Effects of Exercise on Depressive Symptoms in Adults With Arthritis and Other Rheumatic Disease

A Systematic Review of Meta-analyses

George A Kelley; Kristi S Kelley


BMC Musculoskelet Disord. 2014;15(121) 

In This Article


Characteristics of Included Meta-analyses

Of the 95 citations initially identified, 69 (72.6%) remained after removing duplicates. Of the 69 articles that were screened, two aggregate data meta-analyses, both in participants with fibromyalgia, met the criteria for inclusion.[30,37] Post hoc, one study that was initially included was removed because it was not focused specifically on the effects of exercise on depressive symptoms in adults with a specific type of arthritis or other rheumatic disease, but rather, in adults with a variety of chronic illnesses.[43] The precision of the searches was 0.03 while the NNR was 35. The major reasons for exclusion of ineligible studies were an inappropriate study design (48.4%) followed by an inappropriate intervention (17.9%), population (18.9%), outcome (9.5%) and comparison (6.3%). No meta-analysis was deleted because they did not report their results as a SMD. A flow diagram that depicts the search process can be found in Figure 1 while a list of all excluded studies, including the reasons for exclusion, is shown in Additional File 2 .

For the two included meta-analyses,[30,37] one included studies on aerobic or strength training exercise[30] while the second was limited to aerobic exercise studies but also included studies in which participants performed strength training as long as the number of minutes of strength training did not exceed the number of minutes spent performing aerobic exercise.[37] Both meta-analyses included fibromyalgia participants as defined by the diagnostic criteria of the original studies.[30,37] A general description of the characteristics of each meta-analysis is provided in Table 1 .

Figure 1.

Flow diagram for the selection of studies. *, number of reasons exceeds the number of studies because some studies were excluded for more than one reason.

Methodological Quality

Item by item results for each meta-analysis using the AMSTAR instrument is shown in Additional File 3 . The meta-analysis by Busch et al.[30] satisfied 10 of the 11 (91%) of the AMSTAR criteria while the study by Hauser et al.[37] met 9 of the 11 criteria (82%). One meta-analysis was judged as (1) not avoiding the status of publication as an inclusion criterion and (2) not providing a list of excluded studies.[37]

Data Synthesis

The overall results for both included systematic reviews with meta-analysis are shown in Table 2 . As can be seen, SMD reductions in depressive symptoms included non-overlapping confidence intervals for both with one meta-analysis yielding a small SMD[37] and one yielding a medium SMD.[30] While the overall SMD was approximately twice as large for the Busch et al. meta-analysis,[30] the between-meta-analysis 95% CIs for both were overlapping, suggesting no statistically significant difference between the two studies.[30,37] In addition, the pooled SMD for the Busch et al. meta-analysis was the result of the current investigative team combining the results from those studies meeting the ACSM guidelines for aerobic exercise with the one strength training study and one aerobic study not meeting the ACSM recommendations, as reported by the authors.[30] A statistically significant and a large amount of heterogeneity were found for both meta-analyses as well as overlapping 95% PIs.[30,37] Data for the NNT, number who could benefit and percentile improvement are shown in Table 3 . No small-study effects were observed for the overall meta-analysis results of Busch et al.[30]0, -3.8, 95% CI, -10.9 to 3.3, p = 0.11) while statistically significant small study effects were observed for the Hauser et al.[37] meta-analysis (β0, -2.4, 95% CI, -4.8 to -0.01, p = 0.02). With each study deleted from the overall model once for each meta-analysis,[30,37] SMD changes in depressive symptoms remained statistically significant with non-overlapping confidence intervals. Results ranged from -0.49 (95% CI, -0.12 to -0.85, p = 0.009) to -0.72 (95% CI, -0.35 to -1.10, p < 0.0001) in the study by Busch et al.,[30] and -0.23 (95% CI, -0.08 to -0.38, p = 0.003) to -0.36 (95% CI, -0.15 to -0.57, p = 0.001) in the study by Hauser et al..[37] Cumulative meta-analysis, ranked by year, revealed that SMD changes in depressive symptoms have remained statistically significant with non-overlapping confidence intervals since 2001 for the meta-analysis by Busch et al.[30] (range of years, 1996 to 2004) and from 2004 onward in the meta-analysis by Hauser et al.[37] (range of years, 1996 to 2009). No statistically significant difference was found between active control groups and the other types of control groups included in the Hauser et al. meta-analysis (Qb = 1.13, p = 0.29).[37]

Subgroup results were provided for both meta-analyses.[30,37] For the Busch et al. meta-analysis,[30] results for depressive symptoms were reported according to studies meeting the ACSM recommendations for aerobic exercise,[30,31] (4 studies), those not meeting the recommendations (1 study) and those limited to strength training exercise (1 study). For those studies meeting the ACSM recommendations, a statistically significant reduction in depressive symptoms was observed (SMD, -0.40, 95% CI, -0.04 to -0.76, p = 0.003). The SMD for the study not meeting the recommendations was -1.22 (95% CI, -1.90 to -0.54) while the SMD for the study in which strength training was performed was -1.14 (95% CI, -2.08 to -0.20). Subgroup analyses in the study by Hauser et al.[37] included results partitioned according to two studies in which low intensity exercise (50% to 60% of maximum heart rate) was compared to moderate intensity exercise (60% to 80% of maximum heart rate) as well as eight studies in which land-based exercise was compared to water-based exercise. The authors reported no statistically significant differences between either low or moderate intensity exercise (SMD, -0.16, 95% CI, -0.67 to 0.13, p = 0.53) or land versus water-based exercise (SMD, -0.44, 95% CI, -0.88 to 0.01, p = 0.05).