Objective

We report results of a pilot clinical study of a subcutaneous fluorescence affinity sensor (FAS) for continuous glucose monitoring conducted in people with type 1 and type 2 diabetes. The device was assessed based on performance, safety, and comfort level under acute conditions (4 h).

Research Design and Methods

A second-generation FAS (BioTex Inc., Houston, TX) was subcutaneously implanted in the abdomens of 12 people with diabetes, and its acute performance to excursions in blood glucose was monitored over 4 h. After 30–60 min the subjects, who all had fasting blood glucose levels of less than 200 mg/dl, received a glucose bolus of 75 g/liter dextrose by oral administration. Capillary blood glucose samples were obtained from the finger tip. The FAS data were retrospectively evaluated by linear least squares regression analysis and by the Clarke error grid method. Comfort levels during insertion, operation, and sensor removal were scored by the subjects using an analog pain scale.

Results

After retrospective calibration of 17 sensors implanted in 12 subjects, error grid analysis showed 97% of the paired values in zones A and B and 1.5% in zones C and D, respectively. The mean absolute relative error between sensor signal and capillary blood glucose was 13% [±15% standard deviation (SD), 100–350 mg/dl] with an average correlation coefficient of 0.84 (±0.24 SD). The actual average “warm-up” time for the FAS readings, at which highest correlation with glucose readings was determined, was 65 (±32 SD) min. Mean time lag was 4 (±5 SD) min during the initial operational hours. Pain levels during insertion and operation were modest.

Conclusions

The in vivo performance of the FAS demonstrates feasibility of the fluorescence affinity technology to determine blood glucose excursions accurately and safely under acute dynamic conditions in humans with type 1 and type 2 diabetes. Specific engineering challenges to sensor and instrumentation robustness remain. Further studies will be required to validate its promising performance over longer implantation duration (5–7 days) in people with diabetes.

This paper provides an introduction to “human factors engineering,” an applied science that seeks to optimize usability and safety of systems. Human factors engineering pursues this goal by aligning system design with the perceptual, cognitive, and physical capabilities of users. Human factors issues loom large in the diabetes management domain because patients and health care professionals interact with a complex variety of systems, including medical device hardware and software, which are themselves embedded within larger systems of institutions, people, and processes. Usability considerations must be addressed in these systems and devices to ensure safe and effective diabetes management.

This article discusses the implications of the new Food and Drug Administration’s draft guidance on human factors and usability engineering for the development of diabetes-related devices. Important considerations include the challenge of identifying users, when the user population is so dramatically broad, and the challenge of identifying use environments when the same can be said for use environments. Another important consideration is that diabetes-related devices, unlike many other medical devices, are used constantly as part of the user’s lifestyle—adding complexity to the focus on human factors and ease of use emphasized by the draft guidance.

Despite disparities in health problems and outcomes, people with disabilities are underrepresented in diabetes research. This results in a lack of evidence-based knowledge regarding best approaches in caring for this population. This article addresses the need for research that includes people with disabilities and describes the common reasons persons with disabilities are not included in research, including scientists’ concerns regarding threats to a study’s internal validity and cost. Arguments are provided as to how involving people with disabilities in research will improve our science and reduce disparities in this population. In addition to the ethical reasons for including persons with disabilities in research, the ability to generalize study findings to this population and thus speed our development and translation of this knowledge for use by clinicians is discussed. The bias in study conclusions that arise from study samples that do not include persons with disabilities and its possible effect on care delivery are presented. Two strategies that researchers can use to increase the inclusion of persons with disabilities in research are described: (1) Universal Design of Research and (2) intervention optimization study designs. Universal Design of Research includes research design processes such as the use of multisensory formats for recruiting participants, approaches to designing and presenting research instruments and interventions, and methods of data collection to promote the inclusion of participants with a wide range of abilities in research studies. Intervention optimization study designs offer an efficient way for scientists to rapidly build the most potent interventions for a wide range of people, including those with disabilities participating in mainstream research.

In 2007, five blood glucose meters (BGMs) were introduced with integrated speech output necessary for use by persons with vision loss. One of those five meters had fully integrated speech output, allowing a person with vision loss independence in accessing all features and functions of the meter. In comparison, 13 BGMs with integrated speech output were available in 2011. Accessibility attributes of these 11 meters were tabulated and product design features examined.

All 13 meters were found to be usable by persons with vision loss to obtain a blood glucose measurement. However, only 4 of them featured the fully integrated speech output necessary for a person with vision loss to access all features and functions independently.

The vast majority of diabetes-related self-management technology utilizes small visual displays (SVDs) that often produce a low level of contrast and suffer from high levels of reflection (glare). This is a major accessibility issue for the 3.5 million Americans with diabetes who have reduced vision. The purpose of this article is to gather comparative data on the key display attributes of the SVDs used in blood glucose meters (BGMs) and home blood pressure monitors (HBPMs) on the market today and determine which displays offer the best prospect for being accessible to people with reduced vision.

Nine BGMs and eight HBPMs were identified for this study on the basis of amount of devices sold, full-functionality speech output, and advanced display technologies. An optical instrumentation system obtained contrast, reflection (glare), and font height measurements for all 17 displays.

The contrast, reflection, and font-height values for the BGMs and HBPMs varied greatly between models. The Michelson contrast values for the BGMs ranged from 11% to 98% and font heights ranged 0.39–1.00 in. for the measurement results. The HBPMs had Michelson contrast values ranging 55–96% and font height ranging 0.28–0.94 in. for the measurement results.

Due largely to the lack of display design standards for the technical requirements of SVDs, there is tremendous variability in the quality and readability of BGM and HBPM displays. There were two BGMs and one HBPM that exhibited high-contrast values and large font heights, but most of the devices exhibited either poor contrast or exceptionally high reflection.

Introduction

Nearly 20% of the adults with diagnosed diabetes in the United States also have visual impairment. Many individuals in this group perform routine diabetes self-management tasks independently, often using technology that was not specifically designed for use by people with visual impairment (e.g., insulin pumps and pens). Equitable care for persons with disabilities requires providing instructions in formats accessible for nonreaders. However, instructions in accessible formats, such as recordings, braille, or digital documents that are legible to screen readers, are seldom available.

Method

This article includes a summary of existing guidelines for creating accessible documents. The guidelines are followed by a description of the production of accessible nonvisual instructions for use of insulin pens used in a study of dosing accuracy. The study results indicate that the instructions were used successfully by 40 persons with visual impairment.

Discussion and Conclusions

Instructions in accessible formats can increase access to the benefits of diabetes technology for persons with visual impairment. Recorded instructions may also be useful to sighted persons who do not read well, such as those with dyslexia, low literacy, or who use English as a second language. Finally, they may have important benefits for fully sighted people who find it easier to learn to use technology by handling the equipment while listening to instructions. Manufacturers may also benefit from marketing to an increased pool of potential users.

This article discusses human factors (HF) processes and how they are applied during the development of a medical device to minimize the risk that the user interface design could lead to patient errors, adverse events, and product recalls. This process is best defined as “prevention through design.” The HF design process is exemplified by three distinct phases: (1) preliminary analysis, (2) formative design evaluation and modification, and (3) design validation. Additional benefits of employing HF principles during medical device development are briefly reviewed, including reduced patient risk by eliminating design flaws, increased patient adherence through the reduction in the complexity of therapeutic regimes, and reduced likelihood for product recalls.

Background

Aggregation of insulin into insoluble fibrils (fibrillation) may lead to complications for diabetes patients such as reduced insulin potency, occlusion of insulin delivery devices, or potentially increased immunological potential.

Even after extensive investigation of fibril formation in regular human insulin, there are little published data about the intrinsic fibrillation of fast-acting analogs. This article investigates and compares the intrinsic fibrillation of three fast-acting insulin analogs—lispro, aspart, and glulisine—as a function of their primary protein structure and exclusive of the stabilizing excipients that are added to their respective commercial formulations.

Methods

The insulin analogs underwent a buffer exchange into phosphate-buffered saline to remove formulation excipients and then were heated and agitated to characterize intrinsic fibrillation potentials devoid of excipient stabilizing effects. Different analytical methods were used to determine the amount of intrinsic fibrillation for the analogs. After initial lag times, intrinsic fibrillation was detected by an amyloid-specific stain. Precipitation of insulin was confirmed by ultraviolet analysis of soluble insulin and gravimetric measurement of insoluble insulin. Electron microscopy showed dense fibrous material, with individual fibrils that are shorter than typical insulin fibrils. Higher resolution kinetic analyses were carried out in 96-well plates to provide more accurate measures of lag times and fibril growth rates.

Results

All three analogs exhibited longer lag times and slower intrinsic fibrillation rates than human insulin, with glulisine and lispro rates slower than aspart. This is the first study comparing the intrinsic fibrillation of fast-acting insulin analogs without the stabilizing excipients found in their commercial formulations.

Conclusions

Data show different intrinsic fibrillation potentials based on primary molecular structures when the formulation excipients that are critical for stability are absent. Understanding intrinsic fibrillation potential is critical for evaluating insulin analog stability and device compatibility.

Insulin is susceptible to thermal fibrillation, a misfolding process that leads to nonnative cross-b assembly analogous to pathological amyloid deposition. Pharmaceutical formulations are ordinarily protected from such degradation by sequestration of the susceptible monomer within native protein assemblies. With respect to the safety and efficacy of insulin pumps, however, this strategy imposes an intrinsic trade-off between pharmacokinetic goals (rapid absorption and clearance) and the requisite physical properties of a formulation (prolonged shelf life and stability within the reservoir). Available rapid-acting formulations are suboptimal in both respects; susceptibility to fibrillation is exacerbated even as absorption is delayed relative to the ideal specifications of a closed-loop system. To circumvent this molecular trade-off, we exploited structural models of insulin fibrils and amyloidogenic intermediates to define an alternative protective mechanism. Single-chain insulin (SCI) analogs were shown to be refractory to thermal fibrillation with maintenance of biological activity for more than 3 months under conditions that promote the rapid fibrillation and inactivation of insulin. The essential idea exploits an intrinsic incompatibility between SCI topology and the geometry of cross-b assembly. A peptide tether was thus interposed between the A- and B-chains whose length was (a) sufficiently long to provide the “play” needed for induced fit of the hormone on receptor binding and yet (b) sufficiently short to impose a topological barrier to fibrillation. Our findings suggest that ultrastable monomeric SCI analogs may be formulated without protective self-assembly and so permit simultaneous optimization of pharmacokinetics and reservoir life.

Background

Postprandial hyperglycemia contributes to poor glucose control and is associated with increased cardiovascular risk in type 2 diabetes mellitus (T2DM). The objective of the study was to determine the effect of postprandial self-monitoring of blood glucose (pp-SMBG) on glucose control, lipids, body weight, and cardiovascular events.

Method

Subjects with T2DM hemoglobin A1c (A1C) between 6.5 to 7.0% were randomized into the study group (at least two pp-SMBG a day and dietary modification based on glucose readings) and control group (dietary modification based on glucose readings but no mandatory pp-SMBG) for a 6-month, observational study. Oral antidiabetic drugs or insulin regimen was unchanged in either group if A1C remained less than 7.0% during the study. End points included A1C, lipids, body weight, and cardiovascular events.

Results

One hundred sixty-nine subjects, mean age 63 years, and body weight 88 kg were recruited. Hemoglobin A1c, weight, low-density lipoprotein (LDL), and triglycerides (TGs) were similar in the groups at baseline. By the end of 6 months, A1C (6.7 ± 0.1 to 6.4 ± 0.1%, p < .05), body weight (88.5 ± 7.3 to 85.2 ± 6.3 kg, p < .05), LDL (92.3 ± 2 8.4 to 81.1 ± 22.6 mg/dl, p < .05), and TGs (141 ± 21 to 96 ± 17 mg/dl, p < .05) decreased in the study group, but did not change in the control group. No cardiovascular events were observed in either group during the 6-month study period.

Conclusions

In T2DM subjects who had already reached their A1C goal, pp-SMBG at least twice a day was associated with further improvement in glycemia, lipids, and weight, as well as exercise and dietary habit. We assume that lifestyle modification promoted by postprandial hyperglycemia awareness may underlie these findings. These results substantiate the importance of implementing pp-SMBG into lifestyle modification, and emphasize that pp-SMBG is critical in the control of T2DM.

Objective

Glycemic control is a rapidly developing field in intensive care medicine with the aim of reducing mortality, morbidity, and cost. Current intensive care unit (ICU) glucose measurement technologies are susceptible to interference from medications, volume expanders, and other substances present in critically ill patients. We hypothesized that a fixed-wavelength mid-infrared (mid-IR) spectroscopy system would be accurate for measuring glucose levels of ICU patients.

Research Design and Methods

This is a prospective investigation of plasma samples from two different institutions treating a heterogeneous population of ICU patients. The first 292 samples were collected from 86 patients admitted to Stamford Hospital, and the next 352 samples were collected from 75 patients from three ICUs at the University of Maryland. Plasma samples were measured on a Fourier-transform infrared or a proprietary spectrometer, with a glucose prediction algorithm to correct for spectral interference, which were compared with reference measurements taken using a YSI 2300 glucose analyzer.

Results

Glucose values ranged from 24 to 343 mg/dl. Numerous medications and injury/disease states were observed in the patient populations, with metoprolol, fentanyl, and multiple organ failure the most prevalent. Despite these interferents, there was a high correlation (r ≥ 0.94) and low standard error (≤12.8 mg/dl) between the predicted glucose values and those of the YSI 2300 STAT Plus reference instrument in the three studies. A total of 95.1% of the 644 values in the three studies met International Organization for Standardization 15197 criteria.

Conclusion

These results suggest that a fixed-wavelength mid-IR spectrometer can measure glucose accurately in the plasma of ICU patients.

Background

Prolonged severe hypoglycemia (SH) in hospitalized patients is associated with increased morbidity and mortality. This study was undertaken to identify risk factors for SH, to apply that knowledge to the development of a prediction algorithm, and to institute a prevention program at a tertiary medical center.

Methods

We analyzed SH events for 172 patients and developed computer algorithms to predict SH that were tested on a population of 3028 inpatients who were found to have blood glucose (BG) <90 mg/dl during their hospital stay. Variables with significant bivariate associations were entered into partition analyses to identify interactions. Logistic regression was performed by calculating parameters related to the odds of hypoglycemia below each cut point. Sensitivity and specificity were determined at various cut points. The cut points resulting in 50% sensitivity for each hypoglycemia level were determined. These algorithms were tested against the initial 172 adjudicated patients.

Results

Variables related to the BG <40 mg/dl cut off point were basal and adjustment scale insulin doses, weight, and creatinine clearance, while variables related to the 60 mg/dl and 70 mg/dl cut points were basal, prandial, and adjustment scale insulin doses, weight, creatinine clearance, and sulfonylurea use. The 50% sensitivity cut point developed using the <70 mg/dl algorithm correctly identified 71% of the adjudicated cases, while the <60 mg/dl and <40 mg/dl algorithms identified 70% and 55% respectively.

Conclusions

A validated prediction algorithm for SH can aid in the identification of patients at risk for SH and may be useful in the development of prevention strategies.

Background

This study examined whether mobile phone-based, one-way video messages about diabetes self-care improve hemoglobin A1c (A1C) and self-monitoring of blood glucose (SMBG).

Methods

This was a 1-year prospective randomized trial with two groups. The active intervention lasted 6 months. The study enrolled 65 people with A1C >8.0% who were established (>6 months) patients in the endocrinology clinics of the Walter Reed Health Care System. Participants were randomized to receive “usual care” or self-care video messages from their diabetes nurse practitioner. Video messages were sent daily to cell phones of study participants. Hemoglobin A1c and SMBG data were collected at 0, 3, 6, 9, and 12 months.

Results

Participants who received the messages had a larger rate of decline in A1C than people who received usual care (0.2% difference over 12 months, adjusting for covariates; p = .002 and p = .004 for the interaction between time and group and for the quadratic effect of time by group, respectively). Hemoglobin A1c decline was greatest among participants who received video messages and viewed >10 a month (0.6% difference over 12 months, adjusting for covariates; p < .001 for the interaction between time and group and the quadratic effect). Self-monitoring of blood glucose metrics were not related to the intervention.

Conclusions

A one-way intervention using mobile phone-based video messages about diabetes self-care can improve A1C. Engagement with the technology is an important predictor of its success. This intervention is simple to implement and sustain.

Background

Fast-acting insulin analogs have been available since 1996. The absorption rate of these insulins is still too slow to mimic the physiological insulin action in healthy subjects. This study investigates the clinical performance of InsuPatch™, a local skin-heating device, on postprandial glucose excursion.

Methods

Twenty-four type 1 diabetes mellitus subjects on continuous subcutaneous insulin infusion were included in this crossover study [10 male, 14 female, age: 43.5 ± 11.3 years, diabetes duration: 18.3 ± 10.5 years, glycosylated hemoglobin: 7.4 ± 0.8%, body mass index: 25.0 ± 3.0 kg/m² (mean ± standard deviation)]. The impact of local skin heating was measured by dividing the two-hour area under the curve by integration time (AUC/t120) for blood glucose (BG) above baseline after two standardized breakfast and dinner meal pairs (with and without heating) per subject. For the first breakfast pair, venous insulin concentration was also measured.

Results

A significant reduction was found for the AUC/t120 after breakfast and after dinner meals (42 breakfast meal pairs, AUC/t120 not heated 66.4 ± 32.8 mg/dl vs heated 56.8 ± 34.0 mg/dl, p = .017; 38 dinner meal pairs, AUC/t120 not heated 30.8 ± 31.0 mg/dl vs heated 18.4 ± 23.9 mg/dl, p = .0028). The maximum venous insulin concentration with heating was 27% higher than without heating (n = 23). The number of hypoglycemic events on days with heating (n = 9) was similar to the number of days without heating (n = 13).

Conclusions

Local heating of the skin around the infusion site significantly reduced postprandial BG by enhancing insulin absorption. The heating device was well tolerated, and it could facilitate development of closed-loop systems.

Background

Multiple factors impact subcutaneous insulin injection pain. Injection devices [e.g., syringe or pen needle (PN)] affect pain due to needle length, diameter, needle polishing and lubrication, and needle tip geometry.

Methods

We evaluated a modified 5-bevel PN tip in 32 G × 4 mm 31 G × 5 mm and 8 mm PNs vs the equivalent marketed 3-bevel PNs in laboratory penetration force testing, as well as in insulin-taking subjects for overall acceptability, comparative pain, and preference. The clinical tests were done in three ways: paired insertions with the subjects blinded to PN tip geometry, after brief at-home use of 5-bevel PNs, and again with subjects informed about each needle’s tip geometry in paired insertions.

Results

Average penetration force in a skin substitute was 23% lower with the 5-bevel PNs vs similar 3-bevel PNs (p ≤ 0.01). In blinded testing and after at-home use, patients rated the 5-bevel needle as acceptable. After short-term home use, patients rated the 5-bevel PN less painful and preferable to their usual PN (both p < 0.01). In paired, informed testing, the 5-bevel PN was less painful and preferred to subjects’ currently used needles (p ≤ 0.01) and to other marketed PNs (p < 0.01).

Conclusions

Needle tip geometry affects penetration force. When blinded, patients did not distinguish differences in PN tip geometry with fine-gauge PN insertions. A 5-bevel needle tip is perceived as less painful and is preferred by subjects following home use for usual injections. Similar results occurred when patients were informed that they were using a needle with a modified tip.

Improved needle designs could increase patient compliance with insulin therapy. In this issue of Journal of Diabetes Science and Technology, Hirsch and colleagues assessed patient pain and preference for a 5-bevel needle design among diabetes patients. A blinded comparison with traditional 3-bevel needles yielded no significant difference, but patients preferred the 5-bevel needle in unblinded home injection and clinical insertion studies. This suggests that important subjective/contextual factors contribute to preference in conjunction with the fundamental needle design change. While 5-bevel needles may increase patient acceptance, more dramatic changes of needle design, such as microneedles, could enable still greater patient acceptance through reduced pain as well as improved insulin pharmacokinetics.

Background

Patients consider multiple parameters in adjusting prandial insulin doses for optimal glycemic control. Difficulties in calculations can lead to incorrect doses or induce patients to administer fixed doses, rely on empirical estimates, or skip boluses.

Method

A multicenter study was conducted with 205 diabetes subjects who were on multiple daily injections of rapid/short-acting insulin. Using the formula provided, the subjects manually calculated two prandial insulin doses based on one high and one normal glucose test result, respectively. They also determined the two doses using the FreeStyle InsuLinx Blood Glucose Monitoring System, which has a built-in, automated bolus calculator. After dose determinations, the subjects completed opinion surveys.

Results

Of the 409 insulin doses manually calculated by the subjects, 256 (63%) were incorrect. Only 23 (6%) of the same 409 dose determinations were incorrect using the meter, and these errors were due to either confirmed or potential deviations from the study instructions by the subjects when determining dose with meter. In the survey, 83% of the subjects expressed more confidence in the meter-calculated doses than the manually calculated doses. Furthermore, 87% of the subjects preferred to use the meter than manual calculation to determine prandial insulin doses.

Conclusions

Insulin-using patients made errors in more than half of the manually calculated insulin doses. Use of the automated bolus calculator in the FreeStyle InsuLinx meter minimized errors in dose determination. The patients also expressed confidence and preference for using the meter. This may increase adherence and help optimize the use of mealtime insulin.

Since the early 2000s, there has been an exponentially increasing development of new diabetes-applied technology, such as continuous glucose monitoring, bolus calculators, and “smart” pumps, with the expectation of partially overcoming clinical inertia and low patient compliance. However, its long-term efficacy in glucose control has not been unequivocally proven. In this issue of Journal of Diabetes Science and Technology, Sussman and colleagues evaluated a tool for the calculation of the prandial insulin dose. A total of 205 insulin-treated patients were asked to compute a bolus dose in two simulated conditions either manually or with the bolus calculator built into the FreeStyle InsuLinx meter, revealing the high frequency of wrong calculations when performed manually. Although the clinical impact of this study is limited, it highlights the potential implications of low diabetes-related numeracy in poor glycemic control. Educational programs aiming to increase patients’ empowerment and caregivers’ knowledge are needed in order to get full benefit of the technology.

Aim

The aim was to study the longitudinal relationship between plantar fascia thickness (PFT) as a measure of tissue glycation and microvascular (MV) complications in young persons with type 1 diabetes (T1DM).

Methods

We conducted a prospective longitudinal cohort study of 152 (69 male) adolescents with T1DM who underwent repeated MV complications assessments and ultrasound measurements of PFT from baseline (1997–2002) until 2008. Retinopathy was assessed by 7-field stereoscopic fundal photography and nephropathy by albumin excretion rate (AER) from three timed overnight urine specimens. Longitudinal analysis was performed using generalized estimating equations (GEE).

Results

Median (interquartile range) age at baseline was 15.1 (13.4–16.8) years, and median follow-up was 8.3 (7.0–9.5) years, with 4 (3–6) visits per patient. Glycemic control improved from baseline to final visit [glycated hemoglobin (HbA1c) 8.5% to 8.0%, respectively; p = .004]. Prevalence of retinopathy increased from 20% to 51% (p < .001) and early elevation of AER (>7.5 µg/min) increased from 26% to 29% (p = .2). A greater increase in PFT (mm/year) was associated with retinopathy at the final assessment (ΔPFT 1st vs. 2nd–4th quartiles, χ2 = 9.87, p = .02). In multivariate GEE, greater PFT was longitudinally associated with retinopathy [odds ratio (OR) 4.6, 95% confidence interval (CI) 2.0–10.3] and early renal dysfunction (OR 3.2, CI 1.3–8.0) after adjusting for gender, blood pressure standard deviation scores, HbA1c, and total cholesterol.

Conclusions

In young people with T1DM, PFT was longitudinally associated with retinopathy and early renal dysfunction, highlighting the importance of early glycemic control and supporting the role of metabolic memory in MV complications. Measurement of PFT by ultrasound offers a noninvasive estimate of glycemic burden and tissue glycation.

Background

In glycemic control, postprandial glycemia may be important to monitor and optimize as it reveals glycemic control quality, and postprandial hyperglycemia partly predicts late diabetic complications. Self-monitoring of blood glucose (SMBG) may be an appropriate technology to use, but recommendations on measurement time are crucial.

Method

We retrospectively analyzed interindividual and intraindividual variations in postprandial glycemic peak time. Continuous glucose monitoring (CGM) and carbohydrate intake were collected in 22 patients with type 1 diabetes mellitus. Meals were identified from carbohydrate intake data. For each meal, peak time was identified as time from meal to CGM zenith within 40–150 min after meal start. Interindividual (one-way Anova) and intraindividual (intraclass correlation coefficient) variation was calculated.

Results

Nineteen patients were included with sufficient meal data quality. Mean peak time was 87 ± 29 min. Mean peak time differed significantly between patients (p = 0.02). Intraclass correlation coefficient was 0.29.

Conclusions

Significant interindividual and intraindividual variations exist in postprandial glycemia peak time, thus hindering simple and general advice regarding postprandial SMBG for detection of maximum values.

Background

Optimizing a closed-loop insulin delivery algorithm for individuals with type 1 diabetes can be potentially facilitated by a mathematical model of the patient. However, model simulation studies that evaluate changes to the control algorithm need to produce conclusions similar to those that would be obtained from a clinical study evaluating the same modification. We evaluated the ability of a low-order identifiable virtual patient (IVP) model to achieve this goal.

Methods

Ten adult subjects (42.5 ± 11.5 years of age; 18.0 ± 13.5 years diabetes; 6.9 ± 0.8% hemoglobin A1c) previously characterized with the IVP model were studied following the procedures independently reported in a pediatric study assessing proportional–integral–derivative control with and without a 50% meal insulin bolus. Peak postprandial glucose levels with and without the meal bolus and use of supplemental carbohydrate to treat hypoglycemia were compared using two-way analysis of variance and chi-square tests, respectively.

Results

The meal bolus decreased the peak postprandial glucose levels in both the adult-simulation and pediatric-clinical study (231 ± 38 standard deviation to 205 ± 33 mg/dl and 226 ± 51 to 194 ± 47 mg/dl, respectively; p = .0472). No differences were observed between the peak postprandial levels obtained in the two studies (clinical and simulation study not different, p = .57; interaction p = .83) or in the use of supplemental carbohydrate (3 occurrences in 17 patient days of closed-loop control in the clinical-pediatric study; 7 occurrences over 20 patient days in the adult-simulation study, p = .29).

Conclusions

Closed-loop simulations using an IVP model can predict clinical study outcomes in patients studied independently from those used to develop the model.

Background

The Afinion HbA1c (Axis-Shield) is a newer point-of-care device for measurement of hemoglobin A1c (A1C) using a boronate affinity method unlike the more commonly used DCA immunoassay method (Siemens Medical Solutions Diagnostics). The Afinion’s accuracy and precision, when compared with high-performance liquid chromatography (HPLC) and DCA methods, have not been established in pediatric practice settings.

Methods

Capillary blood was collected from 700 subjects with diabetes mellitus at seven Pediatric Diabetes Consortium sites. Each subject’s A1C was measured locally using Afinion and DCA devices, and by a central laboratory (University of Minnesota) using a Tosoh HPLC method. In addition, repeated measurements on six whole blood samples provided by the National Glycohemoglobin Standardization Program (NGSP) were taken at three clinical centers using the Afinion and DCA methods and centrally using the Tosoh HPLC method to assess the precision of each device.

Results

The coefficient of variation for measurements of whole blood samples for precision analysis was 2% for Afinion, 3% for DCA, and 1% for HPLC. In the patient samples measured at the seven clinic sites, the Afinion generated higher A1C results than the HPLC (mean difference = +0.15; p < 0.001), while the DCA produced lower values (mean difference = -0.19; p < 0.001). The absolute differences with HPLC were similar for the Afinion and DCA (median 0.2%) with a slight advantage for the Afinion when compared with DCA (p < 0.001 by rank test). The DCA tended to read lower than HPLC, particularly at high A1C levels (p < 0.001), while the Afinion’s accuracy did not vary by A1C.

Conclusions

When compared to the central laboratory HPLC method, the differences between the results of the Afinion and DCA devices are clinically insignificant, and the Afinion and DCA have similar accuracy and precision when used in pediatric practice settings.

Point-of-care (POC) hemoglobin A1c measurement is now used by many physicians to make more timely decisions on therapy changes. A few studies have highlighted the drawbacks of some POC methods, e.g., poor precision and lot-to-lot variability. Evaluating performance in the clinical setting is difficult because there is minimal proficiency testing data on POC methods. In this issue of Journal of Diabetes Science and Technology, Wood and colleagues describe their experience with the Afinion method in a pediatric clinic network, comparing these results to another POC method as well as to a laboratory high-performance liquid chromatography method. Although they conclude that the Afinion exhibits adequate performance, they do not evaluate lot-to-lot variability. As with laboratory methods, potential assay interferences must also be considered.