Use of an insulin bolus advisor facilitates earlier and more frequent changes in insulin therapy parameters in suboptimally controlled patients with diabetes treated with multiple daily insulin injection therapy: results of the ABACUS trial

BACKGROUND

We assessed the impact of using an automated bolus advisor integrated into a blood glucose meter on the timing and frequency of adjusting insulin therapy parameter settings and whether the availability of this technology would increase blood glucose test strip utilization in diabetes patients treated with multiple daily insulin injection (MDI) therapy.

SUBJECTS AND METHODS

The Automated Bolus Advisor Control and Usability Study (ABACUS) trial, a 26-week, prospective, randomized, controlled, multinational study that enrolled 218 type 1 and type 2 diabetes patients, demonstrated that use of an automated insulin bolus advisor helps improve glycemic control in suboptimally controlled, MDI-treated patients. Patient data were assessed to determine when and how often changes in insulin parameter settings occurred during the study. Patient meters were downloaded to determine blood glucose monitoring frequency.

RESULTS

One hundred ninety-three patients completed the study: 93 control arm (CNL) and 100 intervention (experimental) arm (EXP). Significantly more EXP (47.5%) than CNL (30.7%) patients received one or more changes in their insulin sensitivity factor (ISF) settings during the study (P=0.0191). Changes in ISF settings occurred earlier and more frequently in EXP than CNL patients throughout the study. A similar trend was seen in changes in insulin-to-carbohydrate ratios. There were no differences in daily self-monitoring of blood glucose frequency [mean (SD)] between CNL and EXP patients: 4.7 (1.5) versus 4.6 (1.3) (P=0.4085).

CONCLUSIONS

Use of an automated bolus advisor was associated with earlier, more frequent changes in key insulin parameters, which may have contributed to subsequent improvements in glycemic control but without increased glucose test strip utilization.

Use of an insulin bolus advisor improves glycemic control in multiple daily insulin injection (MDI) therapy patients with suboptimal glycemic control: first results from the ABACUS trial

OBJECTIVE

Use of automated bolus advisors is associated with improved glycemic control in patients treated with insulin pump therapy. We conducted a study to assess the impact of using an insulin bolus advisor embedded in a blood glucose (BG) meter on glycemic control and treatment satisfaction in patients treated with multiple daily insulin injection (MDI) therapy. The study goal was to achieve >0.5% A1C reduction in most patients.

RESEARCH DESIGN AND METHODS

This was a 26-week, prospective, randomized, controlled, multinational study that enrolled 218 MDI-treated patients with poorly controlled diabetes (202 with type 1 diabetes, 16 with type 2 diabetes) who were 18 years of age or older. Participants had mean baseline A1C of 8.9% (SD, 1.2 [74 mmol/mol]), mean age of 42.4 years (SD, 14.0), mean BMI of 26.5 kg/m(2) (SD, 4.2), and mean diabetes duration of 17.7 years (SD, 11.1). Control group (CNL) patients used a standard BG meter and manual bolus calculation; intervention group (EXP) patients used the Accu-Chek Aviva Expert meter with an integrated bolus advisor to calculate insulin dosages. Glucose data were downloaded and used for therapy parameter adjustments in both groups.

RESULTS

A total of 193 patients (CNL, n = 93; EXP, n = 100) completed the study. Significantly more EXP than CNL patients achieved >0.5% A1C reduction (56.0% vs. 34.4%; P < 0.01). Improvement in treatment satisfaction (Diabetes Treatment Satisfaction Questionnaire scale) was significantly greater in EXP patients (11.4 [SD, 6.0] vs. 9.0 [SD, 6.3]; P < 0.01). Percentage of BG values <50 mg/dL was <2% in both groups during the study.

CONCLUSIONS

Use of an automated bolus advisor resulted in improved glycemic control and treatment satisfaction without increasing severe hypoglycemia.

Automated bolus advisor control and usability study (ABACUS): does use of an insulin bolus advisor improve glycaemic control in patients failing multiple daily insulin injection (MDI) therapy? [NCT01460446]

BACKGROUND

People with T1DM and insulin-treated T2DM often do not follow and/or adjust their insulin regimens as needed. Key contributors to treatment non-adherence are fear of hypoglycaemia, difficulty and lack of self-efficacy associated with insulin dose determination. Because manual calculation of insulin boluses is both complex and time consuming, people may rely on empirical estimates, which can result in persistent hypoglycaemia and/or hyperglycaemia. Use of automated bolus advisors (BA) has been shown to help insulin pump users to more accurately meet prandial insulin dosage requirements, improve postprandial glycaemic excursions, and achieve optimal glycaemic control with an increased time within optimal range. Use of a BA containing an early algorithm based on sliding scales for insulin dosing has also been shown to improve HbA1c levels in people treated with multiple daily insulin injections (MDI). We designed a study to determine if use of an automated BA can improve clinical and psychosocial outcomes in people treated with MDI.

METHODS/DESIGN

The Automated Bolus Advisor Control and Usability Study (ABACUS) is a 6-month, prospective, randomised, multi-centre, multi-national trial to determine if automated BA use improves glycaemic control as measured by a change in HbA1c in people using MDI with elevated HbA1c levels (#62;7.5%). A total of 226 T1DM and T2DM participants will be recruited. Anticipated attrition of 20% will yield a sample size of 90 participants, which will provide #62;80% power to detect a mean difference of 0.5%, with SD of 0.9%, using a one-sided 5% t-test, with 5% significance level. Other measures of glycaemic control, self-care behaviours and psychosocial issues will also be assessed.

DISCUSSION

It is critical that healthcare providers utilise available technologies that both facilitate effective glucose management and address concerns about safety and lifestyle. Automated BAs may help people using MDI to manage their diabetes more effectively and minimise the risk of long-term diabetes related complications. Findings from a recent study suggest that BA use positively addresses both safety and lifestyle concerns; however, randomised trials are needed to confirm these perceptions and determine whether bolus advisor use improves clinical outcomes. Our study is designed to make these assessments.

TRIAL REGISTRATION

NCT01460446.

Hypoglycemia impairs quality of blood glucose simulation in a clinical decision support system

BACKGROUND

Clinical decision support systems allow for decisions based on blood glucose simulations. The DiasNet simulation tool is based on accepted principles of physiology and simulates blood glucose concentrations accurately in type 1 diabetes mellitus (T1DM) patients during periods without hypoglycemia, but deviations appear after hypoglycemia, possibly because of the long-term glucose counter-regulation to hypoglycemia. The purpose of this study was to evaluate the impact of hypoglycemia on blood glucose simulations.

METHOD

Continuous glucose monitoring (CGM) data and diary data (meals, insulin, self-monitored blood glucose) were collected for 2 to 5 days from 17 T1DM patients with poor glycemic control. Hypoglycemic episodes [CGM glucose <63 mg/dl (3.5 mmol/liter) for ≥20 min] were identified in valid (well-calibrated) CGM data. For 24 hours after each hypoglycemic episode, a simulated (DiasNet) glucose profile was compared to the CGM glucose.

RESULTS

A total of 52 episodes of hypoglycemia were identified in valid data. All subjects had at least one hypoglycemic episode. Ten episodes of hypoglycemia from nine subjects were eligible for analysis. The CGM glucose was significantly (p < .05) higher than simulated blood glucose for a period of 13 h, beginning 8 h after hypoglycemia onset.

CONCLUSIONS

The present data show that hypoglycemia introduces substantial and systematic simulation errors for up to 24 h after hypoglycemia. This underlines the need for further evaluation of mechanisms behind this putative long-term glucose counter-regulation to hypoglycemia. When using blood glucose simulations in decision support systems, the results indicate that simulations for several hours following a hypoglycemic event may underestimate glucose levels by 100 mg/dl (5.6 mmol/liter) or more.

The relationship between generic and diabetes specific psychological factors and glycaemic control in adults with type 1 diabetes

259 adults with type 1 diabetes completed measure of anxiety, depression and diabetes specific distress, HbA1c from medical records. Anxiety not depression predicted HbA1c, this association was mediated by illness specific cognitions. Targeting illness specific cognitions may be more productive than treatment of general dysphoria in type 1 diabetes.

The prevalence of depression and anxiety in adults with Type 1 diabetes

AIM

The aim of the present study was to determine the prevalence of anxiety and depression in a large UK group of people with Type 1 diabetes.

METHODS

Patients aged 16-60 years were invited to complete self-report questionnaires when they attended outpatient clinic appointments. HbA(1c) was recorded from the clinic database.

RESULTS

Analysis was based on 273 complete questionnaires. The mean scores for both anxiety (mean 6.4, sd 4.5) and depression (mean 3.4, sd 3.5) were consistent with normative data. Females reported significantly higher mean anxiety than males, although neither reached the criterion for 'caseness'. Significant differences to the norm were observed for the percentages reporting moderate to severe levels of depression in males (chi(2) = 6.44; d.f. = 2; P = 0.04) and moderate to severe levels of anxiety in females (chi(2) = 7.47; d.f. = 2; P = 0.02). HbA(1c) was positively correlated with HADS scores (anxiety r = 0.2, P = 0.001, depression r = 0.14, P = 0.02).

CONCLUSIONS

While there is no significant difference in the mean anxiety or depression in this cohort compared with those reported for a non-diabetic, healthy population, the results suggest that there is an increased prevalence of clinically relevant anxiety in females and of depression in males with Type 1 diabetes when compared with the normative data.

Influence of alcohol on cognitive performance during mild hypoglycaemia; implications for Type 1 diabetes

AIMS

Alcohol and hypoglycaemia independently affect cognitive function. This may be relevant for insulin-treated diabetic patients who drive motor vehicles. The aim of this study was to examine the effect of mild hypoglycaemia (2.8 mmol/l) with modest alcohol intoxication (levels below UK driving limits) on intellectual performance in patients with Type 1 diabetes.

METHODS

A hyperinsulinaemic glucose clamp (60 mU/m2) was used to study 17 subjects [age 35 +/- 8 years, HbA1c 8.1 +/- 1.4% (mean +/- sd)] on four occasions: (A) euglycaemia (4.5 mmol/l) with placebo, (B) euglycaemia with alcohol, (C) hypoglycaemia (2.8 mmol/l) with placebo, and (D) hypoglycaemia with alcohol. Cognitive performance was assessed using four-choice reaction time (4CRT, primary outcome), measurements of general intellectual skills [trail making B (TMB) and digit symbol substitution (DSST)], and visual information processing [visual change detection (VCD)]. A test related to driving performance (hazard perception) was also used.

RESULTS

In experiments B and D the average blood alcohol level was 43 mg/dl. This was associated with deterioration in 4CRT [+ 35 ms [95% confidence interval (CI) 20, 50]] and TMB, whereas hypoglycaemia without alcohol increased 4CRT only [+ 39 ms (95% CI 5, 73)]. However, when alcohol was combined with hypoglycaemia, there was marked deterioration in all the cognitive function tests [4CRT 74 ms (95% CI 35, 113), TMB, DSST and VCD]. Hazard perception was not affected. The effect of alcohol was no different in euglycaemia than in hypoglycaemia, i.e. there was no interaction. Whereas hypoglycaemia did not reduce the likelihood that the subjects would drive, alcohol did.

CONCLUSIONS

The cumulative effect of alcohol and hypoglycaemia on cognitive function together has implications for driving in patients with Type 1 diabetes. Both independently impair cognitive function and together the effects are additive. Patients with Type 1 diabetes should be educated about hypoglycaemia and driving and should avoid alcohol completely if planning to drive.

Use of the Internet to optimize self-management of type 1 diabetes: preliminary experience with DiasNet

Six patients with type 1 diabetes participated in a pilot trial. Their median age was 36 years (range 29-61) and the median duration of diabetes was 10 years (range 3-29). They were asked to enter, from their home or work PC, blood glucose values, insulin doses and a food diary. From the data entered, a computer model generated a simulation of the blood glucose concentration for the data collection period. It could then suggest alternative insulin doses (or regimes), or meal sizes, to reduce the risk of hypo- and hyperglycaemia. During a six-month study, patients entered a median of five sets of data (range two to eight). Feedback from participants revealed that while the system was helpful, difficulties with data entry hindered its use. Information gained from this exercise is shaping further development of the system.

Preventing glycaemic excursions in diabetic patients requiring percutaneous endoscopic gastrostomy (PEG) feeding after a stroke

AIMS AND METHODS

Enteral feeding for diabetic patients with a stroke is often associated with hyperglycaemia and/or hypoglycaemia, which can adversely influence neurological recovery. We have developed a structured enteral feeding programme aimed at establishing 'normal' feeding patterns and avoiding marked glycaemic excursions.

RESULTS

Of 332 consecutive patients admitted to an acute stroke unit, 20 of 41 diabetic patients required PEG feeding. Over the initial 24-48 h, patients had 22- h feeds with continuous intravenous infusion of soluble insulin. Thereafter, they were established on three feeds per day. Soluble insulin was given prior to each feed with isophane insulin at 2200 h. Average duration of feed was 13 +/- 8 days with an achieved glucose level of 8.7 +/- 2.5 (mean +/- SD) mmol/l associated with 0.8 episodes of biochemical hypoglycaemia (< 3 mmol/l) each week.

CONCLUSIONS

A structured enteral feeding programme for PEG-fed diabetic patients can improve the quality of care with avoidance of marked glycaemic excursions.

Performance of a continuous glucose monitoring system during controlled hypoglycaemia in healthy volunteers

It has been suggested that the continuous glucose monitoring system may be a useful tool for detecting unrecognised hypoglycaemia, especially at times when finger prick testing is difficult or impossible (e.g., at night). Studies suggest that subcutaneous glucose levels closely mimic blood glucose levels with a lag time of only a few minutes. However, no studies have been published to show how well the sensor performs during sustained or in recovery from hypoglycaemia. This study involved using a hyperinsulinaemic glucose clamp (60 mU/m2) in nine healthy volunteers. Each subject had two sensors inserted the day before the study. Blood glucose levels were maintained at euglycaemia for the first 60 min, then decreased to 45 mg/dL (2.5 mmol/L) for 60 min, and finally restored to euglycaemia. Blood glucose measurements were compared with interstitial values recorded by the sensor. Sensor profiles showed acceptable agreement with blood glucose levels at each of the three plateaus with a correlation coefficient of 0.79, slope of 0.85, and mean absolute error of 7%. The sensor drop closely matched the drop in blood glucose, but the recovery from hypoglycaemia was delayed by an average of 26 min. Continuous glucose sensing provides a useful means of detecting unrecognised hypoglycaemia in type 1 diabetes, although the duration of hypoglycaemia may be overestimated.

DiasNet--a diabetes advisory system for communication and education via the internet

Intensive diabetes treatment can lead to a substantial reduction of the rate of the complications associated with diabetes. However, a number of patients may have poor control despite specialist care, and this along with devolution of care to non-specialists suggests that alternative interventions should be developed. The present paper describes an Internet based system where more emphasis is put on patient empowerment, the keywords being education and communication. The DiasNet system is based on a well documented decision support system, Dias, designed for use by clinicians. The scope of DiasNet has been widened from being used by clinicians to give advice on insulin dose, to also being used by patients as a tool for education and communication. Patients can experiment with their own data, adjusting insulin doses or meal sizes. In this way different therapeutic and dietary alternatives can be tried out, allowing the patient to gain experience in achieving glycaemic control. DiasNet is implemented in JAVA according to the client/server principle, enabling a new way of communication between patient and clinician: in case of any problems, the patient simply phones the clinician, who immediately, using his or her office PC, can take a look at the data the patient has entered.

The effect of evening alcohol consumption on next-morning glucose control in type 1 diabetes

OBJECTIVE

Alcohol is associated with acute hypoglycemia in patients with type 1 diabetes. After drinking alcohol in the evening, delayed hypoglycemia has also been described, although its cause is unknown. We performed a controlled study to investigate this phenomenon.

RESEARCH DESIGN AND METHODS

We admitted six men with type 1 diabetes (aged 19-51 years, HbA(1c) 7.0-10.3%) on two occasions, from 5:00 P.M. to 12:00 noon the following day. They received regular insulin injections before standardized meals, at 6:00 P.M. and 8:00 A.M., and a basal insulin infusion (0.15 mU x kg(-1) x min(-1)) from 11:00 P.M. They drank either dry white wine (0.75 g/kg alcohol) or mineral water at 9:00 P.M. over 90 min. Blood glucose, alcohol, insulin, cortisol, growth hormone, and glucagon levels were measured.

RESULTS

Blood ethanol reached a mean (SEM) peak of 19.1 (1.2) mmol/l and was undetectable by 8:00 A.M. There were no significant differences in evening or overnight blood glucose levels between the studies. In the morning, fasting and postprandial blood glucose levels were significantly lower after consumption of wine (postprandial peak 8.9 [1.7] vs. 15 [1.5] mmol/l, P < 0.01), and from 10:00 A.M., five subjects required treatment for hypoglycemia (nadir 1.9-2.9 mmol/l). None of the subjects had hypoglycemia after consumption of water. After consumption of wine, growth hormone secretion was significantly reduced between midnight and 4:00 A.M. (area under the curve 2.1 [1.1] vs. 6.5 [2.1] microg. l(-1) x h(-1), P = 0.04). There were no differences in insulin or other hormone levels.

CONCLUSIONS

In type 1 diabetes, moderate consumption of alcohol in the evening may predispose patients to hypoglycemia after breakfast the next morning. This is associated with reduced nocturnal growth hormone secretion. Patients should be informed of this risk and advised regarding appropriate preventative measures.

DiasNet--an Internet tool for communication and education in diabetes

Although studies have shown that intensive diabetes treatment with the goal of maintaining blood glucose concentrations close to the normal range lead to a substantial reduction of the rate of the complications (e.g. blindness, kidney failure, amputations and circulatory diseases), this can be difficult to achieve using conventional means. It is recognised that a number of patients may have poor control despite specialist care, and this along with devolution of care to non-specialists suggests that alternative interventions should be developed. DiasNet, which is used by patients as a tool for education and communication, is based on a well documented decision support system used by clinicians to give advise on insulin dose. In DiasNet the patients can experiment with their own data, adjusting insulin doses or meals sizes, and thereby learning how to cope with various situations. The system can run both as an ordinary application on a standard PC, and as an Internet application using a standard web browser. The latter enables a new way of communication between patient and clinician. DiasNet is currently being tested on a small group of diabetes patients.

Reducing hospital inpatient length of stay for patients with diabetes

AIMS

To ascertain the effect of routine review by a diabetes nurse advisor on length of stay for medical and surgical inpatients with diabetes.

METHODS

Inpatients with diabetes were identified prospectively from January 1997 until December 1998 (792 in 1997 and 819 in 1998). A new post of diabetes nurse advisor was introduced in January 1998 to optimize diabetes management. Length of stay was calculated retrospectively from hospital computer records.

RESULTS

Median length of stay in 1997 was 11 days in medicine and 8 days in surgery. In 1998, the nurse advisor made 1936 visits to 819 patients; median length of stay fell to 8 days in medicine and 5 days in surgery (P < 0.001). Bed occupancy by patients with diabetes fell from 6.8 to 4.0%. Mean length of stay across the hospital remained unchanged.

CONCLUSIONS

The introduction of a ward-based diabetes nurse advisor was associated with significant reductions in length of stay in inpatients with diabetes. Since this study was not a randomized study, other factors may have contributed to this change. However, the consistency of the reduction across specialities suggests the post itself had an important effect.

Impaired absorption and omission of insulin: a novel method of detection using the diabetes advisory system computer model

The Diabetes Advisory System (DIAS) is a decision-support program developed to assist insulin dose adjustment in type 1 diabetes. In this paper, we show how it might be used to identify impaired absorption or omission of insulin in patients with poorly controlled blood glucose. An evaluation of glucose results from four outpatients with persistent hyperglycemia is presented (age 19-48 years with type 1 diabetes for 13-18 years of duration, HbA1c 9.4-13.6%). Each had completed a 4-day record of blood glucose (BG, pre-meal and bedtime), dietary (carbohydrate) intake, and insulin doses (with injection sites). From these data, DIAS modeled a glucose profile (simulated glucose, SG) for the same period. Qualitative assessments were made of differences between BG and SG, and selective reduction or complete removal of insulin doses where BG >> SG. Large improvements in modeling were attributed to either impaired absorption or omission of insulin. Confirmation of these problems was sought from the patients by detailed consultation and physical examination. Impaired insulin absorption was suspected in two patients, both having significant injection site abnormalities. Insulin omission was suspected in the other two subjects. Both had normal injection sites, and one admitted to missing doses. Following retraining, data from three patients showed noticeable improvements in overall modeling as well as glucose control. Using DIAS in the evaluation of patients with type 1 diabetes may highlight previously unrecognized injection site abnormalities or insulin dose omission. This could assist rational optimization of insulin therapy in cases of persistently poor glucose control.

A detailed examination of the clinical terms and concepts required for communication by electronic messages in diabetes care

The wider electronic exchange of clinical information between heterogeneous information systems in the delivery of diabetes care demands a common structure in the form of a message standard. A European Standard electronic diabetes message is being developed in conjunction with CEN TC251. This paper describes the methodologies that the 1998 DO IT Workshop has used to identify potential areas of difficulty in the design and implementation of the preliminary message model. To facilitate implementation and to avoid ambiguity in electronic messaging it is particularly important that there is standardisation of the definitions of the clinical terms specifically used in diabetes care across systems. Comprehensive lists of such terms to describe all areas of diabetes care do not exist and there is a lack of harmonisation of definitions in many areas. Thus, to better understand the user requirements of diabetes messaging several approaches were adopted. A review of the clinical terms and concepts contained in pre-existing datasets was undertaken with detailed study of a number of specific areas of diabetes care, analysing the conceptual structure of all the clinical terms that they comprised. Consideration of several worst case clinical scenarios for messages to communicate was also made to identify deficiencies in the message structure. This activity confirmed the importance of creating a Standard for a superset or thesaurus of diabetes specific terms, with appropriate definitions, to harmonise data communication in different IT systems to facilitate messaging. A substantial number of new terms were identified in the workshop and these will form an important first step to accomplishing a first draft superset once fully analysed. It was also apparent that certain specific areas within diabetes care, but most particularly in nursing, dietetics and podiatry, need urgent work to further develop the concepts and terms. This needs to be facilitated for an appropriate group of such professionals. To achieve such a Standard, continued co-operation with CEN/ISSS was recognised to be very important.

What do patients with diabetes know about their tablets?

AIMS

To assess knowledge about oral hypoglycaemic agents amongst patients with diabetes and non-specialist healthcare professionals.

METHOD

An anonymous questionnaire was used in two centres to assess knowledge about oral agents amongst 261 patients with Type 2 diabetes mellitus (mean age 64 years) and 102 health professionals (including doctors, nurses and pharmacists).

RESULTS

Only 15% of patients knew the correct mechanism of action of their medication and 62% took tablets correctly in relation to food. Only 10% of those taking a sulphonylurea knew it may cause hypoglycaemia and 20% of those taking metformin were aware of its gastrointestinal side-effects. Twenty per cent forgot to take their tablets at least once a week and 5% omitted tablets because of hyperglycaemia. Only 35% of patients recalled receiving advice about their medication with only 1% receiving written advice. The healthcare professionals showed important gaps in their knowledge on dosage timing and mechanism of action, particularly with respect to metformin and acarbose.

CONCLUSION

It is concluded that patients' and professionals' knowledge of oral hypoglycaemic agents is poor. More appropriate advice and information to patients from prescribers may improve patient understanding and hence compliance. Community pharmacists are ideally placed to reinforce this information.

Screening for diabetic retinopathy using computer based image analysis and statistical classification

Diabetic retinopathy is one of the most common causes of blindness in Europe. However, efficient therapies do exist. An accurate and early diagnosis and correct application of treatment can prevent blindness in more than 50% of all cases. Digital imaging is becoming available as a means of screening for diabetic retinopathy. As well as providing a high quality permanent record of the retinal appearance, which can be used for monitoring of progression or response to treatment, and which can be reviewed by an ophthalmologist, digital images have the potential to be processed by automatic analysis systems. We have described the preliminary development of a tool to provide automatic analysis of digital images taken as part of routine monitoring of diabetic retinopathy in our clinic. Various statistical classifiers, a Bayesian, a Mahalanobis, and a KNN classifier were tested. The system was tested on 134 retinal images. The Mahalanobis classifier had the best results: microaneurysms, haemorrhages, exudates, and cotton wool spots were detected with a sensitivity of 69, 83, 99, and 80%, respectively.

Dynamic updating in DIAS-NIDDM and DIAS causal probabilistic networks

Diabetes advisory system (DIAS) is a decision support system, which has been developed to provide advice on the amount of insulin injected by subjects with insulin-dependent diabetes mellitus (IDDM). DIAS employs a temporal causal probabilistic network (CPN) to implement a stochastic model of carbohydrate metabolism. The CPN network has recently been extended to provide also advice to subjects with noninsulin-dependent diabetes mellitus (NIDDM). However, due to increased complexity and size of the extended CPN the calculations became unfeasible. The CPN network was, therefore, simplified and a novel approach employed to generate conditional probability tables. The principles of dynamic CPN's were adopted and, in combination with the method of conditioning, learning, and forecasting, were implemented in a time- and memory-efficient way. An evaluation using experimental data was carried out to compare the original and revised DIAS implementations employing data collected by patients with IDDM, and to assess the a posteriori identifiability of model parameters in patients with NIDDM.

Beyond retinal screening: digital imaging in the assessment and follow-up of patients with diabetic retinopathy

Many screening methods are available for detecting diabetic retinopathy. However, once patients develop retinopathy, it is unclear as to what method should be used for their review. We describe a novel and integrated system for the screening and treatment of diabetic retinopathy using high street optometrists for primary screening and digital imaging as a secondary screening tool, with referral to a joint retinal clinic only where ophthalmological intervention may be required. Of 3586 patients screened by optometrists, 328 were classified as having moderate/severe pre-proliferative retinopathy or diabetic maculopathy. Patients with proliferative retinopathy (1% of the total) were recalled directly to the joint retinal clinic. A consecutive sample (281) of these patients, together with a further 100 classified by the optometrists as having no or background retinopathy were compared using digital images and standard 35 mm colour transparencies. These, together with the original optometrist reports, were reviewed independently and individually by an ophthalmologist. A further sample of 124 patients who had undergone both digital imaging and ophthalmologist slit-lamp examination were also compared. Comparison of 35 mm colour transparencies with optometrist reports showed the latter had a sensitivity for detecting sight-threatening retinopathy (STR) of 62%, a specificity of 84%, and a kappa score of 0.62. The results for digital images were 90%, 97%, and 0.90, respectively, although the extent of retinopathy was under-reported in 10 patients. With ophthalmologist slit-lamp examination as the gold standard, the sensitivity of digital imaging was 90% with a substantial level of agreement between them (kappa 0.61). We conclude that digital images provide an efficient method for the follow-up of patients with established or previously treated retinopathy.

Preliminary experience of the DIAS computer model in providing insulin dose advice to patients with insulin dependent diabetes

The Diabetes Advisory System (DIAS) is a model of human glucose metabolism which predicts hourly blood glucose concentrations and provides advice on insulin dose. Its ability to provide appropriate advice was assessed in 20 well-controlled IDDM patients (mean (SD) age 38 (11), duration 17 (9) years; HbA1 8.8 (0.9)%, reference range 5.4-7.6%). Patients recorded blood glucose measurements, insulin dose and food intake for 4 days. These data were used to generate insulin dose advice by both DIAS and a diabetes specialist nurse. Patients were then allocated to follow either DIAS or nurse advice for a further 4 days. There was no significant difference in mean recorded blood glucose values or frequency of reported hypoglycaemia between the DIAS and nurse groups either before or after insulin dose adjustment. The DIAS model, however, generated significantly lower insulin dose advice than the nurse (median (range)% change in insulin dose: DIAS group -13.3% (-25.0 to +11.6) versus nurse group 0% (-8.7 to +2.5), P < 0.05). We conclude that, in the patients studied, DIAS provided insulin dose advice which maintained good short term control of diabetes, despite significant reductions in dose in some cases.

DIAS-NIDDM--a model-based decision support system for insulin dose adjustment in insulin-treated subjects with NIDDM

A decision support system has been developed, Diabetes Insulin Advisory System for patients with non-insulin dependent diabetes mellitus (DIAS-NIDDM), assisting in the adjustment of insulin doses in insulin-treated subjects. DIAS-NIDDM uses a causal probabilistic network (CPN) model of carbohydrate metabolism to make stochastic predictions of blood glucose (BG) excursions. The CPN model is an extension of an existing model with an added component representing endogenous insulin secretion. A linear relationship between BG and insulin concentration due to BG stimulated insulin secretion is assumed. Model parameters (pancreatic sensitivity, insulin sensitivity, and time-to-peak of NPH insulin) are estimated by Bayesian probability updating from patient's specific data (food intake, insulin doses, BG measurements) recorded over a period of 4 days. The estimated parameters allow the system to be potentially used as a diagnostic tool to identify abnormalities of carbohydrate metabolism: impaired insulin secretion, insulin resistance and the severity of the impairments. DIAS-NIDDM was used to predict patient-specific BG profiles and advise on insulin doses during a pilot study in eight patients with NIDDM of whom five were treated with insulin. Compared to the administered insulin amount, daily insulin amount advised by DIAS-NIDDM was similar (within 4 U) in three patients, higher by 20% (19 U) in one patient and lower by 40% (18 U) and 50% (11 U) in two patients, respectively. The inter-day coefficient of variation of the daily insulin advice suggests that, at least according to DIAS-NIDDM criteria, day-to-day adjustment of insulin doses is necessary to maintain optimum control.

Using a double blind controlled clinical trial to evaluate the function of a Diabetes Advisory System: a feasible approach?

This paper assesses the feasibility of using a double blind controlled clinical trial to evaluate the function of a decision support system by applying such a design to the evaluation of a Diabetes Advisory System (DIAS). DIAS is based on a model of the human carbohydrate metabolism and is designed an interactive clinical tool, which can be used to predict the effects of changes in insulin dose or food intake on the blood glucose concentration in patients with insulin dependent diabetes. It can also be used to identify risk periods for hypoglycaemia. and to provide advice on insulin dose. The latter feature was evaluated in the present study. We believe double blind controlled clinical trials are prerequisites for clinical application of many decision support systems, and conclude that the present double blind controlled clinical trial is a suitable evaluation method for the function of DIAS.

DIAS--the diabetes advisory system: an outline of the system and the evaluation results obtained so far

The present paper gives a description of the Diabetes Advisory System (DIAS), and the evaluation results obtained so far. DIAS is a decision support system for the management of insulin dependent diabetes. The core of the system is a compartment model of the human carbohydrate metabolism implemented as a causal probabilistic network (CPN or Bayesian network), which gives it the ability to handle the uncertainty, for example, in blood glucose measurements or physiological variations in glucose metabolism. The evaluation results suggest that, at least in our hands, DIAS can generate advice that is safe and of a quality that is at least comparable to what is available from experienced clinicians.

Alcohol ingestion and glycaemic control in patients with insulin-dependent diabetes mellitus

It is widely accepted that alcohol consumption by patients with insulin-dependent (Type 1) diabetes mellitus is associated with an increased risk of hypoglycaemia. This association has been the subject of few studies, however, and there is not much evidence to support advice currently given to patients. Available information suggests that moderate alcohol consumption by healthy, fed subjects does not cause acute hyper- or hypoglycaemia although there may be a delayed risk of hypoglycaemia the morning after evening alcohol intake. Alcohol can lead to potentially hazardous hypoglycaemia in fasted individuals or in those dependent upon alcohol and has been associated with hypoglycaemic unawareness.

Use of the DIAS model to predict unrecognised hypoglycaemia in patients with insulin-dependent diabetes

The Diabetes Advisory System (DIAS) is a model of human glucose metabolism implemented in a causal probabilistic network. It handles data on insulin dose, carbohydrate intake and blood glucose concentration to predict hourly blood glucose concentrations and thus provide an indication of blood glucose values between home blood tests. DIAS was used to predict blood glucose profiles in eight patients with well-controlled insulin-dependent diabetes, who are at increased risk of hypoglycaemia (abnormally low blood glucose levels). DIAS predicted nocturnal hypoglycaemia in six patients and daytime hypoglycaemia in one patient. The occurrence of nocturnal hypoglycaemia was not recognised by the patient or suspected by their doctor but was subsequently confirmed by blood testing in five patients. It is known that unrecognised nocturnal hypoglycaemia is common in patients with well-controlled diabetes. The ability of DIAS to identify such periods of hypoglycaemia with reasonable accuracy illustrates how the advanced technology it employs may provide reliable decision support to clinicians.

Analysing the hypoglycaemic counter-regulation: a clinically relevant phenomenon?

This paper describes an analysis of the temporal relation between episodes of low blood glucose (hypoglycaemia) and counter-regulations, i.e., episodes of elevated blood glucose (hyperglycaemia), in patients with insulin dependent diabetes. The relation was assessed by statistical methods based on a metabolic computer model of the human glucose metabolism. The study material was standard collected clinical data on meals, insulin injections, and measured blood glucose from hospitalised patients. We have found that a typical hypoglycaemic counter-regulation begins 6-8 h after the hypoglycaemia, that it lasts 16-18 h, giving a total duration of 24 h, and that it elevates the blood glucose by 4-10 mmol/l. The phenomenon was demonstrated in the data from more than half of the patients with hypoglycaemic episodes.

Genetic susceptibility to multiple sclerosis in a Shanghai Chinese population. The role of the HLA class II genes

The association of MS with the HLA class II loci DR and DQ was investigated in subjects of Shanghai Chinese and British Caucasian origin. Our aim was to determine whether common alleles predispose to the disease in both races. In the Caucasian population MS was significantly positively associated with the putative haplotype DRB1*1501, DQA1*0102, DQB1*0602. In contrast, HLA class II alleles were not found to predispose to the disease in the Shanghai Chinese, suggesting that this haplotype is unlikely to be a universal susceptibility determinant. The absence of a disease association with the HLA-DR and -DQ genes in the Chinese population has a number of possible explanations. Our study suggests that other genetic and/or environmental components may be more important in determining susceptibility to MS in this race.

Analysis of a Chinese population suggests that the TNFB gene is not a susceptibility gene for Graves' disease

Graves' disease is associated with different HLA genes in different races. The TNFB gene lies between the class I and class II HLA genes and has two alleles, TNFB*1 and TNFB*2. Studies in Caucasians have suggested that the TNFB gene might be a susceptibility gene for Graves' disease. To investigate further the role of TNFB in predisposition to Graves' disease, we determined whether the TNFB disease associations in the Chinese were similar to those in Caucasians. A total of 57 patients with Graves' disease (32 male) were studied. A TNFB gene fragment was amplified from genomic DNA by using the polymerase chain reaction and digested with Nco I to distinguish the TNFB alleles (TNFB*1 and TNFB*2). Genotype frequencies were compared with those in a racially matched group of 92 controls. TNFB*1 homozygosity occurred in 15 (26%) Graves' and 22 (24%) control subjects. TNFB*1/TNFB*2 heterozygosity occurred in 29 (51%) and 48 (52%) and TNFB*2 homozygosity in 13 (23%) and 22 (24%), respectively (NS). There were gender differences in the frequencies of TNFB*1 homozygosity (13 male [41%], 2 female [8%]). TNFB*1/TNFB*2 heterozygosity (13 male [41%], 16 female [64%]) (chi 2 = 7.3, p = 0.02), and TNFB*2 frequency (19 male [59%], 23 female [92%]; pc = 0.04) in Graves' patients. We conclude that the TNFB associations with Graves' disease in the Hong Kong Chinese differ between the genders and from those described in Caucasians. The TNFB gene is not a susceptibility gene for Graves' disease.

The Bgl II RFLP associated with type 1 diabetes in DR3-positive subjects is not due to a DQA1 promoter region polymorphism

Type 1 (insulin-dependent) diabetes is strongly associated with the HLA genes encoding DR3 and DR4 and their associated DQ alleles. While 70% of all Caucasian diabetic patients carry the DR3-associated allele DQA1*0501, this allele also occurs in up to 40% of the healthy population. A DQA1 Bgl II 7.2 kb RFLP has been shown to identify a disease-associated subset of DR3-positive subjects. We examined the frequency of this RFLP pattern in 43 diabetic and 25 control DR3-positive subjects and found it to be present in 27 (65%) and 5 (20%) respectively (p = 0.0012). The promoter of the DR3-associated DQA1*0501 allele was amplified in four diabetic subjects who were positive, and four control subjects who were negative, for the 7.2 kb band. The promoter was digested with Bgl II to determine whether polymorphism within the promoter created a disease-associated Bgl II restriction site, which might influence disease susceptibility by an effect on gene transcription. No amplified promoter fragment contained a Bgl II restriction site, suggesting that the disease-associated 7.2 kb band does not result from DQA1 promoter region polymorphism but may be due to polymorphism elsewhere on the DR3 haplotype.

The HLA association with Graves' disease is sex-specific in Hong Kong Chinese subjects

OBJECTIVE

Graves' disease is associated with different HLA genes in Caucasians and the Chinese, in whom the HLA associations may be stronger in males than females. Common HLA-associated susceptibility in both races may occur at the HLA-DQ loci. The aims of this study were to examine the HLA-A, B, DR and DQ associations with Graves' disease in a Hong Kong Chinese population and to determine whether the HLA associations differ between the sexes and between subjects with and without thyrotoxic periodic paralysis.

DESIGN

HLA-A, B and DR types were determined by serological typing and DQA1 and DQB1 alleles by oligonucleotide probing of the respective enzymatically amplified gene.

PATIENTS

Ninety-seven Chinese patients with Graves' disease (31 males with, 35 males without and 31 females without thyrotoxic periodic paralysis) and 105 racially matched healthy controls.

MEASUREMENTS

Frequencies of HLA types/alleles at each locus were compared between patients and controls and between the Graves' subgroups using the chi 2-test.

RESULTS

HLA-B46, DR9 and DQB1*0303 were associated with Graves' disease in males only; these associations were weaker in males with thyrotoxic periodic paralysis. DR12, DQA1*0401 and DQB1*0301 were protective, regardless of sex or the presence of thyrotoxic periodic paralysis. The positive HLA associations in the Hong Kong Chinese were distinct from those in Caucasians whereas the protective haplotype was similar to that described in Caucasians.

CONCLUSIONS

These findings call in question the role of HLA genes in disease susceptibility but suggest a role for HLA in protection from Graves' disease.

An investigation of the association between HLA-DQ heterodimers and type 1 (insulin-dependent) diabetes mellitus in five racial groups

The association between HLA-DQ alpha Arg52-HLA-DQ beta non-Asp57 heterodimers and type 1 (insulin-dependent) diabetes was compared in Japanese, Chinese, Caucasian, North Indian Asian, and Afro-Caribbean patients to determine their importance in disease susceptibility. The potential to encode four Arg52-non-Asp57 DQ heterodimers, two in cis and two in trans, was significantly associated with increased risk of type 1 diabetes in all races except the Japanese. The possibility of encoding two Arg52-non-Asp57 heterodimers was also significantly associated with increased risk of the disease in all races except the Japanese. The possibility of encoding one heterodimer was not significantly associated with type 1 diabetes in any of the races studied. Heterogeneity testing revealed significant differences in RR values for four, two, and one heterodimers in all races except the Japanese and significant differences in RR for four and two heterodimers when compared across the races. This, together with the lack of an association between Arg52-non-Asp57 heterodimers and type 1 diabetes in the Japanese, suggests that, assuming the same genetic basis for disease in all races, the heterodimer is unlikely to be of primary importance in susceptibility to the disease.

The influence of HLA-DR and -DQ alleles on progression to multiple sclerosis following a clinically isolated syndrome

A 5-year follow-up study was performed on 70 Caucasian patients presenting with isolated neurological syndromes of the optic nerve, brain stem, or spinal cord to assess the risk of progression to MS. The influence on patient prognosis of HLA-DR and -DQ alleles and presentation with disseminated brain lesions, demonstrated by MRI scanning, was determined. Clinical progression to MS was observed in 61% of optic neuritis patients, 50% of patients with a brain-stem syndrome, and 35% of patients with a spinal cord disturbance. MS and the isolated clinical syndromes were positively associated with DRB1*1501, DQA1*0102, and DQB1*0602; the frequency of these alleles in the latter group was intermediate between that seen in MS patients and healthy controls. Conversion to MS was positively associated with the DRB1*1501.DQA1*0102.DQB1*0602 haplotype, but the influence of HLA was only significant in patients with disseminated brain lesions at presentation (MRI positive); MS developed in 86% of MRI-positive, DRB1*1501-positive patients compared with 55% of MRI-positive, DRB1*1501-negative patients (p < 0.025). The data suggest that these HLA alleles are involved in susceptibility to initial demyelinating lesion formation and are important in the subsequent development of MS in MRI-positive patients.

Both DQA1 and DQB1 genes are implicated in HLA-associated protection from type 1 (insulin-dependent) diabetes mellitus in a British Caucasian population

Inherited susceptibility to Type 1 (insulin-dependent) diabetes mellitus is partly determined by HLA genes. It has been suggested that protection from disease may be conferred by HLA-DQB1 genes which encode molecules with aspartate at position 57. We investigated the contributions of HLA-DRB1, DQA1 and DQB1 genes to protection from disease. Restriction fragment length polymorphism and sequence specific oligonucleotide analysis in 156 British Caucasian Type 1 diabetic and 116 control subjects showed protection from disease was associated with DR2, DRw6 and DR7 haplotypes. The most protective DQA1 allele was DQA1*0102 which occurred on both DR2 and DRw6 haplotypes. The DQB1 alleles DQB1*0303, DQB1*0602 and DQB1*0603 were associated with protection, as was DQB1*0604, which encodes an Asp-57 negative DQ beta molecule. Heterozygosity for both protective and predisposing HLA markers was reduced in diabetic compared with control subjects. We conclude that both DQA1 and DQB1 genes are implicated in HLA-associated protection from Type 1 diabetes in this British Caucasian population. The overall structure of the DQ heterodimer is critical and DQ beta-Asp 57 is of secondary importance in determining protection from disease. The effect of protective HLA types may predominate over that of predisposing markers.

Susceptibility to IDDM in a Chinese population. Role of HLA class II alleles

MHC associations with IDDM in a Chinese population were studied to investigate genetic susceptibility to the disorder. The frequency of HLA-DR3 was significantly higher in the diabetic patients (19/49 [38.7%] vs. control subjects, 11/105 [10.5%], Pc less than 1.3 x 10(-3), RR = 5.3 [CI 2.3-12.1]), whereas DR4 was not (11/49 [22.4%] vs. 28/105 [26.7%], NS). The frequency of DR3/4 heterozygosity was higher in the diabetic patients (6/49 [12.2%] vs. control subjects, 0/105 [0%], P = 1.7 x 10(-3), RR = 31.5 [CI 3.8-263.6]). The frequency of DR3/9 heterozygosity also was higher in the diabetic patients (6/49 [12.2%] vs. control subjects, 2/105 [1.9%], P = 0.03, RR = 6.2 [CI 3.0-12.7]). No significant associations were noted between DQB1 alleles and IDDM. Among DR4-positive subjects, the frequency of DQB1 allele DQB1*0302 was higher in the diabetic patients (10/11 [90.0%] vs. control subjects, 12/24 [50%], Pc less than 0.05, RR = 7.0 [CI 1.3-38.0]), and the frequency of DQB1*0401 was significantly lower in the diabetic patients (2/11 [18.2%] vs. control subjects, 16/24 [66.7%], Pc = 0.04, RR = 0.1 [CI 0.02-0.46]). No DR4 subtype was associated significantly with IDDM. The frequency of DQA1*0501, a DQA1 allele, was higher in diabetic patients (22/41 [53.7%] vs. control subjects, 20/95 [21.1%], Pc less than 3 x 10(-3), RR = 4.3 [CI 2.0-9.3]). The frequency of DQA1*0301, which has been associated consistently with IDDM in other ethnic groups, was not significantly higher in the diabetic patients in this study (27/41 [65.9%] vs. control subjects, 53/95 [55.8%], NS).(ABSTRACT TRUNCATED AT 250 WORDS)

Diabetic impotence: risk factors in a clinic population

Analysis of questionnaire replies from 292 diabetic men aged 20-59 revealed impotence in 23% (66), compared to 9% (7) of a sample of 81 non-diabetic outpatients. Further investigation of 45 impotent diabetic men and 45 matched non-impotent diabetic controls revealed greater psychological stress and poorer glycaemic control in the impotent group. There were no significant differences in the frequency of peripheral neuropathy, retinopathy, nephropathy or large vessel disease in the 2 groups. 20% (9) of the impotent diabetic group had definite autonomic neuropathy which was not present in any of the non-impotent group. The frequency of early autonomic damage was, however, similar in both groups. Impotence in diabetic men is associated with hyperglycaemia, stress and definite autonomic neuropathy. Our data suggest that men with early autonomic neuropathy are frequently potent.