Diagnostic Accuracy of a Device for the Automated Detection of Diabetic Retinopathy in a Primary Care Setting.

OBJECTIVE: To determine the diagnostic accuracy in a real-world primary care setting of a deep learning-enhanced device for automated detection of diabetic retinopathy (DR). RESEARCH DESIGN AND METHODS: Retinal images of people with type 2 diabetes visiting a primary care screening program were graded by a hybrid deep learning-enhanced device (IDx-DR-EU-2.1; IDx, Amsterdam, the Netherlands), and its classification of retinopathy (vision-threatening [vt]DR, more than mild [mtm]DR, and mild or more [mom]DR) was compared with a reference standard. This reference standard consisted of grading according to the International Clinical Classification of DR by the Rotterdam Study reading center. We determined the diagnostic accuracy of the hybrid deep learning-enhanced device (IDx-DR-EU-2.1) against the reference standard. RESULTS: A total of 1,616 people with type 2 diabetes were imaged. The hybrid deep learning-enhanced device's sensitivity/specificity against the reference standard was, respectively, for vtDR 100% (95% CI 77.1-100)/97.8% (95% CI 96.8-98.5) and for mtmDR 79.4% (95% CI 66.5-87.9)/93.8% (95% CI 92.1-94.9). CONCLUSIONS: The hybrid deep learning-enhanced device had high diagnostic accuracy for the detection of both vtDR (although the number of vtDR cases was low) and mtmDR in a primary care setting against an independent reading center. This allows its' safe use in a primary care setting.

Multivariable risk-based patient selection for prostate biopsy in a primary health care setting: referral rate and biopsy results from a urology outpatient clinic.

BACKGROUND: According to their guidelines, Dutch general practitioners (GPs) refer men with prostate-specific antigen (PSA) level ≥3.0 ng/mL to the urologist for risk-based patient selection for prostate biopsy using the Rotterdam Prostate Cancer Risk Calculator (RPCRC). Use of the RPCRC in primary care could optimize the diagnostic pathway even further by reducing unnecessary referrals. To investigate this, we calculated the risk and assessed the rate of men referred to the urologist with PSA level ≥3.0 ng/mL by implementing the RPCRC in a primary health care setting. METHODS: In January 2014, an exploratory study was initiated in collaboration with the primary health care facility of the GP laboratory in Rotterdam. GPs were given the possibility to refer men with a suspicion of prostate cancer (PCa) or a screening wish to this primary care facility (STAR-SHL) where further assessment was performed by specially trained personnel. Risk-based advice on referral to the urologist was given to the GP on the basis of the RPCRC results. If requested, advice on the treatment of lower urinary tract symptoms (LUTS) was provided. All men signed informed consent. RESULTS: Between January 2014 and September 2017, a total of 243 men, median age 64 [interquartile range (IQR), 57-70] years were referred for a consultation at the primary care facility. Of the 108 men with PSA level ≥3.0 ng/mL and a referral related to PCa, GPs were advised to refer 58 men to the urologist (54%). Of the men with available follow-up (FU) data [n=187, median FU, 16 (IQR, 9-25) months] 54 men were considered high-risk (i.e., had an elevated risk of PCa as calculated by the RPCRC). Of these men, 51 (94%) were actually referred to secondary care by their GP, and so far 38 men underwent biopsy. PCa was detected in 30 men [47% had Gleason score (GS) ≥3+4 PCa], translating to an overall positive predictive value (PPV) of 79%. Within the available FU time, 2 out of 38 (5%) men with PSA level ≥3.0 ng/mL which were considered low-risk have been diagnosed with GS 3+3 PCa. CONCLUSIONS: Risk-stratification with the RPCRC in a primary health care setting could prevent almost half of referrals of men with PSA level ≥3.0 ng/mL to the urologist. In more than three-quarters of men referred for prostate biopsy, the suspicion of PCa was confirmed and almost half of men had clinically significant PCa (GS ≥3+4 PCa). These data show a huge potential for multivariable risk-stratification in primary care.