Setting and Patients
The clinical laboratory is located in a 370-bed suburban university community hospital that serves the population of the health department (HD) (234,551 inhabitants).
Primary care laboratory requests are made electronically from the patient's electronic medical record (EMR) by the GPs. Laboratory requests are made through an electronic system, the CPOE, which prints the recommendations automatically in the primary care doctor's office, according to the requested tests, and offers the GPs a field to complete information on the reason for the request (clinical history/diagnostic suspicion), using the International Classification of Diseases, Ninth Revision, Clinical Modification (CIE-9-MC) codes . It is not mandatory, but is optional.
The primary care patients are phlebotomized in different primary care centers (PCCs); the samples are collected by couriers and delivered to the laboratory sample reception desk. Samples arrive at the laboratory ≤ 3 h after phlebotomy .
The reports are automatically sent from the LIS to the EMR. The tests related to diabetes can be requested in an individualized manner.
We included all primary care patients of the HD covered by the clinical laboratory.
Regarding the first objective (detecting new cases of type 2 diabetes and prediabetes through the LIS automatic HbA1c register), a prospective study was designed from 1 May 2017 to 31 May 2018.
From 1 July 2016 to 31 May 2018, a prospective study was also designed to study the second aim (improvement of diabetes monitoring through the LIS automatic HbA1c register, lipid tests and urinary albumin analyses).
Three-milliliter blood samples were collected from each of the subjects into BD Vacutainer® K2EDTA tubes (Becton, Dickinson and Co., Franklin Lakes, NJ, USA) to analyze HbA1c levels on a Tosoh G8 HPLC Analyzer, ion-exchange HPLC instrument [NGSP-certified, anchored to the Diabetes Control and Complication Trial (DCCT) reference study and to the IFCC Reference Method].
Glucose concentration was measured from serum collected in BD Vacutainer® serum separating tubes II Advance Tube (SST) (Becton, Dickinson and Co., Franklin Lakes, NJ, USA) using the hexokinase method with a Cobas 8000® Chemistry System (Roche Diagnostics, Basel, Switzerland).
Cholesterol, cHDL and triglyceride levels were also measured (Cobas 8000, Roche, Mannheim, Germany) in serum samples through an enzymatic colorimetric method, and cLDL was calculated using the Friedewald formula.
The quantitative urinary albumin was measured (Cobas 8000, Roche, Mannheim, Germany) with an immunoturbidimetric assay. The albumin-to-creatinine ratio (ACR) strip test is based on the protein error of the pH indicator (tetrabromophenol blue) and measured in UC-3500 (Sysmex, Kobe, Japan).
In a first stage, a meeting was held between the laboratory professionals and the GP in charge of one of the PCCs to discuss strategies—first to find new interventions to detect new cases of type 2 diabetes and prediabetes and second to adhere to laboratory diabetes monitoring guidelines.
Aim 1: Detection of New Cases of Diabetes
Regarding the first study purpose, since a strategy for patients > 45 years old was already established in 2013 , the team agreed that the LIS would automatically register HbA1c values on the laboratory request of any primary care patient > 25 and < 46 years old with cHDL values < 35 mg/dl (0.90 mmol/l) and/or triglyceride levels > 250 mg/dl (2.83 mmol/l) [1, 2] when the former was not requested in either the current order or during the previous year and fasting glucose was > 100 mg/dl. We chose the fasting glucose threshold, in consensus with GPs, because of economic constraints, since our laboratory belongs to the public sector, and because of our prior experience  in type 2 diabetes detection. The final diagnosis of type 2 diabetes was based on HbA1c values ≥ 6.5% (48 mmol/mol) and fasting glucose ≥ 126 mg/dl (7 mmol/l) and prediabetes when HbA1c = 5.7–6.4% and fasting glucose = 100–125 mg/dl. When just one of the two tests was abnormal, the LIS would recommend a second glucose and HbA1c request to confirm the presence, or not, of diabetes. In the screened population and for the detected cases, median values for HbA1c and fasting glucose were reported, to compare results between patients with diabetes and prediabetes. Finally, we counted the number of newly diagnosed cases (diabetes and prediabetes) and calculated the total economic cost per identified patient, taking into account the total number of additional HbA1c tests (N HbA1c) and the cost of reagents (1.15€ per HbA1c test) (cost per identified patient = N HbA1c × 1.15€/number of identified patients).
Aim 2: Improvement of Diabetes Monitoring
To improve diabetes monitoring, the pathologists and GPs decided that HbA1c, cholesterol, cHDL, cLDC, triglyceride and ACR values would be automatically registered by the LIS in primary care patients with a laboratory order for diabetes (CIE-9-MC codes) (6) when those tests were not demanded in the current request, and also had not been previously requested in the recommended period of time, according to guideline recommendations. We reported the number of additionally registered tests according to the strategy design and their median values. We also calculated the total economic cost, taking into account the total number of additional HbA1c, lipid and ACR tests and the cost of reagents (1.15€ per HbA1c test; 0.09€ per cholesterol test; 0.68€ per cHDL test, 0.24€ per triglyceride test and 1.31€ per ACR test when quantified and 0.04€ when just measured through strip analysis).
Thereafter, a report was sent to the GPs coordinators of the different PCCs informing them about the proposed strategies, and a 2-month reflection period was established to communicate the strategies to every GP in the nine different PCCs. In the last meeting, these strategies to detect new cases of type 2 diabetes and prediabetes and to adhere to laboratory diabetes monitoring guidelines were approved. The study was approved by the Hospital Research Committee, and the need for consent was waived.
All procedures performed in studies involving human participants were in accordance with Hospital San Juan Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
Data were statistically processed and analyzed using the SPSS 22.0 software package, and Microsoft Excel. Continuous data were tested for normal distribution using the Kolmogorov-Smirnov test. Descriptive statistics were presented as median and interquartile range (IQR) and percentages for continuous data and categorical data. The comparative study by two percentages was done by way of a U Mann-Whitney test or Kruskal–Wallis test analysis, as appropriate. A two-sided P ≤ 0.05 rule was utilized as the criterion for rejecting no difference.