SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

Background  Few data are available on the incidence, risk factors and contamination pathways involved in acute indigenous hepatitis E in developed countries.

Aims  To draw up an overall picture of hepatitis E cases, to confirm whether or not the majority of the cases were indigenous and to attempt to identify the risk factors and contamination pathways involved in hepatitis E.

Methods  This study was performed in the framework of a national network (ANGH) including 96 participating centres. The 19 centres with at least one case of acute HEV reported a total number of 53 cases.

Results  A decreasing South-to-North geographic gradient was observed. A nonspecific clinical profile was observed in many cases. Acute hepatitis E was of indigenous origin in 90% of the patients. The most relevant and/or frequent possible risk factors among the 47 indigenous metropolitan cases were water consumption from a personal water supply, uncooked shellfish consumption and the recent acquisition of a pet pig.

Conclusions  This national survey confirmed that acute indigenous hepatitis E is an emerging disease in developed countries such as France, and suggests that various risk factors are responsible for acute indigenous hepatitis E contamination in non-endemic countries.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

The hepatitis E virus (HEV) is an enterically transmitted RNA virus responsible for large waterborne epidemics of acute hepatitis in endemic regions. Although acute hepatitis E is relatively rare in developed countries, an increasingly large number of indigenous cases have been recently reported in Western Europe, especially in the UK,1 the Netherlands2 and France.3 Until now, few studies on the incidence of acute hepatitis E have been available, and no complete national surveys on acute hepatitis E have ever been conducted to our knowledge in developed countries. Although water is a well-known vector in endemic regions, the potential causes of indigenous contamination in non-endemic countries such as France still remain to be elucidated.

The aims of this retrospective French survey were to draw up an overall picture of recent acute hepatitis E cases, to confirm whether or not a majority of these cases were indigenous, to document the clinical and epidemiological features of HEV infection, and to identify the risk factors involved in HEV exposure and the contamination pathways responsible for indigenous acute hepatitis E.

Patients and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

An email was sent to 96 French General Hospitals affiliated to the ANGH (Association Nationale des Hôpitaux Généraux) in January 2007, with a view to collecting information about all the cases of acute hepatitis E observed at the respective gastroenterological departments since January 2004. All practitioners dealing with acute hepatitis E patients had to complete an anonymous questionnaire in which they were asked to assess the epidemiological, clinical and biological characteristics of the acute infection and the risk factors possibly involved in HEV contamination.

Case definition

Acute hepatitis E diagnosis was based on alanine aminotransferase (ALT) levels more than 10 times higher than the normal value and the presence of increasing anti-HEV IgG titres and/or the presence of anti-HEV IgM and/or HEV RNA in the serum and/or stools. All the patients were found to be negative in the following tests: IgM antibodies to hepatitis A virus; AgHBs, anti-HBc IgM and DNA to hepatitis B (Cobas Amplicor HBV Monitor; Roche diagnostics, Branchburg, NJ, USA); antibodies and RNA to hepatitis C (Cobas Amplicor HCV Monitor; Roche diagnostics); IgM antibodies for Cytomegalovirus (Vidas CMV IgG and IgM; BioMérieux, Lyon, France) and Epstein Barr Virus (ImmunoDOT EBV MONO G and M; BioMédical Diagnostics, Marne la Vallée, France). Antinuclear (except for two patients with low positive titres of 1/40), antismooth muscle and anti-LKM-1 antibodies were also negative in all the patients.

Hepatitis E testing

Serum anti HEV IgG and IgM were measured by performing an enzyme immunoassay (HEV ELISA; Genelabs Diagnotics, St Ingbert, Germany): the same method was used by all the referral laboratories. The positivity of the serum antibodies was defined by an optical density > cut-value. HEV RNA was detected in the serum samples using reverse transcription-PCR for open reading frame (ORF-2). The amplified fragments were sequenced directly and aligned with those of other HEV strains (Centre National de Référence, Hôpital du Val de Grâce, Paris; Laboratoire de Virologie, Hôpital Purpan, Toulouse, France).

Statistical analysis

The chi-squared test was used to compare paired proportions. Cochran–Mantel–Haenszel test was used to compare multiple proportions. Normally distributed data were compared using either the parametric chi-squared test or Student’s t-test. Non-normally distributed data were compared using the nonparametric Wilcoxon test. The Shapiro–Wilks test was used to test normality. Significance level was defined as P < 0.05, based on two-sided tests.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

Geographic distribution of the ANGH centres

Ninety-four of the 96 French General Hospitals participating were located in metropolitan France (53% in the North and 47% in the South) and two in overseas departments. The response rate from the hospitals in metropolitan France and the two overseas departments was 78% (75/96), including 56 centres without any cases of acute hepatitis E and 19 centres with at least one case. The response rates obtained from the South and North of France were nearly identical, amounting to 78% in the north and 77% in the south (Figure 1a). The response rate from the hospitals in the two overseas departments was 100%: the first one (in French Guiana) reported one case of acute hepatitis E, whereas the second one reported that no cases had been noted (Island of Reunion). Apart from the overseas departments, the negative response rate (no cases of acute hepatitis E) was 90% (39/43 centres) in the North and 53% (16/30 centres) in the South (P = 0.0009). Conversely, the positive response rate (at least one case of acute HEV) was 10% (4/43 centres) in the North and 47% (14/30 centres) in the South, which indicates that the incidence of acute hepatitis E obeys a decreasing South-to-North geographic gradient. Including the overseas departments, the 19 centres giving positive responses reported a total number of 53 cases (12, 10, 9, 5, 2 and 1 cases in 1, 1, 1, 1, 2 and 13 hospitals, respectively). All the three centres reporting the largest number of cases were located in the South of France: the first two were in South-Eastern France, in the Var (12 and 9 cases, respectively) and the third one in the South-West, in Pyrénées-Atlantiques (10 cases). The existence of a decreasing South-to-North geographic gradient was confirmed by the fact that 44 cases were reported by 14 centres in the South, whereas only eight cases were reported by four centres in the North (Figure 1b).

image

Figure 1.  (a) Response rate reported by hospitals in southern and northern France. (b) Number of cases of acute hepatitis E reported by hospitals in southern and northern France.

Download figure to PowerPoint

The number of cases increased during the 3-year retrospective study period: 10 cases occurred in 2004, 14 cases in 2005 and 24 cases in 2006. Five prospective cases were reported during the first 6 months of the year 2007.

Patients’ epidemiological, clinical and biological features

Epidemiological data were available on all 53 patients: these data showed that 47 patients had not travelled outside Europe or had any contact or relationships with persons who had been to an endemic country 6 months or less before the onset of the symptoms. However, six patients had travelled outside Europe to an HEV-endemic country, including three to Pakistan, one to Algeria, one to Morocco and the last one to Uzbekistan and Tajikistan. The last patient, who had put the health recommendations into practice during the entire journey, had a 3f genotype, corresponding to French indigenous strains. This brought the proportion of indigenous cases of acute hepatitis E up to 90% (48/53 patients).

The mean age (±s.d.) of the 53 patients was 56.1 ± 15.6 years (36 males, 17 females; Table 1). Forty-nine were of Caucasian origin, three were from Asia and one from the Middle East (Algeria). Twenty-seven patients lived in rural areas and 17 had retired (Table 1). High rates of alcohol consumption (>40 g/day) were noted in the case of 11 patients, but none of the 53 patients was found to have chronic liver disease.

Table 1.   Comparison of the epidemiological, clinical, biological parameters and the risk factors of hepatitis E virus contamination between the serological and the sero-virological groups of patients
  Total (n = 53)Serologic group of HEV patients (n = 39)Sero-viraemic group of HEV patients (n = 14)P
  1. HEV, hepatitis E virus; AST, aspartate aminotransferase; ALT, alanine aminotransferase.

Age (mean ± s.d.)56.1 ± 15.652.8 ± 14.965.5 ± 13.80.001
Male gender n (%)36 (65)27 (69)9 (64)0.99
Urban area n (%)26 (49)19 (49)7 (50)0.80
Drug-induced liver injury consumption n (%)18 (34)10 (26)8 (57)0.08
Jaundice as circumstance of discovery n (%)25 (47)17 (44)8 (57)0.74
Jaundice n (%)36 (68)27 (69)9 (64)0.99
Duration of jaundice (days, mean ± s.d.)17.0 ± 9.815.3 ± 11.015.1 ± 9.60.90
Pruritis n (%)14 (26)12 (31)2 (14)0.38
Abdominal pain n (%)16 (30)13 (33)3 (21)0.62
Vomiting n (%)9 (17)9 (23)0.12
Diarrhoea n (%)11 (21)8 (21)3 (21)0.70
HPM n (%)14 (26)8 (21)6 (43)0.21
SPM n (%)2 (4)2 (5)0.98
Hospitalization n (%)32 (60)22 (56)10 (71)0.51
Hospitalization (day number)10.1 ± 7.79.3 ± 3.512.0 ± 12.90.54
AST (N)42.6 ± 41.439.5 ± 34.751.1 ± 56.70.98
ALT (N)53.1 ± 38.954.4 ± 41.349.7 ± 32.60.83
Total bilirubin (μmol/L)150.5 ± 150.6128.8 ± 114.1211.0 ± 214.30.35
Alkaline phosphatase (IU/L)233.8 ± 104.5241.7 ± 120.6216.9 ± 56.40.74
Prothrombin index (%)83.1 ± 18.687.4 ± 14.371.6 ± 24.30.02
Risk factors n (%)   0.61
 No personal water or shellfish consumption26 (49)17 (44)9 (64) 
 Shellfish consumption12 (23)10 (26)2 (14) 
 Personal water consumption11 (21)8 (21)3 (22) 
 Personal water and shellfish consumption4 (7)4 (9) 

The circumstance of discovery of the acute hepatitis E was nonspecific in many of the cases diagnosed (26 cases, 49%): the symptoms included flu-like symptoms (nine cases), abdominal pain with vomiting and diarrhoea (nine cases), arthralgia with myalgia but no fever (four cases), isolated asthenia (three cases) and an isolated skin rash in the last case. Twenty-five patients (47%) were referred for jaundice either alone (21 cases) or associated with fever, flue-like symptoms or vomiting (four cases; Table 1). The last two patients were asymptomatic; in the first case, acute HEV was discovered as the result of cardiology investigations and in the second case, as the result of a blood test carried out prior to chemotherapy. During the follow-up period, the patients’ clinical features ranged from no symptoms to fulminant hepatic failure, i.e. with encephalopathy and coagulation disorders. Thirty-two patients required hospitalization, mostly for an aetiologic assessment of the jaundice (Table 1). Two of the 53 patients developed signs of fulminant liver failure and died. The first one, a 76-year-old man known to have a high rate of alcohol consumption (60 g/day) died 34 days after the day of the diagnosis. A postmortem liver biopsy showed the existence of acute cholestatic hepatitis associated with cholangitic features and a bridging fibrosis with no cirrhosis compatible with an alcoholic origin, but without the features typical of alcoholic hepatitis. This patient had not been exposed to any risk factors obviously associated with HEV contamination. The second patient, a 59-year-old man died 52 days after the diagnosis, subsequent to two liver transplantations. A liver biopsy on the native liver brought to light histological features typical of fulminant hepatitis without any specificity and ruled out the possibility of an underlying chronic liver disease. The two patients had not travelled outside France during the previous 6 months. In addition, they had a positive HEV RNA in the serum and stools, whereas the phylogenetic analysis showed that the strains involved belonged to the 3f genotype, thus confirming that an indigenous source of contamination was responsible.

The main biological features of the 53 patients are given in  Table 1. All the patients showed biochemical evidence of hepatitis, including ALT levels more than 10 times higher than the normal values. Three patients had a prothrombin index of <50%. Two of them had ALT levels 51 and 109 times above the normal values and died as the result of fulminant liver failure. The third one, who had a prothrombin index of 45% and an ALT level 95 times higher than the normal values, recovered spontaneously. Eleven patients had normal total bilirubin levels, whereas all but two had elevated alkaline phosphatase levels.

Patients’ serological and virological characteristics

HEV RNA was detected in the serum and/or stools of 14 of the 30 patients tested (47%). In the sero-viraemic group of 14 patients, the anti-HEV IgG tests were positive in the case of 10 patients and negative in that of the other four patients, whereas the anti-HEV IgM tests were positive in all the patients tested (six of six patients). The genomic sequences of all 14 viraemic patients were available, and genotype 3f was identified in all these cases, which indicates that an indigenous HEV contamination process was involved. None of them had visited an endemic country 6 months or less before the first symptoms occurred. The HEV diagnosis on the remaining 39 patients was based on an increasing anti-HEV IgG titre or a strong reactivity to IgM, or dual IgG and IgM positivity. Positive anti-HEV IgG or anti-HEV IgM or anti-HEV IgG and IgM antibodies were obtained in 16, 3 and 20 patients, respectively. The presence of HEV RNA was investigated in 16 of the 39 patients, but the results were negative in all the patients tested. Comparisons were made between the serological and sero-viraemic groups of patients (Table 1). A significant difference was found to exist between the two groups only in terms of age and the prothrombin index: the mean age was higher and the prothrombin index was lower in the viraemic group (mean age: 65.5 ± 13.8 vs. 52.8 ± 14.9, P = 0.001; mean prothrombin index: 71.6 ± 24.3% vs. 87.4 ± 14.3%, P = 0.02). All the other epidemiological, clinical and biological parameters and the risk factors possibly responsible for HEV contamination tested in this study were similar in the two groups. It is worth noting in addition that a lower rate of patients with drug-induced liver injury occurred in the serological group (26% vs. 57%, P = 0.08) (Table 1).

Rate of acute indigenous hepatitis E and risk factors possibly associated with the infection

Forty-eight of the 53 patients had acute indigenous hepatitis E. The patient in French Guiana was a 9-year-old girl who had not been exposed to any obvious HEV contamination risk factors. The indigenous metropolitan cases amounted to 89%. The most relevant and/or frequent possible risk factors responsible for the 47 indigenous metropolitan cases were water consumption from a spring or private well or a nearby river (15 cases), uncooked shellfish consumption (15 cases) and the recent acquisition of a pet pig (one case) (Table 2). Geographical comparisons between the most relevant and/or frequent possible risk factors were made to explain the decreasing South-to-North geographic gradient. No differences were found to exist between the two geographical areas in terms of either personal water consumption or shellfish consumption (Table 2). Nor were any particular epidemiological factors identified in the group of 31 patients involved in the various small-scale outbreaks in comparison with the group consisting of the 16 other indigenous metropolitan patients (Table 3).

Table 2.   Combination and separate comparisons of the major metropolitan indigenous possible risk factors between North and South of France
 SouthNorthTotal*
  1. P = 0.40, Cochran–Mantel–Haenszel test.

  2. P = 0.64, Fisher test.

  3. * Five non-indigenous cases and one nonmetropolitan case of acute hepatitis E were excluded of the analysis.

Combination comparison
 No personal water or shellfish consumption18321
 Personal water and/or shellfish consumption24226
 Total42547
Separate comparison
 No personal water or shellfish consumption18321
 Shellfish consumption10111
 Personal water consumption11011
 Personal water and shellfish consumption314
 Total42547
Table 3.   Combination and separate comparisons of major metropolitan indigenous possible risk factors between the cases of the three small-scale outbreak and the other cases
 Cases of small-scale outbreak (n = 31) Other cases (n = 16) Total* (n = 47)
  1. P = 0.49, Cochran–Mantel–Haenszel test.

  2. P = 0.55, Fisher test.

  3. * Five non-indigenous cases and one nonmetropolitan case of acute hepatitis E were excluded of the analysis.

Combination comparison
 No personal water or shellfish consumption15621
 Personal water and/or shellfish consumption161026
 Total311647
Separate comparison
 No personal water or shellfish consumption15621
 Shellfish consumption6511
 Personal water consumption8311
 Personal water and shellfish consumption224
 Total311647

Seven patients were accustomed to drinking water either directly from a spring (two patients) or from a private well (five patients), whereas eight consumed vegetables from their own vegetable garden, which was irrigated with water from a river or a private well. All these 15 patients, but one, were reported by centres located in the South of France. The 59-year-old male patient who died after two liver transplantations had drunk water from a spring during a bicycle ride 35 days before the onset of the symptoms. In addition, this patient had consumed wild boar 29 days prior to his admission. However, no HEV RNA was detected in the remaining frozen portion of wild boar. The patient’s wife, who had eaten the same meal, had positive anti-HEV IgG and IgM but negative HEV RNA. The fact that she had not drunk any spring water as her husband had suggests that the wild boar may have been responsible for the infection. However, HEV RNA was detected once in the 35 water samples collected at different points in the spring from which the patient had drunk. Unfortunately, no HEV genotyping was carried out, as the fragment amplified by RT-qPCR was too short to be sequenced.

Uncooked mussels and/or oysters were consumed by 15 patients 2–9 weeks before the onset of the symptoms. The origin of the shellfish was the Mediterranean in 13 cases (they came from a grocery shop in five cases and were gathered personally in eight cases) and the Atlantic Ocean in the other two cases (the shellfish came from a grocery shop in both cases). The consumption of shellfish was reported by most of the centres located in the south of France (13 of the 15 centres).

One of the 47 patients had acquired a 3-month pet pig 2 months before the onset of the symptoms. A genotype 3f HEV strain was identified in both the pig and its owner. The isolates obtained from the patient and his pet pig showed 93% nucleotide homology and 98% amino acid homology.4 No other possible source of HEV contamination of the owner was identified. It can therefore be reasonably assumed that the owner’s acute hepatitis E was directly transmitted during regular contacts between the owner and his pet pig.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

During the last decade, an increasingly large number of cases of acute hepatitis E have been reported in patients who have not recently travelled to endemic countries. Most of the studies available so far, which were performed in the UK,1, 5 the Netherlands,2 France3, 6 and the US,7 have included only a few indigenous cases of acute hepatitis E. Until now, no national surveys have been conducted in non-endemic countries on acute hepatitis E during a specific period of time. The main aim of this retrospective study was therefore to draw an overall picture of the acute cases of hepatitis E recorded since January 2004 in France. This study was performed in the framework of a national network (ANGH) that included 96 participating centres. A total number of 53 acute cases of hepatitis E were reported to have occurred between January 2004 and June 2007. Acute hepatitis E was found to show a decreasing South-to-North gradient in this country, as a majority of the positive responses were from hospitals in the South (47% vs. 10%) and most of the negative responses were from those in the North (90% vs. 53%), although the response rates from the South and North were similar. Several of the cases reported in this study occurred in roughly the same area in the South of France (12 and 9 cases occurred in the Var and 10 cases in Pyrénées-Atlantiques), but no particular epidemiological factor was identified between the group of 31 patients involved in the three different small-scale outbreaks and the remaining group consisting of the 16 other indigenous patients.

The presentation at diagnosis was atypical in nearly half the patients: symptoms such as flu-like symptoms or vomiting with diarrhoea or arthralgia with myalgia but no jaundice occurred. In two asymptomatic patients, acute hepatitis E was diagnosed as the result of biological investigations carried out because these patients were being treated for other diseases. On the other hand, there is a high prevalence of the anti-HEV antibodies (ranging from 3% to 18%, among blood donors in Europe and the US) underlying this increasingly frequent hepatic viral disease.8–10 The fact that this disease is often not diagnosed may be not only because of the high rate of asymptomatic forms of acute hepatitis E, which are usually estimated to account for about 90%,9 but also because of the fact that a large number of atypical symptomatic patients with acute hepatitis E do not undergo HEV tests. As the ALT levels of all the patients in this study were more than 10 times higher than the normal values, routine liver tests should have been carried out on patients with persistent flu-like symptoms or vomiting with diarrhoea or arthralgia with myalgia, and these tests should have been completed by serological and virological HEV tests on patients with significant levels of hypertransaminasaemia.

Acute hepatitis E is known to be responsible for fulminant hepatic failure in endemic areas. In addition, the overall mortality rate because of acute hepatitis E complicated with fulminant liver failure in endemic regions has been reported to range between 0.5% and 4%.10 However, the incidence of fatal fulminant hepatitis secondary to acute hepatitis E has not yet been established in non-endemic regions such as the Western countries.11 In this study, the overall fatality rate of indigenous fulminant hepatitis E was 4.2% (two of 48 patients), which was similar to the rate usually reported in endemic countries.10 In the two lethal cases included in this study, the onset of fulminant hepatitis was because of genotype 3 HEV, which may therefore be as virulent as genotypes 1 and 2. Among the risk factors associated with fulminant hepatitis E, active alcohol abuse or chronic alcoholic liver disease has been reported.12 The fact that one of the two patients with fatal fulminant hepatitis had chronic alcohol consumption and alcoholic liver features strongly suggested that an underlying chronic alcohol liver disease was a risk factor contributing to fulminant liver failure in the course of HEV infection. On the other hand, the second patient had none of the risk factors usually associated with chronic hepatitis and showed no chronic hepatic features at liver biopsy. HEV should therefore be recognized as a potential cause of non-A, non-B fulminant hepatitis in industrialized countries, even in patients with no chronic liver disease and no excessively high alcohol consumption rates.

In endemic countries, the main routes of HEV contamination are the ingestion of contaminated water and that of soiled food products.13 Contamination by water is not yet an established pathway in non-endemic countries, as the viral sequences responsible have not been identified so far in water supply samples. However, HEV RNA has been amplified from environmental water samples, especially from urban wastewater in several European countries, which suggests that HEV is circulating in environments of this kind.14, 15 In addition, as shellfish consumption was found to be possibly one of the risk factors involved in acute hepatitis E, it is now suspected that exposure to sewage or wastewater may contribute to HEV endemicity.5 Lastly, the HEV genotype 3 was recently detected in Corbicula japonica from Japanese rivers, which supports the idea that HEV may also contaminate river waters.16 The epidemiological results obtained in this study suggest that several HEV transmission routes may contribute to maintaining the endemicity of acute hepatitis E in France. The two most relevant and frequent risk factors involved in contamination with acute hepatitis E were found to be the direct or indirect consumption of water from a private well or a nearby river, and shellfish consumption. Because of lack of sanitation, the water from these various private supplies may have been regularly contaminated by wastewater or sewage, thus exposing the owners either directly or indirectly to contamination via vegetables irrigated with soiled water. Although no definite evidence is available so far that HEV transmission can occur via the consumption of shellfish and little is known about HEV resistance in environments with high salt levels, uncooked mussels and/or oysters were found to have been consumed by one–third of the patients included in this survey.17 Further studies focusing on patients’ eating and drinking habits are now required to confirm whether shellfish consumption and the use of personal water supplies are significantly correlated with the risk of acute hepatitis E.

The main limitations of this observational survey are its retrospective nature and the relatively heterogeneous patient population studied. In fact, the definition of acute hepatitis E used in this study was based on two-fold serological and sero-virological criteria for acute HEV diagnosis. However, after statistical analysis, all the serological and virological parameters tested between the two groups of patients were similar, except for age and prothrombin index, which suggests that there were no major differences between the two groups. Moreover, this suggests that the serological diagnostic criteria used to define HEV infection in this study may provide a valid means of diagnosing acute hepatitis E. It is worth noting that a higher rate of patients with drug-induced liver injury was found to occur in the virological group of patients. This shows that it is necessary to test patients’ HEV RNA to avoid wrongly diagnosing drug-induced liver injury, as well as to check the accuracy of the various serological tests used to diagnose acute HEV in the context of elevated ALT levels. Although hepatitis E serological investigations have their limitations, diagnosis of hepatitis E is often based in practice on the presence of specific antibodies, as standard biological criteria for acute HEV diagnosis still remain to be approved by the scientific community. The incidence of HEV among all cases of acute hepatitis needs to be assessed, as it was not possible to address this important issue in this study.

In conclusion, the results of this national survey confirm that acute indigenous hepatitis E is an emerging disease in developed countries such as France. In addition, the mortality rate recorded in this study, which was similar to those frequently reported in developing countries, shows the potential severity of the HEV genotype 3. Lastly, this study supports the existence of several risk factors possibly responsible for contamination, which may partly explain the outbreaks of acute indigenous hepatitis E currently occurring in non-endemic countries.

Acknowledgement

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

Declaration of personal and funding interests: None.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References