The clinical questionnaire devised for all HCW included a list of 33 symptoms linked to COVID-19, exposure to proven infected persons, precautional isolation measures, results of RT-PCR tests, data about underlying health conditions and routine or recent treatments. The list of symptoms (Supplementary Table 1) was drawn from the existing literature on March–April 2020 (1, 8–11) or inferred from the reports of the self-help Facebook “Coronavirus, CoV-2 and COVID-19 group for doctors only” created at the beginning of the European outburst and joined by more than 100.000 Italian and European doctors. A simplified list comprising the 18 more objectifiable symptoms (Supplementary Table 1) was completed by their doctors for all residents considered less able to report their symptoms. Symptoms referring to the period of SARS-Cov2 diffusion in Italy only were recorded, starting in February 2020. Exclusion of usual symptoms related to chronic conditions or to allergy was asked.

The degree of suspicion of COVID-19 was graded using a self-made scoring algorithm obtained by the combination of relevant symptoms, exposure to infected persons and, when available, chest computerized tomography-scan or ultrasound (Table 3). Scores ≥2 were considered suspicious and followed-up independently of test results.

All the 286 HCW underwent RT-PCR at the time of our integrative screening. Previous RT-PCR tests results were available in 41 workers and 146 residents (Table 4). A single nursing home (AS) had all 88 residents screened for SARS-CoV-2, as a systematic screening was performed after a hotspot of COVID-19 was detected. All remaining HCV and residents had been tested because of evocative symptoms or contact with affected people.

All RT-PCR tests were run in the same laboratory using RADI COVID-19 detection Kit (KHMedical), on ABI 7500 Bio-Rad CFX 96 platform. The lower limit of detection of this kit is 0.66 copies/microliter.

Serology screening used the rapid point-of-care test^® Core tests COVID-19 IgM/IgG Ab, whose instructions report IgM&IgG sensitivity 97.6% and specificity 100%. Tests reading and photographic recording were performed 15 min after deposing the finger blood sample just collected. Weak lines for IgM and IgG were considered positive, according to the report by Li et al. (12) and in consideration of the principle of maximum safety.

All those screened positive with immunoassay with no or negative RT-PCR were immediately isolated and a further swab test was taken.

Serology with the identical lot rapid Core test was repeated after 1 week in 4 HCW with discrepant negative serology screening and previous positive RT-PCR and at 60-day interval in 67 of 86 HCW with positive serology screening or with recent/suspicious clinical scores and negative tests.

Unlike HCW, residents with discrepant serology and positive virology tests did not repeat a serology test due to difficulty in getting an informed consent for further assessments. A 60-day serological follow up was also not performed.

Data were reported as counts and percentages. Comparison between independent proportions (groups) were performed using Fisher Exacts-test. Trends in proportions were tested using Chi-square test for trends. All analysis assumed a 5% significance level and were performed using R software (version 4.0.5).

Informed consent was obtained from all HCW and residents or their representatives.

Formal review and approval by the ethic committee of Spedali Civili of Brescia (NP 4378) was obtained for this study.

Voluntary donations and some residual funding from previous different research studies.

A thorough virology screening was not available for the 246 residents.

Previous RT-PCR results were available for 146 residents (59.3%—Table 1).

32/59 residents positive for serology had also been previously identified with RT-PCR while 5 cases with previous positive RT-PCR turned out negative at serology (Table 4).

Paired positive IgM and IgG bands were present in 57 of 59 seropositive cases, five of which showing weak bands. Two asymptomatic cases showed isolated IgM. No isolated IgG positivity was found.

Overall, 18/64 residents (28.1%) with positivity to either test reported complete absence of symptoms. Mild symptoms and a clinical score 0–1 were present in further 2 cases. A clinical history suspicious of COVID-19 scoring ≥2 was associated with at least one positive test in 44 residents but was also present in 31 residents with both tests negative (Table 4).

In our study the virus was rapidly spreading between residents with Alzheimer's disease or other dementias associated with wandering (30/58), while the contagion amongst all other residents (28/188) was significantly less frequent (51.7 vs. 15.9%—p < 0.001 Fisher exact-test). Actually, the rate of infection in residents with dementia unassociated with wandering was comparable to the group with no cognitive impairment (7/42–16.9%—p-value NS).

The number of infections amongst the 20 residents in persistent vegetative state was instead inferior in comparison with all other residents (1/20–5% vs. 57/226–25.2%—p = 0.051 Fisher exact-test).

No correlation was found between the rate of infected residents and the different size, in terms of number of residents, of these three nursing homes: 23/107 (21.5%) vs. 28/88 (31.8%) vs. 7/51 (13.7%)—p = 0.58 (Test for trends in proportions—p value NS).

All 286 workers had SARS-CoV-2 RT-PCR, clinical and serological screening (Tables 2, 4).

Previous RT-PCR results were available for 41 HCW (14.3%).

14/44 HCW initially seropositive also had a previous positive RT-PCR result. Repeated RT-PCR in those identified with specific antibodies was positive in one additional HCW, who was thereafter isolated.

Paired positive IgM and IgG bands were present in all cases except for a single asymptomatic case showing a weak IgM band only.

Four symptomatic workers with previous positive RT-PCR and discrepant negative serology, were re-tested after 1 week with the same test-kit and all showed very weak bands for both IgM and IgG (Table 4). These four serology tests, although ultimately evaluated as positive in Table 2, have been considered false negative in Table 4, where all results are reported by chronological phases.

Follow up serology after 60 days, performed in 78% (67 of 86) of seropositive or clinically suspicious/recent HCW, yielded two additional serology-positive cases whose suspicious clinical history was recent at the time of the screening. These two additional cases have been considered serology-positive in Table 2, based on final information provided by this study.

HCW in close contact with the residents (including doctors, nurses, health care assistants, personal care-givers, physiotherapists, hairdressers, and beauticians) had significantly more infections (20.79 vs. 1.66% in remaining HCW; p = 0.00013 Fisher exact-test—Table 2). Familial contact with ill relatives was present in 11 HCW only, including the single administrative employee that tested positive.

The seroprevalence in the single nursing home with no COVID-related deaths amongst residents was markedly inferior (8%) to that found in the two nursing homes with known cases (19%) and appeared comparable to 7.5% seroprevalence of the general population in Lombardy at the time.¹

The rate of positive tests was also greater amongst HCW with additional external activities in comparison to those with exclusive activity within their nursing home (20 vs. 16%—Table 2).

Among all 50 HCW with ultimately proven positivity for SARS-CoV2 (serology and/or RT-PCR) in our study, 40% only had fever and 50% reported cough (Supplementary Table 1).

At final evaluation of all test results, 9 seropositive HCW (18%) had been considered clinically non-suspicious (clinical score 0–1), 5 of which (10%) reported complete absence of symptoms. All 5 totally asymptomatic HCW had negative RT-PCR test while serology was frankly positive for both IgM and IgG in 2 cases, showed a weak IgM band only in a third case, and paired weak IgM and IgG bands in the last 2 cases. Close contact with COVID-19 patients was reported in two of them. A suspicious clinical score ≥2 was present in 41 HCW with at least one positive test but also in 61 workers who finally tested negative (Table 4).

Follow up serology after 60 days, performed in 78% (67 of 86) of seropositive or clinically suspicious/recent cases, provided identical either paired IgM and IgG or isolated weak IgM positivity patterns to those previously obtained, with no case showing disappearance of IgM or appearance of IgG. As mentioned, paired IgM and IgG bands were also demonstrated in two of the four workers with recent suspicious symptoms and negative initial serology screening.

Viral RNA was still present in two workers and two residents re-tested after being found seropositive, three of whom were completely asymptomatic.

On the basis of the available RT-PCR ± data of participants at the time of symptoms or contacts with infected people, sensibility and specificity for on-site serology was 83 and 84%, sensibility and specificity for our clinical score was 91 and 56%, while integrated clinical scoring and on-site serology had sensibility 89% and specificity 87%.

Symptoms of COVID-19 are poor indicators of the disease as people affected may experience a wide range of symptoms from mild to a myriad of different symptoms, and a proportion of cases between 6.3 and 96.0% (13) is asymptomatic.

Despite our scoring system intentionally devised to be highly sensitive, unexpected positive tests were still present in 9 workers and 20 residents with clinical scores below the threshold, and entirely asymptomatic disease was more frequent amongst residents (18/64–28.1% vs. 5/50–10.0%; p = 0.019). A similar difference has been observed elsewhere in long-term facilities (14, 15) and is most probably justified by a lesser degree of awareness and/or by difficulty in communicating with older people is particularly concerning when planning strategies for transmission containment in care homes.

In our study, the list of symptoms had been simplified since the vast majority of participants were not able to adequately report more subtle and subjective signs. This has allowed homogeneity and comparison within the same group, but it has made it impossible to assess atypical COVID-related symptoms described in older people elsewhere (16, 17). The significant paired difference in proven infection rates (64/246–26.0% vs. 50/286–17.5%; p = 0.020 Fisher exact-test) might be correlated with more common asymptomatic or atypical forms amongst residents.

Asymptomatic transmission is held responsible for failure of symptoms-based screenings and monitoring strategies in nursing homes (18–21). The awareness that a larger proportion of infected residents have no symptoms increases the urgency for their frequent regular testing (22).

It is considered the gold standard to identify COVID-19: however, nasal and oropharyngeal swabs are charged with a considerable quote (30–40%) of false negative results (23, 24).

In our study 7–8% of residents and workers tested at the time of symptoms/possible contact showed single or multiple false negative virology tests on nasopharyngeal swabs. Repetition of tests proved worthwhile, allowing detection of viral RNA in four cases previously testing negative.

In our study, the residents had an easier access to the virological tests in comparison to HCW. This was due to specific dispositions of our Regional Health System at the beginning of the pandemics, when the availability of tests was still not meeting the demand. In fact, access to swab testing was guaranteed to all the residents with suspicious symptoms or possible contact with an infected person after 6 weeks of the pandemics. The same option was offered to those HCW that were by profession in direct contact with the residents and directly employed by the nursing home, not including the shift workers.

Antibody testing misses early infections but can be useful to understand the prevalence in a community, particularly when no tests have been previously available, and it may disclose false negative cases at antigen testing (25). The advent and the gradual expansion of the population vaccinated with SARS-CoV-2 vaccines, negatively impacts on the informative value provided by serology testing in the absence of associated clinical information and dynamic evaluation of antibody level changes.

Accurate point-of-care virology and immunology tests are particularly needed for effective monitoring of an infection in vulnerable and enclosed settings (12, 26–29).

A variety of serology tests has been internationally approved since the start of COVID-19 pandemics. The diagnostic performance of quantitative serology is generally considered superior to rapid qualitative tests, but its supply has been late and is still not meeting the demand.

Point-of-care antibody tests have demonstrated efficiency comparable to that of laboratory tests (30), have been successfully used in a series of studies so far (27, 28, 31–33), and represented a valid alternative to unavailable quantitative serology in our study.

Point-of care serology tests are cheap and easy to use in number of different clinical situations (12, 26) and are particularly useful in the case of poorly cooperative older residents. Despite controversy over their accuracy (31, 34), the authors decided to follow the encouragement of WHO to use these tests in research studies to establish their usefulness. Their results require a nuanced interpretation and correlation with RT-PCR and clinical information (27, 35).

In our study, rapid on-site serology has been able to confirm symptoms in 35 persons, 20 of whom (6 HCW and 14 residents) had previous negative RT-PCR results and to unveil a previous COVID in 24 further persons with no suspicious symptoms and negative or no previous RT-PCR.

Four HCW and 5 residents had a positive RT-PCR but a negative first serology test.

Interpretation of IgM results appears particularly challenging (36). Similar to other studies of the same period (34), no case with isolated IgG was found.

Although high levels of IgM have been reported 22–24 days after symptom onset at serial serology testing (37) and several studies have shown simultaneous production of IgM and IgG against SARS-CoV-2 (12, 24, 37), the unchanged paired IgM and IgG positivity in all tests repeated after 60 days with no case with IgM disappearance is disturbing. Similarly, the two asymptomatic cases with isolated IgM and with identical positivity pattern at re-testing after 60 days raise doubts about a possible false positive result.

The combination of qualitative and quantitative antibody testing could represent a safer strategy to obtain reliable data.

Comparative evaluations of rapid tests for COVID-19 infection in care homes are urgently needed too (28).

Similar to other studies (7, 38), serology allowed a retrospective evaluation of the control strategies adopted. Besides continuous attention to maintain adherence to the measures put in place and regular testing, identifying weaknesses is crucial to safety in nursing homes.

Laboratory testing capacity was lagging for several weeks after the COVID-19 outbreak in Italy (6). While our regional government focused on hospitals and locating available ICU beds, Lombardy's nursing homes were in many ways left to fend for themselves with no access to RT-PCR testing for weeks. This has negatively interfered with the ability to contain the spread of disease.

The size of a care home is related to the probability of viral contamination, in relation to the number of residents and staff and the presence/ absence of shift workers. This seems confirmed by our results, but the small number of care homes studied does not allow significant conclusion.

The structure of the building itself and the proportion of available single rooms are determinant factors in containment of infection. Although the data of our three nursing homes do not provide confirmation of this hypothesis, a greater representative sample size would be needed to assess it.

The high rate of infected staff in close contact with residents is probably linked to lack of using PPE during the initial days of the outbreak, when no cases had yet been identified or no PPE was available.

Tracing contacts and clarifying the chain of viral transmission in the confined crowded environment of a nursing home might be difficult (35), especially in the absence of laboratory testing capacity. The quick viral spread observed amongst residents with wandering dementia, should be considered when planning strategies to contain the spread of the disease. Moreover, the scientific community has been questioning the ethical implications of implementing testing and restrictive measures such as isolation for residents with dementia. These people are unable to understand the significance of such measures that deeply modify their daily routine (39–41). It is well-documented that reduced social interactions have a greater impact on the residual capacities of these subjects compared to the general population. Therefore, future guidelines relating to restrictive measures must also take this into consideration (22).

Our study highlights the difficulty to gain access to RT-PCR for SARS-CoV-2 infection at the beginning of the pandemics in Italy, the potential contribution of clinical score associated with RT-PCR for SARS-CoV-2 infection diagnosis, and the usefulness of rapid serology tests to assess the level of the COVID-19 outbreak diffusion in nursing homes.

Our data show that a combined strategy with nasopharyngeal swabs, rapid on-site serology and accurate anamnestic monitoring has contributed to reveal the real diffusion of SARS-CoV-2. Our approach has increased the awareness of a previous infection in people that could not get a virology test or had a false negative in the past, clarifying the chain of transmission and identifying symptomatic and asymptomatic persons actually carrying the virus.

The contribution of rapid serology tests in unveiling previous infections is now less in the more developed countries where vaccines have been already administered to the people in nursing and care homes. However, our proposed triple screening strategy could still be a valuable option for daily practice in countries with reduced or no national health service. It could also guide the prioritization of the more vulnerable residents for access to vaccines or the decision to omit the second dose in case of a previous documented immunization (42).

However, continuous monitoring of symptoms and implementation of protective measures is mandatory given the frequent appearance of new variants of SARS-CoV-2 worldwide that might escape the protection provided by the vaccines currently available.