Course of glycaemia in poorly controlled type 2 diabetes patients 2.5 years after optimizing oral treatment in general practice

Introduction

In general practice, glycaemic control in poorly controlled type 2 diabetes patients (DM2) can be improved without insulin therapy by optimizing oral treatment [1], [2]. It is not well known, however, how long normoglycaemia can be maintained without exogenous insulin [3]. It is also uncertain whether tight glycaemic control can be maintained in the long run with insulin therapy [4–6].

The aim of this prospective study was to investigate whether poorly controlled DM2 patients in general practice, both those who were optimized with only oral hypoglycaemic agents (OHA) treatment and those who were switched immediately to insulin therapy, had acceptable glycaemic control 2.5 y after optimization.

Methods

A protocol for optimizing oral treatment in poorly controlled (HbA_1c ≥8%) patients was introduced in 18 general practices [1]. It included lifestyle advice and, if necessary, insulin therapy. Both the OHA scheme (first step: sulphonylurea [if BMI > 27 kg/m², then metformin]; second step: metformin-sulphonylurea combination [in case of adverse effects or contraindication, acarbose; no triple therapy]) and indications for insulin therapy (mixed insulin 30/70 twice daily; if BMI ≥ 30 kg/m² with metformin) followed the guidelines on type 2 diabetes of the Dutch College of General Practitioners [7].

Fifty-nine patients who maintained acceptable control (HbA_1c<8%) after oral treatment was optimized (OHA group) originally continued with OHA therapy. Thirty patients with poor control despite maximal feasible oral medication were immediately started on insulin (insulin group).

HbA_1c, height, body weight, diabetes duration and use of OHA were measured after OHA optimization (T₀). Identical variables were measured 2.5 y later (T_2.5). Means at T₀ and T_2.5 were compared using the paired-samples t-test with 95% confidence intervals (95% CI). Since the natural course of a cohort study is known to lose patients at follow-up [8], we performed sensitivity analysis by completing missing HbA_1c values with the best- and worst-known percentages in order to create a best- and worst-case outcome.

Results

Ten patients in the OHA group (n = 59; male 37%; mean age 64.9 y; mean diabetes duration 5.7 y) were lost to follow-up: six died, two moved to another practice and two for unknown reasons. At T_2.5, 59% (n=35) still had HbA_1c<8%, although mean HbA_1c had increased from 6.9% to 7.8% (Table I). Sensitivity analysis showed in the best-case scenario an increase to 7.4% (p<0.05) and in the worst-case to 8.6% (p<0.001) (data not shown). OHA treatment remained unchanged in 18 patients and had been adjusted in 21, while 10 patients finally switched to insulin. This resulted in HbA_1c<8% in 11 (23%), 16 (33%) and eight (16%) patients, respectively (Table II). Twenty patients had maximal oral medication, 10 of whom had HbA_1c ≥8%. Additional analysis with χ² tests showed no significant correlation between adjustment of treatment and glycaemic control at follow-up.

Table I.  Course of glycaemia and body weight in patients with type 2 diabetes who, after optimization, either continued oral treatment originally (OHA group) or switched to insulin therapy (insulin group). Means (SD) and mean difference (95% CI) at T₀ and T_2.5 (after 2.5 y).

	OHA group			Insulin group
	T0 (SD)	T2.5 (SD)	Difference T2.5 - T0 (95% CI)	T0 (SD)	T2.5 (SD)	Difference T2.5 - T0 (95% CI)
HbA1c (%)	6.9 (0.6)	7.8 (1.3)	0.9 (0.6, 1.3)^b	9.8 (1.7)	8.3 (1.6)	−1.5 (−0.4, −2.6)^a
Body weight (kg)	79.0 (14.9)	79.1 (14.0)	0.1 (-1.3, 1.4)	85.0 (14.5)	92.5 (17.0)	7.5 (5.0, 10.0)^b
BMI (kg/m^2)	27.9 (4.3)	28.1 (4.1)	0.2 (-0.4, 0.7)	30.5 (6.3)	33.2 (7.5)	2.7 (1.8, 3.6)^b

^ap<0.05, ^bp<0.001.

Table II.  Results of adjustment of oral hypoglycaemic agents ^a during a 2.5-y follow-up period.

		Result
Intervention	Number of patients n(%)	Acceptable (HbA1c<8%) n(%)	Poor (HbA1c ≥ 8%) n(%)
Medication unchanged	18 (37)	11 (23)	7 (14)
Dose of OHA raised, new agent added or combination of adjustments	21 (43)	16 (33)	5 (10)
Switched to insulin therapy	10^b (20)	8 (16)	2 (4)
Total	49 (100)	35 (72)	14 (28)

^aSulphonylurea, metformine and/or acarbose.

^bFour patients continued using one oral hypoglycaemic agent after switching to insulin.

Nine patients in the insulin group (n=30; male 33%; mean age 59.5 y; mean diabetes duration 7.6 y) were lost to follow-up: four refused insulin, three were referred to an internist, one died and one moved to another practice. At T_2.5, 33% (n=10) had HbA_1c<8%. Mean HbA_1c had decreased from 9.8% to 8.3% (Table I). Sensitivity analysis showed in the best-case scenario a decrease to 7.4% (p<0.001) and in the worst-case to 9.3% (p=0.32) (data not shown). Mean body weight increased 7.5 kg (p<0.001).

Exemplar patient OHA group Mr H, born 13 May 1932, diabetes duration 7 y, BMI 29 kg/m², had HbA_1c 8.3%. After optimization of oral treatment (weight reduction of 2 kg and increasing dosage of metformin), HbA_1c of 7.0% was achieved. At this point, he was included in the OHA group of our study. After 2.5 y, HbA_1c was 7.9% and body weight increased by 1 kg. In this period, gliclazide (sulphonylurea group) was started, but the oral medication was not maximal at the end . Exemplar patient insulin group Mrs A, born 11 June 1927, diabetes duration 15 y, BMI 31 kg/m², had poor glycaemic control (HbA_1c 9.9%) despite maximal oral medication. However, HbA_1c remained at 9.7% after education and instruction in self-regulation by the diabetes nurse. After this attempt at optimization, she was included in the insulin group. Mixed insulin 30/70 twice daily was started with continuation of metformin only. After 1 y, an HbA_1c percentage of 7.9 was achieved with a 5.1 kg increase in body weight; after 2.5 y, HbA_1c was 8.2% and body weight increased 7.2 kg in total.

Discussion

This study showed that 59% of DM2 patients who continued originally with only OHA because of an acceptable HbA_1c<8.0% after optimization of oral treatment because of poor glycaemic control still had an acceptable HbA_1c after 2.5 y. The mean HbA_1c had increased significantly, but remained < 8%. Insulin therapy was not started in 10 poorly controlled patients with maximal OHA, which may correspond with the finding that some GPs are reluctant to start insulin therapy [9]. Adjustment of treatment was not significantly correlated with acceptable glycaemic control, possibly since the study population was relatively small and since 11 out of 49 patients’ acceptable glycaemia could be maintained even without adjustment of treatment.

Of the patients receiving insulin therapy immediately after failure of optimization, 33% had acceptable glycaemic control. Moreover, mean HbA_1c had decreased by 1.5%, although still > 8%. Body weight had increased by a mean of 7.5 kg, clearly more than in the UK Prospective Diabetes Study (UKPDS). This difference may have been caused by selection bias, because UKPDS patients were newly diagnosed with a lower baseline HbA_1c [10].

In conclusion, 2.5 y after oral treatment was optimized in poorly controlled DM2 patients, six out of 10 of those who originally used only OHA still had acceptable glycaemic control. In three out of 10 patients who originally started on insulin immediately, acceptable glycaemic control was maintained, although with considerable weight gain. GPs seemed to be reluctant to switch to insulin. In both groups there may be room for improvement of glycaemic control. Further studies on optimization of poorly controlled diabetes patients are warranted to specify subgroups who should continue with oral therapy and subgroups who should be started on insulin straight away.

This study was supported financially by the Agis Health Care Assurance Group. We thank Th. van den Bos, M. C. V. M. Luykx, general practitioners; B. W. van den Berg and R. P. Verhoeven, internists; the diabetes nurses of the Home Care Oost-Veluwe; and all participating GPs.