Crowded Household, Subdivided Flats, and Dilapidated Housing Are Risk Factors of Bedbug (Cimex Lectularius and Cimex Hemipterus) Infestation: a Cross-sectional Study of Hong Kong Households

Background: Bedbugs have been a neglected issue globally, disproportionately affecting low-income households. The features of many deprived housing units in Hong Kong provide suitable habitats for bedbug infestations. This study aims to identify the housing risk factors for bedbug infestations in Hong Kong. Methods: Using a cross-sectional study design, online self-reported questionnaires in Chinese were distributed between June 2019 to July 2020. Data collected were socio-demographics, crowded household condition, housing type, dilapidated housing features, and frequency of noticing bedbugs in the participant’s place of residence in the past year. The latter was transformed into a dichotomous dependent variable, “bedbug infestation”. Weighted bivariate and multivariate analysis using binary logistic regression were performed on SPSS 24. Results: The study sampled N=696 participants, 63.7% have had bedbug infestations. Bivariate analysis shows a positive correlation between the number of dilapidated housing features and bedbug infestation (OR=1.28, 95% CI 1.18-1.39, p<0.001). N=663 were included in the multivariate analysis. Those aged 45-64 (OR=2.53, 95% CI 1.30-4.91, p=0.006) and have primary education or below (OR=9.43, 95% CI 3.12-28.44, p<0.001) have signicantly greater odds of bedbug infestation compared to their respective reference groups, ≥ 65 and tertiary education. Having monthly household income ≤ HKD30,000 (OR=1.69, 95% CI 1.15-2.5, p=0.008) and living in subdivided ats (OR=16.53, 95% CI 1.01-269.72, p=0.049) and crowded household (OR=1.55, 95% CI 1.06-2.28, p=0.024) increases the odds of bedbug infestation. Dilapidated housing features that signicantly increase the odds of bedbug infestation are having second-hand furniture (OR=2.97, 95% CI 1.16-7.58, p=0.023), housing cleanliness issues (OR=2.66, 95% CI 1.13-6.25, p=0.024), presence of bedbugs in neighbouring residential units (OR=3.32, 95% CI 1.57-7.04, p=0.002), and presence of bedbugs on the streets (OR=1.9, 95% CI 1.12-3.23, p=0.018).

Conclusions: Crowded household, subdivided ats, and dilapidated housing are risk factors for bedbug infestations. To better control bedbug infestations, there needs to be a shift from viewing infestations as a personal hygiene to a public health issue. Efforts and policies should focus on addressing the housing risk factors identi ed in this study and prioritise vulnerable groups such as the elderly, low education level, low-income groups, and occupants of subdivided ats. Background Bedbugs (Cimex lectularius and Cimex hemipterus) are nocturnal ectoparasites that feed on human blood (1). The United States Environmental Protection Agency (US EPA) has deemed bedbugs to be a "pest of signi cant public health importance" (2). One inseminated female bedbug can start an infestation alone by laying 200-500 eggs in a lifetime at a rate of 5-15 eggs a day (1,3,4). Bedbug infestations occur when their population grows out of control causing adverse health effects (2,5,6,7). Bedbug bites usually occur in a linear pattern on exposed skin while the host is asleep or still, these usually result in multiple itchy sores where bites occur (7,8,9,10). In severe cases, the bites may result in bullous eruptions (11) and excessive blood lost to bloodmeals may result in anaemia (7,9). Bedbugs have the potential to act as vectors for over 40 diseases, although no outbreaks have been attributed to them so far (7,12,13). Bedbug infestations may result in a broad range of psychosocial disorders including anxiety, depression, and insomnia (14).
Bedbug infestations pose a signi cant economic burden to households and businesses (5,15). In 2019, the median monthly household income for all households in Hong Kong is HKD28,700 (around USD3,700) (16). Hiring a professional exterminator in Hong Kong per household infestation typically ranges from HKD3,000 to 30,000 (around USD390 to 3,900) depending on the infestation severity, treatment types, living oor area, and other factors. For businesses or facilities such as hotels or hospitals, the cost may be upwards of HKD200,000 (around USD26,000). Low-income households may not afford to hire exterminators or replace infested belongings.
Reported since the 1990s, the global bedbug resurgence has been attributed to several factors including human population growth and urbanisation (6,7,17). These factors strain housing systems leading to more deprived housing with dilapidated housing features that provide favourable conditions for bedbug infestation and spread such as cracks in walls, peeling wallpaper, and crowded housing (5,18,19). The effect of different building types on the risk of bedbug infestation is worth further investigation since certain building characteristics may pose different risks (20).
Having the world's most expensive housing market (21), the housing situation in Hong Kong offers a unique set of environmental factors that are hypothesized to facilitate the local and international spread of bedbug, these include crowded living environments, dilapidated housing, and vastly different housing situation for those who can afford it versus those who cannot (22,23,24).
Bedbug infestations are neglected in Hong Kong despite being a public health threat since 1. Bedbugs are perceived to pose an insigni cant health concern compared to other pests such as mosquitos; 2. Those affected by bedbugs are unlikely to report or seek help for several reasons such as shame, and the lack of means or know-how; and 3. Bedbug infestations are perceived as a personal hygiene instead of a public health issue, shifting the burden of action to individual households rather than the collective efforts of society (25,26).
This study aims to identify the housing risk factors associated to bedbug infestations in Hong Kong.

Methods
This study used a population-based cross-sectional study design and was conducted in the Hong Kong Special Administrative Region (HKSAR), China.
Data collection and sampling method Data was collected using online self-reported questionnaires in Chinese, it was later translated into English (Additional le 1). Data collection occurred between June 2019 to July 2020. Google forms was used to create the questionnaire, its electronic link was broadcast on discussion forums and social media pages of different districts in Hong Kong. Participants were eligible to participate if they lived in Hong Kong and were aged 18 or above. N = 696 participants voluntarily completed the questionnaire. G*Power 3.1 was used to calculate the minimum required sample size (n = 617) for the multivariate binary logistic regression (27).

Measurements
For the question, "In the past year, how often did you see bedbugs in your place of residence?", responses ranged from "never" to "very often" on a ve-point Likert scale. This variable was transformed into a dichotomous dependent variable, "bedbug infestation", with "never" being "no" and all other responses being "yes". A picture of a bedbug was provided to remind participants of its appearance and minimise its erroneous recognition.
Crowded household was measured using a pseudo quantitative method. Participants were asked whether they felt that their residence lacked space or is crowded (given the variable name "feeling crowded"), their living oor area (ft 2 ), and household size. Data from these variables were used to compute the dichotomous variable "crowded household" de ned as those who felt that their residence lacked space or is crowded, or those with living oor area per capita ≤ 120 ft 2 /person. The cut-off of ≤ 120 ft 2 /person was chosen since less than 25% of the sample met the criteria. Living oor area per capita was computed by taking the upper bounds of each interval responses for living oor area and dividing that by the household size. For living oor area of > 900 ft 2 , the upper bound was taken as 1200 ft 2 , and household size ≥ 5 was taken as 5.
Participants selected their housing type and dilapidated housing features from lists created based on the literature. Participants' sex, age, education level, monthly household income (HKD), and district were also collected. All variables were collected as categorical variables.
Participants had the option to leave their contact information if they were willing to participate in future bedbug related research. N = 7 participants were contacted, and the researchers visited their residence to make observations and take photographs of their housing situation within the study period.

Statistical analysis
A choropleth map of self-reported bedbug infestation cases by district was made. Data analysis was performed using IBM SPSS 24. Weighting by age and sex was applied to the analysis using census data for the end of 2019.
Bivariate logistic regression using chi-square test for categorical variables was used to identify variables associated with bedbug infestation. All variables were considered for inclusion in the multivariate logistic regression to investigate their effects on the odds of the bedbug infestation. Covariates were entered using the forward conditional method if p < 0.05 and retained if p < 0.1. Effect estimates for the covariates in the bivariate and multivariate analysis are presented as odds ratio (OR) with their corresponding 95% con dence interval (CI). Statistical signi cance was considered if p < 0.05.
Hosmer-Lemeshow goodness-of-t test and multicollinearity diagnostics were performed on the nal model of the multivariate regression. The model does not violate the goodness-of-t assumption if p > 0.05. Multicollinearity was considered if the covariates had variance in ation factors (VIF) ≥ 10, or their absolute value of the Pearson correlation coe cient |r|≥0.7 (28).

Results
The questionnaire received N = 696 participants; they were all included in the analysis. The sample size included in the multivariate regression after listwise deletion of missing variables is N = 663 (95.3%), this is beyond the minimum required sample size of 617. N = 7 underwent follow-up visits, Figs. 6 to 10 were selected photographs taken at their place of residence. In Table 1, responses for the question "In the past year, how often did you see bedbugs in your place of residence?" were transformed into the variable "bedbug infestation" with "never" being "no" and all other responses being "yes", n = 422 (63.7%) have experienced bedbug infestation in the past year. Bivariate analysis Socio-demographic variables signi cantly associated with bedbug infestation are age (p = 0.006), education level (p < 0.001), monthly household income (p < 0.001), and region (p = 0.013). Sex is not signi cantly associated with bedbug infestation. Age has a positive trend with percentage of bedbug infestation while education level and monthly household income have negative trends. Only those in income groups < HKD10,000 (p = 0.001) and HKD10,000-30,000 (p < 0.001) have signi cantly greater ORs compared to the reference category, >HKD80,000. Thus, monthly household income was recoded into a dichotomous variable "monthly household income ≤ HKD30,000" (p < 0.001) and included in the multivariate regression. Compared to living in the New Territories region, living in the Hong Kong Island region is not signi cantly different, but living in the Kowloon region (p = 0.005) has signi cantly greater OR of bedbug infestation.
Crowded household (p < 0.001) and the variables that were used to derive it i.e. feeling crowded (p = 0.004), living oor area per capita ≤ 120 ft 2 /person (p = 0.024), living oor area (p = 0.001), and household size (p < 0.001) are signi cantly associated with bedbug infestation. There is a negative trend between living oor area and percentage of bedbug infestation (Fig. 2). Those living in < 300 ft 2 (p = 0.005) have signi cantly greater OR compared to the reference category, > 900 ft 2 . However, the relationship between household size and percentage of bedbug infestation appears to peak at the extremes (Fig. 3). When dividing the upper bounds of the intervals for living oor area by that of household size to compute living oor area per capita, the negative trend with percentage of bedbug infestation is retained (Fig. 4).
The housing types signi cantly associated to bedbug infestation are public rental housing (p < 0.001), home ownership scheme (p = 0.011), private housing (whole unit) (p = 0.004), and subdivided ats (p = 0.017). Public rental housing and subdivided ats signi cantly increases the odds of bedbug infestation whereas home ownership scheme and private housing (whole unit) signi cantly decreases its odds.

Discussion
This is the rst empirical study to investigate the bedbug issue and its associated housing risk factors in Hong Kong.

Crowded household
This study nds that crowded household is a more important factor for bedbug infestations than household size. The data from this study shows a nonlinear relationship between household size and the percentage of bedbug infestation (Fig. 3), however living oor area per capita is linear (Fig. 4).
Furthermore, crowded household and the variables that it was computed from i.e. feeling crowded, living oor area, and household size were all considered for inclusion in the nal model using the forward conditional method for variable selection, but only crowded household was entered.
This study disagrees with Gounder et al.'s 2014 ndings that household size is a more important factor than crowdedness (29). The disagreement may be due to differences in study design and methodology.
Unlike this study, Gounder et al. 2014 de ned crowded housing as having ≥ 2 household members for every living room and bedroom and did not measure the living oor area of the participants.
It is the crowdedness of the living situation which facilitates the propagation of bedbug infestations as human hosts become accessible by living in close proximity (5).

Housing type
Living in subdivided ats is a risk factor for bedbug infestation. Subdivided ats are formed from the splitting of a residential unit into two or more subdivisions, thus subdivided units often neighbour several others (23,30). Over 50% of subdivided ats are located in the Kowloon region (30), this coincides with the choropleth map showing that the number of self-reported bedbug cases are concentrated in the Kowloon region (Fig. 1). Subdivided ats are usually crowded and have many dilapidated housing features, its marginalised residents often possess many health related risk factors and socioeconomic disadvantages (22,23,31,32). In 2016, households living in subdivided ats have median living oor area per capita of 56.5 ft 2 and median monthly household income of HKD13,500, both are lower than their respective medians for all domestic households (30). Other studies have found similar results, that living in poor neighbourhoods and buildings with many adjacent housing units facilitate the spread of bedbugs (5,6,20). Bedbug infestation in subdivided ats have been reported to be a cause for other social issues such as people sleeping at 24-hour fast food restaurant to avoid bedbug bites (25,33). The combination of building and resident characteristics of subdivided ats make their occupants especially vulnerable to bedbug infestations.

Dilapidated housing features
This study nds that participants who report more dilapidated housing features are more likely to report bedbug infestations (Fig. 5). This study has identi ed having second-hand furniture, housing cleaning issues (besides having bedbugs), presence of bedbugs in neighbouring residential units, and presence of bedbugs on the streets to be independently associated with bedbug infestations.
Second-hand furniture has been suggested as a risk factor in other studies as they may harbour bedbugs from the previous owner (5,6,20,29,34). Housing cleanliness issues may allow bedbugs to hide and be di cult to detect and eradicate which agrees with previous literature (5,6,7).
Having bedbugs in neighbouring residential units and on the streets may indicate spreading of bedbugs in a community setting via hitchhiking or egress points such as cracks in walls or electrical conduits. Sheele et al. 2019 found that knowing someone with bedbugs is also a risk factor for bedbug infestation (35). This complicates bedbug management as bedbugs may return from the wider community, even if adjacent neighbouring units are treated for bedbugs. Addressing bedbugs may require the collective efforts of the wider community, not just the neighbourhood or individuals.

Socio-demographics
Having higher education level is a protective factor against bedbug infestation, it may re ect knowledge on bedbug infestation management or the ability to access such information. Older adults (45-64) are at greater risk since they are more active, thus are more likely to be in contact with infested places or persons, facilitating the spread of bedbugs (20,35). This study nds that the elderly (≥ 65) has the greatest proportion of bedbug infestation, they may be more likely to suffer from disabilities and nancial di culties resulting in their inability to maintain household cleanliness and not afford bedbug management services (29,31). Having monthly household income ≤ HKD30,000 is a risk factor for bedbug infestation. In comparison, the 2019 median monthly household income of all economically active households in Hong Kong is HKD35,500 (16), and the typical cost of hiring exterminators ranges from HKD3,000 to HKD30,000. Low-income households may not afford to hire bedbug exterminators or replace infested furniture and personal belongings. Committing to these costs may result in perpetual poverty as bedbugs may return, requiring multiple treatments (5). Furthermore, low-income households are more likely to participate in risky behaviours such as trading second-hand furniture or using communal laundries (5,20,29).

Limitations
Although age and sex weighting were applied to the analysis, the sample may be non-representative of the Hong Kong population as the sampling method used was volunteer sampling using online selfreported questionnaires. Responses from disadvantaged or marginalised groups with limited internet access such as primary education or below, elderly (≥ 65 year olds), and occupants of subdivided ats may have been barred from participating, resulting in the reduced representativeness of these groups and their larger con dence intervals (36). Furthermore, there was no way to con rm the existence of bedbug infestations or any of the participants responses, except for a few cases (N = 7) who underwent follow-up visits.
Online data collection made it di cult to comprehensively evaluate the participants' housing situation.
The presence of certain housing factors depended on the participant's subjective view of their existence, for example the same housing unit may be considered to have housing cleanliness issues by one participant but acceptable to another. Participants selected dilapidated housing features from a list, although an "others (please specify)" option was available, protective factors were not investigated.
Although steps were taken to minimise the erroneous recognition of bedbugs by providing a picture on the questionnaire to remind them of its appearance, bedbug sightings by older participants may be inaccurately reported since previous studies have found that the elderly (> 60 year olds) are more likely to wrongly identify bedbugs from a picture compared to younger people in self-reported questionnaires (35). Furthermore, participants may be predominantly reporting adult bedbug sightings and failing to identify smaller bedbugs in earlier instars, resulting in under-reporting (37).
Social desirability may skew the responses towards lower reported bedbug infestations and housing risk factors since having them are associated with negative stereotypes such as being poor, uneducated, and unhygienic (14,26). However, people who do not have bedbugs may not report their situation since they may nd the voluntary online questionnaire irrelevant to them, and vice versa for those who have bedbugs, resulting in an arbitrarily higher percentage of reported bedbug infestations.
The cross-sectional study design was unable to establish the temporal sequence of events between bedbug infestations and the variables being investigated. Socio-demographic and housing factors are likely to have existed before the occurrence of the bedbug infestation. However, having bedbug infestations may result in some of these factors arising. For example, the signs of bedbugs (their faeces, carcass, and exuviae on walls or furniture) may be interpreted as having housing cleanliness issues.
Crowded and dilapidated housing features are likely to be manifested similarly in other settings, however certain features of Hong Kong's housing situation such as the housing related policies, housing types, and their speci c building features may limit the generalisability of the results to other countries.

Policy recommendations
There needs to be a shift in viewing bedbug infestations as a personal hygiene to a public health issue. Efforts and policies should be focused on alleviating crowded and dilapidated housing and providing adequate standards of living. This will directly address the global bedbug resurgence by removing its environmental facilitators and reverberate improvements to other aspects of life related to housing such as employment, education, and health. Efforts and policies should also prioritise vulnerable groups such as the elderly, low education level, low-income groups, and occupants of at-risk housing types such as subdivided ats.
Faced with the global threat of bedbug resurgence, simultaneous top-down and bottom-up approaches are required. Examples of top-down approaches are anti-poverty policies, increasing the supply and shortening the waiting time of public housing, and relief and cleaning services for those in deprived housing (22,31,32,38). Bottom-up approaches focus on empowering and building resilience of the public to address bedbugs themselves, especially vulnerable groups at risk or already suffering from bedbugs.
Educating low-income households to identify the early signs of bedbug and to self-manage using integrated pest management (IPM) or affordable non-chemical control methods when infestation rates are still low prevents infestations from exacerbating and spreading, thus mitigates the expensive costs of hiring exterminators or replacing furniture and personal belongings (39,40,41,42,43,44 Ethics and consent to participate Written informed consent was obtained from all participants in digital form. After accessing the link to the online survey, participants were shown a statement of consent which explains the purpose of the study, type of questions to be asked, eligibility criteria, data security, participant rights, and risks involved.
The questions to the online questionnaire were only shown after participants voluntarily select "Agree" then "next".

Consent for publication
Not applicable Availability of data and materials The dataset used in this research is available in as an additional le (Additional le 3). Description: Linelist dataset containing participants' responses used in data analysis. The rst row is the variable name and corresponds to the variables presented in the results section, the spaces have been replaced with an underscore. In "Sheet 1", each row represnts a participant and each column a variable.
Missing variables are entered as "999". The column labelled "case_weight_age_sex" contains the case weightings by age and sex. "Sheet 2" shows the coding scheme for each variable.    Darkened corridoor of a subdivided at This gure belongs to us and was taken during the follow-up visits.

Figure 7
Ceiling paint peeling with rebar showing through walls above a rusty and leaking pipe This gure belongs to us and was taken during the follow-up visits.
Page 27/29 Figure 8 Sleeping area next to a wall covered with blood streaks from dead bedbugs This gure belongs to us and was taken during the follow-up visits.

Figure 9
Bedbugs coming out and hidden in cracks in walls This gure belongs to us and was taken during the follow-up visits.