Dengue, chikungunya, and scrub typhus are important etiologies of non-malarial febrile illness in Rourkela, Odisha, India

Rourkela is located in Sundargarh district close to the northern border of the state of Odisha, India, and has a population of over 0.5 million. It has a tropical climate, with high temperatures and heavy rainfall between June–October and December–January. Subjects were enrolled at the outpatient department (OPD) of Ispat General Hospital (IGH), which is a 685-bed hospital for employees of Rourkela Steel Plant, Steel Authority of India, Inc., and for people living in and around Rourkela. A total of 90,867 patients attended the OPD at IGH Rourkela in 2016, of which 13,534 patients (~ 15%) reported fever.

Between January 2016–January 2017, 954 OPD patients aged 12 months - 70 years (with children categorized as ≤15 years and adults as ≥16 years) who reported having fever up to 48 h prior to enrollment or who were febrile on day of enrollment (body temperature ≥ 37.5 °C) were enrolled in our study. These patients had already been through triage for malaria in the OPD, where probable malaria cases were identified based on presence of high fever, chills, and rigor. Such cases, along with individuals with symptoms of influenza and pregnant women (due to the possible risks associated with enrolling these subjects), were not enrolled in our study. Written consent or assent was obtained from each enrolled patient prior to a physical examination and administration of a detailed clinical questionnaire, and each patient provided a blood sample.

All 954 enrollees were tested for malaria by at least one diagnostic test; a subset of enrollees (293/954) from whom a larger volume of blood was collected was also tested for dengue, chikungunya, and scrub typhus by either RDT, ELISA, or both tests. Patient blood samples were collected either in a microvette (small volume ~ 100 ul) and/or a vacutainer (large volume ~ 3-5 ml) at the time of enrollment and tested by Hemocue to assess haemoglobin levels. Whole blood was used for thin and thick blood smear preparations, as well as malaria RDTs. Microvette and vacutainer samples were centrifuged to separate plasma and red blood cells, and these fractions were stored at − 80 °C before use. DNA was extracted from the red blood cells using QIAamp DNA blood Mini Kits (Qiagen Inc., Valencia, CA).

Malaria: Three tests -- rapid diagnostic test (RDT), blood smear microscopy, and polymerase chain reaction (PCR) -- were used for malaria diagnosis. All 954 participants were tested P. falciparum and P. vivax by RDT (FalciVax, Zephyr Biomedicals, India, a commercially available lateral flow kit that detects P. falciparum histidine-rich protein 2 and P. vivax lactate dehydrogenase) at time of enrollment. All participants also had a thick and thin blood smear made with whole blood from a finger prick, stained with Giemsa and examined by microscopy using a 100x oil immersion objective. Thick smear parasitemia was calculated against 200–500 leukocytes and expressed as parasites per microliter of blood, using the white blood cell (WBC) count if determined, or assuming 8000 WBCs per microliter blood. A semi-nested PCR assay (modified from [14]) targeting the 18S small subunit ribosomal protein gene (SSU rDNA) was used for species-specific molecular detection of four Plasmodium species (P. falciparum, P. vivax, P. malariae, and P. ovale) using DNA extracted from a subset of patients (852/954, 89.3%) from whom a microvette (~100ul) or vacutainer (~ 3-5 ml) of blood had been collected.

Dengue, chikungunya, and scrub typhus: Testing for dengue, chikungunya, and scrub typhus was done on a subset of patients (293/954, 30.7%) enrolled from July 2016 to January 2017 from whom larger volumes of blood were available (since larger volumes of plasma are required for dengue and chikungunya RDTs). A pilot study was first conducted with blood samples from approximately 100 of these 293 patients, to test two commonly used dengue RDTs (Dengue Day 1 test from J. Mitra and Co., India, and SD Bioline Dengue Duo test from Alere Medical, India) and two chikungunya RDTs (Advantage Chikungunya IgM card RDT from J. Mitra and Co., New Delhi, India; and the SD Bioline Chikungunya IgM RDT from Alere Medical, India) [15‐17]. The Dengue Day 1 test and the SD Bioline Dengue Duo test recognize the dengue non-structural protein 1 antigen (NS1) and DENV-specific IgM and IgG antibodies in patient sera. The Advantage Chikungunya IgM card RDT and the SD Bioline Chikungunya IgM RDT detect chikungunya-specific IgM in patient sera. The Dengue Day 1 kit and the Advantage Chikungunya IgM card RDTs, both from J. Mitra and Co., worked more efficiently in our hands and were subsequently chosen for all downstream diagnoses.

Since the sensitivity and specificity of ELISA-based diagnostic kits are better than RDTs, we also used ELISA kits to test extracted plasma samples for dengue and chikungunya. The Panbio Dengue Early ELISA kit for detection of NS1 antigen (Alere Medical, India), and the Panbio Dengue IgM Capture ELISA (Alere Medical, India) were used for diagnosis of dengue, and the Chikungunya IgM ELISA kit (SD Bioline) was used for diagnosis of chikungunya.

The Scrub Typhus Detect IgM ELISA kit (InBios International, Inc., Seattle, WA, USA) was used to detect Orientia tsutsugamushi. Optical density values > 0.5 were considered as a positive test result [18, 19].

A flowchart describing samples tested for malaria, dengue, chikungunya, and scrub typhus using the above diagnostic tests is shown in Fig. 1.

Patient clinical and demographic questionnaire data were collected on paper case report forms (CRFs) that were entered and uploaded into the CSCMi’s REDCap database [20]. Results for all assays carried out on plasma, blood, and DNA samples were also uploaded into CSCMi REDCap. Data sets were exported into Stata (version 14.2). A patient was considered to have malaria if any malaria test (RDT, microscopy, or PCR) was positive. Based on national guidelines [21] our case definitions for dengue diagnosis were as follows: 1) patients with NS1 or IgG/IgM positive RDT results were considered probable dengue fever cases, and 2) patients with NS1 or IgM positive ELISA results were considered current or past primary or secondary early dengue infection cases. Patients with IgM positive RDT or ELISA results were considered probable chikungunya cases [22]. Patients with febrile illness, with or without eschar, who had plasma positive for IgM (O.D. > 0.5) by ELISA were considered probable scrub typhus cases [18]. The period from June–October was considered the rainy season; age was categorized into ≤15 (child) and ≥ 16 years (adult). Symptoms evaluated for their association with laboratory results included documented fever (body temperature ≥ 37.5 °C at the time of enrollment), headache, aches, chills, vomiting, diarrhea, cough, dizziness, sweating, abdominal pain, chest pain, eye or ear pain, or rash. The association between laboratory test results and symptoms, age, gender, season, and anemia (for the definition see Additional file 1: Table S1) were examined using the chi-square test in univariate and generalized linear regression models for multivariate models with a log link and binomial distribution [23]. A p-value < 0.05 was considered significant. The concordance between diagnostic test results was examined using the McNemar test and visualized using proportional Venn diagrams in Stata (version 14.2).

We enrolled 954 participants from patients attending the outpatient clinic at Ispat General Hospital, in Rourkela, Odisha, India from January 2016 to January 2017. A total of 58.4% patients were male and 66.4% > 15 years of age, with a median age of 25 (interquartile range 12–44). Most patients reported having fever in the last 48 h (96.5%) and 276 patients (30.2%) were febrile at the time of enrollment; the use of fever suppressant medication was common (76.6%). The most common clinical symptoms presented by patients included headache (49.7%), chills (48.3%), and aches or pains (39.8%); symptoms differed significantly by age group for eight symptoms, with documented fever, vomiting, and cough more common in children (≤15 years), and chest pain, lower back pain, dizziness, aches, and headache more common among persons aged 16 years and above (Fig. 2). There were no differences in symptoms by gender; documented fever was more common in the rainy season (33.9% vs. 24.9% in dry season, p = 0.004) and chills were more common in the dry season (43.6% in rainy vs. 55.5% in dry, p < 0.001). Sixty-four patients (6.7%) reported being diagnosed with malaria at least once in the past year, and 58 of them (90.6%) reported receiving antimalarial therapy. Anaemia was detected among 38.8% of the study population.

Fifty-four out of 954 patients (5.7%) were positive for P. falciparum, P. vivax, or P. malariae by at least one diagnostic method (RDT, microscopy, or PCR): 46 by PCR (46/852, 5.4%), 30 by RDT (30/954, 3.1%) and 25 by microscopy (25/954, 2.6%; Table 1). Among the 852 patients tested by all three tests, three patients were positive by RDT but not by another malaria test: two of these (both with P. falciparum) had a history of antimalarial treatment within the last month, but this was absent in the third patient who had both P. falciparum and P. vivax (Table 2). No patient was positive by microscopy and negative by both other tests (Table 2). Three patients were positive by RDT and microscopy but not by PCR, with species matching (2 P. vivax and 1 P. falciparum). Twenty-three patients were positive by PCR but not by RDT or microscopy (sub-patent infections, all P. falciparum); when only considering microscopy, 25 patients had submicroscopic infections (positive by PCR, but negative by microscopy, 1 P. vivax, and 24 P. falciparum, one of each was also detected by RDT). A proportional Venn diagram (Fig. 3) shows the concordance between the malaria diagnostic test results.

Twenty patients were positive by all three tests (14 P. falciparum) with only one species discrepancy whereby RDT and microscopy detected P. vivax but PCR detected P. malariae. Mixed Plasmodium species (defined as more than one species detected by the same test) were identified in only one patient (P. falciparum and P. vivax). When using PCR as the gold standard, RDT had a sensitivity of 47.8% and a specificity of 99.3%; for microscopy this was 45.7 and 99.6%, respectively. By microscopy, men were more likely to be positive for malaria (3.9% vs. 1.4% among women, p = 0.029), whereas by PCR adults were more likely to be positive (6.5% vs. 3.0% among children, p = 0.036); sub-patent infections or Plasmodium species were not associated with age, gender, or season. Vomiting and lack of body aches were associated with malaria for all tests, e.g., for RDT, vomiting was 3.80 times more likely (95% CI 1.77–8.16, p = 0.001) among patients with a positive RDT compared to patients without, whereas the prevalence ratio of report of body ache was 0.35 (95% CI 0.15–0.82, p = 0.016).

Seventy-four patients of 293 patients tested were positive (25.3%) for dengue by either RDT or ELISA (Table 3). Thirty-seven patients out of 285 patients tested were positive (13%) for NS1 antigen by both RDT and ELISA, and 8 patients out of 285 patients tested were positive (2.8%) for dengue virus (DENV)-specific IgM by both RDT and ELISA (Table 3). Ten patients of 288 tested were positive (3.5%) for both NS1 antigen and DENV-specific IgM by ELISA. Men were more likely to be positive than women for dengue IgM by the J. Mitra RDT (13/169 or 7.7% vs. 3/121 or 2.5%, p = 0.06). Rainy season was associated with a positive result for the IgM ELISA (p = 0.004) and NS1- specific ELISA (p < 0.001) tests and the J. Mitra rapid test for NS1 (p < 0.001). Documented fever was more common among patients who tested positive for NS1 antigen, but not for the other tests used (documented fever was 56.4% among 39 patients positive for NS1 antigen by RDT, and 24.4% among 246 negative patients, p < 0.001). A proportional Venn diagram (Fig. 4) shows the concordance between the dengue diagnostic test results.

Eight patients out of 290 tested were positive (2.8%) for chikungunya virus (CHIKV)-specific IgM by either the J. Mitra or SD Bioline RDT, and 15 of 288 patients tested were positive (5.2%) by ELISA for CHIKV-specific IgM. Seventeen of 293 patients tested positive (5.8%) by either ELISA or RDT for CHIKV-specific IgM, while 6 of 285 patients tested were positive (2.1%) by both CHIKV-specific IgM RDT and ELISA (Table 3). No significant associations were detected between being positive for chikungunya and gender, age, season, or symptoms, but persons positive for chikungunya by any test were more likely to be anemic (see Additional file 1: Table S1).

Of 287 samples tested, 10 (3.5%) were positive for Orientia tsutsugamushi-specific IgM antibodies by ELISA. Of these, 3 samples had an O.D. value > 1 and were likely to have active or recent infection; O.D. values between 0.5 and 1, observed in 7 patients in our study, may be indicative of early or past infections (Table 4). Scrub typhus was only detected among persons > 15 years during the rainy season; patients positive for scrub typhus were more likely to be anemic (Additional file 1: Table S1).

Of the 290 patients tested for at least three infections, 105 (36.2%) were positive for at least one pathogen, 17 (6.1%) for two, and one patient (0.4%) for three pathogens (Table 5) by any test for malaria, dengue, chikungunya, or scrub typhus. This last patient was positive for dengue, chikungunya, and P. falciparum. Eight out of ten patients positive for scrub typhus were also positive for dengue, chikungunya or malaria; six of these were dengue positive by any test (p = 0.009 for association between scrub typhus and dengue). Almost half of chikungunya positive patients (by any test) were positive for dengue, malaria or scrub typhus (8/17 or 47%); five of these were malaria positive (p < 0.001 for the association between malaria detected by any test and chikungunya, N = 280, 2 sub-patent malaria infections). No association was seen between dengue and chikungunya positives (4 co-infections with dengue positives among 17 chikungunya positive patients or 23.5%, versus 69 dengue positives among persons negative for chikungunya or 25.8%, p = 0.83). Patients tested for at least three infections were more likely to be adults and enrolled in the rainy season compared to patients tested for malaria only (p < 0.001 for both). All patients positive for multiple infections visited in the rainy season versus 93.2% of the persons with one detected infection (82/88) and 69.2% (128/185) of the patients with no infection detected (p < 0.001). No significant differences were detected by age, gender or symptoms compared to patients with a single or no infection; however, anemia was higher among the 17 patients positive for ≥2 infections (58.3%), compared to 26.8% in patients positive for one infection, and 22.1% in patients where none of these four infections were detected (N = 266, p = 0.018, Additional file 1: Table S1).

There were 1,090,724 cases of malaria and 331 deaths reported in India in 2016, of which 449,467 cases (41.2%) and 77 deaths (23.3%) were reported from Odisha (NVBDCP). These case numbers are likely underestimates since only ~ 8% of cases may be detected by the surveillance system, as noted by the World Health Organization (W.H.O.) [24]. Even going by these low estimates, India accounts for ~ 6% of all malaria cases and 51% of P. vivax cases globally. Malaria transmission occurs in both urban and rural India and is especially high among tribal populations in the mountainous, inaccessible regions of eastern and northeastern India [25].

Of the three commonly used methods for malaria diagnosis (RDTs, microscopy, and PCR), microscopy remains the standard for malaria diagnosis in India. In Asian countries, males have a higher risk of malaria (as diagnosed by microscopy) compared to females (as diagnosed by microscopy) [26]; the reason is not completely clear but may relate to differences in exposure by gender. PCR-based methods are more sensitive, but require expensive equipment that is unavailable in most regions. RDTs, while less sensitive and specific than microscopy, are available and used for first-line malaria diagnosis in remote regions. Our results confirm findings of previous studies across India of higher sensitivity of PCR over microscopy when used for detection of Plasmodium [27, 28]. Submicroscopic malaria has been reported to be more common among older people and possibly related to the development of immunity, and also with the recent use/suboptimal use of antimalarial drugs [29]. Three patients were positive by RDT and negative by microscopy and PCR: the antigens produced by recently-cleared malaria parasites can persist in the blood after treatment for a period of time, explaining the positive RDT in the presence of a negative PCR test [30]. In our previous treatment studies, we have noted that for P. falciparum (although not for P. vivax), some RDTs can be positive up to 28 days post treatment, whereas the PCR test result for the same patient was negative in 1 week. No patient identified as Plasmodium positive by microscopy was negative by both RDT and PCR. However, there were 23 sub-patent infections among the 46 patients positive for Plasmodium by PCR. Sub-patent infections are of concern, because malaria infections not detected by microscopy or RDT may not be treated appropriately; patients may continue to carry the parasite and act as a pool for further transmission. Overall, our finding that 54 out of 954 patients (5.7%) were positive for P. falciparum, P. vivax, or P. malariae by at least one diagnostic method despite all 954 patients having undergone malaria triage in the OPD leads us to recommend the use of malaria diagnostic test(s) rather than relying upon patient symptoms.

The incidence of dengue has grown dramatically around the world in recent decades. The WHO predicts an incidence of 50–100 dengue million cases and ~ 20,000 dengue-related deaths per year, and nearly 40% of the world’s population lives in areas with a risk of DENV transmission [31]. India is estimated to contribute 34% of the total global burden of dengue [32], and suffered several outbreaks of severe dengue over the last few years [33‐36]. There were 129,166 reported cases with 245 deaths in 2016 (NVBDCP). Dengue is endemic in Odisha since 2010, and there has been a four-fold increase in the number of reported cases, from 2450 in 2015 to 8380 in 2016, accounting for 6.5% of the total cases in India [35].

Dengue diagnosis in our study was carried out by detection of either the NS1 antigen or antibodies specific to DENV envelope proteins. We used two techniques for detection of DENV: RDTs to detect NS1, DENV-specific IgG and IgM, and ELISA targeting either NS1 or DENV-specific IgM. NS1 antigen is present in the bloodstream immediately after onset of fever for up to 18 days, and detection of NS1 can thus be used as a proxy for early infection. DENV-specific IgM antibodies may be present in the serum as early as 3–5 days following onset of fever, and persist for several months, in the case of a primary infection, but appear earlier in a secondary infection. IgG antibodies appear 14 days after a primary dengue infection, and can persist for life, but appear within the first couple of days following onset of symptoms in a secondary infection.

We observed a 100% concordance of Dengue NS1 positivity between 101 samples tested by RDTs from two different manufacturers (J. Mitra and SD Bioline) in India, with a positivity rate of ~ 13%. A slightly larger but comparable number of samples (~ 14.2%) were positive when tested by ELISA for NS1 antigen. Four samples were NS1 antigen positive by ELISA but not RDT, and two samples were positive by RDT but not ELISA. There was a significant difference in the number of samples positive by ELISA for Dengue NS1 versus DENV-specific IgM, as expected since these methods test chronologically different stages of the infection, with NS1 antigen presence in the serum preceding DENV-specific IgM by 3–5 days. However, there were significantly fewer samples identified as positive for DENV-specific IgM by RDT than ELISA (20%), which may be due to the increased sensitivity of the capture ELISA IgM over the RDT. Only 2 samples were positive for DENV-specific IgG by the SD Bioline RDT but they were also positive by the J. Mitra RDT. There was also a large variation in positivity in our dengue test results, ranging from 1 to 14.2% by any single test, and 25.3% positive across all dengue tests combined, so it is likely that some positives will be missed if just one diagnostic test is used. Dengue prevalence rates varied from 7 to 38% across other hospital-based febrile illness studies in India and depending upon the assay used; the highest prevalence was observed in a study that used RDTs for detection of DENV (Table 6).

Chikungunya transmission is concentrated in Southeast Asia, the Americas, Pacific Islands, and Africa. The virus has a high infection rate, with up to 75% of a population infected in areas of virus circulation [37]. Early documented outbreaks of chikungunya in India in the 1960s and 1970s were followed by a major outbreak in Kenya in 2004, which spread to the Indian subcontinent and southeast Asia by 2010 [8, 38]. Since then, there has been continuous transmission of chikungunya in India [8, 39], with an increase in number of cases in India from 27,553 in 2015 to 64,057 in 2016, although in Odisha the number of cases decreased from 81 to 51 during the same period (NVBDCP).

A definitive diagnosis can be accomplished using three laboratory tests: virus isolation, serology, and PCR. We used both an immunochromatography-based RDT as well as an ELISA assay to measure chikungunya-specific IgM levels in patient plasma. It is important to note that chikungunya-specific IgM is usually detected in the plasma from day 4–7 following the onset of symptoms, so we may have missed early infections. In our study, 2.8% patients tested were positive for chikungunya-specific IgM by the RDT (J. Mitra and Co.), but twice as many (5.2%) were found to be positive by the more sensitive ELISA based assay. We did not have the resources to confirm the test results with virus culture and RT-PCR, so it is unclear which test result is more reliable, and thus, further examination of these results is warranted. Similar proportions of chikungunya IgM-positive individuals (2–6%) were observed in other hospital-based studies from India (Table 6) [4, 5, 11].

Scrub typhus is a re-emerging infection in India in that its prevalence has been increasing in recent years across various parts of the country [11]. It is underreported [11, 40], and prevalence data are not easily available as routine field and hospital-based surveillance has only recently begun [41]. It rose to prominence as a dominant cause of fever among soldiers along the Indo-Burmese border during World War II, and the Indo-Pakistan border in the 1990s [42]. India belongs to the ‘tsutsugamushi triangle’, the postulated zone of highest prevalence of scrub typhus cases [42], and the illness has been found to be prevalent ubiquitously across the country [40, 43‐47].

Scrub typhus diagnosis may be carried out by one or more of the following methods: IgM and IgG ELISA to test for the presence of anti-O. tsutsugamushi antibodies in patient serum, and PCR targeting O. tsutsugamushi antigenic genes. We used the Scrub Typhus Detect ELISA kit from InBios international to identify IgM antibodies targeting the O. tsutsugamushi-derived recombinant 56-kDa antigen, as an indicator of exposure to the bacterium. We observed a positive result in 10/287 (3.5%) patient sera tested. Other hospital-based surveillance studies in India have identified scrub typhus as a frequent etiology of acute febrile illness, with frequencies varying from 9 to 48% [4, 11, 44, 48, 49] (Table 6). We were unable to test for the presence of Orientia bacteria by PCR and did not collect convalescent sera from patients to test an increase in antibody titers by IFA. We may also have missed some scrub typhus cases if the plasma sample was collected too early in the infection. Eschar was not observed in any patients.