The first patient in this cluster who had laboratory-confirmed MERS was a 27-year-old woman who worked as a janitor in a women-only university in Riyadh. She reported experiencing dry cough and fatigue on October 1, 2015; she sought care at a private healthcare clinic on October 4 and was provided treatment and sent home the same day. On October 7, after signs and symptoms worsened to include fever, shortness of breath, productive cough, and signs of pneumonia, she again sought care in the same healthcare clinic, and a diagnosis of MERS was suspected. On October 8, a nasopharyngeal sample was collected and the patient was transferred to a public hospital in Riyadh, designated for isolation and treatment of MERS patients. MERS-CoV infection was confirmed on October 9. A second case in this cluster has recently been described (22).

Figure 1

Figure 1. Schematic of expatriate dormitory (the residence, buildings 1–12) and MERS-CoV infection attack rates (IARs), Riyadh, Saudi Arabia, 2015. Each building contained 2 villas on 3 floors. The distance between buildings is...

The first patient resided in an enclosed, women-only, expatriate dormitory composed of 24 villas (Figure 1). Each villa is a 3-story building with 7 bedrooms (2 on the ground floor, 3 on the first floor, and 2 on the second floor) and is inhabited by 24–50 women. On inspection of the living quarters, the field team found that most of the windows in the bedrooms were closed and sealed and that ventilation within the bedrooms was poor. Initial open-ended interviews with some residents informed the study team that residents shared the same kitchen and dining room within the villa but did not typically eat together or share food at mealtimes. There were no designated social spaces; however, residents reported gathering around laptops to watch movies together.

A total of 828 women who lived in the residence complex were included in the seroepidemiologic study; none of the eligible women refused to participate. All participants were female, and median age was 35.1 (26.6–41.3) years. None were Saudi Arabia nationals; they were from the Philippines (84.6%), Sri Lanka (6.4%), Indonesia (2.9%), Nepal (1.6%), and India (1.1%) (Table 1). A total of 49 participants (1 case-patient and 48 non–case-patients) reported having >1 chronic condition (e.g., asthma, diabetes, heart disease, hypertension, breast cancer) (Table 1). The MERS case-patient reported having asthma; among non–case-patients, the most common chronic conditions reported were asthma (31%), diabetes (25%), and hypertension (18%).

In terms of occupation, almost half (49.1%) of participants reported working at the women-only university in Riyadh, including 17 (89.5%) of the MERS case-patients (Table 1). Participants reported working in 1 of 4 hospitals as either their primary or secondary occupation (Table 1).

Figure 2

Figure 2. Epidemiologic curve for symptomatic laboratory-confirmed case-patients with Middle East respiratory syndrome coronavirus infection, Riyadh, Saudi Arabia, 2015. The curve includes only the 12 case-patients for whom symptom onset was reported, not...

Contact tracing of the initial patient and molecular and serologic laboratory test results identified an additional 18 MERS-CoV infections (Figure 2; Table 2). Of the 19 total case-patients, 12 (63.2%) were from villa 2A; 2 (10.5%) were from a facing villa (1B); and 1 case (5.3%) was reported from each of 5 villas either close to the mostly affected villa (2A) or 2 other villas (10A and 7A) populated with residents from the Philippines (Figure 1).

Among the 8 MERS-CoV cases positive by PCR, 8 were also serologically positive for MERS-CoV (Table 2). According to PRNT₉₀ or ppNT serology results for either the first or second serum sample, an additional 11 persons were serologically positive for MERS-CoV infections. Therefore, a total of 19 of the 828 dormitory residents had evidence of MERS-CoV infection by molecular or serologic testing or both; the infection attack rate [IAR] for the cohort was 2.3%.

Of the 9 patients from whom a second sample was collected in March 2016, a total of 8 had ppNT titers of >1:20, and only 2 of these had PRNT₉₀ titers of >1:20. For 2 of these 8 patients, ppNT indicated a >4-fold fall in antibody titer; for the others, ppNT antibody levels remained within 2-fold that of the initial serum sample.

Bivariate analyses indicated significant assocations between MERS and the following risk factors: having direct contact with a known MERS patient (RR 10.9, 95% CI 6.7–17.6); sharing a bedroom (RR 25.5, 95% CI 10.3–63.1), kitchen (RR 15.5, 95% CI 5.4–44.2), bathroom (RR 25.5, 95% CI 10.3–63.1), meal (RR 19.4, 95% CI 7.5–50.3), or transportation vehicle (RR 11.8, 95% CI 4.9–28.5); and having indirect contact with a known patient (RR 15.5, 95% CI 5.4–44.2) (Table 3). The presence of a chronic condition did not vary by MERS infection status. According to multivariate analyses, direct contact with a known MERS patient (OR 27.6, 95% CI 8.4–91.0) and sharing a bedroom with a MERS patient (OR 5.7, 95% CI 1.5–22.5) remained statistically significant. Having a functioning air conditioner in the bedroom was protective (OR 0.15, 95% CI 0.03–0.82). None of the women reported traveling outside of Saudi Arabia in the 14 days before symptom onset (data not shown).

This study details the comprehensive investigation of a cluster of MERS cases reported outside a healthcare-associated or camel industry–associated occupational setting. In this women-only, expatriate worker dormitory in Riyadh, Saudi Arabia, the overall IAR of 2.3% is similar to that found in a household contact study conducted in 2014 (IAR of 4.3%) (9). However, in this outbreak, the residential setting was more crowded than typical single-family households. Although we found the IAR in some villas to be low, we identified IARs as high as 35.3% (12/34) in 1 villa (2A), probably because of the exceptionally crowded living and sleeping conditions. Within this villa, 12 women were infected with MERS-CoV but only 10 reported any symptoms. Rates of IAR were not affected by the presence or absence of underlying conditions or the median age of residents by villa.

This study identified the independent risk factors for infection to be direct contact and sharing a bedroom with a MERS patient. Findings from other serologic studies have been similar (23). We hypothesize that the increased human-to-human transmission within villas resulted from the clustering of the women’s activities. For example, the same women who lived together typically ate and socialized together, worked together, and traveled to and from work together. These activities added to the likelihood of intense direct physical contact among the women and probably led to limited but effective human-to-human transmission within their residence.

Globally, the extent of human-to-human transmission outside of healthcare facilities is uncertain, and whether MERS-CoV has the potential for sustained community transmission is unclear. Transmission among family members seems to be limited but can be amplified in healthcare settings (24,25) among persons with underlying medical conditions and to healthcare workers. Contributors to propagation of MERS-CoV infection in healthcare facilities include aerosol-generating procedures such as intubation, suction, and collection of nasopharyngeal swabs (26). Compared with the total number of MERS-CoV infections reported to WHO to date, patients in our study cohort were significantly younger (median age 32 vs. 52 years, respectively), healthier (6.3% vs. 41.0% reporting >1 chronic condition), and more likely to be female (0 vs. 68.1% male) (27).

Healthcare staff can prevent human-to-human transmission of MERS-CoV through stringent adherence and implementation of detailed and clear protocols for standard, droplet, and aerosol infection prevention and control (IPC) measures among the various persons within a healthcare setting (i.e., healthcare workers, patients, and visitors) (28). Such IPC measures were not followed by the inhabitants of the dormitory in this study.

Although we were able to rule out a connection to dromedary camels, we were not able to specifically determine the source of this outbreak. Of the 19 laboratory-confirmed case-patients, 17 reported working at the same women-only university in Riyadh and the other 2 worked primarily as cleaners at the same healthcare facility in Riyadh (hospital B). Of these 19 case-patients, 3 also reported having a secondary place of employment, including working as cleaners at 2 other hospitals in Riyadh (hospitals A and D). We hypothesize that 1 of the 19 infected women identified in this investigation may have been exposed to and infected with MERS-CoV while working as a cleaner in a healthcare facility where persons with undiagnosed MERS had been cared for. In August 2015, hospital B, reportedly the primary occupation location for 2 women who were MERS-CoV positive according to PCR, was the location of a small cluster of laboratory-confirmed MERS cases (n = 5). Unfortunately, viral genetic sequencing was conducted on only 1 of those patients (22); without further epidemiologic and sequencing data from other patients in this cluster, or from the laboratory-confirmed patients in the small cluster in hospital B in August 2015, we cannot surmise further.

The time lag between identification of MERS patients in hospital B in August 2015 and the timing of this outbreak in October 2015 suggests that persons with subclinical cases may have been in or working in this hospital during August–October 2015; however, because testing for MERS-CoV in Saudi Arabia was substantial (29), missing symptomatic cases was unlikely. A subject of some debate and recent focus has been the potential role of mildly symptomatic or asymptomatic infections and possible environmental contamination in the spread of MERS-CoV in healthcare facilities (22,30–33). The rapid initiation of this investigation and use of an existing protocol (34) (developed for such use after the rapid isolation of close contacts regardless of the development of symptoms and the implementation of a no-fly policy among residents of the compound until the full 14-day follow-up was completed) probably limited further human-to-human transmission inside and potentially outside of Saudi Arabia.

Our study highlights the potential role of healthcare workers not responsible for direct patient care (e.g., hospital cleaners) in the spread of MERS-CoV. Often, hospital cleaning staff may be from other countries, may speak several languages, and may be missed by efforts to increase IPC specific to MERS-CoV. Specific MERS-CoV IPC training should be directed to cleaning staff in healthcare facilities, in addition to healthcare providers, in appropriate languages, particularly to protect them from infection and from facilitating virus spread within the healthcare facility.

For the 8 women with RT-PCR–confirmed infection, antibody titers ranged from 1:10 to 1:80 by PRNT and from 1:20 to 1:160 by ppNT. For 9 of the 19 women with confirmed evidence of infection by RT-PCR, serologic testing, or both, for whom follow-up serum samples were available 3 months after the putative exposure, 7 women had PRNT titers of <1:20 and 1 woman had ppNT titers of <1:20. Thus, the ppNT antibody test was somewhat more sensitive for detecting evidence of past infection. A ppNT titer of 1:20 is therefore an optimal indicator of past infection in seroepidemiologic assays. The ppNT, although more sensitive, correlated well with PRNT among persons with RT-PCR–confirmed MERS-CoV infection (35) and was uniformly negative in serum from persons in areas where MERS-CoV is not endemic (e.g., Hong Kong [36]). For this study, we categorized those without RT-PCR evidence of MERS-CoV infection but PRNT or ppNT antibody titers >1:20 as being MERS-CoV infected.

Of the 8 women who had RT-PCR–confirmed infection, 2 were asymptomatic, as were 6 of the 11 women whose diagnosis was made solely by serologic testing. Serologic studies of cohorts of patients positive for MERS-CoV by RT-PCR have shown that milder disease and asymptomatic infections may not be associated with detectable serologic responses (37). Thus, our serologic testing probably underestimates the true number of MERS-CoV infections that may have occurred. However, our data provide evidence that even asymptomatic infections can sometimes lead to detectable serologic responses and that such investigations are useful. Furthermore, the serologic results at 5 months after putative exposure show evidence of antibody titers waning to below diagnostic limits in some patients but also show that antibodies may remain detectable in others. This information is useful when interpreting seroepidemiologic studies in high-risk populations.

Our study had several limitations. Because of multicollinearity of the exposure variables (38), the accuracy of individual predictors may be compromised. The lack of collection of acute blood samples during the outbreak limited our ability to detect seroconversion. In addition, we were not able to conduct sequencing for patients of this outbreak and therefore were not able to use this information to potentially confirm that all 19 infected women acquired their infection from a common source or to identify the source of the outbreak.

The rapid initiation of contact tracing, isolation, and subsequent investigation probably contributed to the quick halt of human-to-human transmission in this outbreak. On the basis of the possible source of infection, to reduce secondary human-to-human transmission outside the occupational setting, our study indicates that IPC measures introduced in healthcare facilities should focus on not only healthcare personnel but also those working within the wider facility, including cleaners.