Are general practice characteristics predictors of good glycaemic control in patients with diabetes? A cross-sectional study

Finding ways to better care for people with diabetes and other chronic conditions is a priority for the Australian health system. It is notable that five of the nine national health priority areas established by Australian governments1 are chronic conditions (diabetes, mental health, asthma, arthritis, obesity and dementia). Increasing survival of patients with cancer (one of the three other priority areas, together with cardiovascular health and injury prevention) means that this could also be considered a chronic condition.

Diabetes is one of the fastest growing chronic conditions in Australia, with National Health Survey reports showing an increase in prevalence from 1.3% in 1989–90 to 4.5% in 2011–12.2 This increase is probably linked with the ageing of our population, rising levels of obesity, and greater life expectancy for those with diabetes.2 Diabetes expenditure is also growing rapidly, with the annual costs (after adjusting for inflation) rising from $811 million in 2000–01 to $1507 million in 2008–09, an average annual growth rate of 10%.2

Recent studies have shown that there are opportunities for improving care for those with diabetes. For example, the CareTrack Australia study, which explored the appropriateness of the care provided in general practice, found that only 63% of patients with diabetes received optimal care.3 Further, data from the Australian Institute of Health and Welfare Institute indicate that in 2009–10 only 18% of patients with diabetes had completed the recommended annual cycle of care (ACC).4

It is clear that there are many possible reasons why an individual patient might not receive optimal care. For example, a patient might have several comorbid conditions that militate against the use of a recommended medication. Patients themselves might not adhere to the recommended therapy, or be able to afford a prescribed medication. The workloads of individual general practitioners differ, and the amount of training in chronic disease management is variable.

Practice characteristics, such as the employment of a practice nurse or the use of medical software, might also affect the appropriateness of care. Studies of the relative contributions of these factors have produced mixed results. For example, a study of 422 general practices providing care for 154 945 patients with diabetes in the United Kingdom found no relationship between practice size and the quality of diabetes management (defined as achieving HbA_1c levels of 53 mmol/mol or less).5

An American study of 40 medical groups found that the practice health systems (including formal quality assurance activities and data feedback), clinical information systems (including a diabetes registry), and decision support (including clinical guidelines, clinician prompts for diabetes) were all associated with patients with diabetes having HbA_1c levels of 64 mmol/mol or less.6 It is notable that they found no association between glycaemic control and the practice having a diabetic educator, a dedicated specialist diabetic nurse, a primary care team, patient reminders, or self-management training for patients.

A systematic review of the impact on glycaemic control of 11 different categories of practice quality improvement found that most produced small to modest benefits for glycaemic control.7

A study of 10 health centres in the Netherlands (45 GPs) found that the quality of care for patients with diabetes was higher if the centre had a diabetes education program, yearly medical check-ups by both the GP and a nurse practitioner, and structured follow-ups.8

A 2009 Australian study found that a GP service incentive payment (SIP) scheme for improving chronic disease management increased the probability of a GP requesting an assessment of HbA_1c levels by 20%.9 A similar study found that practices receiving practice nurse support were more likely to claim SIP payments for diabetes care, whereas practice size, diabetes activities, and the sex and age of the GPs were not associated with SIP payments.10

In this article, we present the results of our analysis of baseline data collected as part of the Diabetes Care Project (DCP), a federal government-funded project conducted between 2011 and 2014 to evaluate whether changes to the way in which primary care is organised and funded might improve care for people with diabetes.11 The DCP was a pragmatic clustered, three-armed, randomised controlled trial that evaluated the efficacy of coordinated patient care and flexible funding for GPs in the management of patients with diabetes. Clustering was at the level of the general practice. The three arms of the study were the control group (usual care), intervention group 1 (installation of the online chronic disease management support tool cdmNet12 and access to training and capability building), and intervention group 2 (installation of cdmNet, altered funding arrangements, provision of a care facilitator, and access to training and capability building). The primary outcome measure was defined as a change in patient HbA_1c levels at the end of the trial. Participants were included if they were at least 18 years old, had established type 1 diabetes mellitus (diagnosed at least 12 months ago) or newly diagnosed or established type 2 diabetes mellitus, and had the capacity to provide informed consent. Exclusion criteria included having a terminal illness or dementia, being pregnant or planning to become pregnant, and participating in the Coordinated Veterans Care program.

Our aim was to assess whether the characteristics of GP practices were associated with good glycaemic control in patients with diabetes and with completing an ACC. The dataset for our cross-sectional study, in contrast to that used in the DCP, was not risk-stratified, and we excluded newly diagnosed patients. In addition, our study includes the results of a separate survey of GP practices that was not part of the DCP study protocol.

Methods

A full medical history was obtained and patient demographic data were collected at baseline, as well as data on a number of clinical indicators, including HbA_1c levels. This information was obtained from the medical records in each practice. Practice characteristics were collected by telephone interview with the relevant staff of each practice (Box 1). Glycaemic control was defined as having an HbA_1c level of 53 mmol/mol or less.

The ACC is a measure of the clinical management of diabetes according to Australian national guidelines.4 Completion of the ACC enables GPs to submit a claim to Medicare for the costs of managing a patient with diabetes. For the ACC to be complete, all of the following actions need to have been undertaken: annual measurement of HbA_1c level; comprehensive eye examination every two years; twice yearly assessments of body mass index, blood pressure and feet; annual measurement of total cholesterol, triglyceride and high-density lipoprotein cholesterol levels; and annual test for microalbuminuria. Information on whether the ACC had been completed in the 18 months before starting the DCP was also available from Medicare as part of the baseline dataset.

The original sample size for the DCP was based on its being a three-armed, clustered, randomised controlled trial that compared changes in HbA_1c levels in the three groups during the study period. We excluded all newly diagnosed patients in this sample, as they would not have had time to have a completed ACC, leaving a sample size of 5455 patients. Logistic regression of a binary response variable (Y) on a binary independent variable (X) with a sample size of 1835 observations (of which 50% are in the group X = 0 and 50% are in the group X = 1) would achieve 80% power at P < 0.05 for detecting a difference corresponding to an odds ratio of 1.3. To allow for the clustered design, we assumed an intraclass correlation coefficient of 0.05, providing a design effect of approximately 3; this was equivalent to a required sample size of 5505 (ie, 3 × 1835) individuals, close to the actual sample size.

Practice predictors of good glycaemic control were assessed at an unadjusted univariate level, and after adjustment for patient age, sex, duration of diabetes, patient complexity (a dichotomous variable based on macro- and microvascular complications of diabetes, and comorbid renal disease and depression), and the Australian Bureau of Statistics Socio-Economic Indexes for Areas (SEIFA) score for relative socio-economic disadvantage.13 Odds ratios and 95% confidence intervals were calculated using clustered logistic regression models.

Practice predictors of a completed ACC were initially assessed at the univariate level, and then in a multivariate model that included all variables significant at P ≤ 0.05 in the univariate analyses. Adjustment for age, sex, duration of diabetes, patient complexity and SEIFA score was undertaken for all models, including the multivariate model, but these adjustments had little effect on the results; these analyses are therefore not presented in this article. All models involving patient-level data were adjusted for clustering by practice. Stata 14 (StataCorp) was used for all statistical analyses.

The study protocol was approved by the human research ethics committees of the Department of Health and Ageing (application number 15/2011), the Department of Human Services (2010/Co09591), the Australian Institute of Health and Welfare (EC 2011/4/38), SA Health (HREC 474/09/2014) and Queensland Health (HREC/11/QTDD/65), and by the Aboriginal Health Research Ethics Committee (Aboriginal Health Council of South Australia; reference 04-11-471). On the basis of the above approvals, the Department of Health, Victoria, determined that the study did not need review by their Human Research Ethics Committee.

Results

Data were available for 5455 patients and for 147 of 150 medical practices (98%). The mean age of the patients was 65.3 ± 11.6 years; 55.9% were men. The median time since diagnosis of diabetes was 9 years (interquartile range, 4–15 years). In general, these data were consistent with data for the general Australian population of people with diabetes. The enrolled GP practices had fewer solo GPs and more practice nurses than the national average.11

Of the 5455 patients, 55.0% (95% CI, 53.7%–56.4%) had good glycaemic control (HbA_1c ≤ 53 mmol/mol); 37.0% (95% CI, 35.8%–38.3%) had completed an ACC in the past 18 months.

Using logistic regression of glycaemic control on practice identification number treated as a dummy variable, the McKelvey and Zavoina pseudo-R² indicated that GP practice accounted for 5% of the variability in glycaemic control. The patient characteristics of age, sex, duration of diabetes, condition complexity and SEIFA score together accounted for 13.5% of variability in glycaemic control.

Box 2 presents data on the relationship between each of the practice characteristic variables and glycaemic control. Although the patient having a completed ACC is, strictly speaking, a patient characteristic, it is nonetheless a measure of practice quality and has therefore been included in this table.

In the unadjusted analysis, only the practice having regular multidisciplinary team meetings (odds ratio [OR], 1.16; 95% CI, 1.03–1.31) and the patient having completed an ACC (OR, 1.21; 95% CI, 1.07–1.37) were significantly associated with patients being in glycaemic control. After adjustment for age, sex, duration of diabetes, condition complexity and SEIFA, only the association with the patient having completed an ACC remained statistically significant.

Data for predictors of the patient having completed an ACC in the past 18 months are shown in Box 3. At the univariate level, the practice having a chronic disease-focused nurse (OR, 2.01; 95% CI, 1.07–3.77), the practice having regular staff educational events (OR, 1.68; 95% CI, 1.03–2.73), and the practice offering diabetes education events for patients (OR, 1.92; 95% CI, 1.21–3.06) were all statistically significantly associated with completion of an ACC. In the multivariate analysis, however, only the practice having a chronic disease-focused nurse and the practice running diabetes education events for patients were statistically significant.

Discussion

It is notable that neither having at least one practice nurse nor a chronic disease-focused practice nurse were directly associated with good glycaemic control in patients with diabetes. A review of the literature identifies mixed findings about the value of practice nurses. For example, a 2003 Cochrane review concluded that the availability of a diabetes specialist nurse or nurse case manager may improve patients’ diabetic control for short periods, but there was not enough evidence to demonstrate this effect in the longer term.14 On the other hand, a retrospective cohort study of 397 patients with type 2 diabetes recruited from five general practices in the Netherlands found that delegating diabetes care to a practice nurse led to improvements in diabetes care, including a statistically significant drop in HbA_1c levels.15 Finally, an observational study of 193 Danish general practices and 12 960 patients with type 2 diabetes found that the proportion of patients with HbA_1c levels of 64 mmol/mol or more was lower in practices with well implemented, nurse-led type 2 diabetes consultations.16

Having completed an ACC was the only variable that was significantly associated with glycaemic control after adjustment for patient characteristics. It is interesting that the only published study that examined the association between the completion of an ACC and health outcomes found that the physical functioning of women with diabetes who had completed an ACC was poorer than in women who had only had their HbA_1c levels measured.17 Our finding that patients who had completed an ACC were more likely to be in glycaemic control is the first to indicate the usefulness of this quality indicator.

The practice having a chronic disease-focused practice nurse, the practice offering diabetes education sessions for staff, and the practice organising self-management activities for patients with diabetes were features significantly associated with completing an ACC in the univariate model, although the relationship with staff education was not significant in the multivariate model. This highlights the need to have practice nurses trained in chronic disease management. This recommendation is also supported by the finding of a Cochrane review that the addition of patient education or elevating the role of the nurse to involvement in complex intervention strategies seems important for improving both patient outcomes and process outcomes.18

It would appear that practice characteristics, apart from those described above, are only weak predictors of the patient having completed an ACC. It is unfortunate that we could not obtain information about individual GPs in each practice, and it is likely that their individual characteristics would have a greater influence on the completion of the ACC than information about the practice as a whole, just as patient characteristics explained more of the variability in glycaemic control than did practice characteristics.

As with all cross-sectional studies, it is unwise to attribute causality to any of the detected associations. Practices nominated themselves to participate in the program, and were thus possibly more interested in diabetes management than those that did not. Finally, practice data, including whether the nurse was a practice nurse or a chronic disease-focused nurse, were self-reported, introducing the possibility of bias; further, the analysed data were based on responses to a checklist of practice attributes completed by practice staff, rather than a direct assessment of these features.

The findings of this study underline the importance of having practice nurses trained in chronic disease management, as well as the practice providing education to its diabetic patients. It is also important that this is the first study to find an association between completion of an ACC and good glycaemic control, and GP practices are therefore encouraged to ensure that their patients with diabetes complete an ACC.

Box 1 –
Practice characteristics assessed for this study

Practice location
Practice size
Practice has a practice nurse
Practice has a chronic disease-focused practice nurse
Practice currently uses chronic disease management software
Corporate practice (ie, part of a larger group of practices)
Allied health professionals located in the practice
Practice participated in the Australian collaborative quality improvement program in the past 24 months
Practice takes part in audit and feedback
Practice has dedicated staff member who coordinates care or manages cases
Practice has regular multidisciplinary team meetings
Practice has regular staff education events
Practice uses shared electronic medical records with care team
Practice has education events for patients with diabetes
Practice has formal motivation and self-management education activities for patients with chronic diseases

Box 2 –
Practice characteristics associated with good glycaemic control: univariate models

Variable	Category	HbA_1c > 53 mmol/mol		HbA_1c ≤ 53 mmol/mol		Odds ratio	95% CI for odds ratio	P*	P^†
Variable	Category	n	%	n	%	Odds ratio	95% CI for odds ratio	P*	P^†

Metropolitan practice	No	706	46.3%	820	53.7%	1.00
Yes	1743	44.4%	2186	55.6%	1.08	0.93–1.25	0.310	0.470
Practice size	1–2 GPs	559	45.5%	670	54.5%	1.00
≥ 3 GPs	1882	44.7%	2327	55.3%	1.03	0.88–1.21	0.704	0.413
Practice nurse	No	408	44.7%	504	55.3%	1.00
Yes	2002	44.9%	2462	55.1%	1.00	0.82–1.21	0.965	0.575
Chronic disease-focused nurse	No	440	48.1%	474	51.9%	1.00
Yes	2009	44.2%	2532	55.8%	1.17	0.95–1.44	0.135	0.375
Chronic care planning software already used	No	1589	45.3%	1918	54.7%	1.00
Yes	830	44.3%	1042	55.7%	1.04	0.90–1.20	0.593	0.383
Corporate practice	No	2142	45.1%	2612	54.9%	1.00
Yes	240	43.2%	316	56.8%	1.08	0.91–1.29	0.392	0.466
Co-located allied health professionals	No	1375	45.3%	1660	54.7%	1.00
Yes	1074	44.4%	1346	55.6%	1.04	0.90–1.19	0.596	0.634
Practice involved in quality improvement collaboration	No	1626	45.6%	1939	54.4%	1.00
Yes	816	43.7%	1051	56.3%	1.08	0.93–1.26	0.329	0.839
Practice has audit and feedback	No	1184	44.4%	1483	55.6%	1.00
Yes	1255	45.5%	1502	54.5%	0.96	0.83–1.09	0.512	0.692
Practice has dedicated case management	No	1146	44.2%	1448	55.8%	1.00
Yes	1293	45.7%	1537	54.3%	0.94	0.82–1.08	0.376	0.436
Practice has regular multidisciplinary team meetings	No	1726	46.0%	2028	54.0%	1.00
Yes	618	42.4%	839	57.6%	1.16	1.03–1.31	0.027	0.226
Practice has regular staff education	No	1067	45.2%	1291	54.8%	1.00
Yes	1372	44.8%	1694	55.2%	1.02	0.89–1.17	0.774	0.877
Practice uses shared electronic medical records	No	1551	44.9%	1901	55.1%	1.00
Yes	894	44.9%	1099	55.1%	1.00	0.86–1.16	0.969	0.944
Practice has patient diabetes education events	No	1719	44.8%	2117	55.2%	1.00
Yes	703	45.2%	852	54.8%	0.98	0.83–1.16	0.851	0.198
Practice has self-management activities	No	1518	45.2%	1840	54.8%	1.00
Yes	921	44.6%	1145	55.4%	1.03	0.89–1.18	0.729	0.913
Completed annual cycle of care	No	1603	46.6%	1833	53.4%	1.00
Yes	846	41.9%	1173	58.1%	1.21	1.07–1.37	0.002	0.011

* Based on clustered logistic regression. † Adjusted for age, sex, duration of diabetes, Socio-Economic Indexes for Areas (SEIFA) score and level of condition complexity.

Box 3 –
Practice characteristics associated with a completed annual cycle of care (ACC)

Variable	Category	ACC completed		ACC not completed		Odds ratio	95% CI for odds ratio	P*	P^†
Variable	Category	n	%	n	%	Odds ratio	95% CI for odds ratio	P*	P^†

Metropolitan practice	No	890	58.3%	636	41.7%	1.00
Yes	2546	64.8%	1383	35.2%	0.76	0.46–1.24	0.202
Practice size	1–2 GPs	854	69.5%	375	30.5%	1.00
≥ 3 GPs	2568	61.0%	1641	39.0%	1.46	0.82–2.59	0.202
Practice nurse	No	654	71.7%	258	28.3%	1.00
Yes	2739	61.4%	1725	38.6%	1.60	0.89–2.88	0.120
Chronic disease-focused nurse	No	690	75.5%	224	24.5%	1.00
Yes	2746	60.5%	1795	39.5%	2.01	1.07–3.77	0.029	0.036
Chronic care planning software already used	No	2216	63.2%	1291	33.3%	1.00
Yes	1151	61.5%	721	34.5%	1.08	0.63–1.84	0.792
Corporate practice	No	2922	61.5%	1832	38.5%	1.00
Yes	436	78.4%	120	21.6%	0.44	0.18–1.08	0.073
Co-located allied health professionals	No	1951	64.3%	1084	35.7%	1.00
Yes	1485	61.4%	935	38.6%	1.13	0.69–1.86	0.623
Practice involved in quality improvement collaboration	No	2326	65.3%	1239	34.8%	1.00
Yes	1093	58.5%	774	41.5%	1.33	0.77–2.31	0.312
Practice has audit and feedback	No	1707	64.0%	960	36.0%	1.00
Yes	1698	61.6%	1059	38.4%	1.11	0.68–1.80	0.677
Practice has dedicated case management	No	1548	59.7%	1046	40.3%	1.00
Yes	1857	65.6%	973	34.4%	0.78	0.48–1.26	0.301
Practice has regular multidisciplinary team meetings	No	2499	66.6%	1255	33.4%	1.00
Yes	803	55.1%	654	44.9%	1.62	0.97–2.72	0.068
Practice has regular staff education	No	1639	69.5%	719	30.5%	1.00
Yes	1766	57.6%	1300	42.4%	1.68	1.03–2.73	0.038
Practice uses shared electronic medical records	No	2239	65.0%	1213	35.0%	1.00
Yes	1187	59.6%	806	40.4%	1.25	0.76–2.08	0.381
Practice has patient diabetes education events	No	2591	67.5%	1245	32.5%	1.00
Yes	808	52.0%	747	48.0%	1.92	1.21–3.06	0.006	0.004
Practice has self-management activities	No	2193	65.3%	1165	34.7%	1.00
Yes	1212	58.7%	854	41.3%	1.33	0.81–2.17	0.262

* Based on clustered logistic regression. † Multivariate model including two variables significant at P < 0.05 in univariate models.

Vitamin D testing: new targeted guidelines stem the overtesting tide

Bilinski and Boyages have previously reported that the frequency of vitamin D testing had risen dramatically in Australia between 2000 and 2010.1,2 Further, testing did not translate to improved health outcomes.3 Since that report,1 Medicare Benefits Schedule (MBS) expenditure on vitamin D testing rose from $109.0 million in the 2009–10 financial year to $151.1 million in 2012–13, falling slightly in 2013–14 to $143.1 million.

An MBS review for vitamin D testing in 2014 recommended targeted testing for high-risk groups only, and against population screening.3 High-risk patients include those with osteoporosis, osteomalacia, disorders of calcium and parathyroid hormone, malabsorption, chronic renal disease, patients with darker pigmented skin or reduced sun exposure, those under 16 years of age and patients taking drugs known to reduce vitamin D levels. Five new Medicare item numbers (66833, 66834, 66835, 66836 and 66837) were introduced in November 2014 to replace the two previous numbers (66608 and 66609) and improve quality use of testing. This report analyses data from the Medicare statistical reporting tool for the first 8 months (November 2014 to June 2015) since the introduction of the new MBS numbers and guidelines.

The present study found that there had been a marked reduction in benefits paid for vitamin D testing (Box). In absolute terms, there was a saving of about $39.46 million (an average fall of 42%) compared with the same time period the year before. The greatest fall occurred in the summer month of February 2015 but the trend continued in the winter months. The number of services for vitamin D per 100 000 population fell from 18 140 in 2013–14 to 14 415 in 2014–15. The savings to the end of the 2014–15 financial year equate to about $42 million and, if the trend continues (ie, a reduction of 42%), the annual savings will be close to $64 million, reducing the annual spend to $60 million.

The new policy has almost halved expenditure in a short period of time and, if sustainable, will result in a large amount of funds to be reinvested. Before this intervention there had been an unsustainable growth in vitamin D testing. This report highlights the impact of various strategies including analysis of general practitioner test-ordering patterns and quality use of pathology testing policy based on good clinical practice and evidence-based medicine. The report highlights the value of regular monitoring and publication of all high-cost and high-volume pathology test item numbers, which will allow professional societies as well as individual clinicians to monitor trend data to look for opportunities to reinvest the scarce health dollar. New real-time business intelligence and Big Data tools have made this task easier.4

The study is limited by the nature of the MBS data, which capture only the number and dollar benefit of service. Further, patient-level data analysis could shed light on appropriateness of testing.

Although a large proportion of Australians (between 31% and 58%) are estimated to have vitamin D deficiency (defined as serum 25-hydroxyvitamin D levels < 50 nmol/L5), according to season, moderate to severe deficiency is uncommon and only present in about 4% of people.2,6 The new testing requirements should allow better targeting of those at greatest risk and those who will benefit most.

Box –
Medicare benefits paid for all vitamin D testing in Australia: July 2013–June 2015

2013–14* ($)

2014–15^† ($m)

Difference ($)

Difference (%)

July

12 772 332

10 901 057

−1 871 275

−14.7

August

11 713 108

10 840 342

−872 766

−7.5

September

12 958 166

14 157 002

1 198 836

9.3

October

12 306 916

11 422 808

−884 108

−7.2

November

11 661 161

9 744 773

−1 916 388

−16.4

December

12 265 479

7 653 015

−4 612 464

−37.6

January

8 276 481

5 133 015

−3 143 466

−38.0

February

13 777 880

6 294 101

−7 483 779

−54.3

March

12 315 477

7 097 905

−5 217 572

−42.4

April

11 416 530

6 301 455

−5 115 075

−44.8

May

12 454 074

5 954 850

−6 499 224

−52.2

June

11 207 630

5 732 827

−5 474 803

−48.8

Total

143 125 234

101 233 150

−41 892 085

−29.3

* MBS item numbers 66608 and 66609. † MBS item numbers 66833, 66834, 66835, 66836 and 66837 for November 2014 to June 2015, and 66608 and 66609 for July 2014 to October 2014, and then drop effectively to zero in the remaining months.

Low stress resistance leads to type 2 diabetes: study

A recent study published in Diabetologia (the journal of the European Association for the Study of Diabetes) has found 18-year-old men with low stress resistance have a 50% higher risk of developing type 2 diabetes in their lifetime.

The population based study examined all 1,534,425 military conscripts in Sweden during 1969–1997 who underwent psychological assessment to determine stress resilience. They had to have had no previous diagnosis of diabetes.

They were followed up for type 2 diabetes from 1987–2012 with the maximum attained age being 62.

After adjusting for body mass index, family history of diabetes, and individual and neighbourhood socioeconomic factors, the research found 34,008 men had been diagnosed with type 2 diabetes.

The study found the 20% of men with the lowest resistance for stress were 51% more likely to have been diagnosed with diabetes than the 20% with the highest resistance to stress.

Authors Dr Casey Crump, Department of Medicine, Stanford University, Stanford, CA, USA, and colleagues in Sweden and the USA admit lifestyle behaviours related to stress including smoking, unhealthy diet and lack of physical activity could be related to the increased risk of diabetes. The study also could not make any assertions about women as it only included male army cadets.

The authors conclude: “These findings suggest that psychosocial function and ability to cope with stress may play an important long-term role in aetiological pathways for type 2 diabetes. Additional studies will be needed to elucidate the specific underlying causal factors, which may help inform more effective preventive interventions across the lifespan.”

Latest news:

Potato consumption linked to gestational diabetes

A study published in the BMJ has found a link between a woman’s pre-pregnancy consumption of potatoes and her chances of suffering gestational diabetes.

The researchers from the Eunice Kennedy Shriver National Institute of Child Health and Human Development and Harvard University tracked 15,632 women over a 10-year period, which resulted in 21,693 singleton pregnancies.

Of these pregnancies, 854 were affected by gestational diabetes.

After taking into account risk factors such as age, family history of diabetes, diet quality, physical activity and BMI, researchers found that higher total potato consumption was significantly associated with a risk of gestational diabetes.

The researchers found that if women substituted two servings of potatoes a week with other vegetables, wholegrains or legumes, there is a 9-12% lower risk of contracting gestational diabetes.

They say one explanation of the findings is that potatoes have a high glycaemic index which can trigger a rise in blood sugar levels thanks to the high starch content.

The most recent Australian dietary guidelines released in 2015 say Australians need to eat less starchy vegetables.

The authors of the study admit that the observational nature of their study means no definite conclusions can be drawn about cause and effect.

However, they conclude: “Higher levels of potato consumption before pregnancy are associated with greater risk of GDM, and substitution of potatoes with other vegetables, legumes, or whole grain foods might lower the risk.”

Latest news:

Can patients presenting with acute coronary syndrome be screened for diabetes using glycosylated haemoglobin?

The prevalence of diabetes in Australia is 7.4%.1 However, it is three times higher in patients admitted to hospital with acute coronary syndrome (ACS).2 It often remains undetected, and the prevalence of unrecognised diabetes in ACS populations is estimated to be 4%–22%.3 Diabetes is an independent predictor of increased mortality risk after myocardial infarction,4 so that early detection is of particular importance. In Australia, the incorporation of elevated glycosylated haemoglobin (HbA_1c) levels (≥ 48 mmol/mol) into the diagnostic criteria for diabetes in 2012 has facilitated its diagnosis in hospital admissions for ACS.5 HbA_1c levels are not affected by the acute stress of the ACS event, and their assessment does not require a fasting sample. We assessed the feasibility of routinely collecting HbA_1c data as part of a prospective cohort study of consecutive ACS admissions to Monash Health, Victoria. We enrolled patients from 1 January 2013 to 30 June 2014 who were over 21 years of age and fluent in English. HbA_1c was routinely assayed by high-performance liquid chromatography (Arkray Adams Glycohaemoglobin Analyzer HA-8160). The study participants were relatively young, and most were men (Box). Assessment of the prevalence of diabetes was based on self-reports or an HbA_1c value of at least 48 mmol/mol. The overall prevalence of diabetes in the sample was 31% (128/414 patients), with a trend towards lower prevalence in those presenting with ST elevation myocardial infarction (STEMI). Of the 373 patients for whom HbA_1c measurements were available, 102 (27%) had values of at least 48 mmol/mol at the time of presentation. Of the 128 patients classified as having diabetes, 12 cases (9%) had previously been unrecognised. Of the patients with diabetes for whom the relevant data were available, HbA_1c was > 53 mmol/mol in 73 of 117 cases (62%), low-density-lipoprotein cholesterol ≥ 1.8 mmol/L (calculated using the Friedewald formula) in 48 of 75 cases (64%), high-density cholesterol ≤ 1.0 mmol/L in 64 of 84 cases (76%), and triglyceride levels ≥ 2 mmol/L in 41 of 109 cases (38%). Our data confirm a very high prevalence of diabetes in patients with ACS, most of whom had suboptimal diabetes and lipid management on admission. We confirm that it is practicable to measure HbA_1c in consecutive ACS hospital admissions. HbA_1c assessment will have supplementary value in the optimal management of ACS hospital admissions. Hyperglycaemia associated with the acute stress of the ACS event would, however, require measurement of fasting plasma glucose levels.

Box –
Selected baseline characteristics of the 414 participants with acute coronary syndrome

Characteristic

Patients with diabetes

Patients without diabetes

Number

128

286

Age, years (median, IQR)

60 (53–66)

58 (49–66)

Sex (male)

101 (78.9%)

226 (79.0%)

Diagnosis*

ST elevation myocardial infarction

35 (27.3%)

117 (40.9%)

Non-ST elevation myocardial infarction

44 (34.4%)

93 (32.5%)

Unstable angina

48 (37.5%)

72 (25.2%)

HbA_1c, mmol/mol (median, IQR)

60.7 (50.8–76.0)n = 117

38.8 (36.6–41.0)^†n = 256

Lipid levels, mmol/L (median, IQR)

Total cholesterol (TL < 3.5 mmol/L)

3.9 (3.3–5.2)n = 110

4.8 (3.8–5.6)^†n = 256

Triglycerides (TL < 2.0 mmol/L)

1.7 (1.2–2.6)n = 109

1.5 (1.0–2.2)^‡n = 256

High-density-lipoprotein cholesterol (TL > 1.0 mmol/L)

0.9 (0.8–1.0)n = 84

1.0 (0.9–1.2)^†n = 213

Low-density-lipoprotein cholesterol (TL < 1.8 mmol/L)

2.0 (1.6–3.0)n = 75

2.9 (2.1–3.8)^†n = 205

IQR = interquartile range. TL = target level. * Missing diagnosis data: one for the diabetes group, four for the non-diabetes group. Diabetes group v no diabetes group: † P < 0.001; ‡ P < 0.002.

Continuous quality improvement and metabolic screening during pregnancy at primary health centres attended by Aboriginal and Torres Strait Islander women

Attending to perinatal risk factors, such as diabetes and hypertension during pregnancy, obesity and excess gestational weight gain,1–5 is important for optimising maternal and infant health outcomes. Pregnancy is also a key period for implementing strategies that prevent long-term adverse health outcomes, as excess gestational weight gain and gestational diabetes mellitus (GDM) are respectively predictors of long-term obesity6 and the development of type 2 diabetes.7

Screening for and follow-up of metabolic risk factors are components of recommended pregnancy care in Australia.8 Ensuring that Aboriginal and Torres Strait Islander (respectfully referred to in this article as Indigenous) women receive such care is expected to contribute to giving babies a healthy start to life and to improving the health of their mothers. In Australia, low birth weight, premature birth and perinatal death are substantially more frequent in Indigenous than in non-Indigenous pregnancies.9 Obesity, pre-existing diabetes and GDM are some of the risk factors that are more common in Indigenous women.3,4,10 Later in life, cardiovascular disease and diabetes are major contributors to the difference in life expectancy between Indigenous and non-Indigenous Australians.11

As care can differ between health centres with different characteristics, such as urban and rural or remote locations,8 effective long-term strategies are needed across a range of settings to facilitate the provision of all components of recommended pregnancy care.12 The Audit and Best Practice for Chronic Disease (ABCD) National Research Partnership13,14 aims to improve the provision of care by primary health care centres (PHCs) serving mainly Indigenous populations. It uses a continuous quality improvement (CQI) framework to increase the efficiency and effectiveness of organisational systems. Previous ABCD Partnership research indicates that increases in self-ratings of organisational systems are associated with improvements in the delivery of health care for those with type 2 diabetes.15

We investigated screening for metabolic risk factors during pregnancy and follow-up actions by PHCs participating in the ABCD partnership. We also investigated associations between self-ratings by organisational systems and the proportion of women who undergo metabolic screening.

Methods

The study was approved by human research ethics committees in the relevant states and territories, and by Indigenous subcommittees where required.16 The analyses were approved by the Monash University Human Research Ethics Committee (CF12/3434-2012001670).

Study design and setting

The ABCD National Research Partnership study protocol has been described in detail elsewhere.13,16 This partnership links multiple PHCs and stakeholders across the health system in collaborative CQI research.14 One21seventy, the National Centre for Quality Improvement in Indigenous Primary Health Care, supports CQI in PHCs by providing evidence-based practical tools and training.14 The ABCD Partnership has access to One21seventy data from PHCs that have volunteered to participate in research.13,14 This article reports longitudinal analysis of data from 76 PHCs (2592 health records) involved in the ABCD Partnership across five Australian states and territories. The PHCs conducted up to four CQI cycles, comprising 58.5% (168 of 287) of the One21seventy maternal health audits conducted between 2007 and 2012. Twenty-one of the 76 PHCs began maternal health auditing in 2007; 13 commenced in 2008, 13 in 2009, 11 in 2010, 10 in 2011, and 8 in 2012. Depending on their needs, PHCs may focus in some years on CQI activities in other clinical areas; of 50 PHCs that had completed two or more maternal health audits, 11 (22.0%) conducted audits in non-consecutive years.

Intervention: continuous quality improvement cycles

At baseline, systems assessments and audits of health records were conducted and the results provided to PHCs in real-time by an automated CQI reporting system. PHCs use the reports for participatory interpretation and goal setting, and this is followed by the initiation of relevant actions. Data collection was repeated in subsequent years to assess success in improving care (end of cycle 1), and to identify new priorities for improvement (start of cycle 2). PHCs are encouraged to complete one cycle each year.

Maternal health audit tool

Recorded pregnancy care was assessed by auditing the health records of women with a recent pregnancy (mothers with an infant aged 2–14 months, who resided in the community during their pregnancy and attended for pregnancy care at least once).13,16 Audits were conducted by trained auditors (local PHC staff, staff from other PHCs, or CQI facilitators) supported by a standard protocol and regional CQI facilitators. The audit tool and parameters of the outcome measures were based on best practice guidelines, policy and research reports, and stakeholder consultations.16 At each PHC, the auditor used a standard sampling protocol to select a random sample of at least 30 records to audit (if fewer than 30 eligible records were available, all were audited).13

The Systems Assessment Tool

Structured assessments of PHC system strengths and weaknesses were conducted by PHC staff together with a trained external CQI facilitator using the Systems Assessment Tool (SAT).13,15 This consensus process produces a self-reported overall mean score (range, 0–11) for the state of development of PHC organisational systems, and five subscale scores (delivery system design, information systems and decision support, self-management support, external links, and organisational influence and integration).

Key outcome measures

The audit tool collected information on documentation of the following items in each health record:16

body weight, body mass index (BMI) and blood pressure (BP) screening in women attending at earlier than 13 weeks’ gestation;
BP checks at any point during the pregnancy;
a 50 or 75 gram glucose challenge test (GCT) and, if indicated, an oral glucose tolerance test (OGTT) at 20–30 weeks’ gestation;
for women with a BMI under 20 or over 30 kg/m²: development of a BMI management plan;
for women with high BP (≥ 140/90 mmHg): repeated BP measurements, urine tests for protein, examination by or referral to a general practitioner or obstetrician, or prescription of anti-hypertensive medication;
an OGTT for those with an abnormal GCT result (plasma glucose concentration ≥ 7.8 mmol/L 1 hour after a 50 g glucose load (morning, non-fasting), or ≥ 8.0 mmol/L after a 75 g glucose load).

“Follow-up” in this article refers to taking the next appropriate action after an abnormal screening result.

Statistical methods

Analyses were conducted using Stata version 12.1 (StataCorp). P < 0.05 (2-sided) was defined as statistically significant. Differences in screening proportions at baseline and at the final audit were assessed with respect to PHC governance, location, population size (t tests or Mann–Whitney U tests) and state or territory (one-way analysis of variance or Kruskal–Wallis tests). Paired t tests assessed differences between the first and last SAT scores. Using each health record as the unit of analysis, random effects logistic regression analysis (generating odds ratios) assessed any associations between metabolic screening and CQI cycle number (Stata xtlogit command). Random effects logistic regression allowed for repeated measures of each outcome (eg, did a patient receive a BP check: yes or no) at each cycle per PHC. This method also allowed for adjustment for similarities in women within each PHC. The reference group comprised audit data from the PHCs before they had conducted a CQI cycle (ie, cycle 0 or baseline). We also tested for a trend to increased metabolic screening with each additional CQI cycle (Stata nptrend command).

For each PHC, the proportion of women receiving screening after each CQI cycle was calculated. Treating each PHC as the unit of analysis, univariable linear regression (generating β coefficients) assessed associations between:

the average proportion of women who underwent screening across all cycles, and average overall or subscale SAT scores;
the total change (from first to final cycle) in the proportion of women who underwent screening, and the total change in overall or subscale SAT scores.

Results

A range of PHC settings were included in the study. Most women who attended these PHCs for pregnancy care were Indigenous Australians (87.9%) (Box 1).

While most women who attended during the first trimester were weighed, the BMI was calculated for less than a third; but women attending after the PHC had conducted at least one CQI cycle were more likely to have had their BMI assessed than women attending PHCs that had not done so. Similar patterns were observed for BP checks at any point during the pregnancy and diabetes screening. Improvements in screening appeared to be sustained over sequential CQI cycles, and there were trends for additional improvements with each additional cycle (Box 2).

At baseline, the only significant differences in screening were those between states and territories for first trimester BP checks (P = 0.04), BP checks at any stage of the pregnancy (P = 0.02) and diabetes screening (P = 0.002). These differences were not significant at the PHCs’ final audits (all P > 0.05).

There were also indications of sustained improvements in the provision of follow-up actions after CQI participation, but the sample sizes were too small for statistical analysis. Follow-up actions for high BP included repeated BP assessment (pre-26 weeks, 88.1%; post-26 weeks, 91.9%), urine tests (pre-26 weeks, 88.1%, post-26 weeks, 83.9%), referral (pre-26 weeks, 85.7% post-26 weeks, 94.3%) and antihypertensive medication (pre-26 weeks, 42.9%, post-26 weeks, 26.4%). Follow-up OGTTs were reported for most women who received an abnormal GCT result. Few women with an abnormal BMI, however, had a documented BMI management plan (Box 3).

Systems assessment data were available for 35 PHCs (46.1%); data were available for more than one time point for 21. The mean overall SAT score at the final cycle (7.36) was statistically significantly higher than at the first cycle (6.23; P = 0.009), but there were no significant differences in SAT subscale scores between the first and final cycles (data not shown). Higher average self-ratings of some organisational systems were associated with greater provision of metabolic screening (Box 4). For example, the average provision of first trimester BP screening was 3.7 percentage points higher for each additional point scored on the SAT information systems and decision support domain. Diabetes screening was associated with higher overall self-ratings, as well as with higher ratings of self-management support systems, and of organisational influence and integration.

In addition, there was a statistically significant association between a one-point increase from first to final assessment in information systems and decision support scores and an increase of 5.7 percentage points in the proportion of women receiving diabetes screening between the first and final audits (β = 5.7; 95% CI, 0.6–10.9; P = 0.03). However, no other significant associations between changes in SAT scores and screening were detected (data not shown).

Discussion

This large longitudinal study of PHCs found substantial improvements in routine metabolic screening in pregnancy associated with participation in a CQI initiative. Improvements were sustained over multiple cycles, with evidence for additional improvements with each consecutive CQI cycle. Initiation of follow-up actions also improved after CQI participation. Higher self-ratings of some organisational systems were significantly associated with greater metabolic screening.

Screening at baseline was incomplete for all the metabolic risk factors investigated, consistent with reports from other Indigenous communities.17 It is unclear whether metabolic screening coverage in other maternity care settings is incomplete, as this information is not reported in other routine perinatal data collections. However, improvements associated with CQI participation were observed with respect to BMI and BP assessment and screening for diabetes during pregnancy. Measurement of BMI early in pregnancy is important because maternal and neonatal morbidity increases with maternal BMI,3 and the recommended gestational weight gain depends on the BMI category.1 Measurement of BMI may be influenced by both the mothers’ and health professionals’ understanding of the importance of healthy gestational weight gain and awareness of weight gain guidelines, and by the confidence of health professionals that they can discuss weight with women without causing undue concern.18 It is encouraging that we encountered no instances of women who declined to be weighed. Similarly, first trimester BP assessment and universal second trimester GDM screening are also recommended in Australia, and these remain areas for improvement. It is important to explore potential barriers to GDM screening, both because the prevalence of diabetes during pregnancy is higher among Indigenous women than in non-Indigenous women4 and because of the importance of diabetes management during pregnancy.4

Pregnancy is an opportune time for health practitioners to discuss weight management with women.19 However, few women in this study with an abnormal BMI had a management plan, which may reflect suboptimal action taken, a lack of documentation of the actions taken, or both. Excess weight gain increases pregnancy risks, such as macrosomia, preterm birth and the need for caesarean delivery,1 as well as the long-term risk of obesity,6 making active management vital for the wellbeing of mother and child. Potential barriers to developing weight management plans include limited resources for referral, food security concerns, and inadequate staff time, especially in remote communities. Development of resources or programs for gestational weight management tailored to the needs of Indigenous women may assist.

Most women with an abnormal GCT result subsequently underwent a diagnostic OGTT. Recent controversy about diabetes screening20 may have created barriers to screening and follow-up. While large-scale implementation of the International Association of Diabetes in Pregnancy Study Group guidelines, starting in 2015,21 may partially resolve these problems, the number of women diagnosed with GDM will also increase,22 with potential resource implications for PHCs.

The positive associations between self-ratings of organisational systems and first trimester BP and diabetes screening in our study support targeting of organisational systems as a strategy for improving the provision of metabolic screening during pregnancy. However, further large-scale improvements in systems and processes that support health professionals in conducting metabolic screening and management are vital if the long-term consequences of these complications in pregnancy are to be reduced. We hope that our findings encourage further discussion about how pregnancy care for Indigenous women might be improved. All levels of the health system have roles to play, and systems-based research networks, such as the ABCD Partnership, are ideally placed to develop appropriate strategies.

Our study was limited by the fact that SAT data were available for only some PHCs (35 of 76, 46.1%), reducing the statistical power of our analysis to detect associations. Selection bias was also possible, as this study included only the One21seventy PHCs that volunteered their data for research (58.5% of the audits conducted overall). Our data may not be representative of PHCs not participating in the One21seventy initiative, but this extensive network includes a large population, and there are currently no other comparable data sources in Australia. Bias caused by the possibility that PHCs with lesser improvement would be less likely to remain in the CQI initiative is difficult to gauge, as commencement years varied and PHCs may have conducted maternal health audits in non-consecutive years. However, the generalisability of our results may have been enhanced by the fact that PHCs used the audit tool according to their needs, rather than as a research requirement. As we performed multiple statistical tests, there was a risk of finding significant associations by chance. This possibility was reduced by not undertaking statistical tests for follow-up actions, as the small numbers involved were inadequate for meaningful comparisons.

The CQI initiative continues, and further assessment of its effects on service delivery and health outcomes is planned as the sample size increases. Future directions include investigating the effects on service provision of the audit year, the year of commencement, and the duration of CQI participation. A cluster randomised controlled trial is an alternative study design that could be used to test hypotheses arising from the current findings.

Despite the limitations, our study has significant strengths that increase the generalisability of its findings. Most previous CQI research in pregnancy care has been hospital-based, implemented in a single service, not focused on metabolic screening, or not conducted in Australia.23–25 Our research applied a unique system-wide participatory approach to assess systemic issues commonly affecting provision of care.14 It used a detailed, longitudinal dataset to investigate long-term sustainability, and included many PHCs across several settings.

Our study shows the potential of a CQI initiative supported by a systems-based research network to improve the provision of recommended pregnancy care at PHCs attended by Indigenous women. These findings are encouraging, and suggest a successful approach for achieving further improvement in pregnancy care provision.

Box 1 –
Characteristics of the 76 primary health care centres included in the study, and of the 2592 women whose records were audited

Characteristics of the primary health care centres

Governance structure

Government-operated

49 (64.5%)

Community-controlled

27 (35.5%)

Location

Remote

56 (73.7%)

Urban or regional

20 (26.3)

Service population size

≥ 1000 people

39 (51.3%)

< 1000 people

37 (48.7%)

State or territory

Northern Territory

28 (36.8%)

Queensland

27 (35.5%)

Western Australia

11 (14.5%)

New South Wales

6 (7.9%)

South Australia

4 (5.3%)

Characteristics of the women

Indigenous status^∗

2141 (87.9%)

Aboriginal

2028 (83.3%)

Torres Strait Islander

57 (2.3%)

Aboriginal and Torres Strait Islander

56 (2.3%)

Age^†

Median, years

24.4 (IQR, 20.6–29.6)

< 20 years

545 (21.1%)

20–34 years

1807 (69.9%)

≥ 35 years

233 (9.0%)

First attendance for pregnancy care occurred before 13 weeks’ gestation^‡

1321 (51.0%)

Median number of pregnancy care visits

7 (IQR, 5–10)

IQR = interquartile range. ∗n = 2435 (data missing for 157 women). †n = 2585 (data missing for 7 women). ‡n = 2591 (data missing for 1 woman).

Box 2 –
Documented metabolic screening during pregnancy after completion of each continuous quality improvement (CQI) cycle, and associations between metabolic screening and primary health care centre (PHC) participation in each CQI cycle

Metabolic screening

CQI cycle

P (for trend)

076 PHCs

150 PHCs

228 PHCs

38 PHCs

46 PHCs

Weight measured in first trimester (1321 women)

440/562 (78.3%)

344/418 (82.3%)

153/202 (75.7%)

49/65 (75.4%)

56/74 (75.7%)

Odds ratio (95% CI)

1.0

1.4 (0.9–2.0) P = 0.10

1.0 (0.6–1.6) P = 0.89

1.2 (0.6–2.4) P = 0.59

1.4 (0.7–2.8) P = 0.34

0.38

BMI calculated in first trimester (1321 women)

132/562 (23.5%)

126/418 (30.1%)

63/202 (31.2%)

25/65 (38.5%)

31/74 (41.9%)

Odds ratio (95% CI)

1.0

2.4 (1.6–3.5) P < 0.001

3.4 (2.0–5.6) P < 0.001

5.1 (2.4–10.7) P < 0.001

9.4 (4.6–19.4) P < 0.001

< 0.001

Blood pressure check in first trimester (1321 women)

485/562 (86.3%)

370/418 (88.5%)

180/202 (89.1%)

56/65 (86.2%)

59/74 (79.7%)

Odds ratio (95% CI)

1.0

1.3 (0.8–1.9) P = 0.27

1.5 (0.9–2.7) P = 0.15

1.6 (0.7–3.7) P = 0.24

1.1 (0.5–2.3) P = 0.78

0.51

Blood pressure check at any point during the pregnancy (2592 women)

1123/1201 (93.5%)

745/758 (98.3%)

383/388 (98.7%)

131/135 (97.0%)

110/110 (100.0%)

Odds ratio (95% CI)

1.0

3.7 (1.9–7.3) P < 0.001

7.0 (2.5–19.4) P < 0.001

2.0 (0.6–6.5) P = 0.25

—

< 0.001

Diabetes screening (2541 women)^∗

669/1192 (56.1%)

469/736 (63.7%)

234/380 (61.6%)

86/135 (63.7%)

74/98 (75.5%)

Odds ratio (95% CI)

1.0

1.3 (1.0–1.6) P = 0.04

1.2 (0.9–1.7) P = 0.15

1.7 (1.1–2.6) P = 0.02

3.4 (1.9–5.9) P < 0.001

< 0.001

BMI = body mass index. ∗In 2010, the audit tool was refined to include “not applicable” if women had already been diagnosed with diabetes, or were offered but declined BMI or blood pressure assessment or diabetes screening. Since 2010, 26 women were recorded as having pre-existing diabetes, and 25 women declined diabetes screening. This reduced the denominator for diabetes screening to 2541. There were no recorded instances of women declining BMI or blood pressure checks.

Box 3 –
Recorded metabolic abnormalities during pregnancy and subsequent follow-up after each continuous quality improvement (CQI) cycle

Metabolic risk factors and follow-up

CQI cycle

76 PHCs

50 PHCs

28 PHCs

8 PHCs

6 PHCs

Abnormal BMI in first trimester (377 women)

39/132 (29.6%)

34/126 (27.0%)

17/63 (27.0%)

5/25 (20.0%)

8/31 (25.8%)

BMI management plan (103 women)

6/39 (15.4%)

10/34 (29.4%)

6/17 (35.3%)

4/5 (80.0%)

4/8 (50.0%)

High blood pressure in first trimester (1150 women)

11/485 (2.3%)

12/370 (3.2%)

5/180 (2.8%)

1/56 (1.8%)

0/59

Blood pressure follow-up < 26 weeks (73 women)

13/32 (40.6%)

17/27 (63.0%)

7/9 (77.8%)

2/2 (100.0%)

3/3 (100.0%)

High blood pressure at any time during pregnancy (2492 women)

72/1123 (6.4%)

51/745 (6.8%)

25/383 (6.5%)

2/131 (1.5%)

8/110 (7.3%)

Blood pressure follow-up ≥ 26 weeks (110 women)

34/49 (69.4%)

30/35 (85.7%)

17/20 (85.0%)

no cases

6/6 (100.0%)

Abnormal GCT result (1530 women)

120/667 (18.0%)

92/469 (19.6%)

41/234 (17.5%)

15/86 (17.4%)

9/74 (12.2%)

Follow-up OGTT (277 women)

104/120 (86.7%)

81/92 (88.0%)

40/41 (97.6%)

14/15 (93.3%)

7/9 (77.8%)

PHC = primary health care centre; BMI = body mass index; GCT = glucose challenge test; OGTT = oral glucose tolerance test.

Box 4 –
Associations between the average proportions of women undergoing metabolic screening and average Systems Assessment Tool scores (across all cycles) for 35 primary health care centres (β-coefficient, 95% CI)

Overall score

Delivery system design

Information systems and decision support

Self-management support

External links

Organisational influence and integration

BMI calculated in first trimester

4.2 (−3.5 to 11.9)

2.7 (−4.8 to 10.2)

5.5 (−1.3 to 12.2)

3.5 (−1.6 to 8.6)

1.9 (−4.5 to 8.4)

1.2 (−5.1 to 7.4)

Blood pressure check in first trimester

2.6 (−0.6 to 5.8)

1.9 (−1.3 to 5.0)

3.7^∗ (0.9 to 6.4)

1.5 (−0.6 to 3.7)

−0.6 (−3.4 to 2.1)

2.5 (−0.0 to 5.1)

Blood pressure check at any point during pregnancy

0.9 (−0.9 to 2.6)

0.5 (−1.2 to 2.2)

1.3 (−0.2 to 2.9)

0.3 (−0.9 to 1.5)

0.3 (−1.2 to 1.8)

0.7 (−0.8 to 2.1)

Diabetes screening

5.3^∗ (0.6 to 10.1)

4.6 (−0.1 to 9.3)

3.8 (−0.6 to 8.2)

3.4^∗ (0.2 to 6.7)

1.2 (−3.1 to 5.4)

4.9^∗ (1.1 to 8.6)

BMI = body mass index. ∗P < 0.05.

Testing times! Choosing Wisely when it comes to monitoring type 2 diabetes

Harnessing the value of self-monitoring of blood glucose among people with non-insulin-treated type 2 diabetes

Affecting over one million Australians, type 2 diabetes (T2D) costs our country an unsustainable $15 billion annually,1 and is predicted to be the nation’s leading cause of disease burden by 2017.2 It is therefore essential to engage people with this condition in cost-effective therapy to reduce these costs, which arise mostly from treating the long-term complications (eg, blindness, stroke, amputation).

Self-monitoring of blood glucose levels (by means of a finger-prick blood sample analysed with an ambulatory blood glucose meter) is an essential part of managing type 1 diabetes and insulin-treated T2D; however, the clinical benefit for people with T2D who are not using insulin is, and we believe remains, a matter of debate.

The Government decides …

On 29 May 2015, the Australian federal government announced that access to testing strips for self-monitoring of blood glucose (SMBG) would be limited for most people with T2D. This announcement followed a much anticipated 2-year review process and extensive consultation. The Pharmaceutical Benefits Scheme (PBS) now stipulates that:

unrestricted access to SMBG strips will continue for people with T2D who are using insulin or other medicines (eg, corticosteroids, sulfonylureas) to detect asymptomatic hypoglycaemia, or during illness that may cause fluctuations in blood glucose level;3 and
access will now be restricted for those with T2D “who are not using insulin and who have their blood glucose level under control. The PBAC [Pharmaceutical Benefits Advisory Committee] recommended that these patients be limited to a six month supply [100 strips] following changes to their diabetes management, with a further six months’ supply available at the prescriber’s discretion.”3

This second stipulation is more specific but consistent with another recommendation released just a month earlier.

Choosing Wisely Australia recommends …

Among its 25 recommendations published on 29 April 2015, Choosing Wisely Australia (an initiative of NPS MedicineWise), in collaboration with the Royal Australian College of General Practitioners, made only one about diabetes. This was: “Don’t advocate routine self-monitoring of blood glucose for people with type 2 diabetes who are on oral medication only.”4

Originating in the United States, Choosing Wisely is a laudable global movement encouraging clinicians and consumers to question the use of unnecessary medical tests, treatments and procedures.

In the US and Canada, Choosing Wisely recommendations for diabetes have focused similarly on restricting SMBG strips for people with non-insulin-treated T2D; in the United Kingdom, recommendations are expected in late 2015.

Despite nuances of language in these international recommendations, SMBG among people with non-insulin-treated T2D is clearly a “hot topic”.

The evidence base indicates …

In 2012, two highly influential systematic reviews — a Cochrane review and a meta-analysis — were published.5,6 Based largely on the same set of randomised controlled trials (RCTs), their conclusions were comparable: “clinical benefit is limited” for SMBG in people with non-insulin-treated T2D.

The Cochrane review5 included 12 RCTs (3249 participants). Among these, nine trials of 6 months’ duration found that glycated haemoglobin (HbA_1c) levels were reduced on average by 0.3% (a statistically, but not clinically, significant improvement).5 There was no significant reduction in HbA_1c levels in trials with 12 months of follow-up. Overall, no benefit was shown for patient satisfaction, emotional wellbeing or health-related quality of life, and SMBG was considered unlikely to be cost-effective.5

Challenging the evidence and assumptions

Our own critical appraisal revealed too much variation in trial methods and populations to draw firm conclusions about the value of SMBG overall.7 In particular, in some trials, participants were not given instructions about when or how often to check their blood glucose level (or this was not reported). Among trials where frequency was reported, it varied enormously — from four times per month to six times per day, 7 days per week. In most cases, the SMBG conducted was insufficient to provide interpretable blood glucose patterns that could inform diabetes self-management and lifestyle choices (eg, food intake or physical activity). Some studies incorporated feedback and education about self-management, but others did not.

We refer to this random, low frequency, routine SMBG as “unstructured”, and suggest it is ineffective because it does not enable people with T2D or health professionals to detect blood glucose level patterns or act upon them. Indeed, people with non-insulin-treated T2D reported that their GPs rarely refer to their glucose diary data, and perceive this to mean that SMBG is worthless.7 They experience SMBG as “frustrating”, “painful”, “inconvenient” and “expensive”, they lack motivation for it, and report “feelings of failure or anxiety in response to high blood glucose readings”.7

However, in studies where the protocol for a “structured” approach to SMBG was clearer, the findings were more positive — reduced HbA_1c levels, less glycaemic variability overall, less time spent in hyperglycaemia.5,7

Structured monitoring is effective, economical and engaging

After the systematic reviews were concluded, an RCT of structured SMBG was published.8 In the STeP study, structured SMBG was defined as seven checks per day over 3 consecutive days in the week before their consultation with a doctor about their diabetes.8 STeP showed that structured SMBG was associated with a statistically significant reduction in HbA_1c level (− 0.3%; P < 0.001; intention-to-treat analysis), and a per protocol analysis (focused on those who conducted structured SMBG as intended) showed a clinically significant reduction in HbA_1c level (− 0.5%).8

Notably, trials of structured SMBG have also shown important psychological benefits — increased satisfaction with treatment, reduced diabetes-related distress, improved general emotional wellbeing and greater confidence in, and motivation for, diabetes self-care.7–9

The findings of the STeP study suggest that SMBG does not have, as such, a dose-related response, and needs to be viewed, rather, in terms of quality rather than quantity of monitoring.8 The protocol suggests that a person with non-insulin-treated T2D using structured SMBG could use as few as 84 test strips per year (ie, 21 in the week before each quarterly general practitioner visit). This, in fact, compares very favourably with the current Australian average of 300 strips per annum per person with non-insulin-treated T2D, and suggests great potential for the federal government’s restricted access policy (100 strips over 6 months) to be applied sensibly.

Our recent observational study, STeP-IT-UP, (involving 98 people with non-insulin-treated T2D attending 22 general practices across our eastern seaboard), showed that structured SMBG is feasible in Australia.10 Furthermore, our findings support US and European evidence showing significant reductions in HbA_1c levels (without increasing hypoglycaemia) and diabetes-related distress.

What is structured SMBG?

Structured SMBG is more than just 21 finger pricks. It involves meaningful (rather than random) glucose checks at set times (eg, pre-meal and 2 hours post-meal, and before bedtime) to generate a pattern over at least 3 consecutive days. The person with T2D also notes their meal sizes and energy levels to provide context for the readings. While most trials have evaluated SMBG as though it were an active agent, it is actually just one aspect of a complex intervention, requiring:

agreement between the person with T2D and their health professional on glucose targets and the timing and frequency of SMBG;
a supportive and enthusiastic health professional trained in the interpretation of SMBG data;
appropriate feedback to, and education for, the person with T2D;
collaborative review of the SMBG pattern to determine areas for improvement and to discuss what contributed to low, high or within-target glucose levels;
a plan for how to change diet, activity levels or medication to improve glucose levels;
action (ie, actual change in diet, activity levels or medication); and
motivation on the part of the person with T2D, which is likely to be contingent on much of the above being in place.

A closer look at the Choosing Wisely Australia recommendation

We take issue with Choosing Wisely’s initial statement, that there “is no evidence that self-monitoring of blood glucose (SMBG) affects patient satisfaction, general well-being or general health-related quality of life.”4 There is compelling evidence on both sides of this debate, depending on whether SMBG is structured or unstructured.7

Choosing Wisely claimed that Australian Government spending on glucose monitoring strips was $143 million in 2012.4 This is true, but misleading. Only 35% of this spending was for people with non-insulin-treated T2D.3 Most of this funding was for SMBG essential for informing insulin dosing and detecting hypoglycaemia in people with type 1 diabetes and those with T2D using insulin or sulphonylureas. Substantial cost savings therefore seem unlikely.

The most positive aspects of the Choosing Wisely recommendation are the exceptions, in particular the usefulness of SMBG for “short-term education about diet influencing blood sugar”, although we would expand this to include physical activity.

Choosing more Wisely Australia

We appreciate absolutely the need for evidence-based medicine — and have described the complexity of this evidence base. Nevertheless, we remain concerned that restricting access to glucose monitoring strips conveys the wrong message philosophically. At face value, it implies that some forms of diabetes require less monitoring and are, therefore, less serious than others. Yet all diabetes is serious and all diabetes leads to complications if not monitored and managed appropriately: conveying any other message is confusing, inaccurate and potentially dangerous.

As with most behaviour, if individuals do not value it, or perceive more costs than benefits, they are unlikely to instigate or maintain the behaviour. This applies not only to people with non-insulin-treated T2D, but also to health professionals. While the government is undoubtedly interested in potential costs savings, the PBS final report also recognises the need to emphasise to clinicians and people with T2D that “changes are being implemented to encourage better practice and direct more attention to appropriate use of test strips”.3

Far from recommending against routine SMBG, which may unintentionally deter any SMBG in people with non-insulin-treated T2D, we believe Choosing Wisely Australia should positively advocate structured SMBG for all people with T2D not using insulin or other hypoglycaemia-inducing medications. This would be more consistent with its mission not only to reduce unnecessary medical tests, but also to promote evidence-based clinical practice. Structured SMBG offers an evidence-based model for effective blood glucose monitoring and engagement in diabetes self-management.

Life expectancy improvements for people with Type 1 diabetes

A team of Endocrinologists say that despite the incidence of Type 1 diabetes having doubled in the last 20 years, there is significantly improved life expectancy for sufferers.

Prof Peter Colman, Dr Mervyn Kyi and a team at the Royal Melbourne Hospital wrote about their findings in a clinical focus in the Medical Journal of Australia.

They wrote that new technologies, such as more convenient blood glucose meters with built-in bolus dose calculators, smartphone applications, insulin pumps, continuous glucose-monitoring systems and closed-loop insulin systems have helped patients better manage their condition.

They also said that trials such as using type 2 diabetes medications to help reduce cardiovascular disease in T1D are underway and transplantation of the pancreas or the islets of Langerhans are being performed at hospitals in Sydney and Melbourne.

Kyi and colleagues are hopeful that one day the disease might be able to be prevented.

“The ability to predict T1D on the basis of genetic, immunological and metabolic markers has provided opportunities for prevention at different preclinical stages”, they noted. “Much attention has focused on interventions at diagnosis and in the preclinical antibody positive stage,” they wrote.

Despite the advancements, they said that type 1 diabetes is still associated with considerable premature mortality.

They commented that: “T1D is still associated with considerable premature mortality caused by acute and chronic complications, particularly ischaemic heart disease.

“Recent reports of improved life expectancy … nonetheless provide great hope for persons with T1D and their clinicians”, they concluded.

Read the full report in the Medical Journal of Australia.

Recent advances in type 1 diabetes

Type 1 diabetes (T1D) affects around 120 000 Australians, half of whom are diagnosed in adulthood.1 It is caused by the immune-mediated destruction of pancreatic beta cells, leading to insulin deficiency, hyperglycaemia and the risk of ketoacidosis. Antibodies directed against the beta-cell antigens insulin, glutamic acid decarboxylase 65 (GAD65), insulinoma-associated protein 2 (IA-2) and zinc transporter 8 (ZnT8) are markers of T1D autoimmunity used to confirm the diagnosis of T1D and to identify normoglycaemic people at high risk of progressing to T1D.2 The incidence of T1D has doubled in Australia during the past 20 years,3 leading to speculation that it is caused by environmental or epigenetic factors. Environmental changes that may play a pathogenic role include viral infections, a more hygienic environment, and increased caloric intake with associated weight gain.4

Some individuals with insulin deficiency presenting as T1D do not express the typical beta-cell autoantibodies, and alternative diagnoses, such as early-onset type 2 diabetes or idiopathic (type 1b) diabetes, should be considered. Another possibility is monogenic diabetes, particularly in children with a strong family history and unusual clinical features, such as renal impairment or exocrine pancreatic insufficiency.

General management principles

Since the 1990s, the most common treatment strategy has been the combination of once- or twice-daily injections of long-acting insulin (eg, insulin detemir or glargine) and short-acting insulin (eg, neutral insulin or the aspart, lispro and glulisine insulin analogues) taken with meals. An individual’s required dose of long-acting (basal) insulin is relatively constant, but can vary with exercise and physiological stressors, such as illness. Glargine and detemir insulin are commonly preferred to NPH (neutral protamine Hagedorn) insulin because their use is associated with lower rates of nocturnal and severe hypoglycaemia. Detemir is also associated with less weight gain than insulin NPH. All three long-acting insulins achieve similar reductions in glycated haemoglobin (HbA_1c) levels.5 Novel ultralong-acting insulin analogues are being developed. Insulin degludec provides basal insulin coverage for more than 40 hours, and achieves similar glycaemic control with less overnight hypoglycaemia than glargine.6 Degludec is approved for use in Europe, but is yet to be approved by the United States Food and Drug Administration or the Australian Therapeutic Goods Administration.

Bolus (meal-time) insulin requirements in T1D are more variable than basal insulin requirements, depending primarily on carbohydrate intake. Accordingly, many people with T1D learn to determine the carbohydrate content of their meals (“carbohydrate counting”) to guide their prandial insulin dose. Studies of advanced carbohydrate counting in T1D suggest that it reduces HbA_1c levels and the frequency of hypoglycaemia.7

Ideally, people with T1D learn to self-adjust their bolus insulin dose to achieve optimal glycaemic control, using calculations based on personal experience and clinical observation. The specific formulas, incorporating the insulin-to-carbohydrate ratio and the insulin sensitivity factor, are provided in the Box. Self-adjustment of insulin dose can be learned informally or during structured training courses, such as the Dose Adjustment for Normal Eating (DAFNE) program (www.dafne.org.au).8 Central to self-adjustment of insulin dose is the self-monitoring of blood glucose levels. This requires a minimum of four glucose checks each day: once before each meal, and at bedtime. Glucose checks 2 hours after a meal can also be used to assess the adequacy of the bolus insulin dose for the preceding meal.

Variability in both insulin absorption after subcutaneous injection and of carbohydrate absorption from the gut can contribute to glycaemic variation in T1D. Areas of lipohypertrophy that can result from repeated insulin injections at the same site may further impair insulin absorption. Development of lipohypertrophy is less likely if the injection site is rotated regularly.

Exercise and illness are also potent modifiers of glycaemia. The acute glycaemic effects of exercise can vary according to the intensity of physical activity, but there is generally a tendency to delayed hypoglycaemia and a reduced insulin requirement.

All individuals should have a thorough knowledge of sick-day management, including more frequent glucose monitoring, ketone testing, self-adjustment of insulin dose, maintaining oral intake of food and liquid, and having a low threshold for seeking medical attention.

As with any chronic disease, psychosocial factors strongly influence T1D self-management. An individual may fluctuate between active management and total disengagement, depending on social circumstances and psychological factors, such as anxiety and depression. Health care providers should recognise these challenges. Taking social factors into account, referring people with T1D to appropriate psychological services, and gently helping the individual re-engage with self-care usually achieves better outcomes than focusing only on medical management.

Hypoglycaemia is a common and often a feared adverse event that is a major barrier to attaining near-normal blood glucose levels. Adults with T1D experience, on average, two mild hypoglycaemic episodes each week. More severe hypoglycaemia (requiring external assistance for recovery) occurs with an annual prevalence of 30%.9 The risk of hypoglycaemia increases with the duration of disease and is inversely correlated with HbA_1c levels. Hypoglycaemia is a recognised cause of seizure and coma, and recent evidence suggests it may also cause serious cardiac ischaemia and arrhythmia. In addition, recurrent severe hypoglycaemia in children with early-onset T1D is associated with lower cognitive test scores in adulthood.9 Severe hypoglycaemia and hypoglycaemic unawareness also underlie diabetes-related distress and poorer emotional well-being.10 Glycaemic targets should therefore be individualised, and must always be balanced against the risk of hypoglycaemia. It is encouraging that the incidence of severe hypoglycaemia in a cohort of Australian children with T1D has decreased over the past decade from 17.3 to 5.8 per 100 patient-years.11

New technologies that assist with glucose control

There has been an explosion in the availability of new technologies that assist the self-management of T1D. New blood glucose meters are more convenient than older models, with streamlined electronic data upload and connectivity to health care teams. Blood glucose meters that include a built-in bolus dose calculator (also called “smart meters”) integrate the insulin-to-carbohydrate ratio and insulin sensitivity factor when calculating insulin doses for each meal. There is, however, no strong evidence for improved glucose control in individuals using these devices.12 Several smartphone applications are now available to assist with various aspects of self-management, including carbohydrate counting, glucose monitoring and insulin dose calculations.

Insulin pumps are becoming increasingly popular, and are used by about 10% of Australians with T1D.13 Insulin pumps deliver short-acting insulin through a subcutaneous cannula that the patient re-positions every 3 days. Insulin is delivered continuously, and the rate can be varied to better mimic endogenous insulin production and to match physical activity. Prandial and correctional insulin boluses are administered via the pump under the control of the user. Compared with multidose insulin therapy, pump therapy is associated with a reduction in HbA_1c levels of about 3.3 mmol/mol (0.3%), but, more importantly, also with reduced rates of severe hypoglycaemia and improved quality of life.14 However, their safe and successful operation requires the close engagement of the individual with their pump, their willingness to undertake adequate glucose monitoring, and their proficiency in carbohydrate counting. Insulin pumps cost up to $9500, which can generally be partially or fully reimbursed by private health insurers; the government provides a subsidy for children from low-income families.

Continuous glucose monitoring systems measure interstitial fluid glucose concentrations every 5 minutes via a thin glucose-sensing subcutaneous cannula. Continuous glucose monitors can be used in individuals with T1D for a week at a time to retrospectively assess the 24-hour blood glucose profile and to diagnose unrecognised hypo- and hyperglycaemia. They can also be employed for longer periods, together with an insulin pump, to provide real-time glucose readings. These sensor-augmented pumps enable individuals to adjust insulin doses in response to glucose level trends, and to safely reduce HbA_1c levels by about 3.3 mmol/mol (0.3%).15 Sensor-augmented pumps also incorporate a “low-glucose-suspend” feature that temporarily stops insulin infusion if the blood glucose level is moving towards the hypoglycaemic range.16

Initial trials of closed-loop insulin systems (the so-called “artificial pancreas”) that combine a continuous glucose monitor, a pump delivering insulin alone or in combination with glucagon, and a computer algorithm that determines the insulin dose, suggest that this strategy will deliver further improvements in glucose control.17 These systems are still in development, and are currently available in Australia only through participation in a clinical trial.

Non-insulin treatments that assist with glucose control

In type 2 diabetes (T2D), metformin is used to reduce the required insulin dose, prevent weight gain and reduce cardiovascular risk. More than a third of children and half of adults with T1D are overweight or obese,18,19 and a meta-analysis has shown that metformin therapy is associated with reduced levels of total cholesterol and low-density lipoprotein,20 suggesting that metformin might help prevent cardiovascular disease in T1D. This hypothesis is being tested in the REducing with Metformin Vascular Adverse Lesions in type 1 diabetes (REMOVAL) study currently underway in Europe, Canada and Australia (https://www.clinicaltrials.gov/ct2/show/NCT01483560).

Other T2D medications are being tested for their ability to reduce blood glucose variability in T1D. Pilot trials in T1D populations have found that the glucagon-like peptide-1 analogue liraglutide suppresses glucagon levels and reduces insulin requirements,21 that the dipeptidyl peptidase-4 inhibitor sitagliptin showed a trend to reducing hyperglycaemia in a subset of patients,22 and that the sodium–glucose cotransporter 2 inhibitor dapagliflozin achieved a dose-related (but non-significant) reduction of glycaemic excursions and insulin requirements.23 The use of these drugs in T1D is still experimental, pending further studies.

Treatment targets and complications

The importance of glucose control in T1D was shown by the landmark Diabetes Control and Complications Trial,24 which found that tighter glucose control (HbA_1c level of 54 mmol/mol [7.1%] v 76 mmol/mol [9.1%]) greatly diminished the risk of microvascular complications (retinopathy, nephropathy and neuropathy). Accordingly, the general HbA_1c target for adults is ≤ 53 mmol/mol, although the target is lower in pregnancy (if this can be achieved safely) and less stringent for children and adolescents (< 58 mmol/mol) and those prone to hypoglycaemia.25,26 The HbA_1c target also needs to be tailored to the clinical circumstances of each patient, including their disease duration, life expectancy and comorbidities. Screening for retinopathy (fundal examination), nephropathy (urine albumin/creatinine ratio) and neuropathy (clinical examination) should be performed at least annually, together with assessment of important cardiovascular risk factors, including smoking, blood pressure and lipids. Abnormalities should be treated according to national guidelines.26 Genetic predisposition appears to be an important determinant of nephropathy, although the relevant genes are yet to be identified.

Pregnancy and T1D

Women with T1D can experience healthy pregnancies with excellent outcomes for their infants. However, optimal outcomes require effective preparation and excellent glycaemic control throughout the pregnancy; this generally requires support from an experienced diabetes-in-pregnancy team. Pregnancy preparation begins with educating young women with T1D about safe sexual practices (including contraception) and the need to plan for their pregnancy. All forms of contraception are safe for women with T1D, but hormonal contraception can adversely affect glucose control.

Preconception care includes optimising glucose control, assessing and treating diabetes complications, ensuring adequate folate (2.5 mg–5 mg daily) and iodine (150 μg daily) intake, stopping or substituting potentially harmful medications, reviewing lifestyle factors (including diet, smoking and alcohol), and checking thyroid function, anti-transglutaminase antibodies for coeliac disease, and rubella and varicella immune status. Glucose control should be as close to target as is feasible. It is recommended that HbA_1c be 53 mmol/mol or less for 2–3 months prior to conception, and as low as is safely possible during the pregnancy. This reduces the risks of congenital malformations and miscarriage, the risk for each of which is roughly four times that for the general population.27,28 Insulin pumps can assist women achieve their glucose target, but similar control can be attained with an intensified multidose insulin regimen. Continuous glucose monitors are sometimes used to fine-tune management.

There is an increased risk of hypoglycaemia in early pregnancy, and hyperemesis may complicate glucose management. Insulin requirements usually increase substantially during the second half of pregnancy, when hyperglycaemia accelerates fetal growth. Hypertension, oedema, proteinuria and signs of evolving pre-eclampsia should be carefully monitored and treated aggressively, with consideration given to expediting delivery when clinically appropriate. If pregnancy is progressing well, delivery should be planned for around 38 weeks’ gestation to reduce the risk of stillbirth and to minimise the risk of delivery complications. Although there have been significant improvements in neonatal outcomes, rates of macrosomia and caesarean delivery are higher for mothers with T1D than for the general population.28

After delivery, breastfeeding is encouraged. Insulin requirements change dramatically postpartum, and adjusting the insulin dose with the help of the health care team after discharge from hospital is essential. All outcomes are improved if management is undertaken at a tertiary centre with a specialist diabetes-in-pregnancy team and high-level neonatal care.29

Prospects for the prevention of T1D

The ability to predict T1D on the basis of genetic, immunological and metabolic markers has provided opportunities for prevention at different preclinical stages. Most attention has focused on interventions at diagnosis, a stage when significant amount of endogenous insulin is still produced. Treatment with anti-CD3, anti-CD20 or abatacept have shown clear promise and may be employed in the future to preserve beta-cell function. Trials are also underway investigating strategies that prevent T1D in high-risk populations characterised by detectable islet autoantibodies but with normal glucose tolerance. The intranasal insulin trial, led by an Australian team but including additional New Zealand and German sites, for instance, is examining whether a 12-month course of nasal insulin vaccine is an effective prevention strategy (https://clinicaltrials.gov/ct2/show/NCT00336674). An even more experimental strategy would be to prevent people with an elevated genetic risk from developing autoimmunity, a process that usually commences in utero and during the first few years of life. Central to this effort will be identifying the environmental triggers of autoimmunity, a question being investigated by the Environmental Determinants of Islet Autoimmunity Study.30

T1D cure with pancreas or islet transplantation

Whole pancreas transplant, usually performed simultaneously with a kidney transplant in people with T1D and renal failure, is a well-established therapy at the Westmead National Pancreas Transplant Unit in Sydney and the Monash Medical Centre in Melbourne. This procedure normalises blood glucose levels without requiring exogenous insulin, and also partially repairs established nephropathy, retinopathy, vascular disease and hypoglycaemic unawareness. Pancreatic graft survival is estimated to be greater than 60% at 5 years, but the benefits must be weighed against the morbidity associated with the surgery and associated immunosuppression.31

The concept of transplanting the islets of Langerhans to cure T1D was proposed more than 30 years ago, and remains an attractive but logistically difficult long-term solution. In Australia, islet isolation is undertaken at Westmead Hospital in Sydney and at St Vincent’s Hospital in Melbourne, and transplants are performed at these centres and at the Royal Adelaide Hospital. Islet transplants are offered to those who have life-threatening hypoglycaemia. The results have been impressive, with reductions in both insulin requirement and hypoglycaemia; 44% of recipients still did not require insulin 3 years after transplant (usually an endovascular procedure, not a surgical procedure). These benefits come, again, at the expense of lifelong immunosuppression and its attendant risks.32 Limited tissue supply is a key barrier to more widespread use of islet transplantation. Advances in stem-cell technologies or in the production of porcine islets for human transplant may overcome this problem in the future.

Prognosis in T1D

Despite recent advances, T1D is still associated with considerable premature mortality caused by acute and chronic complications, particularly ischaemic heart disease.33 The presence and severity of chronic kidney disease and ischaemic heart disease predict all-cause mortality in T1D.34,35 Recent reports of improved life expectancy in the US36 and in Denmark37 nonetheless provide great hope for persons with T1D and their clinicians.

Box –
Bolus insulin self-titration calculation

Carbohydrate ratio (CHR): grams of carbohydrate covered by 1 unit of insulin.
Insulin sensitivity factor (ISF): drop in blood glucose level (in mmol/L) achieved by each unit of insulin.

Sample calculation:

Tim has T1D. His CHR is 10 : 1 and his ISF is 2 : 1.
He is about to have a lunch that includes 50 g of carbohydrate.
His current blood glucose level (BGL) is 14 mmol/L. His target BGL is 6 mmol/L.
Prandial insulin dose = carbohydrate load ÷ CHR or 50 g ÷ 10 = 5 units of insulin to cover the carbohydrate in the meal.
Correctional insulin dose = variance from target glucose ÷ ISF or (14 − 6) ÷ 2 = 4 units of insulin to correct BGL from 14 mmol/L to 6 mmol/L.
He should therefore take 9 units of short-acting insulin with his meal.

Hypoadrenalism secondary to topical corticosteroid-containing skin-lightening cream: danger of over-the-counter cosmetic agents

A 26-year-old Sudanese woman was referred to the endocrinology clinic for investigation of low serum cortisol detected during investigations for fertility difficulties. There were no symptoms or signs to suggest adrenal insufficiency, nor did she have any overt Cushingoid features to suggest exogenous glucocorticoid exposure. Her blood pressure was 120/84 mmHg without postural drop. Although she had a generalised dark complexion, her face was a lighter shade compared with the rest of her body.

On further questioning, she admitted to using two “skin-lightening” creams for many years. These contained fluocinonide 0.075% and hydrocortisone acetate 1% and were purchased over the counter from a suburban store selling African goods in Sydney.

Repeat pathology (Box) confirmed the low cortisol on several separate mornings — 62 nmol/L, 116 nmol/L, < 28 nmol/L (reference interval [RI], 200–600 nmol/L), with a low-normal adrenocorticotropic hormone (ACTH) level of 3.1 pmol/L (RI, <10 pmol/L) and low 24-hour urine cortisol of 33 nmol (RI, 50–250 nmol/24 hours). These were consistent with the corticosteroid-containing creams causing suppression of her hypothalamic–pituitary–adrenal (HPA) axis. Cream usage was carefully weaned over a few weeks, with no symptoms of adrenal insufficiency. ACTH stimulation test performed a month later showed recovery of adrenal gland cortisol reserve.

Systemic absorption of the corticosteroid-containing creams resulted in suppression of the HPA axis in our patient. Studies have found a linear relationship between use of topical clobetasol propionate, which is used in skin-lightening creams, and the development of HPA axis suppression, occurring as early as a few days after commencement.1

Despite use of exogenous corticosteroids, the patient’s serum and urine cortisol levels were low. Routine cortisol immunoassays are most specific for cortisol (hydrocortisone), with significant cross-reactivity to prednisolone, methylprednisolone and prednisone. There is less cross-reactivity with other steroids, including cortisone and dexamethasone.2 Reduced cross-reactivity of the other synthetic corticosteroids in the creams used by this patient is the most likely reason for the low cortisol levels measured. This underscores the need for clinicians to be aware of possible falsely low cortisol results due to limitations of currently available immunoassays for measuring synthetic steroids.

The use of skin-lightening creams in women with dark skin tone is a common practice in some countries, with 67% prevalence in parts of Africa.3 While the prevalence is not known in Australia, it is possible that it is becoming more common with increasing migrant populations.4 Use of these corticosteroid-containing creams can cause unrecognised glucocorticoid excess syndromes and secondary adrenal insufficiency.1 Symptoms of hypoadrenalism can occur after cream cessation.1 Increased awareness of the potentially harmful consequences of these seemingly harmless “cosmetic agents” in our increasingly multicultural population will help minimise complications.

Box –
Patient’s pathology results

Test

During cream use

After stopping cream use

Reference interval

Serum sodium

140 mmol/L

135–145 mmol/L

Serum potassium

4.2 mmol/L

3.5–5.0 mmol/L

Serum cortisol (repeated samples on separate days)

62 nmol/L (7.30 am)116 nmol/L (8.40 am)< 28 nmol/L (8.50 am)

200–600 nmol/L

Serum ACTH

3.1 pmol/L

< 10 pmol/L

24-hour urine free cortisol

33 nmol/24 hours

50–250 nmol/24hours

Serum cortisol after 250 μ synacthen (short synacthen test)

153 nmol/L (0 min) (9.00 am)448 nmol/L (30 min)621 nmol/L (60 min)

> 550 nmol/L after synacthen stimulation

ACTH = adrenocorticotropic hormone.