With written informed consent, eligible patients were randomly assigned in a 1:1 manner to receive either structured care or usual care using a computer-generated allocation code with equal numbers in each group within the same hospital. The sealed envelope was opened sequentially by nonstudy personnel. The structured care group was followed up at the Diabetes Centre by the specialist team using a predefined protocol, whereas the usual care group was managed in their original clinic but returned for assessments at 2 years. The study was approved by the research ethics committee of each institution.

Inclusion criteria included a diagnosis of type 2 diabetes (8), age 35–75 years (inclusive), and plasma creatinine level 150–350 μmol/l (inclusive). Exclusioncriteria included malignancy or life-threatening diseases; nondiabetes renal disease (e.g., biopsy-proven glomerulonephritis orobstructive uropathy); unstable psychiatric illnesses, and renal function (≥20% difference in two consecutive plasma creatinine values within 3 months before recruitment).

The primary composite end point was death and/or ESRD, defined as the need for dialysis, or plasma creatinine level ≥500 μmol/l. The secondary end point was the percentage of patients attaining ≥3 treatment targets and their impacts on clinical outcomes.

In both groups, the composite end point occurred in 24 patients (RR 0.96 [95% CI 0.50–1.84]) with similar rates of clinical events, hospitalization, and emergency room visits (Table 1). Among the nine study sites, the structured care group had fewer event rates than the usual care group in five hospitals, higher rates in two hospitals, and similar rates in two hospitals.

At 2 years and after adjustment for age, sex, and study sites, the structured care group had lower diastolic blood pressure and A1C levels and were more likely to attain ≥3 treatment goals (61% [ n = 63] vs. 28% [ n = 28]) (Table 2). With per protocol analysis, patients who attained ≥3 treatment goals (n = 91) had a 60% risk reduction in reaching the primary end point compared with those who did not attain ≥3 treatment goals (n = 114) (14 vs. 34; RR 0.43 [95% CI 0.21–0.86]) (Table 3; Fig. 2).

To date, most clinical studies are “mechanism research,” with the aim of understanding disease causality and efficacy of interventions under controlled environments. This type of research does not address issues such as barriers in translating such evidence to daily clinical practice. “Translational research” is more focused on outcomes of practical relevance to patients, health care providers, and health care systems. It examines the relationships between structural factors (e.g., practice structure and health care personnel), care processes (e.g., frequencies of measurements of A1C), and health outcomes in the real-life setting. Thus, whereas translational research has higher external validity (i.e., the possibility of generalizing study results), it has lower internal validity (i.e., the strength of the causal relationship between intervention and outcome of interest) than mechanism research. Nevertheless, these two types of research are complementary because once the evidence base of an intervention is established, there is a need to translate it to practice. With the growing popularity of chronic disease management, translation research is important in evaluating the effectiveness of these programs.

Several factors may explain our failure to prove that structured care was better than usual care in improving clinical outcomes. These include underpower of the study because of general improvement in the care standard, contamination by specialist care in the usual care group, and insufficient auditing to ensure adherence to the structured care protocol. When we first conceptualized the study in 2000, we used data available then to estimate sample size. In the Asian subgroup analysis of the Reduction of End Points in Type 2 Diabetes With the Angiotensin II Antagonist Losartan (RENAAL) study, which examined the renoprotective effects of losartan in type 2 diabetic patients with plasma creatinine (110–260 μmol/l) levels lower than those of the present cohort (150–350 μmol/l), the 3-year cumulative incidence of the primary end point of death and ESRD was 48% (9). In a subsequent proof-of-concept study involving patients with renal function comparable to that of the present cohort, the 2-year cumulative incidence of death and ESRD was 50% in the usual care group and 25% in the structured care group (6). Compared with these two studies (1997–2002), the 2-year cumulative incidence of the primary end point in the present cohort (2003–2007) was 24% in both the structured care and usual care groups. These findings suggest that increasing awareness of the beneficial effects of intensive risk factor control and inhibition of the renin-angiotensin system (10) might have led to improvement in the care standard and the reduced rate of clinical end points, at least in the Hong Kong setting.

Although this improvement in the care standard might have attenuated the benefits of structured care in our study, consistent with our predefined hypothesis, more patients in the structured care group attained ≥3 treatment goals (61%) than in the usual care group (28%). This difference was translated to 60–70% risk reduction in premature death and ESRD. In an observational study of 6,386 type 2 diabetic patients, attainment of ≥2 treatment goals (A1C, blood pressure, or LDL cholesterol) was associated with 30–50% risk reduction in new onset of CHD (11). These findings strongly support the need to attain multiple treatment targets to improve cardiorenal complications in type 2 diabetes. Because of the small sample size, we were unable to stratify patients by severity of renal function to perform subgroup analysis.

Although the structured care group was three times (61%) more likely than the usual care group (28%) to attain ≥3 treatment targets, there was considerable heterogeneity in event rates among hospitals, possibly due to different care systems. In two study sites, event rates were higher in the structured care group; in one, the result was due to a higher baseline plasma creatinine level in the structured care group, and in another, the usual care group was managed by the specialist team with regular review meetings. These findings have provided important insights into the potential effects of care organization on clinical outcomes. Thus, to fully realize the benefits of protocol-driven care, trained personnel or information technology is needed to ensure adherence to protocol and attainment of treatment targets through regular audits and feedback, a setup not dissimilar to a clinical trial setting (12). However, because of resource limitation, we were unable to implement these measures to complete the cycle of quality planning, assurance, and improvements (13).

There are also multiple barriers to translating evidence to practice at the levels of patients, care providers, and health care systems (14,15). In most clinical audits, <10% of type 2 diabetic patients attained ≥3 treatment goals of blood pressure, LDL cholesterol, and A1C (15). In addition, 20–50% of patients were noncompliant with life-saving drugs such as statins (16,17). Whereas clinical inertia of physicians might delay commencement or escalation of therapy (18), frequent testing of risk factors may not translate to changes in treatment regimens (19). Although use of e-mails and telephone calls improved doctors' adherence to some performance indexes (e.g., blood tests), the effects of these behavioral changes on clinical outcomes have not been explored (20). Conversely, in a 2-year randomized study, regular telephone calls by a pharmacist to reinforce compliance and ensure continuation of care reduced mortality and hospitalization by 30–50% in patients receiving ≥5 chronic medications including drugs for diabetes and CHD (17).

Thus, in the management of chronic diseases such as diabetes, an integrated approach including early diagnosis, risk stratification, use of protocols with predefined follow-up schedules, assessments, prompts for intervention further augmented by audits, and patient empowerment is needed to achieve desired outcomes (13). It has been suggested that if the specialist-led team care in the Steno-2 study can be translated to the primary care setting, such a model can be cost-saving (21). However, the challenge lies in identifying effective measures to ensure integration between different care providers with adequate communication and quality assurance. To this end, change in the practice environment to promote multidisciplinary care and self-management (22), a community-based shared care program (23), and pay-for-performance schemes (24) have improved adherence to treatment guidelines and metabolic control in type 2 diabetes.