Clinical Presentation, Risk Factors, and Comparison of Laboratory Diagnostics for Seasonal Influenza Virus Among Cambodians From 2007 to 2020

Abstract

Background

Despite its global significance, challenges associated with understanding the epidemiology and accurately detecting, measuring, and characterizing the true burden of seasonal influenza remain in many resource-poor settings.

Methods

A prospective observational study was conducted in Cambodia at 28 health facilities between 2007 and 2020 utilizing passive surveillance data of patients presenting with acute undifferentiated febrile illness (AUFI) to describe the prevalence of influenza A and B and characterize associated risk factors and symptoms using a questionnaire. A comparison of rapid influenza diagnostic tests (RIDTs) and real-time reverse transcription polymerase chain reaction (rRT-PCR) results was also conducted.

Results

Of 30 586 total participants, 5634 (18.4%) tested positive for either influenza A or B, with 3557 (11.6%) positive for influenza A and 2288 (7.5%) positive for influenza B during the study. Influenza A and B were strongly associated with the rainy season (odds ratio [OR], 2.30; P < .001) and being from an urban area (OR, 1.45; P < .001). Analysis of individual symptoms identified cough (OR, 2.8; P < .001), chills (OR, 1.4; P < .001), and sore throat (OR, 1.4; P < .001) as having the strongest positive associations with influenza among patients with AUFI. Analysis comparing RIDTs and rRT-PCR calculated the overall sensitivity of rapid tests to be 0.492 (95% CI, 0.479–0.505) and specificity to be 0.993 (95% CI, 0.992–0.994) for both influenza type A and B.

Conclusions

Findings from this 14-year study include describing the epidemiology of seasonal influenza over a prolonged time period and identifying key risk factors and clinical symptoms associated with infection; we also demonstrate the poor sensitivity of RIDTs in Cambodia.

While the severity and impact of pandemic influenza are well recognized, the ongoing morbidity and mortality of seasonal influenza remain underappreciated, particularly in terms of its impact on young, elderly, and vulnerable populations [1–4]. Circulating throughout the world, seasonal influenza A and B viruses are highly contagious acute respiratory infections transmitted readily between humans through infectious respiratory droplets [5–7]. It is estimated that seasonal influenza A and B result in 3–5 million cases of severe acute respiratory disease and 290 000 to 650 000 deaths each year globally, with estimated mortality rates in Southeast Asia ranging between 3.5 and 9.2 per 100 000 population [5, 6].

At a global level, the capacity to implement influenza surveillance has improved in recent years, including the increased use of laboratory diagnostics, sentinel surveillance, and information sharing through formalized international networks such as the Global Influenza Surveillance and Response System (GISRS) [1, 8, 9]. Despite these overall improvements, frontline health systems continue to face persistent challenges accurately detecting, characterizing, and measuring the true impact of influenza in certain high-risk populations, especially in less-resourced areas [5, 10].

Throughout Southeast Asia and the greater Asia-Pacific region in general, the epidemiology and burden of influenza are complex and globally significant. Large population centers and animal markets lead to higher proportions of human–animal interactions when compared with other regions of the world and an associated potential for the mutation, emergence, and seeding of influenza virus outbreaks [11–15]. Understanding the transmission dynamics and epidemiological characteristics of influenza is essential to support health program planning, preparedness, vaccine strategies, and control interventions, particularly since the emergence of the coronavirus disease 2019 (COVID-19) pandemic. In 2006, the Cambodian National Influenza Center (NIC), under the Institute Pasteur du Cambodge (IPC), was established to support the monitoring of influenza strains within the country as part of the GISRS network [10]. Since the NIC's establishment, the seasonal and ongoing circulation of multiple influenza strains, including the pandemic influenza A (H1N1) 2009 strain, are routinely detected in Cambodia, generally peaking during the rainy season [10, 16, 17].

As part of a collaborative effort to strengthen febrile illness management in Cambodia, the National Institute of Public Health (NIPH), Cambodia Ministry of Health, and the United States Naval Medical Research Unit INDO PACIFIC (NAMRU-IP) established a long-term prospective health facility–based passive surveillance program utilizing laboratory diagnostics to test, treat, and characterize patients presenting with acute undifferentiated febrile illness (AUFI). The aims of the study included utilizing AUFI surveillance data to describe the longitudinal epidemiology of influenza A and B in Cambodia and characterize associated risk factors and symptoms of influenza among AUFI patients. Additionally, the study's design allowed for a diagnostic comparison between rapid influenza diagnostic tests and real-time reverse transcription polymerase chain reaction (rRT-PCR).

METHODS

Study Design and Participation

Data were collected between January 2007 and December 2020 as part of a long-term surveillance program titled “Surveillance and Etiology of Acute Undifferentiated Febrile Illnesses (AUFI) in Cambodia,” which has been previously described [18, 19]. We used a prospective observational design to identify and record the outcomes and etiologies of eligible participants from 28 study sites throughout Cambodia. Study site health facilities were spread across 9 provinces throughout Cambodia, comprising 2 provincial referral hospitals, 3 military referral hospitals, and 23 public health centers. The estimated total health facility population catchment area was around 900 000, encompassing ∼6% of Cambodia's total population. Eligibility criteria included individuals aged 2 years or older presenting at 1 of the study site health facilities with AUFI symptoms, including a measured oral or tympanic temperature >38°C or axillary temperature >37.5°C and fever of ≥24 hours but <10 days. In the study, AUFI was defined as the acute onset of fever lasting <10 days, where no pathognomonic finding was identified following a full clinical history and physical examination by a trained study physician that would suggest a specific diagnosis [18, 20]. Participants were not followed longitudinally during this study and could re-enroll if returning to a study site health facility and meeting the eligibility criteria.

Upon enrollment and consent, physicians completed a questionnaire to record the study participant's clinical data including temperature, respiratory and pulse rates, blood pressure, and information pertaining to the onset and history of the current illness, associated symptoms and their duration, and current medications. Participants also completed a questionnaire to collect demographic data including age, gender, place of residence, select risk factors, and contact details. Participant enrollment and acute clinical visit questionnaires used to capture these data are included as Supplementary Files 1 and 2. All participants were tested on site with a rapid influenza diagnostic test (RIDT) using the QuickVue Influenza A + B Test (Quidel Inc., San Diego, CA, USA). Nasal and throat swab specimens were also collected from all enrolled participants and sent for laboratory analysis.

Specimen Collection, Transportation, and Analysis

Nasal and throat swabs were collected by trained medical personnel using standard clinical procedures. Swabs were placed into separate vials containing 2–3 mL of viral transport medium and stored at 4°C. Samples were transported to the NAMRU-IP laboratory, located within the Cambodian National Institute of Public Health (NIPH) in Phnom Penh, for testing. To conduct rRT-PCR, ribonucleic acid (RNA) was extracted from the swab specimens using QIAamp viral RNA mini kits (QIAGEN, Hilden, Germany) as per the manufacturer's instructions and stored at −70°C. One-step rRT-PCR assays were performed on each sample to detect the influenza A or B virus genomes as previously described [15].

Definition of a Positive Influenza Virus Infection

For this study, a positive influenza case was defined as any participant testing positive by either rapid test or rRT-PCR assay.

Statistical Analysis

Pearson's chi-square test was used to analyze categorical variables, with the Wilcoxon rank-sum test applied for numerical variables that did not meet the assumptions of normality and equal variance between groups. Binomial generalized linear models were used to model the relationship of demographic and environmental characteristics with influenza status or symptom cluster. Factor analysis of mixed data (FAMD) was performed to conduct symptom cluster analysis and describe the association between variables. Hierarchical cluster analysis of the participants based on the FAMD-transformed matrix was performed. Symptoms recorded in the acute clinical visit questionnaire were included in the model, with the first 9 selected and retained for cluster analysis as these accounted for >80% of the total variance. A diagnostic test accuracy meta-analysis was conducted to analyze sensitivity and specificity outcome measures of the RIDT compared with rRT-PCR results. Logit transformation and Clopper-Pearson methods were additionally applied to the sensitivity and specificity analysis models to generate meta-analysis forest plots as previously described [21]. Data were analyzed using R statistical software (version 4.3.0; R Foundation). All statistical tests were set at a significance level of P < .05. Factor and cluster analyses were conducted using the FactoMineR (version 2.9; Husson et al.) and factoextra (version 1.0.7; Kassambara and Mundt) R packages.

Coronavirus Disease 2019 Pandemic Implications

As a result of the emergence of the COVID-19 pandemic, individuals with fever and influenza-like illness (ILI) symptoms during 2020 were categorized as suspected COVID-19 patients and referred directly to Cambodian Government–appointed COVID-19 testing facilities. While study-related influenza testing for patients at study site facilities occurred throughout 2020, eligible participant enrollment numbers were impacted. Data collected during 2020 were excluded from the symptom cluster analysis due to participant enrollment and sample collection disruptions associated with the COVID-19 pandemic.

Patient Consent

Enrollment in the study was voluntary. Eligible participants ≥18 years of age provided written informed consent, while the parent or legal guardian of nonadult participants (<18 years of age) provided permission and the dependent provided assent for participation. The study was approved by the National Ethics Committee for Health Research (NECHR) of the Royal Kingdom of Cambodia (reference number 208 NECHR) and the Naval Medical Research Center's Institutional Review Board (project number NAMRU2.2012.0001) in compliance with all applicable federal regulations governing the protection of human subjects.

RESULTS

A total of 30 586 participants were enrolled and tested for influenza throughout the study period, comprising 16 393 (53.6%) children under the age of 18 years and 14 193 (46.4%) adults aged 18 years or older. Over the 14-year period, there were an additional 473 patients who met the study eligibility criteria but elected not to participate. Table 1 provides a summary of study participant characteristics and individuals testing positive for influenza. Figure 1 illustrates a flowchart of study participant enrollments and influenza test results by year. A location map illustrating the number of participants per study site from each participating health facility is included as Supplementary File 3.

Table 1.

Characteristics of Overall Study Participants and Individuals Testing Positive for Influenza

			Influenza A		Influenza B		Influenza A or B
	Category	Testeda	Positiveb	P Valuec	Positive	P Valuec	Positive	P Valuec
Participants	Total	30 586 (100)	3557 (11.6)		2288 (7.5)		5634 (18.4)
Age	Median (IQR), y	15 (7–30)	10 (6–18)	<.001	9 (6–14)	<.001	10 (6–17)	<.001
Age group	Adult (≥18 y)	14 193 (46.4)	940 (6.6)	<.001	472 (3.3)	<.001	1366 (9.6)	<.001
	Children (<18 y)	16 393 (53.6)	2617 (16.0)		1816 (11.1)		4268 (26.0)
Age breakdown	2–4 y	3802 (12.4)	563 (14.8)	<.001	352 (9.3)	<.001	876 (23.0)	<.001
	5–14 y	11 354 (37.1)	1885 (16.6)		1366 (12.0)		3128 (27.5)
	15–24 y	5011 (16.4)	498 (9.9)		293 (5.8)		766 (15.3)
	25–34 y	4220 (13.8)	295 (7.0)		141 (3.3)		422 (10.0)
	35–44 y	2570 (8.4)	146 (5.7)		57 (2.2)		197 (7.7)
	45–54 y	1993 (6.5)	103 (5.2)		41 (2.1)		142 (7.1)
	55–64 y	1100 (3.6)	43 (3.9)		23 (2.1)		65 (5.9)
	65+ y	536 (1.8)	24 (4.5)		15 (2.8)		38 (7.1)
Gender	Female	15 021 (49.1)	1709 (11.4)	.2	1048 (7.0)	.001	2657 (17.7)	.001
	Male	15 565 (50.9)	1848 (11.9)		1240 (8.0)		2977 (19.1)
Education	Lower primary	19 825 (64.8)	2447 (12.3)	<.001	1647 (8.3)	<.001	3933 (19.8)	<.001
	Primary	6076 (19.9)	687 (11.3)		416 (6.8)		1071 (17.6)
	Lower secondary	2520 (8.2)	224 (8.9)		118 (4.7)		335 (13.3)
	High school	2026 (6.6)	185 (9.1)		99 (4.9)		275 (13.6)
	Diploma or university	139 (0.5)	14 (10.1)		8 (5.8)		20 (14.4)
Area	Rural	14 206 (46.4)	1071 (7.5)	<.001	819 (5.8)	<.001	1806 (12.7)	<.001
	Urban	16 380 (53.6)	2486 (15.2)		1469 (9.0)		3828 (23.4)
Season	Dry (Nov–Apr)	11 613 (38.0)	586 (5.0)	<.001	722 (6.2)	<.001	1265 (10.9)	<.001
	Rainy (May–Oct)	18 973 (62.0)	2971 (15.7)		1566 (8.3)		4369 (23.0)

Abbreviation: IQR, interquartile range.

^aNo. (% by category).

^bNo. (% test positive).

^cPearson's chi-square test, Wilcoxon rank-sum test.

Figure 1.

Flowchart of participant enrollment by year and influenza rapid test and rRT-PCR assay results. Abbreviation: rRT-PCR, real-time reverse transcription polymerase chain reaction.

Of the 30 586 participants tested for influenza by rapid test and/or rRT-PCR, a total of 5634 (18.4%) individuals tested positive for either influenza A or B, with 3557 (11.6%) positive for influenza A and 2288 (7.5%) positive for influenza B. A summary of positive influenza cases by type and diagnostic method is illustrated in Supplementary File 4. Of the 5427 rRT-PCR-detected cases, influenza subtypes included 169 (3.1%) H1N1, 1612 (29.7%) Pandemic H1N1, 1555 (28.7%) H3N2, 2 (<0.1%) H5N1, 316 (5.8%) B/Victoria, and 178 (3.3%) B/Yamagata, with the remaining 1595 (29.4%) untyped due to resource constraints. Notably, influenza B lineages were not determined by additional RT-PCR testing before 2016. A breakdown of influenza subtypes recorded by year is provided in Supplementary File 5 (Supplementary Table 1).

Most influenza cases among AUFI patients were recorded in children (n = 4268, 75.8%), notably 5–14 years of age (n = 3128, 55.5%), and among participants living in urban settings (n = 3828, 67.9%) and were detected during the rainy season from May through to October (n = 4369, 77.5%). The highest test positivity proportions were also recorded in children (Table 1). During the study period, the highest test positivity proportions among AUFI patients recorded by year were in 2007 for influenza A and in 2011 for influenza B (Figure 2).

Figure 2.

Test positivity proportions of influenza by type and age group per year throughout the study duration.

Table 2 provides an analysis of the association between positive influenza cases and key characteristics. During the study, the odds of AUFI patients testing positive for either influenza A or B were 2.3 times higher during the rainy season as compared with the dry season (odds ratio [OR], 2.30; 95% CI, 2.15–2.47; P < .001), with a large association between influenza A and the rainy season (OR, 3.30; 95% CI, 3.01–3.63; P < .001). Participants from urban areas had a 45% greater likelihood of testing positive for either influenza A or B compared with participants from rural areas (OR, 1.45; 95% CI, 1.34–1.57; P < .001). This association was greater for influenza A (OR, 1.58; 95% CI, 1.44–1.75; P < .001) than for influenza B (OR, 1.16; 95% CI, 1.04–1.31; P = .009).

Table 2.

Association of Key Participant Characteristics and Positive Influenza Cases

	Influenza Type A			Influenza Type B			Influenza Type A or B
	aOR	95% CI	P Value	aOR	95% CI	P Value	aOR	95% CI	P Value
Year	0.99	0.98–1.00	.2	1.02	1.01–1.04	<.001	1	0.99–1.01	.5
Age	0.98	0.98–0.99	<.001	0.98	0.97–0.98	<.001	0.98	0.98–0.98	<.001
Gender
Female	—	—		—	—		—	—
Male	1.01	0.94–1.09	.8	1.12	1.03–1.22	.011	1.06	1.00–1.13	.051
Education
Lower primary school	—	—		—	—		—	—
Primary school	1.17	1.06–1.29	.002	1.13	1.00–1.27	.05	1.18	1.08–1.28	<.001
Lower secondary school	0.98	0.84–1.14	.8	0.87	0.71–1.06	.2	0.94	0.82–1.06	.3
High school	1.17	0.98–1.39	.082	0.99	0.79–1.24	>.9	1.09	0.94–1.26	.2
Diploma or university	1.37	0.74–2.35	.3	1.3	0.58–2.53	.5	1.25	0.75–2.01	.4
Area
Rural	—	—		—	—		—	—
Urban	1.58	1.44–1.75	<.001	1.16	1.04–1.31	.009	1.45	1.34–1.57	<.001
Season
Dry (Nov–Apr)	—	—		—	—		—	—
Rainy (May–Oct)	3.30	3.01–3.63	<.001	1.23	1.12–1.35	<.001	2.30	2.15–2.47	<.001

Abbreviation: aOR, adjusted odds ratio.

Among the 5634 participants testing positive for either influenza A or B, the most common symptoms reported in addition to fever were cough (n = 4846, 86.0%), headache (n = 4504, 79.9%), and sore throat (n = 3908, 69.4%). Of symptoms specifically recorded, the least common reported among influenza-positive participants were bloody stools (n = 11, 0.2%), diarrhea (n = 156, 2.8%), and shortness of breath (n = 365, 6.5%). Analysis of individual symptoms identified cough (OR, 2.8; 95% CI, 2.58–3.07; P < .001), chills (OR, 1.4; 95% CI, 1.30–1.47; P < .001), and sore throat (OR, 1.4; 95% CI, 1.29–1.48; P < .001) as having the strongest positive associations with influenza among patients with AUFI during the study. A breakdown of symptoms recorded by participants testing positive and negative to influenza and measures of association of individual symptoms of participants testing positive to either influenza A or B are provided in Supplementary File 5 (Supplementary Tables 2 and 3).

An analysis of symptom combinations associated with standardized World Health Organization (WHO) and the Centers for Disease Control and Prevention (CDC) definitions of ILI and positive influenza cases recorded in the study by total participants, age, gender, and season is provided in Supplementary File 5 (Supplementary Table 4). Of 21 481 participants reporting fever and cough symptoms, 4846 (22.6%; P < .001) tested positive to influenza, accounting for 86.0% of total cases; 3670 influenza cases were detected from 15 340 participants reporting fever, cough, and sore throat (23.9%; P < .001), and 3908 cases from 17 540 participants reporting fever and sore throat (22.3%, P < .001), accounting for 65.1% and 69.4% of positive influenza cases, respectively.

Table 3 summarizes the results of the symptom cluster analysis by total participants, children, and adults. Seven symptom clusters were auto-selected from the cluster analysis modeling. Frequencies of the symptoms analyzed in the 7 symptom clusters are provided in Supplementary File 5 (Supplementary Table 5). Among total participants, besides cough (Cluster 1; OR, 2.24; 95% CI, 2.04–2.46; P < .001), positive associations between influenza infection and Cluster 2 (headache, sore throat, and cough; OR, 1.64; 95% CI, 1.51–1.79; P < .001) and Cluster 3 (headache, joint pain, muscle aches, cough, and malaise; OR, 1.26; 95% CI, 1.11–1.42; P < .001) were identified. Among children with AUFI, positive associations were found in Cluster 1 (OR, 2.51; 95% CI, 2.25–2.82; P < .001) and Cluster 2 (OR, 1.42; 95% CI, 1.29–1.56; P < .001). Results of the symptom cluster analysis for influenza A and B are provided in Supplementary File 5 (Supplementary Tables 6 and 7). A summary matrix describing the association of symptom clusters varied by RIDT and rRT-PCR study test result combinations is provided in Supplementary File 5 (Supplementary Table 8).

Table 3.

Summary Matrix Describing Association of Auto-selected Symptom Clusters With Positive Influenza (A and/or B)

Symptoms (Including Fever)	Total Participantsa (29 467)			Children <18 ya (15 900)			Adults ≥18 ya (13 567)
	Negative (24 019) vs			Negative (11 769) vs			Negative (12 250) vs
	Positive (5448)			Positive (4131)			Positive (1317)
	ORb	95% CIb	P Valueb	ORb	95% CIb	P Valueb	ORb	95% CIb	P Valueb
Cluster.S1	2.24	2.04–2.46	<.001	2.52	2.25–2.82	<.001	1.78	1.48–2.15	<.001
Cough
Cluster.S2	1.64	1.51–1.79	<.001	1.42	1.29–1.56	<.001	2.07	1.69–2.54	<.001
Headache + sore throat + cough
Cluster.S3	1.26	1.11–1.42	<.001	1.19	0.97–1.46	.088	1.31	1.11–1.54	.001
Headache + joint pain + muscle aches + cough + malaise
Cluster.S4	1.04	0.87–1.25	.6	1.01	0.78–1.28	>.9	1.10	0.84–1.43	.5
Headache + malaise + cough + “other”
Cluster.S5	0.81	0.73–0.88	<.001	0.78	0.70–0.87	<.001	0.84	0.70–1.02	.076
Headache + cough + malaise + sore throat
Cluster.S6	0.94	0.76–1.14	.5	0.85	0.62–1.15	.3	1.01	0.77–1.32	>.9
Shortness of breath + headache + malaise + cough + sore throat
Cluster.S7	0.27	0.07–0.75	.030	0.38	0.06–1.36	.2	0.20	0.01–0.95	.12
Bloody stool + abdominal cramp + malaise

Abbreviation: OR, odds ratio.

^aExcludes 2020 data.

^bRegression model were adjusted with age, gender, and enrollment year.

An analysis comparing rapid influenza tests and rRT-PCR assay results (Supplementary File 6) over the study period identified an overall sensitivity of the rapid influenza tests of 0.492 (95% CI, 0.479–0.505) and a specificity of 0.993 (95% CI, 0.992–0.994) for both influenza type A and B among AUFI patients, based on a positive rRT-PCR assay considered the true-positive result. No significant difference was identified from the random effects modeling regarding the rapid tests’ capacity to diagnose influenza A or B among AUFI patients, including both sensitivity (P = .24) or specificity (P = .29).

DISCUSSION

This study utilized surveillance data over a 14-year period across 28 health facilities to describe the long-term epidemiology of influenza A and B among individuals seeking treatment for febrile illness throughout Cambodia. The detailed level of prospective laboratory, clinical, and epidemiological data collected has provided an opportunity to analyze and characterize key risk factors and associated symptoms of influenza and better understand influenza transmission within a developing country. The outcomes of this study provide insight for guiding influenza management, planning and targeting appropriate interventions, and supporting outbreak preparedness at local, national, and regional levels both within Cambodia and throughout Southeast Asia.

While fluctuations of influenza test positivity rates over time were observed, almost 1 in 5 (18.4%) study participants presenting with fever tested positive for influenza A or B over the course of the 14-year study period. Consistent with previous research describing the significant burden of influenza in East and Southeast Asia among febrile patients [22], this study demonstrates the ongoing public health effects of influenza in Cambodia and the need to understand and characterize the key drivers of transmission. Of note, positive associations between AUFI patients living in urban settings and primary school–aged children and the likelihood of testing positive for influenza, in particular influenza A, were identified even in the model adjusting for age and location. These results highlight the need for these populations to be prioritized when developing and implementing public health mitigation strategies for influenza virus and future surveillance efforts. Similarly, and consistent with other influenza surveillance results, this study recorded seasonal differences in percent positive rates of infection and identified a strong association between the likelihood of testing positive for influenza, particularly influenza A, and the rainy season, which highlights and builds evidence surrounding the seasonality of influenza transmission in Cambodia [10, 17] and within similar tropical and subtropical geographies of Southeast and Southern Asia [23]. Understanding these combined risk factors provides health programs with key data to support the effective targeting and timing of influenza interventions, including the possible implementation of nonpharmacologic measures such as masking or vaccination programs, as a means to disrupt transmission during expected peak periods [14].

During most of the study, the test positivity proportion was almost always higher for influenza A than influenza B. Yet, during parts of 2011, 2013, 2016, and 2019, this was not the case. Additionally, influenza A had a test positivity rate of at least 4% throughout the 14 years of the study, while influenza B occasionally decreased to as low as 0.1%. These results suggest that influenza A, unlike influenza B, while associated with the rainy season, remains a likely etiology for febrile illness throughout the entire year in Cambodia. However, both pathogens must be considered when developing preventive efforts to provide protection throughout a prolonged period.

A strength of this study was that it occurred before, during, and after the 2009 H1N1 pandemic. This novel type of influenza A entered Cambodia in 2009 and quickly became the most prevalent subtype in both 2009 and 2010 and again in 2016 and 2018. It continues to be detected each year, representing 6.6%–59.8% of rRT-PCR-detected positive influenza cases annually between 2009 and 2020 and 29.7% (1612/3353) of all influenza A cases since 2009. In 2020, this study was closed due to restrictions on movement during the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic, and cases were not evaluated. Therefore, a description of the COVID-19 pandemic's impact on influenza epidemiology including the changes in percentage of positive tests, lineages, and subtypes is beyond the scope of this manuscript. However, these prepandemic results can serve as a comparator to determine how SARS-CoV-2 influenced the type of influenza in circulation and clinical presentation since the pandemic's inception.

In addition to assessing the epidemiology of influenza virus in Cambodia, the second objective was to compare RIDTs and PCR as diagnostic modalities. Consistent with previous research assessing the performance of RIDTs [24–27], we found that the RIDTs were a nonsensitive but very specific test throughout the course of the study. Based on the overall prevalence of influenza in this study, together with the results of the comparative analysis of RIDTs vs PCR, for every 1000 patients presenting with febrile illness, 93 out of 184 suspected cases would not be identified if RIDTs were used alone without PCR. Using only RIDTs in a resource-limited setting without PCR accentuates the challenges in identifying individual cases and detecting the true burden of influenza. Since this population required symptoms including fever severe enough to present for clinical care to be a part of the study, in the general population the results could be worse than missing over half of the cases. For example, the results might be worse when considering asymptomatic and pauci-symptomatic individuals who do not present for care. Additionally, as reflected in the diagnostic results of this study, in the cases of dual influenza A and B infection, >73% of the time the results between PCR and RIDT were discordant. Using mixed methodologies will identify more instances when infection with both types of influenza occurs. Results from this study reinforce the understanding of the need to utilize RIDTs as diagnostic tools in conjunction with informed clinical judgment and ideally confirmatory laboratory testing where possible to adequately detect, treat, and control influenza within a target population.

Practical access to laboratory-confirmed influenza testing, especially in resource-poor settings, remains somewhat limited due to various prohibitive factors including cost, infrastructure needs, and operational capacity [28]. Difficulties associated with the clinical diagnosis of influenza due to the nonspecific signs and symptoms making the presentation similar to a large number of respiratory pathogens also present a significant challenge for frontline clinicians [6]. The current WHO case definition for ILI is an acute respiratory infection that includes a measured fever of ≥38°C and cough, with onset within 10 days [29], while the CDC definition includes fever of ≥37.8°C, together with cough and/or sore throat [30]. Results from this study identified cough as the most common symptom reported among AUFI patients with influenza, and cough and fever were the symptoms with the greatest association with a positive diagnosis, while malaise, nausea, and diarrhea were negative predictors. The results from Clusters 1 and 2 in our analysis support current WHO and CDC definitions, as well as previous research conducted in Cambodia [15]. An individual presenting with either of these symptom clusters will likely have a high pretest probability for influenza virus, which should then guide diagnostic interpretation even in the setting of a negative RIDT. Unlike in children, Cluster 3, which included joint pain, muscle aches, and malaise, was predictive of influenza in adults. Additionally, a significant association was identified between RIDT false-positive test results (ie, RIDT positive but PCR negative) and Cluster 3, which could lead to a potential misdiagnosis among febrile adults presenting with these specific symptoms when reliant on RIDT diagnostics alone. Of note, seasonal influenza vaccines and antiviral medications are rarely used in Cambodia and are unlikely to influence the clinical presentation of disease [16], allowing these results to reflect the natural history of infection. Evidence-based data that aid in characterizing the clinical symptoms both positively and negatively associated with influenza can support informing and strengthening effective frontline diagnosis and management.

This study does have some limitations. The study's design relied upon the analysis of passive surveillance data collected from febrile patients. Thus, the potential of missed asymptotic or mildly symptomatic infections among individuals not seeking health care is present and may lead to the underestimation of the true burden of influenza in the population as well as the inability to assess diagnostic test accuracy in this setting. Clinical severity was not assessed using a screening tool, also limiting the associations that can be made between symptoms among those presenting in our ambulatory setting and those with more severe disease who were unlikely to enroll. Other respiratory viruses were also not assessed due to cost constraints. Seasonal influences associated with a potential perceived understanding that the rainy season is associated with higher rates of infectious diseases such as mosquito-borne disease may have influenced the likelihood of someone with AUFI symptoms seeking care during this period. Future studies that incorporate active case finding surveillance approaches may lead to a more comprehensive understanding of the transmission dynamics of influenza within the population and identification of additional risk factors. Additionally, challenges associated with the translation of common medical words and a potential for cross-cultural differences in meanings may be seen as a limitation of the study, particularly in relation to the self-reporting of symptoms. Finally, almost one-third of the participants were 5–14 years old, creating a bias toward that age bracket.

Results from this long-term, prospective surveillance study utilized laboratory-confirmed influenza data, together with the collection of associated clinical systems, epidemiological data, and rapid diagnostic results, to characterize influenza transmission throughout Cambodia between 2007 and 2020. These results provide insight into the nature and burden of influenza throughout this pre–COVID-19 pandemic period and have aided in building an understanding of the key factors driving transmission and the populations most at risk of infection and identifying key clinical symptoms associated with influenza. Given the ongoing challenges associated with the swift and accurate frontline diagnosis of influenza, particularly in resource-poor settings, results from this study provide useful evidence-based data on the limited sensitivity of RIDTs. Influenza A and B are principal drivers of febrile illness, especially among children presenting for clinical care. Ongoing research is needed to improve our understanding of the clinical and epidemiological characteristics of influenza in Cambodia and serve to guide the development and implementation of targeted seasonal influenza prevention and control policies and programs within the region.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.