Search

We used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, a checklist of 27 items that ensure the quality of systematic review or meta-analysis. ^{Reference Moher, Liberati, Tetzlaff and Altman40} Inclusion criteria were studies of patients with acute coronary syndromes (with data on rates of subsequent invasive coronary procedures or mortality) which reported data for those with defined mental illness as well as those without mental ill health. We only included those with severe mental illness provided there was a subgroup with schizophrenia. We excluded non-comparative studies, and also those concerning depression or anxiety only. We searched Medline/PubMed and EMBASE abstract databases from inception to 20 July 2010. In these databases the keywords/MeSH terms (‘ACS or REVASCULARISATION or REVASCULARIZATION or PROCEDURES or GRAFT or ANGIOPLASTY or PERCUTANEOUS or CATHETERIZATION or CARDIA* or HEART’) and (‘PSYCHI* or MENTAL or PSYCHOSIS or PSYCHOTIC or SCHIZOPHR* or SEVERE MENTAL ILLNESS or SMI’) were used. In addition, four full text collections were searched: ScienceDirect, Ingenta Connect, SpringerLink and Wiley Online Library. In these online databases the same search terms were used as a full text search and as a citation search. The abstract databases Web of Knowledge and Scopus were searched, using the above terms as a text word search, and using key papers in a reverse citation search. Finally, a number of journals were hand-searched and several experts contacted. We excluded studies where the event rate was measured in the general population rather than in those with cardiovascular disease. ^{Reference Lawrence, Holman, Jablensky and Hobbs41} Where 30-day and 365-day mortality rates were cited, we examined only 365-day rates. Data were extracted using a standard form (available from the authors on request) by A.J.M. and checked by D.L.

Meta-analysis

We used summary meta-analysis, pooling hazard ratios where reported. All hazard ratios were entered adjusted rather than unadjusted (where reported and except where indicated). Odds ratios were converted into relative risks (hazard ratios) using the reported control event rate. ^{Reference Zhang and Yu42} Confidence intervals were obtained from all studies or calculated from the data provided. Heterogeneity was reduced by stratifying using type of mental illness and procedure type; despite this, heterogeneity (defined by I ^{Reference Murray and Lopez2} ) remained high and so random effects meta-analysis was preferred. We required a minimum of three independent studies to justify pooling by procedure type. Any potential sources of bias were reported. Publication bias was assessed using the Begg–Mazumdar statistic. ^{Reference Begg and Mazumdar43}

Results

The initial PubMed search generated 572 hits. Of these, only 16 discussed cardiac procedures and 58 discussed mortality (online data supplement). Searches in the four full-text collections generated 1572 hits (Fig. 1). Using these strategies we identified 241 references of interest but only 74 were primary data studies. After excluding studies with no relevant outcome, no comparison group or other methodological issues, we identified 9 papers relating to cardiac procedures following acute coronary syndrome and 6 papers relating to mortality following acute coronary syndrome.

Studies examining procedure rate after acute coronary syndrome

Our search identified 9 publications relating to cardiac procedures following acute coronary syndrome and contained in these reports were 22 analyses using broadly defined mental disorder or severe mental illness and 10 using an adequate definition of schizophrenia or related psychosis (online Table DS1). There was no evidence of publication bias using the Begg–Mazumdar statistic (Fig. DS1). The total sample size was 825 754 (mean 91 750, s.d. = 120 158). Druss et al examined cardiovascular care following an acute myocardial infarction. ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz44} After adjusting for demographic, clinical, hospital and regional factors, those with mental disorders were only 41% (for schizophrenia) to 78% (for substance use) as likely to undergo cardiac catheterisation as those without mental disorder. In a further study, Druss et al found patients hospitalised for myocardial infarction with mental health diagnoses were less likely to have reperfusion conducted. ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz45} Young & Foster identified people with mental illness who had experienced a myocardial infarction: this group had significantly lower levels of all three revascularisation procedures – cardiac catheterisation, percutaneous transluminal coronary angioplasty (PTCA) and coronary artery bypass graft (CABG) – compared with those without mental illness, with the lowest rates seen in those over 64 years old. ^{Reference Young and Foster46} Petersen et al examined the records of 4340 male veterans discharged after a clinically confirmed myocardial infarction: those with mental illness were less likely to have undergone in-patient diagnostic angiography (age-adjusted RR = 0.90, 95% CI 0.83–0.98) but there was no difference in CABG. ^{Reference Petersen, Normand, Druss and Rosenheck47} Kisely et al carried out a population-based record-linkage analysis of related data from 1995 through 2001 compared with the general public for each outcome (n = 215 889): in psychiatric in-patients the adjusted rate ratios for cardiac catheterisation, PTCA and CABG were 0.41, 0.22 and 0.34 respectively. ^{Reference Kisely, Smith, Lawrence, Cox, Campbell and Maaten48} However, Plomondon et al found no difference in cardiac procedure rates after acute coronary syndromes presenting to Veterans Health Administration (VHA) hospitals. ^{Reference Plomondon, Ho, Wang, Greiner, Shore and Sakai49} Similarly, Jones & Carney found no difference in the rates of revascularisation. ^{Reference Jones and Carney50} Laursen et al followed 605 649 patients admitted with heart disease in Denmark between 1994 and 2007: people with admissions for severe mental disorder had higher mortality rates from heart disease following cardiac procedures but received lower rates of cardiac revascularisation. ^{Reference Laursen, Munk-Olsen, Agerbo, Gasse and Mortensen51} Abrams et al examined the rate of PCI or CABG within 30 days of admission. ^{Reference Abrams, Vaughan-Sarrazin and Rosenthal52} They examined mixed mental disorders, defined by ICD–9 codes, in 21 745 patients admitted following acute myocardial infarction. They reported results in two related samples: patients with psychiatric comorbidity had lower receipt of coronary revascularisation (hazard ratio, HR = 0.92, 95% CI 0.85–0.99) in the out-patient sample but equal rates (HR = 1.00, 95% CI 0.91–1.1) in the in-patient sample. ^{Reference Abrams, Vaughan-Sarrazin and Rosenthal52}

Fig. 1 Study selection. CABG, coronary artery bypass graft; PCI, percutaneous coronary intervention; PTCA, percutaneous transluminal coronary angioplasty.

Quantitative differences in procedure rates

For those with mental illness or severe mental illness there was significant heterogeneity (I = 98.1%). The meta-analytic random effects size was 0.86 (95% CI 0.80–0.92, P<0.0001), suggesting that those with any mental illness received a 14% lower rate of cardiac procedures (Fig. 2). Looking at each procedure individually, there was significantly lower receipt of CABG (RR = 0.85, 95% CI 0.72–1.00), cardiac catheterisation (RR = 0.85, 95% CI 0.76–0.95) and PTCA/PCI (RR = 0.87, 95% CI 0.72–1.05). For those with schizophrenia there was moderate heterogeneity (I = 77.6%). The meta-analytic relative risk was 0.53 (95% CI 0.44–0.64, P<0.0001), suggesting that those with schizophrenia received about half the comparable rate of cardiac procedures (Fig. 3). Looking at each procedure individually, there was significantly lower receipt of CABG (RR = 0.69, 95% CI 0.55–0.85) and PTCA/PCI (RR = 0.50, 95% CI 0.34–0.75).

Studies examining mortality after acute coronary syndrome

We identified ten studies relating to mortality following cardiac events but two reported mortality as the general population rate, one had a significant methodological issue and one had insufficient data for analysis. The total sample size from six valid studies was 596 368 (mean 99 394, s.d. = 132 344). In Druss et al's 2000 study, patients with mental disorders had a small but statistically significantly lower risk of mortality at baseline, and in unadjusted analysis 12.8% of those with schizophrenia died within 30 days compared with 10.8% in the comparator population; however, this was not significant after adjustments. ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz44} Yet in their replication study, Druss et al found that mental disorder of all types was associated with a 19% increase in mortality at 1 year. ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz45} Importantly, when the five quality indicators were added to the model the association was no longer significant, suggesting that elevated mortality is related to poor quality of care. Petersen et al examined the records of 4340 male veterans discharged after a clinically confirmed myocardial infarction and found a trend towards higher rate of death at 1 year in those with mental illness; the risk of death within 1 year was 1.25 (95% CI 1.00–1.53). ^{Reference Petersen, Normand, Druss and Rosenheck47} Plomondon et al studied 14 194 patients (including 18% with severe mental illness) with acute coronary syndromes presenting to VHA hospitals between October 2003 and September 2005. ^{Reference Plomondon, Ho, Wang, Greiner, Shore and Sakai49} One-year mortality was lower for patients with severe mental illness (15.8% v. 19.1%, P<0.001). However, in multivariable analysis there was no significant difference in mortality (HR = 0.91, 95% CI 0.81–1.02) between patients with and without severe mental illness. ^{Reference Plomondon, Ho, Wang, Greiner, Shore and Sakai49}

Four additional studies were noteworthy but could not be entered into the meta-analysis. Young & Foster found that in the older cohort (≥65 years old) with mental illness there was a 21% lower risk-adjusted likelihood of death (P<0.001) compared with those without mental illness. ^{Reference Young and Foster46} In the younger cohort those with schizophrenia and substance misuse had higher in-patient mortality rates (both P<0.001). Unfortunately, data were inadequately reported for extraction. Kisely et al examined mortality and revascularisation in a general population sample. ^{Reference Kisely, Smith, Lawrence, Cox, Campbell and Maaten48} They conducted a population-based record-linkage analysis of related data from 1995 through 2001 compared with the general population for each outcome (n = 215 889): the age-standardised mortality rate ratio for psychiatric patients was 1.31 (95% CI 1.25–1.36). In the study from Lawrence et al, ischaemic heart disease was the major cause of excess mortality in psychiatric patients. ^{Reference Lawrence, Holman, Jablensky and Hobbs41} The standardised mortality rate from ischaemic heart disease in mental health users was almost twice that in the overall population (1.91 in total ischaemic heart disease, 1.74 in acute myocardial infarction). However, in the latter two studies the risk was measured at the population level, not specifically in those with acute coronary syndrome. Recently, Blecker et al examined mortality in 341 individuals with heart failure and severe mental illness compared with 1460 with heart failure and no severe mental illness. ^{Reference Blecker, Zhang, Ford, Guallar, Dosreis and Steinwachs53} Mortality was 29.9% in the severe mental illness group compared with 31.7% in those without severe mental illness. However, we excluded this study from the meta-analysis because it was not clear whether the follow-up period was identical in the two groups.

Fig. 2 Meta-analytic summary for receipt of cardiac procedures in patients with mental ill health v. those with no mental illness. CABG, coronary artery bypass graft; CC, cardiac catheterisation; IP, in-patient; OP, out-patient; PTCA, percutaneous transluminal coronary angioplasty; PCI, percutaneous coronary intervention.

Fig. 3 Meta-analytic summary for receipt of cardiac procedures in patients with schizophrenia v. those with no mental illness. CABG, coronary artery bypass graft; CC, cardiac catheterisation; PTCA, percutaneous transluminal coronary angioplasty.

Fig. 4 Meta-analytic summary of mortality rates following acute coronary syndrome in patients with mental illness v. those with no mental illness.

Quantitative differences in mortality following acute coronary syndrome

For those with severe mental illness there was significant heterogeneity (I = 91.6%), therefore random effects meta-analysis was preferred. The pooled relative risk for mortality was 1.15 (95% CI 1.02–1.29). Excluding the studies by Kisely et al and Lawrence et al which did not focus on acute coronary syndrome, the pooled relative risk for mortality was 1.11 (95% CI 1.00–1.24; P = 0.05), suggesting an 11% increase in mortality rates (Fig. 4).

Possible mechanisms underlying elevated mortality rates

In the general population prolonged QT_c interval and low heart rate variability have been associated with increased cardiovascular mortality and sudden death, particularly in people with prior cardiovascular disease and diabetes. ^{Reference Montanez, Ruskin, Hebert, Lamas and Hennekens61,Reference Elming, Brendorp, Kober, Sahebzadah and Torp-Petersen62} Schizophrenia appears to be associated with similar QT abnormalities, possibly even in the absence of antipsychotic medication, and it is possible that these factors are influential following acute coronary syndrome. ^{Reference Bär, Koschke, Boettger, Berger, Kabisch and Sauer63–Reference Jindal, Keshavan, Eklund, Stevens, Montrose and Yeragani65} It is also well known that metabolic syndrome and diabetes increase the risk of mortality, ^{Reference Lakka, Laaksonen, Lakka, Niskanen, Kumpusalo and Tuomilehto66,Reference Ray, Rao, Seshasai, Sivakumaran, Nethercott and Preiss67} and these conditions are often more common in those with severe mental illness. There is also a higher rate of sudden cardiac death in those taking antipsychotics; ^{Reference Ray, Chung, Murray, Hall and Stein68} that said, sudden cardiac death accounts for only a small proportion of excess mortality in schizophrenia. ^{Reference Mitchell69} Unfortunately none of the studies cited here examined use of psychotropic drugs. A second mechanism might be the influence of severe mental illness on the effectiveness of cardiac treatment, for example through low engagement in rehabilitation. Preliminary evidence suggests similar uptake but lower completion of physical exercise programmes in those with known mental ill health. ^{Reference Crone, Johnston, Gidlow, Henley and James70} Confounding factors such as alcohol and drug use and medical comorbidities may also be influential. Most concerning is whether deficits in quality of care influence high mortality in this population. Li et al analysed New York's publicly released Cardiac Surgery Report of surgeons’ risk-adjusted mortality rates. ^{Reference Li, Glance, Cai and Mukamel71} After adjustments, patients with both substance use and psychiatric disorders were more likely to receive care from surgeons in the high-mortality group (OR = 1.76, P = 0.024). Druss et al found that the excess mortality following myocardial infarction was negated when five quality measures were added to the model, suggesting that poor quality of care may be an important explanatory variable. ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz45} More recently, Copeland et al analysed whether patients’ reduced primary care use over time was a significant predictor of mortality over a 4-year period among VHA patients; those with schizophrenia were likely to have low primary care use decreasing with time, and this was linked with inferior survival. ^{Reference Copeland, Zeber, Wang, Henley and James72}

Despite the large sample size we acknowledge several limitations in this meta-analysis. Our results may not be representative of all healthcare systems as all studies following acute coronary syndrome originated in the USA. It is not clear that inequalities would exist in other healthcare systems where there are fewer barriers to care for those who are most socially disadvantaged. One limitation is that the definition of severe mental illness varied in some studies, and although most used ICD–8 or ICD–9 coding the definition employed by one group was not entirely clear. ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz44} A second limitation is that all studies were retrospective and none distinguished current from historical mental ill health and thus it is uncertain if risk applies to those with prior as well as current diagnoses. A third limitation in relation to mortality was that the studies by Lawrence et al and Kisely et al reported risk in population samples not specifically following acute coronary syndrome and were therefore excluded from the analysis. ^{Reference Lawrence, Holman, Jablensky and Hobbs41,Reference Kisely, Smith, Lawrence, Cox, Campbell and Maaten48} Additionally, Druss et al and Young & Foster did not report 1-year mortality rates, ^{Reference Druss, Bradford, Rosenheck, Radford and Krumholz44,Reference Young and Foster46} adding to the heterogeneity in mortality analysis. Finally, there was poor data quality regarding hospitalisation and patient–provider factors underlying poor outcomes.

Improving inequalities in cardiac care

The disparities in cardiac care noted here are consistent with the wider literature documenting disparities in treatment of other medical domains including diabetes, general medicine and cancer care. ^{Reference Mitchell, Malone and Carney Doebbeling12,Reference Desai, Rosenheck, Druss and Perlin73,Reference Mateen, Jatoi, Lineberry, Aranguren, Creagan and Croghan74} Individuals with schizophrenia and severe mental illness receive as little as half of the monitoring offered to people without mental ill health. ^{Reference Roberts, Roalfe, Wilson and Lester75,Reference Vahia, Diwan, Bankole, Kehn, Nurhussein and Ramirez76} For example, in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study, patients with schizophrenia had limited access to or received suboptimal medical care. ^{Reference Nasrallah, Meyer, Goff, McEvoy, Davis and Stroup77} In a retrospective analysis of 1998–2003 Medicaid claims, fewer than 20% of people starting treatment with antipsychotic medication received baseline glucose testing and fewer than 10% received baseline lipid testing. ^{Reference Morrato, Newcomer, Allen and Valuck78} Unfortunately, professional responsibility for comorbid medical disorders is often unclear. Poor mental health status is linked with poor general practitioner accessibility and perceived barriers to medical treatment. ^{Reference Al-Mandhari, Hassan and Haran79} Hence medical comorbidity often is overlooked in those with severe mental illness, with up to half of all chronic conditions remaining unrecognised. ^{Reference Kilbourne, McCarthy, Welsh and Blow80–Reference Bernardo, Banegas, Canas, Casademot, Riesgo and Varela84} Yet there is great interest in effective interventions that might reduce cardiovascular mortality in schizophrenia as well as in mental illness in general. Several groups have developed screening and monitoring guidelines. ^{85–Reference Waterreus and Laugharne88} However, implementation of these has been inconsistent. ^{Reference Citrome and Yeomans89–Reference Mackin, Bishop and Watkinson93} In the general population lifestyle interventions can significantly influence mortality. ^{Reference Iestra, Kromhout, van der Schouw, Grobbee, Boshuizen and van Staveren94} In diabetes glycaemic control has similar benefits. The magnitude of the effect of intensive glycaemic control in diabetes is of the same order as the excess risk documented here. ^{Reference Ray, Rao, Seshasai, Sivakumaran, Nethercott and Preiss67} Interventions specifically targeting weight control and eating habits in people with chronic mental illness have shown some promising results. ^{Reference Khazaal, Fresard, Rabia, Chatton, Rothen and Pomini95–Reference Ganguli99} However, it is uncertain if benefits are maintained and whether there is any measurable effect on mortality. Ultimately it has been suggested that a reorganisation of mental health services would help redefine responsibility for physical health. ^{Reference Horvitz-Lennon, Kilbourne and Pincus100} There is some support for a collaborative model of care, co-locating psychiatric and primary care. ^{Reference Dombrovski and Rosenstock101} Yet to improve cardiac care in severe mental illness, interventions must be effective at the secondary-care level for hospital specialists with limited interest in mental illness.

In conclusion, following cardiac events individuals with mental illness appear to receive about 14% less frequent therapeutic cardiac procedures (47% in the case of schizophrenia), and they have about an 11% increased mortality rate. Further work is required to explore whether these factors are causally linked and whether improvements in medical care might improve survival in those with mental ill health.