Per-Henrik Groop, Folkhälsan Research Center, Biomedicum Helsinki, University of Helsinki, FIN-00014 Helsinki, Finland. (fax: +358-9-191 25452; e-mail: firstname.lastname@example.org).
Abstract. Forsblom C, Thomas MC, Moran J, Saraheimo M, Thorn L, Wadén J, Gordin D, Frystyk J, Flyvbjerg A, Groop P-H, on behalf of the FinnDiane Study Group (Folkhälsan Institute of Genetics, Helsinki; Department of Medicine Helsinki University Central Hospital, Helsinki, Finland; The Baker IDI Heart and Diabetes Institute, Melbourne, Vic.; The Queen Elizabeth Hospital, Woodville, SA, Australia; and Aarhus University Hospital, Aarhus C., Denmark). Serum adiponectin concentration is a positive predictor of all-cause and cardiovascular mortality in type 1 diabetes. J Intern Med 2011; 270: 346–355.
Background. Adiponectin is widely regarded as an anti-atherogenic, antioxidant and anti-inflammatory molecule. However, adiponectin concentration is paradoxically increased in individuals with type 1 diabetes, in whom it is positively associated with adverse clinical outcomes.
Objective. To explore the association between serum adiponectin concentration and mortality outcomes in adults with type 1 diabetes.
Design. Multicentre prospective cohort study.
Setting. Primary and tertiary care.
Subjects. Finnish adults with type 1 diabetes (n = 2034).
Main outcome measures. All-cause and cardiovascular mortality. Independent predictors of mortality were determined using the Cox and the Fine and Gray competing risks proportional hazards models.
Results. During a median of 11 years of follow-up, there were 173 deaths (8.5%, 1.0 per hundred person-years). Adiponectin was linearly associated with all-cause mortality [Cox model: hazard ratio (HR) 1.02, 95% confidence interval (CI) 1.01–1.03, P < 0.001] and cardiovascular mortality (Fine and Gray model: HR 1.02, 95% CI 1.00–1.04, P = 0.035); patients with the highest adiponectin concentrations had the shortest survival. The mortality risk associated with adiponectin was independent of glycaemic and lipid control, pre-existing cardiovascular disease, markers of inflammation and the presence and severity of kidney disease.
Conclusions. Although adiponectin is generally considered to be a protective molecule, increased concentrations of adiponectin in type 1 diabetes are independently associated with all-cause and cardiovascular mortality. Moreover, the fact that this association was observed for the first time in patients with normal urinary albumin levels, who have few comorbidities, suggests that adiponectin is specifically linked with vascular damage in type 1 diabetes.
Serum adiponectin concentration is elevated in individuals with type 1 diabetes [1, 2]. It has been suggested that this increase in concentration might be compensatory or vasculoprotective, as adiponectin has been shown to have a range of anti-inflammatory, antioxidant and anti-atherosclerotic actions [3–6]. We have previously reported that adiponectin levels are positively associated with all-cause mortality in Finnish adults with type 1 diabetes, with the highest mortality observed in those with the highest circulating concentration [2, 7]. Adiponectin concentration has also been positively associated with mortality outcome in patients with type 1 diabetes and overt nephropathy in a previous small study . However, this association is potentially confounded by the fact that elevated adiponectin concentration is associated with increased risk of microangiopathy and progressive nephropathy in individuals with type 1 diabetes [2, 9, 10], and the majority of deaths occur in individuals with established chronic kidney disease (CKD) . By contrast, in two small case–control studies in diabetic patients, increased adiponectin concentration was associated with a lower risk of coronary artery disease or progression of coronary artery calcification [12, 13]. Consequently, the true (and renally independent) nature of the association between adiponectin and cardiovascular outcomes in individuals with type 1 diabetes remains to be clearly determined. In the present study, we have investigated the association between serum adiponectin concentration and all-cause and cardiovascular mortality in a large nationwide multicentre cohort of Finnish adults with type 1 diabetes (the FinnDiane study).
The FinnDiane study has been described in detail previously [2, 7]. In brief, the study was established at Helsinki University Central Hospital, Finland, to study the clinical, biochemical, environmental and genetic risk factors for type 1 diabetes and its complications. All adults with type 1 diabetes attending diabetic and/or renal outpatient clinics at 21 university and central hospitals, 33 district hospitals and 26 primary healthcare centres across Finland were asked to participate in the study. The ethical committees of all participating centres approved the study protocol. As previously reported, approximately 78% of diabetic patients in these centres were recruited. Written informed consent was obtained from each patient, and the study was performed in accordance with the Declaration of Helsinki as revised in 2000. For this study, consecutive patients who had been recruited into the FinnDiane prospective cohort between 1995 and 2005 were included, and outcomes were ascertained for those participants without end-stage kidney disease (ESKD) (n = 2034).
Details of clinical status, including age at diagnosis, presence and severity of diabetic complications, insulin therapy and other regular medications were obtained from the attending physician using a standardized questionnaire. Data on smoking habits, alcohol intake, level of education and social class were obtained using a patient questionnaire. Fasting blood samples were obtained for measurement of levels of Haemoglobin A1c (HbA1c) lipids, high-sensitivity C-reactive protein and serum creatinine. The estimated glomerular filtration rate (eGFR) was calculated using the abbreviated Modification of Diet in Renal Disease equation . Urinary albumin excretion rate (AER) was stratified for each patient according to the International Diabetes Federation guidelines  on the basis of three consecutive timed urine collections.
Serum adiponectin concentration was determined by an in-house time-resolved immunofluorometric assay based on two monoclonal antibodies and recombinant human adiponectin (R&D Systems, Abingdon, UK), as previously described . Adiponectin has a molecular weight of ∼30–36 kDa depending on the degree of glycosylation, but the molecule is known to form a wide range of polymers in vivo. The predominant polymers include trimers, hexamers and highly congregated multimers of ∼300 kDa. Both of the antibodies were able to detect several adiponectin polymers in serum, including the three major molecular forms . Within- and between-assay coefficients of variation averaged <5% and 10%, respectively, for both standards and unknown samples. The recovery of exogenously added adiponectin to serum was 101 ± 1% (mean ± SEM based on 10 samples) . Repetitive thawing and freezing of serum for up to seven cycles did not alter the immunoreactive levels of serum total adiponectin (data not shown) .
Follow-up of patients
All patients have now been followed and their outcomes until 17 March 2010 recorded (median follow-up 10 years). For this study, all-cause mortality was identified via a search of the Finnish National Death Registry and centre databases. All deaths were confirmed with death certificate data. Death from cardiovascular disease (CVD) was defined as a death because of an underlying or immediate cardiovascular cause either according to the death certificate or Statistics Finland. At Statistics Finland, any inadequate, contradictory or difficult to classify deaths are assessed by independent physicians who can require more information if necessary to verify the event. The International Classification of Diseases codes I20–25 and I60–69 were used for classification of CVD mortality.
We used stepwise regression analysis to identify variables independently associated with adiponectin concentration, using the log of the concentration as the dependent variable, as its distribution was skewed and kurtotic (Figure S1). The proportion of the total variance that was attributed to predictor variables in a multiple regression model was calculated as the regression effect size (eta2; stata™ statistical software, V11.1, 2010; College Station, TX, USA). To evaluate the independent predictors of all-cause mortality in individuals with type 1 diabetes, we used the Cox proportional hazards model. The predictors of the cumulative incidence of cardiovascular mortality, accounting for the competing event of noncardiovascular death, were ascertained using the Fine and Gray model , which extends the Cox proportional hazards model to competing risks data by considering the subdistribution hazard . The strength of the association between each predictor variable and the outcome was assessed using the subhazard ratio , which is the ratio of hazards associated with the cumulative incidence function in the presence and absence of a prognostic factor. Model selection from candidate variables was accomplished by minimization of the Akaike and Bayesian information criteria . In both models, covariate functional form (including assessment of nonlinear effect) was adjudged by residual-by-time analysis and (cubic) regression splines , where appropriate. The potential for multiple colinearity was tested using the variance inflation factor and condition number, where a variance inflation factor <10 and condition number <30 is desirable . Overall, the Cox model fit was assessed by (i) approximation of cumulative Cox–Snell residuals to (−log) Kaplan–Meier estimates and residual plots (including a formal test of proportional hazards and ‘added-variable’ goodness-of-fit tests  and (ii) the Harrell’s C statistic . The Cox model performance was judged by the explained variation (R2) using 1000 bootstrap repetitions of the whole data set, adjusting for covariates via the ‘str2d’ Stata™ module . The Fine and Gray model was implemented in Stata™ statistical software using the ‘stcrreg’ module. As standard errors of the Fine and Gray model are robust (Huber–White type), model selection was guided by information criteria, as mentioned earlier. The model specification was established by residual (score and Schoenfeld) analysis, DFBETA measures of influence across each model covariate and formal check of proportionality by the use of time-varying covariate effects.
The FinnDiane cohort, in which serum adiponectin concentrations were estimated, comprised 2034 adult patients with type 1 diabetes. The baseline cohort characteristics are summarized in Table 1. Briefly, approximately half of the participants were men (51%). The mean age of the participants was 39 years, with a median duration of diabetes of 20 years. Forty-seven per cent of patients had hypertension (defined by the use of anti-hypertensive agents and/or blood pressure >140/90 mmHg). At baseline, 15% had microalbuminuria (assessed by urinary AER) and 15% had macroalbuminuria; 65% had a urinary AER in the normal range. A further 5% of study participants were unclassified because of an inadequate number of urine collections. Only 12% of patients had an eGFR <60 mL min−1 per 1.73 m2, most of whom also had macroalbuminuria.
Table 1. Baseline clinical characteristics of 2034 patients with type 1 diabetes from the FinnDiane study, stratified according to serum adiponectin concentration
Lower quartile <8.6
Middle quartiles 8.6–16.5
Upper quartile >16.5
Values are mean ± SEM, unless otherwise indicated. *Versus middle quartiles, univariate P < 0.05.
35 ± 1*
37 ± 1
41 ± 1*
Male gender (%)
Duration of diabetes (years)
17 ± 1*
20 ± 1
25 ± 1*
Insulin dose (IU kg−1)
0.8 ± 0.1*
0.7 ± 0.1
0.6 ± 0.1*
8.4 ± 0.1*
8.5 ± 0.1
8.5 ± 0.1*
Systolic blood pressure (mmHg)
131 ± 1
131 ± 1
136 ± 1 *
Diastolic blood pressure (mmHg)
81 ± 1
80 ± 1
79 ± 1
Antihypertensive medication use (%)
Angiotensin-converting enzyme inhibitor
Angiotensin receptor blocker
Calcium channel blocker
Lipid-lowering therapy (%)
Total cholesterol (mmol L−1)
4.8 ± 0.1*
5.0 ± 0.1
5.2 ± 0.1*
Low-density lipoprotein cholesterol (mmol L−1)
3.1 ± 0.1*
3.2 ± 0.1
3.3 ± 0.1*
High-density lipoprotein cholesterol (mmol L−1)
1.1 ± 0.1*
1.2 ± 0.1
1.4 ± 0.1*
Triglycerides (mmol L−1)
1.5 ± 0.1*
1.3 ± 0.1
1.2 ± 0.1*
Any retinopathy (%)
Retinopathy requiring laser therapy (%)
Current smoker (%)
Established macrovascular disease (%)
Glomerular filtration rate (mL min−1 per 1.73 m2)
87 ± 1*
82 ± 1
68 ± 1*
Median C-reactive protein (mg L−1)
Determinants of serum adiponectin concentration in patients with type 1 diabetes
In this cohort, the mean (±SD) adiponectin concentration in serum was 13.6 ± 8.2 mg L−1. The concentration distribution across the FinnDiane cohort has been previously described  and is presented as Figure S1. Serum adiponectin concentration was independently associated, by multivariate regression analysis, with eGFR, AER, high- and low-density lipoprotein cholesterol concentrations, waist–hip ratio, insulin dose, duration of diabetes and diuretic use (Table 2). As previously described, the presence and severity of CKD were the major determinants of adiponectin concentration. In particular, the eGFR was negatively correlated with the adiponectin concentration, consistent with the glomerular filtration of this low molecular weight protein . However, for any level of eGFR, adiponectin concentration was higher in patients with increased urinary albumin excretion (AER). Adiponectin concentration was also associated with markers of insulin resistance, including dyslipidaemia, insulin dose and waist–hip ratio (Table 2), consistent with reduced concentrations seen in type 2 diabetes and in insulin-resistant patients with the metabolic syndrome . Previous studies have suggested that some anti-hypertensive medications might impact on adiponectin production [28, 29]. We found no association between treatment with an angiotensin-converting enzyme inhibitor or an angiotensin receptor blocker and adiponectin, although diuretic users had modestly increased concentrations (Table 2).
Table 2. Variables independently associated with serum adiponectin concentrations in multivariate regression analysis. The proportion of the total variance attributable to predictor variables is calculated as the regression effect size (eta2)
Regression effect size (% change)
Estimated glomerular filtration rate (mL min−1 per 1.73 m2)
Albumin excretion rate (log)
High-density lipoprotein cholesterol (mmol L−1)
Low-density lipoprotein cholesterol (mmol L−1)
Insulin dose (units kg−1)
Duration of diabetes (years)
Diuretic use (yes/no)
C-reactive protein (mg L−1)
Adiponectin and mortality outcomes
Median follow-up was 11.0 years, during which time there were 173 deaths (8.5%, 1.0 per hundred person-years). Serum adiponectin concentration was significantly associated with all-cause mortality. After adjusting for factors associated with serum adiponectin concentration, as well as other factors independently associated with all-cause mortality (including age, duration of diabetes, the presence and severity of CKD, the presence of established CVD, triglycerides and glycaemic control) (Table 3), adiponectin concentration remained significantly associated with all-cause mortality in multivariate Cox regression analysis. The association between adiponectin and all-cause mortality appeared to be linear, such that individuals with the highest adiponectin concentrations had the highest mortality, and vice versa (Table S1).
Table 3. Variables independently associated with all-cause mortality in patients with type 1 diabetes from the FinnDiane cohort in a Cox regression model
95% confidence interval
aNonlinear relationships were established by fractional polynomials (FP); HbA1c is modelled as a third-degree FP and triglycerides as a -2 FP. For interpretation of their functional form see Figure S2.
Macrovascular disease (yes/no)
Current smoker (yes/no)
Glomerular filtration rate (mL min−1 per 1.73 m2)
Fasting triglycerides (mmol L−1)a
C-reactive protein (mg L−1)
Adiponectin (mg L−1)
No convincing evidence of nonlinear effects was demonstrated by (martingale) residual-by-time analysis, fractional polynomials and (cubic) regression splines. There was no evidence of significant interactions. In particular, there was no interaction between the severity of CKD, adiponectin concentration and mortality (P = 0.29). This relationship was not confounded by the known association between adiponectin and ESKD, as adiponectin concentration was also associated with survival in patients with normoalbuminuria at baseline, of whom none developed ESKD during follow-up (hazard ratio 1.04; 95% confidence interval 1.00–1.07, P = 0.03, Table S2).
The Cox model was well specified (Harrell’s C = 0.87; goodness-of-fit test, P = 0.94), satisfied the proportional hazards assumption (P = 0.49) and had good explanatory power (adjusted R2 = 0.83). Although previous studies have suggested that the association between adiponectin concentration and adverse outcomes might be different in men and women , no gender-dependent interaction was identified in our model. In addition, the relationship between adiponectin and mortality was independent of glycaemic control and fasting triglyceride levels, both of which were nonlinearly associated with mortality (Table 3), as previously described in this cohort .
Adiponectin and cardiovascular mortality
In this study, we specifically examined cardiovascular deaths whilst taking into consideration (in an estimation sense) the competing risk of noncardiovascular death (Fine and Gray model) . This strategy might be especially important in patients with diabetes, as this disease is also associated with increased noncardiovascular mortality , which might potentially confound cause-specific analysis. In our competing risk model, the circulating concentration of adiponectin was also significantly associated with the cumulative incidence of cardiovascular mortality (Fig. 1, Table 4). Again, this association was linear and independent of other predictive variables including age, duration of diabetes, the presence and severity of CKD and the presence of macrovascular disease. The model was well specified, and the proportional hazards assumption was not violated. The cumulative incidence of cardiovascular mortality across the adiponectin percentiles is demonstrated in Fig. 1. Although the effect of adiponectin was a (simple) linear function on the (log) subhazard scale (Table 4), the Fine and Gray model is nonlinear on the cumulative incidence scale and covariates might have different effects on the component cause-specific hazards (of cardiovascular and noncardiovascular deaths). This is illustrated by the different effect slopes and separation of the percentile lines in Fig. 1.
Table 4. Competing risk model of variables associated with the cumulative incidence of cardiovascular mortality in patients with type 1 diabetes from the FinnDiane cohort, taking into account the competing risk of noncardiovascular death. No time-varying covariates were demonstrated, P ≥ 0.10
95% confidence interval
Macrovascular disease (yes/no)
Adiponectin (mg L−1)
Adiponectin concentration is elevated in individuals with type 1 diabetes [1, 2]. Here, we demonstrate that individuals with the highest adiponectin concentrations have the highest all-cause and cardiovascular mortality. These data are consistent with recent studies showing a positive association between adiponectin level and mortality outcomes in other settings, including in patients with CKD , chronic heart failure  and established coronary artery disease  and in the elderly , all of whom share features of comorbid vascular disease.
Our data contrast with those from two case–control studies, in which increased adiponectin concentration in individuals with type 1 diabetes was associated with a lower risk of coronary artery disease or progression of coronary artery calcification [12, 13]. However, these studies were small and did not correct for the presence and severity of CKD. This is particularly important in patients with type 1 diabetes, as renal function is a major determinant of adiponectin concentration , and renal disease is strongly associated with cardiovascular risk . In larger diabetic cohorts, adiponectin concentration has been positively associated with mortality outcomes . However, this association was only observed in diabetic patients with overt nephropathy, in whom adiponectin was also associated with an increased incidence of ESKD, potentially confounding this analysis . No association between adiponectin and mortality was demonstrated in a subset of diabetic patients without kidney disease (P = 0.43) , although with small numbers and few events this could represent a type II error. In our much larger cohort, the mortality risk associated with adiponectin was independent of the presence and severity of kidney disease. In addition, our findings were not confounded by the known association between adiponectin concentration and progression to ESKD , and it was observed in patients with normoalbuminuria, none of whom progressed to ESKD during the study follow-up. It is possible that through its association with renal dysfunction in type 1 diabetes, adiponectin might also be associated with increased risk of mortality associated with CVD, infection and malignancy, as the incidence of these adverse outcomes in patients with microalbuminuria is more than twofold greater than in those without nephropathy . However, we have previously shown that adiponectin is not associated with the progression of normoalbuminuria to microalbuminuria in this cohort  and <7% of patients with normal levels of albumin developed albuminuria during follow-up . This suggests that any relationship between adiponectin and mortality is potentially independent of any association with occult kidney damage. Moreover, the fact that patients with normal albumin levels share none of the comorbidities in which an association between adiponectin and mortality has been previously documented (e.g. advanced age, chronic heart failure and CKD) [32–35] suggests that adiponectin is specifically linked with vascular damage leading to all-cause and cardiovascular mortality in type 1 diabetes.
An increased adiponectin concentration is observed very early in the course of type 1 diabetes , correlating with the loss of β-cell function . It has been suggested that the loss of C-peptide that is normally cosecreted with insulin might increase adiponectin production in type 1 diabetes . By contrast, its over-secretion in the presence of impaired glucose regulation and early type 2 diabetes might suppress adiponectin [26, 38, 39]. Notably, duration of type 2 diabetes is associated with an increase in circulating adiponectin concentration , supporting an association with the loss of β-cell function. These complicated changes might be a reason why previous studies have failed to demonstrate a consistent association between adiponectin and cardiovascular outcomes in type 2 diabetes, although low adiponectin concentration is clearly associated with an adverse cardiovascular risk profile and obesity .
A positive association between adiponectin and mortality has also been ascribed to ‘reverse causality’, whereby silent or overt vascular disease leads to compensatory rises in adiponectin concentration to counteract metabolic and vascular stress . Interestingly, in our cohort, adiponectin was not associated with the presence of overt CVD, and the association between adiponectin and mortality risk was no stronger in those with prevalent vascular disease nor in those with overt CKD, in whom silent CVD is known to be more prevalent. This is similar to findings in elderly patients, in whom the association between adiponectin and mortality is also seen in both the presence and the absence of established CVD . It has also been suggested that elevated adiponectin concentration is a marker of increased catabolism. Indeed, poorly controlled type 1 diabetes might be associated with increased catabolism. However, the association between all-cause mortality and adiponectin concentration was independent of glycaemic control and other markers of catabolism including weight and waist–hip ratio. This is similar to findings in other clinical settings, in which the association between adiponectin and mortality appears to be independent of markers of catabolism [35, 42].
The strengths of our study include its very large cohort of individuals with type 1 diabetes, high participation rate, long follow-up and access to subsidized care (75–100% of costs), contemporary treatment regimens (including a range of insulin regimens, statins and blockers of the renin–angiotensin system and self-monitoring technologies. We used validated methods to identify deaths, and all deaths in our cohort were confirmed through death records. Surveillance bias is unlikely given the uniform vital status follow-up procedures used. In our questionnaire, we collected broad data on tobacco or alcohol use, diet, education, socio-economic status, other possible confounders (e.g. insulin resistance) or the severity of disease. There have been few changes in diabetes treatment and healthcare over the study period, so the results are unlikely to have been confounded. In our statistical analysis, we specifically incorporated the nonlinear effects of predictive variables, modelled for interactions and, in the case of cardiovascular mortality, modelled within the paradigm of a formal competing risks (Fine and Grey) model, which looked at the cumulative incidence of cardiovascular deaths whilst taking into consideration (in an estimation sense) the competing risk of other causes of death, which might otherwise have confounded our results .
Several study limitations need to be considered. First, these studies were limited by reliance on data obtained from a single blood collection. Second, although our assay of total adiponectin incorporated the detection of the major polymeric molecular forms of adiponectin in serum, we did not differentiate between specific molecular forms. It is now clear that different forms of adiponectin might play distinct roles in the regulation of metabolism. Moreover, defects in adiponectin multimerization are associated with diabetes which might confound interpretation of these results. Nonetheless, high molecular weight adiponectin has the most potent insulin-sensitizing activity and is closely correlated with total adiponectin concentration in patients with the metabolic syndrome and diabetes . Third, these data are essentially observational, and although observational studies have a number of potential advantages , it is also possible that associations demonstrated in this study may be due to confounding by unmeasured factors or those that are difficult to quantify. For example, adiponectin concentration might be associated with a range of differences in metabolism that might impact on adverse outcomes in diabetic individuals.
In summary, an increased concentration of adiponectin in type 1 diabetes is associated with poor clinical outcome, with respect to all-cause and cardiovascular mortality. This finding cannot be explained by renal dysfunction, catabolism, inflammation or pre-existing CVD in diabetic patients and suggests an independent link between adiponectin and cardiovascular outcomes in type 1 diabetes. Whether adiponectin is a marker or mediator of compensatory pathways to try to stem the progression of vasculopathy remains to be established.
Conflict of interest statement
The authors declare no conflicts of interest.
This study was supported by the Danish Medical Research Council, the Danish Diabetes Association, the Novo Nordisk Foundation and the Clinical Institute, Aarhus University, Aarhus, Denmark. The FinnDiane Study was supported by grants from the Folkhälsan Research Foundation, Wilhelm and Else Stockmann Foundation, Liv och Hälsa Foundation and the Finnish Medical Society (Finska Läkaresällskapet). We thank all the patients who participated in the study and gratefully acknowledged the skilful laboratory assistance of Maikki Parkkonen, Anna Sandelin, Anna-Reetta Salonen, Tuula Soppela and Jaana Tuomikangas, as well as Karen Mathiassen, Hanne Petersen and Anette Mengel at the Medical Research Laboratories, Aarhus University Hospital, Aarhus, Denmark. We also gratefully acknowledge all the physicians and nurses at each centre participating in the recruitment of patients: Central Finland Central Hospital: A. Halonen, A. Koistinen, P. Koskiaho, M. Laukkanen, J. Saltevo and M. Tiihonen; Central Hospital of Kanta-Hame: P. Kinnunen, A. Orvola, T. Salonen and A. Vähänen; Central Hospital of Kymenlaakso: R. Paldanius, M. Riihelä and L. Ryysy; Central Hospital of Lansi-Pohja: P. Nyländen and A. Sademies; Central Ostrobothnian Hospital District: S. Anderson, B. Asplund, U. Byskata and T. Virkkala; City of Vantaa Health Center: (Rekola) M. Eerola and E. Jatkola, (Tikkurila) R. Lönnblad, J. Mäkelä, A. Malm, and E. Rautamo; Helsinki University Central Hospital (Department of Medicine, Division of Nephrology): H. Rosvall, M. Rosengård-Bärlund, M. Rönnback and J. Wadén; Iisalmi Hospital: E. Toivanen; Kainuu Central Hospital: S. Jokelainen, P. Kemppainen, A.-M. Mankinen and M. Sankari; Kerava Health Center: H. Stuckey and P. Suominen; Kouvola Health Center: E. Koskinen and T. Siitonen; Kuopio University Hospital: M. Laakso, L. Niskanen, I. Vauhkonen, T. Lakka, E. Voutilainen, L. Mykkänen, P. Karhapää, E. Lampainen and E. Huttunen; Kuusamo Health Center: E. Vierimaa, E. Isopoussu and H. Suvanto; Kuusankoski Hospital: E. Kilkki and L. Riihelä; Lapland Central Hospital: L. Hyvärinen, S. Severinkangas and T. Tulokas; Länsi-Uusimaa Hospital, Tammisaari: J. Rinne and I.-M. Jousmaa; Mikkeli Central Hospital: A. Gynther, R. Manninen, P. Nironen, M. Salminen and T. Vänttinen; North Karelian Hospital: U.-M. Henttula, P. Kekäläinen, A. Rissanen and M. Voutilainen; Paijat-Hame Central Hospital: H. Haapamäki, A. Helanterä and H. Miettinen; Palokka-Vaajakoski Health Center: P. Sopanen, L. Welling and K. Mäkinen; Pori City Hospital: K. Sävelä, P. Ahonen and P. Merensalo; Riihimäki Hospital: L. Juurinen and E. Immonen; Salo Hospital: J. Lapinleimu, M. Virtanen, P. Rautio and A. Alanko; Satakunta Central Hospital: M. Juhola, P. Kunelius, M.-L. Lahdenmäki, P. Pääkkönen and M. Rautavirta; Savonlinna Central Hospital: T. Pulli, P. Sallinen, H. Valtonen and A. Vartia; Seinajoki Central Hospital: E. Korpi-Hyövälti, T. Latvala and E. Leijala; South Karelia Hospital District: E. Hussi, T. Hotti, R. Härkönen and U. Nyholm; Tampere University Hospital: I. Ala-Houhala, T. Kuningas, P. Lampinen, M. Määttä, H. Oksala, T. Oksanen, K. Salonen, H. Tauriainen and S. Tulokas; Turku Health Center: I. Hämäläinen, H. Virtamo and M. Vähätalo; Turku University Central Hospital: M. Asola, K. Breitholz, R. Eskola, K. Metsärinne, U. Pietilä, P. Saarinen, R. Tuominen and S. Áyräpää; and Vasa Central Hospital: S. Bergkulla, U. Hautamäki, V.-A. Myllyniemi and I. Rusk.