Association between chronic diseases and falls among a sample of older people in Finland

Falls of older adults are globally a major health, economic and societal problem. One third of people over 65 years old fall at least once every year [1‐5] and the number of falls per year increases with age and frailty level [6, 7]. Falls are associated with increased mortality, morbidity, reduced functioning, impaired quality of life and premature nursing home admissions [8]. Fear of falling may increase even after a non-injurious fall [9]. It can lead to a negative cycle where the activity of the person reduces, followed by a reduction in functionality. That, in turn, increases the need for help in daily activities [10, 11]. The world’s population is ageing rapidly; the number of people aged 65 or older is expected to grow from 524 million in 2010 to nearly 1.5 billion in 2050 [12]. Rapidly increasing ageing populations are a challenge to limited social and health care systems. At the same time, the costs resulting from falls and their consequences are increasing rapidly, and the importance of fall prevention grows. Screening for high fall risk persons consumes resources and public funding. Efficient methods for screening fall risk are needed to find the persons at most risk to target preventive actions towards them. Identifying recurrent fallers is especially important as they are more prone to severe consequences compared to non-recurrent fallers. For example, in a prospective study by Pluijm et al. 11.9% of recurrent fallers (at least two falls within 6 months) had a fall-related fracture compared to 3.4% of non-recurrent fallers [13]. One-time fallers (1 fall within 1 year) and recurrent fallers are reported to have distinct characteristics with respect to age, physiology, cognition, fear of falling and ability to perform daily activities [14‐16].

Several chronic conditions have been identified to be associated with the higher prevalence of falling, including Parkinson’s disease [17, 18], heart failure [19], cognitive impairments [1, 20], diabetes [21], arthritis [22], depression [7, 23] and chronic kidney disease [24, 25]. In addition, conditions such as osteoporosis further increase the risk of fractures [26]. Along with the age, the prevalence of multiple chronic conditions increases [27‐29], and earlier studies have identified co-occurrence of diseases and multi-morbidity patterns among older adults [30‐32]. An association between history of falls and the number of diseases and the number of health conditions have been found, and the amount of falls rises with the number of diseases [32, 33]. Only a few studies have assessed, if there is a relationship between falls and multiple chronic conditions or multi-morbidity patterns [32, 34, 35]. Related to multi-morbidity studies, an agreement on which diseases should be included in the multi-morbidity assessment is missing. Based on a systematic literature review, Diederichs et al. recommended to include a minimum of 11 diseases (cancer, diabetes mellitus, depression, hypertension, myocardial infarction, chronic ischaemic heart disease, heart arrhythmias, heart insufficiency, stroke, chronic obstructive pulmonary disease, and arthritis) that are highly prevalent in the older population in the multi-morbidity assessment [36]. Moreover, Fortin et al. recommended that investigators designing future studies to assess the prevalence of multi-morbidity should consider the number of diagnoses to be assessed (with 12 or more frequent diagnoses of chronic diseases appearing ideal) and should attempt to report results for differing definitions of multi-morbidity (both 3 or more diseases and the classic 2 or more diseases) [37]. Recently, Griffith et al. concluded that the suitability of existing multi-morbidity measures for use in a specific research study depends on study-related factors such as purpose of the study, outcomes examined and preferences of the involved stakeholders [38]. In an extensive study using self-reported falls and disease data from 16,357 community dwelling older adults, Sibley et al. found falls to associate with a number of chronic conditions and with chronic disease patterns [32]. Another recent study with 149,876 respondents resulted in two “at risk” populations characterized by different chronic disease combinations and age. The authors concluded that these groups might benefit from different risk stratification and interventions [35]. In a sample of 2566 Swedish older adults, unhealthy lifestyle and high burden of chronic diseases were found to increase the risk of falling over long follow-up periods (5 and 10 years) [34]. Identifying diseases and possible individual clusters of certain conditions associated with risk of falling would help to develop solutions for screening those most at risk among older populations. Utilizing existing data saved on electronic health record provides a possibility for easy and cost-effective screening for high fall risk persons. Even though specific diseases have been identified to increase the risk of falling, and the number of chronic conditions have been linked to higher risk of falling, current fall prevention guidelines do not instruct measuring the amount of chronic conditions. The amount of earlier studies linking the amount or individual clusters of chronic conditions with higher fall risk is still very small [32, 34, 35]. Verifying the results with another sample help in improving fall prevention guidelines by recommending comprehensive fall assessment, or interventions aiming for fall prevention, among older adults with combinations of certain conditions that are connected with higher risk of falling. Thus, the aim of this study was to reveal the prevalence of falls and the association between number of chronic conditions and occurrence of recent falls, and the association between individual or multiple chronic conditions and occurrence of recent falls in a population-based sample of older people. In addition, demographic differences and lifestyle characteristics between non-fallers, one-time fallers and recurrent fallers were studied.

A survey was conducted between November 2014 and January 2015 in Oulu, Finland. Oulu has a population of approximately 200,000 and it is Finland’s fifth largest city, located in Northern Finland. Of the population, 14% was aged 65 years or more at the end of 2014 [39]. The survey was part of a research study: Tailored Services for Elderly - Gamified remote service concept for promoting health of older people (GASEL-project, 2014–2016). Eleven volunteer seniors piloted the study questionnaire, and their feedback was used to make minor usability-related changes. A random sample of 1500 people living in Oulu was obtained from the Population Register Centre of Finland. The sampling criteria were: 1) born during 1.1.1900–31.12.1949 (aged 65–114 by the end of 2014); 2) spoke Finnish as a native language; and 3) had a permanent living address in Oulu in November 2014. A reminder and another copy of the questionnaire were sent to the non-responders 4 weeks after the first survey. The sample of 1500 persons represents approximately 5.5% of the over 65-year-old persons living in the city of Oulu and 918 persons (61.2%) responded to the survey (3.3% of the over 65-year-olds).

The study protocol was approved by the Ethics Committee of Human Sciences at the University of Oulu (statement 6/2014). The extensive questionnaire included items related to the participants’ socio-demographic background, perceived loneliness, stress, health, medication, social participation, personality, physical activity, functional ability, occurrence of recent falls, fear of falling, Internet and communications technology (ICT) use, and gaming. The questionnaire is available as supplementary material (Supplementary file 1, GASEL questionnaire (English version)).

To determine if the respondent had diseases diagnosed by a medical professional the question: “Do you currently have any of the following conditions as diagnosed by a medical professional?” and the answer options “yes” or “no” were used. The complete list of asked diseases can be found from the Supplementary file 1.

Frequency of recent falls was determined by the question: “Have you fallen during the last three (3) months? (A fall is defined as an event that results in a person coming to rest inadvertently on the ground or floor or other lower level).” The answer options were: “no” or “yes, ____ times.” Three month recall time was chosen, because in the target group the possibility of poorer cognitive function may lead to potential bias with longer recall time [40, 41].

The respondents were categorized according to their self-reported alcohol use into three categories: no alcohol, light drinker (1–3 drinks per week), moderate (4–7 drinks per week for women and 4–14 drinks per week for men), and heavy (≥8 drinks per week for women and ≥ 15 drinks per week for men) [42]. Smoking was categorized according to their selection of options: 1) I don’t smoke at all, 2) I quit smoking in year____, 3) I currently smoke ____ cigarettes a day, 4) I use snuff or other tobacco products other than cigarettes. Activity level was determined by the question: “How long are you active each day? (e.g. biking/walking, housework, walking to the store, exercise hobbies, etc.)? 1) < 1 hour per day, 2) 1-2 hours per day, and 3) > 2 hours per day” [43]. Sleep quality was determined by the question: “How do you feel about the quality of sleep you are currently getting (regardless of how much you sleep)?” and the answer options: 1) good 2) satisfactory, or 3) poor. “Can you rise from a chair independently without using your hands?” was used as a question to determine if the person was able to stand up from a chair. The answer options were: 1) I am able to rise independently without using my hands, 2) I am able to rise independently using my hands on first try, 3) I am able to rise using my hands after several tries, 4) I need minimal assistance to rise, and 5) I need moderate or maximal assistance to rise.” A modified SOF Frailty Index (Study of Osteoporotic Fractures) [44] described in our earlier study [45] was used to identify those with frailty symptoms. The modified items were: 1. Weight loss irrespective of intention to lose weight (if weight loss in 3 months is equal or more than 1 kg, score = 1), 2. “Can you rise from a chair independently without using your hands?” (if unable to rise from a chair, score = 1), and 3. Poor energy as identified by “Which of the following best describes how energetic you have felt in the last month?” (with the response “I feel moderately/very/extremely exhausted, tired or powerless”, score = 1). The subject’s status was categorized as frail if the SOF index was 2 or 3; intermediate if the SOF index was 1, and robust if the SOF index was 0. The Mini Nutritional Assessment (MNA) [46] includes seven questions about decrease in nutrition, weight loss, mobility, stress or acute illness, neuropsychological problems (dementia or depression), and Body Mass Index (BMI). Compared to the original MNA questionnaire, in our study weight loss was determined by the question: “In the last 3 months, has your weight 1) decreased by____ kg (if the decrease is more than 3 kg, score = 0, if the decrease is between 1 kg and 3 kg, score = 2), 2) stayed the same (score = 3), 3) increased by _____ kg (score = 3), and 4) I don’t know (score = 1).” Moreover, the question about psychological problems was modified: “Do you currently have any of the following conditions as diagnosed by a medical professional?” was used as a question to discover if the respondent had memory disorders (mild or moderate to severe), other neurological illness, or depression, with the answer options “yes” or “no.” In our modified MNA, 0 points is given if the respondent had moderate or severe memory disorder and/or depression, 1 point is given for mild memory disorder and 2 points is given if the respondent didn’t have memory disorders or depression. An individual is considered to have malnutrition if their total MNA score is 0–7 points, to be at risk of malnutrition at 8–11 points, and to have normal nutrition at 12–14 points.

Eligible data on 872 participants were available. Of the respondents, 116 (13.3%) had fallen during the last 3 months. Seventy-one (8.4%) had fallen once, 27 (3.1%) two times, 10 (1.1%) three times, 3 (0.3%) four times and 5 (0.6%) more than five times. The characteristics of the study participants are presented in Table 1. Non-fallers were younger than the one-time or recurrent fallers (mean age 72.7, 76.6 and 76.0 years, respectively, p < 0.001). Categorization as frail was more common among recurrent fallers than non-fallers or one-time fallers (40.0, 5.7 and 7.6%, respectively, p < 0.001) and fallers were less frequently able to get up from a chair without using hands or without assistance than non-fallers or one-time fallers (39.7, 86.2, 79.4%, p < 0.001). Fallers had symptoms of malnutrition more often than non-fallers (37.5% for recurrent fallers, 22.5% for one-time fallers and 12.3% for non-fallers, p = 0.005). Recurrent fallers also had poor self-reported sleep quality more often than non-fallers or one-time fallers (19.5, 7.1 and 12.5%, respectively, p = 0.036).

The occurrence of chronic diseases and common geriatric conditions, urinary incontinence and dizziness, in the groups of fallers and non-fallers are shown in Table 2. The median number of diseases differed significantly in the groups of non-fallers, one-time fallers and recurrent fallers (medians 2, 3 and 4, respectively; p < 0.001). The frequency of rheumatoid arthritis, blood circulation problems in the brain and legs, cardiac failure, osteoporosis, Parkinson’s disease, moderate or severe cognitive impairment, mild cognitive impairment, depression, urinary incontinence, and dizziness were higher among recurrent fallers than non-fallers or one-time fallers (Table 2).

The number of chronic diseases and their relationship to recurrent falls are shown in Table 3. According to age and gender -adjusted multinomial logistic regression, the respondents with five or more chronic diseases significantly more often reported having two or more falls during the last 3 months than the participants with less chronic diseases. The odds ratio for having experienced recurrent falls was 11.70 (95% CI: 2.99–46.44) with five chronic diseases and 8.60 (95% CI: 2.10–35.17) with six or more chronic diseases compared to no chronic diseases.

Agglomerative hierarchical clustering with Ward’s linkage resulted in eight clusters. The formed clusters are shown in Table 4. Cluster 1 participants had a low number of chronic conditions. Cluster 2 participants had the highest prevalence of elevated blood pressure. Cluster 3 participants had coronary artery disease, and heart failure as well as diabetes. Cluster 4 participants had asthma and COPD, and Cluster 5 participants hypothyroidism and diabetes. Cluster 6 in the main included participants with osteoporosis, but the prevalence of elevated blood pressure was not seemingly high compared to the other clusters. Cluster 7 consisted of participants with blood circulation problems in the legs. The multiple chronic disease Cluster 8 included 32 participants who had a higher prevalence of multiple chronic conditions; elevated blood pressure, diabetes, coronary heart disease, heart failure, cognitive impairments, blood circulation problems in brain and legs, asthma, COPD and rheumatoid arthritis. The formed clusters were labeled after the disease, which prevalence in the cluster exceeds the prevalence in the total sample and the prevalence in the other clusters, except for the elevated blood pressure. In the split sample for validation, agglomerative hierarchical clustering formed similarly eight clusters, of which one cluster included healthy participants and one cluster included participants with multiple chronic conditions.

The identified clusters also differed with regard to demographic characteristics and geriatric conditions, as shown in Table 5. The participants in Cluster 1 were younger, more physically active, more frequently non-frail, and had smaller prevalence of geriatric conditions such as urinary incontinence and dizziness. In contrast, multi-morbid Cluster 8 participants were older, less physically active, more often overweight (BMI > 30), at risk of malnutrition, and had higher prevalence of geriatric conditions such incontincence and dizziness.

Multinomial logistic regression analyses (Table 6) showed that Cluster 6 (osteoporosis, OR = 5.65, 95% CI: 1.23–25.85, p = 0.026) and Cluster 8 (multiple chronic disease, OR = 13.42, 95% CI: 2.47–72.96, p = 0.002) were significantly associated with recurrent falling (≥2 falls during the last 3 months) relative to Cluster 1 (low-chronic). None of the clusters was associated with higher risk of one-time falling. In the validation sample, cluster 8 (multiple chronic disease), was also significantly associated with recurrent falling (p = 0.006) and none of the clusters formed with the validation sample were associated with one-time falling.

This study examined the association between chronic diseases and falls in a sample of older people (age 65–98) and demonstrated that fall incidence among community-dwelling older people is highly linked to presence of chronic disease. Individuals with five, six or more chronic diseases had significantly higher fall rates compared to those without diseases and several individual chronic diseases were associated with increased fall rates. The study showed, by clustering chronic diseases, that combinations of chronic diseases play a role in falls. The group with multiple chronic conditions had significantly higher risk of recurrent falling relative to the group of low chronic disease subjects. Osteoporosis did not cluster clearly with other diseases, but the group with osteoporosis was associated with increased risk of recurrent falls when compared with the group with low chronic disease.

The prevalence of recurrent fallers increased with age, which aligns with previous findings in the literature [3, 4, 6, 8, 48]. We also found that the percentage of subjects who reported poor sleep quality is higher in the recurrent fallers group and one-time fallers group compared to non-fallers (19.5, 12.5, 7.1%, respectively), and similar findings were made regarding frailty according to the modified SOF-index (40, 7.6 and 5.7%), and malnutrition according to MNA (37.5, 22.5 and 12.3%). The prevalence of recurrent fallers is higher among older adults with self-reported urinary incontinence and dizziness, which supports the findings in other studies [30, 49]. The number of falls differed between subjects with diabetes, rheumatoid arthritis, blood circulation problems in the brain and legs, cardiac failure, Parkinson’s disease, osteoporosis, moderate or severe cognitive impairment, mild cognitive impairment, depression, dizziness, and urinary incontinence compared to subjects without the specific condition. These diseases and geriatric conditions have been found to increase fall risk also in previous studies [1, 7, 17‐26].

Our results showed a correlation between multiple chronic diseases and recurrent falls. Compared to the no disease group with age adjustment, increase in the risk of recurrent falls was found in the group having five or more diseases. This finding is supported by the findings in previous studies [32, 50]. We also found by clustering diseases that the multiple chronic disease group had a significantly higher risk of having fallen recurrently during the last 3 months than the group of low chronic diseases. Subjects in the osteoporotic cluster also had a high prevalence of elevated blood pressure and were more likely to have fallen during the last 3 months. It is also important to note that not all clusters were associated with high fall risk and none of the disease clusters was significantly associated with higher risk of one-time falling. This finding supports the previous findings of different characteristics of one-time fallers and recurrent fallers [14‐16]. All clusters, except the low chronic cluster, had a notable percentage of subjects with elevated blood pressure (43.9–87.5%). The highest prevalence was in the multiple chronic disease cluster and elevated blood pressure was, in fact, the most common disease within that cluster. Our results partly align with previous studies. Sibley et al. found in their Canadian population-based study that clusters of hypertension and COPD had higher risk of falling than the low chronic diseases cluster [32]. In an earlier study by Paliwal et al., a positive association was found between the number of falls and certain diseases; heart attack, angina, stroke, asthma, COPD, chronic kidney disease, arthritis, depression, and diabetes [35]. The combinations of diseases associated with higher prevalence of falling differ between our study and the studies by Sibley et al. and Paliwal et al. This may be due to slightly different questionnaires used to collect information on medical history and diagnoses and different lists of asked diseases. In addition, we had a relatively low sample size compared to previous two studies and we had to exclude some of the diseases from clustering due to low prevalence of these diseases in the total sample. Similarly as in the previous studies, we did not collect any information on the treatment of the diseases, which may have had an impact on the results. National differences may also exist. In another study, Ek et al. used sociodemographic and lifestyle-related characteristics in forming clusters and had a follow-up period of 5 years. Their finding was that participants with unhealthy lifestyle and a high burden of chronic disease had higher risk of falling over a longer period of follow-up (5 and 10 years) compared to those in the reference group [34].

The strength of this study is its population-based sample and analysis as well as its rather unique approach of clustering diseases and studying their association with falls. However, some limitations of this study need to be taken into account. Firstly, the questionnaire was completed only once and no follow-up period was included. Furthermore, we have no information if the diagnosis of the certain chronic disease was given before or after the occurrence of falls, and no specific information about the diseases, such as the type or location of cancers. We also did not study how severe the diseases were or how the diseases or conditions were controlled or medicated. Cognitive impairments may also affect the results. The prevalence of certain diagnosis of some pathologies such as dementia or osteoporosis may be underestimated. Some diseases have low prevalence and, for example, Parkinson’s disease was not included in the cluster analysis even though it is commonly known to produce a high risk of falls. The three-month recall time may also lead to absence of actual recurrent falls in our study sample and the actual amount of recurrent fallers may be higher [51]. A larger population sample, for example from health records, would give more information about the chronic diseases and falls and show the potential of using health records in automatic screening of persons with high fall risk.

Our study gives promising results about the association between diseases and falls and further implicates that using data sources such as patient health records for screening for high fall risk individuals is important. In addition, the use of machine learning methods in health records for screening for high-risk persons is still unexploited. Future studies should utilize available data sources from patient records to study the association between combination of diagnoses and falls. Studies with large data sets should consider utilizing machine learning when analyzing the relationship between clusters of chronic conditions and falls.

This study supports the earlier findings that risk of recurrent falls increases with age. The results further indicate that persons with certain combinations of chronic diseases may have elevated fall risk and the risk is highest with multiple chronic conditions. Multiple chronic disease cluster participants were, in general, older, less physically active, overweight, at risk of malnutrition, and had more geriatric conditions compared to the low chronic cluster. In addition to the multiple chronic disease cluster, clusters where osteoporosis and elevated blood pressure were the defining factors had significantly higher risk of falls. Notably, persons with five or more chronic conditions reported significantly more often having fallen at least two times within the last 3 months. Although further studies with a larger population are needed, these results implicate the potential of identifying high-risk persons by analyzing health record data. Finally, the results show that certain combinations of chronic diseases, especially multiple chronic conditions, increase the risk of recurrent falls.