECTIVES: To determine the diagnostic performance of an AI-based clinical decision support tool for cutaneous melanoma detection, operated by a smartphone application (app), when used prospectively by primary care physicians to assess skin lesions of concern due to some degree of melanoma suspicion. METHODS: This prospective multicentre clinical trial was conducted at 36 primary care centres in Sweden. Physicians used the smartphone app on skin lesions of concern by photographing them dermoscopically, which resulted in a dichotomous decision support text regarding evidence for melanoma. Regardless of the app outcome, all lesions underwent standard diagnostic procedures (surgical excision or referral to a dermatologist). After investigations were complete, lesion diagnoses were collected from the patients' medical records and compared with the app's outcome and other lesion data. RESULTS: In total, 253 lesions of concern in 228 patients were included, of which 21 proved to be melanomas, with 11 thin invasive melanomas and 10 melanomas in situ. The app's accuracy in identifying melanomas was reflected in an area under the receiver operating characteristic (AUROC) curve of 0.960 [95% confidence interval (CI) 0.928-0.980], corresponding to a maximum sensitivity and specificity of 95.2% and 84.5%, respectively. For invasive melanomas alone, the AUROC was 0.988 (95% CI 0.965-0.997), corresponding to a maximum sensitivity and specificity of 100% and 92.6%, respectively. CONCLUSIONS: The clinical decision support tool evaluated in this investigation showed high diagnostic accuracy when used prospectively in primary care patients, which could add significant clinical value for primary care physicians assessing skin lesions for melanoma.

The use and usefulness of point-of-care tests in patients with pharyngotonsillitis - an observational study in primary health care.

BACKGROUND: Rapid antigen detection tests (RADT) for Group A streptococci (GAS) and point-of-care tests for C-reactive protein (CRP) are commonly used in patients with pharyngotonsillitis in Sweden and Denmark although CRP testing is not supported by guidelines. We aimed to describe (1) the proportion of patients tested with RADT and/or CRP, (2) the relation between test results and antibiotic prescribing, and (3) the association between CRP level and microbial aetiology. METHODS: We used a post-hoc-analysis of data collected in primary health care in a prospective aetiological study of 220 patients 15-45 years old diagnosed with pharyngotonsillitis. The outcomes of RADTs and CRP tests were related to antibiotic prescribing and microbial aetiology. RESULTS: A RADT was used in 94% of the patients. A CRP test was used in 50% of the patients but more commonly in those with a negative RADT (59%) than in those with a positive RADT (38%) (p = 0.005). Most (74%) CRP tests were used in patients with a negative RADT. Antibiotic prescribing differed greatly between patients with a positive RADT (96%) and patients with a negative RADT (17%) (p < 0.001). In patients with a negative RADT, there was a positive association between CRP value and antibiotic prescribing (OR 1.05; 95% CI 1.02-1.07; p < 0.001). Patients with CRP values ≤ 30 mg/l were seldomly prescribed antibiotics. Patients with GAS in culture had the highest median CRP (46 mg/l), which was higher than in patients without GAS (8 mg/l; p < 0.001). However, the positive predictive value for GAS never exceeded 0.60 (95% CI 0.31-0.83) at the investigated CRP levels. CONCLUSIONS: The widespread use of tests is a major deviation from national guidelines. Most CRP tests were used in patients with a negative RADT, suggesting a belief in the added value of a CRP test, and the CRP result seemed to influence antibiotic prescribing. However, as an aetiological test, CRP is not useful for predicting GAS.

Seasonal variations in use and outcome of rapid antigen detection tests and cultures in pharyngotonsillitis: a register study in primary care.

BACKGROUND: Diagnosis and treatment of pharyngotonsillitis are commonly focused on group A streptococci (GAS), although the disease is often associated with other pathogens. While the incidence of pharyngotonsillitis is known to vary with season, seasonal variations in the prevalence of potential pathogens are sparsely explored. The aim of this study was to explore any seasonal variations in the use and outcome of rapid antigen detection tests (RADTs) for GAS and throat cultures among patients diagnosed with pharyngotonsillitis in primary care. METHODS: We retrieved and combined retrospective data from the electronic medical record system and the laboratory information system in Kronoberg County, Sweden. Primary care visits resulting in a diagnosis of tonsillitis or pharyngitis were included, covering the period 2013-2016. The monthly rate of visits was measured, along with the use and outcome of RADTs for GAS and throat cultures obtained on the date of diagnosis. The variations between calendar months were then analysed. RESULTS: We found variations between calendar months, not only in the mean rate of visits resulting in a diagnosis of pharyngotonsillitis (p < 0.001), but in the mean proportion of RADTs being positive for GAS among the diagnosed (p < 0.001), and in the mean proportion of visits associated with a throat culture (p < 0.001). A lower mean rate of visits in August and September coincided with a lower proportion of RADTs being positive for GAS among them, which correlated with a higher proportion of visits associated with a throat culture. CONCLUSIONS: This study suggests that the role of GAS in pharyngotonsillitis in Sweden is less prominent in August and September than during the rest of the year.

The aetiology of pharyngotonsillitis in primary health care: a prospective observational study.

BACKGROUND: Few studies on pharyngotonsillitis have examined the clinical presentation of different aetiologies where pathogens have been detected using molecular methods. We aimed to assess how well clinical signs and symptoms can predict (1) the presence or absence of a broad range of viruses and bacteria, and (2) reconsultations for a sore throat or a complication. METHODS: In this descriptive observational prospective study in primary health care 220 patients aged 15-45 with suspected pharyngotonsillitis were sampled from nose, throat and blood and screened for 20 bacteria and viruses using polymerase chain reaction (PCR), culture and serology. Odds ratios (OR) and predictive values with 95% confidence intervals (CI) were used to show association between microbiological findings and clinical signs and symptoms. Patients were followed up after 3 months by reviewing electronic medical records. RESULTS: Both cough and coryza were more common in patients with only viruses (67%) than in patients with only bacteria (21%) (p < 0.001), whereas tonsillar coating was more common in patients with only bacteria (53%) than in patients with only viruses (29%) (p = 0.006). Tonsillar coating (adjusted OR 6.0; 95% CI 2.5-14) and a lack of cough (adjusted OR 3.5; 95% CI 1.5-8.0) were significantly associated with Streptococcus pyogenes (group A streptococci; GAS) and with any bacterial finding. A Centor score of 3-4 had a positive predictive value of 49% (95% CI 42-57) for GAS and 66% (95% CI 57-74) for any bacterial findings. The use of rapid antigen detection test for GAS increased the positive predictive value for this group to 93%. CONCLUSIONS: Signs and symptoms, both single and combined, were insufficient to rule in GAS or other pathogens. However, both cough and coryza were useful to rule out GAS. The results support the clinical approach of restricting rapid antigen detection testing to patients with 3-4 Centor criteria. The low carriage rate of bacteria among asymptomatic controls implied that most detections in patients represented a true infection.

Association between bacterial finding, antibiotic treatment and clinical course in patients with pharyngotonsillitis: a registry-based study in primary healthcare in Sweden.

BACKGROUND: The role of non-group A streptococci and Fusobacterium necrophorum in pharyngotonsillitis has been disputed and few prospective studies have evaluated the effect of antibiotic treatment. This study uses registry data to investigate the relation between antibiotic prescription for pharyngotonsillitis in primary healthcare and return visits for pharyngotonsillitis, complications, and tonsillectomy. METHODS: Retrospective data were extracted from the regional electronic medical record system in Kronoberg County, Sweden, for all patients diagnosed with pharyngotonsillitis between 2012 and 2016. From these data, two cohorts were formed: one based on rapid antigen detection tests (RADT) for group A streptococci (GAS) and one based on routine throat cultures for ß-haemolytic streptococci and F. necrophorum. The 90 days following the inclusion visit were assessed for new visits for pharyngotonsillitis, complications, and tonsillectomy, and related to bacterial aetiology and antibiotic prescriptions given at inclusion. RESULTS: In the RADT cohort (n = 13,781), antibiotic prescription for patients with a positive RADT for GAS was associated with fewer return visits for pharyngotonsillitis within 30 days compared with no prescription (8.7% vs. 12%; p = 0.02), but not with the complication rate within 30 days (1.5% vs. 1.8%; p = 0.7) or with the tonsillectomy rate within 90 days (0.27% vs. 0.26%; p = 1). In contrast, antibiotic prescription for patients with a negative RADT was associated with more return visits for pharyngotonsillitis within 30 days (9.7% vs. 7.0%; p = 0.01). In the culture cohort (n = 1 370), antibiotic prescription for patients with Streptococcus dysgalactiae ssp. equisimilis was associated with fewer return visits for pharyngotonsillitis within 30 days compared with no prescription (15% vs. 29%; p = 0.03). CONCLUSIONS: Antibiotic prescription was associated with fewer return visits for pharyngotonsillitis in patients with a positive RADT for GAS but with more return visits in patients with a negative RADT for GAS. There were no differences in purulent complications related to antibiotic prescription.

Presence of microorganisms in children with pharyngotonsillitis and healthy controls: a prospective study in primary healthcare.

PURPOSE: Most studies on paediatric pharyngotonsillitis focus on group A streptococci. This study, however, analyses a broad spectrum of bacteria and viruses related to paediatric pharyngotonsillitis and evaluates their associated clinical symptoms and courses. METHODS: This observational prospective study in primary healthcare includes 77 children aged < 15 with a sore throat and 34 asymptomatic children, all of whom were sampled from the tonsils with an E-swab® for analysis with culture and PCR for 14 bacteria and 15 viruses. Patients were evaluated clinically, and their symptoms recorded in diaries for 10 days. Participants were followed up for 3 months by reviewing medical records. RESULTS: A pathogen was detected in 86% of patients and in 71% of controls (P = 0.06). Bacteria were found in 69% of patients and 59% of controls (P = 0.3), and viruses in 36% and 26%, respectively (P = 0.3). Group A streptococci was the most common finding, with a prevalence of 49% and 32%, respectively (P = 0.1). Clinical signs were not useful for distinguishing pathogens. None of the controls and 16% of the patients reconsulted for a sore throat within 3 months. CONCLUSION: Bacteria were more common than viruses in both study groups. The high rate of pathogens in asymptomatic children interferes with diagnoses based on aetiology.

A 2-year follow-up study of patients with pharyngotonsillitis.

BACKGROUND: Longtime follow-up studies on patients with pharyngotonsillitis are rare. We aimed to describe the patterns of new visits for a sore throat, complications and tonsillectomy during 2 years in a cohort of patients with pharyngotonsillitis and non-infected controls. METHODS: A retrospective chart review was performed on a cohort of patients with acute sore throat (n = 207), and non-infected controls (n = 108). New visits, complications and tonsillectomy within 2 years was recorded and analyzed in relation to microbiological findings at inclusion. RESULTS: Patients with Group A streptococci (GAS) (12/66) reconsulted more often within 30 days than patients with no GAS (9/141) (p = 0.009) and patients with F. necrophorum (2/29). After 2 years, we observed no significant differences in reconsultations with regard to aetiology at inclusion. A single complication was recorded and 5 patients were planned for tonsillectomy. CONCLUSIONS: Group A streptococci were the sole aetiological agent associated with recurrent sore throat while F. necrophorum did not distinguish itself as a major cause of either recurrent infection or complications in this cohort. More studies, preferably with the focus on adolescents, are needed before F. necrophorum can be considered an important cause of pharyngotonsillitis.