Article Text

Download PDFPDF

Research paper
Metabolic syndrome is related to polyneuropathy and impaired peripheral nerve function: a prospective population-based cohort study
  1. Rens Hanewinckel1,2,
  2. Judith Drenthen2,3,
  3. Symen Ligthart1,
  4. Abbas Dehghan1,
  5. Oscar H Franco1,
  6. Albert Hofman1,4,
  7. M Arfan Ikram1,
  8. Pieter A van Doorn2
  1. 1Department of Epidemiology, Erasmus University Medical Centre Rotterdam, Rotterdam, The Netherlands
  2. 2Department of Neurology, Erasmus University Medical Centre Rotterdam, Rotterdam, The Netherlands
  3. 3Department of Neurophysiology, Erasmus University Medical Centre Rotterdam, Rotterdam, The Netherlands
  4. 4Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, USA
  1. Correspondence to Dr M Arfan Ikram, Department of Epidemiology, Erasmus University Medical Centre, P.O. Box 2040, Rotterdam 3000 CA, The Netherlands; m.a.ikram{at}


Objective Diabetes mellitus is a known risk factor for polyneuropathy, but the role of pre-diabetes and metabolic syndrome remains unclear. We aimed to investigate the role of these factors in a community-dwelling middle-aged and elderly population.

Methods 1256 participants of the population-based Rotterdam Study (mean age 70.0, 54.5% females) were screened for polyneuropathy with a questionnaire, neurological examination and nerve conduction studies. Data on type 2 diabetes and components of metabolic syndrome were also collected. Logistic regression was used to investigate associations of diabetes, pre-diabetes and metabolic syndrome and its separate components with polyneuropathy. Linear regression was used to investigate associations with nerve conduction parameters in participants without polyneuropathy.

Findings Diabetes was associated with polyneuropathy (OR 3.01, 95% CI 1.60 to 5.65), while impaired fasting glucose was not (OR 1.55, 95% CI 0.70 to 3.44). Metabolic syndrome was associated with polyneuropathy (OR 1.92, 95% CI 1.09 to 3.38), with a stronger association when more components of the syndrome were present. Analysing separate components of metabolic syndrome revealed associations for elevated waist circumference (OR 2.84, 95% CI 1.35 to 5.99) and elevated triglycerides (OR 2.01, 95% CI 1.11 to 3.62). Similar associations were found after excluding participants with diabetes. In participants without polyneuropathy, metabolic syndrome associated with lower sural sensory nerve action potential amplitudes.

Conclusions Metabolic syndrome, abdominal obesity and dyslipidaemia, are strongly associated with polyneuropathy, irrespective of the presence of diabetes. Metabolic syndrome also associates with impaired nerve function in people without polyneuropathy. Our study therefore suggests that cardiometabolic disturbances have an impact on peripheral nerve function that extends beyond clinically manifest disease.

View Full Text

Statistics from


Diabetes mellitus is a major risk factor for polyneuropathy, a disabling condition that is associated with falls, fractures, ulcers and mortality.1 Impaired fasting glucose (IFG) and impaired glucose tolerance (IGT), together commonly called pre-diabetes, might also increase the risk of polyneuropathy.2 With the globally increasing prevalence of both diabetes and pre-diabetes, incidence of polyneuropathy and its consequences is also expected to increase.3 However, evidence supporting the association between pre-diabetes and neuropathy has so far been inconclusive. Several studies showed a higher prevalence of pre-diabetes in patients with idiopathic neuropathy compared to previously published prevalence numbers in the general population. However, since these studies did not include a control group, a true association cannot be inferred.4–8 Results from studies that did investigate the association between pre-diabetes and neuropathy using a control group have been inconsistent.9–15 The uncontrolled nature, often retrospective study design, lack of appropriate adjustments and potential referral bias of most of these studies have previously been indicated.16

More recently, attention has shifted from hyperglycaemia as a single cause of polyneuropathy to a more multifactorial hypothesis suggesting an interaction of glucose metabolism with other metabolic factors.17 ,18 Obesity, another global epidemic, is one of these factors. Hyperglycaemia and obesity are central components of the metabolic syndrome (MetS).19 This syndrome comprises a combination of interrelated risk factors for cardiovascular diseases and diabetes and also includes high-blood pressure, elevated triglycerides and reduced high-density lipoprotein cholesterol.19 Prevalence of MetS has also been rising and nowadays almost 50% of the US population meets the criteria. MetS is related to cardiovascular disease, neurodegenerative disease and cancer18 and several components of this syndrome have also been implicated in the development of neuropathy in patients with diabetes.20–23 A few case–control studies even suggested an association between MetS and idiopathic polyneuropathy,12 ,24 ,25 but more well-developed, extensive epidemiological studies are necessary to confirm this association.17 ,18 ,26

Therefore, we investigated the association of pre-diabetes and MetS with chronic polyneuropathy in an unselected sample of community-dwelling middle-aged and elderly people.

Materials and methods


The current study was part of the Rotterdam Study, a large prospective, population-based cohort study in the Ommoord district of Rotterdam, the Netherlands.27 The study was initiated in 1990 when all inhabitants aged 55 years or older were invited to participate. There were no other eligibility criteria besides minimum age and postal code. The study was expanded in 2000 and 2006, this last time inviting all persons aged 45 years or older, living in the Ommoord district. Currently, 14 926 participants have been enrolled in the Rotterdam Study. At baseline, and at follow-up every 4 years, participants undergo extensive interviews and examinations at home and at the research centre. The Rotterdam Study has been approved by the Medical Ethics Committee of the Erasmus Medical Centre and by the Ministry of Health, Welfare and Sport of the Netherlands, implementing the “Wet Bevolkingsonderzoek: ERGO (Population Studies Act: Rotterdam Study)”. All participants provided written informed consent to participate in the study and to obtain information from their treating physicians.

Since June 2013, a polyneuropathy screening has been implemented in the core protocol of the Rotterdam Study. The current study includes all participants that have been invited for this assessment between June 2013 and October 2015. During this period, 1544 participants were invited for the polyneuropathy screening. Of these participants, 262 were excluded: 93 did not undergo the screening and 141 participants were not sufficiently screened, mainly due to logistic reasons (shortage of time to complete all examinations or due to a lack of personnel or on some occasions). Of the 1310 participants who were sufficiently screened, information on diabetes and MetS was present in 1256 participants. These 1256 participants were included in the analyses.

Assessment of type 2 diabetes, pre-diabetes and metabolic syndrome

Diabetes mellitus type 2 was diagnosed using serum glucose measurements performed at the research centre, data on medication use (through linkage with pharmacy dispensing data) and general practitioners’ records. Participants with a fasting glucose ≥7.0 mmol/L, a non-fasting glucose level ≥11.1 mmol/L (if fasting samples were not available) and participants who used antidiabetic treatment were considered as having diabetes. Additionally, when participants were diagnosed with incident diabetes in between previous follow-up measurements of the Rotterdam Study, as identified through the link with the general practitioner records, participants were also considered as having diabetes.28 We defined IFG using the WHO 2006 criteria (fasting glucose level ≥6.1 mmol/L and <7.0 mmol/L, in the absence of diabetes).2 Glucose tolerance test were not performed, precluding the possibility to investigate the effect of impaired glucose tolerance.

MetS was defined according to the harmonised criteria published in 2009.19 The presence of at least three of the following five components defined MetS: elevated waist circumference (≥94 cm for males, ≥80 cm for females), elevated triglycerides (≥1.7 mmol/L, or drug treatment for elevated triglycerides), reduced high-density lipoprotein cholesterol (HDL-C; <1.0 mmol/L in males, <1.3 mmol/L in females, or specific treatment for reduced HDL-C), elevated blood pressure (systolic ≥130 mm Hg and/or diastolic ≥85 mm Hg, and/or use of antihypertensive treatment) and elevated fasting glucose (≥5.6 mmol/L, or use of glucose lowering medication). Medication use was assessed by self-report and by going through the medication cabinets at the house of the participants during the home interview. Participants using lipid-lowering medication (the drugs that are mainly prescribed for lipid abnormalities in the Netherlands are statins) were considered to fulfil the triglyceride criterion, but not the HDL-C criterion, since statins only have a marginal effect on HDL-C. Specific drugs that are meant to raise HDL-C levels are very rarely prescribed in our population sample. Waist circumference, height, weight and blood pressure (mean of two consecutive measurements) were measured at the research centre. Information on glucose, cholesterol and triglyceride levels was derived from blood samples taken as close to the polyneuropathy screening as possible. For 730 participants data on these cardiometabolic factors was collected on average 2 months before the polyneuropathy screening. In the remaining 526 participants evaluation of these factors took place at a previous visit (mean 4.9 years before polyneuropathy screening).

Polyneuropathy screening

Participants were screened for the presence of polyneuropathy with an in-person screening consisting of a symptom questionnaire, neurological examination and nerve conduction studies. The questionnaire consisted of sensory and motor symptoms of the legs or feet, like weakness, tingling, numbness, burning, allodynia, cramps and pain. The neurological examination comprised assessment of vibration (Rydel-Seiffer tuning fork) on the hallux of both feet, pain sensation (wooden pin) on the lower legs, muscle strength for dorsiflexion of the feet (Rasch-MRC) and knee and ankle tendon reflexes. Nerve conduction studies were performed on the sural nerve bilaterally and on the peroneal nerve unilaterally. The distal peroneal compound muscle action potential (CMAP) amplitude (mV, baseline-peak) and distal motor latency (ms) were recorded while stimulating 8 cm proximal to the recording electrode, which was placed on the extensor digitorum brevis muscle. The sural sensory nerve action potential (SNAP) amplitude (µV, baseline-peak) and sensory nerve conduction velocity (m/s) were measured by applying stimuli 14 cm proximal to the recording electrode, which was placed behind the lateral malleolus. Standard methods of supramaximal stimulation were applied. Examination took place at room temperature. Skin temperature was documented, but maintaining the temperature above a certain degree was not possible in the current setting.

Medical records were scrutinised when participants scored abnormal on any of the three elements in the screening procedure (symptoms, signs or abnormal nerve conduction parameters) to investigate whether participants received a previous specialist’s diagnosis of polyneuropathy. Participants could also self-report a previous diagnosis of polyneuropathy, which was subsequently checked in records. All collected data, both from the screening and medical records, from each individual participant was evaluated by an expert panel with extensive experience in diagnosing neuromuscular diseases. The panel consisted of an experienced neuromuscular specialist, a neurophysiology specialist and a medical doctor trained in epidemiology with a special interest in neuromuscular diseases. The panel categorised participants into no, possible, probable and definite polyneuropathy, depending on the certainty of the diagnosis. Participants were discussed until unanimity was reached. When all three elements of the screening were abnormal participants were categorised as definite polyneuropathy, and when no elements were abnormal participants were categorised as no polyneuropathy. The remaining participants were categorised as possible and probable polyneuropathy, depending on the level of abnormality. Typically, one abnormal element or two slightly abnormal elements yielded a possible polyneuropathy categorisation, and two abnormal elements a probable categorisation. All participants with a previous diagnosis of polyneuropathy made by a neurologist were categorised as definite polyneuropathy, since we consider a complete clinical work-up performed by a neurologist as superior to our screening. More details on the screening and diagnostic work-up can be found elsewhere.29

Data analysis

Logistic regression analyses were performed to investigate the association of diabetes, pre-diabetes, continuous glucose levels and MetS with (categories of) polyneuropathy. Analyses involving continuous glucose levels were performed using restricted cubic splines regression to assess potential non-linear associations. Analyses involving MetS were repeated after excluding participants with diabetes.

Additionally, linear regression analyses were used to investigate the association of diabetes, pre-diabetes and MetS and its individual components with nerve conduction parameters. These analyses were performed in participants without polyneuropathy in order to investigate whether associations could already be found in the absence of polyneuropathy. For the sural nerve, the side with the highest SNAP amplitude and conduction velocity was used in the analyses.

All analyses were adjusted for age, sex, height and time between assessment of cardiometabolic factors and polyneuropathy screening. Logistic regression analyses involving diabetes and pre-diabetes were additionally adjusted for weight, diastolic blood pressure, systolic blood pressure, blood pressure lowering medication, smoking, serum triglyceride level, serum HDL-cholesterol level and lipid lowering medication in a second model. Analyses involving components of MetS were additionally adjusted for the other components of the syndrome. Interaction terms for sex were explored in all models to investigate effect modification. Interaction between components of MetS in analyses involving these components was also investigated.

To further exclude the possibility that the time between polyneuropathy screening and assessment of cardiometabolic factors influenced the results, we performed a sensitivity analyses in which we repeated the logistic regression analyses into diabetes, impaired fasting glucose and (components of) MetS, restricted to the 730 participants with assessment of cardiometabolic factors on average 2 months before the polyneuropathy screening.

Splines regression for continuous glucose levels was performed in R V.3.2.0, all other analyses were performed in SPSS statistical package, V.21 for Windows (IBM Corp, Armonk, New York, USA).


In total, 1256 participants were included in the analyses. The sample consisted of 685 females (54.5%) and 571 males (45.5%), and the mean age was 70.0 years (see table 1). Type 2 diabetes was present in 175 participants (13.9%) and IFG in 153 participants (12.2%). MetS was present in 659 participants (52.5%). Elevated waist circumference and elevated blood pressure were the most common components, present in 67.5% and 78.0% of participants respectively. Sixty-four participants (5.1%) were diagnosed with a definite polyneuropathy, 92 (7.3%) with a probable polyneuropathy and 218 (17.4%) with a possible polyneuropathy.

Table 1

Population characteristics

Diabetes was associated with definite polyneuropathy (OR 3.01, 95% CI 1.60 to 5.65, see table 2). This association was mostly confined to males and slightly attenuated after adjusting for cardiovascular risk factors. We did not observe an association between IFG and polyneuropathy (OR 1.55, 95% CI 0.70 to 3.44). When investigating fasting glucose levels continuously, there was no association after exclusion of participants using antidiabetic treatment. No associations were found with possible or probable polyneuropathy (table 2).

Table 2

Association of diabetes mellitus and impaired fasting glucose with polyneuropathy

MetS, defined as the presence of at least three of five criteria, also associated with definite polyneuropathy (OR 1.92, 95% CI 1.09 to 3.38, figure 1). This association was stronger when more components of the syndrome were present (at least four components: OR 2.64, 95% 1.40 to 4.98 and all five components OR 3.23, 95% CI 1.22 to 8.55). Of the individual components, elevated waist circumference (OR 2.84, 95% CI 1.35 to 5.99) and elevated triglycerides (OR 2.01, 95% CI 1.11 to 3.62) were both related to definite polyneuropathy (table 3). Reduced HDL-C levels were associated with possible polyneuropathy (OR 1.50, 95% CI 1.01 to 2.24). A similar, although not significant effect estimate was found for probable, but not for definite polyneuropathy. After excluding participants with diabetes a similar pattern of associations, especially between elevated waist circumference and definite polyneuropathy (OR 2.36, 95% CI 1.04 to 5.32), was found. There was no significant difference between genders in these analyses, nor was there significant interaction between different components of the syndrome.

Table 3

Association of MetS components with polyneuropathy

Figure 1

Association of number of MetS components with polyneuropathy. In (A) the association of MetS with polyneuropathy in the total study population is shown. In (B) participants with diabetes have been excluded to investigate whether the association is also present in the population without diabetes. Grey squares represent the presence of MetS (at least 3 components present). The white triangles and black dots represent the presence of at least four and five components respectively. All groups are compared to no MetS (<3 components). MetS, metabolic syndrome; PNP, polyneuropathy.

Restricting the analyses to the 730 participants with assessment of cardiometabolic factors on average 2 months before the polyneuropathy screening yielded even stronger associations, be it with wider CIs (see online supplementary table S1). Diabetes (OR 5.98, 95% CI 2.11 to 16.93) strongly associated with definite polyneuropathy, as did the number of MetS components, elevated waist circumference and elevated triglycerides.

supplementary table S1

Sensitivity analyses of the main analyses in 730 participants with collection of cardiometabolic factors on average 2 months before polyneuropathy screening

In individuals without clinical or neurophysiological suspicion on polyneuropathy, MetS was associated with lower sural SNAP amplitudes in the total and the population without diabetes. Additionally, MetS related to lower peroneal CMAP amplitudes, but this association was present in males only (table 4). When focusing on the separate components of MetS, we found that elevated fasting glucose and reduced HDL-C related to a lower sural SNAP amplitude, especially in males, and that elevated waist circumference related to a lower peroneal CMAP amplitude in males only. There were no associations between any of the metabolic factors and peroneal distal latency or sural sensory nerve conduction velocity.

Table 4

Association of MetS with nerve conduction parameters in males and females categorised as no polyneuropathy


In this population-based study, both diabetes and MetS were strongly associated with the presence of polyneuropathy. Elevated waist circumference and elevated triglycerides were the metabolic factors contributing the most to this association. We also showed that the effect of these factors is independent of the presence of diabetes. Although no association was found between IFG and polyneuropathy, an elevated fasting glucose was related to lower sural SNAP amplitudes in participants without polyneuropathy. Additionally, we showed that MetS in participants without polyneuropathy related to lower sural SNAP amplitudes in males and females and to lower peroneal CMAP amplitudes in males.

The strong association between MetS and polyneuropathy has also been found in some other studies. These studies suggested an association of dyslipidaemia or abdominal obesity with polyneuropathy in cohorts of patients with diabetes20–23 and in case–control studies of patients with idiopathic neuropathy.12 ,24 ,25 ,30 A recent population-based study, similar to ours, also showed an association of MetS with polyneuropathy, with a stronger association in the presence of more components. In contrast to our study, independent associations of specific MetS components were only found with secondary neuropathy outcomes, but not with the primary outcome, which was presence of polyneuropathy.31 Our findings further contribute to the hypothesis that neuropathy in patients with diabetes is not only related to consequences of long-standing hyperglycaemia, but is also influenced by the presence of other metabolic factors, especially abdominal obesity and dyslipidaemia. We also showed that these factors likely are evenly important in persons without diabetes. Potential pathways through which these metabolic factors may lead to neuropathy include oxidative stress, possibly in combination with neuronal and axonal mitochondrial dysfunction, which can lead to nerve injury via chronic metabolic inflammation, insulin resistance and nerve ischaemia.18

Our study is cross-sectional, which makes it difficult to draw firm conclusions about causality. It is possible that persons with polyneuropathy become less active, and consequently gain weight and develop dyslipidaemia. However, we showed that the association with polyneuropathy got stronger when more components of MetS were present, suggesting a dose–response relation. Additionally, we found that MetS, elevated fasting glucose and dyslipidaemia related to lower sural SNAP amplitudes in males and females without any suspicion on polyneuropathy and to lower peroneal CMAP amplitudes in males without polyneuropathy. Impaired function of these nerves, especially the SNAP amplitude of the sural nerves, is a fairly sensitive marker for axonal polyneuropathy.32 Together these findings suggest that these metabolic factors increase the risk to develop polyneuropathy and not the other way around. Still, longitudinal studies are required to further strengthen this hypothesis. The observed differences between males and females in the association of diabetes, pre-diabetes and (components) of MetS with polyneuropathy and peripheral nerve function warrant further investigation.

Our study further showed that diabetes was associated with polyneuropathy. This is a well-established association which has been documented before.33 After adjusting for other cardiovascular risk factors, the association with polyneuropathy remained significant. This suggests that although other metabolic factors contribute to the pathophysiology of polyneuropathy in patients with diabetes, consequences of prolonged hyperglycaemia are probably most important. In contrast with some other studies, we did not find an association of polyneuropathy with IFG. Most studies that suggested an association between pre-diabetes and polyneuropathy lacked a control group,4–8 or did not control for age, which is strongly related to both polyneuropathy and (pre)diabetes.10 ,34 ,35 One controlled study did show an association of IGT and neuropathy,9 while the majority of controlled studies, especially the studies that took the confounding effect of age into account, found no association.11–13 ,15 ,25 ,30 ,31 ,36 Despite these findings, pre-diabetes is still often considered a risk factor for polyneuropathy. Our study does not support this assumption, but it is possible that we lacked sufficient power to show an effect. Moreover, we did not include IGT into the definition of pre-diabetes, since oral glucose tolerance tests were not performed. IGT might be a better measure of pre-diabetes. Additionally, since we included an elderly population, with a mean age of 70 years, we cannot comment on the potential role of pre-diabetes earlier in life.

Our study is one of the very few studies that approaches the question if pre-diabetes and MetS are associated with polyneuropathy using a study design that includes an unselected sample of the middle-aged and elderly general population, without specific sampling techniques for cases or controls. Moreover, we used a rigorous definition of polyneuropathy, which is diagnosed with a protocol that largely resembles clinical practice and also includes nerve conduction studies. However, since we performed our study in a homogeneous, middle-aged to elderly population, we must note that our results may not directly translate to more heterogeneous, or younger populations. Our study also has other limitations. First, for some of the subgroup analyses samples were limited, yielding insufficient power to show small associations. Second, due to logistics of the study, in 526 participants the presence of diabetes and MetS was assessed ∼4 years before the polyneuropathy screening. It is possible that participants developed incident (pre)diabetes during this period and lipid levels may have changed since the blood samples were collected. Therefore, we adjusted all analyses for the time between the assessment of cardiometabolic factors and the polyneuropathy screening to take this into account. Additionally, we performed a sensitivity analyses while excluding these 526 participants, which suggested that the results in the main analyses might even be an underestimation of the true effect. Third, we did not perform an oral glucose tolerance test. Possibly, postload hyperglycaemia in the postprandial phase plays a key role in the pathogenesis of complications such as polyneuropathy.26 Finally, although our screening approach is as close to clinical practice as possible, it has not been validated and it is possible that some participants were misclassified as having polyneuropathy, while instead abnormality was due to focal neuropathies, radiculopathy or non-specific symptoms attributable to osteoarthritis or other locomotor problems. This especially concerns the possible and probable polyneuropathy categories, which might explain the inconsistent findings in these groups. It is reasonable to assume that misclassification in the definite polyneuropathy category is very minimal, because we combined symptoms, neurological examination and nerve conduction studies to make this categorisation. The association with definite polyneuropathy thus provides an accurate estimate of the effect.

In conclusion, this population-based study showed that cardiometabolic disturbances, like abdominal obesity and dyslipidaemia, are strongly related to the presence of polyneuropathy and impaired peripheral nerve function in participants without polyneuropathy, irrespective of the presence of diabetes. Therefore, screening and optimal control of these risk factors may be warranted. Whether this also reduces neuropathic symptoms, or may even prevent the development or progression of polyneuropathy needs to be further evaluated in longitudinal studies.


View Abstract


  • Contributors RH, JD, OHF, AH, MAI and PAvD have made substantial intellectual contributions to conceptualisation and design of the study. RH, JD, SL and AD were involved in acquisition of data. RH, JD, MAI and PAvD were involved in analysis and interpretation of data. RH drafted the article. JD, SL, AD, OHF, AH, MAI and PAvD revised the manuscript critically for important intellectual content. All authors gave final approval of the version to be published. RH has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

  • Funding The Rotterdam Study is funded by Erasmus Medical Centre and Erasmus University, Rotterdam, Netherlands Organization for the Health Research and Development (ZonMw), the Research Institute for Diseases in the Elderly (RIDE), the Ministry of Education, Culture and Science, the Ministry for Health, Welfare and Sports, the European Commission (DG XII), and the Municipality of Rotterdam. The current study was funded by the Prinses Beatrix Spierfonds for neuromuscular diseases (grant number W.OR12-08) and a grant from the Janivo Foundation.

  • Competing interests AD is supported by Netherlands Organization for Scientific Research (NWO) grant (veni, 916.12.154) and the EUR Fellowship. OHF works in ErasmusAGE, a centre for ageing research across the life course, which is funded by Nestlé Nutrition (Nestec); Metagenics; and AXA. AH received grants from the Netherlands Organization for Scientific Research, the Netherlands Genomics Initiative, the Netherlands Ministry of Health and the European Commission; and remuneration as editor of the European Journal of Epidemiology. MAI received grants from the Netherlands Heart Foundation (2009B102 and 2012T008), Netherlands Organization for Health Research and Development (ZonMW: 916.13.054), Internationaal Parkinson Fonds, and Internationale Stichting Alzheimer Onderzoek (number 12533). PAvD received a grant from the Prinses Beatrix Spierfonds for neuromuscular diseases (grant number W.OR12-08) and a grant from the Janivo Foundation to conduct this study. Other, non-related, grants include grants from the Prinses Beatrix Spierfonds, Baxalta, Grifols and from Sanguin Blood Supply (for the conduct of prospective studies into treatment of GBS, CIDP and Pompe Disease).

  • Ethics approval The Medical Ethics Committee of the Erasmus Medical Centre and the Ministry of Health, Welfare and Sport of the Netherlands.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.