The association between specific activity components and depression in nursing home residents: the importance of the social component

Abstract

Objectives

To longitudinally explore the association between activities and depressive symptoms of nursing home (NH) residents, taking into account that each activity may contain multiple components (physical, creative, social, cognitive, and musical).

Method

Study with a baseline and two follow-ups (four and eight months). Participants were forty physically frail residents of four NHs in the Netherlands. Residents were interviewed about depressive symptoms (CES-D) and activities they conducted over the previous week. Three researchers independently rank ordered each activity on the degree to which it could be regarded as having physical, creative, social, cognitive, and musical components. Accounting for the rank score and the time the resident spent on that activity, residents were categorized per activity component into four levels: absent, low, medium, and high.

Results

Mixed models predicting depressive symptoms from individual activity components showed significant associations for the social and cognitive components. Compared with the lowest activity level, the analyses showed fewer depressive symptoms for all higher levels of the social and cognitive components. However, a mixed model adjusted for all activity components showed no unique effect of the cognitive component or other components, while the effects of the social component remained significant. The analyses did not show differences between the time points.

Conclusion

The results suggest that the effects of activities on depressive symptoms might be mainly explained by their social component. It is, thus, important to always stimulate social involvement and interaction when developing and applying depression interventions. However, intervention research is needed to confirm these findings.

Keywords:

• Activities
• activity components
• depression
• nursing home residents

Introduction

Depression is common among residents of nursing homes (NHs) (Djernes, 2006; Jongenelis et al., 2004; Mozley et al., 2000; Seitz, Purandare, & Conn, 2010; Teresi, Abrams, Holmes, Ramirez, & Eimicke, 2001) and contributes to a significant proportion of the disease burden and health care costs (Alexopoulos et al., 2002; Blazer, 2003; Katon, Lin, Russo, & Unutzer, 2003; Unutzer et al., 2000). Understanding protective factors for depression in NH residents is necessary to develop tailored prevention and treatment strategies.

Many studies indicate that in older adults, participation in physical (e.g. physical exercise, walking), creative (e.g. arranging flowers, painting), social (e.g. visiting family or friends, group activities), and musical (e.g. listening to music, singing) activities contributes to improved quality of life, including decreased depressive symptoms (Cherry et al., 2013; Cohen et al., 2006; Greaves & Farbus, 2006; Hassmén, Koivula, & Uutela, 2000; Holtfreter, Reisig, & Turanovic, 2017; Hong, Hasche, & Bowland, 2009; Knapen, Vancampfort, Morien, & Marchal, 2015; Ruuskanen & Ruoppila, 1995; Sarkamo et al., 2014; Seinfeld, Figueroa, Ortiz-Gil, & Sanchez-Vives, 2013; Tavares, Moraes, Deslandes, & Laks, 2014; Vogel et al., 2009; Wassink-Vossen et al., 2014; Willemse, Depla, & Bohlmeijer, 2009; Yuen, Mueller, Mayor, & Azuero, 2011). Given current insights that a better state of well-being can be achieved if one is able to adapt to a challenging disease by making a dynamic balance between opportunities and limitations (de Vugt & Dröes, 2017; Dröes et al., 2017; Huber et al., 2011), participation in such activities may contribute to a person’s social health as it may support their adaptation to age-related challenges such as physical illness and loss of close social contacts. Activities focused on cognitive stimulation such as reasoning or remembering (e.g. puzzles, memory groups; hereafter referred to as cognitive activities) might also be effective, although literature regarding their effects on different health outcomes such as mood and quality of life remains limited (Williams & Kemper, 2010).

For NH residents, the effects of participation in activities on mental health outcomes are less well documented than for a general population of older adults. Due to frequently occurring chronic physical illness, cognitive decline, and loss of close social contacts, elderly people living in institutions seem particularly prone to having major depression and depressive symptoms (Llewellyn-Jones & Snowdon, 2007). Accordingly, prevalence rates of major depression and depressive symptoms are known to be higher within care facilities than in community-dwelling samples (Djernes, 2006; Jongenelis et al., 2004), which underlines the importance of research on potential protective factors in this specific population.

To date, the research on activities in older adults has tended to focus mainly on an activity as such without taking into account that an activity may address different aspects of individuals’ experience or their life domains (Cherry et al., 2013; Cohen et al., 2006; Greaves & Farbus, 2006; Hassmén et al., 2000; Holtfreter et al., 2017; Hong et al., 2009; Knapen et al., 2015; Ruuskanen & Ruoppila, 1995; Sarkamo et al., 2014; Seinfeld et al., 2013; Tavares et al., 2014; Vogel et al., 2009; Wassink-Vossen et al., 2014; Willemse et al., 2009; Yuen et al., 2011). For example, physical exercise in groups can be regarded as an activity that has at least two components, namely one that affects physical experience, and a social component that regards verbal or non-verbal communication with others. A better understanding of the effects of specific activity components on depressive symptoms is needed to develop tailored prevention and treatment interventions. Furthermore, due to the limited longitudinal and experimental research in this field, conclusions on a potential causal relationship between participation in activities and depressive symptoms are difficult to draw. To develop person-centered interventions, however, this knowledge is a prerequisite.

The aim of this study was, therefore, to contribute to the knowledge about the effects of different components of activities on depressive symptoms in NH residents. This study longitudinally explored the association between the degree to which an activity can be regarded as having physical, creative, social, cognitive, and musical components on the one hand, and the extent of depressive symptoms on the other hand, among NH residents.

Method

Study design

A longitudinal observational study with a baseline measurement (t₀) and two follow-ups, at four months (t₁), and eight months (t₂) was conducted between January and December 2017.

Study population

The study population included adults aged 60 years and older living in Dutch NHs. Residents from NHs with ongoing or planned relocations, changes in care methods, or participation in other studies on similar topics were not invited to participate in the study. Only residents from NH units that provide medical-somatic care could participate, meaning that residents from units with predominantly psychogeriatric, palliative, and rehabilitation care were not included. Residents from these somatic care units (further somatic units) who, as indicated by the treating elderly care physician (Koopmans, Lavrijsen, & Hoek, 2013), were severely cognitively impaired, had a mental disorder, or a life expectancy of less than six months were also excluded. Based on an effect size of 0.25, alpha of 0.05, and power of 0.80, a total sample size of 28 NH residents was needed (Erdfelder, Faul, & Buchner, 1996).

Ethics statement

The study has been reviewed by the research ethics committee (cETO) of the Open University (reference number: U2016/06589/FRO). The cETO judged the ethical aspects positively. The study was conducted in accordance with the Declaration of Helsinki (World Medical Association, 2013) as well as the rules applicable in the Netherlands.

Procedure

NHs were recruited using convenience sampling through the University Network of Care Homes Nijmegen (UKON; a Dutch infrastructure for academic long-term care) (Koopmans et al., 2013; Leontjevas et al., 2013) as well as through networks of the researchers. After the NH management provided institutional informed consent, the treating elderly care physicians were asked to indicate which residents were eligible to participate in the study. Eligible residents received an information letter and were visited by one of the researchers who provided additional information verbally about the study’s purpose and confidentiality issues. Written informed consent was obtained from all residents who were willing to participate in the study.

To collect all data, individual, structured face-to-face interviews were conducted with the residents at three time points (t₀, t₁, t₂). Residents were asked to answer questions about socio-demographic factors (at t₀ only), participation in activities, and depressive symptoms. The elderly care physicians provided information about use of antidepressant medication during the study.

Outcomes

Socio-demographic factors

A questionnaire assessing gender, age (years), marital status, and educational attainment was used. For educational attainment, educational levels were aggregated into three categories according to the International Standard Classification of Education (ISCED): low (levels 0 [early childhood education] – 2 [lower secondary education]), medium (levels 3 [upper secondary education] – 4 [post-secondary non-tertiary education]), and high (levels 5 [short-cycle tertiary education] – 8 [doctoral or equivalent level]) (UNESCO, 2012).

Depressive symptoms

A Dutch translation of the 20-item Center for Epidemiologic Studies Depression Scale (CES-D) was used to assess depressive symptoms. The CES-D is a self-reported questionnaire showing high internal consistency, acceptable test–retest reliability, and excellent concurrent validity with respect to clinical and self-report criteria (Radloff, 1977). The total score ranges from zero to 60 (with 0 [rarely or never] to 3 points [most or all of the time] for each item). A higher total score indicates more depressive symptoms. A cut-off score of 16 has typically been recommended to indicate depression (Weissman, Sholomskas, Pottenger, Prusoff, & Locke, 1977). For each measurement point in this study, the Cronbach’s alpha of the CES-D was regarded as good (α = .92).

Activities

Residents were interviewed about their participation in activities over the previous week. For all activities, residents were asked to point out the group size, the number of days during the previous week that they participated, and the average time spent on the activity per day. Three researchers independently rank ordered each activity reported by the residents on the degree of physical (i.e. physical exertion), creative (i.e. creating something new), social (i.e. social interaction), cognitive (i.e. multiple mental abilities involving memory, attention, problem-solving, orientation, planning, and decision-making), and musical (i.e. musical involvement) components using rank scores absent or very low, low, medium, high, and very high. After combining the component rank scores of the three researchers, the scores were inspected by clustering the activities (like reading books, reading papers, etc.) and by sorting activities per rank score. Discrepancies regarding equivalence (equivalent activities, e.g. reading a paper and reading a book, should have equivalent scores on specific components) and distinctiveness (e.g. a more physically intensive activity should have a higher score on the physical component than a less intensive activity) were solved based on consensus of three researchers by using intermediate rank scores. These intermediate rank scores indicate ordering within a cluster of comparable activities or when one or two researchers ordered the component to a higher rank score (e.g. one researcher scores Low and two researchers score Medium which resulted in a Low++ rank score, see Table 1 for example).

Table 1. Examples of activities with their corresponding component rank scores (levels: A = absent or very low, L = low, M = medium, H = high).

Activities	Physical	Creative	Social	Cognitive	Musical
Reading in their own room (book, magazine, newspaper)	A	A	A	M	A
Discussing news or sports in a group	A	A	H+	M++	A
Exercise alone in their own room by a workout program on television	H++	A	A	L+	A+
Walking outside in a group (with or without using a rollator)	M++	A	M	A+	A
Handcrafts alone in their own room (knitting, embroidering)	A++	M+	A	L+	A
Painting alone in their own room	A+	H++	A	M	A
Painting in a group	A+	H++	L++	M	A
Memory exercises in a group	A	A+	H	H++	A
Memory exercises alone in their own room by a program on television	A	A+	A	H++	A
Listening to music alone in their own room	A	A	A	A+	M+
Playing the keyboard alone in their own room	A++	M+	A	M++	H++

Note. ‘+’ and ‘++’ reflect intermediate rank scores that were added when the activity was ranked to a higher rank by one or two researchers, or when discrepancies regarding equivalence or distinctiveness were solved by re-ordering the activities on the component.

For each resident, five activity component scores were calculated. For this, first, rank scores were rescaled on the range from 0 = absent/very low to 4 = very high, with 1/3 and 2/3 for the intermediate scores. These scores were multiplied by the number of days during the previous week that the resident participated in the activity and the average time spent on the activity per day. Subsequently, for each resident, five total component scores were calculated and categorized using cut-off scores determined at baseline into four levels: absent (scores of 0), low (first 33.3% residents with scores higher than 0), medium (second 33.3% with scores higher than 0), and high (last 33.3% of residents with scores higher than 0).

Use of antidepressant medication

The use of antidepressant medication was scored dichotomously (0: no antidepressant use, 1: use of one or more antidepressants).

Statistical analysis

Statistical analyses were conducted using IBM SPSS version 22.0 (IBM Corporation, 2013). To prevent type 2 errors, missing values concerning antidepressant use (10 out of 104 records) were imputed to zero (no antidepressant use). The depressive symptom scores were centered and standardized using the SD and the mean score at baseline.

We compared mixed models with fixed effects for time points (t₁ and t₂ compared with t₀) with models without time points using likelihood ratio tests. The model with the best fit was used for estimating the effects. A p-value of .05 was used. However, it is argued that statistical significance examines whether the findings are likely to be due to chance, whereas effect size is important to understand the magnitude of differences found (Sullivan & Feinn, 2012). Therefore, we also reflected on the size of estimated effects, representing the number of standard deviations of depressive symptoms that a component level differed from the lowest component level (i.e. absent for all components but cognitive component with the reference low level). Given that the depressive symptoms were standardized, we used Cohen’s d criteria to describe the effect size: the effect of 0.2 and higher was regarded as small, 0.5 and higher as medium, and 0.8 and higher as large (Cohen, 1988). An effect of 0.2 and larger was regarded as clinically relevant (Kazdin & Bass, 1989).

Mixed models accounting for repeated measurements with unstructured covariance matrix and corrected for antidepressant use were built for activity components (lowest component activity level as reference category) predicting depressive symptoms. A fixed and a random intercept were specified. Several models were built for activity components: one model for each component without adjustments for other activity components, and one model with all components corrected for each other.

Results

Residents from 12 somatic units in four NHs located in two Dutch provinces were recruited for the study. In these NHs, 93 residents (of whom 57 female [61.3%]) met the inclusion criteria and were invited to participate in the study. Of these, 40 (43%) provided informed consent (of whom 26 female [65%]; mean age 79.8 years [SD, 8.9]). All 40 participants completed t₀, 33 participants (82.5%) completed t₁, and 31 participants (77.5%) completed all three measurements. Reasons for those who did not complete all three measurements were death (N, 4), withdrawal (N, 3), or relocation (N, 2).

Table 2 presents socio-demographical data and relevant outcomes at t₀, t₁, and t₂. Most participants were widowed (N, 26 [65%]), had a low level of educational attainment (N, 25 [62.5%]), and did not use antidepressant at baseline (N, 25 [62.5%]). In total, 20 participants (50%) had a depression score of ≥16 at baseline (range from 1 to 46). No significant change over time in depressive symptoms and antidepressants use was found (p > .05). Table 3 shows the number of NH residents categorized per activity component in four levels.

Table 2. Socio-demographic factors, depressive symptoms, and antidepressant use.

	t0 (N, 40)	t1 (N, 33)	t2 (N, 31)
Age, mean (SD) [range]	79.8 (8.9) [60–96]	80.1 (9.4) [60–96]	79.5 (9.4) [60–96]
Sex, female, N (%)/male, N	26 (65)/14	23 (69.7)/10	21 (67.7)/10
Marital status
Unmarried, N (%)	2 (5)	2 (6.1)	2 (6.5)
Married or partnered in a registered partnership, N (%)	8 (20)	6 (18.2)	6 (19.4)
Widowed (after marriage or registered partnership), N (%)	26 (65)	21 (63.6)	19 (61.3)
Divorced (after marriage or registered partnership), N (%)	4 (10)	4 (12.1)	4 (12.9)
Educational attainment
Low, N (%)	25 (62.5)	20 (60.6)	19 (61.3)
Medium, N (%)	14 (35)	12 (36.4)	11 (35.5)
High, N (%)	1 (2.5)	1 (3)	1 (3.2)
Depressive symptoms, mean (SD) [range]	19.6 (13.4) [1–46]	18.9 (13.3) [1–50]	19.6 (13.1) [2–51]
Antidepressant use, yes, N (%)/no, N^a	15 (37.5)/25	12 (36.4)/21	10 (32.3)/21

Note. ^aMissing values (N_t0 = 5, N_t1 = 3, N_t2 = 2) were replaced by antidepressant use is ‘no’.

Table 3. Number of nursing home residents per activity component level, N (%).

	t0 (N, 40)	t1 (N, 33)	t2 (N, 31)
Physical
Absent	16 (40)	11 (33.3)	11 (35.5)
Low	8 (20)	7 (21.2)	6 (19.4)
Medium	8 (20)	5 (15.2)	9 (29.0)
High	8 (20)	10 (30.3)	5 (16.1)
Creative
Absent	18 (45)	16 (48.5)	13 (41.9)
Low	7 (17.5)	8 (24.2)	9 (29)
Medium	7 (17.5)	4 (12.1)	5 (16.1)
High	8 (20)	5 (15.2)	4 (12.9)
Social
Absent	4 (10)	3 (9.1)	6 (19.4)
Low	12 (30)	7 (21.2)	6 (19.4)
Medium	12 (30)	7 (21.2)	7 (22.6)
High	12 (30)	16 (48.5)	12 (38.7)
Cognitive
Absent	0 (0)	0 (0)	0 (0)
Low	13 (32.5)	14 (42.4)	16 (51.6)
Medium	13 (32.5)	6 (18.2)	6 (19.4)
High	14 (35)	13 (39.4)	9 (29)
Musical
Absent	23 (57.5)	18 (54.5)	17 (54.8)
Low	5 (12.5)	8 (24.2)	6 (19.4)
Medium	6 (15)	1 (3)	4 (12.9)
High	6 (15)	6 (18.2)	4 (12.9)

Note. For each resident, five total component scores were calculated and categorized using cut-off scores determined at baseline into levels: absent (scores of 0), low (first 33.3% of scores higher than 0), medium (second 33.3% of scores higher than 0), and high (last 33.3% of scores higher than 0).

Comparison of models with fixed effects for time points (t₁ and t₂ compared with t₀) with models without time points revealed that the latter was the better fitting model. Therefore, the models for the effects of activities did not include the three time points.

Table 4 presents the effects of activities predicting depressive symptoms. The social (F(3, 57.73) = 7.92, p = .000) and cognitive components (F(2, 62.21) = 4.23, p = .019) were associated with depressive symptoms in individual models corrected only for antidepressant use but not for other activity components. Compared with the lowest component activity level, the analyses showed fewer depressive symptoms for all higher levels of the social and cognitive components. The physical (F(3, 53.61) = 2.73, p = .053) and creative components (F(3, 52.44) = 2.52, p = .068) appeared to be marginally significant in individual models showing clinically relevant effect sizes.

Table 4. Mixed models of activity components predicting depressive symptoms.

Parameter	Physical component	Creative component	Social component	Cognitive component	Musical component	All activity components
AD, yes^a	0.02 (0.22)	−0.06 (0.21)	0.05 (0.19)	−0.15 (0.21)	−0.10 (0.22)	0.07 (0.21)
Activity components
Physical^b	F(3, 53.61) = 2.73, p = .053					F(3, 54.55) = 0.29, p = .831
Low	−0.23 (0.13)					0.13 (0.19)
Medium	−0.23 (0.17)					0.15 (0.21)
High	−0.47 (0.17)					−0.00 (0.23)
Creative^b		F(3, 52.44) = 2.52, p = .068				F(3, 51.11) = 1.16, p = .336
Low		−0.19 (0.14)				−0.25 (0.18)
Medium		−0.44 (0.17)				−0.42 (0.23)
High		−0.35 (0.21)				−0.17 (0.26)
Social^b			F(3, 57.73) = 7.92, p < .001			F(3, 54.96) = 4.73, p = .005
Low			−0.39 (0.20)			−0.36 (0.25)
Medium			−0.82 (0.21)			−0.78 (0.27)
High			−0.82 (0.20)			−0.75 (0.27)
Cognitive^c				F(2, 62.21) = 4.23, p = .019		F(2, 65.52) = 1.02, p = .368
Medium				−0.37 (0.15)		−0.21 (0.16)
High				−0.40 (0.15)		−0.23 (0.19)
Musical^b					F(3, 52.72) = 0.94, p = .426	F(3, 58.09) = 0.17, p = .917
Low					−0.16 (0.14)	−0.07 (0.16)
Medium					−0.12 (0.20)	−0.07 (0.20)
High					−0.32 (0.21)	−0.15 (0.23)

Note. Six models were built for activity components: [five individual models] for each component without adjustments for other activity components, and one model with all components corrected for each other. All models are adjusted for antidepressant use. The depressive symptom scores were centered and standardized using the standard deviation and the mean score at baseline. Standard errors are in parentheses. AD = Antidepressant use. p < .05 are in boldface.

^a AD was dummy coded, with ‘no’ as reference category.

^bPhysical, creative, social, and musical activity components were ordinal variables with 4 categories (absent, low, medium, and high), with ‘absent’ as reference category.

^cThe cognitive activity component was an ordinal variable with 3 categories (low, medium, and high), with ‘low’ as reference category.

In the combined model with all components as predictors, a significant association of depressive symptoms with the social component (F(3, 54.96) = 4.73, p = .005) was found, but not with other activity components.

In terms of effect sizes, the medium effect sizes with clinical relevance in individual models for physical, and musical components dropped below the threshold of clinical relevance in the combined model with all components. Although not significant, almost all reduced effect sizes for creative and cognitive components can still be considered as clinically relevant in the combined model. The effect sizes for the social component can be considered as clinically relevant ranging from medium to large effects (e.g. estimated effect for the highest level compared to absent, −0.75 [95% confidence interval, −1.29 to −0.21], p = .007, model corrected for other activity components).

Discussion

A longitudinal observational study of NH residents was conducted to explore the association between depressive symptoms and five components of undertaken activities: physical, creative, social, cognitive, and musical. The analyses showed that participating in activities with social components is associated with fewer depressive symptoms, regardless of whether or not other components were present. To note, the effect sizes ranged from ‘medium’ to ‘high’ and can be interpreted as clinically relevant (Kazdin & Bass, 1989). The findings regarding this association are consistent with previous studies among older people (Greaves & Farbus, 2006; Holtfreter et al., 2017; Hong et al., 2009). From a theoretical perspective, there might be several reasons for the effect of the social component of activities. For instance, social identity theory proposes that a person’s sense of who they are depends on the groups to who they belong (Tajfel, 1979). It is argued that social identification is the mechanism through which social relationships affect depression (Cruwys, Haslam, Dingle, Haslam, & Jetten, 2014). In our study, it is possible that participants who join group sessions, identify themselves with this group, and subsequently, influence one’s sense of belonging which is needed for health maintenance and well-being. Furthermore, self-determination theory (SDT), a theory of human motivation and personality that addresses three universal, innate and psychological needs (competence, autonomy, and psychological relatedness) might explain the mechanism through which the social component of activities affects depression (Ryan & Deci, 2000). SDT has been identified as a way to understand motivations and behavior change in older adults, particularly as it relates to aging and leisure (Altintas, Guerrien, Vivicorsi, Clement, & Vallerand, 2018; Lin & Yen, 2018). Regarding our study, participating in activities with social components may contribute to fulfillment of psychological relatedness (and possibly other needs as well), and subsequently, a decrease in depressive symptoms. Also, in accordance with the concept of social health (de Vugt & Dröes, 2017; Dröes et al., 2017; Huber et al., 2011), participation in such activities may entail adapting to age-related challenges.

Furthermore, the analyses revealed that the cognitive component of activities is associated with fewer depressive symptoms. However, the association disappeared when we corrected for the presence of other components. The effect sizes of both the individual model and the model corrected for other activity components can be classified as ‘medium’. This suggests that the effects are in line with other studies that find effects of cognitive activities and do not account for other activity components including social interaction or involvement. For example, ‘memory exercises in a group’ may be effective because of the social interaction involved, not due to the cognitive aspects.

Although clinically relevant effect sizes were found for most levels of the physical and creative components in models without corrections for other components, nearly all effect sizes decreased to below the threshold of 0.2 for clinical relevance when the model was adjusted for other components. Again, this may suggest that the effects of physical and creative components can be – at least partially – explained by other components. For example, an ‘exercise intervention’ might not or not solely be effective due to body movements as such but due – for a substantial part – to social interaction with other people during the exercises.

The analyses did not show statistically significant effects for the musical components of activities. Furthermore, the analyses suggested that if clinically relevant effects for such activities were statistically significant (e.g. in a study with a larger sample), these effects might also be explained by other components in these activities (e.g. social). It is possible that effects for musical activities in other studies may – at least partially – be attributed to a social component. For example, a musical intervention with patients making music within a group may be effective because of the social interaction involved, not the music itself. However, intervention research is needed to confirm these findings. Exploring and comparing the (combined) effects of specific components in experimental research that varies the components within an activity (such as more social versus less social involvement during a physical activity) might thus be advised.

Analyses implied a lack of temporary changes in depressive symptoms (i.e. no differences were found between the baseline and the two follow-ups). It is unclear whether there may be temporary changes when there is a longer follow-up period. It is possible that a low level of depressive symptoms makes residents more inclined to participate in activities, or that a low level of depressive symptoms is being maintained by participating in activities. Future research is needed to explore the causalities.

This study assessed the effectiveness of activities rather than efficacy, as it focused on activities performed by residents on their own initiative, while participants in experimental studies are randomized between an intervention condition and a control condition. For example, in the study by Sarkamo et al. (2014), participants were randomized into a singing group, a music listening group, and a usual care group. Different findings between the current observational study and randomized controlled trials (RCT) might, therefore, be explained by the differences between those who already participate in activities and those who do not participate in activities but are stimulated to participate in a RCT. For example, some residents may be physically inactive, and a new intervention can be effective while self-chosen physical activities (as in our observational study) can be ineffective. Furthermore, different findings between this study and other studies among older adults might be explained by the intensity of the activities. The intensity of activities among NH residents may be lower than those among community-living elderly people due to physical conditions or because of the intensity or number of activities offered within the NHs.

Strengths and limitations

The present study has gone some way towards enhancing our understanding of the association of specific components of activities with depressive symptoms in NH residents since we were able to take into account that a specific activity can contain multiple components (i.e. physical, creative, social, cognitive, and musical). Another strength of the current study is that residents were followed for eight months, giving the study more power.

In order to limit missing data in self-reports, we conducted face-to-face interviews, which is common practice in research among NH residents (Leontjevas et al., 2013). To reduce social desirability bias, we used structured interviews and a validated measurement instrument with forced-choice items for depressive symptoms (Nederhof, 1985). We used open-ended questions to measure participation in activities because we considered it important to allow residents to mention all of the activities that were meaningful to them.

Another strength of this study is that, by using face-to-face interviews, we were able to account for observed differences in personality (e.g. extraversion vs. introversion). For example, some residents indicated that they had not participated in activities during the past week. Nevertheless, after prompting them with examples, they reported several activities. It was also noted that, while some participants were eager to communicate all activities in which they had participated, others initially seemed reluctant to share this information. Questions about participation in activities were repeatedly asked until the participant indicated that he or she did not participate in further activities. By applying this method with all participants, we minimized the possible effect of differences in openness.

This study had several limitations. First, the sample size was not large enough to correct for the hierarchical structure of the data regarding NHs, since a small sample size at NH level (≤50) may lead to biased estimates of the second-level standard errors (Maas & Hox, 2005). A recommendation for future research would, therefore, be to recruit more residents and more NHs to account for the clustered structure of the data.

Another limitation of this study lies in the determination of the five component scores for each activity reported by the residents. Since no validated measuring instrument was available, determining these scores was done in a pragmatic manner by three researchers. However, since the researchers were blind to each other’s rank scores, and reached consensus for discrepancies regarding equivalence and distinctiveness, we were able to reduce the degree of subjectivity. Nevertheless, it should be noted that the component scores were not based on participant observation, neither on self-reports of the residents. Although some activities were performed in a group, we could not determine whether participants were actually engaged in social interactions. It is possible that self-reports could result in more pronounced effects because reports of others show a tendency to attenuate self-reports (Leontjevas, Teerenstra, Smalbrugge, Koopmans, & Gerritsen, 2016). Therefore, self-reported scores or observations need to be explored in future research. Furthermore, in our study, we could not examine which specific aspects of social components contribute to fewer depressive symptoms. For example, older people with dementia living in care homes state that they particularly found meaning in activities that address psychological and social needs, which related to the quality of the experience of an activity rather than specific types of activities (Harmer & Orrell, 2008). Therefore, future research could explore specific aspects of social components including meaningfulness and sense of belonging related to depression.

In addition, since it may be expected that residents with a more active lifestyle were more attracted to the study and consequently more inclined to participate in the study, residents with a more active lifestyle may have been overrepresented in the sample. Also, as assessed by the CES-D (cut-off score of 16) half of the participants in our study probably suffered from depression. Djernes (2006) notes that the prevalence of major depression ranges from 14% to 42% among elderly people living in NHs. Although CES-D is not a diagnostic instrument, it is possible that our sample had slightly more residents with depression than could be expected. Furthermore, since we used a convenience sample from Dutch NHs, it is important to be cautious when generalizing the results of this study.

Some residents indicated spontaneously that there were too few opportunities to participate in activities within their NH unit. Because no information was collected about determinants of participation, we could not test specific barriers to nor reveal facilitators of participating in activities. Some barriers and facilitators for engagement in activities may also have influenced our results and could be included as potential confounders in future research.

Conclusions and implications

The results of this study suggest that the effects of activities on depressive symptoms might be mainly explained by their social component. For the purpose of depression prevention and treatment, it can thus be advisable to develop and apply interventions that specifically include social interaction and contribute to improvements of a person’s social health. For example, group sessions aimed at flower arranging or painting could be preferred over individual sessions. It is important to realize that each activity residents can be engaged in has multiple aspects. The current study contributes to better insight into the effects of activity-based interventions by showing different effects of specific activity components. Intervention research is needed to understand whether effects of activity-based interventions in NH residents can be explained by their social aspects, and how the effects of specific activities can be enhanced for the purpose of depression prevention and treatment.

Acknowledgment

The authors wish to thank all nursing home residents and personnel for participation.

Disclosure statement

The authors report no conflict of interest.

Funding

This research did not receive any funding from agencies in the public, commercial, or not-for-profit sectors.