Wiener klinische Wochenschrift

, Volume 124, Issue 17–18, pp 639–646 | Cite as

Effectiveness of the Austrian disease management programme “Therapie Aktiv” for type 2 diabetes regarding the improvement of metabolic control, risk profile and guideline adherence: 2 years of follow up

  • Maria Flamm
  • Sigrid Panisch
  • Henrike Winkler
  • Tim Johansson
  • Raimund Weitgasser
  • Andreas C. Sönnichsen
original article

Summary

Background

We evaluated the disease management programme (DMP) “Therapie Aktiv” for diabetes mellitus type 2 during the first year of implementation in a cluster-randomised controlled trial (RCT), which revealed an improvement of process quality, but only insignificant effects on HbA1c. To further analyse the effects of the DMP we followed up participants of the RCT for another year.

Methods

After completion of the RCT, the study was continued as an open observational trial. By patient’s choice, three groups were formed. Group 1 (DMP/DMP): interventions of the RCT (DMP), who remained in the DMP; group 2 (Control/DMP): controls of the RCT (usual care), who participated in the DMP during the second year; group 3 (Control/Control): controls of the RCT (usual care), who remained as controls. Primary outcome measure: HbA1c. Secondary outcome measures: BMI, lipids, blood pressure and measures of process quality.

Results

Only 801 of 1,489 RCT participants completed the study (53.8 %). Group 1:n = 355; group 2:n = 335; group 3:n = 111. After 2 years, pre-post-analysis revealed a significant reduction of HbA1c (0.37 %, p  < 0.001) in both DMP groups, and a reduction of only 0.03 % (p = 0.975) in the control group. However, the difference between the groups was not significant (p = 0.065). There was a significant improvement of process quality in the DMP groups compared with controls.

Conclusion

The DMP clearly enhances process quality. Furthermore, the DMP yields a reduction of HbA1c within groups after 2 years, but significance is lost in between-group analysis. We conclude that the DMP has only a minor effect on metabolic control.

Keywords

Diabetes mellitus type 2 Disease management HbA1c Metabolic control Effectiveness 

Effektivität des österreichischen Disease Management Programms „Therapie Aktiv“ für Diabetes mellitus Typ 2 hinsichtlich metabolischer Kontrolle, Risikoprofil und Leitlinienadhärenz: 2 Jahre Follow up

Zusammenfassung

Grundlagen

Im ersten Jahr der Implementierung haben wir das Disease Management Programm (DMP) „Therapie Aktiv“ für Diabetes mellitus Typ 2 im Rahmen einer cluster-randomisierten kontrollierten Studie (RCT) evaluiert. Es zeigte sich eine deutliche Verbesserung der Prozessqualität, jedoch keine signifikanten Effekte hinsichtlich HbA1c Reduktion. Um die Effektivität des DMPs über einen längeren Zeitraum zu untersuchen, beobachteten wir die RCT-Teilnehmer ein weiteres Jahr.

Methodik

Nach Beendigung des RCT wurde die Studie als offene Beobachtungsstudie weitergeführt. Basierend auf Patientenwahl ergaben sich 3 Gruppen: Gruppe 1 (DMP/DMP): Interventionsteilnehmer des RCT (DMP), die im DMP blieben; Gruppe 2 (Kontrolle/DMP): Kontrollen des RCT (Usual Care), die im zweiten Jahr ins DMP eintraten; Gruppe 3 (Kontrolle/ Kontrolle): Kontrollen des RCT (Usual Care), die als Kontrollen in der Studie blieben. Primäres Zielkriterium: HbA1c. Sekundäre Zielkriterien: BMI, Lipide, Blutdruck und Prozessqualitätsparameter.

Ergebnisse

Daten von lediglich 801 der 1.489 RCT-Teilnehmer (53,8%) konnten nach 2 Jahren analysiert werden. Gruppe 1:n = 355; Gruppe 2:n = 335; Gruppe 3:n = 111. Die Vorher-Nachher-Analyse ergab eine signifikante Reduktion des HbA1c (0,37 %, p < 0,001) in beiden DMP Gruppen, und lediglich eine Reduktion von 0,03 % (p = 0,975) in der Kontrollgruppe. Jedoch waren die Unterschiede zwischen den Gruppen nicht signifikant (p = 0,065). In beiden DMP Gruppen fanden sich, verglichen mit der Kontrollgruppe, signifikante Verbesserungen der Prozessqualität.

Schlussfolgerung

Die Prozessqualität wird durch das DMP deutlich verbessert. Weiters führt das DMP zu einer signifikanten HbA1c Reduktion innerhalb der Gruppe, jedoch geht die Signifikanz im Vergleich zwischen den Gruppen verloren. Wir schließen daraus, dass das DMP lediglich geringe Effekte auf die metabolische Kontrolle hat.

Schlüsselwörter

Diabetes mellitus Typ 2 Disease Management HbA1c metabolische Kontrolle Effektivität 

Introduction

Type 2 diabetes presents a growing challenge for health care systems as the prevalence in all age groups is rising worldwide [1, 2]. About 4.2–4.6 % of the adult population is affected in Austria [3] and the annual incidence of new treatment-dependent type 2 diabetes is estimated to be within the range of 3.8–4.4 per 1,000 person per year [4]. The prevalence of late diabetic complications in Austria corresponds to the European average as demonstrated in the CODE-2 study [5]. Despite large efforts to improve diabetes prevention and treatment, there are still substantial deficits in the implementation of standard care [6, 7]. Hence, there is a strong demand for management optimisation. As a possible strategy to improve diabetes care in Austria, the disease management programme (DMP) for diabetes mellitus type 2 “Therapie Aktiv” has been designed and implemented by the main stakeholders in the field and by statutory public health insurance. So far the evidence regarding the effectiveness of such structured approaches remains inconclusive as has been shown by several systematic reviews [8, 9, 10, 11]. We therefore evaluated the Austrian DMP regarding its effectiveness on metabolic control parallel with the programme implementation in the state of Salzburg in 2007 within a cluster-randomised controlled trial (RCT). The observation period was 1 year. The main results of the RCT have been published recently [12]. In summary, we concluded that the DMP significantly improves process quality, but does not remarkably enhance metabolic control within the first year. HbA1c reduction was 0.41 % in the intervention group, but the DMP-associated net-effect after subtracting the improvement of the control group was only 0.13 %. This effect was significant, however only in unadjusted univariate analysis.

Thus, the achievements of the Austrian DMP “Therapie Aktiv” regarding metabolic control during the RCT fell somewhat short of expectations. However, 1 year of observation might have been too short to measure significant differences between the groups. After completion of the RCT, we therefore continued the study for another year as an open observational trial to assess the prolonged impact of the DMP regarding HbA1c reduction and process quality.

Methods/design

Setting and study population

In 2007, participation in the RCT was offered to all 275 general physicians (GPs) and internists in the province of Salzburg having a contract with public health insurance. Salzburg consists of Salzburg city with about 150,000 inhabitants and five districts (suburban and rural areas) with a total of about 350,000 inhabitants. The prevalence of type 2 diabetes in this population is estimated to be about 2.5–3 % [3].

Primarily 92 physicians (33.5 % of those eligible) signed up and consecutively recruited patients (older than 18 years) with diabetes mellitus type 2 who fulfilled the WHO/ADA-criteria for diabetes diagnosis and were willing to enrol in the DMP “Therapie Aktiv.” The recruitment period of the RCT (first year of observation) extended from July to November 2007. We obtained informed consent according to the declaration of Helsinki from all patients. The RCT has been approved by the ethics committee of Salzburg and has been registered with Current Controlled Trials Ltd. (ISRCTN27414162). Details of the study protocol [13] and results of the RCT were published elsewhere [12, 14].

After 1 year, the study was continued as an open observational trial. Once again, physicians and patients were asked to give informed consent. All physicians and patients who had participated in the RCT were invited to enrol in the follow-up study. The patients of the control group were offered to enrol in the DMP, thus yielding a total of three groups under observation over a period of 2 years. Group 1 (DMP/DMP): intervention patients of the RCT (DMP), who continued to participate in the DMP; group 2 (Control/DMP): control patients of the RCT (usual care), who participated in the DMP in the second year; group 3 (Control/control): control patients of the RCT (usual care), who deliberately did not want to participate in the DMP but were willing to remain in the study as controls.

Intervention

The DMP “Therapie Aktiv” intervention included the following components (details were published elsewhere [12]):
  • A mandatory 10-hour training course for physicians, designed by the Austrian Diabetes Association (ÖDG), the Austrian Medical College (Ärztekammer), and the Austrian Society for General Practice (ÖGAM). All physicians of the intervention group in the RCT had participated in the course at the beginning of the RCT. All physicians of the control group in the RCT wishing to participate in the follow-up study were obliged to take the course at the beginning of the follow-up study.

  • Nine hours of patient education in four modules. Patient education was organised by the Working Group for Preventive Medicine Salzburg (AVOS).

  • Standardized documentation of physical examination, laboratory findings, and diabetes complications in a DMP-CRF once a year.

  • Structured interdisciplinary care according to the guidelines of the ÖDG [15]. Agreement on therapeutic goals in a shared patient–physician decision-making process at quarterly intervals.

Control patients were treated according to usual care.

Data acquisition

At baseline, after 1 and after 2 years the following parameters were examined and recorded: laboratory: HbA1c (primary outcome measure), total cholesterol, triglycerides, HDL- and LDL-cholesterol, creatinine; anthropometric measurements: height and body weight, BMI, systolic and diastolic blood pressure; process quality measures: eye and foot examinations, patient education and regular HbA1c checks.

All data were recorded in the surgeries of the participating physicians and then transferred to the Institute of General Practice, Family Medicine and Preventive Medicine of Paracelsus Medical University Salzburg (PMU) for further processing and evaluation.

Statistical methods

All statistical analyses were performed with IBM® SPSS® Statistics19.0. The analysis was carried out including the data of all patients with at least two measured HbA1c values, and at least 600 days difference between the first and the last available HbA1c value. We evaluated differences within groups in a pre- post-analysis using a Wilcoxon test, and differences between groups at baseline as well as after 2 years using ANOVA (F test). The Chi2 test was used to check for significant differences regarding parameters of process quality (i.e. percentage of patients with guideline adherent care).

Results

Only 68 of the original 92 physicians participating in the RCT were willing to continue the DMP in their surgery (73.9 %), and only 1,072 of the 1,489 patients participating in the RCT remained in the follow-up study (71.99 %). By patient’s choice, 414 patients were allocated to group 1 (DMP/DMP), 440 to group 2 (Control/DMP) and 218 to group 3 (Control/control). Only 801 patients recruited by 55 of the 68 physicians entering the follow-up study completed the study, had at least two distinct HbA1c-measurements at least 600 days apart (74.7 %), and could thus be analysed [group 1:n = 355 (53.0 % male); group 2:n = 335 (51.3 % male); group 3:n = 111 (56.8 % male)]. The flow of physicians and patients through the study is shown in Fig. 1.
Fig. 1

Flow of physicians and patients. The sum of physicians (asterisk) recruiting patients in groups 1, 2 and 3 is > 68, because some physicians do treat patients in more than one group

Baseline data are illustrated in Table 1. None of the baseline data including primary and secondary outcome measures showed significant differences between groups except for age (p = 0.016) which was significantly higher in patients of group 3 (+ 3.27 years, p = 0.012; post-hoc test: Bonferroni) compared with patients of group 2. However, there was no significant difference between group 1 and the other groups (data not shown).
Table 1

Baseline

 

Group 1

Group 2

Group 3

p-valuea

 

N; (mean ± SD)

N; (mean ± SD)

N; (mean ± SD)

Male (%)

355; 53.0

335; 51.3

111; 56.8

0.611b

Age (years)

355; 65.13 ± 10.20

335; 64.26 ± 10.61

111; 67.53 ± 10.24

0.016

Primary outcome measure

HbA1c (%)

355; 7.40 ± 1.48

335; 7.32 ± 1.37

111; 7.14 ± 1.11

0.226

Secondary outcome measures

Creatinine (mg/dL)

351; 0.93 ± 0.25

330; 0.92 ± 0.28

108; 0.96 ± 0.28

0.465

Triglycerides (mg/dL)

353; 185.90 ± 182.14

332; 170.79 ± 130.22

109; 182.10 ± 174.47

0.462

Cholesterol (mg/dL)

353; 197.33 ± 45.55

332; 193.08 ± 38.50

109; 196.77 ± 39.69

0.391

HDL (mg/dL)

352; 52.37 ± 16.19

332; 51.70 ± 14.04

109; 51.76 ± 12.22

0.831

LDL (mg/dL)

330; 108.80 ± 37.24

322; 109.70 ± 32.61

103; 111.25 ± 33.04

0.817

Systolic blood pressure (mmHg)

351; 140.19 ± 19.31

332; 138.42 ± 16.90

109; 139.23 ± 15.14

0.446

Diastolic blood pressure (mmHg)

351; 82.15 ± 11.46

332; 82.14 ± 10.17

109; 82.97 ± 9.99

0.756

BMIc (kg/m2)

351; 30.61 ± 4.91

332; 30.08 ± 4.87

108; 29.50 ± 4.91

0.093

Per protocol analysis (PPA): Two measured values, at least 600 days difference between the first and the last value

ap-value: ANOVA (F test); Welch test

bChi2 test

bMean of height-data was used to calculate BMI

Within group 1 we found significant improvements after 2 years regarding the following outcome measures: HbA1c (− 0.37 %), triglycerides (− 28.44 mg/dL), cholesterol (− 12.21 mg/dL), HDL-cholesterol (3.79 mg/dL), LDL-cholesterol (− 10.01 mg/dL) and BMI (− 0.3 kg/m²). Significant improvements in group 2 were: HbA1c (− 0.37 %), cholesterol (− 8.92 mg/dL), HDL (1.96 mg/dL), LDL-cholesterol (− 9.28 mg/dL), systolic and diastolic blood pressure (− 2.41; − 1.52 mmHg) and BMI (0.18 kg/m²). In group 3, only cholesterol (− 13.34 mg/dL), LDL-cholesterol (− 10.11 mg/dL) and diastolic blood pressure (− 3.25 mmHg) improved significantly. Pre- and post-test changes within the groups over the period of 2 years are shown in Tables 2.
Table 2

Mean changes within the groups (pre-post-analysis) over 2 years

 

Group 1 (mean ± SD)

p-valuea

Group 2 (mean ± SD)

p-valuea

Group 3 (mean ± SD)

p-valuea

p-valuec

Primary outcome measure

HbA1c (%)

0.37 ± 1.28

< 0.001

0.37 ± 1.25

< 0.001

0.03 ± 1.15

0.975

0.065e

Secondary outcome measures

Creatinine (mg/dL)

− 0.01 ± 0.21

0.154

− 0.00 ± 0.30

0.248

− 0.04 ± 0.21

0.476

0.393e

Triglycerides (mg/dL)

28.44 ± 147.11

0.001

14.42 ± 108.81

0.125

16.45 ± 133.49

0.343

0.496e

Cholesterol (mg/dL)

12.21 ± 42.24

< 0.001

8.92 ± 42.94

< 0.001

13.34 ± 41.82

0.001b

0.834e

HDL (mg/dL)

− 3.79 ± 10.79

< 0.001

− 1.69 ± 9.10

0.001

− 0.3 6 ± 10.78

0.728b

0.001d

LDL (mg/dL)

10.01 ± 39.19

< 0.001

9.28 ± 37.68

< 0.001

10.11 ± 29.71

0.001b

0.849e

Systolic blood pressure (mmHg)

2.64 ± 19.88

0.112

2.41 ± 18.94

0.023

2.05 ± 17.36

0.244

0.740e

Diastolic blood pressure (mmHg)

1.17 ± 12.92

0.164

1.52 ± 11.29

0.043

3.25 ± 11.46

0.008

0.338e

BMIf (kg/m2)

0.3 ± 1.94

0.009

0.18 ± 2.2

0.036

− 0.18 ± 1.90

0.259

0.171e

Reduction is shown as a positive number

ap value: Wilcoxon test

bp value: Paired T test

cMean changes between groups within 2 years (adjusted for baseline); p value: F test; PPA two measured values, at least 600 days difference between the first and the last value

dHDL: Difference (group 1–2) = -  2.20 (p = 0.012); difference (group 1–3) = -  3.52 (p = 0.004)

eFurther post-hoc tests were not applicable as F test for changes between groups was not significant

fMean of height-data was used to calculate BMI

We found a total reduction of HbA1c over 2 years by 0.37 % points in both DMP groups (group 1 and 2) compared with a marginal change in the control group (HbA1c 0.02 %). The trend of our primary outcome measure HbA1c is shown in a 3-level graph (Fig. 2). In group 1 (baseline level: 7.4 %), we detected a small rise in the second year (+ 0.08 %) after an initial decrease of 0.45 %. In group 2 (baseline level: 7.32 %), the HbA1c level remained stable in the second year after an initial decrease of 0.37 % in the first year. HbA1c decreased in the first year in group 3 (baseline level: 7.14 %) by 0.12 % and increased in the second year by 0.1 %. The difference between the groups regarding the primary outcome measure was not significant after 2 years (Table 3).
Fig. 2

HbA1c levels (in %) at baseline, after 1 year and after 2 years

Table 3

Mean changes between the groups within 2 years. (Adjusted for baseline)

 

p-valuea

Group1–2

p-valueb

Group1–3

p-valueb

Group2–3

p-valueb

Primary outcome measure

HbA1c (%)

0.065

− 0.05

n.a.

0.20

n.a.

0.25

n.a.

Secondary outcome measures

Creatinine (mg/dL)

0.393

− 0.01

n.a.

0.03

n.a.

0.04

n.a.

Triglycerides (mg/dL)

0.496

4.09

n.a.

9.49

n.a.

5.40

n.a.

Cholesterol (mg/dL)

0.834

0.83

n.a.

− 1.46

n.a.

− 2.29

n.a.

HDL (mg/dL)

0.001

− 2.20

0.012

− 3.52

0.004

− 1.32

0.684

LDL (mg/dL)

0.849

1.27

n.a.

1.38

n.a.

0.11

n.a.

Systolic blood pressure (mmHg)

0.740

− 0.88

n.a.

− 0.01

n.a.

0.87

n.a.

Diastolic blood pressure (mmHg)

0.338

− 0.36

n.a.

− 1.47

n.a.

− 1.11

n.a.

BMIc (kg/m2)

0.171

0.09

n.a.

0.42

n.a.

0.33

n.a.

Per protocol analysis (PPA): Two measured values, at least 600 days difference between the first and the last value

n.a. not applicable

aF test

bPost-hoc tests

cMean of height-data was used to calculate BMI

We also analysed differences between the groups after 2 years for all secondary outcome measures (Table 3). No significant differences regarding any secondary outcome measures could be detected except for HDL-cholesterol after 2 years. HDL-cholesterol increased significantly more in group 1 compared with groups 2 and 3 (p  =  0.001; Table 3).

Results regarding process quality are depicted in Tables 4 and 5. We found a significantly higher proportion of patients participating in patient education in the DMP groups (group 1 > group 2 > group 3, p < 0.001; Table 4). Information regarding the participation in patient education was obtained by both the physicians and the patients separately. The participation rate was reported remarkably higher by the physicians than by the patients.
Table 4

Process quality regarding patient education after 2 years

Proportion of patients (in %)

Group 1

Group 2

Group 3

p-valuea

Patient educationb

63.9

41.2

32.4

< 0.001

Patient educationc

93.2

73.1

67.9

< 0.001

Years since last patient educationc

2.47

2.76

2.99

0.081d

aChi2 test

aPatient education is not planned to be carried out yearly; therefore, we differentiated between “at least once” and “never, or no information available” (according to the information provided by the patients)

bAccording to the information provided by the physicians

cKruskal–Wallis test

Table 5

Process quality regarding regular examinations over 2 years

Proportion of patients (in %)

First year

Second year

Group 1

Group 2

Group 3

p-valuea

Group 1

Group 2

Group 3

p-valuea

Eye examination

89.9

68.7

64.0

< 0.001

76.9

79.7

68.5

0.052

Foot examination

88.7

66.9

55.0

< 0.001

77.2

76.7

58.6

< 0.001

Regular HbA1c checks

91.8

89.3

78.4

< 0.001

80.6

88.7

82.9

0.013

We differentiated between “yes” and “no, or no information available” (according to information provided by the patients)

aChi2 test

Concerning the screening examinations to avoid diabetic complications, we detected a significantly higher rate of eye-and-foot examinations in group 1 in the first year (p < 0.001) with a distinct rise of the examination frequency in group 2 after patients entered the DMP. The frequency of regular HbA1c checks dropped in the second year of DMP participation (Table 5).

Discussion

In pre-post-analysis, our study shows that participation in the DMP leads to a significant reduction in the primary outcome measure HbA1c, while HbA1c remains unchanged in non-participants. The difference between the DMP groups and the controls regarding HbA1c-reduction was not significant due to a number of possible reasons. The large drop-out rate and consequently a fairly small study sample may have led to a lack of statistical power. Also, patients were already under fairly good metabolic control at baseline and thus only a minor decrease in HbA1c values could be achieved. Similarly, secondary outcome measures showed some significant reductions in pre-post-analysis, but no significant differences between groups. The stronger HDL-cholesterol increase in the DMP groups may rather be significant by chance due to multiple testing than reflect a true between-group difference.

Even though we detected a promising HbA1c reduction in both DMP groups after 2 years, this result has to be interpreted with caution. The analysis of HbA1c trend including all the three measurements (baseline, after 1 year, and after 2 years) clearly showed that the decrease in HbA1c largely takes place within the first year of observation in all groups. During the first year, the participation in the RCT and the so-called study effect and/or regression to the mean probably play an important role. Thus, the previously conducted evaluation of “Therapie Aktiv” in the RCT revealed an actual net effect regarding HbA1c of only − 0.13 % in the unadjusted per-protocol analysis of the first year (pre-and post-interventions group: − 0.41 %; control group: − 0.28 %; [12, 16]). This effect is not statistically significant in multivariate analysis, and—even more important—most likely not clinically relevant. Furthermore, the small improvements of HbA1c in the two DMP groups compared with no change in the control group are taken from a non-randomized study with all its draw backs and risks of bias (see the section “Limitations”). On the other hand, even HbA1c stabilization may be a relevant success in long-term diabetes care. Diabetes mellitus type 2 represents a chronic progressive disease with a strong tendency of increasing HbA1c level over the years. Long-term observations within the UK Prospective Diabetes Study (UKPDS) show a steady increase of HbA1c of about 0.1 % per year in diabetic patients under conventional treatment as well as intensive drug therapy [17]. Remarkably, the difference between the groups caused by an initial decrease within the intensive treatment group during the first year could be maintained throughout the whole observation period of 10 years [17]. This tendency of increasing HbA1c is due to the progressive character of diabetes and reflects the normal course of the disease, thus certainly counteracting the effect of the DMP. As the association between increasing glycosylated haemoglobin and cardiovascular disease in diabetic patients is well established [18], an initial HbA1c-decrease of about 0.4 % achieved by the DMP may very well be relevant for clinical outcome if it can be maintained over the years. It will be an important issue of further investigation in the continuing observational study in Salzburg, whether the Austrian DMP can stabilize HbA1c-value in the long run.

Both in the first and the second year, the process quality clearly improved due to the DMP: the proportion of regular foot and eye examinations in the DMP groups was remarkably higher in both years, albeit less distinct in the second year. Regarding regular HbA1c checks, process quality in the controls in the second year is largely comparable to the DMP patients. The proportion of patient participating in patient education is significantly higher in the DMP groups. The finding that frequency of patient education provided by the physicians differed massively from frequency reported by the patients needs to be discussed. Physicians declared almost one-third more participations in patient education than patients. This discrepancy might be due to the patients or the physicians recall bias. Another possible explanation might be that patients did not participate in some of the “medically instructed” and therefore “registered/documented” patient education sessions without knowledge of the advising physician.

Limitations

As not all patients that were primarily recruited for the RCT remained in the open observational study, the reduced number of participants has negatively affected the power of the study. Furthermore, selection bias exerts a relevant influence on the results. Selection bias probably occurred at several levels: first the DMP “Therapy Aktiv” is an optional programme. We only achieved an estimated participation rate of less than 10 % of all diabetic patients in Salzburg and we found evident deficits in care within the included population [6]. However, the DMP and the RCT may have led to the selection of a motivated sample, both regarding providers as well as patients. This putative positive selection might on the one hand imply less detectable room for improvement by an intervention such as the DMP. Hence, an obligatory programme including more patients with poor metabolic control and possibly larger deficits in care might lead to a more consistent improvement. On the other hand, a disease management programme might have less effect or not work at all in less motivated patients and physicians. This question is left to be investigated by further appropriate research. As already stated in our previous work [8], another risk for bias may be assumed due to the incomplete concealment of allocation which was assured only on the level of physicians due to methodical reasons of this pragmatic study.

Furthermore, the groups of the follow-up study presented here were not allocated by a randomisation process but rather by patient’s choice. Thus, more motivated patients may have stayed in or entered the DMP while the patients neglecting physician’s advice more likely entered the control groups. Another limitation of our study that needs to be discussed is that in the second year most GPs took care of both DMP patients and controls. All physicians received the DMP training either in the beginning of the first (RCT intervention) or in the beginning of the second year (RCT controls). Thus, all groups were at least indirectly exposed to an important part of the intervention which certainly led to contamination effects. On the other hand, the unchanged HbA1c in the control group demonstrates that these effects may have been small. Moreover, it is has to be mentioned that a substantial number of patients quit due to drop-out of the attending physician rather than self-motivated.

The large drop-out rate between the RCT and the follow-up study is in itself an important result of our trial. It demonstrates the difficulties of implementing a disease management in the pragmatic setting of primary care in Austria. A large proportion of the GPs who initially signed up for the programme dropped out after the RCT, thus making their patients drop out as well. The reasons for this reluctance of the physicians to take part in the DMP need to be investigated and incentives for participation must be given to both the physicians and the patients.

Conclusion

Process quality of diabetes care is clearly improved by the DMP while its effects on metabolic control and other secondary outcome measures are only minor. Further research is necessary to investigate whether the initial reduction in HbA1c which can be seen in pre-post-analysis can be maintained over the years. Only then a possible effect on diabetic complications and clinical outcome can be estimated. The low proportion of patients with diabetes mellitus reached by the programme, and the high dropout rates of physicians and patients raise serious doubts regarding the population effect of the DMP in its current form. It appears to be advisable to intensify the current programme and provide further incentives for participation to physicians as well as patients.

Acknowledgements

This study was financed by the Jubiläumsfonds of the Austrian National Bank and the Institute of General Practice, Family Medicine and Prevention of the Paracelsus Medical University, Salzburg. We received additional funding by the project “Developing and validating disease management evaluation methods for European health care systems” (DISMEVAL), funded by the European Commission’s Seventh Framework Programme (grant agreement 223277) which addresses this issue by pooling evidence on DMP evaluation of different European countries. We would like to thank our sponsors for these grants. The sponsors were not involved in the realisation of the study, data processing or evaluation. We would also like to thank all the participating physicians of the province of Salzburg for their participation in the study. Furthermore, we express our gratitude to the Medical College of Salzburg and the Salzburger Gebietskrankenkasse (SGKK, Salzburg statutory health insurance) for continuously supporting our work.

Author’s contributions

Maria Flamm contributed substantially in writing this paper. Andreas C. Sönnichsen developed the study idea and contributed substantially in writing the study protocol as well as supervising the writing of this paper. He is in charge of carrying out the study. Henrike Winkler and Raimund Weitgasser contributed to the development of the study protocol and to writing this paper. Sigrid Panisch contributed to the evaluation of the data and to writing this paper. Moreover, she oversaw the maintenance and quality assurance of the data. Tim Johansson contributed to the generation of the data and to writing this paper. Raimund Weitgasser contributed to the development of the DMP and patient as well as GP education and to the scientific evaluation as a specialist in type 2 diabetes.

Notes

Conflict of interest

The authors declare that there is no actual or potential conflict of interest in relation to this article.

References

  1. 1.
    King H, Aubert RE, Herman WH. Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections. Diabetes Care. 1998;21(9):1414–31.PubMedCrossRefGoogle Scholar
  2. 2.
    Wild S, Roglic G, Green A, Sicree R, King H. Global prevalence of diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care. 2004;27(5):1047–53.PubMedCrossRefGoogle Scholar
  3. 3.
    Dorner T, Rathmanner T, Lechleitner M, Schlogel R, Roden M, Lawrence K, et al. Public health aspects of diabetes mellitus—epidemiology, prevention strategies, policy implications: the first Austrian diabetes report. Wien Klin Wochenschr. 2006;118(17–18):513–9.PubMedCrossRefGoogle Scholar
  4. 4.
    Winkelmayer WC, Stedman MR, Pogantsch M, Wieninger P, Bucsics A, Asslaber M, et al. Guideline-conformity of initiation with oral hypoglycemic treatment for patients with newly therapy-dependent type 2 diabetes mellitus in Austria. Pharmacoepidemiol Drug Saf. 2011;20(1):57–65.PubMedCrossRefGoogle Scholar
  5. 5.
    Williams R, Van GL, Lucioni C. Assessing the impact of complications on the costs of type II diabetes. Diabetologia. 2002;45(7):S13–7.PubMedGoogle Scholar
  6. 6.
    Flamm M, Winkler H, Panisch S, Kowatsch P, Klima G, Furthauer B, et al. Quality of diabetes care in Austrian diabetic patients willing to participate in a. Wien Klin Wochenschr. 2011; 123(13–14):436–43.PubMedCrossRefGoogle Scholar
  7. 7.
    Rakovac I, Plank J, Jeitler K, Beck P, Seereiner S, Mrak P, et al. Health status of type 2 diabetics in Austria—perspective of a quality improvement initiative. Wien Med Wochenschr. 2009;159(5–6):126–33.PubMedCrossRefGoogle Scholar
  8. 8.
    Weingarten SR, Henning JM, Badamgarav E, Knight K, Hasselblad V, Gano A, et al. Interventions used in disease management programmes for patients with chronic illness—which ones work? Meta-analysis of published reports. BMJ. 2002 Oct 26;325(7370):925–28F.PubMedCrossRefGoogle Scholar
  9. 9.
    Ofman JJ, Badamgarav E, Henning JM, Knight K, Gano AD, Levan RK, et al. Does disease management improve clinical and economic outcomes in patients with chronic diseases? A systematic review. Am J Med. 2004 Aug 1;117(3):182–92.PubMedCrossRefGoogle Scholar
  10. 10.
    Knight K, Badamgarav E, Henning JM, Hasselblad V, Gano AD, Jr., Ofman JJ, et al. A systematic review of diabetes disease management programs. Am J Manag Care. 2005;11(4):242–50.PubMedGoogle Scholar
  11. 11.
    Norris SL, Nichols PJ, Caspersen CJ, Glasgow RE, Engelgau MM, Jack L, et al. The effectiveness of disease and case management for people with diabetes—a systematic review. Am J Prev Med. 2002;22(4):15–38.PubMedCrossRefGoogle Scholar
  12. 12.
    Sonnichsen AC, Winkler H, Flamm M, Panisch S, Kowatsch P, Klima G, et al. The effectiveness of the Austrian disease management programme for type 2 diabetes: a cluster-randomised controlled trial. BMC Fam Pract. 2010 Nov 5;11(1):86.PubMedCrossRefGoogle Scholar
  13. 13.
    Sonnichsen AC, Rinnerberger A, Url MG, Winkler H, Kowatsch P, Klima G, et al. Effectiveness of the Austrian disease-management-programme for type 2 diabetes: study protocol of a cluster-randomized controlled trial. Trials. 2008;9:38.PubMedCrossRefGoogle Scholar
  14. 14.
    Flamm M., Winkler H, Panisch S, Kowatsch P, Klima G, Furthauer B, et al. Effektivität des österreichischen Disease-Management-Programms “Therapie Aktiv” für Diabetes mellitus Typ 2. ZFA. 2011;87(3):116–22.Google Scholar
  15. 15.
    Austrian Diabetes Association (ÖDG). Diabetes mellitus—guidelines for the practice. Revised and expanded 2007 edition. Wien Klin Wochenschr. 2009;121 Suppl 5:S1–87.Google Scholar
  16. 16.
    Flamm M, Panisch S, Winkler H, Sonnichsen AC. Impact of a randomized control group on perceived effectiveness of a disease management programme for diabetes type 2. Eur J Public Health. 2011 Oct 11.Google Scholar
  17. 17.
    UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet. 1998 Sep 12;352(9131):837–53.Google Scholar
  18. 18.
    Selvin E, Marinopoulos S, Berkenblit G, Rami T, Brancati FL, Powe NR, et al. Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Ann Intern Med. 2004 Sep 21;141(6):421–31.PubMedGoogle Scholar

Copyright information

© Springer-Verlag Wien 2012

Authors and Affiliations

  • Maria Flamm
    • 1
  • Sigrid Panisch
    • 1
  • Henrike Winkler
    • 1
  • Tim Johansson
    • 1
  • Raimund Weitgasser
    • 2
    • 3
  • Andreas C. Sönnichsen
    • 1
  1. 1.Institute of General Practice, Family Medicine and Preventive MedicineParacelsus Medical UniversitySalzburgAustria
  2. 2.Department of Internal Medicine, Diabetes UnitDiakonissen HospitalSalzburgAustria
  3. 3.Austrian Diabetes Association (ÖDG)ViennaAustria

Personalised recommendations