Respiratory Viral Infections and the Risk of Rheumatoid Arthritis

Young Bin Joo; Youn-Hee Lim; Ki-Jo Kim; Kyung-Su Park; Yune-Jung Park

Disclosures

Arthritis Res Ther. 2019;21(199) 

In This Article

Results

Baseline Characteristics

From January 2012 to December 2013, the total number of patients newly diagnosed with RA was 24,117. Of these, 18,688 (77.5%) were females, and the mean age at RA diagnosis was 54.7 (SD 13.2) years (Table 1). Almost all patients (95.0%) had a national health insurance. The proportion of institutions that reported patients who were diagnosed with a new RA was 44.6%, 23.8%, and 31.6% in clinics, general hospitals, and tertiary hospitals, respectively (Additional file 1: Table S1).

Seasonal Tendency of Incident RA

The analysis revealed a seasonal tendency of incident RA in each year (Figure 1). In 2012, the number of incident RA increased from January (n = 781) to July (n = 1274) then decreased to December (n = 775) (Additional file 1: Table S2). July had the highest number of incident RA cases, and December had the lowest. We observed a similar seasonality in 2013. In that year, the number of incident RA cases increased from January (n = 1057) to July (n = 1212) then decreased to December (n = 853). The differences in the weekly mean number of incident RA cases were compared between the top peaks in July and other months. A significantly lower number of weekly incident RA cases were observed from September to February in both 2012 and 2013 (Additional file 1: Table S2). Men and women showed the same seasonal pattern for the number of incident RA (Additional file 1: Figure S1).

Figure 1.

Incident rheumatoid arthritis trends in 2012 and 2013. The x-axis represents the study period from 2012 to 2013. Because respiratory virus data (exposure) are provided weekly, the number of incident RA cases (outcome) is also represented weekly from the first week of January 2012 to the last week of December 2013. The y-axis represents the weekly number of incident RA cases in each week

Figure S1.

Incident rheumatoid arthritis trends according to sex in 2012 and 2013.
X-axis represent the time of study period from 2012 to 2013. Y-axis represent the monthly number of incident RA.

Effects of Respiratory Viral Infections on Incident RA

Most viruses showed seasonal tendencies between 2012 and 2013 (Additional file 1: Figure S2). Outbreaks of parainfluenza virus and bocavirus tended to occur in late spring and early summer. RSV tended to be diagnosed in the autumn. Outbreaks of influenza occurred in winter, and coronavirus tended to be diagnosed from November to December. Rhinovirus was prevalent from spring to autumn, but only exhibited an outbreak in the autumn of 2012, not in the autumn of 2013. Metapneumovirus was infrequent but tended to be diagnosed in late winter and spring. Adenovirus did not exhibit any seasonal tendencies.

Figure S2.

The weekly detection rate of eight respiratory virus in 2012 and 2013
The X-axis represents the study period from 2012 to 2013. As the detection rates for respiratory viruses are collected on a weekly basis, week units from the first week of January 2012 to the last week of December 2013 are represented. The Y-axis represents the weekly detection rate for the virus in each week.

The associations between the detection rate of respiratory viral infections and the number of incident RA were investigated using GAM. At lag1w, fluctuations in associations were observed for all viruses except bocavirus (Additional file 1: Figure S3). As the lag structure changed from lag1w to lag1–8w, however, the fluctuations in the associations gradually disappeared and the association became more apparent and more linear for some viruses. This suggests that ambient respiratory viral infections are associated with an increased number of incident RA after a moderate time interval, rather than having an immediate effect.

Figure S3.

Risks of incident RA associated with infection with eight respiratory viruses over 8 lag weeks.
The X- and Y-axes represent the weekly virus detection rate as a percentage and difference from the mean log relative risk of incident RA, respectively. Solid lines represent associations between the weekly virus detection rate and weekly number of incident RA cases and dashed lines represent 95% confidence intervals for the risk.
Each row represents various types of respiratory virus (adenovirus, parainfluenza virus, respiratory syncytial virus, influenza virus, corona virus, rhinovirus, bocavirus, and metapneumovirus).
Each column represents various moving averages for lag weeks of viral infections. For example, 'Lag1w' represents the previous one week of viral infections and 'Lag1-2w' the moving average of the previous two weeks of viral infection.

Three viruses were significantly associated with the number of incident RA in time-series analysis (Table 2). The number of incident RA increased by 4.8% with an incremental increase of 1% in the parainfluenza virus detection rate at lag1–7w (P = 0.003). At lag1–6w, coronavirus and metapneumovirus were associated with respective increases in the number of incident RA of 9.2% (P < .001) and 44.0% (P = 0.038) with an incremental increase of 1% in virus detection rate. The visualized associations with integrated smoothness estimation of these respiratory viruses are shown in Figure 2a–c.

Figure 2.

Effects of ambient respiratory viral infections on incident RA. Parainfluenza (a), coronavirus (b), and metapneumovirus (c) infection. The x- and y-axes represent the weekly virus detection rate as a percentage and difference from the mean log relative risk of incident RA, respectively. Solid lines represent the associations between the weekly virus detection rate and weekly number of incident RA cases, and the gray area represents 95% confidence intervals for the risk

Subgroup Analysis

In women, all three viruses were significantly associated with incident RA (P < .05), but in men, only parainfluenza virus was (P = 0.019) (Table 3). Elderly-onset RA in patients aged ≥ 60 years was significantly associated with all three viruses (P < .05). Only parainfluenza virus was significantly associated with incident RA in patients aged under 40 years (P = 0.034), and in patients aged 40–59 years, none of the three viruses was significantly associated with incident RA. The presence or absence of preexisting respiratory diseases showed a different association with each virus; that is, coronavirus was only associated with the number of incident RA cases in the absence of other respiratory conditions, while metapneumovirus showed the opposite results, i.e., a significant association (79.18% change in risk per 1-unit change in exposure, P = 0.034) in the presence of a respiratory disease, but not in disease absence (23.87% change in risk per 1-unit change in exposure, P = 0.352).

processing....