Can Self-Reported Symptoms Be Relied on for COVID-19 Screening in Low-Resource Settings?

Abstract

Background: Understanding the extent of the disease penetration and assessing its impact is critical during a pandemic. However, laboratory-based COVID-19 estimation can be resource-intensive and may not be feasible during an emergency, particularly in low-resource settings. Aim: To investigate whether self-reported symptoms can be used for COVID-19 screening to estimate the burden among individuals aged 18 years and above in a rural setting. Methods: A community-based cross-section study was conducted in a rural district of Haryana, a state in north India, using a self-reported semi-structured questionnaire developed on a digital platform. Information on COVID-19 manifestations as essential and non-essential, confirmed laboratory tests, and disability data using Washington Groups of Short Set were obtained. The sensitivity of the COVID-19 symptoms was estimated against laboratory-confirmed true positives. A chi-square or Fisher exact test for association and a multivariable regression to determine the predictors of the prevalence was carried out. Results: In total, 2954 respondents (79.8%), out of 3700 enumerated, were interviewed. The mean age of respondents was 42 years (SD 17.2), with 54.8% female respondents. The prevalence of COVID-19 based on self-reported symptoms was 6.2% (95%CI: 5.3–7.1). The age-adjusted prevalence was 6.04% (95%CI: 5.9–6.1). Of the total COVID-19 cases, 170 (5.7%, 95%CI: 4.9–6.5) revealed a laboratory-confirmed test. Given three essential symptoms to declare provisionally COVID-19 cases, the sensitivity was 82.9% (141/170), but considering two or more essential symptoms along with two or more non-essential, the sensitivity reached up to 91.8% (156/170). The multivariable analysis showed that increased age, higher education attainment, students, entrepreneurs, persons working in private sectors, and participants with poor hygiene were predictors. Conclusions: A symptoms-based identification of COVID-19 cases can give a reliable estimate and valuable insight into the extent of the penetration, especially in low-middle-income countries, and can be a supplement, not a replacement, to a laboratory test.

1. Introduction

The novel coronavirus disease COVID-19 was first reported in Wuhan, Hubei Province of China in the latter part of December 2019 [1]. Within weeks of its outbreak, the COVID-19 disease spread rapidly worldwide and later was declared a global health emergency by the WHO on 11 March 2020 [2]. The disease has multiple clinical manifestations ranging from asymptomatic infection to a severe form of illness, including death [3]. Due to its unprecedented nature since its appearance, the disease has caused more than 770 million confirmed cases and nearly seven million deaths globally till August 2023 [4]. In India, the first case was reported on 27 January 2020, and there were over 45 million cases and more than 533 thousand deaths till June 2023 [5,6]. Meanwhile, the first case of COVID-19 in the Indian state of Haryana, the state being planned for the present study, was reported on 4 March 2020 [7].

In the months following the emergence, many studies investigated the population burden or prevalence of COVID-19 disease, employing various specimens such as nasopharyngeal, oropharyngeal, or nasal swabs or blood sample testing with RT-qPCR, rapid antigen tests, or serological testing [8]. Among many studies, the prevalence of COVID-19 varies widely depending on the types of biological specimens or samples being collected for tests. For example, in a population-based study among 3658 in Indiana, the United States in April 2020 with nasopharyngeal swabs using the RT-qPCR test, the prevalence was 1.7% [9]. The Italian population-based study among 2812 people conducted in February 2021 showed that the COVID-19 prevalence was 2.6% using RT-qPCR of nasopharyngeal swabs.

In India, the majority of reports regarding the estimation of the burden or COVID-19 prevalence were a part of hospital routine data collection, particularly conducted among healthcare workers during the pandemic. On the other hand, the report from the population-based data was limited. Additionally, most of these studies have been based on seroprevalence and are cost-intensive. For example, of the very few population-based studies available, one such study conducted in a Mumbai non-slum area among 2702 people with their serum sample in July 2020, the seroprevalence (IgG) of COVID-19 was 16.1% [10]. Another study in Delhi during the second wave, January 2021, reported a seroprevalence of 50.5% (IgG) [11]. In a similar population-based study in the South Andaman district soon after the first wave of the pandemic, the seroprevalence was reported to be 39.3% [12].

Considering the difficult time due to the pandemic, studies with blood samples for seroprevalence and nasopharyngeal swabs for RT-qPCR tests will many challenges in terms of cost, supplies for logistics, and equipment when planning to investigate the COVID-19 burden or prevalence in rural settings, especially in lower-middle-income countries (LMICs). To understand the real extent, it is important to assess the disease magnitude, especially in Asian countries, such as India, where a significant proportion of the population is rural or semi-urban. Moreover, nearly 75% of the Indian population are rural inhabitants; therefore, the assessment of the prevalence of COVID-19 among the rural population is worthy, employing less resource strategies. Further, there is a potential risk of recurrent COVID-19 or a new variant wave, such as Omicron, Alpha, Beta, and Delta, over the years ahead; such a self-reported symptoms-based screening study to assess the magnitude, especially in lower-middle-income countries, will help to generate data quickly [13]. Therefore, this study aimed to investigate the COVID-19 cases based on self-reported symptoms present during the illness in a rural district of Haryana so that the disease burden could be established [14,15]. Additionally, we also collected information on COVID-19 laboratory tests from each participant.

2. Materials and Methods

2.1. Study Setting and Design

A community-based cross-sectional study was conducted in a rural predominant district (Jhajjar) of Haryana (Figure 1), a state in the northern part of India, from 5 January 2023 to 28 February 2023. Based on the 2011 census, the district Jhajjar has a population of more than 1 million living across its four sub-districts. There are 862 females per 1000 males with an overall literacy rate of nearly 71%. Agriculture is the primary occupation of the residents of the district. This study was conducted in one randomly selected sub-district of the chosen district.

2.2. Sample Size and Sampling Technique

The study participants were individuals aged 18 years and above. The detailed methodology, including sampling technique, was published in a previous paper [16]. Briefly, the estimated sample size was 3700, and estimation was made based on the seroprevalence of COVID-19 among the adult population (8.5%), absolute precision of 1.5%, with 80% power of the study at 95% confidence limits, and a design effect of 1.7 along with the expected non-response rate of 20%. A multistage cluster random sampling technique with probability proportional to size was used in selecting a minimum of forty clusters from the selected sub-district. For this, we obtained a list of electoral polling booths (named here as clusters) from The Election Commission of India of the chosen sub-districts, which constituted the sampling frame. Each electoral polling booth has an adult population of 1000 (500–2000) aged 18 years and above.

For further selection of households, we employed the compact segment sampling technique (Figure 1), keeping the segment size of 20 to 25 households (70 to 100 individuals aged 18 years and above). One segment from one cluster was randomly chosen and the team conducted a door-to-door survey covering all households.

2.3. Study Tool and Data Collection

A semi-structured questionnaire developed on the SurveyMonkey digital platform was used for the present study (https://www.surveymonkey.com/, accessed on 14 December 2024). To align with study objectives, additional questions were included such as information regarding COVID-19 illness in the past along with clinical symptoms, any laboratory COVID-19 virus test conducted, and the Washington Groups of Short Set (WG-SS) on functioning tool for the disability measurements [17]. The team for data collection included one supervisor, two field investigators, and four field assistants. The team took help from local Accredited Social Health Activities (ASHA) workers from the community. The study tool was pretested on the non-study group before the finalization.

We also screened all participants for diabetes using a random blood sugar point of care kit (OneTouch Verio Glucometer LifeScan Inc., New Brunswick, NJ, USA), blood pressure with an OMRON device (Omron Corporation, Kyoto, Japan), height in cm with a stadiometer (SECA Model 214, Seca Gmbh Co., Hamburg, Germany), and weight in kg using an electronic weighing scale (SECA Model 807, Seca Gmbh Co., Hamburg, Germany).

2.4. Study Definitions

COVID-19 individuals (self-reported): A person who declared that he or she had COVID-19 illness in the past. We also obtained information on the presence of clinical symptoms from these infected individuals as well as any laboratory confirmation tests conducted on them. This additional information was collected for cross-verification and sensitivity estimation for symptoms based on COVID-19 case detection. The prevalence was defined based on responses received from the participants (self-reported prevalence).

2.4.1. COVID-19 Illness Symptoms

The symptoms were dichotomized as essential and non-essential symptoms. To dichotomize the COVID-19 symptoms for the present study, we performed an extensive literature review on the COVID-19 acute illness symptoms using both the published scientific literature and official government websites and expert consultations from the relevant fields. Based on the frequency of reported symptoms, we categorized symptoms as essential (more frequent or common symptoms) and non-essential COVID-19 symptoms (less frequent or less common).

2.4.2. Essential COVID-19 Symptoms

• Sore throat;
• Fever (with any degree);
• Myalgia;
• Cough (new onset or in the absence of any previous history of persistent cough or no known history of exposure to allergens);
• Fatigue;
• Shortness of breath;
• Loss of smell or taste.

2.4.3. Non-Essential COVID-19 Symptoms

• Chills or cold;
• Nasal congestion or blocked;
• Joint pain;
• Headache;
• Chest pain or discomfort;
• Nausea or vomiting;

2.4.4. Case Definition Based on COVID-19 Illness Symptoms

1. If a person presented at least three or more essential symptoms, then we concluded provisionally that the person was a COVID-19 case.

2. If a person presented at least two essential symptoms, with at least two or more non-essential symptoms, then we concluded the person was a COVID-19 case.

Hypertension: A person with hypertension was defined when the systolic blood pressure was equal to or more than 140 mmHg with or without high diastolic blood pressure ≥ 90 mmHg (WHO) or receiving medicine for hypertension.

Diabetic Mellitus: It was defined as when the random blood sugar ≥ 200 mg/dL (American Diabetes Association Criteria) or self-reported history of diabetes.

Body Mass Index was calculated as a person’s weight in kilograms divided by the square of height in meters. It was categorized (WHO) as underweight < 18.5 kg/m²; normal: 18.5–24.9 kg/m²; overweight: 25.0–29.9 kg/m²; Obesity: ≥30.0 kg/m².

2.5. Ethics Clearance

This study was conducted under the declaration of Helsinki. The institute ethics committee approved this study (Ref. no. IEC-260/04.03.2022, RP36/2022). We also discussed this study with the district health officials such as chief medical officers and district program managers and later received approval for smooth implementation. Every participant was given an information sheet in their local vernacular about this study, and later, e-consent was obtained from them.

2.6. Data Management and Analysis

All data were transferred from the SurveyMonkey server to STATA version 15 (StataCorp 2015, Stata Statistical Software: Release 15, College Station, TX, USA: StataCorp LP). Data were cleaned and checked before transfer. The server was protected using a password. A descriptive analysis was used to present the data in terms of mean and standard deviation, percentage for prevalence along 95% confidence interval for prevalence. p-value was calculated using chi-square or Fisher’s exact test to determine the factors associated with COVID-19 prevalence. A multivariable logistics regression model was used to determine the predictors (independent variables) of COVID-19 prevalence. A statistical significance was set at a p-value < 0.05 level. The sensitivity of the symptoms-based COVID-19 definitions was also calculated employing the laboratory-confirmed test as true positive.

3. Results

3.1. Characteristics of the Sample Population

In total, 2954 participants were included in this study out of 3700 enumerations. The response rate was 79.8% (

Table 1. Characteristics and COVID-19 prevalence (self-reported) among study participants (N = 2954).

Characteristics	Participants (n, %)	COVID-19 Cases (n, %)	Prevalence %(95%CI:)
Total	2954	183	6.2(5.3–7.1)
Sex
Male	1335 (45.2)	96 (52.5)	7.2 (5.8–8.5)
Female	1619 (54.8)	87 (47.5)	5.4 (4.2–6.4)
Age (in years)
18–25	589 (19.9)	22 (12.0)	3.7 (2.1–5.2)
26–35	636 (21.5)	47 (25.7)	7.4 (5.3–9.4)
36–45	573 (19.4)	33 (18.0)	5.8 (3.8–7.6)
46–55	450 (15.2)	34 (18.6)	7.6 (5.1–10.0)
56–65	360 (12.2)	24 (13.1)	6.7 (4.0–9.2)
66–75	226 (7.6)	14 (7.7)	6.2 (3.0–9.3)
Age ≥ 76	120 (4.1)	9 (5.0)	7.5 (2.7–12.2)
Education
Illiterate	427 (14.5)	15 (8.2)	3.5 (1.7–5.2)
Primary school certificate	284 (9.6)	15 (8.2)	5.3 (2.6–7.8)
Middle school certificate	359 (12.2)	14 (7.7)	3.9 (1.8–5.9)
High school certificate	491 (16.6)	22 (12.0)	4.5 (2.6–6.3)
Senior secondary school	787 (26.6)	50 (27.3)	6.3 (4.6–8.0)
Graduate and above	606 (20.5)	67 (36.6)	11.1 (8.5–13.5)
Occupation
Unemployed/homemaker	1750 (59.2)	92 (50.3)	5.3 (4.2–6.3)
Agriculture	310 (10.5)	20 (10.9)	6.4 (3.7–9.2)
Business/shopkeeper	170 (5.7)	21 (11.5)	12.3 (7.3–17.3)
Government job	100 (3.4)	8 (4.4)	8 (2.5–13.4)
Private job	333 (11.3)	29 (15.9)	8.7 (5.7–11.7)
Student	270 (9.1)	13 (7.1)	4.8 (2.2–7.3)
Not disclosed	21 (0.7)	0	0
Body Mass Index
Underweight	254 (8.6)	7 (3.8)	2.8 (0.7–4.7)
Healthy weight	1762 (60.0)	109 (59.6)	6.2 (5.0–7.3)
Overweight	742 (25.3)	54 (29.5)	7.3 (5.4–9.1)
Obesity	177 (6.0)	12 (6.6)	6.8 (3.0–10.5)
Hypertension (n = 2927) *
Yes	857 (29.0)	67 (36.6)	7.8 (6.0–9.6)
No	2070 (70.1)	115 (62.8)	5.6 (4.5–6.5)
Diabetes Mellitus (n = 2282) *
Yes	131 (4.4%)	11 (7.4)	8.4 (3.5–13.2)
No	2151 (72.8)	137 (92.6)	6.4 (5.3–7.4)
Seeing difficulty
No difficulty	2663 (90.1)	162 (88.5)	6.1 (5.1–6.9)
Some difficulty	268 (9.1)	20 (10.9)	7.5 (4.2–10.6)
Severe or more	23 (0.8)	1 (0.5)	4.3 (4.6–13.3)
Hearing difficulty
No difficulty	2800 (94.8)	178 (97.3)	6.4 (5.4–7.2)
Some difficulty	140 (4.7)	05 (2.7)	3.6 (0.4–6.6)
Severe or more	14 (0.5)	0	0
Walking difficulty
No difficulty	2536 (85.9)	155 (84.7)	6.1 (5.1–7.0)
Some difficulty	341 (11.5)	23 (12.6)	6.7 (4.0–9.4)
Severe or more	77 (2.6)	05 (2.7)	6.5 (0.8–12.1)
Cognition difficulty
No difficulty	2814 (95.3)	173 (94.5)	6.1 (5.2–7.0)
Some difficulty	113 (3.8)	9 (4.9)	8.0 (2.8–13.0)
Severe or more	27 (0.9)	1 (0.5)	3.7 (3.9–11.3)
Hygiene and self-care difficulty
No difficulty	2832 (95.9)	171 (93.4)	6.0 (5.1–6.9)
Some difficulty	113 (3.8)	10 (5.5)	10.6 (4.2–16.9)
Severe or more	28 (0.9)	2 (1.1)	7.1 (3.0–17.3)
Communication difficulty
No difficulty	2873 (97.3)	178 (97.3)	6.2 (5.3–7.0)
Some difficulty	66 (2.2)	04 (2.2)	6.1 (0.1–11.9)
Severe or more	15 (0.5)	1 (0.5)	6.7 (3.6–20.9)

Note: * confirmed using point of care kit/apparatus.

). Among them, 54.8% (1619) were female respondents. The mean age of the study participants was 42 years (standard deviation 17.2). The highest number of respondents was in the 26–35 years age group (21.5%, 636) followed by 18–25 years (19.9%, 589). Around one-quarter of the participants were either illiterate or studied only up to primary school, whereas 20% of them were graduates and above.

Further, 60% (1750) were unemployed, which means they were homemakers or farmers by occupation. Two-thirds of participants (60%, 1762) were in the healthy weight category (body mass index is within 18.5–24.9). While most of the participants did not have any underlying health comorbidities, hypertension was the most frequently found health problem among them (29%, 857). As per the WGSS (Washington Group of Short Survey questionnaire), of the six functional difficulties included in the survey, difficulty in walking (at least some or more) was the most frequently reported (13.1%, 418), followed by difficulty in vision (9.9%, 291) and hearing impairment (5.2%, 154), respectively.

3.2. The COVID-19 Prevalence

Of the total, 183 (6.2%, 95%CI: 5.3–7.1) respondents had COVID-19 disease based on self-reporting in the past along with clinical symptoms, whereas the age-adjusted prevalence was found to be 6.04% (95%CI: 5.9–6.1). Of the total self-reported cases, 170 (5.7%, 95%CI: 4.9–6.5) of them had a laboratory confirmation test as reported by them, with the majority being sampled for nasopharyngeal swabs. The remaining participants did not undergo laboratory tests when they had symptoms or during self-reported illness.

The mean age of the participants who had COVID-19 disease was 44.1 years (standard deviation 16.6). The prevalence of COVID-19 among males was higher than among female participants (7.2% v/s 5.4%). The lowest prevalence was seen in the age group 18 to 25 years; on the other hand, participants in the age group of 46–55 years had the highest prevalence of COVID-19 (7.6%) followed by 76 years and above (7.5%).

Given the prevalence among the literate, the most frequent prevalence of COVID-19 was seen in graduates and above (11.1%), whereas it was 3.5% in illiterate participants. In the case of occupation, the maximum prevalent was reported (12.3%) in entrepreneurs (business/shopkeeper) followed by 8.7% among persons working in various private sectors. Considering the body mass index (BMI), respondents who had unhealthy BMI had higher a prevalence of COVID-19 compared to lower BMI participants (Table 1).

Considering the comorbidities and COVID-19 prevalence, the prevalence of COVID-19 among respondents with hypertension and diabetes was 7.8% (95%CI: 6.0–9.6) and 8.4% (95%CI: 3.5–13.2), respectively. Given the difficulties in functioning, participants with severe or more self-care problems along with being unable to maintain personal hygiene had the highest prevalence of COVID-19 (7.1%, 95%CI: 3.0–17.3), followed by those with difficulty in communication (6.7%, 95%CI: 3.6–20.9, Table 1).

3.3. Sensitivity of the Symptoms-Based COVID-19 Screening

Of the total self-reported cases (183), 141 had three or more essential symptoms that were provisionally considered COVID-19 cases. Given the true positive that had a confirmed positive test result (170), the sensitivity was 82.9% (141/170,

Table 2. Presences of COVID-19 symptoms among self-reported cases.

COVID-19 Symptoms	Participants with Symptoms, (n)	Percentage
Essential symptoms (ES) among self-reported participants (183)
Cases with ≥3 ES	141	77.0
Cases with 2 ES	27	14.8
Cases with <2 ES	15	8.2
Non-essential symptoms (NS) among participants with 2 ES (27)
Cases with ≥2 NS	15	-
Cases with <2 NS	12	-

Note: Participants with confirmed laboratory test (true positive) = 170.

), whereas considering two or more essential symptoms and two or more non-essential, the sensitivity reached up to 91.8% (156/170, Table 2).

3.4. COVID-19 Illness Symptoms

All self-reported COVID-19 individuals were asked presence of clinical symptoms during the illness. The most frequent clinical symptoms that manifested during illness in study participants were fever (165, 90.2%), followed by weakness (118, 64.5%), loss of smell (109, 59.6%), cough (96, 52.5%), myalgia (84, 45.9%), breathlessness (75, 41.0%), and headache (61, 33.3%, Figure 2).

3.5. Factors Associated with COVID-19

Females had 1.2 times (OR 1.2, 95%CI: 0.75–2.00) more prevalence than males, but it was not statistically significant (p = 0.412). The multivariable regression revealed that the COVID-19 prevalence increased as the age of the participants increased compared to the baseline age group (18 to 25 years). The analysis showed that the prevalence was approximately four times more likely to be higher in the age group 26–35 years (OR 3.6, 95%CI: 1.3–10.3) and age 36–45 years (OR 4.3, 95%CI: 1.5–12.8, p = 0.008), six times more in the age group 46–55 years (OR 6, 95%CI: 2.0–18.3) and age 56–65 years (OR 6.1, 95%CI: 1.9–19.6, p = 0.002), and seven times or more in aged 65 and above (

Table 3. Factors associated with the prevalence of COVID-19.

Characteristics	COVID-19 Prevalence		p-Value	Odds Ratio (95% C.I.)
	No	Yes
Sex	n,%	n,%
Male	1239 (92.8)	96 (7.2)		1
Female	1532 (94.6)	87 (5.4)	0.412	1.20 (0.75–2.00)
Age (in years)
18–25	567 (96.3)	22 (3.7)		1
26–35	589 (92.6)	47 (7.4)	0.016	3.61 (1.27–10.33)
36–45	540 (94.2)	33 (5.8)	0.008	4.31 (1.46–12.80)
46–55	416 (92.4)	34 (7.6)	0.001	6.02 (1.98–18.32)
56–65	336 (93.3)	24 (6.7)	0.002	6.12 (1.91–19.61)
66–75	212 (93.8)	14 (6.2)	0.002	7.28 (2.06–25.67)
≥76	111 (92.5)	9 (7.5)	0.000	12.82 (3.31–49.51)
Education
Illiterate	412 (96.5)	15 (3.5)		1
Primary school certificate	269 (94.7)	15 (5.9)	0.066	2.14 (0.95–4.83)
Middle school certificate	345 (96.1)	14 (3.9)	0.206	1.72 (0.74–4.01)
High school certificate	469 (95.5)	22 (4.5)	0.035	2.37 (1.06–5.30)
Senior secondary school	737 (93.7)	50 (6.4)	0.000	4.39 (2.00–9.65)
Graduate and above	539 (88.9)	67 (11.1)	0.000	7.26 (3.29–16.02)
Occupation
Unemployed/ homemaker	1658 (94.7)	92 (5.3)		1
Government job	92 (92)	8 (8)	0.739	1.16 (0.47–2.89)
Private job	29 (8.7)	29 (8.7)	0.163	1.56 (0.84–2.91)
Business/shopkeeper	21 (12.4)	149 (87.7)	0.035	2.14 (1.05–4.35)
Agriculture	20 (6.5)	290 (93.6)	0.311	1.39 (0.73–2.64)
Student	13 (4.8)	257 (95.2)	0.022	3.79 (1.20–11.73)
Body Mass Index
Underweight	247 (97.2)	7 (2.8)		1
Healthy weight	1653 (93.8)	109 (6.2)	0.146	1.84 (0.81–4.18)
Overweight	688 (92.7)	54 (7.3)	0.154	1.87 (0.79–4.43)
Obesity	165 (93.2)	12 (6.8)	0.291	1.75 (0.62–4.95)
Hypertension (2927)
No	1955 (94.4)	115 (5.6)		1
Yes	790 (92.2)	67 (7.8)	0.074	1.46 (0.96–2.20)
Diabetes Mellitus (2282)
No	2014 (93.6)	137 (6.4)		1
Yes	120 (91.7)	11 (8.4)	0.622	1.19 (0.59–2.41)
Seeing difficulty
No difficulty	2501 (93.9)	162 (6.1)		1
Difficulty	270 (92.8)	21 (7.2)	0.815	1.07 (0.60–1.91)
Hearing difficulty
No difficulty	2622 (93.6)	178 (6.4)		1
Difficulty	149 (96.7)	5 (3.2)	0.059	0.33 (0.10–1.04)
Walking/Climbing steps difficulty
No difficulty	2381 (93.9)	155 (6.1)		1
Difficulty	390 (93.3)	28 (6.7)	0.666	1.15 (0.61–2.19)
Cognition difficulty
No difficulty	2641 (93.9)	173 (6.2)		1
Difficulty	130 (92.9)	10 (7.1)	0.57	1.33 (0.49–3.60)
Hygiene and self-care difficulty
No difficulty	2661 (94.0)	171 (6.0)		1
Difficulty	110 (90.2)	12 (9.8)	0.048	2.68 (1.01–7.14)
Communication difficulty
No difficulty	2695 (93.8)	178 (6.2)		1
Difficulty	76 (93.8)	5 (6.2)	0.256	0.41 (0.09–1.90)

).

In the case of literacy, participants with a higher education status were more likely to become infected when compared with illiterate participants as the odds ratio increased from 2.3 at high school level certificates to 7.2 at graduation and higher degrees (Table 3). Among the occupations, participants working in shops or involved in some business and students had a much higher prevalence (two to three times) compared to unemployed/retired/homemakers.

Considering the types of difficulties as predictors, the prevalence of COVID-19 was noted to be 2.6 times (OR 2.6, 95%CI: 1.1–8.3) more in persons who had some difficulty in hygiene or self-care than persons with no difficulty. The remaining parameters like vision or hearing impairment, walking difficulty, difficulty in cognition, and communication showed no association with COVID-19 prevalent with their respective counterparts, i.e., no difficulty (Table 3).

4. Discussion

A community-based surveillance system is critical when a sudden outbreak or epidemic is happening both in rural and urban areas to understand the extent of morbidity and mortality burden. For example, in the last few years, several respiratory-related viral epidemics such as SARS-CoV, H1N1 influenza, and Middle East Respiratory Syndrome (MERS)-CoV have occurred in various parts of the world [18]. Conducting a rapid door-to-door survey during epidemics helps to generate data quickly, thereby informing the findings to healthcare planners and authorities to make decisions promptly.

Since the emergence of SARS-CoV-2, followed by the rapid spread of the disease, the WHO has developed a survey protocol for nationwide population-based serosurveys to facilitate the data collection and sharing of the data to understand the burden of COVID-19 [19]. Conducting seroprevalence, rapid antigenic test, or swab-based test surveys in rural settings of LMICs may encounter numerous potential challenges, including cost, logistics, and poor power supply and equipment during the emergency crisis, particularly when there is an absence of prophylactic or therapeutic choices for the disease. Therefore, alternative methods or solutions are extremely crucial when trying to generate community-based data during a difficult time [20].

Furthermore, given geographical areas, rural populations are less likely prepared to deal with such highly uncertain situations during an unprecedented pandemic compared to urban regions. The impact in such areas could be compounded by skewed distribution of healthcare infrastructure, shortage of personal protective equipment, and a lack of healthcare staff who can adversely affect rural communities. In a nutshell, conducting COVID-19 epidemiological studies in rural areas will help local health authorities understand the overwhelming problems and also help them to manage the risk of transmission and initiate control measures without much delay. This signifies the importance of conducting a survey that helps to generate quick data about disease status in rural areas.

One primary aim of the present study is to investigate COVID-19 cases in rural areas of the Indian state of Haryana using COVID-19 illness symptoms. For this, we classified them as essential and non-essential symptoms. To our knowledge for the first time, we proposed a case definition of COVID-19 illness for a rapid population-based survey to understand the burden where the resources are constrained to conduct laboratory-based surveys. We also calculated the sensitivity against the confirmed laboratory test, i.e., true positive, and that showed a reasonably high sensitivity in both case definitions, three or more essential symptoms (82.9%) and two or more essential and two or more non-essential symptoms (91.9%). This indicates that we can rely on a community-based survey employing such a case definition, especially during the health crisis time. This dichotomization of COVID-19 illness manifestations was carried out after an extensive review of the published materials and various official webpages. In due course of time, such categorization may be revised when new and atypical manifestations appear in the future similar outbreaks or epidemics.

Furthermore, this study showed that the prevalence of COVID-19, as self-reported by participants, was 6.2%. The drawback of a self-reported study is that participants may prefer not to disclose if they were infected with COVID-19 because of various reasons.

Such an alternative or complementary method with fewer resources for data collection to the traditional or classical laboratory test procedure may be enormously useful when the testing capacity is exceeded or faces difficulty due to the overwhelming COVID-19 cases in a region. Perhaps such a study with a symptoms-based case definition can help in terms of generating data within a short period, particularly in low-resource settings.

The present study also revealed that the prevalence of laboratory-confirmed cases, self-reported, but verified, is 5.7%. Compared with other population-based studies conducted elsewhere with a sample size of more than 2000 that used nasopharyngeal swabs for confirmation, the prevalence of COVID-19 ranged from 0.6% to 22.6%, as reported in a systemic review [8]. In another nationwide community-based study conducted in Bangladesh among the general population across all age groups, the prevalence of COVID-19 was reported to be 6.4% [21]. The same study revealed that the prevalence was higher in older age groups [21]. This finding is concurrent with our study. In a similar study conducted in Jakarta, though a single-center study, the prevalence in ages above 60 years was 29.6% [22]. An Ethiopian study showed that the COVID-19 positive rate increases from 1.1% in age less than 20 years to 2.2% in age 40 years and above [23]. In our study, the prevalence increases from 3-fold to 12-fold as the age of the participants increases from 26 to 35 years to more than 76 years.

Similar to other studies conducted elsewhere, fever was the most common symptom followed by weakness and cough reported in our study [24,25]. For instance, in a Bangladesh study, 74% of COVID-19 cases had a fever [21]. The same study reported 55% of participants had a cough, whereas in the present study, coughing was present in 52.5% of the cases.

Further, the current study showed that participants with higher education are greatly more likely to become infected with COVID-19 than participants without any education. At the same time, those involved in business or shopkeeping have a twofold higher prevalence, while students have a threefold higher prevalence compared to those without any employment in the rural community.

There are a few limitations in this study. The primary aim of the study is not about specific diagnosis, i.e., specificity, of the people infected with the SARS-CoV-2 virus or any other virus causing similar symptoms but meant to assess the burden of COVID-19 based on self-reported symptoms in difficult times. The symptoms-based case definition of the present study may not apply to asymptomatic cases or participants with one or less than one essential symptom. Additionally, further research can be conducted incorporating paper-based testing or rapid antigen tests to identify the asymptomatic cases. We did not collect information on the laboratory test from those who did not declare COVID-19 cases, which may help to judge the self-reported estimations. There is still a scope to improve such a case definition in future studies to calculate all screening indicators. As we excluded participants aged below 18 years, the true prevalence among the eligible population is not reflected.

5. Conclusions

Conducting a door-to-door survey for COVID-19 in a rural area not only provides us the information on how much the virus penetrated during the epidemic but also helps in generating disease burden, thereby assisting healthcare authorities in making a decision promptly. Although there are ways of identifying COVID-19 cases in rural communities, such a study with a symptom-based case definition may perhaps be a worthy and reliable method for estimating COVID-19 prevalence during difficult times, especially in low-resource settings. Furthermore, such symptom-based COVID-19 identification acts as a supplement, not a replacement, to a laboratory test.