This was a prospective case-ascertained study which involved the identification of household contacts of a laboratory-confirmed COVID-19 infection. The study was carried out in the northeast and central districts of Delhi and the samples were processed in the laboratory of Microbiology, Maulana Azad Medical College, New Delhi. The non-hospitalized confirmed cases of COVID-19 of age more than five years and their close contacts in their household were enrolled in the study. Once a confirmed COVID-19 case was identified in at least one member of the household such households were included in the study. Households were subsequently followed up to observe development of secondary infections. A household with only one primary case and their identified household contacts were also included in the study after a written informed consent. For the purpose of this investigation, the primary case was identified through the Delhi Government surveillance system. The study was carried out for a duration of six months from 28 ^th December 2020 to 28 ^th June 2021. Enrolled households should have four home visits, including the enrolment visit (Day 1) and three follow-up visits on 7, 14 and 28 days of enrolment. Those who had severe COVID-19 disease or were hospitalized, co-primary cases and those living in congregation settings like hostels, orphanages, old age homes and prisons were excluded.

Centre for Evidence Based Medicine (CEBM), University of Oxford, reported that between 5% and 80% of people testing positive for SARS-CoV-2 were asymptomatic. Various government reports in India had also estimated the proportion of asymptomatic infection to be 70%. ¹⁹ Thus, considering the prevalence of asymptomatic cases as 0.70, the sample size was calculated by the following formula: N = 2 × p 1 − p m 2 × deff + c ; where, t=confidence level at 95% (standard value of 1.96), p=0.70 (prevalence of asymptomatic cases), m=margin of error at 5% (standard value of 0.05), design effect (considered as 1.5 in this case), c=non-response rate (considered as 0.10). The sample size was calculated as 484 and rounded off to 500. Therefore, the total number of study participants included in the study were 500. Considering that the average size of household in Delhi is estimated to be approximately 5, a total of 100 households with a laboratory confirmed COVID-19 infected primary case were covered to achieve the desired sample size (i.e., 100 laboratory confirmed cases and their 400 household contacts).

Household: A group of individuals (two or more) living together in the same house with a common household space and shared kitchen was referred to as a household.

Household contacts: A household contact was defined as any person who had resided in the same household (or other closed setting) as a confirmed COVID-19 case. ¹⁹

The incubation period was defined as the time between an exposure that caused COVID-19 infection and the manifestation of the first clinical symptoms of the disease (from infection or exposure to disease).

The serial interval was defined as the time between the primary case's development of symptoms and the onset of symptoms in a contact case.

The average number of infections that an infected person causes during the early stages of an epidemic, when almost all contacts are vulnerable, is known as the basic reproduction number (R0). It should be noted that very little to no immunity to COVID-19 is anticipated.

A positive COVID-19 result determines the secondary infection rate, which is a measurement of the frequency of new COVID-19 infections among contacts of confirmed cases over a certain period of time. In other words, it is the proportion of contacts who contract the infection as determined by polymerase chain reaction (PCR)/serological testing on matched samples.

The secondary clinical attack rate (SAR) is the measure of the frequency of new symptomatic cases of COVID-19 infection among the contacts of confirmed cases in a defined period of time, as determined by a positive COVID-19 result. In other words, it is the rate of clinical manifestation of the infection in contacts.

The term “overcrowding” refers to a situation when the number of people exceeds the capacity of the available living space, whether measured in terms of rooms, bedrooms, or floor area, with detrimental effects on both physical and mental health. It depends on the number of people per room, the tenants' age, sex, and floor area. ^{20 , 21}

A pretested semi-structured questionnaire in alignment with the WHO unity protocols was used for the purpose of data collection. Separate questionnaires were used for primary cases and separate for household contacts. Questionnaires for primary cases included sociodemographic details and questions on housing, disease symptoms, comorbidity status and laboratory reporting. Similarly, household contacts were asked questions on their type of contact with the primary case from their day of laboratory confirmation or the manifestation of symptoms. A daily symptom diary for a period of 28 days was also maintained both by the study participants and the field investigators. The detailed algorithm is shown in Figure 1.

List of all the laboratory confirmed cases in the two districts was obtained from the office of the chief district medical officer (CDMO) along with their addresses on day 1 of the initiation after prior approval from the district administration. Once the list of all positive cases was obtained, simple random sampling using a computer-generated random number table was used to select four households for that data collection day. The field investigator then contacted the cases and took the information such as, is the case living with her/his family, how many members are in the family and how many members have confirmed COVID 19 cases. According to exclusion and inclusion criteria, the cases were selected. The selected households were visited by the survey team consisting of a field investigator, phlebotomist or laboratory technician (LT) and the accredited social health activist (ASHA) /auxiliary nurse midwives (ANM) of the respective areas. In case the household members did not give consent, or the phone number was unreachable then the next household from the available list was selected randomly.

The data collection was done in four phases; in the preparation phase, protocol development, recruitment, training and pretesting in the field were done. The next phase included primary data collection, data quality checking, specimen collection and transportation and refresher training and team meetings at the various intervals. The monitoring and evaluation were conducted at various stages by the investigators for quality assurance. The last phase included data processing. During the study, community engagement was necessitated with the help of ASHA/ANM to convince the patient as some patients were very stubborn and rigid, and initially denied giving the sample.

The nasopharyngeal swabs and blood samples (3 mL) were collected from the patient’s house on day 1, 7, 14 and 28. The specimen box was carried gently without shaking, preventing jerks in the vehicle and provided to the team. The lab technician checked the name and other details and validated with the questionnaire and also checked the subject ID number in the data tool (questionnaire) and labelled that number on the vials. Then, the sample was transported and promptly transferred to the to the technician on duty of the virology lab and serology lab, Department of Microbiology, MAMC. During the process of field data collection and sample collection, the biomedical waste generated i.e., used PPE kits, gloves, needles, syringes, cotton swabs etc. was disposed of in the associated Delhi government dispensary of the catchment area.

Serology

Upon receipt in the laboratory, the identifiers mentioned on the vacutainer tube were cross-checked with the details mentioned in the sample transportation sheet. A unique laboratory accession ID was assigned to each sample. The blood samples were centrifuged to separate serum. Serum was transferred to a pre-labelled microcentrifuge tube and stored at -20°C until further testing. WANTAI SARS-CoV-2 Ab ELISA, an enzyme-linked immunosorbent assay (ELISA) for qualitative detection of total antibodies to SARS-CoV-2 virus in human serum or plasma specimens was used to check for the presence of antibodies. This assay is a two-step incubation antigen “sandwich” enzyme immunoassay, which uses polystyrene microwell strips pre-coated with recombinant SARS-CoV-2 antigen. The results were analysed and validated and provided as a pdf as “reactive” or “non- reactive”.

Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) Testing

The cold boxes containing nasopharyngeal swabs in viral transport media (VTM) vials were cross-checked for the information mentioned on the line list and sample vial. Any discrepancy was reported immediately i.e., any leaked sample, and name/enrolment number mismatch was informed and rectified at the earliest. The real time process was divided into three parts: reaction plate set up, template and PC addition and amplification. The complete process was carried out between 15–25°C. The RT-PCR was put using ICMR approved National Institute of Virology (NIV) RT- PCR kit according to the literature of kit insert.

For accuracy, relevancy, and completeness, the data checking was done weekly by project coordinator and all the investigators of study. After that data was managed regularly and double data entry was done in Microsoft Excel (RRID:SCR_016137).

The data were scrutinized properly by the investigators. All the filled forms were physically verified for their completeness and accuracy. The raw data was entered in MS Excel software and code book for questionnaires on day 1 and follow-up questionnaires for 7, 14 and 28 day was also made. The data was organized and crossed checked twice. The final data were analyzed in SPSS PC Version 26 (RRID:SCR_019096) and STATA16 (RRID:SCR_012763).

Qualitative data have been expressed in proportions and percentages while quantitative data have been expressed in mean standard deviation (SD) and median interquartile range (IQR). Each sample in the secondary contact data were assigned a sampling weight, which was equal to the total number of observations made. Similarly, a finite population correction was introduced in the calculation, which is equal to the inverse of sampling weight as sampling was done without replacement. This helped to weigh the population composition of the collected data to account for different composition of the same. Then, a sampling survey module was designed using STATA 16 and accordingly proportions are calculated. For calculation of R _o, the library (R _o) of R software using secondary attack rate was calculated. To find out predictor variables for the occurrence of secondary cases, binary logistic regression was used. p value <0.05 has been considered statistically significant.

The study was conducted after approval from the WHO ERC and Institutional Ethics Committee of Maulana Azad Medical College with IEC registered as F.1/IEC/MAMC (77/05/2020/No20). Informed consent was taken, and data have been anonymized.

The total primary cases were 109 in both the districts, of which 53 (48.6%) cases were from central Delhi and 56 (51.4%) cases were from the northeast Delhi. The male primary cases 67 (61.5%) were more than female cases 42 (38.5 %). 82 (75%) cases were in the category of 18–44 years and only 4 (3.7%) cases were ≥60 years of age. The median (IQR) age of primary cases was 32 years (25,40) while median (IQR) age of household contacts was 35 years (24,47). The minimum age of primary case was 7 years while the minimum age enrolled for the contacts was 6 years. Similarly, the maximum age among primary cases and contacts was 68 years and 84 years, respectively. The proportion of household contacts across all the age categories was more in the central district as compared to the northeast district and this difference was statistically significant (p=0.043). The distribution of religion and occupation of household contacts were similar across both the districts ( Table 1).

The mean (±SD) household size was 3.9±1.8 and the median household size was 4 (3,5) across both the districts. Majority (85%) households had less than five family members in their house. The family size of more than five members was observed in north-east district 10(63%) as compared to Central district 6(37%). However, this difference in proportion was not significant (p=0.335). Across both the districts, majority (62%) households had ≥3 rooms with 70% having <3 bedrooms. The median number of rooms and bedrooms was found to be 3(2,4) and 2(1,3) respectively. Overcrowding was observed among 75% of the households which was calculated by persons per room criteria (US AHS criteria). Though, overcrowding was observed marginally higher in the central district (51%) as compared to northeast district (49%) as per the persons per room criteria; the difference in proportion was not statistically significant (p>0.05). However, by using persons per bedroom criteria, overcrowding was observed more in the northeast district (66%) as compared to the central district (34%) and this difference in proportions was statistically significant (p=0.039) ( Table 2). About 39% household contacts of primary cases had to share a toilet during the period of illness and difference in proportion was found between those who did not share a toilet in the central (65%) and northeast (35%) districts. This difference in proportion was statistically significant (p=0.031). Majority (79%) of household contacts who took care of the primary case during their COVID-19 illness were from northeast district as compared to only 21% from central district and this difference in proportion was statistically significant (p<0.001). The sharing of utensils and sharing of the same glass with the primary case was seen among 3% of household contacts and all of them belonged to the central district. (p=0.043) Across both the districts, sharing of room (17%), sleeping in the same room (16%) and shaking hands (8%) were the other types of contacts between primary cases and their household contacts ( Table 3).

The most common symptom of presentation was fever (53%) followed by constitutional symptoms (45%) and respiratory symptoms (37%). Influenza-like illness was seen in 71% of primary cases and gastrointestinal symptoms were present in 15% of the cases. Co-morbidity such as obesity (28%), self-reported diabetes (2%), heart disease (2%) and asthma (1%) were present. Other co-morbidities viz renal disease, autoimmune diseases, liver disease, immunosuppression were not reported by the primary cases. Three out of 42 women were pregnant. Of these, two were in the second trimester while one was in the third trimester.

All the primary cases who were RT-PCR positive initially, 20 (18.3%) remained laboratory positive by PCR on 7 ^th day from the day of enrolment, while 4 (3.7%) remained positive even on day 28 from the day of enrolment. However, if the virus had a potential for spread, it could not be ascertained since RT-PCR could not differentiate between the live and the dead viral debris ( Figure 3).

A total of 69 secondary cases were reported among the household contacts during the study period. This was the cumulative total of the number of infections (RT-PCR positive) amongst the household contacts within the 4-week period of observation. It was found that out of these secondary cases, 24 (34.78%) had symptoms while remaining 45 (65.22%) were asymptomatic. The SAR estimated was 13.86% (95% C.I. 9.71%,19.39%) and the secondary infection rate was 33.16% (95% C.I. 26.97%, 40.00%). The median time taken for symptom onset in primary case to symptom onset in secondary cases is 3.5 (0,5.25) days and mean serial interval is 3.6±5.73 days status. Incubation period was calculated as 3.33±7.99 days for the household contacts to become cases. The generation time from infection in primary case to the development of infection in contacts was 2.13±4.46 days. The number of infections produced by a primary case among their susceptible household contacts was 1.26 [95% C.I. 1.21,1.31]. Thus, on an average, one primary case infected 1.26 persons. It was also observed that one in 89 cases needed hospitalization among the total cases in households (2%) in both the districts. Most of the subjects hospitalized were in the age group between 45–59 years ( Table 4).

SAR was higher (20%) in the central district of Delhi than the northeast (4.87%) and the difference was found to be statistically significant (p=0.002), and was higher among females (14.3%) in both the districts combined, however, the difference was not statistically significant. The SAR was maximum in the age groups 45 to 59 year and 18 to 44 years. Those who travelled out for work had a higher SAR as compared to those who stayed indoors, indicating the role of workplace, travelling and social distancing like factors playing a role in transmission too, which might be confounding factors in these households. The risk was 1.58 times as compared to those at home. Those who shared a room, slept in the same room, shared utensils, glasses, kissed and hugged with the primary case had a higher secondary attack rate and these factors were statistically significant. Presence of comorbid conditions and being symptomatic also increased the risk of transmission (OR=2.07 and 29.6 respectively). On univariate analysis, the significant predictors/risk factors of the infection were location of household in the central district (SAR=20% [13.75,28.16]) vs the northeast district (SAR=4.87% [1.83–12.35]) p=0.002, sharing of utensils (SAR=42.85% [14.26–77.11], p=0.02), and using the room to sleep where the case had been isolated (SAR=25% [12.97–42.71], p=0.047). On multivariate analysis, the model correctly predicted 86.79% (goodness of fit), and the only factor which was significant was the presence or absence of symptoms while other factors which were significant in univariate analysis did not predict the SAR.

Secondary attack rate was higher in females in the central district of Delhi, being maximum in the age groups 45 to 59 year and 18 to 44 years. Those who shared a room, slept in the same room, shared utensils, glasses, kissed and hugged with the primary case had a higher secondary attack rate and these factors were statistically significant. Presence of comorbid conditions and being symptomatic also increased the risk of transmission. On univariate analysis, the significant predictors/risk factors of the infection were location of household in the central district (SAR= 20% [13.75,28.16]) vs the northeast district (SAR=4.87% [1.83–12.35]) p=0.002, sharing of utensils (SAR=42.85% [14.26-77.11], p=0.02), and using the room to sleep where the case had been isolated (SAR=25% [12.97–42.71], p=0.047). On multivariate analysis, the only factor which was significant was the presence or absence of symptoms while other factors which were significant in univariate analysis did not predict the SAR ( Table 5).

The household characteristics predicting the occurrence of secondary cases among the households were living in the central district (p=0.006), age of primary case between 18 to 49 years, overcrowding, less no. of rooms, smaller family size and the presence of symptoms in primary case (p=0.009) being associated with higher secondary attack rate among the contacts. On multivariate analysis, the model correctly predicted 71.56% (goodness of fit) and RT-PCR positivity status of primary case and the presence of symptoms was found to be a predictor of occurrence of secondary cases in the household (p<0.001) ( Table 6).

The household contacts who were taking care of the primary case were at a higher risk of secondary infection and this was found to be statistically significant (p=0.019). No other factors like sharing of room or toilet, shaking hands, sleeping in the same room or sharing of the utensils were associated with transmission of infection ( Table 7).

Secondary cases presented with a varied range of symptoms. The most common presentation fatigue was in 15 (21.7%), cough and sore throat in 13 (18.8%) and constitutional symptoms of fever, headache, myalgia in 9 (13%) of them. Diarrhoea was seen among 5 (7.3%) of them while a few of them suffered from loss of smell or taste 4 (5.8%). Comorbidity was present only in 24 (35%), while the majority did not have any comorbidity. Obesity was present among 37% of them, 4.3% had diabetes and 1.4% had heart disease while none of them had any liver disease, kidney disease, autoimmune disease, rheumatological or hematological disease. The RT-PCR pattern of the secondary cases is shown in Figure 4.

Those who were living in the central district had 2.62 times risk of developing secondary infection. Male gender, age more than 45 years and presence of comorbidity in contacts were also associated with the development of secondary cases. On univariate analysis, living in the central district (p=0.003), absence of antibodies on day 1 (p=0.005), presence of symptoms (p<0.001) and presence of respiratory symptoms 14 days before onset in primary case (p<0.001) had higher odds of developing COVID-19 infection and was found to be significantly associated. On multivariate analysis, model fits 70.8% and presence or absence of symptoms (p=0.024), residing in the central district (p=0.048) and absence of antibodies (p<0.001) were the only significant factors found out to be associated with the secondary cases among the household contacts ( Table 8).

It was observed that symptomatic cases were more in the central district 22 (43.1%) while asymptomatic were more in the northeast district (89%). The risk of symptomatic transmission was 6.07 times in the central district as compared to the northeast district and was found to be statistically significant (p=0.024). Female gender had 1.47 times risk of both symptomatic and asymptomatic transmission. Similarly, age group 18 to 44 years also had higher risk of symptomatic transmission as compared to other groups. Presence of comorbidity among the contacts also determined symptomatic transmission for the occurrence of secondary cases and the risk was 3.02 times compared to those who did not have any comorbidity (p=0.032). Other factors like sharing of room, toilet, utensils and seropositivity status did not determine the presence or absence of symptomatic transmission among the secondary cases ( Table 9).