Community-based interventions for preventing COVID-19 transmission in low- and middle-income countries: A systematic review

Master R.O. Chisale Mzuzu University, Mzuzu City, Malawi Sheena Ramazanu The Hong Kong Polytechnic University, Hong Kong Joseph Tsung-Shu Wu Luke International, Mzuzu City, Malawi Frank W. Sinyiza Mzuzu Central Hospital, Mzuzu City, Malawi Thokozani Bvumbwe Mzuzu University, Mzuzu City, Malawi Billy Nyambalo Ministry of Health, Research Unit, Lilongwe City, Malawi Balwani Chingatichifwe Mbakaya (  bcmbakaya@gmail.com ) St John's Institute for Health, Mzuzu City, Malawi https://orcid.org/0000-0002-4705-9064


Information source /search strategy
The following eight database sources were used to gather the required research articles: PubMed, EMBASE, PsycINFO, AMED, CINAHL, DOAJ, Medline and Google Scholar. MeSH database search in PubMed was conducted to identify search terms and keywords. These keywords combined with Boolean operations OR and AND were used to search and retrieve articles from the databases. The search strategy was con ned to research articles published from December 2019 to August 2020. The following key search words were used; Transmission OR spread AND COVID-19 OR COVID OR corona virus, prevent* OR avoidance OR control OR limit AND COVID-19 OR COVID OR corona virus, community AND transmission OR spread AND COVID-19 OR COVID OR corona virus, community AND prevention OR avoidance OR control OR limit AND COVID-19 OR COVID OR corona virus, community OR family OR school OR workplace OR neighbour? OR household OR local AND transmission OR spread AND COVID-19 OR corona virus, community OR family OR school OR workplace OR neighbourhood OR household OR local AND prevention OR avoidance OR control OR limit AND COVID-19 OR COVID OR corona virus AND developing countries OR low-income countries OR middle-income countries. Knowledge OR attitude OR practice AND COVID-19 OR corona virus.
Efforts were also made to identify both published and unpublished studies by manually searching conference proceedings and pre-prints to identify additional unpublished studies. Thereafter, identi ed studies were checked to determine their eligibility.

Study selection
Articles identi ed from the databases were imported to Mendeley Reference Management Software (14). Thereafter, the title, abstract and nally full articles were reviewed against the set inclusion criteria.

Data collection process
The process of data extraction started with a database search of relevant articles as described above and following the PRISMA guidelines (see Figure 1).
Titles and/or abstracts of studies were retrieved and studies that potentially met the inclusion criteria as outlined above were identi ed. The full texts of potentially eligible studies were retrieved and independently assessed for eligibility by two authors. The inconsistencies between the two authors (MC & BC) over the eligibility of some studies were discussed and resolved with a third author (SR). A table was used to extract data from the studies included for assessment of study quality and synthesis evidence. The details included; author, year of study, type of participants, age, setting, country, sample size, study design and methods, study purpose/objectives, study outcomes and results. All relevant information was extracted from each study, summarized and documented.

Search outcome
An initial search of the databases and other sources yielded 10,100 articles. The titles of the identi ed articles were assessed and 10,028 articles were removed because they were either duplicates or abstracts only. Of the remaining 68 articles, 62 were excluded because they did not meet the inclusion criteria and were conducted in high income countries or not in community settings. The remaining six articles met our inclusion criteria (see Figure 1).

Quality appraisal
Quality of the design and reporting system were the main focus at this stage. Three authors (MC, SR & BM) reviewed independently to assess the risk of bias in the included studies. The MMAT(13) was used to appraise the six studies included in the review critically. MMAT is a validated checklist used to appraise the quality of studies included in any systematic review with a quantitative, qualitative and mixed methods approach. The MMAT has two general screening questions applicable to all study designs: 1) Are there clear research questions? 2) Do the collected data allow to address the research questions? The MMAT appraises the following study methodologies and designs: qualitative, quantitative randomised controlled, quantitative non-randomized, quantitative descriptive and mixed methods study designs. The tool is divided into ve components and each component is designed to assess the quality of a speci c study design. For example, assessment criteria for assessing for quantitative descriptive studies included: 1) Is the sampling strategy relevant to address the research question? 2) Is the sample representative of the target population? 3) Are the measurements appropriate? 4) Is the risk of nonresponse bias low? 5) Is the statistical analysis appropriate to answer the research question?
According to the latest version of MMAT tool (13), it is discouraged to calculate an overall score from the ratings of each criterion. Instead, it is advised to provide a more detailed presentation of the ratings of each criterion to better inform the quality of the included studies. Furthermore, excluding studies with low methodological quality is also discouraged.
The actual rating uses a "yes" or "no" or "can't tell" with comments where necessary. Overall, the more the "yes" score, the better the quality of the study. MMAT was chosen to appraise studies in this review because it can simultaneously appraise studies of different designs, which was anticipated but turned to have included cross-sectional studies only which quali ed according to our inclusion criteria. Therefore, we used the MMAT section which appraises the quantitative descriptive study designs.

Data synthesis
We identi ed community-based interventions used to prevent COVID-19 in low-and middle-income countries. Community-based intervention in this systematic review was de ned as (1) a setting other than a health clinic or hospital where interventions were conducted (e.g. home, school, church, neighbourhoods, or workplace), and (2) community as a target (people found in the community setting). A subset analysis was done, categorized by type of community-based intervention for preventing COVID-19 (use of masks, social distancing, quarantine, hand washing, hand sanitizers, sanitation, water etc), and level of KAP on community-based intervention for preventing . A narrative synthesis was conducted based on the content analysis of the included articles.

Quality appraisal
Three authors independently reviewed the six articles; (4,7,(15)(16)(17)(18) included in this systematic review. All the included articles were found to be of highest quality (scored "yes" on all the methodological quality criteria) except one study by Usman et al., (2020) which scored a "no" on one criterion (Is the sample representative of the target population?) in their study, the sample representativeness was not clearly stated. Therefore the included studies in this systematic review are overall of high quality and synthesised evidence can be relied upon.

Study characteristics
Six studies (4,7,16,(18)(19)(20) met the inclusion criteria for this systematic review. All the included studies were conducted during the COVID-19 pandemic and published between the months of April and August, 2020. Of these, one study was conducted in Indonesia (18), one study was conducted in Nigeria and Egypt (19), one study was conducted in Democratic Republic of Congo (DRC), Ethiopia and South Korea(21), one study was conducted in Nigeria (20), and two studies were conducted in Uganda (4,7). See table 2. This shows that the majority of the studies were conducted in Africa, except one study which was conducted in Indonesia (18).
In terms of design, all the studies included in this systematic review used quantitative descriptive designs (cross-sectional).
Data collection was primarily through online survey in all the six studies (See Table 2).
A total of three studies were conducted country-wide through online survey (7,15,16). Two studies were conducted in provinces/states (18,20). One study was conducted in a market setting among vendors(4) The study population age ranged from 12 years (18) to more than 60 years (4). The number of participants in each study varied from 248 (4) to 1, 763 (7).

Summary of the ndings
Studies included in this review were analysed based on the following two outcomes: (1) Types of community-based interventions for preventing COVID-19 transmission in LMICs, and (2) Level of KAP on community-based COVID-19 preventive measures in LMICs. These sub-categories were generated from the objectives of the study. The authors of this systematic review created graphs which summarizes the ndings from the included articles. The presentation and interpretation of the results follow these categories as narrated below.
Type of community-based intervention for preventing COVID-19 We identi ed 10 community-based interventions for preventing COVID-19 in the studies included in this systematic review. These are: (1) use of masks; (2) social distance; (3) hand wash; (4) hand sanitizers; (5) isolation; (6) restriction of gathering; (7) cleaning of surfaces; (8) covering of mouth when coughing; (9) avoidance of public; and (10) lockdown (See gure 2). Among the ten types of interventions, use of masks, social distance and hand wash were top three strategies implemented in the LMICs settings. The least implemented interventions were lockdown and avoidance of public transport. All of the identi ed community-based interventions for preventing COVID-19 which were reported in the articles included in our review, were adopted from those recommended by WHO. Studies looked at some or all of the WHO recommended interventions as a package. However, in one study (18) conducted in Indonesia, their investigations only focused at one preventive measure (social distance) (See gure 2). The authors managed to retrieve results for two countries only (Ethiopia and DRC) in a study by (16) who conducted their study in three countries (DRC, Ethiopia, & South Korea). The results from South Korea were not used in this review because South Korea is a high income country (22) and falls in the exclusion criteria for this systematic review.

Level of KAP on community-based interventions for preventing COVID-19
Five studies looked at the level of KAP on community-based intervention for preventing COVID-19 in LMIC (4,7,18-20). However, one study (16) only looked at the level of practice on community-based intervention for preventing COVID-19 (See Figure 3). The level of knowledge ranged from 62% in a study conducted in Egypt and Nigeria (19) to 99% in a study conducted in Indonesia (18). Age (18-39 years), education (college/bachelors), and background of respondents were factors in uencing knowledge levels (19). While in a study conducted by   (20) in Nigeria, the knowledge of the outbreak of COVID-19 was in uenced by the age (p < 0.05), gender (p < 0.05), level of education (p < 0.05), marital status (p < 0.05) and employment status (p < 0.05). In a study conducted in Uganda by Usman et al. (2020) (4), found that the percentage score among those having their phones connected and those not having connected to the internet on knowledge was 77.00% and 74.53%, respectively.
Hager et al. (2020) (19) found that participants within the 18-29 years age range were 1.4 times (95%CI: 0.55-0.89; p = 0.004) more likely to be knowledgeable than other age groups. Respondents with a high school education were at least 4.7 times (95% CI: 0.15-144.7; p = 0.73) more likely to have satisfactory knowledge about COVID-19 than those with no formal education (19). In the same study, Hager et al (2020) (19), found that Egyptians were 1.8 times (95%CI: 0.43-0.74; p < 0.001) more likely to have more satisfactory knowledge than Nigerians. A study conducted in Uganda by Ssebuufu et al. (2020) (7), found that the mean knowledge scores signi cantly differed across genders, marital status, profession and location (p<0.05) but did not signi cantly differ across age groups (p>0.05) in univariate analysis and ordered logistic regression analysis.
In a study conducted in Indonesia by Yanti et al., (2020) (18), found that the level of attitude on community-based interventions for preventing VOVID-19 was low (59%), while a study conducted in Nigeria by Doherty and colleagues (2020) found that the level of attitude was higher (94%) than the rest of the studies included in this review. There was a signi cant difference (p = 0.0055) in the percentage score for attitude among those connected (86.92%) and those not connected (79.41%) to the internet (Usman et al., 2020). The percentage score for attitude was signi cantly different (P = 0.0358) among those with one source (73.04%) and those with four (82.68%) sources of information (4).
Hager et al. (2020) found that age, gender, level of education, background, and nationality had a signi cant impact on the attitude towards COVID-19. The older the respondents, the better their attitude towards the disease with an odds ratio ranging from 1.34 (95% CI: 1.06-1.74; p = 0.019) to 6.65 (95% CI: 0.17-206.9; p = 0.692). The level of education, background, and nationality greatly affected the perception of global and community response to curbing the spread of COVID-19 and preventing the occurrence of any future pandemic. Furthermore, female participants were 1.59 times (95% CI: 1.27-1.99; p < 0.001), more likely to have a positive attitude towards COVID-19 than males, and Nigerians were 11times (95% CI: 7.57-13.47; p <0.001) more likely to have a positive attitude than Egyptians (19). In a study conducted by Yanti et al., (2020) (18) in Indonesia, found that the respondents' educational and occupational attainment, such as bachelor graduate or civil servant, in uenced their positive attitudes towards social distancing. Furthermore, the respondents who had good knowledge, more than half (58%) had a positive attitude (18).
Level of practice was lowest (19%) in a study conducted by  (16) in Ethiopia and DRC. Nigeria and Egypt registered the highest level of practice (96%) in a study conducted by Hager et al., (2020) (19). A study conducted by Usman et al., (2020)(4) in Uganda, found that there was a positive correlation between attitude and practices (r = 0.17, p = 0.007), as well as their knowledge with practices (r = 0.29, p < 0.001).
In a study conducted in Uganda by Usman et al., (2020)(4), found that the percentage score among those having their phones connected and those not having connected to the internet on practices were 78.49% and 75.08%, respectively. Furthermore, they found that the percentage score for practices was signi cantly different (p = 0.0058) among individuals with no formal education (54.29%) and those with primary (75.24%), secondary (77.03%), and tertiary (80.16%) levels of education. Usman et al., (2020)(4), further found that the percentage score for practice was signi cantly different (p < 0.0001) between those with one source (69.19%) and those with two (89.07%), three (87.23%), four (92.12%), and more than four (86.15%) sources of information. The practice of respondents was signi cantly associated with profession and location of participants in the ordered logistic regression (p<0.001) in a study conducted in Uganda by Ssebuufu et al., (2020) (7). For example, being a health worker was signi cantly associated with good practice (aOR: 2.9 (1.95-4.2). In a study conducted in Nigeria by Doherty et al., (2020) (20), found that both age (p = 0.03) and level of education (p < 0.05) in uenced the respiratory and personal hygiene of the respondents Discussion To the best of our knowledge, this is the rst systematic review conducted with an aim to identify workable and implemented community-based interventions for preventing COVID-19 transmission. This systematic review also identi ed the level of KAP on community-based intervention for preventing COVID-19 among participants in LMICs. Authors acknowledge other types of reviews (min-reviews, scoping, etc) conducted in area of COVID-19 (23,24). However, the current systematic review is unique in that its protocol is registered in a PROSPERO database, which is a register for systematic reviews. Furthermore, this systematic review focused on community-based intervention in LMICs unlike the other previous reviews. Authors searched for research articles that used community-based interventions for preventing COVID-19 in LMICs, and also articles on level of KAP on the community-based interventions published from December 2019 to August 2020. This systematic review identi ed six articles and synthesized its ndings based on the type of community-based intervention used to prevent Covid-19 in LMIC. The review further synthesised the ndings based on the level of KAP among the participants in the communities. The review identi ed 10 plus types of community-based interventions, namely: (1) use of masks; (2) social distance; (3) hand wash; (4) hand sanitizers; (5) isolation; (6) restriction of gathering; (7) cleaning of surfaces; (8) covering of mouth when coughing; (9) avoidance of public; and (10) lockdown. MMAT was used to appraise the methodology of the articles included in this systematic review. Overall, all the six studies included in this systematic review were of high/good quality. Therefore, the outcome of this synthesis can be relied upon to inform education, practice, and policy regarding community-based interventions for preventing COVID-19 transmission in LMICs.
The ndings of this systematic review indicate that the most studied community-based interventions identi ed in the six studies included in this systematic review were use of mask, social distance and hand washing. The least studied community-based interventions were lockdown and avoidance of public transport, which were studied in DRC, Ethiopia and Nigeria (16,20). WHO emphasizes on frequent hand washing as one of the most effective intervention to prevent COVID-19 besides other measures such as social distance, use of mask etc (1,5). Hand washing has been relied upon for decades as a cheap and single most effective intervention in reducing respiratory infections (9,(25)(26)(27). Lockdown was reported in one study (20) out the six studies included in this systematic review. The scarcity of lockdown as a community-based intervention could be ascribed to controversies surrounding on its implementation in the LMICs, where people in these countries protested against the intervention due to its anticipated socio-economic implications (28)(29)(30). Looking at the magnitude of COVID-19 pandemic, the six studies so far identi ed may not be enough to provide a clear direction on policy and practice. As such, more studies are needed which should put more emphasis on the effectiveness of the community-based intervention for preventing COVID-19.
All of the community-based interventions identi ed in this systematic review are in line with those recommended by WHO (1,4,5). However, future studies should consider analysing each intervention separately and not as a lamp some/package as is the case in the current studies. If countries in the LMICs scale up these identi ed community-based interventions, it can help to drastically control the spread of COVID-19 and its consequences such as death, social and economic disruptions among people living in LMICs. This is because the identi ed interventions used in the included articles are scienti cally proven, validated and recommended by WHO (2020) (4,5,31).
The level of KAP among community members is key for any intervention to be effective. This is why we thought of investigating the level of KAP on the implemented community-based interventions. This systematic review also found varying levels of KAP on community-based interventions for preventing COVID-19. For example, the review found that the level of knowledge ranged from 62% (19) to 99% (18). The level of knowledge in the identi ed studies was reported to have been in uenced by age (18-39 years), education (college/bachelors), background of respondents, gender, marital status, and employment status (19,20). LMICs should invest more in educating its citizens in order to improves their knowledge on health literacy. There is su cient evidence that well educated people are always healthier and engage in healthier behaviours (32).
The level of attitude in the articles included in this systematic review ranged from as low as 59% (18) to as high as 94% (20). The ndings of this review have also revealed that age, gender, level of education, background, and nationality were reported to have had a signi cant impact on the attitude towards community-based COVID-19 preventive measures. The results from the six articles included in this systematic review, con rms the fact that knowledge affects one's attitude. For example, there was a signi cant difference (p = 0.0055) in the percentage score for attitude among those connected (86.92%) and those not connected (79.41%) to the internet (4), owing to access to information which might have improved their knowledge. Furthermore, the percentage score for attitude was signi cantly different (P = 0.0358) among those with one source (73.04%) and those with four (82.68%) sources of information (4).
Level of practice ranged from 19% (16) to 96% (19). The review found that there was a positive correlation between attitude and practices (r = 0.17, p = 0.007), as well as their knowledge with practices (r = 0.29, p < 0.001) in a study conducted in Uganda (4). Attitude is touted as one of the variables that in uence intention to practice. A positive or negative attitude towards behaviour will either increase or decrease respectively, the intention to behave and recognition of the behaviour (33). Age and level of education were among the factors that in uenced practice on community-based intervention for preventing COVID-19. The ndings of this systematic review agree with those of a study by (34) in which they stated that sex, sources of information, and education level are some of the possible factors that could in uence the knowledge, attitude, and practices. Knowledge is essential in determining people's behaviour and practices especially during any disease outbreak (34). This is because knowledge level in uences their attitude and practices toward COVID-19 (35). It is therefore important for LMICs to prioritise and implement policies that enhance socio-economic drivers affecting knowledge, attitude and practice in order to succeed in the prevention of COVID-19 and other diseases of poverty such as TB.

Limitations of the study
The following are the limitations of this systematic review: Firstly, the review was limited to studies written in English, which might have led to some bias because articles in other languages could have contributed signi cantly to this systematic review in terms of study outcomes. Secondly, all the articles included in this review were conducted in LMICs. As such, the review may not be generalizable to other global settings. Thirdly, data collection methods (online) used in the articles included in this systematic review, may have targeted literate and economically sound people. i.e. only reached out to participants who has access to technology and know-how to ll it. This might have led to potential participant selection bias.
Despite these shortfalls, this review has identi ed community-based intervention for preventing COVID-19 in LMICs. The review has also generated evidence on the level of KAP on community-based intervention for preventing COVID-19 in LMICs.

Implications of the study ndings for practice, research and policy
The ndings of this systematic review inform practitioners, policy makers and researchers about the community-based interventions so far used for preventing COVID-19 transmission in LMICs, and the level of KAP among people on community-based intervention for preventing COVID-19 pandemic. The ndings of this systematic review show direction which can be easily adopted in future when similar outbreaks occur in LMICs. Furthermore, some of the interventions identi ed in this systematic review may also be applied to prevent some of the diseases of poverty such as TB, in uenza, diarrhoeas which are very common in LMICs if it can further be encouraged beyond Covid-19 era. Governments in LMICs are encouraged to promote the interventions identi ed in this systematic review in order to further halt the spread of COVID-19 and other related infections such as TB, in uenza, and diarrhoea.

Conclusion
This systematic review has identi ed the community-based interventions implemented in LMICs to prevent COVID-19 during the pandemic. Furthermore, level of KAP among the population on the community-based interventions for preventing COVID-19 were unveiled. The ndings of this review, re-a rms the importance of effective implementation of the identi ed interventions which are in line with those recommended by WHO (WHO, 2020). More studies need to be conducted in LMIC to establish the effectiveness of the intervention identi ed in this systematic review and those recommended by WHO. Availability of data and material The data and materials used in this systematic review are available from the corresponding author on request.

Competing interests
The authors declare no competing interests.  Three-quarters the respondent reported that th sanitize/wash hands after us an automated machine (ATM/POS); wh those that sanitize/wash hands after touching Niger currency were 5 Majority of the respondents (o 90%) acknowle that staying at home will help reduce the spre of COVID −19. above half of t respondents ch social distanci movement restriction and of face mask a precautionary measures they against COVID infection