Article Text

Download PDFPDF

Valvular heart disease
Predicting outcome after valve replacement
  1. H Rimington1,
  2. J Weinman2,
  3. J B Chambers1
  1. 1
    Cardiothoracic Centre, Guy’s and St Thomas’ NHS Trust, London, UK
  2. 2
    Section of Health Psychology, Institute of Psychiatry, Kings College London, UK
  1. Correspondence to Dr Helen Rimington, Cardiothoracic Centre, Guy’s and St Thomas’ NHS Trust, Westminster Bridge Road, London SE1 7EH, UK; helen.rimington{at}


Objective: To identify the key predictors of performance on a 6-minute walk and health-related quality of life (QOL) one year after cardiac valve replacement and to use the predictors to guide clinical practice and optimise outcome.

Design: Prospective cohort study.

Setting: Tertiary cardiothoracic centre in the UK.

Patients: 225 patients having first time valve replacement with a mean age 67.1 (SD 12.1) years.

Main outcome measures: Mortality, morbidity, NYHA, performance on a 6-minute walk and health-related QOL one year after surgery.

Results: One year after valve replacement 90% of patients were alive and free from a major event related to their surgery. NYHA category fell by 0.6. Performance on a 6-minute walk improved by 42% and QOL improved on all subscales and both composite scores of the SF-36 QOL questionnaire. Although physical QOL scores improved they did not normalise, unlike the mental QOL scores which were near normal on both occasions. Independent baseline predictors of 6-minute walk performance at one year were baseline walk performance, age and belief in surgery as a treatment. Independent baseline predictors of one year physical QOL were baseline physical QOL and walk performance. Independent baseline predictors of one year mental QOL were depression, baseline mental QOL and age, with age having a positive effect.

Conclusions: One year after valve replacement patients can expect a significant improvement in their exercise tolerance and QOL but their physical QOL is unlikely to be normal. Outcome may be improved by treating depression and modifying negative illness beliefs preoperatively.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Valve replacement surgery is becoming safer and improves longevity1 and NYHA status.2 However functional outcomes still vary and symptom relief may not automatically translate into an improved quality of life (QOL), particularly if new health issues arise. These can include stroke, post cardiopulmonary bypass neurocognitive dysfunction and the secondary effects of anticoagulation regimens. Internationally accepted guidelines allow operating on asymptomatic patients to prevent irreversible myocardial damage3 but as Cheitlin states “it is never possible to make a really asymptomatic patient feel better through surgery”.4

Ideally, we should be able to advise patients about their likely postoperative functional status in terms of a measure of their ability to exercise and also their likely QOL. There are many potential predictors of such outcomes including preoperative ventricular function,5 the aetiology of the valve disease,6 preoperative symptoms,2 postoperative neurocognitive dysfunction7 8 and depression.9 10 However, the relative importance of these determinants and the interactions between them are uncertain. Patients’ beliefs about their illness may affect both their mood and QOL,11 although this has not been studied in valve replacement.

The aims of this study were, first, to assess performance on a 6-minute walk and quality of life one year after surgery and, second, to determine their preoperative predictors.



Between January 2003 and June 2005, 225 patients scheduled for first time valve replacement were enrolled prospectively (table 1). Ninety-nine (44%) had concomitant coronary artery bypass grafting (CABG). Biological valves were implanted in 146 (65%) patients and mechanical replacements in 79 (35%). The mean age was 67.1 (range 26–89) years and 155 (69%) were male and 70 (31%) female, 111 were 70 years or older. All patients gave written informed consent as directed by St Thomas’ Hospital research ethics committee.

Table 1

Operations performed


All 225 patients were studied preoperatively and 204 were restudied after one year (15 died, four withdrew consent and two were lost to follow-up). The study consisted of echocardiography, neurocognitive testing, a 6-minute walk and questionnaires to measure quality of life, anxiety and depression, and illness perceptions. The patients were tested on two occasions: preoperatively during their admission for surgery and one year after surgery as an outpatient. At the one year visit they were asked about the type and duration of rehabilitation programme attended.

Transthoracic echocardiography

Echocardiography was performed using an ATL HDI 5000 (Seattle, WA, USA) or a GE VingMed System Five (Horten, Norway). The left ventricular ejection fraction was estimated semiquantitatively.12 Left ventricular long axis systolic motion was recorded in the four-chamber view using M-mode at the lateral side of the mitral valve annulus. The maximum excursion was measured from nadir to peak after the Q wave on the ECG. Pulsed Doppler was used to measure the early diastolic (E) wave of the mitral filling pattern. The tissue Doppler early diastolic (E′) and systolic velocity were measured at the lateral side of the mitral valve annulus.13 Measurements were averaged over three beats in sinus rhythm and five beats in atrial fibrillation. The E/E′ ratio was calculated as an estimate of left ventricular filling pressure.14 Pulmonary artery pressure was estimated using the peak velocity of the tricuspid regurgitant jet and the inferior vena cava size and reactivity.

Neurocognitive tests

Three neurocognitive tests were performed as recommended in the consensus statement from the central nervous system dysfunction after cardiac surgery conference.15

  1. The trail making tests A and B assess visual scanning, motor skills incorporating speed and agility, hand to eye coordination and attention and the ability to swap between two sets of stimuli.

  2. The Rey auditory verbal learning test (AVLT) measures immediate memory span, serial learning and learning following distraction.

  3. The grooved pegboard (Lafayette Instrument Company, Model 32025, Lafayette, IN, USA) measures fine-motor control, psychomotor speed and rapid visual-motor coordination, representing a combined cognitive-motor task.

Six-minute walk

The 6-minute walk test (6MWT) is simple, safe16 and well accepted by patients.17 It is more representative of everyday activity than a treadmill or bicycle test and is less intimidating to unwell or elderly individuals because they control the intensity of the exercise. It is suitable for patients with a range of comorbidities16 and can be performed in any unobstructed corridor.

In this study patients were excluded only if they were unable to walk unsupported or were on clinician-prescribed bed rest. There were no data to guide the use of the test in these patients so the protocol was constructed conservatively. Patients were accompanied during the test, ensuring that they dictated the pace. Those known to be moderately or severely symptomatic (NYHA 3 or 4) were exercised in a corridor within a cardiac ward area with resuscitation facilities available. All other tests were performed in a quiet hospital corridor. A stopwatch was used to time the test and a measuring wheel (Measuremeter, Trumeter, UK) quantified distance walked. Usually, the patient propelled the measuring wheel; however if they were unable to—for example, they used a walking stick, it was wheeled for them, tracking their path. Distance walked in 6 minutes was recorded regardless of pauses or early termination of the test.

Before each test patients were advised that the aim was to “measure what represents a normal amount of exercise for you at the moment” and that they could change speed or pause as required. They were asked to report symptoms as they occurred. Verbal encouragement was not given because it has been shown to influence the distance walked.18

After 30 seconds the patients were asked if they were comfortable and whether they wrere experiencing any symptoms. This inquiry was repeated at 3, 4 and 5 minutes into the test. If they reported a symptom they were asked what they would normally do when such a symptom occurred and the test continued or stopped accordingly. If they were prepared to continue, they were asked about the severity of the symptom every minute and at the end of the test.


Three questionnaires were administered using well established standardised techniques.

  1. The SF-3619 measures health-related QOL across eight domains: physical functioning, role limitations because of physical health problems, bodily pain, general health, vitality, social functioning, role limitations because of emotional problems and mental health. The eight health domains are summated into physical component summary (PCS) and mental component summary (MCS) scores. Higher scores represent better health status.

  2. The Hospital Anxiety Depression Scale (HADS)20 is a 14-item questionnaire that provides separate scores for anxiety and depression. Higher scores represent higher levels of mood disturbance.

  3. The Revised Illness Perception Questionnaire (IPQ-R)21 is a measure of how the patients perceive their illness. It has eight subscales: identity (the symptoms the patient ascribes to their illness), timeline (perceived illness duration), consequences (the effects on their life), personal control (how much influence they feel they have), treatment control (how well they rate their treatment efficacy), coherence (how well they understand their illness), timeline cyclical (the extent to which their illness is perceived as cyclical or permanent) and emotional representations (the emotional impact of the illness). High scores represent more negative beliefs about the controllability or effects of the illness (less control or more effects). High coherence scores represent poor understanding of the illness.

Statistical analysis

Analysis was performed using SPSS for windows version 11.5.1. Non-normal data were transformed by taking the logarithm10. Comparison of means for normal data was performed using paired or unpaired t tests as appropriate. Pearson’s χ2 was used for frequency data in independent categories. Correlation between two variables was tested using a Pearson correlation test. Multiple linear regression analysis was performed stepwise using a significance level of 0.05. Bonferroni corrections were applied so that variables were only entered into multiple regression analysis if they remained significant after dividing 0.05 by the number of variables tested on bivariate analysis. Although strictly a categorical variable NYHA category fulfilled statistical criteria for normality and was treated accordingly. The data are presented as means (SD).


Death, morbidity and NYHA class

There were six (2.7%) early (before 30 days) and nine (4%) late deaths (30 days to one year). Patients who died were older than survivors at the time of surgery (76.1 (12.1) vs 66.5 (9.3) years, p = 0.003), had shorter 6-minute walk distances (222 metres (124) vs 300 (142) metres, p = 0.04), lower preoperative left ventricular ejection fraction (50.0% (20.9%) vs 60.4% (14.1%), p = 0.008), but similar NYHA category (mean 2.1 for both, NS). There were no cases of structural valvular deterioration, non-structural dysfunction, endocarditis or valve thrombosis although there was one reoperation for a homograft in the aortic position because of coronary artery compression. There were six perioperative strokes and, in the first year, five transient ischaemic attacks, one reversible ischaemic neurological deficit and four additional non-fatal strokes. There were three non-fatal bleeding events and 24 minor bleeding events mostly nose-bleeds. The mean NYHA category fell from 2.16 (0.66) preoperatively to 1.58 (0.64) at one year (p<0.0005).


Ejection fraction was 60.0% (14.9%) preoperatively and 61.4% (10.7%) at one year (NS). However Doppler tissue peak lateral systolic velocity increased significantly from 7.1 (2.1) cm/s preoperatively to 8.9 (2.3) cm/s at one year (p<0.0005) and lateral long-axis excursion also increased from 1.2 (0.3) cm preoperatively to 1.4 (0.3) cm at one year (p<0.0005). The lateral E/E′ ratio fell from 13.8 (8.5) preoperatively to 9.8 (4.6) at one year (p<0.0005). Preoperatively, 39 (17%) patients had a raised pulmonary artery pressure, compared with 10 (6%) at one year (p<0.0005).

Neurocognitive tests

For patients who completed the tests on both occasions there was a 2-second or shorter difference between preoperative and one year trail and pegboard times (all NS) and a two-point reduction in the Rey AVLT score (p = 0.01, borderline significance after Bonferroni adjustment). There were moderate negative correlations between performance on all neurocognitive tests and age (r values between 0.2 and 0.4).

Six-minute walk

There were no adverse incidents during the walk tests. The mean distance walked increased from 294 (142) metres preoperatively to 418 (119) metres at one year (p<0.0005). Of the 169 patients who performed the walk on both occasions 145 (86%) walked further at one year and 121 (72%) walked at least 50 metres further.

Patients aged 70 years or more walked 261 (139) metres preoperatively and 374 (127) metres at one year (113 metres improvement) compared with patients under 70 who walked 328 (135) metres and 446 (117) metres, respectively (118 metres improvement).

Distance walked at one year was not affected by the preoperative valve pathology for patients undergoing aortic valve replacement. There were too few patients in the mitral, tricuspid and pulmonary replacement groups for analysis. Concomitant CABG did not significantly influence distance walked at one year. The distance for patients who had valve surgery with CABG was 403 (118) metres and without was 428 (119) metres (NS).

Performance at one year correlated with age, gender and preoperative walk distance, NYHA category, neurocognitive performance, both IPQ-R control subscales, IPQ-R coherence and depression. Preoperative left ventricular systolic and diastolic function were weakly correlated with one year 6-minute walk performance, but did not withstand Bonferroni correction (table 2).

Table 2

Bivariate correlations between baseline variables and 6-minute walk performance at one year

A stepwise multivariate analysis showed an independent effect only from preoperative walk distance, age and the preoperative treatment control subscale of the IPQ-R. These variables explained 44% of the variance. Preoperative walk distance contributed 36.5%, with age adding 4.6% and IPQ-R treatment control a further 3.0% (overall p value <0.0005).

Quality of life

All domains of the SF-36 questionnaire improved after surgery (table 3). For simplicity, the physical (PCS) and mental (MCS) summary scores alone were used for subsequent analysis.

Table 3

Mean (SD) 6-minute walk distance and mean (SD) SF-36 domain and summary scores preoperatively and at one year

The PCS at one year correlated with preoperative PCS and MCS and with preoperative walk distance. One year PCS was inversely correlated with NYHA, with the consequences, coherence and emotional representations subscales of the IPQ-R, and with anxiety and depression (table 4). Only preoperative PCS and preoperative walk distance were independently correlated with one year PCS on multiple regression analysis. Preoperative PCS contributed 30% of the variance and walk distance a further 5% (p<0.0005).

Table 4

Bivariate correlations between baseline variables and QOL at one year

The one year MCS had a direct relation with preoperative MCS and age. It was inversely related to the consequences and timeline cyclical illness perception subscales and with anxiety and depression preoperatively (table 4). Multiple regression analysis showed that 36% of the variance in MCS at one year was independently explained by baseline depression (29%), baseline MCS (5%) and age (2%) (p<0.0005).


Patients should have been offered a place on a local “phase 3” rehabilitation programme starting just after their routine 6-week postoperative outpatient visit. In this study 121 patients attended a rehabilitation programme, 79 did not (unknown for four). There was no difference in age or gender between attenders and non-attenders. Attenders walked further on their 6-minute walk at one year (430 (106) metres versus 393 (143) metres), but the difference was not statistically significant (p = 0.07). Of the 79 non-attenders, 39 (49%) chose not to and nine (11%) were unable for medical and five (6%) for non-medical reasons. In addition, 23 (29%) did not receive an appointment and three (4%) received it too late in the recovery process.

Patients who attended phase 4 rehabilitation or a gym at least once a week (n = 43), so were involved in an ongoing exercise programme after phase 3, walked significantly further at one year than those who did not attend any rehabilitation or only attended a phase 3 (459 (87) metres versus 407 (129) metres, p = 0.005 t test). They also had higher scores on the one year PCS (42.5 (10.6) vs 37.7 (14.0), p = 0.04 t test) but they had no difference in scores for one year MCS, HADS anxiety or depression scores.


Valve replacement produced significant improvements in both 6-minute walk distance and in all measures of QOL (table 3) for the whole population. However there were wide variations in individual responses to surgery. Walk distance and QOL correlated better with psychological and neurocognitive factors than with left ventricular function.

Walk distance at one year was independently correlated with only preoperative walk distance, age and the baseline treatment control subscale of the IPQ-R. On bivariate analysis it was moderately correlated with preoperative NYHA class, but correlations with four of the five neurocognitive scores and the coherence subscale of the IPQ-R were stronger. By contrast preoperative left ventricular systolic and diastolic function were only weakly correlated with walk distance and the relation appeared mainly through longitudinal ventricular systolic and diastolic function because overall the ejection fraction did not change. Few patients had abnormal ejection fractions at the time of surgery which reflects appropriate timing of surgery for the group; however, more subtle indices of systolic function that were low-normal preoperatively improved after surgery and the filling pressure which was abnormal almost normalised over the follow-up period. Thus, even when patients are operated at the correct time according to guidelines, there is still potential for improvement in left ventricular function and thereby on the influence of physical QOL indirectly.

Little previously published work exists, but Weuve et al22 also showed that physical activity was associated with better cognitive function in elderly females and it is possible that depression is the link between impaired neurocognitive performance and QOL.23

Our results showed improvement in all aspects of QOL. However, although physical QOL score (PCS) improved by 7 points to 38.1, it did not return to normal (a score of 50). The mean score improvements were at or above the minimal clinically important difference (MCID), which is an improvement that the patient will have noticed,24 for all domains except the mental health score and the mental component summary score which were borderline.

On bivariate analysis physical QOL at one year was related to baseline exercise performance, symptoms, two of the neurocognitive test scores, the effects of the illness on the patient’s life, their understanding of their illness and the emotional upset it caused and to both anxiety and depression. Age was not a significant factor in postoperative physical QOL which confirms the work of other researchers.2 25 Mental QOL changed much less than physical QOL. The group mean for mental QOL was just below normal before surgery and just above after surgery. It strongly correlated with both baseline anxiety and depression while age had a protective effect—elderly patients tended to have better mental QOL than younger patients and there were correlations with three of the IPQ-R subscales: consequences, timeline cyclical and emotional representations. Multivariate analysis showed that, in addition to baseline QOL score, baseline walk performance independently predicted physical QOL and baseline depression and age independently predicted mental QOL.

NYHA is not a sufficiently sensitive measure of change for informing clinical advice to these patients or for research studies. We found a group improvement in NYHA class of less than one category and, furthermore, 13% of patients were asymptomatic before surgery. There was a 42% increase in 6-minute walk distance one year after valve replacement. The 6-minute walk distance is a better way of monitoring change. The correlation with NYHA class was moderate at one year (r = −0.4) which is in agreement with Demers et al’s study of patients with stable heart failure.26

Overall the group’s neurocognitive results were similar preoperatively and at one year. This is in agreement with some previous studies,25 26 while others showed a decline in neurocognitive function7 27 but the role of other risk factors for neurocognitive decline in these populations is still to be evaluated. A few of our patients spontaneously reported neurocognitive problems (for example, with reading, crosswords or sewing) which resolved over time. Overall this represents a reassuring message for patients about to experience valve surgery: if an individual performs well preoperatively they are still likely to perform well at one year.

It would be useful to be able to identify patients likely to do badly after surgery. Our results show that elderly patients with poor performance on a 6-minute walk before surgery and with weak beliefs that their surgery will treat their valve disease are most at risk of poor walk performance at one year. Baseline physical QOL in conjunction with baseline exercise performance predict one year physical QOL, while younger patients who are depressed preoperatively with poor baseline mental QOL are most likely to have poor one year mental QOL. About 10% of patients reported a worse QOL after surgery, but they had had an adverse event as a result of the surgery (including perioperative stroke), had significant comorbidity or were depressed.


The majority of our patients (90%) had AVR or AVR with CABG. T tests comparing these patients with those having other types of valve surgery found no significant differences in QOL, anxiety and depression or walk distance at one year. However, our results are likely to be most applicable to patients having AVR and could be less relevant in other types of valve surgery. We used a published MCID for the MCS scores, but there is no final consensus on what constitutes a clinically significant score change for the SF-36.28 29 30

Clinical implications

Our results could support the decision not to offer surgery to patients at high operative risk if they also had features suggesting a poor outcome in terms of QOL. Patients can be informed that while they can expect to notice an improvement in their physical QOL, it is unlikely to normalise. However, it is possible to advise most elderly patients that, despite higher operative risks, their postoperative physical QOL is likely to be equivalent to that of younger patients. Most patients having elective valve replacement spend a few weeks on a surgical waiting list, which could afford valuable time to optimise postoperative outcome by a preoperative intervention such as treating depression or modifying negative illness perceptions. While attendance at a short rehabilitation course is likely to be beneficial, it is regular exercise that produces a significant increase in exercise ability and physical QOL. Patients who revert to their usual pattern of exercise after they have “been to rehab” need encouragement to make long-term lifestyle changes.


The conventional outcome measures of mortality and morbidity have served well in the past when valve replacement surgery was high risk and they remain of value for documenting the most negative outcomes. However, in order to give the majority of patients a more realistic idea of what to expect after surgery and to enable their outcomes to be optimised, other types of assessment are needed. Exercise performance and QOL offer this.

Advice given to patients before valve replacement should include individualised information about likely postoperative QOL and exercise tolerance in addition to their risk of death or of an adverse event in the perioperative period.



  • See Editorial, p 97

  • Funding HR received a Research Training Award from Guy’s and St Thomas’ Charity.

  • Competing interests None.

  • Provenance and Peer review Not commissioned; externally peer reviewed.

Linked Articles