Patient Anxiety and Depression Moderate the Effects of Increased Self-management Knowledge on Physical Activity: A Secondary Analysis of a Randomised Controlled Trial on Health-Mentoring in COPD

Abstract

Objective. Anxiety and depression are common comorbidities in people with chronic obstructive pulmonary disease (COPD). While these comorbidities could potentially lead to a higher motivation to learn about self-management, they could also inhibit patients from translating this knowledge into appropriate self-management behaviours. This paper explores the moderating effects of anxiety and depression on a health-mentoring intervention, focusing on mechanisms of change (mediation). Methods. 182 COPD patients participated in an RCT, with anxiety and depression assessed by the Hospital Anxiety and Depression Scale (HADS), self-management knowledge by the Partners in Health Scale, and spontaneous physical activity using accelerometers, all measured at baseline, 6 and 12 months. The moderated mediation model tested the intervention's effect on physical activity, mediated via changes in self-management knowledge, at different levels of anxiety and depression. Results. Knowledge mediated the effect of the intervention on changes in physical activity only for participants reporting low levels of anxiety or depression. Both acted as moderators: Increased knowledge led to more physical activity among participants reporting low anxiety or depression and to less activity among highly anxious or depressed participants. Conclusion. Although health-mentoring interventions can be an effective tool to increase knowledge and physical activity among COPD patients, it is essential to take anxiety and depression into account, as increased knowledge may have detrimental effects in highly anxious or depressed participants. This suggests that patients with elevated anxiety or depression may need to be treated appropriately before engaging in chronic disease self-management interventions.

Keywords:

• Self-management
• anxiety
• depression
• physical activity
• self-management knowledge
• health mentoring
• moderated mediation

Introduction

Chronic obstructive pulmonary disease (COPD) is a leading cause of ­morbidity and mortality worldwide (1). COPD in developed countries is mainly due to cigarette smoking, and is characterised by irreversible airflow limitation (ratio of forced expiratory volume in one second to forced vital capacity (FEV1/FVC) less than 0.70 post bronchodilator) that causes symptoms including breathlessness (dyspnoea). As COPD is a chronic condition leading to progressive impairment, self-management of the disease is essential to secure well-being among patients and to reduce the burden of disease (2). For example, there is evidence of a strong association between undertaking regular physical ­activity, a self-management behaviour, and a reduced risk for hospital readmission in COPD patients (3). In order to optimally manage COPD, it is important for patients to gather knowledge about self-managing their condition, and to learn how to make informed decisions concerning their care. Examples of this include adopting behaviours to improve their health, such as physical activity, immunization, smoking cessation and adhering to treatment regimens (4).

A systematic review on self-management interventions in COPD revealed only modest effects (5): While participation in self-management programmes was significantly related to fewer hospital admissions, there were no consistent effects on any other outcomes. Moreover, self-management interventions can have adverse effects: A recent trial had to be stopped prematurely because treatment was associated with unexpectedly high mortality rates (6). While the lack of reliable effects may partly be due to the heterogeneity of the interventions (5), an alternative explanation may be that the programmes can only be effective for certain subpopulations of people with COPD. What is more, many trials fail to test intervention effectiveness in terms of behavioural changes and changes in mediating factors such as self-efficacy, attitudes, or other cognitive factors such as self-management knowledge (7).

Anxiety and depression are common comorbidities of COPD that can have a detrimental impact on patients’ self-management capacity and health outcomes, such as exercise performance and frequency of exacerbations (8-10). For patients with severe COPD, anxiety prevalence rates of up to 50–75% have been reported, which is much higher than in the general population, or indeed in other chronic conditions such as AIDS or cardiac disease (11). In patient populations with mild to moderate COPD, prevalence rates of up to 55% have been observed for anxiety, and depression prevalence rates of around 40% are not uncommon (12, 13). Factors that have been associated with anxiety and depression in COPD include long-term oxygen therapy, severe dyspnoea, more severe airflow obstruction, living alone, current smoking, and female gender (14). Patients with COPD who experience anxiety report higher levels of symptom perception than patients without clinical anxiety, which is not necessarily related to worse lung function (12). Panic attacks frequently accompany changes in breathing patterns, which in turn worsens dyspnoea. This phenomenon has been described as the dyspnoea-anxiety-dyspnoea cycle (15), and is likely to interfere with self-management behaviours, particularly with adherence to regular physical activity, which itself is accompanied by dyspnoea and faster breathing (9, 16). Thereby, anxiety is a potential moderator of intervention effectiveness: Self-management programmes may be more effective in the subgroup of patients with low levels of anxiety, than in patients reporting high ­clinical anxiety levels.

There are several potential pathways for anxiety to influence self-management of chronic conditions. While anxiety is often associated with increased motivation to obtain information about the chronic condition (17), and thus a greater interest in accumulating self-management knowledge (16), it is also related to selective biasing of threat information, symptom misinterpretation, and failure to engage in protective strategies (17, 18). This can lead to problems with the execution of self-management strategies, despite knowledge uptake about strategies being better than among non-anxious patients. While there is some evidence to support these theoretical assumptions (16), to our knowledge these pathways have never been tested experimentally or in a concurrent statistical model, and not in COPD.

Aims and Hypotheses

This study aimed to explore the influence of anxiety and depression on the effects of a trial intervention aimed at increasing self-management skills and subsequent self-management behaviour for increased physical activity among COPD patients. More specifically, we hypothesised that anxious participants benefit more from the intervention with regard to increases in self-management knowledge than participants with lower levels of anxiety, but that increased self-management knowledge would lead to poorer adoption of a physical activity regimen among highly anxious or depressed participants, and to higher activity levels among non-anxious participants and participants with low levels of depression. Thus, it was hypothesised that both anxiety and depression act as moderators of the intervention's effect on changes in physical activity that is mediated by self-management knowledge.

Methods

Participants and Procedure

The patient sample for this cluster randomized controlled trial was recruited in 31 general practices in rural, remote and metropolitan areas of Tasmania. The procedures, intervention contents and complete sample characteristics are provided in detail elsewhere (19). Briefly, patients aged over 45 years with a current diagnosis of COPD or being treated with tiotropium for COPD were eligible to participate. Inclusion criteria were: smoking history >10 pack-years, post bronchodilator FEV1/FVC ratio <0.7, and FEV1 30-80%. Exclusion criteria were: lack of English literacy, mental or physical incapacity, end-stage cancer, or residency in a nursing home. Participating practices were randomised using a code generated from a random numbers table stratified in blocks of four by the Rural, Remote and Metropolitan Areas (RRMA) classifications appropriate to Tasmania. Concealment of ­allocation was provided by sequentially numbered opaque, sealed envelopes.

92 patients were enrolled to the usual-care control group (UC), 90 patients to the health-mentoring intervention group. Health-mentoring consisted of up to 16 phone calls by trained community nurses over 12 months (19). The calls addressed five core components to support self-management: 1) psychoeducation; 2) self-management skills training including goal-setting, action planning and problem-solving skills; 3) cognitive coping skills training; 4) communication skills training to facilitate discussions with health-mentor or professionals; and 5) promoting self-efficacy to manage chronic conditions. During calls patients were guided in setting medium to long-term goals targeting physical activity uptake, smoking cessation, nutrition, alcohol consumption, psychosocial well-being, and symptom management (20), and to specify appropriate matching action plans to facilitate goal achievement. The control condition consisted of participants receiving usual care as provided by their GPs plus monthly “social” phone calls without specific advice or skills training.

The study was registered with the Australian and New Zealand Clinical Trials Research network (ACTR 12608000112369) and approved by the Human Research Ethics Committee of the University of Tasmania (H0009777). The main effects of the health mentoring intervention have been reported elsewhere (19).

Measures

All outcomes were measured at baseline (T1), 6 months (T2), and 12 (T3) months after baseline. Participants were able to choose whether to complete paper-and-pencil questionnaires at our study centers or at their GPs’ practices.

Physical Activity

Physical activity was measured via accelerometer (ActiGraph GT1M), which was worn during waking hours over one week to determine the mean daily step count. The threshold for a valid day of activity measurement was set at 10 hours of wearing time with days below that threshold excluded during data cleaning. Daily steps were assessed in participants with at least three valid days. At least five valid days were available in 81% of assessments (21).

Self-management Knowledge

Knowledge was measured on a 9-point Likert scale using the four items of the Partners in Health (PIH) knowledge subscale, which assesses knowledge of what to do when symptoms get worse, sharing in decisions, knowledge of illness and treatment (22).

Anxiety

General anxiety was assessed using the seven-item anxiety subscale (range 0–21) of the Hospital Anxiety and Depression Scale (HADS; 24). A cut-off point of 8 out of 21 is usually used to define caseness, with higher scores indicating higher anxiety levels (24).

Depression

Depression was assessed using the seven-item depression subscale (range 0–21) of the HADS (23). A cut-off point of 8 out of 21 is usually used to define caseness, with higher scores indicating higher depression levels (24).

Analyses

Moderated mediation analyses were performed in SPSS 19 using the PROCESS macro (Model 58; 26). In the current moderated mediation model, differences in physical activity between baseline and T3 were specified as an outcome, differences in self-management knowledge between baseline and T2 acted as a mediator, baseline anxiety and depression were specified both as a moderator of the effect of the intervention on differences in self-management knowledge and of the effect of differences in self-management knowledge on differences in physical activity. All predictor variables were mean-centred prior to analysis, and continuous measures were used for all variables. In order to account for the clustering within practices, a fixed effects approach was used by partialling out the cluster effects from estimates of the coefficients and the related standard errors. Bootstrapping was used to determine the confidence intervals and standard errors for the indirect effect of the intervention on physical activity via self-management knowledge. All analyses controlled for baseline smoking status (yes/no), smoking history (mean pack years), baseline FEV1 as per cent of predicted, the number of comorbidities, age, and gender.

Results

At baseline, the sample consisted of 96 (52.7%) male participants. The overall mean age was 67.7 (SD = 7.8). 76 participants (41.8%) reported being a current smoker with the overall sample reporting a mean number of 48.6 (SD = 24.4) pack years. The overall FEV1% predicted mean was 55.2 (SD = 13.4). 39 participants (21.4%) scored ≥ 8 on the HADS depression scale and 74 participants (40.7%) scored ≥ 8 on the anxiety scale. 31 participants (17%) had a score ≥ 8 on both subscales. Table 1 shows descriptive statistics for the study variables.

Table 1.  Means (standard deviations) and observed range of study variables at different points in time in the usual care (UC) and health mentoring (HM) groups

Variable	UC	HM	Overall	Observed Range	n
Anxiety
T1	7.03 (4.07)	6.73 (4.12)	6.88 (4.09)	0–18	181
T2	6.56 (3.78)	6.00 (3.63)	6.33 (3.72)	0–15	140
T3	6.15 (3.98)	6.22 (4.13)	6.18 (4.03)	0–17	138
Depression
T1	5.12 (3.55)	4.64 (3.14)	4.88 (3.36)	0–17	180
T2	4.65 (3.50)	3.79 (2.83)	4.29 (3.25)	0–14	140
T3	4.74 (3.38)	4.07 (2.60)	4.45 (3.07)	0–17	138
Self-management Knowledge
T1	24.17 (5.86)	23.51 (6.50)	23.85 (6.18)	0–32	172
T2	25.20 (5.10)	27.14 (4.71)	26.01 (5.02)	9–32	142
T3	25.29 (5.02)	26.64 (5.48)	25.86 (5.25)	8–32	139
Physical Activity
T1	4874.45 (2573.76)	5094.50 (3175.03)	4977.77 (2864.85)	458.25–15730.00	164
T2	4547.72 (2485.13)	5360.45 (2968.14)	4871.61 (2707.12)	513.00–15345.60	138
T3	4708.74 (2389.44)	4770.16 (2975.48)	4732.79 (2622.53)	473.29–14450.50	120

There was a significant main effect of the intervention on changes in self-management knowledge; B = 0.50, t(108) = 2.39, p = .019. There also was a small, yet significant negative main effect of anxiety on gain in self-management knowledge; B = −0.06, t(108) = −2.22, p = .029. However, contrary to our hypotheses, there was no interaction effect between anxiety and the effect of the intervention on gaining knowledge; B = −0.08, t(108) = −1.50, p = .138¹. While there was no significant main effect of changes in self-management knowledge on changes in physical activity (B = 0.05, t(108) = 0.51, p = .611), there was a significant interaction effect between self-management knowledge and anxiety on physical activity; B = −0.67, t(108) = −2.70, p = 0.008 (see Figure 1). Moreover, there was a significant indirect effect of the intervention on changes in physical activity via changes in self-management knowledge for participants scoring one SD (or lower) below the mean score of anxiety: indirect effect_ab = 0.24, 95% CI [0.01, 0.70].

Figure 1. Moderated mediation model for anxiety with unstandardized path coefficients (and standard errors). Coefficients for the depression model are given in square brackets.

Following up on the significant interaction effect, the Johnson-Neyman technique was used to identify the values on the continuum of anxiety for which the effect of self-management knowledge on physical activity became significant. Results showed that for participants with a HADS anxiety score below 3.89 (23% of the sample), there was a significant positive effect of changes in self-management knowledge on changes in physical activity, whereas for participants scoring above 12.55 on the HAD anxiety scale (8.3% of the sample), there was a significant negative effect of knowledge on activity (see Figure 2).

Figure 2. Regions of significance for the interaction between self-management knowledge and anxiety on physical activity.

Results for the model testing depression as a moderator are comparable to the model for anxiety, except that there was no main effect for depression on changes in self-management knowledge; B = −0.01, t(108) = −0.46, p = .65. As in the model for anxiety, there was no interaction effect between depression and the effect of the intervention on changes in knowledge; B = −0.09, t(108) = −1.34, p = .182, and no main effect of gains in self-management knowledge on changes in physical activity (B = 0.04, t(108) = 0.44, p = .662). However, once again there was a significant interaction effect between self-management knowledge and depression on physical activity; B = −0.09, t(108) = −3.03, p = 0.003 (see Figure 1). Moreover, there was once more a significant indirect effect of the intervention on changes in physical activity via changes in self-management knowledge for participants scoring one SD (or lower) below the mean score of depression: indirect effect_ab = 0.30, 95% CI [0.03, 0.75].

For participants with a HADS depression score below 2.56 (26.7% of the sample), there was a significant positive effect of changes in self-management knowledge on changes in physical activity, whereas for participants scoring above 7.30 on the HAD depression scale (14.4% of the sample), there was a significant negative effect of knowledge on activity (see Figure 3).

Figure 3. Regions of significance for the interaction between self-management knowledge and depression on physical activity.

Discussion

This study was the first to test the impact of anxiety and depression in a model exploring the mechanisms of an intervention targeting self-management knowledge and subsequent behaviour among COPD patients. Results suggest that the health-mentoring intervention was successful in increasing self-management knowledge, which in turn mediated the intervention's effect on changes in physical activity, but only in participants reporting lower levels of anxiety or depression.

Importantly, the model showed a significant interaction between HADS-score and changes in knowledge, indicating that for participants with low levels of anxiety or depression, high levels of knowledge are related to more physical activity, whereas for highly anxious or depressed participants, more self-management knowledge is associated with less activity. The regions of significance for the interaction effects implied that about a quarter of the sample significantly profited from increases in knowledge, whereas for about one in ten participants the changes in self-management brought about by the intervention seemed to result in detrimental effects on subsequent self-management behaviour.

On a theoretical level, this interaction can help to explain why self-management interventions do not necessarily lead to desired behaviour changes in a diverse sample of COPD patients, as it is likely that potential intervention effects are not the same at different levels of anxiety or depression (26). Moreover, this effect has important implications for clinical practice, as it suggests that in order to prevent potential adverse effects among highly anxious or depressed patients, it is important to treat participants’ psychological comorbidities before facilitating self-management skills development. However, studies show that less than a third of patients with COPD receive treatment for anxiety and/or depression (27, 28).

The cognitive model of anxiety (18) emphasises the importance of constructive thinking whereby intrusive, unhelpful, and anxiety-provoking thoughts are replaced by more helpful thoughts that enhance one's coping with the situation. For people with COPD, this cognitive skill involves a difficult balancing act between regular monitoring of symptoms that could indeed signify real danger and might require medical intervention, and avoiding catastrophic misinterpretation of benign symptoms. A study in patients with heart failure also reported a link between illness perceptions and anxiety and depression, showing that patients who are unable to make sense of their illness and perceive its effects as uncontrollable and threatening, also reported higher levels of anxiety and depression (29). While cognitive behavioural therapy (CBT) was shown to be effective in the treatment of panic and anxiety disorders in physically healthy populations (30, 31), a recent meta-analysis reported only small effects of CBT on anxiety among patients with COPD (32). However, the same meta-analysis reported that interventions including exercise components seem to be more effective in reducing clinical and subclinical levels of anxiety in this patient population. Likewise, participation in pulmonary rehabilitation (PR), and especially the physical exercise component of PR has been associated with decreased symptoms of anxiety and depression among patients with COPD (33).

The empirical results from this study did not support the theoretical assumption that high anxiety leads to an increased motivation to obtain self-management knowledge—rather, there was a small, yet significant negative main effect of anxiety on changes in knowledge, indicating that higher anxiety levels result in fewer improvements in knowledge, independent of participation in the health-mentoring program. This seems to contradict findings from other studies (16, 17), which showed that anxiety does not lead to avoidance of illness information. A potential explanation for this negative main effect of anxiety on knowledge uptake may be that anxiety can impede cognitive processing of information, especially with regard to threatening stimuli (34). However, since the size of this effect was relatively small and was not replicated for depression, future studies should further evaluate this finding and explore potential confounders of this effect such as cognitive performance capacity.

The connection between dyspnoea, panic attacks and anxiety disorder is well supported by theory and evidence (18, 35). However, another aspect of anxiety may be of importance, especially when examining its effect on physical activity. The study by Vögele and von Leupoldt (12) suggests that of all the types of anxiety disorders, panic disorder with agoraphobia is especially prevalent among COPD patients. A patient suffering from agoraphobia is likely to avoid places and situations in which it is difficult to obtain help. If a COPD patient considers their home to be the only safe environment, this leads to severely limited mobility and very restricted possibilities to perform physical activity. Thus, exposure-based treatments during which patients are confronted with avoided situations and physical sensations in a safe environment could help increase mobility—a mechanism likely to play a role in the findings about the beneficial effects of interventions incorporating exercise components on anxiety among patients with COPD. However, it can be assumed that if a patient's anxiety is displayed on a more general, rather than a disease-, or symptom-specific level, more extensive therapy sessions are indispensable.

The current study has a number of strengths and weaknesses. Since the tested moderated mediation model was fairly complex, we avoided introducing anxiety and depression into the same model and reported results for the two models separately. Nevertheless, future studies may seek to disentangle the impact of anxiety and depression on self-management knowledge and behaviours by exploring their independent roles, as well as potential interaction effects. Moreover, as illness perceptions seem to play an important role for both anxiety and depression and self-management behaviours in COPD (8, 29), it may be worthwhile to explore the influence of illness perceptions in more detail. As is the case with all trials based on convenience samples, generalizability of the results to the general population is an issue. Especially in studies among people with COPD, cautious interpretation of the results is required, as the rates of attrition and refusal to participate among this population is particularly high (9). Therefore, it is important to replicate the results in larger population samples, which would also allow for more advanced statistical procedures.

The major strengths of this study are that it tests for the role of anxiety and depression by involving an experimental manipulation of self-management knowledge, and that it explores the two theoretical pathways of how anxiety and depression might influence knowledge uptake and self-management behaviour in a concurrent statistical model. Ideally, a future study should additionally try to target these comorbidities by involving CBT and a guided exercise intervention, so that an experimental-causal-chain design (36) can be established, which will provide stronger evidence for the causality of the processes at work.

Conclusion

This study results in a number of practical implications. First of all, it shows that a health-mentoring intervention delivered by community nurses can successfully influence physical activity as mediated by changes in self-management knowledge, but not in populations reporting clinical or elevated levels of anxiety or depression. More importantly, it also implies that such an intervention may lead to detrimental effects on self-management behaviour in participants with high levels of either comorbidity. Levels of psychological treatment in this population have been shown to be inappropriately low. Therefore, it is essential not only to consider the moderating role of such comorbidities for the effectiveness of self-management interventions, but also to actively target them in this clinical population in order to prevent adverse intervention effects in a subgroup of participants.

Declaration of Interest Statement

We wish to confirm that there are no known conflicts of interest associated with this publication. The authors alone are responsible for the content and writing of the paper.

This work was supported by NHMRC project grant ID490028, NHMRC Centre of Research Excellence grant ID1001062, a Lung Foundation Australia/Boehringer Ingelheim COPD Research Fellowship for J. Walters, a Royal Hobart Hospital Research Foundation grant, and a University of Tasmania Institutional Research Grant.

Acknowledgments

Health-mentor training was developed and delivered by H. Cameron-Tucker, L. Joseph, and E. Cummings.

Notes

¹ The influence of anxiety remains the same when controlling for HADS depression.