Incidence, symptoms and medical care for influenza virus and respiratory syncytial virus illnesses among older adults in Eastern China: Findings from the China Ageing Respiratory Infections Study (CARES), 2015-2017

INTRODUCTION

Respiratory virus infections like influenza virus and respiratory syncytial virus (RSV) are major causes of acute hospitalizations and deaths. In temperate climates infections occur mainly during winter, where in tropical and subtropical climates infections may occur year-round.¹ In older adults, while the potential severity of influenza illness is commonly known,² studies suggested RSV severity may be similar or greater than influenza.³ The World Health Organization recommends prioritizing older adults to receive influenza vaccination but uptake remains low in many locations.⁴ More recently RSV vaccines for older adults were approved in some countries.⁵

Vaccination polices require knowledge of influenza and RSV disease burden. Community-based cohort studies^6–10 can actively identify respiratory illnesses not captured by routine health systems. However, most of these studies have been conducted in households with children^6,11–13 and in high-income countries.¹⁴ Fewer were among older adults^15,16 including those aged ≥80 years, a group vulnerable to severe complications from these infections,¹⁴ and in low-and-middle income countries.^12,16–18 To address this gap, we established a community cohort of older adults in Jiangsu Province, China to characterize the incidence, symptom profiles, and medical care of PCR-confirmed influenza virus and RSV illnesses from December 2015–September 2017.¹⁹

METHODS

Study design

We monitored respiratory illnesses in a prospective longitudinal cohort of older adults aged 60– 89 years in Jiangsu Province, China (Figure 1A) from December 2015–September 2017 (Figure 1B), A detailed cohort description was previously published.¹⁹ In brief, we screened and enrolled older adults in New District and Xiangcheng in Suzhou and Economic Development Zone (EDZ) and Tinghu in Yancheng, representing urban and semi-urban areas with different socio-economic development levels in eastern China. We recruited older adults in three age groups (60–69, 70– 79, and 80–89 years old) with oversampling among those aged 80–89 years. At enrolment, we collected information on demographics, household size, influenza vaccination history, smoking status, medical history, cognitive functioning, and self-rated health and functioning.²⁰ For cognitive functioning, two measures were used: the Mini-Cog (3-item recall) and Standardized Mini-Mental State Examination (SMMSE).^21,22

• Download figure
• Open in new tab

Figure 1.

(A) CARES study sites – Yancheng and Suzhou, Jiangsu province, mainland China. Blue: Jiangsu province; Green: Yancheng city; Pink: Suzhou city. (B) CARES timeline including screening period, enrollment period, and active surveillance for acute respiratory illnesses (ARI). (C) Weekly numbers of influenza and RSV PCR-negative ARI, PCR-confirmed influenza ARI, and PCR-confirmed RSV ARI. Blue: PCR-confirmed influenza ARI; Red: PCR-confirmed RSV ARI; Grey: Influenza and RSV PCR-negative ARI; White: ARI but no respiratory specimen collected.

We conducted year-round active surveillance for acute respiratory illness (ARI) (Appendix Methods) because influenza viruses are known to circulate beyond the typical winter months in eastern China, and we expect RSV to circulate mostly in winter seasons.^1,23 During weekly telephone calls, we asked if the participant was experiencing respiratory symptoms. Illness episodes with onset ≥14 days apart were considered separate episodes. Home visits were arranged for new ARI episodes to collect respiratory specimens (combined mid-turbinate and oropharyngeal swabs) within 7 days of symptom onset for laboratory confirmation of infections; study staff also administered an acute illness questionnaire that recorded illness symptoms and whether medical care was sought. Ten days after illness onset, a follow-up questionnaire on illness outcome was completed by telephone. Respiratory specimens were placed in coolers at 2– 8°C and transported to the Suzhou Center for Disease Prevention and Control (CDC) or Yancheng CDC laboratories within 24 hours of collection. Specimens were then stored at -80°C until testing.¹⁹

RESULTS

We approached 2,280 older adults and screened 1,914 (84%). Of those, 1,573/1,914 (82%) met inclusion criteria, 1,532/1,573 (97%) were enrolled, and 1,527/1,532 (97%) initiated acute respiratory illness surveillance.¹⁹ As planned, distribution across participant age groups was similar: 60-69 years (n=547, 36%), 70-79 years (515, 34%), and 80–89 years (465, 30%) (Table 1). Most participants lived in urban (24%) or semi-urban (43%) areas and had no formal education (51%) or only attended primary school (34%). At enrollment, most participants described overall health status as excellent, very good, or good (59%) compared with fair or poor (40%). Sixty-nine percent of participants self-reported ≥1 chronic health condition and 15% reported hospitalization in the past two years. Assessing cognitive function, almost all participants (98%) recalled three items, with the remaining recalling 1-2 items and having the ability to complete a clock drawing test.

From 27 December 2015—2 September 2017, 1,337/1,527 (88%) participants completed active surveillance follow-up. Among the 190 (12%) lost to follow-up, 84 (44%) died, 7 (4%) withdrew due to deteriorating health and 81 (43%) withdrew for other reasons (Appendix Table 2).

Approximately 30% of follow-up contacts to participants were through a proxy who had daily interaction with the participant. From 29,551 and 29,822 successful contact person-weeks reported from New District and Xiangcheng in Suzhou, and 20,036 and 41,075 person-weeks from EDZ and Tinghu in Yancheng respectively (Appendix Tables 3 – 6), were identified 1,043 episodes of acute illness, of which 898 (86%) reported by 522 participants met the ARI definition. We collected respiratory specimens for 840 (94%) ARI episodes, 95 (11%) were PCR-confirmed influenza virus infections (“influenza ARI”) and 22 (3%) were RSV infections (“RSV ARI”), including 4 co-infections. Two participants (0·1%) had two separate influenza ARIs (Table 2). We identified 56 (59%) influenza and 16 (73%) RSV ARIs from December— February, compared with 12 (13%) and none during June—September (Figure 1C). The temporal pattern of influenza detections in our cohort was consistent with community surveillance data over the same period (Appendix Figure 2). There were 26 (1·7%) ARI- associated hospitalizations; four were influenza ARIs and one was an RSV ARI (Table 2). One death associated with RSV ARI was observed and zero for influenza ARI (Table 2).

• Download figure
• Open in new tab

Figure 2.

(A) Proportion of reported symptoms in participants with PCR-confirmed influenza ARI (N=93), PCR-confirmed RSV ARI (N=20), or influenza and RSV PCR-negative ARI (N=708). (B) Mean symptom severity score for reported symptoms and symptom categories (upper respiratory, lower respiratory, or systemic symptoms) in participants with PCR-confirmed influenza ARI, PCR-confirmed RSV ARI, or influenza and RSV PCR-negative ARI. Blue: PCR- confirmed influenza ARI; Red: PCR-confirmed RSV ARI; Grey: Influenza and RSV PCR- negative ARI. * p-value ≤0·05

Of the 95 influenza virus infections, 86 were influenza A, 8 were influenza B, and one was a co- infection with influenza A and B viruses. Influenza A subtyping was completed for 63 specimens; 9 (14%) were A(H1N1) and 54 (86%) were A(H3N2). Twenty-three specimens were not subtyped due to high PCR cycle thresholds. Four influenza B viruses were B/Victoria, three were B/Yamagata, and two were untyped. The most commonly detected virus in Year 1 was A(H1N1) and in Year 2 was A(H3N2) (Appendix Figure 2).

We estimated the cumulative incidence (Table 3) and incidence rate (Appendix Table 7) by age group, and weighted overall estimate, of PCR-confirmed influenza ARI or RSV ARI in each study year. In Year 1, the overall age-weighted incidence of influenza ARI among older adults ≥60 years was 0·8% (95% CI:0·3–1·4), and that of influenza ARI-associated hospitalizations was 0·1% (95% CI:0–0·3). In Year 2, the age-weighted influenza ARI and influenza ARI- associated hospitalization incidences were 6·1% (95% CI:4·7–7·7) and 0·3% (95% CI:0–0·8) respectively. For RSV, in Year 1 the overall age-weighted incidence of RSV ARI was 0·5% (95% CI:0·1–1·0), with no hospitalizations observed. In Year 2, the age-weighted RSV ARI and RSV ARI-associated incidences were 1·0% (95% CI:0·5–1·6) and 0·1% (95% 0–0·2) respectively. Significant differences between age groups were not observed (Appendix Figure 3).

Acute illness surveys were completed for 822 (92%) ARI episodes from 487 participants. Comparing ARIs of different virus aetiologies (Table 4, Figure 2, Appendix Figure 4), subjective feverishness was more common in influenza ARI (55%) compared to RSV ARI (30%, p=0·03) or PCR-negatives (44%, p=0·04). Influenza ARIs met the ILI case definition (24%, p=0·04) significantly more often than PCR-negative ARIs (16%), but not statistically different for RSV ARIs (15%). The sensitivity of the ILI case definition for influenza ARI detection was 24% (95% CI: 15%, 34%) and 15% (95% CI: 3%, 38%) for RSV ARIs; the specificity was 83% (95% CI: 81%, 87%) and 83% (95% CI: 80%, 86%), respectively. Older adults with influenza ARI reported higher symptom severity scores for subjective fever (Appendix Table 8) and worsened shortness of breath (Appendix Table 9). There were no differences in symptom severity scores by upper respiratory, lower respiratory, or systemic symptom categories. The mean ratings for overall health or ability to do normal activities during the illness were also similar across aetiologies (Table 4). Illness duration for an influenza ARI (12·5 days, 95% CI: 11·4, 13·8) was similar to RSV ARI (14·1 days ,95% CI: 10·9, 18·2), but statistically longer (p=0·02) than PCR-negative ARIs (11·0 days, 95% CI: 10·7, 11·3). The number of days spent in bed for influenza ARI (1·2 days, 95% CI: 0·8, 1·7) was also similar to RSV ARI (1·4 days, 95% CI: 0·5, 2·4), but statistically more (p=0·04) than PCR-negative ARIs (0·7 days, 95% CI: 0·6, 0·8).

Participants sought medical attention in 67% (549/822) of ARI episodes (Table 4). Ambulatory care (outpatient care or visit to an emergency room) was common for both influenza and RSV ARIs, with outpatient care statistically more likely for influenza ARI (80%) compared to RSV ARI (50%, p=0·01) or PCR-negative ARI (62%, p≤0·001). Participants with influenza ARIs (4·3%, p=0·28) and RSV ARIs (5·0%, p=0·32) were hospitalized more frequently than those with PCR-negative ARIs (2·4%) although the difference was not significant. ARI episodes were classified by the highest level of care obtained, the level of care differed significantly by age groups for influenza ARI (p=0·05) and PCR-negative ARI (p=0·03) (Appendix Table 10), with a trend of less hospitalizations in older age groups for influenza ARI and a reverse trend for PCR-negative ARI (Appendix Figure 5).

Among all ARIs, 71% reported using any medication, with significantly higher use among influenza ARI (84%) than other ARI (Table 4). Antibiotic use was reported in 53% of all ARI episodes, with significantly higher use among influenza ARIs (65%) and RSV ARIs (70%) than PCR-negative ARIs (51%).

DISCUSSION

We estimated the incidence of influenza and RSV ARI, described symptom profiles, and medical care seeking behavior in >1,500 older adults in eastern China between 2015–2017. This was one of the largest prospective observational studies to actively identify laboratory-confirmed influenza and RSV illnesses in older adults. We observed a higher incidence (6·1%) of PCR- confirmed influenza ARI in 2016–17 (study year 2) when circulation of influenza A(H3N2) virus predominated; and a lower incidence (0·8%) in 2015–16 (year 1), based on only a partial study year of active surveillance when circulation of influenza A(H1N1) virus predominated. We estimated PCR-confirmed RSV ARI incidence ranged from 0·5–1·0% in both study years. While our study was not powered to estimate hospitalization incidences, our findings suggested cumulative incidences of 0·1–0·3% and 0–0·1% for hospitalization during acute influenza ARI and RSV ARI, respectively in community-dwelling older adults ≥60 years; and no hospitalizations were observed in those ≥80 years for either influenza or RSV ARI in our cohort. Understanding the effect of influenza virus and RSV infections on older adults provides an evidence base to inform prevention and control measures, including vaccine programs.

Community-based studies of influenza incidence have mostly focused on all ages^12,17,18 and few focus on older adults specifically.¹⁵ Tinoco et al. identified ILI, defined as fever or feverishness plus cough or sore throat, by three times weekly home visits in about 500 older adults ≥65 years old from four sites in Peru between 2009–15, with reported influenza vaccination coverage of 26 per 100 person-years.¹⁷ Prasert et al. identified ARI, defined as new onset of cough or worsening of chronic cough regardless of fever, by weekly calls or home visits in about 1,500 unvaccinated adults ≥65 years in a prospective observational cohort on influenza vaccine effectiveness in Thailand between 2015–17.¹⁵ Cohen et al. identified infection by nasopharyngeal sampling twice weekly irrespective of symptoms in about 50 older adults ≥65 years (among 1116 participants of all ages from 225 households with almost no influenza vaccination) in South Africa between 2016–18.¹⁸ Compared to these unvaccinated populations (except Peru), in our study of 1500 older adults ≥60 years, the estimated A(H1N1) ARI incidence (of 1·3/100 person-years) in study year 1 (a partial season) was less than the incidence (based on symptomatic and asymptomatic infections combined) estimated in South Africa (2·3/100 person-years during 2017–18)¹⁸ and higher than influenza illness in Thailand (<0·3/100 person-years in 2015-16 and 2016-17 seasons).¹⁵ In study year 2, our estimated A(H3N2) incidence (of 6·5/100 person-years) was higher than Peru (3·6/100 person-years)¹⁷ and Thailand (<2/100 person-years),¹⁵ but lower than South Africa (13·6/100 person-seasons during 2017–18).¹⁸ Thus, our virologic-confirmed influenza incidence estimates fell within the range previously described for community-dwelling older adults, and on the higher end when A(H3N2) viruses predominated. Our study also provided valuable data on influenza ARI-associated hospitalizations in community-dwelling older adults aged ≥80 years, that despite actively identifying ARI we did not observe more influenza ARI- or RSV ARI- associated (acute) hospitalizations in them than those younger between 60-79 years old, although our study was underpowered to observe statistical differences between age groups.

With RSV vaccines now available for older adults, data on RSV incidence in community- dwelling older adults can inform population impact and models evaluating the potential benefits of RSV vaccination programs in older adults, although few of these studies exist, particularly before the COVID-19 pandemic.^6,8,16,28,29 Prior studies identified RSV illness by passive surveillance only²⁹ or evaluated incidence based on both virologic and serologic confirmed infection.⁸ Korsten et al. (RESCEU study) assessed annual RSV community burden in about 500 older adults :260 years old from three cities in Europe between 2017–19 by weekly emails or calls to identify ARI, which was defined as having at least one of the following symptoms: cough, nasal congestion or discharge, wheezing or shortness of breath.²⁸ Our study, with active surveillance in >1,500 older adults, found that 0·5–1% experienced virologic-confirmed RSV ARI, 0·3–0·6% had medically attended RSV ARI, and 0–0·1% had RSV ARI-associated hospitalization during 2015–17 in China. In comparison, Korsten et al. reported higher cumulative incidences of virologic-confirmed RSV ARI (2·1% and 4·9% during 2017-18 and 2018-19 October–May seasons respectively) and medically attended RSV ARI (0·8% and 1·4% respectively); and even higher estimates if serologic-confirmed infections.²⁸ Our observed lower RSV incidences may be due to differences in surveillance methods, laboratory testing, or true epidemiological variations. Future analyses of our study serum samples¹⁹ could identify additional RSV infections.

Our study suggested older adults with influenza ARI had longer illnesses than those with PCR- negative ARI, consistent with findings in older adults in RESCEU ²⁸ and other working age adults.²⁶ We also observed influenza ARI were more likely to include fever compared to PCR- negative ARI or RSV ARI. More influenza ARI met the ILI case definition; and while the sensitivity of the ILI definition to detect either influenza (24%) or RSV (15%) ARI was low, it was relatively higher for influenza ARI, consistent with earlier findings.¹⁰ This finding suggests the ILI case definition is less effective for identifying RSV in older adults. To capture RSV burden in community-dwelling older adults through routine public health surveillance systems, a new syndromic case definition may be needed.

Over half of our participants with influenza or RSV ARI sought medical care. We observed influenza ARI were more likely to require ambulatory care compared to PCR-negative ARI or RSV ARI, as also reported by Korsten et al.²⁸ Influenza vaccine coverage was very low (0·4%) in this population in Jiangsu Province, similar to those in other Chinese cities where influenza vaccines require out-of-pocket payment. Future studies are needed to assess the potential benefit of influenza and RSV vaccination on reducing infections and burden on the healthcare system, and methods that increase influenza vaccine uptake among older adults in China.

Antibiotic use in 9% of all ARI and overuse in 31% of influenza ARI was reported in Europe.²⁸ Here, antibiotic use was common in all ARI (53%), particularly for influenza (65%) and RSV (70%) ARI, indicating the broad overuse of antibiotics for non-bacterial infections. Although a decline of antibiotic consumption in health care settings particularly in community care was reported in Shandong Province from 2012–17,³⁰ our study suggested antibiotic overuse was still prevalent in the community, and efforts to promote rational use of antibiotics in primary care should be reinforced.

Our study has several limitations. First, there may have been under-detection of illnesses and PCR-confirmed infections, especially in the first study year due to influenza virus circulation before active surveillance began. Second, if a proxy reported on behalf of the participant, some illness episodes may have been missed. However, proxies likely reported most illness episodes as they were required to have daily contact with participants. Finally, we evaluated symptoms and severity within four days of reported illness, potentially missing later-developing symptoms.

Despite these limitations, our study has several important public health implications. We found a higher incidence of PCR-confirmed influenza (mainly from H3N2) in older adults than most previous studies. When the incidence of influenza ARI increased in study year 2, the burden on the healthcare system also increased with more medical care-seeking by older adults, reflecting the effect of seasonal influenza epidemics on healthcare utilisation at the population level. While some cities provide free influenza vaccination for older adults, it is not included in the national immunization program in China and most must pay out-of-pocket. Improving prevention measures such as influenza vaccination in this vulnerable population with extremely low vaccine uptake can reduce infections and healthcare burden. We estimated the incidence of PCR- confirmed RSV illness in older adults in both rural and urban settings in China, providing needed data on RSV infections and burden in older Asian populations to inform the need for RSV vaccination programs. We observed widespread use of antibiotics for viral infections.

CONTRIBUTORS

All authors meet the International Committee of Medical Journal Editors criteria for authorship. This project is a multi-national collaboration between investigators in the US, Hong Kong, and mainland China, and has required substantial contributions from the co-investigators. In terms of the specific contributions, the study protocol was designed by BJC and MGT and drafted by CX. Substantial input was provided by NHLL, HZ, JZ, FT, LL, BZ, CL, QD, YS, RW, PS, CG, FH, LK, and ADI on the study design in the planning phase in a series of meetings in the US, Hong Kong, and mainland China. The study instruments were prepared by NHLL, CX and RW with input from HZ, JZ, LL, BZ, YC, YS, PS, CG, FH, LK, BJC, MGT and ADI. Study staff training were designed and performed by NHLL, HZ, LL, BZ, QD, CX, YC, YS, RZ, RW, CG and HY. Study activities were managed and organized by HZ, JZ, LL, BZ and YC with oversight by NHLL and BJC. The laboratory tests were developed and performed by GC, DKWC, and HZ. Statistical analyses were done by NHLL. The initial draft of the present manuscript was prepared by NHLL and ADI with input from MGT and BJC, and all authors provided critical review and revision of the text and approved the final version.

DECLARATION OF INTERESTS

BJC has consulted for AstraZeneca, Fosun Pharma, GSK, Haleon, Moderna, Novavax, Pfizer, Roche, and Sanofi Pasteur. The authors report no other potential conflicts of interest.