Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Diabetes management

Systematic review of diabetes programmes [1]
Large dataset evaluation [2]
Real world experience [3]

Bandolier 134 looked at glucose self-monitoring in type 2 diabetes. But glucose self-monitoring is only one intervention among many to help people with type 2 diabetes achieve better control and better quality of life. Several recent studies on management make useful additional reading.

Systematic review of diabetes programmes [1]

This review sought English-language studies in a number of electronic databases using a systematic approach to care and including more than one intervention. Guidelines, protocols, algorithms, care plans, or systematic education or provider education programmes were potentially included. Trials should have used an experimental or quasi-experimental study design using Cochrane criteria, mainly randomised trials or before-after studies. Various outcomes were used, like Hb A1c, or screening visits for retinopathy, or patient satisfaction. Only adult patients were included in the review.


Twenty-four studies were included, 19 randomised trials and five nonrandomised controlled studies. There were 6,400 patients in total, so many trials were small.

The results are summarised in Table 1. Other than glucose self monitoring and glycaemic control, programmes could show little benefit. The most likely reason was that studies were small and inadequate.

Table 1: Results of studies of diabetes programmes

Glycaemic control
7 of 20 comparisons had significantly lower HbA1c with program, and overall the mean HbA1c level with program was 0.5% lower than the control intervention
Frequency of glycaemic monitoring
Some evidence that programs could help increase glucose self monitoring
Retinopathy screening
Small benefit
Nephropathy screening
Small benefit
Foot screening
Small benefit
Foot self-care
Mixed results
Systolic blood pressure
No overall benefit
No overall benefit

Large dataset evaluation [2]

Systematic reviews of inadequate data can be less than fulfilling, especially where organisational features are concerned. Indeed, systematic reviews of disparate interventions may be exactly the wrong way to seek knowledge. It is better to have a look at a single large organisation, and ask whether there are consistent features in those parts of the organisation that deliver best care.

There is one organisation that can do this better than any other, and that is the US Veterans' Affairs. It can do this because it looks after 3.7 million people, has centralised record keeping and analysis, but decentralised management and performance. It has the ability to keep up-to-date information on primary care practices where diabetics were treated, have key information on the diabetics, their medicine, laboratory tests, and general history, and their utilisation of healthcare services.

Using these datasets they sought patients with at least two outpatient visits with an associated diagnosis of diabetes, with a filled prescription for insulin, oral hypoglycaemic drug, or blood glucose monitoring supplies, and at least one visit to a VA primary care clinic. The outcome was Hb A1c level, and organisational variables were used to examine which features were associated with lower or higher Hb A1c at the 10% level or better (p≤0.1).


There were 82,000 diabetics in 177 clinics. The average age was 66 years, with 36% taking insulin and 77% oral hypoglycaemic drugs. The average Hb A1c level was 7.6%.

Many variables were associated with lower or higher Hb A1c (Table 2).

Table 2: Organisational characteristics impacting on diabetic control

(HbA1c %)
Features producing better HbA1c results
Greater staffing authority
All physicians involved in quality improvement
Being a large academic practice
Greater authority to establish or implement clinical policies
Notifying patients of assigned provider
Hiring needed new staff
Computerised reminders
Having nurses specific for the programme
Greater use of gatekeeping
Special teams or protocols for clinical issues
Weekly multidisciplinary team meetings
Features producing worse HbA1c results
Quality improvement programmes do not involve all physicians
General internal medicine physicians only
Located in acute care hospital
Patients almost always see assigned provider

Not all of these were obviously related to direct clinical care. The organisational characteristics associated with better diabetic control were these:

None of this is rocket science, and any good professional manager would be expected to jot most of these requirements down on the back of an envelope in a few minutes as core values for the success of any venture. It's just that they are not often used in healthcare.

Real world experience [3]

What happens to type 2 diabetics with inadequate glycaemic control has been evaluated in a real world study from northern California. Kaiser Permanente looks after about three million people. It has excellent electronic database information.

All type 2 diabetics with Hb A1c levels above 8%, with at least one year of membership of the scheme, without renal disease, and initiating a new therapy formed the setting for the study. Prescriptions for the dozen most commonly prescribed monotherapies and combination therapies were examined. There had to be at least one refill, and no evidence of the use of that therapy in the previous year.

The outcome was the Hb A1c values at first occurrence of the end of the study (12 months after the start of therapy), discontinuation of therapy, or modification of therapy. Good control was a level of Hb A1c of 7% or below. Results were adjusted for different case mixes between therapies.


The cohort consisted of 4,775 poorly controlled new users of therapy. Their mean age was 60 years, half were women, and the average starting level of Hb A1c was 9.9%. At this time 30% of the general Kaiser diabetic population had a Hb A1c level of 7% or below.

The most common therapy before starting a new therapy was sulphonylurea monotherapy. The most commonly initiated therapy was sulphonylurea plus metformin. Most diabetics (91%) were started on combinations of two or more oral agents.

Overall, 18% of diabetics starting a new therapy achieved good levels of Hb A1c of 7% or below in the three months after the change. The overall average fell from 9.9% to 8.6%. Several therapies and combinations of therapies achieved significantly higher numbers of patients with good control (Table 3).

Table 3: Therapies for type 2 diabetes associated with good diabetic control

Percent with HbA1c below 7%
Sulphonylureas + metformin + thiazolidinediones
Metformin + insulin
Sulphonylureas + thiazolidinediones

Several behavioural factors were also associated with better control. These included attending more than 70% of outpatient appointments, where 17% achieved good control compared with 11% in those who did not attend at least 70%. The frequency of self-monitoring of blood glucose was also associated with more people achieving good control.


What is really interesting about these three papers is that they make us look at how we amass and assess information on delivering a complex package of care. There are issues about particular interventions, and about how those interventions are delivered.

Despite having 24 randomised trials of various interventions in just 6,000 patients, the meta-analysis failed to provide much in the way of clear direction. By contrast, the VA study of 80,000 patients and 177 clinics was able to give us real direction about the components of packages of care that made a difference. Most had nothing to do with clinical interventions, and were about organisational issues. Targets were not mentioned. Clinics with all of the good features and few or none of the bad ones, obtained average reductions of 2-2.5% in Hb A1c levels more than clinics not having these characteristics. The UKPDS indicated that a 1% reduction in Hb A1c levels leads to a 21% reduction in the risk of diabetes related complications and death, so the implications for patients of these and other clinics is very considerable. The Kaiser study, again made possible because of superb data collection, helps in looking both at particular interventions and behavioural factors when dealing with a difficult group of type 2 diabetics with poor control. There is almost enough information to write the book on how to run an effective diabetic service.


  1. K Knight. A systematic review of diabetes disease management programs. American Journal of Managed Care 2005 11: 242-250.
  2. GL Jackson et al. Veterans affairs primary care organizational characteristics associated with better diabetic control. American Journal of Managed Care 2005 11: 225-237.
  3. AJ Karter et al. Achieving good glycaemic control: initiation of new antihyperglycaemic therapies in patients with type 2 diabetes from the Kaiser Permanente northern California diabetes registry. American Journal of Managed Care 2005 11: 262-270.

previous or next story