Impact of first-wave COronaVIrus disease 2019 infection in patients on haemoDIALysis in Alsace: the observational COVIDIAL study

Abstract Background There are only scarce data regarding the presentation, incidence, severity and outcomes of coronavirus disease 2019 (COVID-19) in patients undergoing long-term haemodialysis (HD). A prospective observational study was conducted in eight HD facilities in Alsace, France, to identify clinical characteristics of HD patients with COVID-19 and to assess the determinants of the risk of death. Methods All HD patients tested positive for COVID-19 from 5 March to 28 April 2020 were included. Collected data included patient characteristics, clinical features at diagnosis, laboratory data, treatments and outcomes. Results Among 1346 HD patients, 123 tested positive for COVID-19. Patients had a median age of 77 years (interquartile range 66–83), with a high number of comorbidities (3.2 ± 1.6 per patient). Symptoms were compatible in 63% of patients. Asthenia (77%), diarrhoea (34%) and anorexia (32%) were frequent at diagnosis. The delay between the onset of symptoms and diagnosis, death or complete recovery was 2 (0–5), 7 (4–11) and 32 (26.5–35) days, respectively. Treatment, including lopinavir/ritonavir, hydroxychloroquine and corticosteroids, was administered in 23% of patients. The median C-reactive protein (CRP) and lymphocyte count at diagnosis was 55 mg/L (IQR 25–106) and 690 Ly/µL (IQR 450–960), respectively. The case fatality rate was 24% and determinants associated with the risk of death were body temperature {hazard ratio [HR] 1.96 [95% confidence interval (CI) 1.11–3.44]; P = 0.02} and CRP at diagnosis [HR 1.01 (95% CI 1.005–1.017); P < 0.0001]. Conclusions HD patients were found to be at high risk of developing COVID-19 and exhibited a high rate of mortality. While patients presented severe forms of the disease, they often displayed atypical symptoms, with the CRP level being highly associated with the risk of death.

on the front line of COVID-19 during the first French wave of the epidemic.

M A T E R I A L S A N D M E T H O D S
Data were collected from 5 March, the first confirmed dialysis case of COVID-19, to 28 April 2020 from adult (!18 years old) patients undergoing long-term HD and testing positive for COVID-19. A positive test for COVID-19 was defined by either a positive reverse transcription-polymerase chain reaction (RT-PCR) test on a nasopharyngeal swab or compatible radiologic findings on a low-dose computed tomography (LD-CT) scan. Eight HD facilities in three cities participated in the study: Strasbourg [Clinique St Anne, Association pour l'utilisation du rein artificiel en Alsace (AURAL) Bergson, Hôpitaux Universitaires de Strasbourg], Colmar (AURAL Colmar, Hôpitaux Civils de Colmar) and Mulhouse (Hôpital Emile Muller, Diaverum private centre, AURAL Mulhouse). These eight facilities provided HD treatment for 1346 patients in Alsace.
The collected data were as follows: (i) General characteristics and demographics.
(ii) Comorbidities, including age >75 years, functional disabilities, obesity (body mass index >30 kg/m 2 ), history of cancer (<5 years), current immunosuppression (history of transplantation, autoimmune disease, chemotherapy <5 years), chronic respiratory disease (e.g. chronic obstructive pulmonary disease, sleep apnoea treated with continuous positive airway pressure, need for long-term oxygen therapy), stroke, peripheral arterial disease, ischaemic heart disease and diabetes. A histroy of high blood pressure was not taken into account, due to its nearly ubiquitous prevalence in this population. (iii) Diagnostic methods: RT-PCR on a nasopharyngeal swab, LD-CT scan indicating the extent of the impairment (minimal if <10% of the pulmonary parenchyma, mild if 10-25%, mild-severe if 25 What this study adds?
• HD patients often displayed an atypical presentation, such as gastrointestinal involvement, less hyperthermia, a much more severe form of COVID-19 and delayed viral clearance. • the incidence and mortality rate of COVID-19 were much higher in HD patients than in the general population. • C-reactive protein and body temperature at admission were predictive of the risk of death.
What impact may this have on practice or policy?
• screening indications should be comprehensive in this population to ensure adequate isolation and repeated reverse transcription-polymerase chain reaction testing would be advisable before ending isolation in COVID-19 HD patients. • we suggest strict implementation of 'barrier gestures' in these patients in whom strict containment cannot be secured due to the regular need for dialysis. • because of the substantial vulnerability of our patients, the prioritized distribution of personal protective equipment in a period of relative shortage or more extensive indications for hospitalization should be considered.
consecutive negative RT-PCRs on nasopharyngeal swabs after a minimum time of 24 days after the onset of symptoms and with at least 48 h without symptoms.

Statistical analysis
Quantitative data are described as median and interquartile range (IQR) or mean and standard deviation (SD) according to the normality of the distribution. Distribution normality was tested graphically and with the Shapiro-Wilk test. Comparisons were performed using the Student's t-test or the Mann-Whitney test, as appropriate. Qualitative data are described according to the frequency for each modality and compared with Fisher's exact test.
Risk factors for death were analysed with a multivariate Cox model including variables with a P-value <0.2 in the univariate comparison between deceased and surviving patients. Statistical significance was defined as P < 0.05. All statistical analyses were performed with STATA/MP 13.1 (StataCorp, College Station, TX, USA).

R E S U L T S
Between 5 March and 28 April, 123 patients undergoing longterm haemodialysis (HD) were infected with COVID-19 (Mulhouse, n ¼ 64; Strasbourg, n ¼ 45; Colmar, n ¼ 14), representing 9.1% of all patients undergoing long-term HD in our centres (n ¼ 1346). The cumulative and incident cases in included HD facilities are presented in Supplementary data, Appendix 1.
Patient characteristics, demographics and comorbidities at the time of diagnosis are summarized in Table 1. Of note, a history of high blood pressure was found in 97.5% of patients.

Diagnosis
The median time between first symptoms and diagnosis was 2 days (IQR 0-5). The origin of the contamination was unknown in the majority of cases (62.5%). Otherwise, the contamination mostly occurred in patients living in institutions (40%) or intrafamilial settings (29%). Of note, 21% of infections were due to nosocomial contamination (already in a hospital unit or rehabilitation unit for at least 7 days for another reason before the onset of symptoms) and 10% were related to an evangelist gathering cluster in Mulhouse. There was no evidence of either vertical or horizontal transmission within the HD units.
The clinical features and severity of the disease at the time of diagnosis are presented in Table 1. Symptoms were found to be compatible with COVID-19 infection (a combination of at least two symptoms among fever, cough and dyspnoea) in 63% of patients.
The primary method of diagnosis was RT-PCR of a nasopharyngeal swab for 112 patients, which was contributive in 88%. LD-CT scan was performed on 64 patients and was contributive in 88% and confirmed the diagnosis in 13 patients that had an initially negative RT-PCR. The extent of the impairment (available for 49/64 patients) on LD-CT scan was minimal in 13 (26.5%), mild in 21 (43%), mild-severe in 6 (12%), severe in 7 (14.5%) and critical in 2 (4%) patients.
Laboratory findings at the time of diagnosis are provided in Table 2. Of note, the median maximal CRP level was 112 mg/L (IQR 56-203) and was reached after a median delay of 7 days (IQR 4-12) after the onset of symptoms.

Treatment
Sixty per cent of the patients required oxygen support therapy at the time of diagnosis. When oxygen therapy was needed, the median time between the onset of symptoms or time of diagnosis and maximal oxygen therapy was 8 days (IQR 3-11) and 2 days (IQR 0-8), respectively (data available in 29 patients).  (9) CIN, chronic interstitial nephropathy; BMI, body mass index; SpO2: pulsatile saturation in oxygen; ENT, ear, nose and throat.
Antibiotic therapy was initiated in 77% of patients after a median delay of 0 days (IQR À1-1) and 3 days (IQR 1-5) after diagnosis and symptoms onset, respectively. A putative antiviral therapy was initiated at the discretion of the physician in 23% of patients after a median delay of 2 days (IQR 1.5-3.5) and 4.5 days (IQR 3-9.5) after diagnosis and symptoms onset, respectively. This treatment was discontinued in 4/23 patients due to intolerance or medical contraindication. The main reasons for non-prescription of a specific therapy were the clinician's choice and the absence of guidelines in HD patients (65%).
An overview of treatments is provided in Table 3.

Outcomes at the time of writing
Hospitalization was required in 71% of patients after a median time of 2 days (IQR 0-5) after symptoms onset, seven of whom were transferred to the ICU. The median hospital stay was 9 days (IQR [4][5][6][7][8][9][10][11][12][13][14]. At the time of writing, follow-up data on 117/123 patients were available: 39 (33%) were discharged from the hospital, 11 (9%) remained hospitalized, 27 (23%) had only ambulatory care, 11 (9%) were transferred to a rehabilitation centre and 29 (24%) had died. Complete recovery was observed in 30 (24%) patients after a median interval of 32 days (IQR 26.5-35) after the onset of symptoms. In survivors at Day 21 (n ¼ 63), 52% had a sustained positive RT-PCR on a nasopharyngeal swab or had no control exam until this time. Supplementary data, Appendix 2 depicts the time interval until complete recovery, death or the end of the observation time. Of note, a mean weight loss of À2.4 6 2.7 kg (SD À13-1.5) was observed.

Deceased patients
Twenty-nine (24%) patients died from COVID-19, three of whom died in the ICU. The median time between symptoms onset and death was 7 days (IQR 4-11). The Kaplan-Meier survival curve is presented in Figure 1 . A comparison of deceased and surviving patients is shown in Table 4.
At diagnosis, deceased patients presented a significantly older age, a more severe presentation, a higher CRP level, a higher body temperature and more often required oxygen therapy and at a higher flow. Patients who died more frequently had a typical presentation of the disease at diagnosis compared with survivors (85% versus 57%; P ¼ 0.011), including dyspnoea (70% versus 45%; P ¼ 0.027). Clinical features at diagnosis were critical in 31% (compared with 2% in survivors) and moderate in 31% (compared with 58% in survivors). Of note, compared with survivors, deceased patients more frequently had diabetic kidney disease (48% versus 32%) or an autosomal polycystic kidney disease (APKD) (15% versus 7%). None of the specific comorbidities was associated with the risk of death, while only a history of ischaemic heart disease tended to be  The results of the multivariate analysis are presented in Table 5, while the Kaplan-Meier survival curves according to CRP quartile are shown in Figure 1.

D I S C U S S I O N
The rapid outbreak of the COVID-19 pandemic represented an unexpected challenge for the medical community. Although the epidemic curve is descending in many countries, the imminent end of containment measures raises the risk of a second wave [6,7], and any new disease-related information would be useful to improve care. Given the lack of knowledge relative to specific and vulnerable patients, we report herein on a series of patients undergoing long-term HD in eight facilities in Alsace, France, where the first cluster of COVID-19 occurred. The incidence of COVID-19 was much higher in our HD patients than in the background population (9.1% versus 0.16%). There are various putative explanations. First, our HD population carried comorbidities at risk of COVID-19 infection. Second, physical distancing is challenging during HD sessions, in the waiting room, during transport to HD facilities and in living facilites (many patients live in nursing homes). Finally, these patients are in contact with healthcare providers three times a week, such that screening was likely more thorough than in the general population. There was no evidence for transmission within our HD units, as the first cases occurred in patienta treated in separate rooms and containment measures were immediately applied for all suspected patients. Indeed, each patient presenting to the HD unit with any symptom suggestive of COVID-19 was admitted to a dedicated isolated room until the RT-PCR result was available. All patients with COVIDIAL study proven COVID-19 remained thereafter in an isolated room to avoid nosocomial transmission.
In the first reports in the Chinese population [2,8], only 25-50% of patients had at least one comorbidity, which is far below our findings. Compared with hospitalized US patients who had more similar clinical characteristics and a significant number of comorbidities (88% with at least one comorbidity) [9], our patients were older and slightly less obese. Notwithstanding, our patients more frequently had diabetes, chronic respiratory disease and a history of cancer. The question of whether end-stage kidney disease (ESKD) per se or the comorbidities associated with ESKD worsens the outcome remains to be determined.
Compared with HD patients of the nationwide French Renal Epidemiology and Information Network (REIN) registry [10], patients in our cohort were older (74 6 15 versus 69 6 15 years) and were more likely to have APKD (8.5% versus 5.8%), obesity (36% versus 29%), chronic respiratory disease (34% versus 19%), cancer (20% versus 12%), ischaemic heart disease (45.6% versus 29%), stroke (22% versus 13%) and peripheral arterial disease (34.7% versus 21%). These observations highlight the substantial vulnerability of our cohort, i.e. the 'frails upon the frails'. Moreover, the higher proportion of patients with partial (51.4% versus 8.5%) or an absence of autonomy (15% versus 8.5%) would explain both the higher vulnerability of these patients and the frequent need for external or institutional help, which could be a vector of contamination. This suggests that 'barrier gestures' should be strictly implemented in these patients in whom strict containment cannot be secured due to the need for dialysis. Screening indications should be comprehensive in this population to ensure adequate isolation when needed. Also, the favoured distribution of personal protective equipment in a period of relative shortage or wider indications for hospitalization should be considered.
Typical symptoms of COVID-19 infection have been initially described as fever, cough and dyspnoea [11]. Fever was inconsistent in our cohort compared with previous reports (57% versus 98%) [8,11]. A lack of fever is frequent in infected HD patients, who tend to be slightly hypothermic in baseline conditions [12]. A considerable proportion of our patients presented with less typical symptoms, mostly asthenia, diarrhoea and anorexia. The latter has been previously described in the general COVID-19 population [8,[13][14][15] as well as in a case report of HD patients [16] and kidney transplants [17]. For example, diarrhoea, initially described as an uncommon symptom [8], was found in 34% of our patients. The extensive screening in our patients likely enabled the detection of atypical forms, although more frequent gastrointestinal involvement in patients with ESKD cannot be excluded.
Lymphopaenia was frequent in our cohort, as also previously described in the general COVID-19 population [8,9,11,15], and associated with the severity of the disease [18]. This was attributed not only to a chronic endothelial dysfunction in ageing patients with chronic disease, but also to a direct cytotoxic action of the virus [15,19]. Lymphopaenia should be considered both as part of the diagnosis and as a marker of the risk level of the disease [20]. In this study, CRP was higher than in previous studies conducted in the general COVID-19 population, even those patients with severe disease [8,9,11]. Of particular note, CRP was predictive of the risk of death in our cohort of COVID-19 HD patients. Such an association between CRP and disease severity was previously reported and associated with the extent of impairment on LD-CT scan (correlation coefficient 0.873, 0.734; P < 0.001) [21,22]. Some authors have suggested the use of a predictive score based on CRP and lung volume involvement on LD-CT scan at the time of diagnosis [21,22]. High PCT (>0.5 ng/mL), an uncommon finding in the general COVID-19 population [8,9,11], was more frequent in our cohort.
The mortality rate in our cohort was much higher than in the general COVID-19 population: 24% compared with the 1-5% mortality in the general COVID-19 population [23] and 8-15% mortality among patients >70 years old [9,23]. An abysmal prognosis was previously described in a small cohort of kidney transplant recipients in the USA [17], with an early mortality rate of 28% at 3 weeks. A similar mortality rate of 26% was found in Italian [24] patients admitted to the ICU who had fewer comorbidities and were younger [median age 63 years (IQR 56-70)]. These findings thus emphasize the detrimental effect of ESKD and comorbidity burden over that of age. We did not find any association between age and risk of death in our cohort, in contrast to the general population. This discrepancy could be explained by the more advanced age of our cohort, as we compared old versus very old patients, which might have attenuated the differential risk. However, one should bear in mind that our HD population had a much higher mortality rate (16%) than that found in the general COVID-19 population (0.91%), with a steeper effect of age [10,25]. Our results suggest an estimated mortality rate of 26.7%, but the exact case fatality rate (CFR) should take into account the asymptomatic patients, whose exact percentage was unknown in our population.
A noteworthy observation was the poor and prolonged viral clearance in our patients. Only 24% of our patients had viral clearance after a median of 32 days after onset of symptoms. while viral clearance was initially reported after a median time of 17-20 days in the general COVID-19 population [26,27]. However, a recent study [28] found a viral clearance after a median of 24 days, with a maximum duration of 42 days, which is more in line with our population. Age and ESKD are wellidentified causes of immune dysfunction, which could delay viral clearance [29]. In addition, the higher prevalence of a severe form of the disease could be associated with prolonged viral shedding [27]. Currently French guidelines [30] recommend the end of isolation 24 days after the onset of symptoms in immunocompromised patients, without control RT-PCR testing. Given the present data, it would seem cautious to perform another RT-PCR test before ending isolation. In our opinion, and considering the false negative rate of $30% [31,32], a minimum of two consecutive negative swabs would appear to be a sound precaution.
Our study has certain limitations. The cohort was modestly sized and we could not perform a proper comparison due to the lack of other reported cohorts in HD patients. The number of missing data also represents a drawback. Finally, the absence of systematic screening in both the general and specific COVID-19 populations such as ours may limit our conclusions, especially regarding the CFR. Further larger studies are necessary to estimate the risk factors in the HD population. The high mortality rate in our patients also points to the need for randomized controlled studies on antiviral therapy.

C O N C L U S I O N
In the present report, we describe a cohort of 123 HD patients presenting with symptomatic COVID-19 together with their clinical profiles and outcomes. HD patients were found to be at high risk of developing COVID-19. Several clinical characteristics, including atypical presentation, the predictive value of elevated CRP and delayed viral clearance, were worth noting and could be specific for this population. The outcomes were abysmal in this particularly vulnerable population, with a mortality rate of 24%.

S U P P L E M E N T A R Y D A T A
Supplementary data are available at ndt online.

A C K N O W L E D G E M E N T S
We express our thanks to Nadia Honoré for compiling the REIN regional data, to the Clinical Research Assistant of AURAL for gathering clinical data and all the Alsace nephrologists on the frontline. This article is dedicated to the memory of our nephrologist colleague, Abdelmajid Ben Aicha, deceased of COVID-19. This observational study was nested in the French REIN registry, which obtained an agreement from the Commission nationale de l'informatique et des libertés for the use of personal data.

C O N F L I C T O F I N T E R E S T S T A T E M E N T
None declared.