Diastolic Dysfunction in Rheumatoid Arthritis: A Meta-Analysis and Systematic Review




To determine if the prevalence of diastolic dysfunction is increased in rheumatoid arthritis (RA) patients.


We conducted a time- and language-restricted literature search to identify studies conducted to compare echocardiographic parameters in patients with RA and controls. The mean difference for echocardiographic variables of interest was calculated using a random-effects model. A systematic review of the literature was performed.


A total of 25 studies reporting on 5,836 subjects (1,614 with RA) were included. Results reflect mean differences, with positive values denoting higher values in RA patients. Patients with RA had larger mean left atrial dimension (mean difference 0.09 cm [95% confidence interval (95% CI) 0.01, 0.17]; P = 0.02), higher left ventricular mass index (mean difference 6.2 gm/m2 [95% CI 1.08, 11.33]; P = 0.02), higher mean systolic pulmonary artery pressure (mean difference 5.87 mm Hg [95% CI 4.36, 7.38]; P < 0.00001), prolonged isovolumetric relaxation time (mean difference 9.67 msec [95% CI 5.78, 13.56]; P < 0.00001), and higher transmitral A wave velocity (mean difference 0.13 meters/second [95% CI 0.07, 0.18]; P < 0.00001) compared to controls. A subanalysis of 2,183 subjects excluding 2 large unmatched studies showed the same results, with the exception that patients with RA had a lower mitral E/A ratio (mean difference −0.17 [95% CI −0.25, −0.09]; P < 0.00001), suggestive of diastolic dysfunction. There were no differences in left ventricular ejection fraction (%), transmitral E wave velocity (meters/second), and mitral deceleration time (msec).


Patients with RA were more likely to have echocardiographic parameters of diastolic dysfunction, and have higher systolic pulmonary artery pressures and larger left atrial sizes.


Patients with rheumatoid arthritis (RA) have an approximately 2-fold higher incidence of congestive heart failure (CHF) as compared to the general population (1). CHF is an independent risk factor for mortality in RA patients and is responsible for 1 in 8 deaths of patients with RA (2).

Left ventricular diastolic dysfunction (LVDD) encompasses mechanical abnormalities involving decreased distensibility, impaired relaxation, and abnormal diastolic filling of the left ventricle, irrespective of the ejection fraction (EF) (3), and has been associated with diabetes mellitus (DM), old age, coronary artery disease (CAD), hypertension, and obesity. Increased left atrial (LA) dimension is also suggestive of DD in the absence of arrhythmias (4, 5). Evidence supports an association between increased LV mass and impaired DD (6, 7). DD is an echocardiographic diagnosis made via transthoracic echocardiography (TTE), and cannot be made clinically (8). The importance of DD lies in the fact that it may serve as a precursor to systolic and diastolic CHF and also may cause morbidity and mortality on its own.

DD is fairly common. In the general population, a cross-sectional survey of more than 2,000 subjects revealed an overall DD prevalence of 28.1% (9), while another study involving more than 36,000 subjects had a DD prevalence of 65.1%, with the majority (60.0%) having mild DD (10). In the general population, DD is an independent predictor of mortality (4, 9, 10) and incident CHF (11).

The higher incidence of CHF in RA necessitates the research of its precursor forms and preclinical assessment, and therefore the importance of studying DD in patients with RA. In RA, one study showed the prevalence of DD to be approximately 37% (12). The TTE results are conflicting regarding diastolic parameters in RA patients (see the Results). The data have been variable with respect to EF, DD, and LV mass in RA. Therefore, we performed this meta-analysis and systematic review to compare the presence of DD in RA patients as compared to non-RA controls. Our hypothesis was that DD is increased in patients with RA independent of other risk factors.

Significance & Innovations

  • Diastolic dysfunction is more common in rheumatoid arthritis (RA) patients as compared to controls.

  • RA patients have larger left atrial sizes and higher pulmonary artery pressures as compared to controls.


Literature search.

The review was conducted according to the recommendations of the Meta-Analysis of Observational Studies in Epidemiology group (13). An experienced librarian conducted a time-restricted (from January 1, 1990 to July 31, 2012) and language-restricted (English) Medline search with the medical subject headings terms “(heart failure OR ventricular dysfunction OR atrial OR echocardiography) AND (rheumatoid arthritis)” in August 2012. Time restriction was employed to ensure uniformity of echocardiographic techniques. The Cochrane, CINAHL, and Web of Science databases were also searched. We also searched the bibliography of the identified studies as well as general review articles on the topic to identify further relevant studies. Unpublished studies were not included. Bibliographically-identified studies were only included if they could be retrieved from Medline. To ascertain the trend in the non–English language literature, we conducted a similar search restricted to non–English language articles concurrently. A Medline search for non-English studies revealed 128 titles, of which only 2 met our inclusion criteria. Both suggested an increased DD prevalence in RA as compared to controls, based on abstract review (14, 15).

Study eligibility and selection.

Two authors (FA and MA) independently identified studies using our search strategy. We included all studies that 1) were conducted to compare the TTE parameters in RA patients and controls, 2) included patients with RA who met the American College of Rheumatology 1987 classification criteria for RA (16), and 3) had a case–control design. No studies were excluded based on participant number. When multiple studies were published using the same data, the corresponding authors were contacted. When no response was received, the latest study with the highest number of subjects was included.

TTE data.

Based on TTE, there are 3 grades of DD: mild or grade I, i.e., impaired relaxation pattern; moderate or grade II, i.e., pseudonormal pattern; and severe or grade III, i.e., restrictive filling pattern (17). The TTE parameters commonly used to determine DD include isovolumetric relaxation time (IVRT; msec), mitral valve deceleration time (msec), transmitral early filling peak velocity (meters/second), transmitral late filling peak velocity (meters/second), and the E/A ratio (18). There is no gold standard for TTE-based diagnosis, and a combination of the aforementioned variables was used to grade DD, with the more variables suggesting DD, the stronger the diagnosis; however, data were not available for us to assess grading, nor was grading reported in most of the studies. These variables change in different directions with increasing DD grades, e.g., E/A ratio increases and deceleration time decreases (18).

Data were available for most of the routine TTE parameters. For the purpose of our systematic review, studies reporting any single diastolic parameter suggesting DD were classified as having DD. The majority of studies relied on a single abnormal value to classify subjects as having DD. Studies were included irrespective of the TTE technique used and irrespective of the echocardiographic parameters used to classify DD.

Data abstraction and quality assessment.

Data were extracted independently by 2 authors (FA and SJB) on a mutually agreed data extraction sheet to collect information on study characteristics, participant information, RA disease features, TTE parameters, and medication usage. The Newcastle-Ottawa Scale for case–control studies was used to assess the quality of the included studies (19). This quality score was calculated on the basis of 3 major components of the studies: selection of study groups (0–4 points), comparability of study groups (0–2 points), and ascertainment of exposure (0–3 points). A higher score indicated better methodologic quality.

Statistical approach.

Unit discordance for variables was resolved by converting all units to a standard measurement for that variable. Percentages and means ± SDs were calculated to describe the distributions of categorical and continuous variables, respectively. Since individual patient information was not available for all patients, we report weighted means and SDs. A Student's t-test for independent samples was used for continuous variables and the chi-square test with Yates' correction was used for proportions. A 2-tailed alpha level of 5% was used for hypothesis testing. The baseline data were analyzed using the Statistical Package for Social Scientists, version 20.0 (IBM).

The meta-analyses were performed using Review Manager, version 5.1 (Cochrane Collaboration). Mean differences and their 95% confidence intervals (95% CIs) were used to summarize the effect size for each echocardiographic variable using the random-effects model. Measures of heterogeneity, including Cochran's Q statistic, the I2 index, and the tau-squared test, were estimated and reported. Since 2 studies did not provide results after matching cases and controls and therefore a source of heterogeneity, we performed an additional subgroup analysis after exclusion of these unmatched studies. Publication bias was assessed using a funnel-plot; analyses were done with respect to all echocardiographic variables of interest. A random-effects model was used to calculate the mean difference and 95% CI, given variable degrees of data heterogeneity and given the inherent heterogeneity in any systematic review of studies from the published literature.


A total of 520 citations were identified, of which 25 were included in the final analysis (Figure 1). There were 5,836 subjects, of which 1,614 (27.7%) had RA. The mean duration of RA was 8.45 years. When given, the mean Disease Activity Score in 28 joints (DAS28) ranged from 3.8–5.9 for the RA patients. When mean scores were not given, 21–77% of cases had DAS28 scores >3.2. Most studies (68%) used transmitral pulse wave Doppler as the TTE technique, while the remaining studies also employed tissue Doppler imaging. Limited data were available on annular mitral velocities using tissue Doppler imaging for meaningful use in this analysis.

Figure 1.

Search strategy for identification of studies.

The details of the individual studies are given in Table 1. All of the studies had age- and sex-matched subjects and as a general exclusion criterion, excluded subjects with CAD, DM, hypertension, and valvular heart disease. Two studies, however, included patients with such confounding comorbidities (20, 21). These aforementioned 2 studies did not have overlap of echocardiographic data (Gabriel S: unpublished observations). The average quality score was 5.9 of a maximum possible 9 on the Newcastle-Ottawa Scale for the included studies. Data pertaining to racial characteristics in most studies were unreported. Characteristics of the study subjects are given in Table 2. There were baseline differences in age, sex, DM, hypertension, hyperlipidemia, body mass index, history of CAD, and smoking, largely due to the 2 unmatched studies. Data excluding these 2 studies (Table 2) eliminated the baseline differences. The systematic review showed that 21 studies had evidence of DD in RA patients as compared to controls based on prolonged IVRT or deceleration time, lower E/A ratios, and/or increased LA size (Table 1). The presence of any one of these features was considered as evidence of DD.

Table 1. Major features of studies included in the analysis*
Author, year (ref.)CountrySubjectsAge/sex matchedYears with RABlind TTEResultsNotesRA as DD riskQuality score
  • *

    RA = rheumatoid arthritis; TTE = transthoracic echocardiography; DD = diastolic dysfunction; N = no; NR = not reported; RF+ = rheumatoid factor positive; CCP = cyclic citrullinated peptide; Y = yes; DAS28 = Disease Activity Score in 28 joints; EF = ejection fraction; IVRT = isovolumetric relaxation time; LA = left atrial; IL-6 = interleukin-6; LV = left ventricular; DT = deceleration time; MI = mass index; DMARDs = disease-modifying antirheumatic drugs; PAP = pulmonary artery pressure; RV = right ventricular; MPI = myocardial perfusion index.

Crowson et al, 2011 (20)USTotal: 1,961 RA: 231N/NMean 8.8NRDD in 26.5% of RA and 21.7% of non-RA68% were RF+ CCP not givenY3/1/3
Marasovic-Krstulovic et al, 2011 (26)CroatiaTotal: 160 RA: 80Y/YMean 9.3YDD was seen in 36.25% of RA and 15% of non-RA (P = 0.002) CCP-positive patients need earlier TTE examinationRF % and CCP % not  given RA medication use was  not a predictor of DDY4/1/3
Sitia et al, 2012 (52)ItalyTotal: 42 RA: 22Y/YMean 2.9YDT higher in RA (P < 0.05)RF % and CCP % not  given Mean DAS28 5.89Y3/1/3
Abdul Muizz et al, 2011 (53)MalaysiaTotal: 106 RA: 53Y/YMedian 5.0YNo difference in EF, E/A ratio, IVRT, or LA size62.3% were RF+ CCP not given No association with  duration of disease or  DAS28 DAS28 >3.2 in 77% of  casesN2/1/2
Liang et al, 2010 (21)USTotal: 1,692 RA: 244N/NMedian 8.2NRDD more in RA (P = 0.006)71% RF+ and 45% CCP+ DD related to disease  duration and IL-6 levelY3/1/2
Rudominer et al, 2009 (27)USTotal: 178 RA: 89Y/YMean 12.0YLV end DD higher in RA (P < 0.001), LA dimension higher in RA (P < 0.001)56% RF+ and 53% CCP+ RA was an independent  predictor of LV massN3/1/2
Acar et al, 2009 (54)TurkeyTotal: 109 RA: 68Y/YMean 5.9YNo differenceRF % and CCP % not  given Mean DAS28 4.7N3/1/2
Wislowska et al, 2008 (55)PolandTotal: 60 RA: 30Y/YMean 12.5NRRA had higher IVRT (P < 0.0005), higher E/A ratio (P = 0.003), higher DT (P = 0.009), and larger LV mass (P = 0.02)53.4% were RF+ CCP not given No relation with disease  duration Mean DAS28 3.8Y3/1/1
Yazici et al, 2008 (38)TurkeyTotal: 139 RA: 72Y/YMean 7.8NRIn RA, E/A ratio was lower (P < 0.0001) and LVMI was higher (P < 0.01)RF % and CCP % not  given No relation with DMARDs  or disease No change in DD  parameters at 5-year  followup Mean DAS28 3.9Y2/1/0
Udayakumar et al, 2007 (56)IndiaTotal: 90 RA: 45Y/YMean 5.1NRE/A ratio was lower in RA (P = 0.004), while IVRT (P = 0.001), PAP (P = 0.003), and LA dimension (P = 0.01) were higher in RA62% were RF+ CCP not given DD related to disease  durationY4/1/1
Meune et al, 2007 (57)FranceTotal: 54 RA: 27Y/YMean 8.3NRIn RA, higher LVMI (P = 0.009) and higher PAP (P = 0.015)59.3% RF+ and 77.8%  CCP+ No relation with duration  or inflammation Mean DAS28 4.3Y3/1/2
Guler et al, 2007 (30)TurkeyTotal: 57 RA: 30Y/YMean 9.3NRIn RA, E/A ratio was lower (P = 0.011); higher LVMI (P = 0.045), thicker posterior wall (P = 0.013), and bigger LA (P = 0.016)RF % and CCP % not  given Higher P wave dispersion  and duration in RA Mean DAS28 5.4Y2/1/1
Birdane et al, 2007 (22)TurkeyTotal: 100 RA: 60Y/YMean 10.7NRIn RA, DT and IVRT increased (P < 0.05)57% were RF+ CCP not given RVDD in RA Mean DAS28 4.3Y3/1/1
Canturk et al, 2006 (58)TurkeyTotal: 68 RA: 34Y/YMean 11.7YIVRT higher in RA (P = 0.02)RF % and CCP % not  given Some parameters  correlated with disease  duration DAS28 >3.2 in 35% of  casesY2/1/1
Rexhepaj et al, 2006 (23)KosovoTotal: 121 RA: 81Y/YMean 6.3NRLower E/A ratio in RA (P < 0.001)RF % and CCP % not  given RVDD in RAY2/1/1
Arslan et al, 2006 (59)TurkeyTotal: 99 RA: 52Y/YMean 6.4NRLarger LA in RA (P < 0.001), lower E/A ratio and higher DT in RA (both P < 0.001)RF % and CCP % not  given Positive relation with  disease duration for  E/A ratio and DTY2/1/1
Gonzalez-Juanatey et al, 2004 (60)SpainTotal: 94 RA: 47Y/YMean 15.5YIn RA, DT (P = 0.04), IVRT (P = 0.003), and systolic PAP (P = 0.004) were higher, while E/A ratio was lower (P = 0.01)85% were RF+ CCP not given No relation with disease  duration or activity DAS28 >3.2 in 21% of  casesY4/1/1
Levendoglu et al, 2004 (24)TurkeyTotal: 84 RA: 40Y/YMean 10.0NRIn RA, lower E/A ratio (P < 0.05) and higher IVRT and MPI (P < 0.05)RF % and CCP % not  given RV diameters with lower  E/A ratio in RA (P < 0.05) Duration related to E/A ratio and IVRTY2/1/1
Alpaslan et al, 2003 (25)TurkeyTotal: 64 RA: 32Y/YMean 9.0YLower E/A ratio in RA (P < 0.05), higher IVRT and MPI in RA (P < 0.05)RF % and CCP % not  given RV function and values  were sameY2/1/1
Di Franco et al, 2000 (61)ItalyTotal: 65 RA: 32Y/YMean 9.0YLower E/A ratio in RA (P = 0.002), LA diameter higher in RA (P = 0.03)RF % and CCP % not  given Correlation present  between disease  duration and E/A ratio  (r = 0.04)Y2/1/1
Montecucco et al, 1999 (62)ItalyTotal: 108 RA: 54Y/YMean 5.9YLower E/A ratio in RA (P = 0.003)76% were RF+ CCP not given DD related to disease  durationY3/1/1
Wislowska et al, 1998 (63)PolandTotal: 200 RA: 100Y/YMean 9.4YIn RA, higher LV end DD (P < 0.01) and aortic root diameter (P < 0.001)78% were RF+ CCP not givenN4/1/1
Corrao et al, 1996 (64)ItalyTotal: 80 RA: 40Y/YMean 6.9YE/A ratio higher in RA (P = 0.001), LVMI also higher in RA (P = 0.02)RF % and CCP % not  given No association with drug  useY3/1/1
Mustonen et al, 1993 (65)FinlandTotal: 26 RA: 12Y/YMean 8.0NRIVRT was prolonged in RA (P = 0.01)100% were RF+ CCP not given IVRT correlated with  disease durationY4/1/1
Maione et al, 1993 (66)ItalyTotal: 79 RA: 39Y/YMean 7.1YMore DD in RA (P = 0.019)71.8% were RF+ CCP not given No relation to disease  duration or activityY4/1/1
Table 2. Characteristics of study subjects*
 Matched dataUnmatched data
RA subjects (n = 1,614)Non-RA subjects (n = 4,222)PRA subjects (n = 1,139)Non-RA subjects (n = 1,044)P
  • *

    RA = rheumatoid arthritis; NS = not significant; BSA = body surface area; BMI = body mass index; CAD = coronary artery disease; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate.

  • Data available for all subjects.

Sex, no. (%)  < 0.001  NS
 Men347 (22)1,714 (41) 219 (19)213 (20) 
 Women1,267 (78)2,508 (59) 920 (81)831 (80) 
Age, mean years51.659.1< 0.00148.347.4NS
RA duration, mean years8.45 6.0 
BSA, mean m21.81.8NS1.81.8NS
 No. (%)421 (26.1)389 (9.2) 421 (37.0)389 (37.3) 
BMI, mean kg/m226.928.0NS25.425.6NS
 No. (%)963 (59.7)3,616 (85.6) 488 (42.8)438 (42.0) 
CAD, %1.85.1< 0.0010.00.0NS
Diabetes mellitus, %2.35.3< 0.0010.20.1NS
 No. (%)1,311 (81.2)2,425 (57.4) 1,067 (93.7)977 (93.6) 
Hypertension, %20.033.6< 0.0010.90.3NS
 No. (%)1,542 (95.5)4,155 (98.4) 1,067 (93.7)977 (93.6) 
Hyperlipidemia, %18.040.9< 0.0011.01.5NS
 No. (%)554 (34.3)1,709 (40.5) 310 (27.2)261 (25.0) 
Smokers, %28.344.8< 0.0017.89.6NS
 No. (%)951 (58.9)3,605 (85.4) 476 (41.8)427 (40.9) 
CRP level, mean mg/liter7.421.94< 0.00112.92.2< 0.001
 No. (%)925 (57.3)1,999 (47.3) 450 (39.5)269 (25.7) 
ESR, mean mm/hour38.5610.74< 0.00138.5610.74< 0.001
 No. (%)538 (33.3)256 (6.1) 538 (47.2)256 (24.5) 

When reported, 6 studies suggested that DD was not related to disease duration and/or disease activity, whereas 7 studies suggested such a link (Table 1). Unfortunately, there were not enough data given to compare those RA patients with and without DD to determine any association with disease activity or duration for our analysis. Four studies (Table 1) reported on right ventricular (RV) parameters, of which 3 studies (22–24) showed the presence of RVDD in RA patients and one study did not (25).

Most studies reported on medications used in RA patients (data not shown). Patients receiving known cardiotoxic medicines such as gold and penicillamine were mostly excluded. Since we performed a meta-analysis on the aggregate data and did not have access to individual patient–level data, we were unable to assess the effects of medication used on echocardiographic parameters. Three studies that analyzed the effects of medicine usage on echocardiographic parameters found no association with diastolic parameters (21, 26, 27).

Publication bias was assessed with respect to all TTE variables with no evidence of bias. Data are only shown for the E/A ratio (Figure 2), which showed symmetric distribution of all studies at both extremes of outcomes as well as around the midline consistent with no effect.

Figure 2.

Funnel-plot for the E/A ratio to assess for publication bias. MD = mean difference. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/acr.21861/abstract.

All results reflect the mean difference and 95% CI, with positive values reflecting a higher value in RA patients (Table 3). The mean LA dimension was higher in RA patients as compared to controls (mean difference 0.09 cm [95% CI 0.01, 0.17]; P = 0.02) (Figure 3A). The LV mass index (LVMI) was higher in RA patients (mean difference 6.2 gm/m2 [95% CI 1.08, 11.33]; P = 0.02) (Figure 3B). The pulmonary artery pressure (PAP) was higher in RA patients (mean difference 5.87 mm Hg [95% CI 4.36, 7.38]; P < 0.00001) (Figure 3C). The IVRT was also higher in RA patients (mean difference 9.67 msec [95% CI 5.78, 13.56]; P < 0.00001) (Figure 3D). The A wave velocity was higher in RA patients (mean difference 0.13 meters/second [95% CI 0.07, 0.18]; P < 0.00001). There was no significant difference in the EF, E wave velocity, E/A ratio, and deceleration time between the 2 groups.

Table 3. Meta-analysis outcomes (random-effects model)*
Echo variableMean difference (95% CI)PQI2, %τ2§
  • *

    95% CI = 95% confidence interval; EF = ejection fraction.

  • Cochran's Q statistic for heterogeneity.

  • I2 index for degree of heterogeneity.

  • §

    Tau-squared measure of heterogeneity.

Left atrial dimension, cm (n = 846)0.09 (0.01, 0.17)0.0268.0578.00.02
 Unmatched studies excluded (n = 846)0.09 (0.01, 0.17)0.0268.0578.00.02
Left ventricular mass index, gm/m2 (n = 804)6.2 (1.08, 11.33)0.0265.6985.057.13
 Unmatched studies excluded (n = 560)7.19 (4.38, 10.00)< 0.0000111.3020.04.07
Pulmonary artery pressure, mm Hg (n = 594)5.87 (4.36, 7.38)< 0.0000119.6580.01.92
 Unmatched studies excluded (n = 119)4.02 (2.14, 5.89)< 0.000012.6224.00.68
Isovolumetric relaxation time, msec (n = 670)9.67 (5.78, 13.56)< 0.0000197.6386.044.06
 Unmatched studies excluded (n = 670)9.67 (5.78, 13.56)< 0.0000197.6386.044.06
Transmitral A wave velocity, meters/second (n = 750)0.13 (0.07, 0.18)< 0.00001153.4890.00.01
 Unmatched studies excluded (n = 750)0.13 (0.07, 0.18)< 0.00001153.4890.00.01
Transmitral E wave velocity, meters/second (n = 709)−0.01 (−0.04, 0.02)0.4645.0367.00.00
 Unmatched studies excluded (n = 709)−0.01 (−0.04, 0.02)0.4645.0367.00.00
E/A ratio (n = 1,033)−0.13 (−0.28, 0.02)0.08496.4396.00.10
 Unmatched studies excluded (n = 789)−0.17 (−0.25, −0.09)< 0.00001103.9684.00.02
Mitral deceleration time, msec (n = 481)6.38 (−2.76, 15.51)0.1731.8169.0151.76
 Unmatched studies excluded (n = 481)6.38 (−2.76, 15.51)0.1731.8169.0151.76
Left ventricular EF, % (n = 1,219)−0.22 (−1.19, 0.75)0.65101.7882.03.26
 Unmatched studies excluded (n = 975)−0.01 (−0.97, 0.94)< 0.0000175.6378.02.83
Figure 3.

A, Left atrial dimension comparison between patients with rheumatoid arthritis (RA) and controls. B, Left ventricular mass index comparison between patients with RA and controls. C, Pulmonary artery pressure comparison between patients with RA and controls. D, Isovolumetric relaxation time comparison between patients with RA and controls. 95% CI = 95% confidence interval; IV = inverse variance.

Analysis with the random-effects model excluding the studies by Liang et al (21) and Crowson et al (20) was performed to adjust for the unmatched design of these studies (Table 3). This analysis revealed a lower E/A ratio (suggesting DD), with a mean difference of −0.17 (95% CI −0.25, −0.09; P < 0.00001) in RA patients. All other results remained unchanged at our level of significance, therefore further strengthening the increased DD propensity observed in RA.


Our systematic review and meta-analysis involving 5,836 patients show that patients with RA are more likely to demonstrate features of DD as manifested by increased LA dimension, prolonged IVRT, higher A wave velocity, and lower E/A ratio (due to higher A wave velocity seen in the adjusted analysis), along with increased LVMI and PAP. These findings remained unchanged on the subanalysis excluding the unmatched studies. Taken together, these features suggest a higher prevalence of DD in patients with RA. At least 21 studies showed evidence of DD based on the presence of at least one of increased deceleration time or IVRT or lower E/A ratio (Table 1). In patients with RA, prolonged IVRT is the most likely abnormality found in DD.

The finding of increased LA dimension is also of additional interest because it is a risk factor for atrial fibrillation (28). We have evidence that increased P wave dispersion, a predictor of atrial fibrillation, is seen more often in patients with RA (29, 30). Two recent studies have found increased prevalence of atrial fibrillation in RA patients (31, 32).

The finding of a significantly higher PAP in RA patients is important. There is some evidence that patients with RA have higher PAP and primary pulmonary arterial hypertension than controls, which seems to be related to disease duration and activity (33, 34). This finding has important clinical implications because the cause of fatigue or dyspnea in an RA patient may be CAD, lung disease, pulmonary hypertension, or DD. Further research is needed to study the significance of increased PAP in the RA population and to determine if routine echocardiographic monitoring in patients with RA is needed.

The preservation of EF with the development of DD suggests pathophysiologic mechanisms occurring primarily on the diastolic properties of hearts of patients with RA. One study suggesting an aberrant immune response as a contributing factor identified that moderate to severe DD in RA patients was predicted by a cytokine profile, whereas no such association was seen with mild DD (12). This study, however, lacked a control group of a non-RA population. This will be an important area of research for future studies to assess the impact of disease activity on DD, if there is any. The influence of inflammation in RA along with the other comorbid conditions, which themselves predispose to DD, needs to be explored.

Recently, there has been an interest in RV diastolic parameters in patients with heart failure with normal EF because of a potential pathogenic role (35). Three of the 4 studies reporting RV diastolic parameters suggested RVDD in RA. Another study using tissue Doppler TTE found elevated RVDD in RA patients independent of pulmonary pathology (36). The authors concluded that RVDD was suggestive of subclinical primary cardiac and pulmonary involvement in RA patients. This study was not included in our analysis due to a lack of LV data. The significance of these findings is contingent upon improved understanding of the role of RV diastolic pathology in different classes of heart failure.

The natural history of DD in RA is not well known. In one study of the general population, the 2-year probability of developing systolic CHF was 1.9%, whereas the probability of developing symptoms of dyspnea, edema, or fatigue in patients with DD was 31.1% (37). Another study in the general population showed that DD remained stable in approximately one-half of the patients, improved in 21%, and clinically deteriorated in 27% over 3.6 years of followup, and an improvement in the DD grade resulted in survival benefit (4). The finding that the rate of CHF in RA and non-RA patients remains constant over time is suggestive of a factor in addition to inflammation (1). This was shown by one study of RA patients with DD, where despite inadequate disease control at 5 years of followup, the DD status was preserved (38). Whether the rate of progression of DD in patients with RA is higher, whether DD in RA is associated with greater morbidity and mortality, and whether the severity of DD in RA has a similar prognostic value as in individuals from the general population need to be researched.

Treatment options for DD continue to evolve; data suggest that aggressive blood pressure control may improve diastolic parameters (39). However, current evidence to suggest screening for DD in the general population is lacking (40).

In the general population, the triggers of DD progression to symptomatic stages include ischemia, arrhythmia, hypertension, or hospitalization, but are unknown in approximately 50% of the cases (37). One study of healthy individuals showed that DD was predicted by C-reactive protein levels (41). Our systematic review identified an almost equal number of studies providing evidence for and against the role of disease duration and disease activity as predictors of DD (Table 1). The influence of disease activity or serology on diastology in patients with RA merits further investigation.

A TTE-based meta-analysis showed that the LVMI is increased in patients with RA (42). The increased mass was attributed to vascular stiffening secondary to inflammation (27). However, in a recent study evaluating an LV mass in RA patients with magnetic resonance imaging (MRI), lower LV mass and EF in patients with RA than in controls was identified (43). These findings were associated with anti–cyclic citrullinated peptide titers, but not with rheumatoid factor or radiographic disease. The authors speculated chronic myocarditis and hypoperfusion culminating in mass loss. Another potential cause could be amyloid deposition. Cardiac amyloid involvement in RA is much more common (28% in RA patients versus 3% in non-RA patients) than is diagnosed (44). The divergent findings of the 2 studies may be attributed to differences in baseline characteristics and the imaging techniques used. Since our data were also based on TTE parameters, more MRI-based data are needed.

Women accounted for 80% of the RA population in our study. RA is 3 times more common in women than in men (45), and evidence shows that the rate of systolic CHF is higher in women with RA as compared to men in contrast to the general population, where the incident CHF rate for women is lower (1, 46). The data on sex differences in diastolic heart failure and DD, however, are conflicting depending on the study design. Hypertension leads to DD by increasing interstitial fibrosis, disturbance of calcium homeostasis, and increased deposition of collagen (47, 48). Hypertension is more common in men; however, in the age group ≥65 years, it is notably more common in women (49). In addition, sex-based differences in symptoms perception and differences in ventricular diastolic distensibility, skeletal muscle adaptation to CHF, calcium metabolism, and the nitric oxide system may contribute to this increased incidence. The loss of the inhibitory effects of estrogen on the renin–angiotensin system may also play a role. This topic is of great interest and has been reviewed in detail (50). This is concerning, especially since cardiovascular disease is generally not pursued as aggressively in women as in men.

Our analysis included most patients without any of the known medical conditions that commonly result in DD. We did note that the prevalence of comorbid conditions was higher in the control subjects in contrast to the cases. This may be attributed to the selection bias of excluding RA subjects with these conditions. However, after exclusion of the 2 unmatched studies, these baseline differences were eliminated. More importantly, the results pertaining to TTE data remained essentially unchanged after excluding these unmatched studies. Subjects included in this analysis were mostly without the comorbid conditions that would be present in the real-world RA patients and would influence the cardiac function of RA patients. The interplay of all of these factors and the resultant outcome on DD in RA needs to be assessed further. All of the studies were of fair quality and were planned to evaluate cardiac involvement in RA. The 2 non–English language studies agreed with our general conclusion. Publication bias against negative studies cannot be ruled out, but the funnel-plot did not reveal its presence.

Our study is not without limitations. Unknown factors can result in bias from confounding. Observer bias may have been present, since most studies did not specify if echocardiography was blinded. These results are primarily applicable to women. TTE parameters can be affected transiently, e.g., by use of nonsteroidal antiinflammatory drugs (51), and may have contributed to increased DD in these studies. We should also note that the PAP data were only reported by 5 studies, therefore limiting the generalizability of this finding. In clinical practice, a diagnosis of DD is based on a combination of TTE parameters and not just one abnormal value, and the use of a singular abnormal value may have overestimated the difference. The grade of DD is of prognostic significance; however, sufficient data were lacking to determine that. Due to the presence of a significant degree of heterogeneity in included studies, we applied a random-effects model to minimize its impact on meta-analysis outcomes. This heterogeneity likely stems from different patient populations and settings, TTE performers, disease activity, and control selection. The results of this meta-analysis of aggregate data may not be generalizable to an individual patient due to a lack of patient-level data and therefore our inability to perform subgroup analyses.

In conclusion, our meta-analysis indicates that there is an increased prevalence of TTE parameters of DD in RA patients. RA patients also have a significantly larger LA dimension and higher PAP than the control population. These findings invite further longitudinal studies to assess the evolution of DD and its impact on clinical outcomes in patients with RA.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Alam had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Aslam, Bandeali, Khan, Alam.

Acquisition of data. Aslam, Bandeali, Alam.

Analysis and interpretation of data. Aslam, Bandeali, Khan, Alam.