Schizophrenia Research 159 (2014) 176–181
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Standard cardiovascular disease risk algorithms underestimate the risk of cardiovascular disease in schizophrenia: Evidence from a national primary care database Gary McLean a,⁎, Julie Langan Martin b, Daniel J. Martin b, Bruce Guthrie c, Stewart W. Mercer a, Daniel J. Smith b a b c
Institute of Health and Wellbeing, University of Glasgow, 1 Horselethill Road, Glasgow G12 9LX, UK Institute of Health and Wellbeing, University of Glasgow, Gartnavel Royal Hospital, 1055 Great Western Road, Glasgow G12 0XH, UK Quality, Safety and Informatics Research Group, University of Dundee, Mackenzie Building, Kirsty Semple Way, Dundee DD2 4BF, UK
a r t i c l e
i n f o
Article history: Received 19 February 2014 Received in revised form 14 July 2014 Accepted 14 July 2014 Available online 13 August 2014 Keywords: Cardiovascular disease Risk scores Schizophrenia Cardiovascular risk factors
a b s t r a c t Background: Schizophrenia is associated with increased cardiovascular mortality. Although cardiovascular disease (CVD) risk prediction algorithms are widely in the general population, their utility for patients with schizophrenia is unknown. Methods: A primary care dataset was used to compare CVD risk scores (Joint British Societies (JBS) score), cardiovascular risk factors, rates of pre-existing CVD and age of first diagnosis of CVD for schizophrenia (n = 1997) relative to population controls (n = 215,165). Results: Pre-existing rates of CVD and the recording of risk factors for those without CVD were higher in the schizophrenia cohort in the younger age groups, for both genders. Those with schizophrenia were more likely to have a first diagnosis of CVD at a younger age, with nearly half of men with schizophrenia plus CVD diagnosed under the age of 55 (schizophrenia men 46.1% vs. control men 34.8%, p b 0.001; schizophrenia women 28.9% vs. control women 23.8%, p b 0.001). However, despite high rates of CVD risk factors within the schizophrenia group, only a very small percentage (3.2% of men and 7.5% of women) of those with schizophrenia under age 55 were correctly identified as high risk for CVD according to the JBS risk algorithm. Conclusion: The JBS2 risk score identified only a small proportion of individuals with schizophrenia under the age of 55 as being at high risk of CVD, despite high rates of risk factors and high rates of first diagnosis of CVD within this age group. The validity of CVD risk prediction algorithms for schizophrenia needs further research. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Rates of cardiovascular disease (CVD) morbidity and mortality have fallen substantially across the developed world in the recent years (Nichols et al., 2013; World Health Organisation, 2013). One factor contributing to this is the effective primary prevention of cardiovascular disease resulting from the use of accurate cardiovascular risk assessment algorithms (Anderson et al., 1991a). Within the UK, the most commonly used CVD risk score is the Joint British Societies Score (JBS2), which combines the Framingham Coronary Heart Disease (CHD) and Stroke algorithms, to calculate CVD risk (British Cardiac Society et al., 2005). Individuals with a risk score for a cardiovascular event of N20% within the next 10 years are considered eligible for primary prevention intervention strategies (British Cardiac Society et al., 2005). These strategies include lifestyle changes with a focus on weight reduction, dietary
⁎ Corresponding author. E-mail address: [email protected] (G. McLean).
http://dx.doi.org/10.1016/j.schres.2014.07.022 0920-9964/© 2014 Elsevier B.V. All rights reserved.
advice, increased exercise and smoking cessation, as well as the consideration for drug therapy (SIGN, 2007). However, observed reductions in rates of CVD morbidity and mortality within the general population have not occurred for patients with schizophrenia. Furthermore, evidence suggests that this mortality gap may be widening (Morden et al., 2012; Lawrence et al., 2013). A systematic review assessing the accuracy of the Framingham CVD algorithm identified considerable variation in the performance of the risk score between different clinical populations (Brindle et al., 2006). While it is known that such risk algorithms have limitations in certain chronic diseases such as rheumatoid arthritis (Crowson et al., 2012), little is known about their usefulness for schizophrenia and how this may relate to risk and pre-existing levels of CVD within different age groups. We recently found that individuals with schizophrenia had lower rates of recorded CVD than controls within a large primary care sample (Smith et al., 2013). Given the high rates of cardiovascular mortality in this cohort, this finding might suggest a systematic under-recognition of cardiovascular disease in schizophrenia. Reasons for this are likely to be multifactorial but it is possible that individuals with schizophrenia
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
are not being identified by primary care physicians as being at high risk of cardiovascular disease, either because they are not having their cardiovascular risk factors assessed, or because existing risk calculators do not effectively identify this clinical sub-group as being at high risk of CVD. Although the prevention of CVD requires early identification and recording of risk factors, there has to date been very little study of how levels of recording of risk factors within primary care might differ between patients with schizophrenia compared to the general population (Ketola et al., 2006). Further, it is known that cardiovascular risk increases with age and, as such, age is heavily weighted within the JBS2 risk score (The Joint British Societies, 2006). People with schizophrenia tend to develop CVD (and have higher levels of CVD risk factors) at a much younger age than the general population (Holt, 2013). For example, compared to the general population, they have much higher rates and greater intensity of smoking (Hennekens et al., 2005; Myles et al., 2012), higher blood cholesterol (Redelmeier et al., 1998), and higher rates of diabetes (Sukdeb, 1995; Bai et al., 2013) and metabolic syndrome (Lee et al., 2012). They are also more likely to have risk factors, such as obesity (which may be an independent risk factor because of psychotropic medication) (Allison et al., 1999a; Ratliff et al., 2013), which are not currently included within risk calculators because they have not been found to be independent predictors of CVD risk for the general population. Antipsychotic medication in particular is an important additional factor for CVD risk. It is recognised that antipsychotics increase the total and low-density lipoprotein (LDL) cholesterol, as well as increasing triglycerides and decreasing high-density lipoprotein (HDL) cholesterol (Meyer, 2001). Although there has been some work on the use of Framingham risk scores to estimate CVD risk in people with schizophrenia (Goff et al., 2005; Slomka et al., 2012; Tay et al., 2013) — including by age (Jin et al., 2011), and antipsychotic prescribed (Barak et al., 2005; Daumit et al., 2008) — there have to date been no large scale studies from primary care databases which have examined how estimated CVD risk in people with schizophrenia compares to the general population by age and gender. 2. Methods We used data from the Primary Care Clinical Informatics Unit database (University of Aberdeen), which consisted of all 1,751,841 registered people who were alive and permanently registered with 314 general practices on March 31, 2007. This dataset is a representative sample covering approximately one third of the Scottish population. A more detailed explanation of the dataset is available elsewhere (Barnett et al., 2012). People were identified as having ‘schizophrenia or related nonorganic psychosis’ (from here referred to as ‘schizophrenia’) based on the recording ever of any of the following primary care Read Codes (where % is noted this means “this code and any below it in the code hierarchy”): E10% schizophrenic disorders; E121 chronic paranoid psychosis; E12z paranoid psychosis NOS; E13% other non-organic psychoses; E13z non-organic psychosis/psychotic episode; NOS; E1z non-organic psychosis NOS; Eu20% schizophrenia; Eu22% persistent delusional disorder; or the recording in the last 12 months of Eu23% acute/transient psychotic disorder. Deprivation status was measured using the Carstairs deprivation score, which is widely used in health research (Carstairs, 1991) and estimates of the 10-year risk for Framingham risk scores were generated for each participant. Joint British Societies (JBS2) scores were calculated based on the 1991 Framingham equations (Anderson et al., 1991b). The JBS2 score calculates 10 year CVD risk as the sum of CHD and stroke risk using the relevant Framingham equations containing the risk factors; age, sex, systolic BP, smoking status, diabetes, total cholesterol and HDL cholesterol. As HDL cholesterol figures were not present in the data, we used estimated levels taken from the Scottish Health
177
Survey from 2011 by age group and gender, as used by the ‘ASSIGN’ score (Scottish Government, 2011; Assign, 2013). To comply with JBS guidelines, people were identified as smokers if they had a record of having smoked at any time (British Cardiac Society et al., 2005). The date of first diagnosis for those with a record of pre-existing CVD was also recorded and from this we calculated age at first diagnosis of CVD. 3. Statistical analysis We restricted our analyses to those aged 35 to 74 as this is the age range in which the JBS risk score has been validated. People with a pre-existing diagnosis of CHD, stroke or peripheral vascular disease were excluded (see Table 1). Summary statistics for all study variables were calculated, including means and standard deviations for continuous characteristics and percentages for categorical characteristics. Summary statistics on risk estimates, risk factors, rates of pre-existing CVD and age of first diagnosis of CVD are presented by age and gender for schizophrenia and controls. For all statistical analyses, a p value less than 0.05 was considered statistically significant. All analyses were performed in Stata version 13.0. We also used all variables included in the JBS risk calculation (with the exception of HDL cholesterol) as well as the deprivation and BMI levels in a multivariate regression model using robust standard errors to assess any differences in these variables between schizophrenia and controls, in terms of their association with being identified as high risk under JBS. 4. Results 4.1. CVD rates Table 1 highlights that pre-existing rates of CVD were higher in the 35–44, 45–54 and 55–64 age groups for both men and women with schizophrenia and lower only within the 65–74 age group. 4.2. Data completeness Table 2 shows the number of individuals with missing data for those variables included in the calculation of the JBS score. Age, sex and diagnosis of diabetes were all fully recorded while HDL cholesterol levels were estimated for all individuals. Overall both men (schizophrenia men 27.2% vs. control men 24.5%) and women (schizophrenia women 31.9% vs. control women 25.9%) with schizophrenia were more likely to have data fully recorded. Of the individual variables, most of the missing data were found for the total cholesterol, with around three quarters of individuals having no recorded levels (and levels of missing data higher in the younger age groups). In comparison, only 4.4% of men and 2.2% of women with schizophrenia did not have a record of smoking status and 8.4% of men and 3.4% of women had no systolic blood pressure (SBP) recorded. Rates of individuals with known cardiovascular disease were calculated by age at first diagnosis for the schizophrenia cohort and for controls by gender (Fig. 1). The schizophrenia cohort were more likely to have a first diagnosis at a younger age relative to controls. Although rates in those aged under 55 were higher in men for both schizophrenia and controls, differences between cohorts were larger for men than for Table 1 Pre existing CVD rates by age group and gender. Number (% with CVD) Age group
Schizophrenia Controls Men Men
35–44 45–54 55–64 65–74
15 (1.2) 62 (5.6) 133 (15.9) 126 (27.2)
Schizophrenia Controls Women Women
1171 (0.8) 12 (1.4) 5324 (4.2) 36 (3.7) 14,523 (13.4) 90 (9.6) 23,258 (29.8) 136 (18.4)
1334 (0.9) 3396 (2.7) 8578 (7.9) 16,915 (19.0)
178
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
Table 2 Recording of risk factors used in JBS score by age and gender. Total without CVD
Schizophrenia Controls Men Men
Schizophrenia Controls Women Women
35–44
1260 (37.6)
853 (26.4)
45–54
1049 (31.3)
55–64
703 (21.0)
140,056 (34.1) 121,810 (29.6) 93,812 (22.7)
65–74 Total
337 (10.1) 3349
54,731 (13.3) 411,126
602 (18.6) 3232
117,080 (83.6) 108,897 (89.4) 87,047 (92.8) 52,303 (95.6) 365,327 (89.0)
832 (97.5)
103,296 (73.8) 105,727 (86.8) 86,267 (92.0) 52,014 (95.0) 347,304 (84.6)
812 (95.1)
14,104 (10.1) 27,488 (22.6) 33,867 (36.1) 26,040 (47.6) 101,499 (24.7)
157 (18.4) 274 (29.4) 344 (40.7) 266 (44.2) 1041 (32.2)
13,486 (10.1) 26,059 (21.6) 36,096 (36.0) 35,067 (48.9) 110,708 (26.0)
13,490 (9.6) 27,236 (22.4) 33,742 (36.0) 26,009 (42.4) 100,477 (24.5)
156 (18.3) 269 (28.9) 344 (40.7) 263 (43.7) 1032 (31.9)
13,265 (9.9) 25,942 (21.5) 36,061 (36.0) 35,047 (48.8) 110,315 (25.9)
Smoking status recorded 35–44 1187 (94.2) 45–54
1006 (95.9)
55–64 65–74 Total
676 (96.1) 333 (98.8) 3202 (95.6)
Systolic blood pressure recorded 35–44 1089 (86.4) 45–54
980 (93.4)
55–64 65–74 Total
670 (95.3) 329 (97.6) 3068 (91.6)
Total cholesterol recorded 35–44 229 (18.2) 45–54 290 (27.6) 55–64 256 (36.4) 65–74 145 (43.0) Total 920 (27.4)
Number with data fully recorded 35–44 226 (17.9) 45–54 287 (27.4) 55–64 256 (36.4) 65–74 143 (42.4) Total 912 (27.2)
932 (28.8) 845 (26.1)
916 (98.3) 831 (98.3) 583 (96.8) 3162 (97.8)
898 (96.3) 831 (98.3) 581 (96.5) 3122 (94.9)
133,856 (31.4) 120,592 (28.3) 100,165 (26.1) 71,761 (18.6) 427,521
127,485 (95.2) 116,501 (96.6) 97,437 (97.3) 69,775 (97.2) 411,198 (96.4)
122,894 (91.8) 115,311 (95.6) 96,969 (96.8) 69,455 (96.8) 404,629 (94.9)
Note: Of the other risk factors required for calculation of JBS score age, gender and diabetes were fully recorded while HDL levels were estimated for all people.
women (schizophrenia men 46.1% vs. control men 34.8%; p b 0.001: schizophrenia women 28.9% vs. control women 23.8%; p b 0.001). 4.3. Risk factors We identified 1942 individuals with schizophrenia and 210,792 individuals with no record of schizophrenia (controls). Men and women with schizophrenia who had records of JBS score risk factors were significantly younger than male controls (mean age 52.9 to 56.9; difference −4.0 95% CI −3.4 to −4.7) and female controls (mean age 56.7 to 58.2; difference −1.5 95% CI −0.9 to −2.2) (Table 3). Individuals with schizophrenia were also more likely to smoke, and have diabetes, but had significantly lower mean systolic blood pressure. For risk factors not included in the JBS score, both men and women with schizophrenia were significantly more socially deprived compared to controls and were more likely to have a BMI N 30. (See Table 4.) Fig. 2 shows the percentage of those by JBS risk level, age group and gender for schizophrenia and controls. Age was a major factor being identified as high risk, with 79% of those with schizophrenia aged 65– 74 estimated at high risk in the next 10 years compared with only 1.3% of those aged 35–44. Men were more likely to be identified as high risk for both schizophrenia and controls. This was reflected in all age groups, with only just over 1% of women identified as high risk in
the 45–54 year old age group compared to around 10% of males and only a third of schizophrenia women and a quarter of controls aged 65–74 identified as high risk compared to just under 80% of men for both schizophrenia and controls. The multiple regression model shows that compared to controls, gender (schizophrenia coefficient 6.16 vs. controls 7.17), smoking (schizophrenia coefficient 6.20 vs. controls coefficient 6.86) and diabetes (schizophrenia coefficient 6.62 vs. controls coefficient 7.22) had a smaller effect on being identified as high risk under JBS. In contrast, age had a higher effect for schizophrenia (schizophrenia coefficient 0.54 vs. controls coefficient 0.51). No differences were found for total cholesterol and systolic blood pressure. BMI and deprivation both showed negative associations for schizophrenia and controls, possibly reflecting their exclusion from the JBS algorithm, although results were only statistically significant for controls. 5. Discussion We found that individuals with schizophrenia were more likely to have an early diagnosis of CVD, with over half of men having been first diagnosed under the age of 55. This was not reflected in the CVD risk score estimates, where only a small proportion of those with schizophrenia at risk of CVD were aged b55 years. This highlights a concern that current cardiovascular risk scores might underestimate risk in schizophrenia, particularly in men. As with all cardiovascular risk scores, the absolute risk increases rapidly with age; this leads to a lower identification of younger individuals at high risk and an increased identification of older individuals at low risk (Tunstall-Pedoe, 2011). This issue calls into question the validity of these instruments in conditions such as schizophrenia, where higher mortality rates are a feature at a younger age (Holt, 2013). This could lead to under-prediction of CVD risk and a loss of opportunity for primary prevention interventions, which may contribute to the high rates of cardiovascular disease mortality in individuals with schizophrenia (Smith et al., 2013). From the CVD risk factors included within the JBS2 that were assessed, we found higher rates of smoking in individuals with schizophrenia compared to controls, in keeping with the other studies (Hennekens et al., 2005; Myles et al., 2012). Diabetes prevalence was also significantly higher in those with schizophrenia compared to controls. However, under JBS guidelines, those with diabetes should be treated as at high cardiovascular risk and therefore automatically considered for primary prevention approaches (British Cardiac Society et al., 2005). In contrast to other studies, we found lower levels of systolic blood pressure in individuals with schizophrenia relative to controls which is difficult to account for but which may to some degree reflect adrenergic blockade by antipsychotic medications (Hennekens et al., 2005; Osborn et al, 2008). As an example of potentially important risk factors not included in the JBS score, we found higher rates of obesity (BMI N 30) in individuals with schizophrenia compared to controls. There has been recent evidence to suggest that obesity in people with schizophrenia is associated with an elevated risk for CVD compared to age, gender, race and BMImatched controls (Ratliff et al., 2013). Individuals with schizophrenia were also found to be living in more deprived social circumstances than the general population. Analysis has shown that being in the highest 20% of deprivation rather than the lowest 20% is equivalent in risk score terms to a decade or more in age, or to a concurrent diagnosis of diabetes (Tunstall-Pedoe et al., 2006). In the UK, risk scores such as ASSIGN in Scotland (Assign, 2013) and QRISK in England (Hippisley Cox et al., 2007) have been developed which have incorporated deprivation as a risk factor. However, both give less weight to existing risk factors and have been found to identify fewer individuals in the younger age groups than Framingham, highlighting a further limitation of the current CVD risk prediction algorithms with respect to individuals with schizophrenia (de la Iglesia et al., 2011). However, despite the
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
179
100% 90%
23.90% 32.1%
80% 46.6%
47.4%
% of people
70% 60%
30.0% Age 65 and above 33.0%
50%
Age55-64 25.2%
40% 30%
24.7% 20.1%
17.2%
16.1% 10.1%
0%
Under 45
30%
20% 10%
Age 45-54 28.8%
Schizophrenia Men (n=510)
8.0%
Controls Men (n=73,395)
6.6%
Schizophrenia Control Women Women (n=547) (58,837)
Note: figures represent the percentage of those with established CVD by age of first diagnosis Fig. 1. Rates of known cardiovascular disease for schizophrenia versus controls by age group at first diagnosis. Note: figures represent the percentage of those with established CVD by age of first diagnosis.
identification of high risk factors in schizophrenia, for some risk factors in a multiple regression model only age was found to have a higher effect with respect to being identified as high risk for schizophrenia compared to controls. This may in part be a reflection of higher risk factors in younger patients for those with schizophrenia which are not reflected in CVD risk scores such as JBS due to the weight given to age.
6. Strengths and limitations Our dataset is large, covers all age groups and is representative of the whole Scottish population. We calculated scores for individuals with schizophrenia by age group and sex and compared this to controls, along with age at first cardiovascular disease diagnosis. To our knowledge, this level of detail of JBS2 risk score reporting in people with schizophrenia has not been previously carried out. We were also able to compare the levels of CVD risk estimation with rates of established cardiovascular disease by age and gender in those with schizophrenia
compared to the general population, as a test of the clinical utility of current risk prediction tools. However, our study has some limitations. We used estimated HDL cholesterol levels in all individuals included in our study using age and sex matched means from the 2011 Scottish Health Survey (Scottish Government, 2011) to generate the JBS2 score. Schizophrenia has been associated with lower HDL cholesterol levels than the general population (Huang and Chen, 2005) and by using the mean Scottish levels we may have under predicted cardiovascular risk in individuals with schizophrenia. Around three quarters of the dataset had no information on one or more risk factors and so was excluded from the analysis. Levels of missing data were similar in those with schizophrenia and the general population. It is also possible that people with risk factors recorded were more likely to have a condition included under the Quality and Outcomes Framework (QOF) in the UK, where practices are paid depending on the percentage of people who are successfully controlled for targets such as blood pressure and cholesterol (ISD Scotland, 2013). If this true, then people with missing data may be less likely to have risk
Table 3 Risk factors for cardiovascular disease by gender, schizophrenia versus controls. Variable
Schizophrenia Men
Controls Men
911 (46.9) 52.9 (10.3) 5.2 (1.3)
100,477 (47.7) 56.9 (10.1) 5.2 (1.1)
1.3 (0.4)
Difference (95% CI)
Schizophrenia Women
Controls Women
Difference (95% CI)
−4.0 (−3.4 to -4.7) −0.0 (−0.4 to 0.1)
1031 (53.1) 56.7 (10.5) 5.4 (1.1)
110,315 (52.3) 58.2 (10.5) 5.4 (1.1)
−1.5 (−0.9 to −2.1) −0.0 (−0.1 to 0.1)
1.4 (0.3)
0.0 (−0.1 to 0.0)
1.6 (0.3)
1.6 (0.3)
0.0 (−0.1 to 0.1)
Risk factors (JBS only) Number (%) Age mean (SD) Total cholesterol (mean) HDL cholesterol (mean) Systolic BP (mean) Smoke no. (%) Diabetes no. (%)
130.4 (15.6)
135.5 (15.2)
−5.1 (−3.2 to −6.1)
130.0 (17.0)
133.3 (16.1)
−3.3 (−2.3 to −4.3)
625 (68.9%) 201 (22.1%)
50,959 (51.2%) 18,194 (18.2%)
17.7 (14.4 to 20.9) 3.9 (1.3 to 6.5)
595 (57.9%) 251 (24.3%)
47,316 (43.1%) 15,361 (13.9%)
14.8 (11.6 to17.7) 10.4 (8.2 to 12.5)
Other risk factors Deprivation mean (SD) BMI N 30 no. (%)
0.96 (3.5) 273 (29.9%)
−0.38 (3.3) 25,777 (25.7%)
1.34 (1.13 to 1.55) 4.2 (1.3 to 7.1)
0.36 (3.4) 368 (35.7%)
−0.22 (3.4) 30,983 (28.1%)
0.58 (0.38 to 0.78) 7.6 (4.7 to 10.2)
Note: HDL uses estimated levels taken from the Scottish Health Survey from 2011 by age group and gender.
180
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
Table 4 Regression coefficients for CVD risk factors. Schizophrenia
Controls
Variable
Coefficient
95% CI
Coefficient
95% CI
Gender Age Total cholesterol Systolic BP Smoking Diabetes BMI Deprivation
6.16 0.54 1.93 0.21 6.20 6.62 −0.01 −0.01
5.81 to 6.52 0.52 to 0.56 1.78 to 2.08 0.20 to 0.23 5.83 to 6.56 6.18 to 7.06 −0.02 to 0.00 −0.06 to 0.04
7.17 0.51 1.93 0.21 6.86 7.22 −0.01 −0.01
7.13 to 7.20 0.50 to 0.52 1.91 to 1.94 0.20 to 0.22 6.83 to 6.90 7.17 to 7.27 −0.02 to −0.01 −0.02 to −0.01
factors controlled as the incentive to treat these people is lower for the primary care provider. Changes to the QOF in 2011 incentivised the recording of body mass index, blood pressure, total to HDL cholesterol ratio and blood glucose for people with mental illness potentially leading to improvements in the recording and monitoring of cardiovascular risk factors within this population compared to the levels shown in our study (Holt, 2013). Given the high rates of CVD disease and mortality in those with schizophrenia, it is of concern that NHS England recently decided to withdraw financial incentives through the QOF for monitoring CVD risk factors from 2014 to 2015, although these incentives will remain in place within NHS Scotland. 7. Conclusion In conclusion, we found that JBS2 risk scores identified only a small proportion of individuals with schizophrenia under the age of 55 as being at high risk of cardiovascular disease, alongside high rates of first diagnosis of cardiovascular disease recorded within this age group. This suggests that the current cardiovascular risk prediction algorithms may be under-predicting risk in schizophrenia, particularly for younger men. Existing models used in the UK and elsewhere have ignored schizophrenia or major mental illness in general as a significant predictor (Anderson et al., 1991b; Tunstall Pedoe at al, 1997; Tunstall-Pedoe et al., 2006; Hippisely-Cox et al., 2007). This is likely to contribute to the premature mortality due to cardiovascular disease in this cohort of patients. Similar issues have been found with conditions
such as rheumatoid arthritis (Crowson et al., 2012), which has recently been added to the ASSIGN score in Scotland as a risk factor (ASSIGN, 2013). Further work is therefore required to validate the use of CVD risk scores in individuals with schizophrenia, and possibly to create schizophrenia-specific risk scores, in order to better predict cardiovascular risk. In the meantime, clinicians should obviously continue to manage individual risk factors (such as higher rates of smoking, obesity and diabetes) in this vulnerable clinical group. Role of the funding Source The Chief Scientist Office of the Scottish Government Health Directorates (Applied Research Programme Grant ARPG/07/1); and the Scottish School of Primary Care, partly supported SWM's post and the development of the Applied Research Programme. Contributors Gary McLean carried out the statistical analysis and wrote the first draft. All other authors contributed to subsequent drafts. All authors read and approved the final manuscript. Conflict of interest There are no conflicts of interests. Acknowledgements We thank the Chief Scientist Office of the Scottish Government Health Directorates (Applied Research Programme Grant ARPG/07/1); the Scottish School of Primary Care, which partly supported SWM's post and the development of the Applied Research Programme; and the Primary Care Clinical Informatics Unit at the University of Aberdeen, which provided the data. The views in this publication are not necessarily the views of the University of Aberdeen of University of Glasgow, their agents, or employees. We thank Katie Wilde and Fiona Chaloner of the University of Aberdeen, who did the initial data extraction and management.
References Allison, D.B.,Fontaine, K.R.,Heo, M.,Mentore, J.L.,Cappelleri, J.C.,Chandler, L.P.,Weiden, P.J., Cheskin, L.J., 1999a. The distribution of body mass index among individuals with and without schizophrenia. J. Clin. Psychiatry 60, 215–220. Anderson, K.M., Wilson, P.W., Odell, P.M., Kannel, W.B., 1991a. An updated coronary risk profile: a statement for health professionals. Circulation 83, 356e62. Anderson, K.M.,Odell, P.M.,Wilson, P.W.,Kannel, W.B., 1991b. Cardiovascular disease risk profiles. Am. Heart J. 121, 293–298. ASSIGN, 2013. http://assign-score.com/estimate-the-risk/risk-factors/#more-info (accessed June 2014). Bai, Ya-Mei,Su, Tung-Ping,Chen, Mu-Hong,Chen, Tzeng-Ji,Chang, Wen-Han, 2013. Risk of developing diabetes mellitus and hyperlipidaemia among people with bipolar disorder, major depressive disorder, and schizophrenia: a 10-year nationwide population-
100% 90% 80%
60% 50% 40% JBS
10%
35-44
45-54
55-64 Men
65-74
45-54
55-64
Women
Fig. 2. JBS risk levels by age group and gender.
Controls
Schizophrenia
Controls
Schizophrenia
Controls
Controls
35-44
Schizophrenia
Schizophrenia
Controls
Schizophrenia
Controls
Schizophrenia
Controls
Schizophrenia
Controls
0% Schizophrenia
% of people
70%
65-74
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181 based prospective cohort study. J. Affect. Disord. (ISSN: 0165-0327). http://dx.doi. org/10.1016/j.jad.2013.02.019. Barak, Y., Swartz, M., Plopsky, I., 2005. Assessing cardiovascular risks of olanzapine treatment: a 6-month study versus haloperidol in schizophrenia people. Int. Clin. Psychopharmacol. 20 (6), 315–317. Barnett, K.,Mercer, S.W.,Norbury, M.,Watt, G.,Wyke, S.,Guthrie, B., 2012. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet 380 (9836), 37–43. Brindle, P.,Beswick, D.,Fahey, T.,Ebrahim, S., 2006. The accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart 92, 1752–1759. British Cardiac Society,British Hypertension Society,UK Diabetes,UK HEART,Primary Care cardiovascular Society, The Stroke Association, 2005. JBS 2: Joint British Societies' guidelines on prevention of cardiovascular disease in clinical practice. Heart 91 (Suppl. V), v1–v52. Carstairs, V., 1991. Deprivation and Health in Scotland. Aberdeen University Press, Aberdeen. Crowson, C.S., Matteson, E.L., Roger, V.L., Therneau, T.M., Gabrie, S.E., 2012. Usefulness of risk scores to estimate the risk of cardiovascular disease in people with rheumatoid arthritis. Am. J. Cardiol. 110 (3), 420–424. Daumit, D.C., Goff, J.M., Meyer, V.G., Davis, H.A., Nasrallah, J.P., McEvoy, R., Rosenheck, S.M., Davis, J.K., Hsiao, T.S., Stroup, J.A., 2008. Antipsychotic effects on estimated 10-year coronary heart disease risk in the CATIE schizophrenia study. Schizophr. Res. 105 (1–3), 175–187. de la Iglesia, B., de la Iglesia, B., Potter, J.F., Poulter, N.R., 2011. Performance of the ASSIGN cardiovascular disease risk score on a UK cohort of people from general practice. Heart 97, 491 (e9). Goff, C. Cather, Evins, A.E., Henderson, D.C., Freudenreich, O., Copeland, P.M., Bierer, M., Duckworth, K., Sacks, F.M., 2005. Medical morbidity and mortality in schizophrenia: guidelines for psychiatrists. J. Clin. Psychiatry 66 (2), 183–194. Hennekens, C.H., Hennekens, A.R., Hollar, D., Casey, D.E., 2005. Schizophrenia and increased risks of cardiovascular disease. Am. Heart J. 150 (6), 1115–1121. Hippisley-Cox, .J., Coupland, C., Vinogradova, Y., Robson, J., May, M., Brindle, P., 2007. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. Br. Med. J. 335 (7611), 136–141. Holt, R.I., 2013. Cardiovascular Disease and Diabetes in People With Severe Mental Illness: Causes, Consequences and Pragmatic Management. http://www.rcpsych.ac.uk/pdf/ PCCJ_Holt_FINALONLINE_JAN%5B1%5D.pdf (accessed June 2014). Huang, T.L.,Chen, J.F., 2005. Serum lipid profiles and schizophrenia: effects of conventional or atypical antipsychotic drugs in Taiwan. Schizophr. Res. 80 (1), 55–59. ISD Scotland, 2013. http://www.isdscotland.org/Health-Topics/General-Practice/QualityAnd-Outcomes-Framework (accessed June 2014). Jin, H., Folsom, D., Sasaki, A., Mudaliar, S., Henry, R., Torres, M., Golshan, S., Glorioso, D.K., Jeste, D., 2011. Increased Framingham 10-year risk of coronary heart disease in middle-aged and older people with psychotic symptoms. Schizophr. Res. 125 (2–3), 295–299. Ketola, E., Sipilä, R., Mäkelä, M., Klockars, M., 2006. Quality improvement programme for cardiovascular disease risk factor recording in primary care. Qual. Health Care 9 (3), 175–180. Lawrence, D., Hancock, K., Kisley, S., 2013. The gap in life expectancy from preventable physical illness in psychiatric people in Western Australia: retrospective analysis of population based registers. BMJ 346, f2539. http://dx.doi.org/10.1136/bmj.f2539.
181
Lee, J., Nurjono, M., Wong, A., Salim, A., 2012. Prevalence of metabolic syndrome among people with schizophrenia in Singapore. Ann. Acad. Med. Singap. 41, 457–462. Meyer, J.M., 2001. Novel antipsychotics and severe hyperlipidaemia. J. Clin. Psychopharmacol. 21, 369–374. Morden, N.E., Lai, Z., Goodrich, D.E., MacKenzie, T., McCarthy, J.F., Austin, K., Welsh, D.E., Bartels, S., Kilbourne, A.M., 2012. Eight-year trends of cardiometabolic morbidity and mortality in people with schizophrenia. Gen. Hosp. Psychiatry 34, 368–379. Myles, N., Newall, H.D., Curtis, J., 2012. Tobacco use before, at and after first-episode psychosis: a systematic meta-analysis. J. Clin. Psychiatry 73 (4), 468–475. Nichols, M., Townsend, N., Scarborough, P., Rayner, M., 2013. Trends in age-specific coronary heart disease mortality in the European Union over three decades: 1980–2009. Eur. Heart J. 34, 3017–3027. Osborn, D.P., Wright, C.A., Levy, G., 2008. Relative risk of diabetes, dyslipidaemia, hypertension, and the metabolic syndrome in people with severe mental illnesses: systematic review and meta-analysis. BMC Psychiatr. 8, 84. Ratliff, J.C., Palmese, L.B., Reutenauer, E.L., Srihari, V.H., Tek, C., 2013. Obese schizophrenia spectrum people have significantly higher 10-year general cardiovascular risk and vascular ages than obese individuals without severe mental illness. Psychosomatics 54 (1), 67–73. Redelmeier, D., Siew, H., Booth, G., 1998. The treatment of unrelated disorders in people with chronic medical diseases. N. Engl. J. Med. 338, 1516–1520. Scottish Government, 2011. http://www.scotland.gov.uk/Topics/Statistics/Browse/ Health/TrendMortalityRates (accessed June 2014). SIGN, 2007. SIGN Guideline 97 risk estimation and the prevention of cardiovascular disease. http://www.sign.ac.uk/pdf/sign97.pdf (accessed June 2014). Slomka, J.M., Piette, J.D., Post, E.P., Krein, S.L., Lai, Z., Goodrich, D.E., Kilbourne, A.M., 2012. Mood disorder symptoms and elevated cardiovascular disease risk in people with bipolar disorder. J. Affect. Disord. 138 (3), 405–408. Smith, D.J., Langan, J., McLean, G., Guthrie, B., Mercer, S.W., 2013. Schizophrenia is associated with excess multiple physical-health co morbidities but low levels of recorded cardiovascular disease in primary care: cross-sectional study. BMJ Open 3, e002808. Sukdeb, M., 1995. High prevalence of type II diabetes in schizophrenic people. Schizophr. Res. 15 (1/2), 195. Tay, Y.H.,Nurjono, M., Lee, M.J., 2013. Increased Framingham 10-year CVD risk in Chinese people with schizophrenia. Schizophr. Res. 147 (1), 187–192. The Joint British Societies, 2006. Cardiovascular Risk Assessor (V01.06). http://heartuk.org. uk/health-professionals/resources/risk-calculators/jbs2-cv-risk-assessor (accessed June 2014). Tunstall-Pedoe, H., 2011. Cardiovascular risk and risk scores: ASSIGN, Framingham, QRISK and others: how to choose. Heart 97 (6), 442–444. Tunstall-Pedoe, H., Woodward, M.,Tavendale, R., A'Brook, R.,McCluskey, M.K., 1997. Comparison of the prediction by 27 different factors of coronary heart disease and death in men and women of the Scottish heart health study: cohort study. Br. Med. J. 315 (7110), 722–729. Tunstall-Pedoe, H., Woodward, M., for the SIGN group on risk estimation, 2006. By neglecting deprivation, cardiovascular risk scoring will exacerbate social gradients in disease. Heart 92, 307e1. World Health Organisation, 2013. WHO Mortality Database: 1st May 2013update http:// www.who.int/healthinfo/statistics/mortality_rawdata/en/index.html. World Health Organization, Department of Health Statistics and Information System (accessed June 2014).
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Standard cardiovascular disease risk algorithms underestimate the risk of cardiovascular disease in schizophrenia: Evidence from a national primary care database Gary McLean a,⁎, Julie Langan Martin b, Daniel J. Martin b, Bruce Guthrie c, Stewart W. Mercer a, Daniel J. Smith b a b c
Institute of Health and Wellbeing, University of Glasgow, 1 Horselethill Road, Glasgow G12 9LX, UK Institute of Health and Wellbeing, University of Glasgow, Gartnavel Royal Hospital, 1055 Great Western Road, Glasgow G12 0XH, UK Quality, Safety and Informatics Research Group, University of Dundee, Mackenzie Building, Kirsty Semple Way, Dundee DD2 4BF, UK
a r t i c l e
i n f o
Article history: Received 19 February 2014 Received in revised form 14 July 2014 Accepted 14 July 2014 Available online 13 August 2014 Keywords: Cardiovascular disease Risk scores Schizophrenia Cardiovascular risk factors
a b s t r a c t Background: Schizophrenia is associated with increased cardiovascular mortality. Although cardiovascular disease (CVD) risk prediction algorithms are widely in the general population, their utility for patients with schizophrenia is unknown. Methods: A primary care dataset was used to compare CVD risk scores (Joint British Societies (JBS) score), cardiovascular risk factors, rates of pre-existing CVD and age of first diagnosis of CVD for schizophrenia (n = 1997) relative to population controls (n = 215,165). Results: Pre-existing rates of CVD and the recording of risk factors for those without CVD were higher in the schizophrenia cohort in the younger age groups, for both genders. Those with schizophrenia were more likely to have a first diagnosis of CVD at a younger age, with nearly half of men with schizophrenia plus CVD diagnosed under the age of 55 (schizophrenia men 46.1% vs. control men 34.8%, p b 0.001; schizophrenia women 28.9% vs. control women 23.8%, p b 0.001). However, despite high rates of CVD risk factors within the schizophrenia group, only a very small percentage (3.2% of men and 7.5% of women) of those with schizophrenia under age 55 were correctly identified as high risk for CVD according to the JBS risk algorithm. Conclusion: The JBS2 risk score identified only a small proportion of individuals with schizophrenia under the age of 55 as being at high risk of CVD, despite high rates of risk factors and high rates of first diagnosis of CVD within this age group. The validity of CVD risk prediction algorithms for schizophrenia needs further research. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Rates of cardiovascular disease (CVD) morbidity and mortality have fallen substantially across the developed world in the recent years (Nichols et al., 2013; World Health Organisation, 2013). One factor contributing to this is the effective primary prevention of cardiovascular disease resulting from the use of accurate cardiovascular risk assessment algorithms (Anderson et al., 1991a). Within the UK, the most commonly used CVD risk score is the Joint British Societies Score (JBS2), which combines the Framingham Coronary Heart Disease (CHD) and Stroke algorithms, to calculate CVD risk (British Cardiac Society et al., 2005). Individuals with a risk score for a cardiovascular event of N20% within the next 10 years are considered eligible for primary prevention intervention strategies (British Cardiac Society et al., 2005). These strategies include lifestyle changes with a focus on weight reduction, dietary
⁎ Corresponding author. E-mail address: [email protected] (G. McLean).
http://dx.doi.org/10.1016/j.schres.2014.07.022 0920-9964/© 2014 Elsevier B.V. All rights reserved.
advice, increased exercise and smoking cessation, as well as the consideration for drug therapy (SIGN, 2007). However, observed reductions in rates of CVD morbidity and mortality within the general population have not occurred for patients with schizophrenia. Furthermore, evidence suggests that this mortality gap may be widening (Morden et al., 2012; Lawrence et al., 2013). A systematic review assessing the accuracy of the Framingham CVD algorithm identified considerable variation in the performance of the risk score between different clinical populations (Brindle et al., 2006). While it is known that such risk algorithms have limitations in certain chronic diseases such as rheumatoid arthritis (Crowson et al., 2012), little is known about their usefulness for schizophrenia and how this may relate to risk and pre-existing levels of CVD within different age groups. We recently found that individuals with schizophrenia had lower rates of recorded CVD than controls within a large primary care sample (Smith et al., 2013). Given the high rates of cardiovascular mortality in this cohort, this finding might suggest a systematic under-recognition of cardiovascular disease in schizophrenia. Reasons for this are likely to be multifactorial but it is possible that individuals with schizophrenia
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
are not being identified by primary care physicians as being at high risk of cardiovascular disease, either because they are not having their cardiovascular risk factors assessed, or because existing risk calculators do not effectively identify this clinical sub-group as being at high risk of CVD. Although the prevention of CVD requires early identification and recording of risk factors, there has to date been very little study of how levels of recording of risk factors within primary care might differ between patients with schizophrenia compared to the general population (Ketola et al., 2006). Further, it is known that cardiovascular risk increases with age and, as such, age is heavily weighted within the JBS2 risk score (The Joint British Societies, 2006). People with schizophrenia tend to develop CVD (and have higher levels of CVD risk factors) at a much younger age than the general population (Holt, 2013). For example, compared to the general population, they have much higher rates and greater intensity of smoking (Hennekens et al., 2005; Myles et al., 2012), higher blood cholesterol (Redelmeier et al., 1998), and higher rates of diabetes (Sukdeb, 1995; Bai et al., 2013) and metabolic syndrome (Lee et al., 2012). They are also more likely to have risk factors, such as obesity (which may be an independent risk factor because of psychotropic medication) (Allison et al., 1999a; Ratliff et al., 2013), which are not currently included within risk calculators because they have not been found to be independent predictors of CVD risk for the general population. Antipsychotic medication in particular is an important additional factor for CVD risk. It is recognised that antipsychotics increase the total and low-density lipoprotein (LDL) cholesterol, as well as increasing triglycerides and decreasing high-density lipoprotein (HDL) cholesterol (Meyer, 2001). Although there has been some work on the use of Framingham risk scores to estimate CVD risk in people with schizophrenia (Goff et al., 2005; Slomka et al., 2012; Tay et al., 2013) — including by age (Jin et al., 2011), and antipsychotic prescribed (Barak et al., 2005; Daumit et al., 2008) — there have to date been no large scale studies from primary care databases which have examined how estimated CVD risk in people with schizophrenia compares to the general population by age and gender. 2. Methods We used data from the Primary Care Clinical Informatics Unit database (University of Aberdeen), which consisted of all 1,751,841 registered people who were alive and permanently registered with 314 general practices on March 31, 2007. This dataset is a representative sample covering approximately one third of the Scottish population. A more detailed explanation of the dataset is available elsewhere (Barnett et al., 2012). People were identified as having ‘schizophrenia or related nonorganic psychosis’ (from here referred to as ‘schizophrenia’) based on the recording ever of any of the following primary care Read Codes (where % is noted this means “this code and any below it in the code hierarchy”): E10% schizophrenic disorders; E121 chronic paranoid psychosis; E12z paranoid psychosis NOS; E13% other non-organic psychoses; E13z non-organic psychosis/psychotic episode; NOS; E1z non-organic psychosis NOS; Eu20% schizophrenia; Eu22% persistent delusional disorder; or the recording in the last 12 months of Eu23% acute/transient psychotic disorder. Deprivation status was measured using the Carstairs deprivation score, which is widely used in health research (Carstairs, 1991) and estimates of the 10-year risk for Framingham risk scores were generated for each participant. Joint British Societies (JBS2) scores were calculated based on the 1991 Framingham equations (Anderson et al., 1991b). The JBS2 score calculates 10 year CVD risk as the sum of CHD and stroke risk using the relevant Framingham equations containing the risk factors; age, sex, systolic BP, smoking status, diabetes, total cholesterol and HDL cholesterol. As HDL cholesterol figures were not present in the data, we used estimated levels taken from the Scottish Health
177
Survey from 2011 by age group and gender, as used by the ‘ASSIGN’ score (Scottish Government, 2011; Assign, 2013). To comply with JBS guidelines, people were identified as smokers if they had a record of having smoked at any time (British Cardiac Society et al., 2005). The date of first diagnosis for those with a record of pre-existing CVD was also recorded and from this we calculated age at first diagnosis of CVD. 3. Statistical analysis We restricted our analyses to those aged 35 to 74 as this is the age range in which the JBS risk score has been validated. People with a pre-existing diagnosis of CHD, stroke or peripheral vascular disease were excluded (see Table 1). Summary statistics for all study variables were calculated, including means and standard deviations for continuous characteristics and percentages for categorical characteristics. Summary statistics on risk estimates, risk factors, rates of pre-existing CVD and age of first diagnosis of CVD are presented by age and gender for schizophrenia and controls. For all statistical analyses, a p value less than 0.05 was considered statistically significant. All analyses were performed in Stata version 13.0. We also used all variables included in the JBS risk calculation (with the exception of HDL cholesterol) as well as the deprivation and BMI levels in a multivariate regression model using robust standard errors to assess any differences in these variables between schizophrenia and controls, in terms of their association with being identified as high risk under JBS. 4. Results 4.1. CVD rates Table 1 highlights that pre-existing rates of CVD were higher in the 35–44, 45–54 and 55–64 age groups for both men and women with schizophrenia and lower only within the 65–74 age group. 4.2. Data completeness Table 2 shows the number of individuals with missing data for those variables included in the calculation of the JBS score. Age, sex and diagnosis of diabetes were all fully recorded while HDL cholesterol levels were estimated for all individuals. Overall both men (schizophrenia men 27.2% vs. control men 24.5%) and women (schizophrenia women 31.9% vs. control women 25.9%) with schizophrenia were more likely to have data fully recorded. Of the individual variables, most of the missing data were found for the total cholesterol, with around three quarters of individuals having no recorded levels (and levels of missing data higher in the younger age groups). In comparison, only 4.4% of men and 2.2% of women with schizophrenia did not have a record of smoking status and 8.4% of men and 3.4% of women had no systolic blood pressure (SBP) recorded. Rates of individuals with known cardiovascular disease were calculated by age at first diagnosis for the schizophrenia cohort and for controls by gender (Fig. 1). The schizophrenia cohort were more likely to have a first diagnosis at a younger age relative to controls. Although rates in those aged under 55 were higher in men for both schizophrenia and controls, differences between cohorts were larger for men than for Table 1 Pre existing CVD rates by age group and gender. Number (% with CVD) Age group
Schizophrenia Controls Men Men
35–44 45–54 55–64 65–74
15 (1.2) 62 (5.6) 133 (15.9) 126 (27.2)
Schizophrenia Controls Women Women
1171 (0.8) 12 (1.4) 5324 (4.2) 36 (3.7) 14,523 (13.4) 90 (9.6) 23,258 (29.8) 136 (18.4)
1334 (0.9) 3396 (2.7) 8578 (7.9) 16,915 (19.0)
178
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
Table 2 Recording of risk factors used in JBS score by age and gender. Total without CVD
Schizophrenia Controls Men Men
Schizophrenia Controls Women Women
35–44
1260 (37.6)
853 (26.4)
45–54
1049 (31.3)
55–64
703 (21.0)
140,056 (34.1) 121,810 (29.6) 93,812 (22.7)
65–74 Total
337 (10.1) 3349
54,731 (13.3) 411,126
602 (18.6) 3232
117,080 (83.6) 108,897 (89.4) 87,047 (92.8) 52,303 (95.6) 365,327 (89.0)
832 (97.5)
103,296 (73.8) 105,727 (86.8) 86,267 (92.0) 52,014 (95.0) 347,304 (84.6)
812 (95.1)
14,104 (10.1) 27,488 (22.6) 33,867 (36.1) 26,040 (47.6) 101,499 (24.7)
157 (18.4) 274 (29.4) 344 (40.7) 266 (44.2) 1041 (32.2)
13,486 (10.1) 26,059 (21.6) 36,096 (36.0) 35,067 (48.9) 110,708 (26.0)
13,490 (9.6) 27,236 (22.4) 33,742 (36.0) 26,009 (42.4) 100,477 (24.5)
156 (18.3) 269 (28.9) 344 (40.7) 263 (43.7) 1032 (31.9)
13,265 (9.9) 25,942 (21.5) 36,061 (36.0) 35,047 (48.8) 110,315 (25.9)
Smoking status recorded 35–44 1187 (94.2) 45–54
1006 (95.9)
55–64 65–74 Total
676 (96.1) 333 (98.8) 3202 (95.6)
Systolic blood pressure recorded 35–44 1089 (86.4) 45–54
980 (93.4)
55–64 65–74 Total
670 (95.3) 329 (97.6) 3068 (91.6)
Total cholesterol recorded 35–44 229 (18.2) 45–54 290 (27.6) 55–64 256 (36.4) 65–74 145 (43.0) Total 920 (27.4)
Number with data fully recorded 35–44 226 (17.9) 45–54 287 (27.4) 55–64 256 (36.4) 65–74 143 (42.4) Total 912 (27.2)
932 (28.8) 845 (26.1)
916 (98.3) 831 (98.3) 583 (96.8) 3162 (97.8)
898 (96.3) 831 (98.3) 581 (96.5) 3122 (94.9)
133,856 (31.4) 120,592 (28.3) 100,165 (26.1) 71,761 (18.6) 427,521
127,485 (95.2) 116,501 (96.6) 97,437 (97.3) 69,775 (97.2) 411,198 (96.4)
122,894 (91.8) 115,311 (95.6) 96,969 (96.8) 69,455 (96.8) 404,629 (94.9)
Note: Of the other risk factors required for calculation of JBS score age, gender and diabetes were fully recorded while HDL levels were estimated for all people.
women (schizophrenia men 46.1% vs. control men 34.8%; p b 0.001: schizophrenia women 28.9% vs. control women 23.8%; p b 0.001). 4.3. Risk factors We identified 1942 individuals with schizophrenia and 210,792 individuals with no record of schizophrenia (controls). Men and women with schizophrenia who had records of JBS score risk factors were significantly younger than male controls (mean age 52.9 to 56.9; difference −4.0 95% CI −3.4 to −4.7) and female controls (mean age 56.7 to 58.2; difference −1.5 95% CI −0.9 to −2.2) (Table 3). Individuals with schizophrenia were also more likely to smoke, and have diabetes, but had significantly lower mean systolic blood pressure. For risk factors not included in the JBS score, both men and women with schizophrenia were significantly more socially deprived compared to controls and were more likely to have a BMI N 30. (See Table 4.) Fig. 2 shows the percentage of those by JBS risk level, age group and gender for schizophrenia and controls. Age was a major factor being identified as high risk, with 79% of those with schizophrenia aged 65– 74 estimated at high risk in the next 10 years compared with only 1.3% of those aged 35–44. Men were more likely to be identified as high risk for both schizophrenia and controls. This was reflected in all age groups, with only just over 1% of women identified as high risk in
the 45–54 year old age group compared to around 10% of males and only a third of schizophrenia women and a quarter of controls aged 65–74 identified as high risk compared to just under 80% of men for both schizophrenia and controls. The multiple regression model shows that compared to controls, gender (schizophrenia coefficient 6.16 vs. controls 7.17), smoking (schizophrenia coefficient 6.20 vs. controls coefficient 6.86) and diabetes (schizophrenia coefficient 6.62 vs. controls coefficient 7.22) had a smaller effect on being identified as high risk under JBS. In contrast, age had a higher effect for schizophrenia (schizophrenia coefficient 0.54 vs. controls coefficient 0.51). No differences were found for total cholesterol and systolic blood pressure. BMI and deprivation both showed negative associations for schizophrenia and controls, possibly reflecting their exclusion from the JBS algorithm, although results were only statistically significant for controls. 5. Discussion We found that individuals with schizophrenia were more likely to have an early diagnosis of CVD, with over half of men having been first diagnosed under the age of 55. This was not reflected in the CVD risk score estimates, where only a small proportion of those with schizophrenia at risk of CVD were aged b55 years. This highlights a concern that current cardiovascular risk scores might underestimate risk in schizophrenia, particularly in men. As with all cardiovascular risk scores, the absolute risk increases rapidly with age; this leads to a lower identification of younger individuals at high risk and an increased identification of older individuals at low risk (Tunstall-Pedoe, 2011). This issue calls into question the validity of these instruments in conditions such as schizophrenia, where higher mortality rates are a feature at a younger age (Holt, 2013). This could lead to under-prediction of CVD risk and a loss of opportunity for primary prevention interventions, which may contribute to the high rates of cardiovascular disease mortality in individuals with schizophrenia (Smith et al., 2013). From the CVD risk factors included within the JBS2 that were assessed, we found higher rates of smoking in individuals with schizophrenia compared to controls, in keeping with the other studies (Hennekens et al., 2005; Myles et al., 2012). Diabetes prevalence was also significantly higher in those with schizophrenia compared to controls. However, under JBS guidelines, those with diabetes should be treated as at high cardiovascular risk and therefore automatically considered for primary prevention approaches (British Cardiac Society et al., 2005). In contrast to other studies, we found lower levels of systolic blood pressure in individuals with schizophrenia relative to controls which is difficult to account for but which may to some degree reflect adrenergic blockade by antipsychotic medications (Hennekens et al., 2005; Osborn et al, 2008). As an example of potentially important risk factors not included in the JBS score, we found higher rates of obesity (BMI N 30) in individuals with schizophrenia compared to controls. There has been recent evidence to suggest that obesity in people with schizophrenia is associated with an elevated risk for CVD compared to age, gender, race and BMImatched controls (Ratliff et al., 2013). Individuals with schizophrenia were also found to be living in more deprived social circumstances than the general population. Analysis has shown that being in the highest 20% of deprivation rather than the lowest 20% is equivalent in risk score terms to a decade or more in age, or to a concurrent diagnosis of diabetes (Tunstall-Pedoe et al., 2006). In the UK, risk scores such as ASSIGN in Scotland (Assign, 2013) and QRISK in England (Hippisley Cox et al., 2007) have been developed which have incorporated deprivation as a risk factor. However, both give less weight to existing risk factors and have been found to identify fewer individuals in the younger age groups than Framingham, highlighting a further limitation of the current CVD risk prediction algorithms with respect to individuals with schizophrenia (de la Iglesia et al., 2011). However, despite the
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
179
100% 90%
23.90% 32.1%
80% 46.6%
47.4%
% of people
70% 60%
30.0% Age 65 and above 33.0%
50%
Age55-64 25.2%
40% 30%
24.7% 20.1%
17.2%
16.1% 10.1%
0%
Under 45
30%
20% 10%
Age 45-54 28.8%
Schizophrenia Men (n=510)
8.0%
Controls Men (n=73,395)
6.6%
Schizophrenia Control Women Women (n=547) (58,837)
Note: figures represent the percentage of those with established CVD by age of first diagnosis Fig. 1. Rates of known cardiovascular disease for schizophrenia versus controls by age group at first diagnosis. Note: figures represent the percentage of those with established CVD by age of first diagnosis.
identification of high risk factors in schizophrenia, for some risk factors in a multiple regression model only age was found to have a higher effect with respect to being identified as high risk for schizophrenia compared to controls. This may in part be a reflection of higher risk factors in younger patients for those with schizophrenia which are not reflected in CVD risk scores such as JBS due to the weight given to age.
6. Strengths and limitations Our dataset is large, covers all age groups and is representative of the whole Scottish population. We calculated scores for individuals with schizophrenia by age group and sex and compared this to controls, along with age at first cardiovascular disease diagnosis. To our knowledge, this level of detail of JBS2 risk score reporting in people with schizophrenia has not been previously carried out. We were also able to compare the levels of CVD risk estimation with rates of established cardiovascular disease by age and gender in those with schizophrenia
compared to the general population, as a test of the clinical utility of current risk prediction tools. However, our study has some limitations. We used estimated HDL cholesterol levels in all individuals included in our study using age and sex matched means from the 2011 Scottish Health Survey (Scottish Government, 2011) to generate the JBS2 score. Schizophrenia has been associated with lower HDL cholesterol levels than the general population (Huang and Chen, 2005) and by using the mean Scottish levels we may have under predicted cardiovascular risk in individuals with schizophrenia. Around three quarters of the dataset had no information on one or more risk factors and so was excluded from the analysis. Levels of missing data were similar in those with schizophrenia and the general population. It is also possible that people with risk factors recorded were more likely to have a condition included under the Quality and Outcomes Framework (QOF) in the UK, where practices are paid depending on the percentage of people who are successfully controlled for targets such as blood pressure and cholesterol (ISD Scotland, 2013). If this true, then people with missing data may be less likely to have risk
Table 3 Risk factors for cardiovascular disease by gender, schizophrenia versus controls. Variable
Schizophrenia Men
Controls Men
911 (46.9) 52.9 (10.3) 5.2 (1.3)
100,477 (47.7) 56.9 (10.1) 5.2 (1.1)
1.3 (0.4)
Difference (95% CI)
Schizophrenia Women
Controls Women
Difference (95% CI)
−4.0 (−3.4 to -4.7) −0.0 (−0.4 to 0.1)
1031 (53.1) 56.7 (10.5) 5.4 (1.1)
110,315 (52.3) 58.2 (10.5) 5.4 (1.1)
−1.5 (−0.9 to −2.1) −0.0 (−0.1 to 0.1)
1.4 (0.3)
0.0 (−0.1 to 0.0)
1.6 (0.3)
1.6 (0.3)
0.0 (−0.1 to 0.1)
Risk factors (JBS only) Number (%) Age mean (SD) Total cholesterol (mean) HDL cholesterol (mean) Systolic BP (mean) Smoke no. (%) Diabetes no. (%)
130.4 (15.6)
135.5 (15.2)
−5.1 (−3.2 to −6.1)
130.0 (17.0)
133.3 (16.1)
−3.3 (−2.3 to −4.3)
625 (68.9%) 201 (22.1%)
50,959 (51.2%) 18,194 (18.2%)
17.7 (14.4 to 20.9) 3.9 (1.3 to 6.5)
595 (57.9%) 251 (24.3%)
47,316 (43.1%) 15,361 (13.9%)
14.8 (11.6 to17.7) 10.4 (8.2 to 12.5)
Other risk factors Deprivation mean (SD) BMI N 30 no. (%)
0.96 (3.5) 273 (29.9%)
−0.38 (3.3) 25,777 (25.7%)
1.34 (1.13 to 1.55) 4.2 (1.3 to 7.1)
0.36 (3.4) 368 (35.7%)
−0.22 (3.4) 30,983 (28.1%)
0.58 (0.38 to 0.78) 7.6 (4.7 to 10.2)
Note: HDL uses estimated levels taken from the Scottish Health Survey from 2011 by age group and gender.
180
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181
Table 4 Regression coefficients for CVD risk factors. Schizophrenia
Controls
Variable
Coefficient
95% CI
Coefficient
95% CI
Gender Age Total cholesterol Systolic BP Smoking Diabetes BMI Deprivation
6.16 0.54 1.93 0.21 6.20 6.62 −0.01 −0.01
5.81 to 6.52 0.52 to 0.56 1.78 to 2.08 0.20 to 0.23 5.83 to 6.56 6.18 to 7.06 −0.02 to 0.00 −0.06 to 0.04
7.17 0.51 1.93 0.21 6.86 7.22 −0.01 −0.01
7.13 to 7.20 0.50 to 0.52 1.91 to 1.94 0.20 to 0.22 6.83 to 6.90 7.17 to 7.27 −0.02 to −0.01 −0.02 to −0.01
factors controlled as the incentive to treat these people is lower for the primary care provider. Changes to the QOF in 2011 incentivised the recording of body mass index, blood pressure, total to HDL cholesterol ratio and blood glucose for people with mental illness potentially leading to improvements in the recording and monitoring of cardiovascular risk factors within this population compared to the levels shown in our study (Holt, 2013). Given the high rates of CVD disease and mortality in those with schizophrenia, it is of concern that NHS England recently decided to withdraw financial incentives through the QOF for monitoring CVD risk factors from 2014 to 2015, although these incentives will remain in place within NHS Scotland. 7. Conclusion In conclusion, we found that JBS2 risk scores identified only a small proportion of individuals with schizophrenia under the age of 55 as being at high risk of cardiovascular disease, alongside high rates of first diagnosis of cardiovascular disease recorded within this age group. This suggests that the current cardiovascular risk prediction algorithms may be under-predicting risk in schizophrenia, particularly for younger men. Existing models used in the UK and elsewhere have ignored schizophrenia or major mental illness in general as a significant predictor (Anderson et al., 1991b; Tunstall Pedoe at al, 1997; Tunstall-Pedoe et al., 2006; Hippisely-Cox et al., 2007). This is likely to contribute to the premature mortality due to cardiovascular disease in this cohort of patients. Similar issues have been found with conditions
such as rheumatoid arthritis (Crowson et al., 2012), which has recently been added to the ASSIGN score in Scotland as a risk factor (ASSIGN, 2013). Further work is therefore required to validate the use of CVD risk scores in individuals with schizophrenia, and possibly to create schizophrenia-specific risk scores, in order to better predict cardiovascular risk. In the meantime, clinicians should obviously continue to manage individual risk factors (such as higher rates of smoking, obesity and diabetes) in this vulnerable clinical group. Role of the funding Source The Chief Scientist Office of the Scottish Government Health Directorates (Applied Research Programme Grant ARPG/07/1); and the Scottish School of Primary Care, partly supported SWM's post and the development of the Applied Research Programme. Contributors Gary McLean carried out the statistical analysis and wrote the first draft. All other authors contributed to subsequent drafts. All authors read and approved the final manuscript. Conflict of interest There are no conflicts of interests. Acknowledgements We thank the Chief Scientist Office of the Scottish Government Health Directorates (Applied Research Programme Grant ARPG/07/1); the Scottish School of Primary Care, which partly supported SWM's post and the development of the Applied Research Programme; and the Primary Care Clinical Informatics Unit at the University of Aberdeen, which provided the data. The views in this publication are not necessarily the views of the University of Aberdeen of University of Glasgow, their agents, or employees. We thank Katie Wilde and Fiona Chaloner of the University of Aberdeen, who did the initial data extraction and management.
References Allison, D.B.,Fontaine, K.R.,Heo, M.,Mentore, J.L.,Cappelleri, J.C.,Chandler, L.P.,Weiden, P.J., Cheskin, L.J., 1999a. The distribution of body mass index among individuals with and without schizophrenia. J. Clin. Psychiatry 60, 215–220. Anderson, K.M., Wilson, P.W., Odell, P.M., Kannel, W.B., 1991a. An updated coronary risk profile: a statement for health professionals. Circulation 83, 356e62. Anderson, K.M.,Odell, P.M.,Wilson, P.W.,Kannel, W.B., 1991b. Cardiovascular disease risk profiles. Am. Heart J. 121, 293–298. ASSIGN, 2013. http://assign-score.com/estimate-the-risk/risk-factors/#more-info (accessed June 2014). Bai, Ya-Mei,Su, Tung-Ping,Chen, Mu-Hong,Chen, Tzeng-Ji,Chang, Wen-Han, 2013. Risk of developing diabetes mellitus and hyperlipidaemia among people with bipolar disorder, major depressive disorder, and schizophrenia: a 10-year nationwide population-
100% 90% 80%
60% 50% 40% JBS
10%
35-44
45-54
55-64 Men
65-74
45-54
55-64
Women
Fig. 2. JBS risk levels by age group and gender.
Controls
Schizophrenia
Controls
Schizophrenia
Controls
Controls
35-44
Schizophrenia
Schizophrenia
Controls
Schizophrenia
Controls
Schizophrenia
Controls
Schizophrenia
Controls
0% Schizophrenia
% of people
70%
65-74
G. McLean et al. / Schizophrenia Research 159 (2014) 176–181 based prospective cohort study. J. Affect. Disord. (ISSN: 0165-0327). http://dx.doi. org/10.1016/j.jad.2013.02.019. Barak, Y., Swartz, M., Plopsky, I., 2005. Assessing cardiovascular risks of olanzapine treatment: a 6-month study versus haloperidol in schizophrenia people. Int. Clin. Psychopharmacol. 20 (6), 315–317. Barnett, K.,Mercer, S.W.,Norbury, M.,Watt, G.,Wyke, S.,Guthrie, B., 2012. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet 380 (9836), 37–43. Brindle, P.,Beswick, D.,Fahey, T.,Ebrahim, S., 2006. The accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart 92, 1752–1759. British Cardiac Society,British Hypertension Society,UK Diabetes,UK HEART,Primary Care cardiovascular Society, The Stroke Association, 2005. JBS 2: Joint British Societies' guidelines on prevention of cardiovascular disease in clinical practice. Heart 91 (Suppl. V), v1–v52. Carstairs, V., 1991. Deprivation and Health in Scotland. Aberdeen University Press, Aberdeen. Crowson, C.S., Matteson, E.L., Roger, V.L., Therneau, T.M., Gabrie, S.E., 2012. Usefulness of risk scores to estimate the risk of cardiovascular disease in people with rheumatoid arthritis. Am. J. Cardiol. 110 (3), 420–424. Daumit, D.C., Goff, J.M., Meyer, V.G., Davis, H.A., Nasrallah, J.P., McEvoy, R., Rosenheck, S.M., Davis, J.K., Hsiao, T.S., Stroup, J.A., 2008. Antipsychotic effects on estimated 10-year coronary heart disease risk in the CATIE schizophrenia study. Schizophr. Res. 105 (1–3), 175–187. de la Iglesia, B., de la Iglesia, B., Potter, J.F., Poulter, N.R., 2011. Performance of the ASSIGN cardiovascular disease risk score on a UK cohort of people from general practice. Heart 97, 491 (e9). Goff, C. Cather, Evins, A.E., Henderson, D.C., Freudenreich, O., Copeland, P.M., Bierer, M., Duckworth, K., Sacks, F.M., 2005. Medical morbidity and mortality in schizophrenia: guidelines for psychiatrists. J. Clin. Psychiatry 66 (2), 183–194. Hennekens, C.H., Hennekens, A.R., Hollar, D., Casey, D.E., 2005. Schizophrenia and increased risks of cardiovascular disease. Am. Heart J. 150 (6), 1115–1121. Hippisley-Cox, .J., Coupland, C., Vinogradova, Y., Robson, J., May, M., Brindle, P., 2007. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. Br. Med. J. 335 (7611), 136–141. Holt, R.I., 2013. Cardiovascular Disease and Diabetes in People With Severe Mental Illness: Causes, Consequences and Pragmatic Management. http://www.rcpsych.ac.uk/pdf/ PCCJ_Holt_FINALONLINE_JAN%5B1%5D.pdf (accessed June 2014). Huang, T.L.,Chen, J.F., 2005. Serum lipid profiles and schizophrenia: effects of conventional or atypical antipsychotic drugs in Taiwan. Schizophr. Res. 80 (1), 55–59. ISD Scotland, 2013. http://www.isdscotland.org/Health-Topics/General-Practice/QualityAnd-Outcomes-Framework (accessed June 2014). Jin, H., Folsom, D., Sasaki, A., Mudaliar, S., Henry, R., Torres, M., Golshan, S., Glorioso, D.K., Jeste, D., 2011. Increased Framingham 10-year risk of coronary heart disease in middle-aged and older people with psychotic symptoms. Schizophr. Res. 125 (2–3), 295–299. Ketola, E., Sipilä, R., Mäkelä, M., Klockars, M., 2006. Quality improvement programme for cardiovascular disease risk factor recording in primary care. Qual. Health Care 9 (3), 175–180. Lawrence, D., Hancock, K., Kisley, S., 2013. The gap in life expectancy from preventable physical illness in psychiatric people in Western Australia: retrospective analysis of population based registers. BMJ 346, f2539. http://dx.doi.org/10.1136/bmj.f2539.
181
Lee, J., Nurjono, M., Wong, A., Salim, A., 2012. Prevalence of metabolic syndrome among people with schizophrenia in Singapore. Ann. Acad. Med. Singap. 41, 457–462. Meyer, J.M., 2001. Novel antipsychotics and severe hyperlipidaemia. J. Clin. Psychopharmacol. 21, 369–374. Morden, N.E., Lai, Z., Goodrich, D.E., MacKenzie, T., McCarthy, J.F., Austin, K., Welsh, D.E., Bartels, S., Kilbourne, A.M., 2012. Eight-year trends of cardiometabolic morbidity and mortality in people with schizophrenia. Gen. Hosp. Psychiatry 34, 368–379. Myles, N., Newall, H.D., Curtis, J., 2012. Tobacco use before, at and after first-episode psychosis: a systematic meta-analysis. J. Clin. Psychiatry 73 (4), 468–475. Nichols, M., Townsend, N., Scarborough, P., Rayner, M., 2013. Trends in age-specific coronary heart disease mortality in the European Union over three decades: 1980–2009. Eur. Heart J. 34, 3017–3027. Osborn, D.P., Wright, C.A., Levy, G., 2008. Relative risk of diabetes, dyslipidaemia, hypertension, and the metabolic syndrome in people with severe mental illnesses: systematic review and meta-analysis. BMC Psychiatr. 8, 84. Ratliff, J.C., Palmese, L.B., Reutenauer, E.L., Srihari, V.H., Tek, C., 2013. Obese schizophrenia spectrum people have significantly higher 10-year general cardiovascular risk and vascular ages than obese individuals without severe mental illness. Psychosomatics 54 (1), 67–73. Redelmeier, D., Siew, H., Booth, G., 1998. The treatment of unrelated disorders in people with chronic medical diseases. N. Engl. J. Med. 338, 1516–1520. Scottish Government, 2011. http://www.scotland.gov.uk/Topics/Statistics/Browse/ Health/TrendMortalityRates (accessed June 2014). SIGN, 2007. SIGN Guideline 97 risk estimation and the prevention of cardiovascular disease. http://www.sign.ac.uk/pdf/sign97.pdf (accessed June 2014). Slomka, J.M., Piette, J.D., Post, E.P., Krein, S.L., Lai, Z., Goodrich, D.E., Kilbourne, A.M., 2012. Mood disorder symptoms and elevated cardiovascular disease risk in people with bipolar disorder. J. Affect. Disord. 138 (3), 405–408. Smith, D.J., Langan, J., McLean, G., Guthrie, B., Mercer, S.W., 2013. Schizophrenia is associated with excess multiple physical-health co morbidities but low levels of recorded cardiovascular disease in primary care: cross-sectional study. BMJ Open 3, e002808. Sukdeb, M., 1995. High prevalence of type II diabetes in schizophrenic people. Schizophr. Res. 15 (1/2), 195. Tay, Y.H.,Nurjono, M., Lee, M.J., 2013. Increased Framingham 10-year CVD risk in Chinese people with schizophrenia. Schizophr. Res. 147 (1), 187–192. The Joint British Societies, 2006. Cardiovascular Risk Assessor (V01.06). http://heartuk.org. uk/health-professionals/resources/risk-calculators/jbs2-cv-risk-assessor (accessed June 2014). Tunstall-Pedoe, H., 2011. Cardiovascular risk and risk scores: ASSIGN, Framingham, QRISK and others: how to choose. Heart 97 (6), 442–444. Tunstall-Pedoe, H., Woodward, M.,Tavendale, R., A'Brook, R.,McCluskey, M.K., 1997. Comparison of the prediction by 27 different factors of coronary heart disease and death in men and women of the Scottish heart health study: cohort study. Br. Med. J. 315 (7110), 722–729. Tunstall-Pedoe, H., Woodward, M., for the SIGN group on risk estimation, 2006. By neglecting deprivation, cardiovascular risk scoring will exacerbate social gradients in disease. Heart 92, 307e1. World Health Organisation, 2013. WHO Mortality Database: 1st May 2013update http:// www.who.int/healthinfo/statistics/mortality_rawdata/en/index.html. World Health Organization, Department of Health Statistics and Information System (accessed June 2014).
