Gastrointestinal (GI) symptoms, such as diarrhea, vomiting, abdominal pain and nausea are not an uncommon manifestation of an acute respiratory infection (ARI) [1–8] (Additional file 1) and have been reported as a hallmark of severe influenza in childhood.
Influenza viruses and other respiratory agents such as human rhinoviruses (HRV), have been detected in stools of patients with ARIs (Additional file 2) [5, 10–14], but their correlation with GI symptoms and their viability in stool is still debated [10, 15].
There are several possible explanations for the observed GI symptoms during an ARI. First, each winter, ARIs and gastroenteritis outbreaks overlap, creating a spurious association between ARI and GI symptoms, maybe caused by a co-infection between respiratory agents and enteroviruses. Second, GI symptoms may be a side effect of drug treatment (antibiotic or antiviral) [17, 18] or food consumption (ex: raw shellfish and molluscs). Third, GI symptoms could either be a manifestation of a direct viral effect, or an indirect viral effect of respiratory viruses, such as lung-derived CD4+ cell-induced dysbiosis resulting in intestinal injury.
Insufficient information about the prevalence of GI symptoms in ARIs, their clinical features and their potential risk factors may lead to diagnostic errors and inadequate treatment.
In the context of the above limitations, the main objectives of this two-year (2014–2016) prospective study were to evaluate the demographical, clinical and microbiological factors associated with the presence of GI symptoms in patients presenting to general practitioner (GP) with an ARI.
A representative sample of 60 GPs from the French Sentinelles Network [21, 22] was recruited to enrol ARI patients from all over mainland France.
To ensure that the selection of ARI patients remained random, each GP was required to include, each week, the first two patients seen in consultation who met the inclusion criteria, regardless of their age. The case definition of ARI was “any person with a sudden onset of symptoms and at least one of the following four systemic symptoms: fever (≥ 38 °C or greater) or history of fever (≥ 38 °C or greater) taken at home or feverishness, malaise, headache, myalgia, AND at least one of the following three respiratory symptoms: cough, sore throat, or shortness of breath”. All patients were recruited within 48 h of the onset of symptoms.
Patients were enrolled by their GPs over two consecutive ARI seasons from week 46, 2014 (10–16 November 2014) to week 15, 2015 (06-12 April 2015) and from week 45, 2015 (02-8 November 2015) to 16, 2016 (18–24 April 2016) (Additional file 3).
The GPs completed a case report form (CRF) for all volunteers who met the case definition and agreed to participate, and submitted this by post (all items are listed on Additional file 4). We defined a patient as vaccinated if he/she had received at least one dose of seasonal influenza vaccine at least 15 days before the onset of ARI symptoms.
Two types of samples were obtained for each enrolled patient: a nasopharyngeal swab and a stool sample. The nasopharyngeal specimen was collected by the GP and was sent with the CRF to the test laboratory by post in a triple-packaged Copan universal transport medium (UTM-RT) container (Copan Italia, Brescia, Italy). Included patients were asked to collect stool specimens and send them to the laboratory within 48 h by post in triple packaging as required by the United Nations class 6.2 specifications.
Nucleic acid extraction
For nasopharyngeal specimens, nucleic acids were extracted from 200 μl of UTM-stored sample and eluted in 60 μl of elution buffer using QiAamp MinElute virus spin kits (Qiagen, France) according to the manufacturer’s instructions. Stool specimens were centrifuged at 14,000 xg for 20 min; then nucleic acids were extracted from 200 μl of the UTM-stored sample and eluted in 40 μl of elution buffer using QiAamp MinElute virus spin kits (Qiagen) according to the manufacturer’s instructions. An internal control (T4 and MS2 phages) was added to each extraction tube to assess the quality of the extraction at the end of the amplification.
Detection of respiratory pathogens
All extracted samples (nasopharyngeal and stool) were screened for influenza A and B viruses by real-time quantitative Reverse Transcription PCR (RT-qPCR); influenza A virus-positive specimens were subtyped and influenza B virus-positive samples were analysed for Victoria and Yamagata lineage according to the method developed by the French National Influenza Centre [25, 26]. Then, the presence of 10 non-influenza respiratory pathogen groups was analysed by RT-qPCR and qPCR using a Fast Track Diagnostic (FTD) Respiratory pathogens 21 kit (Fast Track Diagnostic, Luxemburg) using five tubes containing primer and probe mix for different viruses; Tube-1 [Influenza A, Influenza A subtype H1N1 (A(H1N1)pdm09), human Rhinovirus (HRV), Influenza B], Tube-2 [human Coronaviruses NL63 (HCoV-NL63), 229E (HCoV 229E), OC43 (HCoV-OC43), and HKU1 (HCoV HKU1)], Tube-3 [human Parainfluenza viruses, 2, 3, and 4 (HPIV- 2, 3 and 4) & Internal Control], Tube-4 [human Parainfluenza viruses-1, Mycoplasma pneumoniae (M.pneu), human Bocavirus (HBoV), human Metapneumovirus (HMPV A/B)] and Tube-5 [Respiratory Syncytial virus (RSVA/B), human Adenovirus (HAdV), Enterovirus (EV), human Parechovirus (HPeV)].
Detection of enteric pathogens
Extracted stool samples were screened by RT-qPCR and qPCR using the FTD Viral gastroenteritis kit (Fast Track Diagnostic, Luxemburg) according to the manufacturer’s instructions, using three multiplex PCRs to detect six viruses: human norovirus (hNoVG1 and hNoVG2), adenovirus (hAdV), human astrovirus (HAstV), rotavirus (RV) and sapovirus (SaV). The panel FTD Bacterial gastroenteritis kit (Fast Track Diagnostic, Luxemburg) was used following the manufacturer’s procedure, using two multiplex RT-qPCR for six bacteria: Campylobacter coli/jejuni/lari, Escherichia coli verotoxin positive, Salmonella spp., Shigella spp. + enteroinvasive Escherichia coli, Yersinia enterocolitica, Clostridium difficile. Two different positives controls for viral and bacterial multiplex RT-qPCR reactions and a negative control tube are provided in these kits.
Quantitative variables were described by using means [Min-Max] and standard deviations were compared by the Wilcoxon test. Qualitative variables were described by using proportions and compared using a chi-square or Fisher’s exact test if the chi-square test was not applicable; the results were presented as odds ratio with 95% confidence intervals (OR [95% CI]). We used unconditional logistic regression model to study the factors associated with SGI in ARI patients (yes/no) by comparing independent effects of factors that were associated in the bivariate analyses (p-value of <0.20). Variables for the model were chosen through automatic backwards selection using a significance level of 0.05. Bivariate and multivariate analyses were performed on patients with only one pathogen detected in nasopharyngeal swabs and/or in stool sample. All analyses were been performed using the R program (http://www.r-project.org).
Respiratory pathogens identified in nasopharyngeal samples
Overall, the nasopharyngeal specimens of 204 of the 331 (61.6%) patients were positive for at least one of the 12 respiratory pathogen groups analysed in this study (Table 1). Infection with a single virus accounted for 87.2% (194/204) of the positive nasopharyngeal samples, whereas infections with multiple viruses observed in 5% (10/204) of them, including nine double infections: (A(H1N1)pdm09/HCoV, ADV/HBoV, two Influenza B virus/HCoV, two HCoV/HRVS and two HRV/HBoV) and one triple infection (HRV/ADV/HRSV) (Fig. 2a). The most frequently identified pathogen was influenza virus (34.4%, 114/331; consisting of influenza A virus [12.7%, 42/331] and influenza B virus [21.7%, 72/331]), followed by HCoV (10.6%, 35/331), HRV (7.5%, 25/331) and RSV (6.0%, 20/331) (Table 1 and Fig. 2a). Of the 35 samples that tested positives for HCoV, 13 were HCoV-NL63, 10 HCoV-229E, 7 HCoV-OC43 and 5 HCoV-HKU1.
Factors related to GI symptoms
All factors listed in Table 1 have been analysed to investigate association with GI symptoms in ARI patients. Table 2 shows factors significantly related to GI symptoms in ARI patients. Results of RNA/DNA positivity in stools between ARI patients with and without GI symptoms for the respiratory pathogens tested were also reported in Table 2.
ARI patients who reported at least one GI symptom (57.1%; 189/331) were associated with the presence of high fever (>39 °C) (adjusted odds ratio [aOR] = 1.7 95% confidence interval [CI] [1.1–2.7]; p = 0.03), and headaches (aOR = 2.0 [1.2–3.4]; p = 0.003) (Table 2).
ARI patients with GI symptoms were more likely to have at least one enteric infection (aOR = 3.2 [1.2–9.9]; p = 0.02) detected in stool or to have an infection with HCoV detected in the nasopharynx (aOR = 2.7; [1.2–6.8]; p = 0.002) (Table 2). Proportion of GI symptoms in ARI patients with single infection ranged from 33.3% with HRV infection (in nasopharyngeal swab) to 79.2% with enteric pathogens infection (in stool specimens) (Table 3). ARI patients with HCoV detected in the nasopharynx or enteric pathogen detected in stool were statistically more likely to have GI symptoms than ARI patients with other respiratory pathogens infection (Table 3). Among the 104 ARI patients with laboratory-confirmed influenza at least in the nasopharynx, 56.7% (59/104) had GI symptoms (Table 2). Consumption of antipyretic medication before the consultation seemed to reduce the risk of developing GI symptoms for this population (aOR = 0.3 [0.1–0.6]; p = 0.002) (Table 2).
In this study, results showed that the presence of GI symptoms in ARI patients could not be explained by the detection of respiratory pathogens in stools. However, GI symptoms were more common among patients with ARI who were exclusively infected with HCoV detected in nasopharyngeal sample. This association cannot be explained by the presence of HCoVs in stools because the simultaneous detection of HCoV in nasopharyngeal and stool specimens was sporadic.
Even if the association of GI symptoms with enteric infections is not surprising, it is interesting to point out that 13.2% (25/189) of ARI infections with GI symptoms were associated with laboratory-confirmed enteric infections. This result suggests that GI symptoms in patients with ARI could be related to enteric infections, and that the positive correlation between GI symptoms and fever or headache observed in this study increases the difficulty of clinical diagnosis.
We detected, HCoVs in 10.6% of nasopharyngeal samples of patients with ARI. These results are in line with previous studies reporting HCoVs in 2.1%–18% of respiratory samples of ARI patients. In the present study, patients with HCoVs featured 11.6% of ARI patients with GI symptoms. Moreover 78.9% of patients with HCoV infection declared to have GI symptoms. Although HCoVs are recognized as causes of respiratory infection, their role in gastrointestinal infection remains uncertain and a subject of debate [12, 28, 29]. In the present study, GI symptoms were positively associated with single laboratory-confirmed HCoV infection detected in the nasopharynx of ARI patients. This association cannot be explained by the presence of HCoVs in stools because the simultaneous detection of HCoV in nasopharyngeal and stool specimens was observed in four patients only. The four commonly circulating HCoVs (1 HCoV-NL63, 1 HCoV-229E, 1 HCoV-OC43 and 1 HCoV-HKU1) were detected in stool samples, thus none of the four HCoV could be specifically associated with positivity of stools. The proportion of HCoVs in stool specimens was less important than it was in nasopharyngeal specimens (4 versus 28 respectively) which hampered an efficient comparison of the results and limited their interpretation. Moreover there was no ARI patient presenting HCoV in stools in the absence of HCoV in nasopharynx. Therefore the presence of HCoV RNA in stool is likely due to swallowing rather than due to local replication in the GI tract. The presence of HCoVs in nasopharynx seems to be linked to GI symptoms in ARI patients but the biological mechanism remained unclear. In line with previous studies, no association was observed between seasonal influenza virus detection in nasopharyngeal or stool samples and GI symptoms in ARI patients. However, among the 104 patients with influenza infection, 56.7% (59/104) presented GI symptoms. The mouse model used by Wang showed that influenza infection through a mechanism dependent on type I interferons (IFN-Is) can alter the composition of the intestinal microbiota, resulting in immunological dysregulation that may promote inflammatory gut disorders. The number of Escherichia coli (E.coli) in the intestinal tract increased, perhaps leading to intestinal immune injury. A similar study reported that influenza-induced IFN-Is enhance susceptibility to Salmonella intestinal colonization and dissemination during secondary Salmonella-induced colitis through suppression of host intestinal immunity. The systemic role for IFN-Is in altering the intestinal microbial balance after influenza infection need to be explored.
Interestingly, we found that the consumption of antipyretic drugs before consultation seemed to reduce the risk of developing GI symptoms among laboratory-confirmed influenza patients. This result is in line with previous studies that showed that paracetamol dramatically decreases the morbidity associated with influenza, thereby reducing the clinical symptoms associated with influenza virus infection [31, 32]. Therefore, the consumption of antipyretic drugs before consultation may lead to the underestimation of the frequency of GI symptoms in patients with laboratory-confirmed influenza.
The strengths of this study include its prospective multicentrer design and study length spanning two consecutive ARI seasons, standardized patient screening by the participant GPs, centralized confirmation of microbiological data, the simultaneous search of respiratory pathogens in nasopharyngeal and stool samples and the presence of enteric pathogens (viruses and bacteria) in stool, and other confounding factors that might also cause GI symptoms.
This study did have several limitations. First, the main limitation of this study was the lack of culturing of respiratory viruses from stool samples to determine if RT-qPCR detection represented the presence of viable virus. The detection of respiratory viruses in the stool could simply be RNA/DNA from viruses that were swallowed. A recent study showed that a swallowed virus could be detected in stools if protective mechanisms render it resistant to gastric acid and bile/pancreatic juice. High viscosity of mucus could protect influenza viruses from inactivation in the gastrointestinal environment, accounting for detection of the virus in feces. Second, the number of patients included here did not allow the identification of meaningful associations by sub-analyses. Studies with a small-to-moderate sample size that employ logistic regression have been reported to overestimate the effect measure. Third, we did not collect data pertaining to GI symptoms after GP consultation, which hampered the interpretation of the results.
In conclusion, except for ARI patients with enteric pathogens in stool samples, the presence of GI symptoms in ARI patients could not be explained by the detection of respiratory pathogens in stools. However, the detection of enteric pathogens in stool samples could explained by the presence of GI symptoms in some of ARI cases. The biological mechanisms explaining the association between the presence of HCoVs in nasopharynx and GI symptoms need to be explored.