Introduction
Late-life depression is a debilitating illness that contributes to excess morbidity, mortality, and healthcare costs (Bock et al., Reference Bock2016; Hall and Reynolds-III, Reference Hall and Reynolds-III2014; Hawkins et al., Reference Hawkins, Callahan, Stump and Stewart2014; Luppa et al., Reference Luppa, Sikorski, Motzek, Konnopka, König and Riedel-Heller2012). Conversely, medical morbidities increase the risk of late-life depression (Chang et al., Reference Chang, Pan, Kawachi and Okereke2016; Lyness et al., Reference Lyness, Yu, Tang, Tu and Conwell2009). This bidirectional relationship between depression and medical morbidities extends to falls and fall injuries in older adults and becomes stronger with increasing frequency of falls and severity of fall injuries and depressive symptoms (Eggermont et al., Reference Eggermont, Penninx, Jones and Leveille2012; Kvelde et al., Reference Kvelde2013; Lenze et al., Reference Lenze2007). Significant associations also exist between depression and fall worry, otherwise known as fear of falling (Bruce et al., Reference Bruce, Hunter, Peters, Davis and Davis2015; Bryant et al., Reference Bryant, Rintala, Hou and Protas2015; Goh et al., Reference Goh, Nadarajah, Hamzah, Varadan and Tan2016; Hughes et al., Reference Hughes, Kneebone, Jones and Brady2015; Moreira Bde et al., Reference Moreira Bde2016; Painter et al., Reference Painter, Allison, Dhingra, Daughtery, Cogdill and Trujillo2012; Patil et al., Reference Patil, Uusi-Rasi, Kannus, Karinkanta and Sievänen2014; van Haastregt et al., Reference van Haastregt, Zijlstra, van Rossum, van Eijk and Kempen2008). These relationships between depression and falls and fall worry likely result from common/shared risk factors including chronic illnesses, functional impairments, pain, and cognitive and sensory impairments (Iaboni and Flint, Reference Iaboni and Flint2013; Peeters et al., Reference Peeters, Leahy, Kennelly and Kenny2018).
Given the serious negative sequelae of fall injuries, e.g., fractures, hospitalizations, disability, loss of independence, institutionalization, and/or death (Bergen et al., Reference Bergen, Stevens and Burns2016; Burns et al., Reference Burns, Stevens and Lee2016; Tinetti and Williams, Reference Tinetti and Williams1997), it is not surprising that fall worry is common (ranging from 25% to 50%) among older adults regardless of whether they have experienced a fall, and that those who fall tend to have increased fear of falling again (Iaboni and Flint, Reference Iaboni and Flint2013; Parry et al., Reference Parry, Finch and Deary2013). Some fall worry is warranted and may be beneficial for those at high risk of falling due to functional, cognitive, and sensory impairments. However, unhealthy/excessive fall worry is associated with avoiding or restricting physical and recreational activities, which in turn contributes to a sedentary life style, and subsequently, increased postural and gait imbalance, frailty, functional disability, fall risk, and falls (Allison et al., Reference Allison, Painter, Emory, Whitehurst and Raby2013; Auais et al., Reference Auais, French, Alvarado, Pirkle, Belanger and Guralnik2018; Denkinger et al., Reference Denkinger, Lukas, Nikolaus and Hauer2015; Deshpande et al., Reference Deshpande, Metter, Bandinelli, Lauretani, Windham and Ferrucci2008; Lavedán et al., Reference Lavedán2018; Makino et al., Reference Makino2018; Stubbs et al., Reference Stubbs, Patchay, Soundy and Schofield2014).
Depressive symptoms and activity-limiting fall worry are likely to reinforce each other since both lead older adults to reduce health-promoting physical, social, and recreational activities (Hull et al., Reference Hull, Kneebone and Farquharson2013; Kobayashi and Steptoe, Reference Kobayashi and Steptoe2018). For example, anhedonia, a prominent aspect of late-life depression, results in avoiding healthy behaviors, including physical activities, further weakening physical strength reserve among older adults, especially those with other health conditions, which, in turn, increases fall risk and fear of falling (Lee et al., Reference Lee2017; Stubbs et al., Reference Stubbs, Stubbs, Gnanaraj and Soundy2016). Activity-limiting fall worry is also likely to exacerbate depression by further increasing activity avoidance, fall risk, and social isolation. Along with its depressionogenic effects, social isolation can contribute to morbidity (Cacioppo et al., Reference Cacioppo, Hawkley, Norman and Berntson2011), which increases fall worry and falls.
Though preventable, late-life depression, falls, and fall worry remain public health problems (Bock et al., Reference Bock2016; Florence et al., Reference Florence, Bergen, Atherly, Burns, Stevens and Drake2018). A large body of cross-sectional research confirms the relationship between depression and fall worry, but few longitudinal studies have examined whether changes in fall worry are associated with changes in depressive symptoms or depressive illness and vice versa. It is important to examine whether changes in depressive symptoms are associated with changes in fall worry and vice versa over time. By better understanding the longitudinal and potentially reciprocal relationships between fall worry and depression, more targeted interventions to prevent falls and prevent and treat depression can be designed and implemented.
This study examined longitudinal relationships between depression and activity-limiting fall worry using two waves of interview data (time 1 [T1] and time 2 [T2], one year after T1) using a nationally representative sample of U.S. older-adult Medicare beneficiaries. Hypotheses were: (H1) new or continued activity-limiting fall worry between T1 and T2 will be associated with greater odds of having T2 probable major depression, controlling for T1 probable major depression; and (H2) new or continued probable major depression between T1 and T2 will be associated with greater odds of having T2 activity-limiting fall worry, controlling for T1 activity-limiting fall worry. Based on cross-sectional research discussed above, other covariates were sample members’ T2 fall status, T2 health conditions, and demographic characteristics.
Methods
Data and sample
Data for this came from waves 5 and 6 of the National Health and Aging Trends Study (NHATS), conducted in 2015 and 2016, respectively. De-identified, public-use data files were downloaded directly from the NHATS website. The study was approved by the authors’ institutional review board as exempt from human subjects review.
NHATS wave 1 was conducted in 2011 with a representative sample of U.S. Medicare beneficiaries aged 65 or older (as of September 30, 2010) residing in the community (i.e., in their own or another’s home or in residential care settings but not nursing homes) (Montaquila et al., Reference Montaquila, Freedman, Edwards and Kasper2012). The wave 1 cohort was interviewed annually. Using the same stratified three-stage sample design as in wave 1, replenishment occurred in wave 5 with a sample drawn from the Medicare enrollment database serving as the sampling frame as of September 30, 2014. At wave 5, of the total sample of 8,334 persons, 7,499 who resided in the community (429 in residential care facilities), were interviewed (n = 442 via proxy). At wave 6, 364 of these sample persons were deceased, 95 had moved to nursing homes, and 741 were not interviewed for other reasons (refusal, unavailability, inability to locate, illness, etc.), leaving a sample for the present study of 6,299 persons residing in the community at both wave 5 (T1 hereafter) and wave 6 (T2 hereafter, 355 of whom were interviewed by proxy). This study sample represents 39 million older Medicare beneficiaries, 55% of them female and 78% of them non-Hispanic White, 8% non-Hispanic Black, 7% Hispanic, and 7% of another race/ethnicity.
A significantly higher proportion of sample persons who reported activity-limiting fall worry at T1 than those who did not died by T2 (7.2% vs. 3.5%), and a significantly higher proportion of sample persons with probable major depression at T1 than those without died by T2 (7.0% vs. 2.5%). However, multivariable analysis showed that fall worry and depressive symptom scores were not associated with death or relocation to a nursing home once past-year falls, other health conditions, and demographic characteristics were included in the models. Preliminary cross-sectional and longitudinal analyses also confirmed no differences in PMD’s association with no fall worry and non-activity-limiting fall worry.
Measures
Depressive symptoms were measured with the two-item Patient Health Questionnaire-2 (PHQ-2) (Kroenke et al., Reference Kroenke, Spitzer and Williams2003), which captures cognitive/affective symptoms of anhedonia and depressed mood by asking “Over the last month, how often have you/has the sample person (a) had little interest or pleasure in doing things, and (b) felt down, depressed, or hopeless?”. Responses were based on a 4-point scale (0 = not at all; 1 = several days; 2 = more than half the days; 3 = nearly every day). The combined score was used as an overall symptom severity score, and a score >3 was used to indicate probable major depression (PMD hereafter) (Kroenke et al., Reference Kroenke, Spitzer and Williams2003).
Change (or lack thereof) in PMD between T1 and T2 was measured as (1) T1 no PMD to T2 PMD (new PMD); (2) T1 & T2 PMD (continued PMD); (3) T1 PMD to T2 no PMD (no longer PMD); and (4) T1 & T2 no PMD, used as the reference category.
Fall worry and whether the worry limited activities were measured with two questions: “In the last month, did you (or sample person) worry about falling down?” and “In the last month, did this worry ever limit your (or sample person’s) activities?” Based on responses, fall worries were categorized as activity-limiting worry (ALW), non-activity-limiting worry (NALW), and no worry (NW).
Change (or lack thereof) in ALW between T1 and T2 was measured as: (1) T1 no ALW to T2 ALW (new ALW); (2) T1 & T2 ALW (continued ALW); (3) T1 ALW to T2 no ALW (no longer ALW); and (4) T1 & T2 no ALW (i.e., T1 and T2 NW or NALW or fluctuations between NW and NALW), used as the reference category.
Falls were measured with the question, “In the past 12 months, have you (or the sample person) fallen down?” (yes or no). If needed, the following was provided: “By falling down, we mean any fall, slip, or trip in which you lose your balance and land on the floor or ground or at a lower level.” Those who answered yes were asked: “In the last 12 months, have you/has the sample person fallen down more than one time?” A negative response was classified as a single fall, and a positive response was classified as multiple (2+) falls.
T2 Health conditions included (1) the number of chronic illnesses diagnosed by a doctor, ranging from 0 to 8 (high blood pressure, heart attack/heart disease, arthritis, osteoporosis, diabetes, lung disease, stroke, and cancer); (2) whether dementia was diagnosed; (3) the number of activities and instrumental activities of daily living (ADLs/IADLs), ranging from 0 to 11 (feeding, bathing, toileting, dressing, bed transfer, moving inside the house, doing laundry, shopping, preparing meals, taking medication, and managing money), in which the sample person had a little, some, or a lot of difficulty in the past month; (4) whether the person was bothered by body pain in the past month (yes or no); and (5) the number of times the person was hospitalized in the past year.
Demographic variables were age (65–69, 70–74, 75–79, 80–84, 85–89, 90+), gender (female vs. male); race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, all other); marital status (married/partnered, divorced/separated, widowed, never-married); residence (in the community vs. residential care facility); and household income (% of national median income in 2015 [$56,516]: <50%, 50–99%, 100–199%, 200+%).
Analysis
All analyses were conducted with Stata/MP 15’s (Statacorp LLC, College Station, TX) svy function to account for NHATS’ stratified, multistage sampling design. We first described the study sample with respect to T1 and T2 prevalence of and changes in PMD, ALW, and fall status. Then, using χ2 and t tests, we compared the characteristics of those with and without T2 PMD and with and without T2 ALW. Finally, we used multivariable logistic regression models to test hypotheses: (H1) associations between T1-T2 changes in ALW and T2 PMD (dependent variable); and (H2) associations between T1-T2 changes in PMD and T2 ALW (dependent variable), controlling for T1 PMD (H1), T1 ALW (H2), T2 fall status, T2 health conditions (chronic illnesses, dementia diagnosis, ADL/IADL impairments, bothersome pain, and number of past-year hospitalization), and demographic variables. Variance inflation factor diagnostics, using a cut-off of 2.50 (Allison, Reference Allison2015), showed that multicollinearity among the covariates was not a concern. Logistic regression results are presented as adjusted odds ratios (AOR) with 95% confidence intervals (CI). Statistical significance was set at p < .05.
Results
T1 & T2 PMD, ALW, and fall status
Table 1 shows that 11.6% of the sample at T1 and 12.1% at T2 had PMD; T1-T2 changes show that 7.1% did not have PMD at T1 but did so at T2 (new PMD); 4.8% had PMD at both T1 and T2 (continued PMD); 6.7% had PMD at T1 but not at T2 (no longer PMD); and 81.4% did not have PMD at T1 or T2. Table 1 also shows that 8.8% at T1 and 9.7% at T2 reported ALW; T1-T2 changes show that 5.5% did not have ALW at T1 but did so at T2 (new ALW); 4.2% had ALW at both T1 and T2 (continued PMD); 4.6% had ALW at T1 but not at T2 (no longer ALW); and 69.3% had no ALW at T1 or T2 (i.e., either NW or NALW at both T1 and T2 or fluctuations between NW and NALW). At T1 and T2, 30.7% and 31.7%, respectively, reported falls, and a little less than half of fallers reported falling 2+ times.
Table 1. Prevalence of and changes in T1 and T2 depression, fall worry, and fall status among the study sample
Sample characteristics by T2 PMD and T2 ALW statuses
Table 2 shows that compared to those without T2 PMD, those with T2 PMD included higher proportions of those with ALW at T1 and T2 and those who had fallen 2+ times in the past year, (with about a 4 percentage point increase from the rate of T1 2+ falls). Among those without T2 PMD, there was little change in the proportion with 2+ falls. Compared to those without T2 ALW, those with T2 ALW included higher proportions of those with PMD at T1 and T2 and those who had fallen 2+ times in the past year (with a nearly 9 percentage point increase from the rate of T1 2+ falls). Additional analysis (not shown in the table) found that at both T1 and T2, the rates of 2+ falls were highest among those with both PMD and ALW. For example, at T2, the rate of 2+ falls was 47.0% among those with both PMD and ALW, compared to 37.6% among those without PMD but with ALW, 26.9% among those with PMD but without ALW, and 9.3% among those without PMD and ALW (p < .001 in all pairwise χ2 tests). Table 2 also shows that those with T2 PMD and those with T2 ALW also had more health problems and included higher proportions of older, racial/ethnic minority, and nonmarried/nonpartnered, and low-income individuals and those who lived in residential care facilities.
Table 2. Sample characteristics by T2 probable major depression (PMD) and T2 activity-limiting fall worry (ALW) statuses
1 Due to missing values in depressive symptom scores, n = 6,244 for T1 PMD and n = 6,186 for T1-T2 changes in PMD and depressive symptoms scores.
2 ADL/IADL: Activities of daily living and instrumental activities of daily living.
Probability values were calculated using Pearson’s χ2 for categorical variables and t tests for continuous variables and denote differences between two groups.
Associations between T1-T2 changes in ALW and T2 PMD: Multivariable analysis
The second column of Table 3 shows that those with new ALW at T2 had significantly greater odds of T2 PMD compared to those without ALW at either time point (AOR = 2.64, 95% CI= 1 .98−3.51), controlling for T1 PMD and T2 other health conditions. Though at slightly below statistical significance, continued ALW (AOR = 1.64, 95% CI = 0.99−2.69; p = .053) was also associated with greater odds of T2 PMD. Those who no longer had ALW at T2 did not differ significantly from those without ALW at either time point. In addition to T1 PMD, other significant covariates for T2 PMD were 2+ falls in the past year, dementia diagnosis, ADL/IADL impairments, female gender, and low income.
Table 3. Associations between T1-T2 changes in ALW and T2 PMD (Model 1) and between T1-T2 changes in PMD and T2 ALW (Model 2): Logistic regression results
1 PMD = Probable major depression.
2 ALW = Activity-limiting fall worry.
3 ADL/IADL: Activities of daily living and instrumental activities of daily living.
*p < .05; **p < .01; ***p < .001.
Associations between T1-T2 changes in PMD and T2 ALW
The third column of Table 3 shows that those with new PMD at T2 had significantly greater odds of T2 ALW (AOR=2.42, 95% CI=1.66−3.52) compared to those without PMD at either time point, controlling for ALW at T1 and T2 other health conditions. Those with continued PMD also had greater odds of T2 ALW compared to those without PMD at either time point (AOR=2.31, 95% CI = 1.62−3.29). In addition to T1 ALW, other significant covariates for T2 ALW were 2+ falls in the past year, ADL/IADL impairments, and bothersome pain. None of the demographic variables were significant factors.
Discussion
This longitudinal panel observational study of associations between depression and fall worry among a nationally representative sample of U.S. older adults adds to knowledge from cross-sectional studies of these associations. The findings show a high prevalence of depression and fall worry among the study sample, representing 39 million Medicare beneficiaries aged 65+ years. About 12% had PMD at both T1 and T2. A little more than a quarter at T1 and a little less than a third at T2 reported worry about falling, showing that the proportion of those with fall worry increased slightly between T1 and T2. Of those with any fall worry, one-third had ALW at each time point. The study also shows that while a small proportion of the sample had PMD and a small proportion had ALW at both T1 and T2, changes in PMD and ALW status were more common between the two time points.
Our multivariable analyses suggest that PMD and ALW are mutually reinforcing. Onset of ALW between T1 and T2 was significantly associated with greater odds of T2 PMD, after controlling for T1 PMD and T2 health conditions, thus partially supporting H1. Though not statistically significant, continued ALW between T1 and T2 was also associated with greater odds of T2 PMD. However, once ALW subsides, there appears to be no lingering effects on PMD. These findings indicate that increases and continuation in fall worry, ALW in particular, are significant risk factors for increases in PMD, independent of fall status, health conditions, and previous PMD. Our findings also show that onset of and continued PMD were significantly associated with greater odds of T2 ALW, after controlling for T1 ALW and other covariates, thus supporting H2. Once PMD was resolved, ALW also appears to subside. Given the use of survey methodology, causal inferences are difficult to make even with longitudinal data; however, ALW appears to strongly influence PMD and vice versa. In sum, along with previous studies showing cross-sectional association between depression and fall worry (Iaboni et al., Reference Iaboni2015), our results suggest longitudinal, reciprocal relationships between PMD and ALW.As expected, our findings show that PMD and ALW were significantly associated with multiple falls, with the highest rate of multiple falls among those with both PMD and ALW, followed by those with no PMD but with ALW and then by those with PMD but no ALW. This suggests an additive role of PMD and ALW in multiple falls, with ALW perhaps making a stronger contribution to multiple falls. Furthermore, changes (i.e., decreases and increases) in the proportions of those with PMD and ALW between T1 and T2 were correlated with respective changes in experiences of multiple falls between T1 and T2. For example, although those who no longer had ALW at T2 continued to have a substantially higher T2 fall rate compared to those without ALW at either time point, their multiple fall rate decreased from T1 to T2. Conversely, the fall rates increased from T1 to T2 for those with new or continued ALW at T2. These results point to the risk-benefit tension of fall worry. While increased fear of falling stemming from awareness of fall risk may allow individuals to take proactive steps to address modifiable fall risk factors, study results suggest that persistent or unaddressed fall worry that results in activity limitation is not protective and may exacerbate fall risk.
NHATS data did not allow for making causal inferences or for examining reasons or circumstances for changes in PMD and/or ALW between T1 and T2. In addition, although anhedonia and depressed mood are two core symptoms of late-life depression, each was measured with only a single item, and fall worry and associated activity limitations were measured with two questions rather than a validated multi-item scale (e.g., the Survey of Activities and Fear of Falling in the Elderly [SAFE/mSAFFE]; Lachman et al., Reference Lachman, Howland, Tennstedt, Jette, Assmann and Peterson1998) and may not represent the full spectrum of fear of falls. The two-item depression measure was validated (Kroenke et al., Reference Kroenke, Spitzer and Williams2003), and both measures of depression and fall worry have been used in previous NHATS-based research (Choi et al., Reference Choi, Kim, Marti and Chen2014; Gell et al., Reference Gell, Wallace, LaCroix, Mroz and Patel2015). Nevertheless, a full-length depression scale and a multi-item fall worry scale would have enabled more in-depth examination of the relationship between fall worry and depressive symptoms. Recall bias may have also affected respondents’ reports of falls in the past 12-months as older adults tend to underreport falls and fall injuries (Hoffman et al., Reference Hoffman, Ha, Alexander, Langa, Tinetti and Min2018). The study’s strengths lie in the use of nationally representative, longitudinal data that allowed for examining the relationships between changes in both depression and fall worry among a large number of older adults.
Most importantly, this study shows that depressive symptoms and fall worry can change over time, underscoring the importance of efforts to reduce both among older adults. The evidence base for psychosocial interventions that can reduce late-life depression is strong (Kiosses et al., Reference Kiosses, Leon and Areán2011). However, evidence for interventions that can reduce fall worry among community-dwelling older adults is not strong (Kendrick et al., Reference Kendrick2014) and calls for studies of new approaches. One randomized clinical trial demonstrated that cognitive-behavioral therapy-based interventions to reduce fear of falls in older adults did reduce this fear, but it had no effect on falls, injuries, social engagement, loneliness, physical function, anxiety/depression, or quality of life (Parry et al., Reference Parry2016). A meta-analysis of randomized clinical trials did find exercise interventions effective in preventing falls and reducing fall rates (Lee and Kim, Reference Lee and Kim2017); however, most older adults at risk of falling (and thus likely to have ALW) do not receive these services (Frieson et al., Reference Frieson, Tan, Ory and Smith2018). Given significant relationships among falls, fall worry, and depression, preventing falls is also imperative to reducing fall worry and depression. Better approaches are likely to integrate depression screening as an element of fall risk assessment and vice versa (Kao et al., Reference Kao, Wang, Tzeng, Liang and Lin2012), offer interventions dually aimed at reducing fall worry and depression, and ensure that more older adults receive screening and intervention. Behavioral activation approaches (Orgeta et al., Reference Orgeta, Brede and Livingston2017) intended to enhance physical/recreational activities may aid in reducing both ALW and depressive symptoms. Research is needed to test the effectiveness of such approaches in achieving both aims simultaneously. Our findings that PMD and ALW are significantly higher among low income, racial/ethnic minority, and oldest older adults also point to the need for interventions targeted to these groups.
In conclusion, this study based on a nationally representative sample of U.S. older adults allowed for examining longitudinal associations that add to knowledge about the independent and bidirectional (mutually reinforcing) relationships between depression and activity-limiting fall worry. Given the high personal and societal tolls from falls, activity-limiting fall worry, and depression in late life, more research and innovative interventions are needed to reduce these conditions at the same time.
Conflict of interest
The authors declare no conflict of interest.
Description of authors’ roles
All five authors conceptualized the study. NGC conducted statistical analysis and drafted the paper; CNM provided statistical consultation; and NMG, DMD and MEK reviewed and edited it.
