5.1.1. Clinical introduction

People with non ST-egment elevation ACS have a high incidence of recurrent myocardial ischaemia, a similar long term outcome to those with ST elevation MI (STEMI), and a worse outcome than for people with UA¹⁷⁵. A variety of drug (anti–platelet, anti–thrombin) and coronary revascularisation (PCI or CABG) treatment strategies have been investigated for their potential to reduce the frequency of adverse events (death, MI, recurrent myocardial ischaemia).

However, for PCI or CABG to be considered as treatment options, coronary angiography has to be undertaken first to define the extent and severity of the person's coronary disease. Angiography is an invasive procedure, often requiring further anticoagulation, and therefore potentially has some associated risk. This, together with improving drug therapy, has caused investigators to address whether angiography/revascularisation should be performed, and if so, when in the course of an individual's admission it is best undertaken. Angiography may be undertaken early, deferred until later, or undertaken selectively only if the person has evidence of recurrent ischaemia despite appropriate drug therapy.

Supporters of an early invasive strategy reason that the sooner the coronary anatomy can be imaged, the sooner appropriate therapy (including revascularisation) can be given; thereby avoiding lengthy hospital stays and preventing further events ¹⁷⁶. On the other hand, supporters of a conservative management strategy (involving initial antithrombotic and anti-anginal treatment, and angiography performed only if there is evidence of recurrent ischemia) reason that medical therapy can stabilise people and non-invasive stress testing can identify those who require angiography; thereby reducing costs and complications by using angiography more selectively ¹⁷⁶.

A number of clinical trials have been undertaken, but comparison between them is complicated by the:

• era in which they were undertaken (earlier trials involved less aggressive drug therapy and often had a low use of intracoronary stents),
• different time scales used in which angiography could be undertaken,
• frequency of angiography and revascularisation procedures in the conservative arms of the trials, and the
• varying definitions of MI.

In 2007 (the last available year) a total of 77,373 PCI procedures were undertaken in the UK, of which 40.5% were for UA or NSTEMI, and the stent usage overall was 94.7%¹⁷⁷. The use of glycoprotein IIb/IIIa inhibitors (GPIIbIIIa) for people with UA or NSTEMI was 27% and 39% respectively ¹⁷⁸. Thus, in order to provide evidence close to modern day practice older trials where there was a low use of intracoronary stenting were excluded from our analysis. A separate specific analysis was made of those trials reporting on the use of GPIs.

The clinical question posed, and upon which the literature was searched, was:

“In adults with UA or non-ST segment elevation MI does early invasive investigation (i.e. angiography) with intent to assess for (and in those patients deemed suitable, to perform) revascularization improve outcomes in comparison with initial conservative treatment, with or without later angiography?”

5.1.2. Clinical methodological introduction

The literature was searched from 1995 to 2009 for systematic reviews, RCTs, comparative studies, and observational studies comparing conservative management with early invasive management in people with non ST-segment elevation ACS. RCTs were included if they reported on either short (index hospitalisation) or long-term (up to 5 years) outcomes including death, MI, bleeding, stroke, re–hospitalisation.

Four systematic reviews ^{176 179 180,181}, one meta–analysis ³⁹ (an update of the Mehta meta–analysis) and two reports from open RCTs ^{182 183} analysed the effect of an invasive versus conservative approach on death, nonfatal MI (procedural or non-procedural), quality of life, rehospitalisation, bleeding, and stroke.

The Hoenig et al. systematic review included 5 open RCTs (N=7818) in the stenting eral^{178 q} (FRISC II ^{184,185 186,187}, TACTICS-TIMI 18 ¹⁸⁸, VINO ¹⁸⁹, RITA-3 ^{190 191}, ICTUS ^{192 193}). Three analyses were performed pooling trials based on the use of GPIs (stents with GPI use, stents without GPI use, and stents regardless of GPI use). Subgroup analyses were performed according to gender, troponin levels, risk stratification, and ST depression ¹⁷⁶.

The Qayyum et al. systematic review included ten open RCTs (N= 10648; TIMI IIIB ¹⁹⁴, MATE¹⁹⁵, FRISC II ^{184,185 186,187}, TACTICS-TIMI 18 ¹⁸⁸, VINO ¹⁸⁹, RITA-3 ^{190 191}, ICTUS ^{192 193} VANQWISH ¹⁹⁶, NQWMI ¹⁹⁷, and TRUCS ¹⁹⁸). This meta–analysis was excluded from our analysis as it included three RCTs (VANQWISH ¹⁹⁶, TIMI IIIB ¹⁹⁴, MATE¹⁹⁵) that were conducted before the routine use of stents. Also, the inclusion of the TRUCS ¹⁹⁸ RCT was controversial because the patient population (Braunwald class IIIb or IIIc UA) was randomised 48 hours after the index episode of myocardial ischaemia and following a period of stabilization on medical therapy. Thus, these people were managed conservatively for at least 48 hours, making this trial different from the other trials ¹⁷⁹.

The Mehta et al systematic review included 7 open RCTs (N=9208 ; VANQWISH ¹⁹⁶, TIMI IIIB ¹⁹⁴, MATE¹⁹⁵, FRISC II ^{184,185 186,187}, TACTICS-TIMI 18 ¹⁸⁸, VINO ¹⁸⁹, RITA-3 ^{190 191}) and was also excluded from our analysis because pre–stenting era trials were included, and it lacked the ICTUS trial¹⁸⁰.

The O'Donoghue et al. systematic review compared an early invasive strategy with a conservative strategy in men separately from women (8 RCTs; N total = 10412; N women = 3075; N men = 7075). The NCC–CC performed a modified meta–analysis by excluding three pre-stent trials (VANQWISH, TIMI IIIB, and MATE), which also examined the impact of gender on the comparison of invasive and conservative strategies (5 RCTs; FRISC II ^{184,185 186,187}, TACTICS-TIMI 18 ¹⁸⁸, VINO ¹⁸⁹, RITA-3 ^{190 191}, ICTUS ^{192 193}) ¹⁸¹.

The Henriksson et al. meta–analysis was an update of the Mehta et al. meta-analysis. The ICTUS trial and the five year follow-up data from FRISC-II were added. This study was included for consideration by the GDG as it was used in the Henriksson et al. Cost-effectiveness analysis included as economic evidence, although the meta–analysis lacked a rigorous literature search and there was no quality appraisal of the individual trials.

Finally, two open RCTs were appraised that reported quality of life outcomes from the FRISC-II trial¹⁸³ (N=2457; follow-up at 3, 6, 12 months) and the RITA-3 trial ¹⁸² (N=1810; follow-up at 4 and 12 months). Both trials used validated standardised questionnaires to evaluate quality of life in people randomised to routine invasive versus conservative management strategies.

When considering the evidence it is important to consider the heterogeneity in the studies in terms of patient populations, different definitions of MI, different rates of revascularisation (both within each study arm as well as across the different studies), different stent use (stent use was low in older trials), different pharmacological backgrounds (particularly use of GPIs during PCI), and different mortality rates. All the RCTs that randomised people to routine invasive versus conservative management strategies were open due to the nature of the invasive approach.

Two summary tables (see Table 5-1 and Table 5-2) of the characteristics of the trials with stenting during PCI are presented (adapted from Qayyum et al).

Table 5-1

Summary of characteristics of trials comparing early invasive with conservative management strategies in the stenting era.

Table 5-2

Summary of trial characteristics.

In the invasive strategy, (by protocol) time from admission/index pain to randomisation ranged from one to three days and time from randomisation to angiography ranged from four hours to a ‘few days’.

The actual time from randomisation to angiography in the trials ranged from an ‘average’ of 6.2 hours to median of four days and the actual time from randomisation to PCI ranged from an ‘average’ 8.6 hours to a median of four days (in those who underwent PCI).

5.1.3. Clinical evidence statements

Refer to Table 5-3 for a summary of the results from the meta–analyses.

Table 5-3

Summary of outcomes in index hospitalisation: Invasive versus conservative management strategies.

NCC–CC meta–analysis

The NCC–CC conducted a meta–analysis of RCTs with high stent use (range from 50% to 93%) [FRISC II ^{184,185 186,187}, TACTICS-TIMI 18 ¹⁸⁸, VINO ¹⁸⁹, RITA-3 ^{190 191}, ICTUS ^{192 193}]. The four year results of the ICTUS trial and the five year results of FRISC II were used to update the Hoenig et al. meta–analysis. Outcomes were death, MI, or composites of death or MI, and death, MI, or hospitalisation. Effect sizes were reported as relative risks with a random effects model. Interstudy heterogeneity was assessed with the I² statistic.

The NCC–CC used three strategies for conducting the meta–analysis:

• All five RCTs from randomisation to maximum follow-up.
• Three RCTs (ICTUS, FRISC II, RITA-3) from randomisation to maximum follow—up for studies that reported > 1 year follow-up. This was done to update the “late” (> two year follow-up) data in the Hoenig meta–analysis.
• All five RCTs from post-discharge period to maximum follow-up for the outcome of death or MI for the health economics analysis. Note that the index events were not reported in the original published studies. The index events reported in the Hoenig meta–analysis (and the Qayyum et al. meta–analysis for the ICTUS index data only) were subtracted from the entire follow-up events to calculate post-discharge to maximum follow-up outcomes.

Death: Randomisation to maximum follow-up

The NCC–CC meta–analysis of five RCTs (analysis 1) showed a non-significant difference between in randomised to an invasive versus a conservative approach for the risk for death. Results were similar when trials were grouped by the difference in revascularisation procedures between the two arms (either > or < than 50% difference in revascularisation rates). See Figure 4-1.

Figure 5-1

Analysis 1.

MI: Randomisation to maximum follow-up

The NCC–CC meta–analysis (Analysis 2; 5 RCTs) showed no significant difference between those randomised to an invasive versus a conservative approach for the risk of MI at long-term follow-up, however this analysis had significant heterogeneity. An invasive strategy significantly decreased the risk of MI in trials in which there was > 50% difference in revascularisation rates between the two arms (3 RCTs; RR 0.73 [95% CI 0.54 to 0.97], p=0.03). See Figure 5-2.

Figure 5-2

Analysis 2.

Death or MI: Randomisation to maximum follow-up

The NCC–CC meta–analysis (5 RCTs; Analysis 3) showed no significant difference between those randomised to an invasive versus a conservative approach for the risk for death or MI at long-term follow-up, however this analysis had significant heterogeneity. An invasive strategy significantly decreased the risk of death or MI in trials in which there was > 50% difference in revascularisation rates between the two arms (3 RCTs; RR 0.78 [95% CI 0.63 to 0.97], p=0.02). There was NS difference between groups for the risk of death or MI in trials with greater than one year follow-up data (FRISC-II, RITA-3, ICTUS), however this analysis had significant heterogeneity. See Figure 5-3.

Figure 5-3

Analysis 3.

Death, MI, or re-hospitalisation: Randomisation to maximum follow-up

The NCC–CC meta–analysis (Analysis 4; 2 RCTs) showed no significant difference between those randomised to an invasive versus a conservative approach for the risk for death, MI, or rehospitalisation at long-term follow-up, however this analysis had significant heterogeneity. See Figure 5-4.

Figure 5-4

Analysis 4.

Update of Hoenig meta–analysis

To update the Hoenig meta–analysis, a meta–analysis was conducted by the NCC-CC on the three RCTs with follow-up greater than one year (Analysis 5: 5 year results from RITA-3, and FRISC II, and 4 year results from ICTUS). There was no significant difference between those randomised to an invasive versus a conservative approach for the risk of death. See Figure 5-5.

Figure 5-5

Analysis 5.

Death or MI: Post-discharge to maximum follow-up

A meta–analysis was conducted for death or MI post—hospital discharge to maximum follow-up to compare with the Henriksson meta–analysis (that is used in the Henriksson et al. cost– effectiveness analysis included as economic evidence) (Analysis 6). The pre—stent trials (TIMI IIIB, VANQWISH, MATE) were excluded and updated with the long-term follow-up (three years) of ICTUS. None of the original papers reported events in the index hospitalisation. The index events were extracted from the Hoenig and Mehta meta—analyses (both agreed). However, the ICTUS index event data was only reported in the Qayyum meta–analysis, and the reviewers could not see how these numbers were obtained. Index death or MI for the ICTUS trial were obtained from Henriksson who had a personal communication from R. de Winter of the ICTUS trial. See figure Figure 5-6.

Figure 5-6

Analysis 6.

The NCC–CC meta–analysis showed that an invasive strategy significantly reduced chances of death or MI post-hospital discharge [OR 0.71 (95% CI 0.53 to 0.95), p=0.02], however there was significant heterogeneity in this analysis. The Henriksson meta–analysis reported cardiovascular death and MI post-hospital discharge (or death and MI, if there was no data) and the pooled estimate was similar to ours at OR 0.69 (95% CI 0.54 – 0.88). See Figure 5-7 and Figure 5-8.

Figure 5-7

Invasive versus conservative strategy in men (modified from O'Donoghue et al.).

Figure 5-8

Invasive versus conservative strategy in women (modified from O'Donoghue et al).

Summary of outcomes with long-term follow-up

See Table 5-4.

Table 5-4

Summary of Outcomes with Long-term follow-up: Invasive versus conservative management strategies.

5.1.4. Health economic methodological introduction

One relevant cost–effectiveness analysis from a UK perspective was identified ^38,39,199. In addition, one from a Swedish perspective²⁰⁰ and three from a US perspective ^201-203 were identified but not reviewed due to the availability of a directly applicable UK study with only minor limitations.

Henriksson et al.^38,39 reported a cost–utility analysis undertaken from a UK NHS perspective based on effectiveness and resource use data from the five year follow-up of the RITA-3 trial (UK based, n=1810), with a sensitivity analysis where effectiveness data was based on a meta– analysis of all trials in the area. A decision-analytic model was used comprising a short-term decision tree representing the index hospitalisation followed by a Markov model representing the post-index period. The analysis takes into account death, MI, quality of life (EQ5D) and resource use based on data from RITA-3. Relative treatment effect of an early invasive strategy over a conservative strategy was assumed to last only to five years in line with available follow-up in RITA-3 but the impact of alternative assumptions was assessed. Lifetime costs (£ 2003/2004 prices) and QALYs were estimated and stratified by risk. A multivariate predictive model for MI or death in RITA-3 was used to calculate a risk score defining quartiles of risk, with the 4^th quartile subdivided into two groups due to the much higher event rate in the top quartile (risk groups: 1, 2, 3, 4a, 4b).

The primary results of the cost–effectiveness analysis were based on the characteristics of people with the median risk score in each of these five risk groups. Cost effectiveness was expressed in terms of cost per QALY gained. Probabilistic sensitivity analysis was used to evaluate uncertainty. The basecase analysis assumed that the relative effect of an early invasive strategy compared to a conservative strategy was constant across risk groups, but a post hoc analysis of RITA-3 suggested that there was an interaction between treatment effect and risk group. Although the interaction was not statistically significant an alternative analysis was undertaken in which the relative benefit of the early invasive strategy varied with risk group. In another sensitivity analysis pooled effectiveness data were used from a published meta– analysis by Metha et al.¹⁸⁰, which was updated to include results from the ICTUS¹⁹² trial, and the long-term results from RITA-3¹⁹¹ and FRISC-II ¹⁸⁷.

The main potential limitation of the cost–effectiveness analysis is that RITA-3 enrolled 1997-2001 and so may not reflect current practice. Additionally the pooled effectiveness data analysis used in the sensitivity analysis included results from trials where stenting was largely not used (specifically TIMI IIB, VANQWISH and MATE) and does not include all the clinical data identified in the literature review for this guideline.

5.1.5. Health economic evidence statements

Henriksson et al.^38,39 found that an early invasive strategy, compared to a conservative strategy, was generally increasingly cost–effective as risk increased and reported cost–effectiveness ratios of £53,760, £22,949, £21,325, £11,957, £12,750 per QALY gained for risk groups 1, 2, 3, 4a and 4b respectively (1 = lowest and 4b = highest risk).

Allowing the relative treatment effect to vary by risk group improved cost effectiveness in the risk groups 4a and 4b while reducing it in risk groups 1, 2 and 3. Cost effectiveness was also considerably impacted by variations in the assumption regarding duration of treatment effect: assuming that treatment effect was maintained beyond the observed trial follow-up of five years improved cost–effectiveness. Using effectiveness inputs from pooled data instead of from only the RITA-3 trial had a modest impact in terms of reducing cost–effectiveness.

Full results for the basecase analysis and selected alternative scenarios are summarised in Table 5-5 below.

Table 5-5

Mean incremental cost–effectiveness ratio for an early invasive strategy compared to a conservative strategy (% of simulations cost–effective at a threshold of £20,000/£30,000).

Impact of changes in current practice

The main potential limitation of the study is that RITA-3 enrolled 1997-2001 and so may not reflect current practice. Table 5-6 below summarises the key changes in practice identified by the GDG and their potential impact on the cost effectiveness estimates from the Henriksson et al. study.

Table 5-6

Changes in practice and impact on Henriksson cost effectiveness estimates.

Impact of pooled effectiveness estimate excluding pre-stent trials

The Henriksson analysis uses effectiveness data from the RITA-3 trial in the base case analysis but also investigates the impact of using pooled data. The meta–analysis used included trials in the pre-stent era, which were judged not relevant to current practice by the GDG (specifically TIMI IIB, VANQWISH and MATE). Comparable pooled estimates that excluded pre-stent trials and included all relevant published data were generated as part of the clinical review.

Comparing these numbers to the pooled estimates used by Henriksson show that the relative effect in the index hospitalisation is improved and in the post-discharge period is similar although slightly worsened (see Table 5-7 below for figures). As these effects are acting in different directions it is difficult to judge the net impact. In the original analysis using the pooled analysis instead of RITA-3 had a modest impact.

Table 5-7

Comparison of composite endpoints of MI or CV death for early invasive versus initial conservative strategy.

5.1.6. Evidence summary

• When all five trials were included and analysed to the end of the index hospital admission there was no significant overall difference between the invasive and conservative group with respect to death, stroke or non-fatal MI, but an invasive strategy increased the risk of bleeding (mainly minor). However, an invasive strategy significantly decreased the composite of death and MI at 6-12 months follow-up, both late (>2 yrs) death and late MI, and reduced the long-term rate of re-hospitalisation. Procedure- related MI was significantly increased in the invasive arm (the denominator in both arms was the total number of people randomised to each arm).
• There was no difference in mortality at any time whether angiography was undertaken very early (<24 hours from randomisation - ICTUS, TACTICS-TIMI 18, VINO) or when undertaken later (>48 hours - RITA-3, FRISC-II).

The NCC–CC meta–analysis analysed the five RCTs from randomisation to end of maximum follow-up (5 years in RITA-3 and FRISC II, 4 years in ICTUS, 0.5 years in VINO and TACTICS-TIMI 18). Overall, there was a non-significant difference between an early invasive and conservative strategy for death, death or nonfatal MI, or MI. An early invasive strategy significantly reduced MI and death or MI in trials in which there was a greater than 50% difference in revascularisation rates between the trial arms.

• When analysis was undertaken of those trials not involving the routine use of GPIIbIIIa inhibitors (VINO, RITA-3, FRISC-II -use of GPIs ranged from 0-10% in these trials; compared to 94% use in TACTICS-TIMI 18 and ICTUS) an invasive strategy significantly decreased intermediate (6-12 months) MI and refractory angina, but not death at any time point, nor the index admission MI.
• In the FRISC-II trial, an invasive strategy significantly reduced the composite of death or non-fatal MI in those with either ST depression or troponin elevation (higher risk), but not in those without (lower risk), suggesting that the benefit of an invasive strategy was mostly in higher risk people. The FRISC investigators used a risk scoring system (scores 0-7) and showed worsening outcome (death, recurrent MI) as the score increased but greater benefit form the invasive stratgey²⁰⁶.
• Similarly, in the RITA-3 trial there was no difference between management strategies for those at lowest risk, but those at highest risk who were managed by an early invasive strategy had a significantly reduced risk of death or MI up to 5 years follow-up.
• By contrast in the ICTUS trial an early invasive strategy did not confer benefit and there was no evidence that treatment effect was influenced by risk at randomization. Interpretation of the ICTUS trial is influenced by a high rate of early angiography and revascularization in the conservative arm of the trial¹⁹³.
• When analysis was undertaken of those trials with the routine use of GPIs (mainly based on TACTICS-TIMI 18 but including ICTUS - use of GPIs was 94% in the invasive arms of both trials) an invasive strategy significantly reduced in-hospital non-fatal MI, the composite of death or non-fatal MI (but not death alone), suggesting that appropriate use of GPIs reduces in-hospital MI when added to an invasive strategy. It also reduced rehospitalisation over 6-12 months follow-up.
• When analysed by troponin elevation (TACTICS-TIMI 18) there was no difference between invasive and conservative groups who were troponin negative, but there was a reduction in 30 day death or MI in those managed with an early invasive strategy who were troponin positive, again suggesting that the benefit of an invasive strategy is mostly in higher risk people. The TIMI risk score used in this trial was previously developed to stratify people with UA or NSTEMI according to their risk of an adverse outcome^{207 r}and has been modified to allow stratification before the 12-hour troponin is known²⁰⁸.
• When trials with large absolute differences in revascularisation rates between early invasive and conservative strategies (FRISC-II, RITA-3, VINO) were pooled, a significant reduction in death was seen, suggesting that if a strategy of conservative management is associated with a high subsequent rate of revascularisation (as in TACTICS-TIMI 18, ICTUS, TRUCS) the benefit of an early invasive strategy diminishes. Alternatively, the greatest difference between strategies is seen when the conservatively managed group has a low rate of intervention.
• Two RCTs were appraised that reported quality of life outcomes from FRISC-II ¹⁸³ and RITA-3¹⁸². These showed that people randomised to an early invasive strategy had significantly higher quality of life scores at 6 and 12 months follow-up, than those managed by a conservative approach.

5.1.7. Evidence to recommendations

Using methodology described earlier (reference to risk chapter) the GDG plotted the 6-month mortalities for these risk stratified groups in FRISC and RITA, on GRACE graphs (6-month predicted mortality by GRACE score - see Figure 4.9 and Figure 4.10).

Figure 5-9

6-month mortality (y-axis) and GRACE score (x-axis) data from the GRACE Registry. Six month mortality in FRISC-2 for conservative (red) and invasive (blue) groups shown by FRISC risk stratum on the ‘GRACE curve’ (dark blue). FRISC low (more...)

Figure 5-10

6-month mortality (y-axis) and GRACE score (x-axis) data from the GRACE Registry. Six month mortality in RITA-3 for conservative (red) and invasive (blue) groups shown by RITA-3 risk stratum (boxes) on the ‘GRACE curve’ (dark blue). RITA (more...)

Within the trials the benefits of the routine invasive strategy were mainly seen in people at highest risk. The GDG concluded that an early invasive strategy was likely to benefit those people with a predicted six-month mortality of >3.0% (our risk cohorts 2a, 2b, 3 & 4), although evidence to guide treatment of people at very high risk is limited.

The GDG considered the Hoenig and NCC-CC meta-analyses and concluded that:

• Various scoring systems have been used in the trials of early invasive vs. conservative strategies to assess an individual's underlying risk and several (TIMI, FRISC, RITA) have stratified people into high, intermediate and lower risk groups.
• Comparison of the trial populations with an unselected population of people with UA or NSTEMI suggests that the trials enrolled people at low to intermediate levels of risk and people at the highest levels of risk are systematically excluded from the evidence base.
• An early invasive strategy does have benefit, mainly in reducing recurrent ischaemia/infarction in the short term, but also in reducing longer term mortality or reinfarction. However, this benefit appears to be greatest in those people at higher absolute risk of such events (with the most benefit seen in those at the highest risk). This has also been demonstrated in the recently published TIMACS trial. Some studies have attempted to see if there is a difference in relative treatment benefit amongst different risk groups. However the GDG concluded that there was not strong evidence to demonstrate such an effect.
• Conversely, those at lowest risk are likely to have a similar outcome whether initially managed with an early invasive strategy, or one where angiography is undertaken only when recurrent ischaemia is present, either clinically apparent or as demonstrated by non-invasive investigations. This is particularly true for women where there may even be net harm from an early invasive strategy in those who are troponin negative.
• The trials reviewed have compared an early invasive strategy against a selective invasive strategy, with angiography (and, where appropriate, revascularisation) undertaken if there is subsequent evidence of ischaemia (spontaneous or on non-invasive testing). Those in the conservative limb had a high rate of subsequent angiography (16-55% of those in the conservative management groups of 5 RCTs [TACTICS, ICTUS, RITA-3, FRISC II, VINO] underwent angiography during the index admission). Thus, for those in whom a conservative strategy is adopted many would be expected to undergo angiography (and be considered for revascularisation) at a later stage if the potential benefits of this strategy are to be obtained.

5.1.8. Recommendations

R24.

Offer coronary angiography (with follow-on PCI if indicated) within 96 hours of first admission to hospital to patients who have an intermediate or higher risk of adverse cardiovascular events (predicted 6-month mortality above 3.0%) if they have no contraindications to angiography (such as active bleeding or comorbidity). Perform angiography as soon as possible for patients who are clinically unstable or at high ischaemic risk.

R25.

Offer conservative management without early coronary angiography to patients with a low risk of adverse cardiovascular events (predicted 6-month mortality 3.0% or less).

R26.

Offer coronary angiography (with follow-on PCI if indicated) to patients initially assessed to be at low risk of adverse cardiovascular events (predicted 6-month mortality 3.0% or less) if ischaemia is subsequently experienced or is demonstrated by ischaemia testing.

R27.

Offer patients clear information about the risks and benefits of the treatments offered so that they can make informed choices about management strategies. Information should be appropriate to the patient's underlying risk of a future adverse cardiovascular event and any comorbidities.

5.2.1. Clinical introduction

For all those presenting with UA or NSTEMI, other than those considered at lowest risk, coronary angiography has been shown to offer benefit and is recommended (see section 5.1). This benefit arises from the value of knowing the extent and severity of the individual's coronary artery disease, and the important contribution this makes in determining optimum therapy. For some, treatment will be based on drug therapy alone, but for most this will be supplemented by coronary revascularisation, involving either percutaneous coronary intervention (PCI) or surgical coronary artery bypass grafting (CABG). Determining the optimum treatment strategy for an individual patient is a complex matter that takes account of the risk associated with their underlying cardiac condition, their left ventricular function, comorbidity, the distribution of their coronary artery disease and the relative risks of the revascularisation procedure itself. The objectives of both forms of revascularisation are the same - to alleviate symptoms, prolong life and reduce cardiac morbidity - but the two procedures are obviously very different; CABG involves a surgical operation and general anaesthesia, whereas PCI is less invasive and can be done under local anaesthesia.

Broadly speaking, CABG has tended to be preferred for people with more extensive (three vessel), or diffuse, coronary disease (particularly where there is associated poor left ventricular function), and those with significant narrowing of the left main stem coronary artery. PCI, on the other hand, has been favoured for those people with one or more discrete coronary lesions. Thus, randomised trial data comparing PCI and CABG reflect only those people for whom both treatment strategies are felt clinically to be equally appropriate, and therefore address only a subset of all people presenting with coronary disease. Those for whom there are good clinical reasons to favour one treatment strategy over another (for example medical therapy or PCI for those at high surgical risk, PCI for those with single discrete lesions, CABG for those with diffuse triple vessel or complex left main stem disease) have generally not been randomised in trials. However, the interface between revascularisation strategies has changed over the years and has resulted, for instance, in the more recent randomisation of people who would previously have been considered unsuitable for PCI and to require CABG^{212 212} and some who would previously have been considered too old, frail or with too much co-morbidity to undergo CABG. The selection of patient populations, their respective co-morbidity (particularly the prevalence of diabetes and renal disease) and the advances in clinical practice over time complicates data interpretation and trial comparisons.

Considerable clinical trial and registry data comparing PCI and CABG have been used to inform recommendations and guidelines for the management of people with coronary artery disease.

A number of points should be highlighted:

• the data comparing PCI and CABG are predominantly derived from people with stable angina rather than acute coronary syndromes,
• when included in randomised trials those with acute coronary syndromes usually form a minority of the whole group,
• people with ST elevation MI are generally not considered for early CABG because of their high risk, but increasingly undergo immediate (primary) PCI because of its superiority over medical (fibrinolytic) therapy,
• trials have generally not enrolled the elderly (>75 to 80 years) and have varying exclusion criteria, but generally do not include those at highest risk,
• comparisons over time are confounded by advancing surgical and interventional techniques (such as the introduction of coronary stenting, and the use of arterial graft conduits) and changing adjunctive pharmacotherapy (uptake of secondary preventive treatments such as statins and anti-platelet therapy).

The GDG sought data specific to people with UA or NSTEMI in order to determine the place of these two revascularisation procedures (CABG and PCI) in their management. The clinical question posed was:

‘In adults with UA or non-ST segment elevation MI does CABG improve outcomes in comparison with PCI?’

5.2.2. Clinical methodological introduction

The literature was searched from 1995 to 2009 for systematic reviews, RCTs, and observational studies comparing PCI with CABG in people with non ST-segment elevation ACS. There were few RCTs of PCI versus CABG in people with ACS, thus both RCT and observational studies were included. Studies were included if they reported on either short (index hospitalisation) or long-term (up to 5 years) outcomes including death, MI, bleeding, stroke, repeat revascularisation, angina. Studies of angioplasty without stenting were excluded, as were studies in which the NSTEMI/UA population comprised < 60% of the participants, or if the participants had stable coronary artery disease.

Four open RCTs [ERACI-II ^{213 214} AWESOME ²¹⁵, SoS ²¹⁶, and ARTS ²¹⁷] and five cohort studies ^{218 219 220 221}.²²² compared PCI with CABG in people with multivessel coronary artery disease and UA. Three of the cohort studies ^{220 221}.²²² were rejected because they had serious limitations due to high dropout rates and/or lack of adjustment for confounding variables.

Caution should be exercised in combining results of the studies as they are a mix of RCTs and cohort studies with different degrees and types of stent usage. The populations differed in the number and type of diseased vessels. Table 5-8 details differences in the participants recruited to the four open RCTs.

Table 5-8

Summary of baseline characteristics in four RCTs comparing CABG with PCI.

The Palmerini et al cohort study compared PCI and CABG in people with de novo ≥ 50% unprotected left main coronary artery stenosis (N=311; 63% NSTEMI/UA; follow-up at 30 and 430 days). Multivariate analysis identified independent predictors of death ²¹⁸.

The Seung et al cohort study ²¹⁹ assessed 3 year outcomes in people who had PCI or CABG for unprotected left main coronary artery disease (N= 542 propensity score matched pairs of PCI and CABG people; 57% UA; 11% NSTEMI).

5.2.3. Clinical evidence statements

Short term outcomes (index hospitalisation to 30 days) for CABG versus PCI

Refer to summary

Table 5-9 for a summary of short term outcomes in people randomised to PCI or CABG.

Table 5-9

Summary of short-term outcomes in RCTs: CABG versus PCI revascularisation strategies.

MACCE (Major Adverse Cardiac and Cerebrovascular Event²²³) at 30 days (death, Q-wave MI, stroke, or repeat revascularisation)

One RCT (ERACI-II) showed significantly increased MACE at 30 days in the CABG group compared with the PCI group ²¹³.

Level 1+

Death (index hospitalisation to 30 days)

In ERACI-II there was a significantly higher death rate in the CABG group compared with the PCI group ²¹³. However, two RCTs ^{216 215} and a cohort study ²¹⁸ showed non–significant difference for early death.

Level: 1+ and 2+

MI (index hospitalisation to 30 days)

At 30 days, the ERACI-II RCT ²¹³ showed significantly increased MI in the CABG group compared with the PCI group, whereas SoS ²¹⁶ and a cohort study ²¹⁸ showed non–significant difference for early MI.

Level: 1+ and 2+

Repeat Revascularisation (index hospitalisation to 30 days)

The SoS RCT ²¹⁶ and one cohort study ²¹⁸ showed non–significant difference in repeat revascularisations between the PCI and CABG groups.

Level: 1+ and 2+

Bleeding (in hospital)

One RCT ²¹⁶ showed a non–significant difference in bleed rates between the PCI and CABG groups.

Level: 1+

Stroke (index hospitalisation to 30 days)

Three RCTs showed a non–significant difference between PCI and CABG for stroke at 30 days. ^{214 215 216}.

Level 1+

Long-term outcomes: PCI versus CABG

Refer to Table 5-10 for a summary of long-term outcomes for people randomised to PCI or CABG.

Table 5-10

Summary of long-term outcomes: CABG versus PCI revascularisation strategies.

Freedom from MACCE

After long term follow-up, two RCTs ^{214 217} showed that CABG was associated with significantly lower rates of major adverse cardiac events. One cohort study showed significantly lower rates of death/MI/repeat revascularisation ²¹⁸. A propensity score matched cohort showed a non– significant difference between the two groups for death/MI/or stroke (HR 1.10 [0.75, 1.62]) ²¹⁹.

Evidence level 1+ and 2+

Death

Four RCTs ^{216 214,217 215}and two cohort studies ^218,219 showed non–significant difference in death (or survival) between those who received CABG or PCI after long-term follow-up (1 to 5 years).

Level 1+ and 2+

MI

After long-term follow-up, three RCTs ^{214 217 216} and one cohort study ²¹⁸ showed a non– significant difference between those randomised to CABG or PCI for MI rates at one to five years.

Level 1+ and 2+

Angina

Two RCTs showed a non–significant difference between CABG and PCI groups for anginal symptoms at five years ^{214 215}

Level 1+

Bleeding at 1 year

One RCT showed a non–significant difference between CABG and PCI for bleed rates after one year. ²¹⁶

Level 1+

Repeat revascularisation

At long-term follow-up, three RCTs (ARTS ²¹⁷, SoS ²¹⁶, and ERACI-II ²¹⁴) showed significantly higher rates of repeat revascularisation in the PCI group compared with the CABG group. Similarly, a cohort study ²¹⁹ showed that target vessel revascularisation at three years was significantly increased in those receiving PCI [HR 4.76 (2.80, 8.11)] compared with those randomised to CABG.

Level 1+ and 2+

5.2.4. Health economic methodological introduction

Two relevant cost-effectiveness studies were indentified both based on subgroup analyses of resource use and outcomes collected within RCTs^216,217.

Zhang et al.²¹⁶ reported a subgroup analysis of the SOS trial that analysed costs and outcomes for ACS and non-ACS people separately. The study compares within and between the ACS and non-ACS subgroups. Results are presented here for the ACS subgroup. Costs are calculated using UK prices but international resource use is used. The study was judged to be partially applicable (QALYs were not used and there is some uncertainty around the applicability of international resource use to the UK) with potentially serious limitations.

Zhang et al. reported a cost-consequence analysis from a UK NHS perspective based on 1 year effectiveness and resource use data for a subgroup of people with acute coronary syndrome from the SOS trial (n=242). People had multivessel disease eligible for both PCI and CABG. Bare metal stents were used. Patient level resource use collected during the trial was multiplied by unit costs to calculate the average costs per patient (2000 UK costs were used). This included the index hospitalisation costs and one year follow-up costs. Costs and outcomes were presented separately and not aggregated into a cost–effectiveness ratio. Outcomes reported were death, Q-wave MI, bleeding, cerebrovascular accident, repeat revascularisation, health status. No sensitivity analysis was performed.

Interpretation is inhibited as QALYs were not used and there is some uncertainty regarding the applicability of international resource use to the UK setting. The key limitations of the study include the short time horizon (1 year). Additionally, the analysis is based on a single trial that may not reflect the whole body of evidence in this area.

De Feyter et al. ²¹⁷ reported a subgroup analysis of the ARTS trial that analysed costs and outcomes for stable angina and UA people separately. The country perspective of the economic evaluation is unspecified – costs are reported in US dollars, unit costs are from the Netherlands and the place of resources use collection is not reported. This study is judged to have very serious limitations but was included due to the limited evidence available. The study compares within and between the stable and unstable subgroups. Results are presented here for the unstable subgroup.

De Feyter et al.²¹⁷ reported a cost effectiveness analysis from an unspecified healthcare system perspective based on 1 year effectiveness and resource use data for a subgroup of people with UA from the ARTS trial (n=450). People had multivessel disease and were deemed equally treatable with either PCI or CABG. Bare metal stents were used. Patient level resource use collected during the trial was multiplied by unit costs to calculate the average costs per patient (Netherlands costs were used expressed in 2002^s US dollars – presented here converted to 2002 UK pounds using 2002 Purchasing Power Parities¹³⁸). This included the initial procedure and hospitalisation, follow-up event diagnostic tests, rehospitalisation and medication. Cost effectiveness was measured in terms of cost per MACCE-free life year gained (MACCE = major adverse cardiac and cerebrovascular events and included death [all causes], cerebrovascular incident [stroke, TIA, reversible ischemic neurological deficits], non-fatal MI [spontaneous and peri-procedural], repeat revascularisation [PCI, CABG]). No sensitivity analysis was performed.

Key limitations of the study include the non-UK perspective, short time horizon (1 year), choice of outcome measure and unclear costing methods. Additionally, the analysis is based on a single trial that may not reflect the whole body of evidence in this area.

5.2.5. Health economic evidence statements

Zhang et al.²¹⁶ (SoS trial) reported that costs with CABG compared to PCI were significantly higher in the index hospitalisation (£8248 versus £5015), significantly lower post-discharge to one year (£1832 versus £2998) and non–significantly higher overall (£10,080 versus £8014; difference = £2066, CI: -£690, £3487). Various health outcomes were presented disaggregated and were not combined with costs to give a cost–effectiveness ratio – there was significantly more repeat revascularisation with PCI compared to CABG and no significant difference in other outcomes. The key limitation of the study is the 1 year time horizon - if post-discharge costs continue to be lower each year with CABG this could impact conclusions.

De Feyter et al.²¹⁷ (ARTS trial) reported that CABG was associated with a cost of £20,701 per MACCE-free life year gained when compared with PCI with bare metal stents (95% CI: £8,403– £76,769). Without the use of QALYs this result is difficult to interpret. Examination of the breakdown of MACCE in the trial shows that the benefit of CABG is largely derived from the lower rates of repeat revascularisation compared with PCI. Costs were higher with CABG compared to PCI during the initial hospitalisation, lower in the follow up period and non– significantly higher overall (difference = £3267, p=NS). A key limitation of the study is the 1 year time horizon – if post-discharge costs continue to be lower each year the difference between CABG and PCI may diminish.

NHS reference costs for CABG and PCI

Due to the lack of relevant cost-effectiveness analyses, UK costs were sought for CABG and PCI to aid discussions regarding cost effectiveness. Below in Table 5-11 and Table 5-12 are costs extracted from the NHS reference costs ²²⁴.

Table 5-11

NHS reference costs.

Table 5-12

NHS reference costs.

5.2.6. Evidence summary

Four randomised trials were identified in which people with UA or NSTEMI were randomised to PCI or CABG: ERACI-II²¹⁴, AWESOME²¹⁵, SoS²¹⁶, and ARTS²¹⁶.

• ERACI-II randomised 450 people with non-ST elevation acute coronary syndromes, classified by Braunwald's criteria²²⁵.

  ◦

  At 30 days those people undergoing CABG had significantly higher rates of death, acute MI and the composite endpoint of major adverse cardiac and cerebrovascular events (MACCE) than those in the PCI group. The mortality difference persisted to 33 months of follow-up, but after 30 days the number of additional deaths was the same in both groups. The difference became nonsignificant at 5 years. The 30 day mortality in the CABG group in this trial was 5.7% (compared to 0.9% in the PCI group), which the GDG felt to be much higher than would be expected in UK practice.

  ◦

  Those undergoing PCI had a significantly higher rate of further revascularisation procedures at each point of the follow-up period, though most additional revascularisations in the PCI group took place within the first year.

• AWESOME: there was no difference in survival between the PCI (77%) and CABG (74%) groups, but significantly more people undergoing PCI required further revascularisation procedures during the 5 year follow-up period than those in the CABG group.

  ◦

  There was a high drop out rate by the end of the 5 year follow-up period, but this was comparable between groups.

• A subgroup analysis of the SoS trial described events in 242 people with ACS (defined as acute MI or UA) randomised to PCI or CABG. CCS IV angina was diagnosed in 62% of this cohort There was no information on renal function, or LVEF, although 56% had a previous MI.

  ◦

  There was a non-significant difference in any outcome during the index hospitalisation for those randomised to PCI or CABG

  ◦

  After 1 year, there was a non-significant difference in death between the two trial arms (1.6% CABG versus 2.6% PCI) and repeat revascularisation was significantly higher in the PCI arm.

• A subgroup analysis of the ARTS trial compared PCI with CABG in 450 people with UA.

  ◦

  After 1 year, there was non significant difference in death between the two trial arms (2.2% CABG versus 2.7% PCI) and repeat revascularisation was significantly higher in the PCI arm.

The SOS and ARTS subgroup analyses described above also analysed resource use in the trial and estimated costs. They both found that costs in hospital were higher with CABG than PCI but that post-discharge to one-year costs were lower with CABG. The latter is attributable to the reduced rate of repeat revascularisation observed with CABG.

The ARTS study was judged on economic terms to have serious limitations; it was a non-UK perspective, had a short time horizon (1 year), did not estimate QALYs and was unclear in its costing methods. Differences in costs between the two treatment strategies were almost entirely due to the difference in frequency of repeat revascularisation procedures during the follow-up period. Given that the study preceded the era when drug eluting stents became used (which might be expected to reduce this need for repeat procedures) the study was felt of limited applicability to current practice. The GDG discussed this paper but concluded that it did not allow robust conclusions about cost effectiveness to be drawn. The SoS trial was from a UK perspective with clearer methods but shared the other limitations noted above and so the GDG felt this also did not allow robust conclusions to be drawn.

5.2.7. Evidence to recommendations

In its discussions the GDG particularly noted that of the four randomised trials identified:

• All recruited people with multivessel rather than single vessel coronary artery disease
• All excluded people with limited life expectancy due to advanced age or co-morbidity (average age 61-67 yrs across the four trials). Three (ERACI, ARTS, SoS) excluded people who had previously undergone CABG, and only AWESOME recorded including patients with severe left ventricular impairment (LVEF <0.35).
• All will have included troponin positive NSTEMI people but preceded the routine use of this biomarker so it is not possible to subdivide the recruited patient population into UA and NSTEMI
• One trial (AWESOME) had relatively low overall stent usage (55%), much lower than in current practice (used in 94.7% of all PCI procedures in the UK in 2007¹⁷⁸), and had only 11% usage of GPIIbIIIa inhibitors.
• All preceded the use of drug eluting stents and therefore involved only bare metal stents, which are known to have a higher risk of re-stenosis. Given that the most significant difference between the outcome of people undergoing CABG compared to PCI is the increased requirement for further revascularisation procedures during follow up in the PCI group, this difference may be decreased with increasing use of drug eluting stents (55% of all stents inserted in the UK in 2007)⁶.

In addition to the literature above, the GDG also reviewed the results of five cohort studies ^{218 219 220 221}.²²² but felt that few conclusions could be drawn from them because of the degree of selection inherent in their non-randomised nature, often incomplete details and lack of adjustment for confounding factors and other methodological issues. Nevertheless the findings from these registry data were felt compatible with the conclusions the GDG drew from the four randomised studies.

The two revascularisation strategies, CABG and PCI, have been employed in the management of people with UA and NSTEMI for nearly three decades, during which time surgical and PCI procedural techniques have advanced and adjunctive pharmacotherapy has changed. People most suitable for each therapy have been generally agreed (for example, CABG for diffuse triple vessel disease, PCI for single discrete lesions). However, the group of people regarded as potentially equally suitable for both treatment strategies has changed and continues to be the subject of randomised clinical trials, most notable of which recently was SYNTAX. ²¹². Thus, any study comparing these two techniques is inevitably based on a subset of all people admitted with UA/NSTEMI. As outlined above, even allowing for the inevitability of selection. The GDG noted that very few trials have actually specifically addressed people with UA or NSTEMI. Many trials have included these people (33-42% of people with UA/NSTEMI underwent CABG in FRISC II, TACTICS-TIMI 18, and RITA 3) but either not reported their outcome separately or have recruited too few of these people for a meaningful analysis to be undertaken.

Registry data specific to the UA/NSTEMI population has, on the whole, not been particularly useful in drawing conclusions about the applicability of each of these treatment strategies. Also, the definition of outcome events is not always clear between studies; for instance, some trials do not clearly separate those who had myocardial infarcts and those who died, sometimes recording deaths due to an MI simply as a death but not as an MI. The definition of MI is also unclear in some studies. The average age in the randomised trials ranged from 61 to 67 years, thus representing a cohort of people younger than many seen in current practice.

Trials comparing the use of CABG and PCI have generally required equivalent revascularisation; in other words, the cardiologist and cardiac surgeon have to agree that each coronary lesion can be equivalently revascularised by both techniques before randomisation can occur. More recently the potential use of PCI initially just for the perceived ‘culprit lesion’ (with the potential for subsequent further PCI – ‘staged procedures’) has been compared to initial complete PCI revascularisation²²⁶. Safety end points did not differ between groups. This practice may be appropriate when the risk of staged procedures is considered to be lower than one procedure at which full revascularisation is attempted (as might occur in people with renal impairment in whom a reduced single contrast load may be beneficial). Such practice introduces yet another potential variable when clinicians are considering the choice of most appropriate therapy.

The GDG concluded that the evidence supported the use of both revascularisation strategies, with their selection for individual people harmonizing with criteria already recommended in international guidelines, such as the extent and severity of their coronary disease, left ventricular function, the presence of co-morbidity, the estimated risk of each procedure, and patients' informed choice. There may be an early (<30 days) increase in MACCE for people undergoing CABG, as suggested in ERACI-II (not seen in AWESOME) but because of the later increased need for further revascularisations in the PCI group this difference became reversed after five years of follow-up. This is in keeping with the outcome of comparative trials in people with stable angina, where the difference between these two revascularisation techniques is mainly the higher need for repeat procedures in those initially undergoing PCI.

In many people clinical suitability dictates whether PCI or CABG should be undertaken but in a subgroup of people PCI or CABG are equally feasible and appropriate approaches and a relevant concern is which is cost–effective. PCI is a much less expensive procedure than CABG; however the group considered that longer term costs following CABG are likely to be lower than following PCI in particular due to the lower rates of repeat revascularisation as seen in the trials identified, but also potentially due to the greater pharmacological interventions associated with PCI to prevent restenosis. It was noted that a cost–effectiveness analysis in a broader PCI populations (that is, including stable people) has hinged upon whether in the long term a survival advantage accrues with CABG, with results favouring CABG if it does and PCI if it does not ²²⁷. The GDG concluded that there was a lack of evidence regarding long term outcomes and as such great uncertainty as to which was most cost–effective. The group therefore agreed that a research recommendation that addresses both the clinical and cost effectiveness of PCI versus CABG specifically in people with NSTEMI/UA would be useful to help inform the evidence base.

Patient representatives on the GDG stressed the importance of individuals being fully informed of the relative risks, benefits and differences between the two procedures so that they could make informed choice. Clinicians on the group agreed this was of fundamental importance and highlighted the need for appropriate consent processes^228,229, and the value of multi-disciplinary team (MDT) meetings in determining the most appropriate treatment strategy to recommend to people when both seem clinically appropriate²³⁰.

5.2.8. Recommendations

R28.

When advising patients about the choice of revascularisation strategy (PCI or CABG), take account of coronary angiographic findings, comorbidities, and the benefits and risks of each intervention.

R29.

When the role of revascularisation or the revascularisation strategy is unclear, resolve this by discussion involving an interventional cardiologist, cardiac surgeon and other healthcare professionals relevant to the needs of the patient. Discuss the choice of revascularisation strategy with the patient.

5.2.9. Research recommendation

What is the efficacy and cost effectiveness of CABG versus PCI in the management of patients with NSTEACS?

5.3.1. Clinical introduction

Intra-aortic balloon counterpulsation (IABP) was first described in 1962²³¹ and was estimated in 1990 to have been used in over 70,000 cases annually in the USA²³². It has been used as a means of supporting the circulation predominantly in those with failing left ventricles (particularly in cardiogenic shock), or as an adjunct to treatment by cardiac surgery or high risk coronary angioplasty²³³.

The technique involves the insertion of a balloon catheter device, usually via a femoral artery, into the descending thoracic aorta, with the proximal end of the catheter attached to an external pumping device which inflates the intra-aortic balloon during diastole and deflates it just prior to the onset of systole. A full description of the haemodynamic effects of balloon pumping (more precisely termed intra-aortic balloon counterpulsation) is beyond the scope of this guideline, but its haemodynamic benefit arises from its potential to increase diastolic blood pressure (thereby improving coronary blood flow²³⁴, and reduce left ventricular afterload (increasing cardiac output, the amount of blood ejected by the heart). More recently, other percutaneously implanted, circulatory support devices have also been developed and show promise²³⁵.

The technique require invasive intervention, the availability of sophisticated equipment, and staff who are familiar with its implementation and subsequent monitoring, and vascular complications can occur²³⁶. Also, patients with significant peripheral vascular disease may either be unsuitable for insertion of the counterpulsation balloon catheter, or may have ischaemic lower limb complications as a consequence of its insertion. Its use outside cardiac surgical centres has been limited in the past, although the British Cardiovascular Intervention Society reported 983 cases of IABP being used in the UK as an adjunct to coronary angioplasty (PCI) in 2007 (1.7% of all PCI cases), many of which were performed in non-cardiac surgical centres¹⁷⁷, and some of which will have been in patients with UA or NSTEMI.

The GDG therefore wished to review the evidence for its use in patients with UA or NSTEMI to determine whether there was evidence of improved patient outcome. The clinical question asked, and upon which the literature was searched was

‘Does the use of Intra-Aortic Balloon Pump Counterpulsation affect the outcome of patients with non-ST elevation myocardial infarction or unstable angina?’

5.3.2. Clinical methodological introduction

The literature was searched from 1995 to 2009 for systematic reviews, RCTs, comparative studies, and observational studies. There were no relevant studies identified specifically in people with UA or NSTEMI where IABP was used as a form of treatment in its own right to stabilise patients. Studies were excluded if IABP was electively used in stable cases to reduce procedural risk (during PCI or CABG). Studies were excluded if the population comprised mostly STEMI patients or if the population was unclear.

5.3.3. Health economic methodological introduction

No relevant economic studies were identified examining IABP in the population described above.

5.3.4. GDG debate

Whilst IABP has a long track record as a therapeutic intervention in patients with ACS, (including UA and NSTEMI as well as ST elevation MI), its use has been reserved for those who are more severely ill and unstable. Examples of such patients would be those who have severe left ventricular failure, cardiogenic shock or who are haemodynamically unstable. Such patients will often be acutely unwell because of the severity of their myocardial ischaemia, and when unresponsive to medical therapy alone will be considered for IABP. Whilst such clinical scenarios are well described they occur in only a small minority of patients admitted with ACS and therefore a multi-centre study it be reasonable way forward. However there may be ethical issues given the severity of their condition and the relative failure of medical therapy, it may be inappropriate to withhold its use in the control group in those patients who deteriorate haemodynamically and hence equipoise would be difficult.

The GDG were therefore not surprised at the lack of data sufficient to allow a firm recommendation for the use of IABP to be made. However, they were persuaded of the potential for IABP to be beneficial for those patients with severe or recurrent ischaemia whose ischaemia cannot be managed adequately by medical therapy and/or coronary revascularisation alone. It is difficult to assess the size of the population of patients with UA or NSTEMI who may potentially be stabilised by, and may benefit from, the use of IABP but the GDG agreed that it was small (<5%). All cardiac centres undertaking coronary angioplasty (n=98 in 2007) are required, as part of best practice, to be capable of initiating IABP in their catheter laboratories²³⁰. These centres will already have the facility to undertake IABP if believed to be clinically appropriate, and therefore the GDG agreed that even if its use were to increase, the economic impact would be minimal.

Because of the infrequency with which IABP is used, particularly outside surgical or interventional centres, the GDG felt that clinicians should be encouraged to consider the option of IABP for those patients who remain clinically unstable due to recurrent myocardial ischaemia despite medical therapy or early revascularisation, and for those who are haemodynamically unstable prior to undergoing surgical or percutaneous revascularisation. Where IABP is unavailable in their own institution clinicians managing such ‘refractory’ patients should consider discussing the potential for its use in individual cases with a centre able to offer such intervention. The GDG could not, however, make a clear recommendation for its specific use due to a lack of robust evidence.

5.3.5. Research recommendation

What is the efficacy and cost effectiveness of intra-aortic balloon counterpulsation (IABP) in the management of patients with non ST-segment elevation ACS?

5.4.1. Clinical introduction

In people with chronic stable coronary disease there is a strong association between the presence and severity of myocardial ischaemia and an adverse outcome. The ability to perform an exercise test and the exercise time are also predictive^237,238. This adverse outcome can be improved by appropriate treatment whether, medical or revascularisation²³⁹ and revascularisation provides better outcomes when inducible ischaemia involves more than 10% of the myocardium.

In people with unstable and acute coronary syndromes, once the initial unstable episode has stabilised, further spontaneous ischaemia is more frequent in people with NSTEMI than with STEMI and, if present, it increases subsequent mortality. People with NSTEMI also have higher reinfarction rates and mortality at one year than those with UA²⁴⁰. The INSPIRE trial showed that early ischaemia testing after myocardial infarction can identify a low risk group of people who may benefit from early discharge although approximately half of the study group had STEMI and so reliable conclusions could not be drawn about people with NSTEMI alone^t.

The majority of people with UA or NSTEMI will undergo angiography during their acute admission. The extent and severity of their coronary disease is thereby documented and, where appropriate, PCI or CABG can be offered to reduce future risk²⁴¹. Later after hospital discharge when the acute episode has stabilised ischaemia testing can be helpful. The question therefore arises whether ischaemia testing may also be helpful before discharge in people where the coronary anatomy has not already been established by angiography.

The available provocative tests for ischaemia include stress electrocardiography (sECG), myocardial perfusion imaging by scintigraphy (MPS), magnetic resonance imaging (perfusion, viability and stress) and stress echocardiography (sEcho). Each of these has its strengths and weaknesses. MPS has been appraised by NICE and found to be clinically and cost effective for the diagnosis and management of people with angina and MI²⁴².

The clinical question posed was:

In patients with UA/NSTEMI who do not undergo angiography, does investigation prior to hospital discharge for myocardial ischaemia affect outcome?

5.4.2. Methodological introduction

The literature was searched from 1995 to 2009 for systematic reviews, RCTs, comparative studies, and observational studies. There were no RCTs comparing ischaemia testing with no such testing before discharge, but two observational cohort studies were identified ^243,244.

GUSTO IIb ²⁴³ (8011 people with non ST-segment elevation ACS) compared sECG against no sECG and reported death or non-fatal MI at 30 days, death at 30 days and one year, and MI at 30 days. There was no formal comparison between the relevant subgroups (sECG but no angiography, n=1061, and neither sECG nor angiography, n=2402). The NCC-CC therefore conducted a simple statistical analysis but we could not adjust for confounding variables.

The ACOS registry ²⁴⁴ included 5281 people with NSTEMI and compared sECG with no sECG before discharge and reported all cause mortality and revascularisation rates at one year. sECG was also compared with no sECG in 2872 people who did not undergo PCI in hospital.

The applicability of these studies was limited because a high proportion of people also received invasive procedures in both arms (44% angiography ²⁴³ or 77% angiography or PCI in the sECG cohorts ²⁴⁴, and 61% angiography ²⁴³ or 72% angiography or PCI ²⁴⁴ in the no sECG cohort).

5.4.3. Clinical evidence statements

5.4.4. Health economic methodological introduction

No economic analyses were identified that compared ischaemia testing and no such testing before discharge in UA/NSTEMI people who did not undergo angiography.

5.4.5. Evidence summary

• Both cohort studies showed higher one year mortality in those who did not undergo ischaemia testing.
• In the GUSTO-IIb cohort²⁴³ after exclusion of people who had MI, recurrent ischaemia, congestive heart failure, death or angiography within 48 hours of admission (those at highest risk), outcomes (30 day and one year mortality or MI) were significantly worse in people who had not undergone ischaemia testing.
• In the German Acute Coronary Syndrome Registry (ACOS)²⁴⁴ when people who had undergone prior PCI were excluded, those without ischaemia testing had higher one year mortality.

A sub-group of the GUSTO-IIb people²⁴³ did not undergo angiography during index hospital admission and this is the group most relevant to our question. In this subgroup those who had undergone ischaemia testing (n=1061) had a better outcome (mortality at six-months and one-year, or death/MI/revascularisation at six months) unadjusted for risk than those who had not (n=2404). However, the potential for confounding factors was considerable. People who did not undergo ischaemia testing were more likely to be older, female, have hypertension, diabetes or renal impairment, have previously identified coronary artery disease, and were less likely to be treated with aggressive secondary prevention measures. They were therefore at higher risk than those who underwent ischaemia testing and this may have led to their worse outcome.

5.4.6. Evidence to recommendations

The lack of prospective randomisation and the high rate of angiography make these studies only partly relevant to the clinical question asked and therefore the GDG concluded that there was no data upon which they could recommend a routine policy of testing for myocardial ischaemia before hospital discharge in all people who do not undergo angiography during their index admission. Decisions on investigation must take account of individual circumstances and there will be people for whom ischaemia testing may or may not be clinically appropriate. Given this caveat the GDG noted:

• People with UA or NSTEMI with subsequent spontaneous or provocable ischaemia have worse outcomes.
• Myocardial revascularisation can improve outcome (see section 5.2), particularly in those at higher risk of a further ischaemic event.
• Those at higher risk can be identified by using risk scores (see section 2) and also by determining the extent and severity of coronary disease (see section 5.1) and/or the extent and severity of inducible myocardial ischaemia.

5.4.7. Recommendations

R30.

To detect and quantify inducible ischaemia, consider ischaemia testing before discharge for patients whose condition has been managed conservatively and who have not had coronary angiography.

5.4.8. Research recommendation

What is the role of ischaemia testing in people after an acute coronary syndrome and what is the comparative efficacy and cost effectiveness of the different non-invasive tests (for example, stress ECG, echocardiography, radionuclide scanning and magnetic resonance imaging)?

5.5.1. Introduction

Heart failure is a syndrome that develops when cardiac output is insufficient to meet the needs of the body. Impairment of left ventricular function secondary to ischaemic myocardial damage is its commonest cause and it is an important determinant of longer term outcome after an acute coronary syndrome²⁴⁵. NICE and others have published guidance on the detection and management of heart failure^246-248. Medical therapy, myocardial revascularisation and devices such as implantable defibrillators and resynchronisation pacemakers can improve symptoms and outcome^249-251. There is extensive literature on the association between the degree of left ventricular impairment and its effects on clinical outcome. However, much of this is in the setting of stable coronary disease and the findings may be less applicable early after ACS when the myocardium may be temporarily stunned, or before the onset of left ventricular remodelling or the effects of chronic medication. A further relevant question after ACS is the extent and transmurality of infarction, since the association between ST elevation during infarction and the amount of viable myocardium remaining in the infarct territory is poor. All of the commonly available imaging techniques are able to assess the extent and transmurality of infarction and the information is relevant in deciding whether revascularization of the infarct territory is important^252,253. This aspect of investigation after ACS is not covered by this guideline.

The clinical questions asked, and upon which literature searching was undertaken, was:

“In people admitted with UA or NSTEMI, does unselected assessment of left ventricular function before discharge improve clinical outcome?”

5.5.2. Clinical methodological introduction

The literature was searched from 1995 to 2009 for systematic reviews, RCTs, comparative studies, or observational studies. Studies were included if they reported outcomes after 30 days such as death, MI, stroke, bleeding, re-revascularisation and quality of life. Studies that assessed the predictive power of left ventricular function for future events, as opposed to the ability to affect outcome, were excluded.

5.5.3. Clinical evidence statement

No RCTs were found that assessed the effect of measuring left ventricular ejection function (LVEF) compared with not measuring (or delayed measuring) LVEF on outcomes in people with non ST-segment elevation ACS. Most studies were excluded because the populations comprised less than 60% UA or NSTEMI. Most studies were conducted in a more general acute infarction population with a high proportion of STEMI. Therefore, there was no evidence identified as being relevant to the question.

5.5.4. Health economic methodological introduction

No economic analyses were identified that compared measuring LVEF compared with not measuring (or delayed measuring) in people with UA/NSTEMI.

5.5.5. GDG debate

UA, NSTEMI and STEMI are part of a continuous spectrum of pathology and people can move from one state to another. The influence of left ventricular dysfunction on outcome is likely to be independent of clinical presentation and whether the dysfunction arises from the index event or from previous events that may not have been clinically apparent.

During admission with an acute coronary syndrome, some people have echocardiography as part of their assessment, some have left ventriculography at the time of coronary angiography and some have radionuclide imaging for the assessment of myocardial ischaemia. All of these tests provide an assessment of left ventricular function and it is likely that only a minority of people with UA or NSTEMI do not have an opportunity for their left ventricular function to be recorded during their hospital admission or shortly thereafter.

In a previous clinical guideline on secondary prevention after MI (NSTEMI and STEMI) ⁴, it was recommended the assessment of left ventricular function in all people after MI. It would therefore be logical to assess left ventricular function in all people with UA and NSTEMI so that specific treatment for left ventricular dysfunction can be offered to improve symptoms and outcome. There is no evidence that assessment of left ventricular function in the subset of people with UA who have stabilised and who have not already had it assessed in the course of other investigations might improve outcome. It was felt that as this would be a very small number of people, a recommendation was justifiable in the interests of uniformity and simplicity.

Left ventricular function may improve after an acute ischaemic event with the resolution of myocardial stunning and the onset of healing. It may also deteriorate because of myocardial remodelling or progression of coronary disease²⁵⁴. It may therefore also be important to monitor left ventricular function during follow-up, because of this potential for change with time. The frequency of these assessments and the relative merits of the different techniques for assessing function are outside the scope of this guideline.

5.5.6. Recommendations

R31.

Assessment of left ventricular function is recommended in all patients who have had an MI. (This recommendation is from ‘MI: secondary prevention’, NICE clinical guideline 48.)

R32.

Consider assessing left ventricular function in all patients with unstable angina.

R33.

Record measures of left ventricular function in the patient's care record and in correspondence with the primary healthcare team and the patient.

5.6.1. Clinical introduction

The management of ACS has become more complex with increased diagnostic and therapeutic options available to the clinician. Many of these options, for example continuous rhythm monitoring and the administration of specialist drugs, require staff with specialist knowledge and skills, and certain interventions now considered standard practice, such as coronary angiography, can only be delivered in specialist environments by specialist teams. However, many people, including the elderly, are admitted to general wards and managed by general medical or elderly care teams, with referrals to specialist care being dependent on local custom and practice. Specialist care impacts upon the accurate and timely assessment of risk including 12 lead ECG monitoring and early angiography, and interventions such as the use of certain drugs and resuscitation procedures which may be performed by staff with specialist training.

The clinical questions asked, and upon which literature searching was undertaken, was:

“Is there evidence that specialist cardiology care is more clinically and cost-effective than non-specialist care in an UA or NSTEMI population?”

5.6.2. Clinical methodological introduction

The clinical question compared the care provided by specialist and non-specialist teams and not simply the involvement of one particular team member (such as a cardiologist) because the overall care of people requires a collaborative approach. The literature was searched from 1999 to 2009 for systematic reviews, RCTs, comparative studies, and observational studies comparing care of people with NSTE ACS by specialist cardiology teams versus non-specialist teams. The rationale for searching from January 1999 onwards was to reflect current practice, particularly the use of stents for revascularisation.

There were no RCTs that compared the care of people with UA/NSTEMI by specialist cardiology teams versus non-specialist teams. Observational studies were included if they reported outcomes including death, MI, bleeding, stroke, revascularisation, use of appropriate medication, uptake of angiography, uptake of cardiac rehabilitation, uptake of evidence-based practice. Studies were excluded if the NSTEMI/UA population comprised < 60% of the participants. Studies were excluded if the populations had undifferentiated chest pain, or if the population was unclear (such as ‘acute MI’ with no further detail on the proportion of the NSTEMI/UA population). Studies were excluded if the comparison was tertiary care hospitals versus community hospitals, or interventional centres (angiography) versus non-interventional centres because they did not address the specific question being asked, and because of the potentially different, non-randomised, populations.

The studies included focused on specialist care provided by a cardiologist.

A UK observational study (N total =83,599; N NSTEMI = 50,436; MINAP database) compared mortality, prescription of secondary prevention drugs, and angiography in people with acute MI who had received their initial care from a cardiologist or a non-cardiologist Compared with people who were treated by non-cardiologists, people treated by cardiologists were younger, more likely to be male, smoke, have ST elevation, and have lower co-morbidity. Effect sizes were adjusted for patient characteristics/history, and hospital cluster ²⁵⁵.

A US observational study (N non ST-segment elevation ACS= 55,994; CRUSADE database) compared mortality, re-infarction, prescription of secondary prevention drugs, and angiography in people admitted to tertiary hospitals with revascularisation capabilities who had received their initial care from a cardiologist or a non-cardiologist (defined as family practice/internal medicine/other). Compared with people who were treated by non-cardiologists, people treated by cardiologists were significantly younger, more likely to be male and had significantly lower co-morbidity. People cared for by cardiologists were significantly more likely to smoke, and had higher prevalence of a family history of CAD, hyperlipidaemia, prior MI, prior CHF, prior PCI, prior CABG, and significantly more likely to have ST depression. Effect sizes were adjusted for patient characteristics/history, and hospital, and geographic location ²⁵⁶.

5.6.3. Clinical evidence statements

One observational study showed that factors most strongly associated with care by cardiologists were lower presenting heart rate, younger age, male sex, prior PCI, transient ST elevation, lack of renal insufficiency, lack of prior stroke, lack of diabetes, lack of CHF ²⁵⁶.

Refer to Table 5-13 for a summary of outcomes in observational studies comparing care by cardiology versus non cardiology teams.

Table 5-13

Summary of outcomes for cardiology versus non-cardiology care.

5.6.4. Health economic methodological introduction

No economic analyses were identified that compared specialist cardiology care and non-specialist care in an UA or NSTEMI population.

5.6.5. Evidence summary

Those initially seen by cardiologists in the first registry (Myocardial Ischaemia National Audit Project; MINAP) study were:

• More likely to be younger, and less likely to have significant co-morbidity.

  ◦

  This may be because more elderly people (who are those most likely to have co-morbidity) are admitted to hospital and managed by elderly care physicians, and perhaps because of the perception that those who are younger with acute coronary syndromes are somehow more appropriately managed by cardiologists. Given that the elderly are often those at highest risk of an adverse outcome this practice would be counter intuitive, assuming, of course, that specialist care may result in better patient outcome. If this were to be the case one might expect a higher proportion of elderly people being managed by cardiologists than that reported.

• More likely to be male, and more likely to smoke

  ◦

  One explanation for the apparent gender bias may be because female life expectancy is longer than for males and therefore a higher proportion of the elderly population will be female than in the younger cohort Given that the elderly are less likely to be managed by a cardiologist this may explain, at least in part, why they manage fewer women than non-cardiologists.

  ◦

  It has been noted before that, contrary to expectation, current smokers may have better early outcomes following ST elevation MI than non-smokers, the so called ‘smokers paradox’. One proposed explanation for this is that smokers present at an earlier age than non-smokers and that the benefit of relative youth counteracts the worsening prognosis associated with increasing age²⁵⁷. Therefore, it may be that the reason why more smokers are seen initially by cardiologists is a reflection of their younger age.

• More likely to have ST elevation (STEMI) than non-ST elevation (NSTEMI) MI

  ◦

  This may be due to a perception that NSTEMI is a more benign condition than STEMI and that therefore the threshold for referral for specialist cardiological care is higher. This is an erroneous perception because in modern day practice their outcomes are very similar²⁵⁸.

• More likely to be alive at 90 days

  ◦

  Given that people initially seen by cardiologists were younger and had less co-morbidity, both major determinants of outcome, it is impossible to draw any conclusion regarding the effect of specialist care on 90-day outcome. Also, it is important to note that people who died during their hospital admission were excluded from analysis and so the 90-day mortality data relates only to those surviving to hospital discharge. A randomised trial is needed if this question concerning outcome, as it relates to system of care, is to be answered.

• More likely to receive secondary prevention medication

  ◦

  The uptake of aspirin and ACE inhibitors were not significantly different, but initial care under a cardiologist was associated with a higher prescription rate for statins and beta blockers. It is more difficult to explain this difference on reasons of age, gender, or co-morbidity, particularly with regards the statins (age and co-morbidity might influence beta blocker usage) and it may be that this reflects a true difference in adherence to best practice guidelines.

• More likely to undergo coronary angiography

  ◦

  This was reported for people admitted to hospitals without coronary intervention facilities on site, but there was no difference for people admitted to an interventional centre. There are a number of possible explanations for this, including heightened awareness of, and willingness to refer for, angiography amongst those non-cardiologists when services are on site and an appreciation of their use more directly experienced by the referring physician. Little can be concluded from this observation alone.

The second registry (CRUSADE database) reported on nearly 56,000 people from a US population and was different from the MINAP study because it involved only people admitted to hospitals where coronary revascularisation procedures (PCI and CABG) were available on-site (in UK terms a ‘tertiary centre’) whereas MINAP included people admitted to hospitals without revascularisation services. They compared people managed by cardiologists and non-cardiologists and made similar observations to MINAP; people under cardiologists were younger, had less co-morbidity and were more likely to be male, and to smoke. They were more likely to be prescribed secondary prevention medication, which in this study included being more likely to be prescribed aspirin. After various adjustments there was no difference in hospital mortality, but people under cardiological care had less in-hospital reinfarction, and were more likely to undergo coronary angiography, and PCI (but not CABG) than those under non-cardiologists.

5.6.6. Evidence to recommendations

Only observational data is available and conclusions drawn from these must be made with caution because there is a potential for selection bias, and confounding factors, to influence observations. It is unclear if benefits gained reflect the overall care of people within a specialist cardiology service or are attributable to the cardiologist in isolation. The two registries (MINAP, CRUSADE) suggest that differences in practice may exist, particularly with respect to the uptake of secondary prevention therapies, and the use of angiography, and that there may be gender and age-related bias. These observations have also been reported elsewhere^259-262.

There is good evidence, reviewed elsewhere in this guideline, to support the use of various pharmacological agents, revascularisation procedures and cardiac rehabilitation, in the management of people with UA and NSTEMI. Adherence to best practice guidelines is known to vary between institutions²⁶³, and types of healthcare services²⁶⁴, and have shown that better adherence can improve patient outcome^3-6. Evidence also exists for the benefit of systematic implementation of quality assurance processes that encourage guideline implementation²⁶⁵, and recent recommendations by the American College of Cardiology (ACC) and American Heart Association (AHA) have been made concerning the use of performance indicators which can be used to determine adherence to best practice guidelines²⁶⁶.

The GDG concluded that while there was insufficient evidence to make any specific recommendations regarding the systems of multidisciplinary/specialist care in which people with UA or non ST-segment elevation ACS are managed, it felt that the assessment and management of such people by skilled staff in properly equipped settings is the preferred pathway of care. This was supported strongly by the patient representatives on the group.

In summary, the GDG concluded that:

• Adherence to best practice guidelines should be universally applied.
• Adherence to NICE guidelines and mortality should be the subject of regular internal and external process and metric audit
• Audit results should be scrutinised at hospital and network/strategic health authority level to ensure equity of access and quality of care.
• Where a person's care involves more than one institution the whole of the person's in-patient pathway should be considered. Institutions should work together to ensure high performance. This should include the sharing of data and seamless clinical protocols.

5.6.7. Research recommendation

What is the comparative efficacy and cost effectiveness of systems involving specialised care compared to non-specialised care?

5.7.1. Clinical introduction

The World Health Organization has defined cardiac rehabilitation as ‘the sum of activity and interventions required to ensure the best physical, mental, and social conditions so that people with chronic or post-acute cardiovascular disease may, by their own efforts, preserve or resume their proper place in society and lead an active life’ (http://www.who.int/en/).

The National Service Framework for Coronary Heart Disease²⁶⁷ identifies four phases of cardiac rehabilitation: phase 1 (before discharge from hospital); phase 2 and 3 (early post discharge phase); phase 4 (long term maintenance of changed behaviour). Please see Appendix D for further details regarding what comprises the four phases of rehabilitation.

Similarly, The British Association for Cardiac Rehabilitation (BACR) (2007) ²⁶⁸ identify standards and core components for the delivery of cardiac rehabilitation. The core components they identify are (1) lifestyle (physical activity and exercise, diet and weight management, smoking cessation), (2) education, (3) risk factor management, (4) psychosocial support, (5) cardioprotective drug therapy and implantable devices and (6) long-term management strategy. They recommend the core components should be based on a comprehensive assessment, appropriate referral and collaboration with the individual patient, family and carers.

The standard NHS contract for acute hospital services identifies healthcare obligations in relation to discharge communication. Information about this can be found at http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPolicyAndGuidance/DH_081100?IdcService=GET_FILE&dID=158542&Rendition=Web

In addition, The Royal College of Physicians' Health Informatics Unity (HIU) has developed standards for record keeping. These can be accessed at: http://www.rcplondon.ac.uk/pubs/brochure.aspx?e=225.

Patient participation in cardiac rehabilitation following MI (whether an exercise-only programme²⁶⁹, or a more comprehensive approach²⁷⁰) has been shown to reduce all-cause and cardiac mortality when compared to usual care. In 2000 the National Service Framework for Coronary Heart Disease² recommended that more than 85% of people discharged from hospital with a primary diagnosis of acute MI, or after coronary revascularisation, should be offered cardiac rehabilitation. However, less than a third of all people with a prior MI, or who have undergone coronary revascularisation, attend comprehensive cardiac rehabilitation. Uptake is particularly poor among certain groups including ethnic minorities, women, the elderly and those on low incomes or with physical or mental comorbidities⁴.

In 2007 NICE published guidance on secondary prevention following MI ⁴. See Appendix E for all recommendations. No distinction was made in the scope of the MI guideline between non-ST elevation MI and ST-elevation MI. As such, the literature review and recommendations from the MI guideline that pertain to rehabilitation, lifestyle advice and discharge planning are applicable to people with NSTEMI in this guideline.

Given the existing recommendations from the NICE MI Guideline⁴ the GDG addressed the question of whether the psychosocial and educational interventions that constitute the early part of the rehabilitation process should be initiated before hospital discharge, or whether such initiatives could be deferred until after the patient returns to community care.

The clinical question upon which the literature was searched was:

“Do early psychosocial and educational interventions, mobilisation and discharge planning (cardiac rehabilitation - Phase 1) improve emotional and physical wellbeing and long-term outcomes in people with unstable angina or NSTEMI compared to deferred cardiac rehabilitation (Phase 2)?”

5.7.2. Clinical Methodological introduction

The literature was searched from 1999 to 2009 for systematic reviews, RCTs, comparative and observational studies comparing early initation of cardiac rehabilitation with deferred cardiac rehabilitation in people with non ST segment elevation ACS. The rationale for searching from January 1999 onwards was to reflect current practice, particularly the use of stents for revascularisation.

The studies of Phase 1 cardiac rehabilitation provided little detail on the exact type of acute coronary syndrome to which they refer – only that the people had been admitted with MI. As a result, the requirement for a NSTEMI/UA population > 60% was relaxed. The key was that studies had to address early initation of cardiac rehabilitation in an ACS population. Studies were included therefore, if the population had ACS and if the intervention (education, counselling, early mobilisation, discharge planning) occurred in hospital prior to discharge. Outcomes of interest included 30 day and long-term survival, revascularisation, re-infarction, LV function, quality of Life, serious complications (e.g. stroke, GI bleed), therapy concordance, well-being, anxiety, depression, and risk factor profile.

One systematic review of 26 studies (16 controlled clinical trials CCT; 10 before and after studies) compared in-hospital intervention with no in-hospital intervention in people with ACS (STEMI, NSTEMI, or UA). The primary outcome was one year mortality, and secondary outcomes were re-admission rates, smoking cessation, and re-infarction ²⁷¹. This systematic review, while well-conducted, may be difficult to interpret In order to be included, a trial had to have at least an in-hospital intervention that directly targeted the patient (such as education or counselling) However, trials could also be included if the intervention was an in-hospital healthcare provider intervention that tried to change attitudes/knowledge of healthcare providers such as improving physician's skills in counselling through an educational program or education/reminders on benefits of specific therapies. Trials could also be included if the intervention was an in-hospital system-level intervention that involved a global change in the organisation of care (such as critical pathways or facility outcome reporting). The systematic review therefore includes at least a patient-level intervention, with some interventions operating additionally at the provider and/or system levels. Interpretation of the results of this meta-analysis should be tempered by the fact that before and after studies are not randomised. In the before and after studies, outcomes following an intervention (implementation of an in-hospital rehabilitation program, for example) were compared with a control group of people who did not receive the intervention (a historical control cohort).

One open RCT (N=65; 3 months follow-up) randomised people hospitalised for a first-time MI to in-hospital psychological intervention plus standard MI educational material or to standard care involving cardiac rehabilitation nurse in-hospital visits plus standard MI educational material (control). The outcomes were patient perception of illness, angina pain post-discharge, and time to return to work. This study is limited by the small number of participants, short follow-up, and use of mail-in questionnaires ²⁷².

One patient survey was conducted with 20 MI people within 72 hours of their intended discharge from the hospital. In a questionnaire format, people were asked to indicate the importance of 40 information needs. This study is limited by the small sample size, and is most relevant to English-speaking people with an uncomplicated MI ²⁷³.

5.7.3. Clinical evidence statements

5.7.4. Health economic methodological introduction

No economic analyses were identified that examined early psychosocial interventions (in-hospital counselling and patient education, phase 1 cardiac rehabilitation) in an UA or NSTEMI population.

5.7.5. Evidence summary

An extensive literature search returned 1022 possible publications, though all but three were excluded. The most common reasons for exclusion were uncertainty regarding the patient population studied, lack of clarity regarding the time of initiating intervention or low quality of evidence. Of the three that were critically appraised all involved some form of intervention prior to hospital discharge, but none was a randomised comparison between timing of initiation and an initiation of the intervention after discharge from hospital.

In the meta–analysis of Auer et al²⁷¹ most of the studies reviewed were published before the widespread use of PCI, when hospital lengths of stay were much longer. It was also difficult to determine the proportion of people with NSTEMI versus STEMI, making its applicability to the present guideline uncertain. Also, to be included in the meta-analysis studies had to include some in-hospital intervention at a direct patient level, but there could also be interventions that could operate at a hospital level (changing physician practice, or hospital processes). With these caveats the systematic review did suggest that in-hospital intervention reduces the rate of readmission to hospital (but not reinfarction), and resulted in greater smoking cessation. However, when only controlled clinical trials were meta-analysed, there were non-significant differences between in-hospital intervention and no in-hospital intervention for mortality at one year, readmission rates, or re-infarction rates.

One RCT²⁷² compared formal and structured in-hospital psychological intervention in addition to standard educational input, with the latter alone. The study was limited by small number of people included (65 in total) but did show that people receiving psychological intervention felt better prepared for discharge from hospital, had a better understanding of the issues, and had more positive attitudes to the consequences of their myocardial infarct, both at discharge and at 3 month follow-up. They also had a lower frequency of angina at 3 months (14.3% for the psychological group versus 39.3% for the control group).

One survey of people's information needs before hospital discharge²⁷³ demonstrated that people following MI rated receiving information about medication, potential complications, and relevance of symptoms most highly.

5.7.6. Evidence to recommendations

The GDG acknowledged the limitations of the evidence that specifically looked at whether rehabilitation should be initiated early in hospital compared to deferred cardiac rehabilitation. However, an important assumption is made that rehabilitation is initiated after discharge, an assumption that is currently not justified given the patchy nature of rehabilitation services that exists across the country.

The GDG agreed that:

• Good evidence exists for the longer term benefits of a comprehensive rehabilitation process following MI. The post-MI guideline found rehabilitation to be cost effective and the GDG felt that this is good evidence that rehabilitation is cost effective in general.
• Recent NICE guidance ⁴ recommends that people with MI should receive formal rehabilitation and delivery of secondary prevention measures and they do not distinguish between people with NSTEMI and STEMI.
• Although no evidence exists specifically for people with UA, it is part of the same pathophysiological continuum as NSTEMI and so the recommendations would logically apply to both groups.
• It is vital that information and education is delivered in an appropriate format to people prior to discharge from hospital given the importance of establishing people on appropriate medication, and the value of people understanding the indications and actions of these medications, and the underlying nature of their cardiac condition and any effect of co-morbidity.
• Given the continuing importance of education, psychological support and a structured, graded exercise programme after discharge from hospital, systems must be in place to ensure that people are ‘picked up’ by the appropriate rehabilitation services on their return to the community, and that hospitals should work with their primary care colleagues to ensure continuity of care.
• With hospital lengths of stay tending to shorten the time available to deliver appropriate pre-discharge information, ensure adequate discharge planning, and ensure continuity of care in the community is very short Systems need to be put in place to ensure that with the understandable emphasis on returning people home as quickly as possible, the elements of comprehensive rehabilitation that can, and should, be delivered in-hospital should not be overlooked.
• The patient representatives on the GDG stressed very strongly the importance of patient information and education before discharge from hospital, and the need for this to be comprehensive, yet in a form that is appropriate to the individual given ethnic, cultural, gender and psychological differences.
• “Rehabilitation”, in its most general sense, actually starts from the moment of diagnosis because from this time onwards there is potential benefit to people from being well informed and psychologically supported, and therefore the distinction between in-hospital and post-discharge intervention is somewhat arbitrary. The overriding consideration should be to ensure that the process is continuous and that responsibility for delivery of the components of rehabilitation (education, information, psychosocial support, structured exercise etc.) should be clearly attributed.

In conclusion, the GDG were unable to draw evidence-based conclusions specifically regarding the optimum time of delivery of educational and psychosocial intervention. However, the GDG agreed with the cardiac NSF which highlights that ‘cardiac rehabilitation should begin as soon as possible after someone is admitted to hospital with CHD (Phase 1)’ and as such made a consensus recommendation in support of this.

5.7.7. Recommendations

R34.

Before discharge offer patients advice and information about:

• their diagnosis and arrangements for follow-up (in line with 'MI: secondary prevention', NICE clinical guideline 48)
• cardiac rehabilitation (in line with 'MI: secondary prevention', NICE clinical guideline 48)
• management of cardiovascular risk factors and drug therapy for secondary prevention (in line with 'MI: secondary prevention', NICE clinical guideline 48, and 'Lipid modification', NICE clinical guideline 67)
• lifestyle changes (in line with 'MI: secondary prevention', NICE clinical guideline 48).

R35.

Cardiac rehabilitation should be equally accessible and relevant to all patients after an MI, particularly people from groups that are less likely to access this service. These include people from black and minority ethnic groups, older people, people from lower socioeconomic groups, women, people from rural communities and people with mental and physical health comorbidities. (This recommendation is from ‘MI: secondary prevention’, NICE clinical guideline 48.)

R36.

All patients who smoke should be advised to quit and be offered support and advice, and referral to an intensive support service (for example, the NHS Stop Smoking Services) in line with 'Brief interventions and referral for smoking cessation in primary care and other settings' (NICE public health guidance 1). (This recommendation is adapted from ‘MI: secondary prevention’, NICE clinical guideline 48.)