Accidental falls and associated factors among the elderly in Thailand: a national cross-sectional study in 2007, 2011, 2014 and 2017

Abstract

Purpose

This study assesses the prevalence of accidental falls in Thailand's older adult population and the contingent influences surrounding this prevalence.

Design/methodology/approach

Data were drawn from the Cross-Sectional National Surveys of Older Persons in Thailand, pooling of four survey datasets which took place in 2007, 2011, 2014 and 2017. Stratified two-stage sampling was employed. Interviews were conducted with sample sizes of 11,370, 11,061, 13,775 and 12,457 senior citizens, aged 60 and above, in the respective survey years. Further investigation was conducted on subjects who reported to be of good health and without any disability, yet experienced accidental falls. The prevalence of accidental falls was examined, and variable aspects concerning fall risk were assessed with probability-weighted multiple logistic regression.

Findings

The average prevalence of accidental falls from the four surveys was 4.7%. Significant risk factors identified were advanced age, being female, living in a rural residence, having worked in the previous 7 days, lack of/excessive exercise, alcohol consumption, smoking and having an outdoor lavatory.

Originality/value

Accidental falls tend to increase among community-dwelling seniors aged 60 and above. Falls increase with age and are more common among the women in that demographic. Findings suggest the need for government and local agencies to consider tailoring some public health approaches to the prevention of accidental falls. This study also highlights the necessity of proper work environment maintenance to prevent these falls.

Citation

Khongboon, P. and Kespichayawatt, J. (2022), "Accidental falls and associated factors among the elderly in Thailand: a national cross-sectional study in 2007, 2011, 2014 and 2017", Journal of Health Research, Vol. 36 No. 4, pp. 767-780. https://doi.org/10.1108/JHR-07-2020-0308

Introduction

Accidental falls among the elderly have become a critical and global public health issue. It is estimated that a third of the senior population aged 60 and above experience an accidental fall each year [1]. “A fall is defined as an event which results in a person coming to rest inadvertently on the ground or floor or other lower level” [1]. According to the World Health Organization (WHO), 28–35% of elderly persons above age 65 experience accidental falls each year. Additionally, 32–42% of persons above age 70 suffer accidental falls [1]. Naturally, fall statistics concerning the elderly vary between countries [2–4].

A national survey conducted in Thailand revealed that 6.6% of the elderly population had experienced accidental falls in the six months prior to the study [5]. Accidental falls are often the result of morbidity, which serves to aggravate fractures, postfall syndromes and soft tissue injuries [6]. The cumulative influence of these factors negatively affects one's quality of life; thus, it comes as no surprise that falls are a common cause of death for the elderly [7].

Multiple factors increase the risk of seniors experiencing an accidental fall [8, 9]. These aggravating factors can be categorized as intrinsic and extrinsic. Intrinsic factors revolve around pathological and physiological aging processes, which are associated with a slowing of corporal mechanisms that are essential for postural reflexes. Intrinsic factors also encompass cardiac arrhythmias, arthritis, osteoporosis and stroke, as well as neurological and pulmonary conditions [10]. Extrinsic factors relate to the environment of elderly persons, such as public places, personal residences, uneven terrain and public transport. Not all risk factors can be eliminated, but some can be mitigated to reduce the risk of falling. Indeed, most accidental falls arise owing to the interplay of multiple risk factors, and the risk of falls is directly proportional to the number of contingent risk factors [11].

In Thailand, a study on accidental falls suggested that the rate of mortality due to accidental falls will increase gradually, as the general population is progressively aging [12]. Between 2007 and 2017, no studies were conducted on the risk factors of accidental falls in Thai elderly people with good health and no disabilities. This study aimed to determine the risk factors of accidental falls among elderly Thai persons with good health and no disabilities in 2007, 2011, 2014 and 2017. The datasets from the four years were analyzed to categorize different kinds of falls. Fall categories included slipping, falling from a raised floor, falling due to poor lighting and falling on stairs and steps.

Methodology

Participant recruitment phase

This phase was conducted by the National Statistical Office (NSO). In 1994, 2002, 2007, 2011, 2014 and 2017, six cross-sectional surveys were carried out by the NSO. Contextual information concerning the falls was, unfortunately, not gathered in the 1994 and 2002 datasets; hence this study has four sets of data. The primary goal of this survey was to create a database of socioeconomic, demographic, residential and health characteristics of Thai people for national representation. The NSO has formulated general guidelines for survey practice in order to enhance the reliability and validity of NSO surveys. The participants were selected using a multistage sampling technique as show in Appendix 1.

Sampling design and data management phase

This phase was conducted by the authors. The sampling frame included elderly persons aged 60 years and above, in good health and without any disabilities. Exclusion criteria were elderly who fell from presyncope or syncope. After data quality assessment by systematic data editing and cleaning, the final representative sample included 11,370, 11,061, 13,775 and 12,457 individuals in 2007, 2011, 2014 and 2017, respectively.

Since the surveys encompass a broad range of questions concerning the history of accidental falls, the collation of these four cross-sectional sets of data may include accidental falls in the periods between the four study dates, thereby necessitating that caution is exercised when reporting this information.

Data management

The author managed the data to ensure the confidentiality and integrity of the database. The author improved data entry accuracy by double-checking with another researcher. The following methods were employed:

Data cleaning

This concerned the detection and the removal of inconsistencies and errors from the raw data to improve the quality of the data. The coding of missing data was carried out as system-missing in the Statistical Product and Service Solutions software (SPSS) (IBM® SPSS® Statistics).

Data extraction

This was performed to determine the need to change the formatting of the variables. Consistency in the establishment of values for the response category, such as “no” and “yes” was another way of making the dataset friendlier to the users. The original values for the response category for each variable of study are shown in Appendix 2.

Data transformation

The variables were recorded to ensure that the data reflected the purpose of the study. In so doing, we conducted the identification and the recording of variables that were present in all four sets of data. The identifier variables served to ensure that the same individual's values were presented on the same line. Both the independent and dependent variables, particularly at this step, were documented to dummy variables (Table 1).

Data combination

This helped in creating a pool of four cross-sectional survey datasets that facilitated the understanding of changes. Consequently, files from the different years were combined to run a regression that utilized the yearly indicators.

Study variables

Dependent variable

The incidence of accidental falls among the elderly was scrutinized. A fall was defined as “coming to rest on a lower level such as the ground or floor.” Events that resulted from activities such as violence and car accidents were not categorized as accidental falls. The responses from the study were categorized into “no history of fall,” and “at least one accidental fall in the preceding six months.” Accidental falls were attributed to elderly people who reported to be of good health and without any disability but fell over obstacles, while crossing a raised floor, slipped, tripped as a result of poor lighting, and fell on steps and stairs. Falls from presyncope or syncope, which were categorized as spontaneous falls, were excluded from the study.

Independent variables

The independent variables selected were based on a review of the literature [8, 13–15], transformed to dummy variables and considered in multiple logistic regression analysis, Table 1.

Data analysis and statistics

This study used descriptive statistics to help in summarizing the characteristics of the participants. Logistic regression analyses were employed to help assess the extent to which the independent variables selected for the study explained the incidence of accidental falls among the elderly population of Thailand. Confidence intervals (CIs) were calculated at the 95th percentile to facilitate the estimation of the statistical significance.

Sample probability weights were applied to the data for every year [16]. The sample probability weight was calculated in separate datasets before combining the data. It was calculated by the actual sample size divided by the population in that year and multiplied by the sample weight computed from the NSO. Analysis of the statistics was carried out using version 18 of IBM-SPSS (IBM® SPSS® Statistics).

Apart from the ratios of the odds, the modeled absolute prevalence of accidental falls was calculated as a weighted prevalence proportion of individual accidental falls. This procedure helped to measure the effect that each independent variable had on each accidental fall [17, 18].

Independent variable proportions were arranged to facilitate the effective assessment of the comparative effects of the independent variables. This was accomplished by arranging the proportions of the independent variables in ascending order, for each average proportion. This approach is somewhat similar to the calculation and comparison of the risk fractions of the attributable modeled population for the independent variables. It does, however, fail to consider their prevalence. This approach may be considered an evaluation of the “absolute impacts” caused by the risk factors, irrespective of their frequency in the population of the study. Moreover, this approach provides the additional benefit of opting for the absolute prevalence differences related to risk factors, as opposed to mere odd ratios (which vary significantly at dissimilar baseline prevalence for studied results, which can be deceptive) [17, 18].

Ethical consideration

The analyses reported in this study used secondary data originally obtained from the NSO of Thailand, through its countrywide surveys in 2007, 2011, 2014 and 2017. All participants provided their written consent to the NSO, indicating their willingness to participate. The NSO has a file that contains all completed forms of consent received.

Results

The mean ages of the participants were 66.22, 66.39, 66.38 and 65.75 in the 2007, 2011, 2014 and 2017 surveys, respectively (Table 2). Males represented the majority of the population in all four surveys. Approximately 90% of the elderly in the samples lived with others, gradually decreasing from 93.2% in 2007 to 92.1% in 2011, 92.4% in 2014 and 90.1% in 2017.

The number of elderly people who had worked in the seven days prior to the interviews increased gradually, from 52.5% in 2007 to 55.4% in 2011, 57.5% in 2014 and 55.6% in 2017. Approximately 30% lived in the Northeast region, which had the highest proportion of elderly people of all Thai regions. In all four surveys, the majority of the population lived in rural areas. The number of elderly people whose lavatory facilities were outside the house decreased over time, as well, from 28.4% in 2007 to 21.7% in 2011, 18.9% in 2014 and 17.0% in 2017.

The prevalence of accidental falls in the six months prior to the survey was 4.3% in 2007, 4.8% in 2011, 6.2% in 2014 and 3.2% in 2017 (Figure 1). Of the people who fell, 70% experienced their tumble outside their residence. Among these individuals, 4% required subsequent hospital admission.

Logistic regression analysis of associated factors for accidental falls in overall four surveys

Table 3 demonstrates the association of subject characteristics with accidental fall risk. The significant risk factors for accidental falls were age (OR = 1.17, 95% CI: 0.96–1.28), being female (OR = 1.21, 95% CI: 1.08–1.35), living in a rural area (OR = 1.18, 95% CI: 1.06–1.32), having worked in the previous 7 days (OR = 1.64, 95% CI: 1.48–1.81), exercise (OR = 1.44, 95% CI: 1.27–1.64), alcohol consumption (OR = 1.22, 95% CI: 1.08–1.38), smoking (OR = 1.32, 95% CI: 1.16–1.50), lavatory facilities outside the house (OR = 1.10, 95% CI: 0.99–1.23) and the year 2014 (OR = 1.27, 95% CI: 1.31–1.66). On the other hand, elderly people in the 2017 surveys had a negative association with accidental falls (OR = 0.64, 95% CI: 0.55–0.73).

Proportional impacts of the independent variables surrounding the prevalence of accidental falls are presented in Figure 2. On average, the characteristics with the most significant adverse impact on accidental fall prevalence were working in the past seven days (average impact 42.5%), exercise (26.9%), increase in age, per decade (25.0%), smoking (24.3%), alcohol consumption (16.9%), female gender (15.0%) and rural residence (13.3%). The location of lavatory facilities outside the home was associated with a positive proportional impact of 8.0%. Years 2011 and 2014 were associated with a positive proportional impact on the prevalence of accidental falls (24.1% and 10.3%). A lack of handrails in the bathroom and in the bedroom was associated with a positive proportional impact on the prevalence of accidental falls (9.4% and 7.8%). Interestingly, the lack of handrails on stairs was associated with moderate negative proportional impacts on fall prevalence. Being married, but not living as couple, was associated with a positive proportional impact (10.2%). Being married and living together as a couple was associated with a lower risk than being unmarried. The respective proportional impact of the housing type and region was generally modest.

Discussion

Comparing various studies, the average prevalence of accidental falls for the current study is 4.7 %, a figure lower than other studies in Thailand (18.7% and 19.8%, where participants were aged above 60 years, taking previous 6 months' data) [13, 19]. Similarly, the accidental fall rate was 17.2% in Singapore where participants were aged 60 and above considering data from the previous 12 months [20]. In Sri Lanka, people over 65 years and data from the previous 12 months were considered and the fall rate was found to be 34.3% [2]; whereas in Brazil, people over age 60 were surveyed and data from the previous 12 months were considered, revealing an accidental fall rate of 28.1% [21]. In England, a fall rate of 28.4% was revealed after surveying participants aged over 60 years and considering the data from the previous two years [22].

The difference in the fall rate may be attributed to various ways in which falls are defined in different circumstances. The current study defines a fall scenario as one where “a person drops down to the ground due to contact with some obstacle, tumbles downward on the stairs or slips due to inappropriate lighting.” The study did not consider spontaneous falls like those from presyncope or syncope. This implies that only the falls fulfilling the strict criteria were included in the study and hence the current study reported lesser falls than other national and global studies. This low frequency of falls may also be attributed to the difference in population considered in these studies. The current study involved the participants being members of the senior population but not experiencing any disability or weakness and other health conditions.

A connection between older age and accidental falls has been identified in this study echoing earlier studies [23, 24]. Elderly people experience inevitable decline in corporal function. Consequently, they experience difficulty in maintaining an upright posture while they are being transferred [9].

The current study showed correspondence with earlier studies in that the probability of fall was higher among females in comparison to their male counterparts [3] owing to gender-defined differences in terms of bone density, activity levels, rate of mortality and muscle strength [7]. Additionally, both the current and earlier studies suggested that people living in rural areas were more prone to accidental falls [25]. This may be attributed to the engagement in greater work activity due to rural living. Moreover, studies have also been undertaken to explore the connection between urban residence and accident falls [3, 12].

Earlier studies have also explored the risk of falls in older persons in a variety of marital circumstances including single, widowed and divorced subjects [8]; however, such studies failed to adduce any connection between fall risk and marital status. Contrary to earlier studies [21], the current study did not suggest any impact of lower socioeconomic stratification, income and education of people on their risk of accidental fall. Earlier studies suggested that people living alone are more prone to falls [15], but the current study suggests otherwise [26].

Another important detail that this study brought to light was that exercise produced a higher risk of accidental falls. This can be explained by the U-shaped association between exercise and accidental fall risk [27]. This kind of association implies that the fall risk is higher for both extremes (i.e. the highly active and highly inactive members of the population), indicating an intricate interplay among falls, risk and activity level. The intrinsic risk factors of an individual are dependent on the kind of conditions and the intensity of conditions that encompass that subject. This means that if an individual engages in tasks that do not agree with the corporal functionality of his/her body on account of his/her age, an accidental fall is more likely to occur. One of the studies also implied that fall risk is elevated while walking [28], while another indicated that a higher level of physical activity marks reduction of fall risk [29]. Another study posited that higher physical activity would lead to serious injury of the individuals [30].

It was also indicated that certain tasks and actions indirectly elevated fall risk by exposing individuals to greater risk factors in their surroundings. These include environmental factors, like slippery or rough flooring and the absence of even walking paths, as well as other personal factors like frailty or partaking in potentially hazardous exercises [31]. Elderly people who partook in alcohol and tobacco were found at greater risk of falling, as suggested by both the earlier and current research [32]. This study did not find any association between urinary incontinence, visual difficulties, or impaired hearing and the risk of accidental falls. Other studies, however, found that urinary incontinence [22], visual difficulties [3, 12] and impaired hearing [6] were known risk factors for accidental falls.

A correspondence was found between the current study and the previous ones [33] with respect to the relation between fall risk and participants' working conditions during the week before the interview. The greater rate of falls in other studies may be attributed to the more demanding working conditions that characterize the study participants' daily routines, for example, manual labor and/or farming. In the previous decades, the greater portion of the Thai population was heavily involved in agriculture – despite the development of alternative industries [34]. Future studies should extensively investigate the risks associated with work-related accidental falls to devise strategies that safeguard the elderly in the workplace.

The location of washrooms and lavatories outside the house is one of the major reasons behind a significant number of falls among the elderly. These results are in agreement with Sophonratanapokin's study [35], which suggests that most accidental falls involving distant lavatories are caused by improper lighting or rough and wet flooring en route to the bathroom. Such issues are more prevalent in developing countries [2, 36]. Residential layout also plays an important role in fall risk. Since individuals have diverse health and mobility, the current study did not consider the use of squat toilets, lack of handrails and condominium-living as potential contributing causes of accidental falls, unlike earlier studies [37].

The elderly people in the 2017 survey reported a negative association with falls although the prevalence of accidental falls in 2017 was 3.2%, less than the prevalence of accidental falls in 2007 (4.3%) which is the reference year. Although there were fewer falls, there was a higher level of awareness in the elderly concerning risk factors in 2017.

The significance of this study lies in its definition and description of the main characteristics of elderly individuals, as well as its development of a framework related to the history of accidental falls in Thailand. The latter will enhance understanding of issues concerning an aging population, like that of Thailand. This study discussed the circumstances surrounding accidental falls which took place in 2007, 2011, 2014 and 2017 and identified the risk factors related to those events. The study has laid a foundation for further research and will prove essential to the implementation of various plans to mitigate the fall risk for elderly people around the world – or in Thailand at the very least.

This study has limitations. The first involves the use of the cross-sectional survey method for data collection. This method failed to divide the variables based on their cause–effect correlations. Moreover, there is higher probability of recall bias in the study, specifically when participants have an impaired or weak memory. Another limitation is the underreporting of the number of falls since only the data pertaining to the previous six months were considered in each survey; falls that took place beyond this period were excluded.

It is important for other experts to replicate the outcomes of this study since it is likely to involve reporting errors that relate to the context of the falls or their locations. It is recommended that a prospective study is conducted regarding fall events and their contextual factors. The questionnaire was largely focused on the actual fall events, thereby overlooking the contextual fall-related factors specifically associated with the elderly. The overlooked factors included: drugs taken by elderly, work-related falls, and individuals' balance while standing, walking, and turning. The social aspect of accidental falls also calls for additional research. The same is true for the unexplored environmental factors. The lack of focus on the social and environmental components of accidental falls acts as a limitation in this study. Since only healthy and physically fit seniors were involved, the outcomes may not be applied to the general elderly population of Thailand.

Policy recommendations

The elderly may experience accidental falls due to multiple factors. Such factors have been identified as age, occupational position, gender, health conditions and the location of lavatory facilities. A medical expert assigned to evaluate elderly patients and assess fall risk must take all these factors into consideration. This may be followed by studying the problems faced by these patients in order to determine variable risk factors like health and environmental conditions. It is recommended that the Thai public too, consider the aforementioned fall risk factors and mitigate such factors while spreading awareness about them. Moreover, elderly persons in Thailand tend to not retire at age 60, so there is a critical need to introduce fall prevention plans in workplaces. The drafting process of such plans requires further research.

Despite the formulation of a community-based plan for mitigation of fall risks at the hospital level, there is currently a lack of such plans on the national front [38]. The efficiency of existing local plans also requires additional research that evaluates these plans based on the risk factors identified in this study. An effective state-level, evidence-based plan for the prevention of accidental falls is decidedly overdue, particularly considering the increasingly aging Thai population.

Conclusion

This study is the first to explore accidental fall risk factors in the elderly who have good health and no disabilities. Accidental falls have been shown to be more common in elderly people, specifically those aged 60 and above who worked during the week before the research interview. With aging of the population, investigating risk factors for fall-related occupational accidents could contribute to fall prevention programs in workplaces. Moreover, elderly who live in rural areas tend to fall more than their urban counterparts, and women in this demographic have been shown to be more prone to falling than men. Intrinsic and extrinsic factors contribute to the prevalence and incidence of accidental falls. Factors in both categories must be considered by public health authorities. These authorities should, in turn, devise appropriate policies to allocate additional resources to mitigate the risk of accidental falls in elderly people residing in rural areas.

Conflict of Interest: None

Figures

Figure 1

Weighted prevalence (percent) of the accidental fall by year

Figure 2

Proportional impacts of modeled independent variables on prevalence of accidental falls

Table 1

New study variables

Variables	New value label
Dependent variable Accidental falls	0 = no; 1 = yes
Independent variable Age (years)	Continuous
Female	0 = no; 1 = yes
Rural	0 = no; 1 = yes
Living alone	0 = no; 1 = yes
Education	0 = any education; 1 = no education
Marital status	0 = never married (reference), 1 = married, living as a couple, 2 = married, not living as a couple
Income	0 = insufficiency; 1 = sufficiency
Work in the past 7 days	0 = not work; 1 = work
Vision loss	0 = no; 1 = yes
Loss of hearing	0 = no; 1 = yes
Urinary or fecal incontinence	0 = no; 1 = yes
Exercise	0 = no; 1 = yes
Alcohol consumption	0 = no; 1 = yes
Smoke	0 = no; 1 = yes
Type of house	0 = Detached house (reference), 1 = town house, 2 = condominium
Second floor bedroom	0 = no; 1 = yes
Toilet outside home	0 = no, inside home; 1 = yes, outside home
Squat toilets	0 = no; 1 = yes
No handrails on stair	0 = no; 1 = yes
No handrails in bedroom	0 = no; 1 = yes
No handrails in bathroom	0 = no; 1 = yes
Bangkok	0 = no; 1 = yes
Central	0 = no; 1 = yes
North	0 = no; 1 = yes
Northeast	0 = no; 1 = yes
South	0 = no; 1 = yes (reference)
Year 2007	0 = no; 1 = yes (reference)
Year 2011	0 = no; 1 = yes
Year 2014	0 = no; 1 = yes
Year 2017	0 = no; 1 = yes

Table 2

Characteristics of the study population by year, weighted by probability weight

	2007 N (%)	2011 N (%)	2014 N (%)	2017 N (%)	Total N (%)
N	11,370	11,061	13,775	12,457	48,663
Demographic
Mean age	66.22 ± 5.63	66.39 ± 5.79	66.38 ± 5.85	65.75 ± 5.45	66.18 ± 5.69
Age (yrs)
60–69	8,666 (76.2)	8,267 (74.7)	10,352 (75.2)	9,947 (79.9)	37,232 (76.5)
70–79	2,396 (21.1)	2,448 (22.1)	2,901 (21.1)	2,166 (17.4)	9,911 (20.4)
80–89	281 (2.5)	331 (3.0)	499 (3.6)	329 (2.6)	1,440 (3.0)
≥ 90	27 (0.2)	15 (0.1)	23 (0.2)	15 (0.1)	80 (0.2)
Male	6,308 (55.5)	5,941 (53.7)	7,553 (54.8)	6,947 (55.8)	26,749 (55.0)
Rural	8,015 (70.5)	7,197 (65.1)	8,114 (58.9)	7,143 (57.3)	30,469 (62.9)
No education	1,241 (10.9)	822 (7.4)	880 (6.4)	718 (5.8)	3,661 (7.5)
Income insufficiency	3,888 (34.2)	3,389 (30.6)	3,724 (27.0)	4,419 (35.5)	15,420 (31.7)
Work in the past 7 days	5,931 (52.2)	6,123 (55.4)	7,919 (57.5)	6,928 (55.6)	26,901 (55.3)
Living arrangement
Live with other	10,594 (93.2)	10,184 (92.1)	12,730 (92.4)	11,229 (90.1)	44,737 (91.9)
Live alone	776 (6.8)	877 (7.9)	1,045 (7.6)	1,228 (9.9)	3,926 (8.1)
Marital status
Married live together	7,974 (70.1)	7,538 (68.1)	9,785 (71.0)	8,659 (69.5)	33,956 (69.8)
Married but not live together	3,130 (27.5)	3,057 (27.6)	3,407 (24.7)	3,143 (25.2)	12,737 (26.2)
Never married	266 (2.3)	466 (4.2)	583 (4.2)	656 (5.3)	1,971 (4.1)
Geographic
Rural	8,015 (70.5)	7,197 (65.1)	8,114 (58.9)	7,143 (57.3)	30,469 (62.9)
Bangkok	1,258 (11.1)	1,032 (9.3)	1,168 (8.5)	1,163 (9.3)	4,621 (9.5)
Central (excluded Bangkok)	2,397 (21.1)	2,119 (19.2)	3,205 (23.3)	2,903 (23.3)	10,624 (21.8)
North	2,218 (19.5)	2,379 (21.5)	3,085 (22.4)	2,896 (23.2)	10,578 (21.7)
Northeast	4,080 (35.9)	3,878 (35.1)	4,640 (33.7)	3,904 (31.3)	16,502 (33.9)
South	1,416 (12.5)	1,653 (14.9)	1,678 (12.2)	1,592 (12.8)	6,339 (13.0)
Vision loss	3,633 (32.0)	4,045 (36.6)	4,463 (32.4)	4,232 (34.0)	16,373 (33.7)
Loss of hearing	646 (5.7)	676 (6.1)	753 (5.5)	502 (4.0)	2,577 (5.3)
Urinary or fecal incontinence	942 (8.5)	2,072 (18.8)	1,506 (10.9)	722 (5.8)	5,242 (10.8)
Health behavior
Exercise	9,957 (87.8)	9,462 (86.3)	11,905 (86.4)	10,651 (85.6)	41,975 (86.5)
Alcohol consumption	3,355 (29.6)	2,619 (23.8)	3,317 (24.1)	3,231 (25.9)	12,522 (25.8)
Smoking	2,780 (24.5)	1,950 (17.7)	2,486 (18.0)	1,999 (16.1)	9,215 (19.0)
Type of house
Detached house	10,132 (89.1)	6,587 (90.9)	12,272 (89.1)	10,965 (88.0)	39,956 (89.1)
Townhouse	459 (4.0)	215 (3.0)	761 (5.5)	752 (6.0)	2,187 (4.9)
Condominium	779 (6.9)	445 (6.1)	743 (5.4)	740 (5.9)	2,707 (6.0)
Second floor bedroom	4,211 (37.1)	3,730 (33.7)	4,426 (32.1)	3,550 (28.5)	15,917 (32.7)
Outside toilet	3,214 (28.4)	2,401 (21.7)	2,607 (18.9)	2,109 (17.0)	10,331 (21.3)
Squat toilet	8,496 (75.0)	7,440 (67.3)	7,780 (56.5)	6,003 (48.3)	29,719 (61.2)
No handrails on stair	4,218 (37.1)	2,957 (40.8)	5,267 (38.2)	5,022 (40.3)	17,464 (39.0)
No handrails in bedroom	11,177 (99.0)	6,922 (95.5)	13,463 (97.7)	12,138 (97.7)	43,700 (97.7)
No handrails in bathroom	10,958 (97.1)	6,810 (94.0)	12,827 (93.1)	11,509 (92.5)	42,104 (94.1)
Prevalence of accidental fall	489 (4.3)	534 (4.8)	855 (6.2)	393 (3.2)	2,271 (4.7)
Fall place
Inside house	183 (31.9)	190 (32.8)	274 (28.7)	112 (26.4)	759 (30.0)
Outside house	391 (68.1)	389 (67.2)	682 (71.3)	312 (73.6)	1,774 (70.0)
Hospital visit after a fall
Outpatient	544 (94.6)	561 (97.1)	921 (96.2)	408 (96.0)	2,434 (96.0)
Inpatient	31 (5.4)	17 (2.9)	36 (3.8)	17 (4.0)	101 (4.0)

Note(s): Accidental falls in this study were older people (good health and did not have any disability) who fell from obstacles while crossing a raised floor, slipped, fell from poor lighting and fell on steps and stairs

Townhouse includes duplex house or semidetached house

Condominium includes mansion, apartment, dormitory, flats, van and boat

Table 3

Logistic regression in overall four years (sample probability weights)

Independent variable	Coefficient	Odds ratio (95% CI)	p value
Age (years)	0.13	1.17 (0.96–1.28)	<0.001
Female vs male	0.19	1.21 (1.08–1.35)	0.001
Rural vs urban	0.17	1.18 (1.06–1.32)	0.003
Living alone vs living with others	−0.07	0.93 (0.79–1.10)	0.398
No education vs any education	−0.28	0.76 (0.64–0.90)	0.062
Married (couple) vs never married	−0.04	0.96 (0.75–1.23)	0.723
Married (not together) vs never married	0.13	1.13 (0.89–1.45)	0.318
Income sufficiency vs income insufficiency	0.10	1.10 (1.00–1.21)	0.055
Work in the past 7 days	0.49	1.64 (1.48–1.81)	<0.001
Vision loss	−0.06	0.94 (0.86–1.03)	0.187
Loss of hearing vs normal hearing	−0.41	0.67 (0.56–0.79)	0.061
Urinary or fecal incontinence	−0.12	0.89 (0.78–1.00)	0.053
Exercise vs no exercise	0.37	1.44 (1.27–1.64)	<0.001
Drink alcohol vs no drink alcohol	0.20	1.22 (1.08–1.38)	0.001
Smoke vs no smoke	0.28	1.32 (1.16–1.50)	<0.001
Town house vs detached house	−0.09	0.91 (0.71–1.17)	0.464
Condominium vs detached house	−0.05	0.95 (0.77–1.18)	0.654
Second floor bedroom vs first floor bedroom	−0.04	0.96 (0.86–1.07)	0.492
Toilet outside home vs toilet inside home	0.10	1.10 (0.99–1.23)	0.037
Squat toilets vs western sit-down toilets	−0.06	0.94 (0.84–1.05)	0.265
No handrails on stair	−0.02	0.98 (0.89–1.08)	0.698
No handrails in bedroom	0.10	1.11 (0.80–1.54)	0.543
No handrails in bathroom	0.12	1.13 (0.92–1.40)	0.248
Bangkok vs south	−0.16	0.85 (0.68–1.08)	0.185
Central vs south	−0.14	0.87 (0.74–1.01)	0.076
North vs south	−0.08	0.93 (0.79–1.09)	0.349
Northeast vs south	−0.02	0.98 (0.84–1.14)	0.800
Year 2011 vs 2007	0.13	1.13 (0.98–1.31)	0.082
Year 2014 vs 2007	0.39	1.27 (1.31–1.66)	<0.001
Year 2017 vs 2007	−0.45	0.64 (0.55–0.73)	<0.001
Constant	−2.89	0.06	<0.001

Note(s): Married but not living as a couple includes widowed, divorced and separated

Townhouse includes duplex house

Condominium includes mansion, apartment, dormitory, flats, van and boat

Appendix 1

https://drive.google.com/file/d/1s4A_CaldG4WVZvGEMkygExmrs2Tq0B4Q/view

Appendix 2

https://drive.google.com/file/d/1_yHqIgtahCQrp4k6sCuE89GcLwNQ6mV-/view