The effects of age on clinical characteristics, hospitalization and mortality of patients with influenza‐related illness at a tertiary care centre in Malaysia

Abstract Background Age is an established risk factor for poor outcomes in individuals with influenza‐related illness, and data on its influence on clinical presentations and outcomes in the South‐East Asian settings are scarce. The aim of this study was to determine the above among adults with influenza‐related upper respiratory tract infection at a teaching hospital in Malaysia. Methods A retrospective case‐note analysis was conducted on a cohort of 3935 patients attending primary care at the University Malaya Medical Centre, Malaysia from February 2012 till May 2014 with URTI symptoms. Demographics, clinical characteristics, medical and vaccination history were obtained from electronic medical records, and mortality data from the National Registration Department. Comparisons were made between those aged <25, ≥25 to <65 and ≥65 years. Results 470 (11.9%) had PCR‐confirmed influenza virus infection. Six (1.3%) received prior influenza vaccination. Those aged ≥65 years were more likely to have ≥2 comorbidities (P < .001) and were less likely to present with fever (P = .004). One‐third of those aged ≥65 years experienced hospitalization, intensive care admission or death within a year compared to 10% in the ≥25 to <65 years. Age ≥65 years was an independent predictor of hospitalization and death (OR = 9.97; 95% CI = 3.11‐31.93) compared to those aged <25 years. Conclusion Older patients in our cohort were more likely to have comorbidities and present with atypical features, with older age being an independent predictor of poor health outcomes. Our findings will now inform future health policies on older persons and economic modelling of adult vaccination programmes.


| INTRODUC TI ON
Influenza-associated respiratory infection is recognized as a major public health concern globally. 1 There are an estimated 300 000-650 000 deaths annually due to seasonal influenza-associated respiratory infections worldwide, with South-East Asia having the second highest estimated mortality rates (EMR) of 3.5-9.2 per 100 000 individuals, second to sub-Saharan Africa (2.8-16.5 per 100 000). 2 These rates increase significantly in the population aged more than 75 years with rates of 51.3-99.4 per 100 000 individuals. 2 The estimated global influenza death from respiratory infections published in 2018 appears to be higher than a previous estimate of 148 000-249 000 in 2013. 3 Furthermore, the influenza virus is also implicated in a large number of outpatient visits and subsequent hospitalization, accounting for 10%-26% of outpatient visits with influenza-like illness (ILI) and 6%-14% pneumonia hospitalization in East and South-East Asia. [4][5][6][7] Children, older persons, individuals with chronic respiratory diseases and the immunocompromised are considered high-risk groups for complications from influenza viral infections. 5 Clinical manifestations have also been known to differ with age, as older adults are less likely to report classical ILI symptoms. 1,8 Nevertheless, fever, cough and acute illness onset remain predictive of influenza among older adults. 8,9 Malaysia is a middle-income country in South-East Asia with a population of 31 million and a gross national income (GNI) per capita of USD 9800 as of 2017. 10  Disability rates in older Malaysians are significantly higher at 16% compared to developed countries such as Canada which only reported a 6% disability rate when the same assessment tool for activities of daily living was employed. 13 To our knowledge, there is no published report of vaccine uptake among older adults in Malaysia, but vaccine sales figures from the H1N1 pandemic reported sales of 1408 per 100 000. 14 Published data on clinical features and health outcomes associated with influenza infection in Malaysia remains scarce and even more so in older Malaysians with differing disability and vaccination rates. 15 Therefore, we aimed to examine differences in clinical characteristics, clinic visits, hospitalization and death in younger, middle-aged and older Malaysians with laboratory-confirmed influenza diagnosed during a primary care visit with upper respiratory tract symptoms.

| Study setting and population
This was a retrospective cohort study based on a laboratory culture data set collected from February 2012 to May 2014 in University of Malaya Medical Centre (UMMC), a tertiary teaching hospital in Kuala Lumpur, Malaysia. The study had received ethical approval from the local institutional review board (MREC ID: 2016714-4027).
A total of 3935 patients who presented to the primary care clinic during the period of the study with symptoms associated with URTI (sneezing, nasal discharge, nasal obstruction, headache, sore throat, hoarseness of voice, myalgia and cough) were recruited.
Nasopharyngeal swab samples were then obtained and transported in universal transport medium to the laboratory for reverse transcription polymerase chain reaction (RT-PCR) for detection of respiratory viruses. The study methods are described in an earlier article published elsewhere. 16 Individuals who tested positive for influenza viruses were included in this study.

| Clinical data collection
Sociodemographic and clinical data including age, gender, ethnicity, duration of illness at presentation, clinical features, comorbidities, smoking history, vaccination history, subsequent outpatient visits and hospitalization were extracted from the hospital Electronic Medical Record system, using a standardized data collection document (Table S1). Within the primary care department, all patients will have blood pressure measured routinely using automated oscillometric blood pressure monitors on arrival by the clinic nurse prior to their consultation. The blood pressure measurements were then recorded to the nearest mm Hg using an Electronic Medical Record template. Vital status and dates of death were obtained from the National Registration Department.

| Study instruments
Total nucleic acids (NA) from nasopharyngeal swab samples were extracted using the NucliSENS easyMAG automated nucleic acid extraction system (BioMérieux). 17 The xTAG Respiratory Virus Panel (RVP) FAST multiplex RT-PCR assay (Luminex Molecular Diagnostics, USA) was then used to detect the influenza viruses (A and B) and other respiratory viruses. 18 The procedures above were performed according to the manufacturer's protocol.

| Data analysis
The sample size of the study was based on all consecutive patients presenting with ILI to the primary care department of the teaching hospital within the two year period, the minimal period required to study seasonal changes. A retrospective calculation was conducted to determine the power of this study. Assuming that 10% of individuals <65 years of age were either hospitalized or had died within 1 year of presentation, a total sample size of 238 patients will provide 80% power to detect a trebling in relative risk.
The sample population was divided into three age groups: Group 1 (<25 years), Group 2 (25 years-64 years) and Group 3 (65 years and over), to allow for direct comparisons with global influenza surveillance reports. 19 The age group of 0-14 was excluded as this study only involved adults. Data were presented as frequencies with percentages in parentheses, mean with standard deviation in parentheses or median with interquartile ranges where appropriate.
The independent Student's t test, Mann-Whitney U, chi-squared and Fisher's exact tests were used where appropriate for group comparisons in sociodemographic, number of comorbidities, clinical characteristics and outcomes of influenza-associated infection across different age groups. Logistic regression analyses were then conducted to determine the influence of age on clinical outcomes of influenza-associated infection using a composite variable of hospital admission, with or without intensive care unit (ICU) admission or deaths (underlying influenza, pneumonia, respiratory or circulatory deaths), after separate adjustments for potential confounders, such as number of comorbidities and presence of chronic conditions, and the presence of fever. In this analysis, Group 1 was considered the reference category, and comparisons were made between Group 2 and Group 1, and Group 3 and Group 1.

| Sociodemographics
Of the 3935 samples collected during the study period, samples from  (Table 1).
Smoking was only reported in 5.6% of our sample. Only six (1.3%) patients had a history of influenza vaccination, with only one of them being in the at-risk group. No individual over the age of 65 years had previously received any influenza vaccination.

| Clinical characteristics and age
The clinical symptoms and blood pressure of all included cases, according to the three age categories are summarized in Table 2. Four hundred and twenty-four (94.2%) of all cases presented within 7 days of symptom onset. There was no difference in duration of illness with age categories. Cough (84.7%), 63.6% of which was productive, and fever (81.3%) were the most commonly reported clinical symptoms. Nasal discharge and sore throat were reported in 59.6% and 45%, respectively.
Differences in the proportion of cases with fever, headache, reduced oral intake, systolic and diastolic blood pressure exist between  Hospital admission (with or without ICU admission) or death was reported in 47 (10.4%) of the overall sample. Fifteen of the 53 (28.3%) individuals in the aged 65 years and over age group were admitted to hospital or died within 1 year compared to four individuals in the <25 years age group and 28 (9.6%) of those aged 25-64 years (P < .001). All five deaths occurred in those aged 65 years and over.

| Logistic Regression
The odds ratios (OR) and 95% confidence intervals (CI) examining the association between the composite outcome of hospitalization and death are presented in Table 4. As mentioned earlier, comparisons between groups were made using dummy variables with the under 25 age group as the reference category. First, the unad-

| D ISCUSS I ON
Influenza is a major infectious disease that causes substantial morbidity and mortality worldwide every year. Better understanding of the disease burden among clinicians could lead to increased uptake of appropriate vaccinations in high-risk groups and reduce the cases of influenza infection, particularly in South-East Asian countries such as Malaysia, where awareness towards influenza vaccination remains low. 14 In this study, we evaluated the influence of age on clinical features and outcomes of influenza infection particularly among a mostly unvaccinated group with a 1% influenza vaccine uptake.
Our findings overall showed that cough is the most common clinical finding, accounting for approximately 85% of cases whereas fever and nasal congestion were present in >80% and >50% of cases, respectively. These features are consistent with the World These staggering numbers support the call for vaccination and public health policies, and the development of better vaccines that offer enhanced protection for older persons. 34 The likelihood of concurrent comorbidities and chronic conditions increases with age. 35 The overall poor uptake of influenza vaccination by our study population and the older age group is apparent. Influenza vaccination has been shown to be beneficial to older community-dwellers and nursing home residents. 40,41 It also reduces the rate of hospitalization for pneumonia and influenza by 48%-57%, and for all respiratory conditions by 27%-39%. 40 While national guidelines for adult vaccinations do exist, 42 government, insurer or employer funded adult vaccination programmes currently do not.
Despite free health care for older persons in public health facilities, 43 the annual influenza vaccination is funded out-of-pocket, with low awareness for the need to vaccinate, among healthcare workers and patients in high-risk groups. This highlights the reactive nature of the current health service, with lack of emphasis on preventive measures. 44 Therefore, the findings of this study will provide valuable information for economic modelling and to inform future policy makers on future decisions in preventive healthcare funding.
The findings of this study were limited by its retrospective design with data collection based on prior documentation. Hence, if symptoms were not documented, they were assumed absent, potentially underestimating symptom burden. In the analysis of our data, we have included all patients with influenza-positive results, including those with co-infections with other viruses. Where multiple viruses were detected, it would not be possible to discern if the virus was causal or non-causal. Assumptions were also made that patients who did not have any documented influenza vaccinations in our electronic medical records, were unvaccinated. Lastly, all patients were recruited from a single site, and therefore, the findings of this study could not be generalized to the rest of population.

| CON CLUS ION
Older patients with influenza URTI in our cohort were less likely to present with fever, with an increased risk of the adverse outcomes of hospitalization or death. Uptake of influenza vaccination is 1% overall and 0% in the older age group. Future studies should evaluate this relationship prospectively in a representative sample and should also consider evaluating the benefits of vaccination in reducing healthcare burden. Our findings can be used for economic modelling as well as to highlight the major health burden associated with influenza infection within a developing nation with an equatorial climate.