Incidence of risk factors for myocardial infarction and other vascular diseases in patients with psoriasis

Authors


  • Conflicts of interest None declared.

James A. Kaye.
E-mail: jkaye@bu.edu

Summary

Background  Recent cross-sectional studies reported a higher prevalence of diabetes and other risk factors for cardiovascular disease in patients with psoriasis than in the general population.

Objectives  To estimate the cumulative incidences of risk factors for myocardial infarction and other vascular diseases after a first recorded diagnosis of psoriasis and the hazard ratio (HR) for these conditions in patients with psoriasis compared with the general population.

Methods  We used the General Practice Research Database to conduct a cohort study of 44 164 patients with a first-time diagnosis of psoriasis and 219 784 nonpsoriasis comparison subjects psoriasis-matched on age, sex and index date.

Results  HRs were increased among patients with psoriasis vs. the comparison cohort for incident diabetes [HR 1·33; 95% confidence interval (CI) 1·25–1·42], hypertension (HR 1·09; 95% CI 1·05–1·14), obesity (HR 1·18; 95% CI 1·14–1·23) and hyperlipidaemia (HR 1·17; 95% CI 1·11–1·23). Patients with psoriasis also had higher risks of incident myocardial infarction (HR 1·21; 95% CI 1·10–1·32), angina (HR 1·20; 95% CI 1·12–1·29), atherosclerosis (HR 1·28; 95% CI 1·10–1·48), peripheral vascular disease (HR 1·29; 95% CI 1·13–1·47) and stroke (HR 1·12; 95% CI 1·00–1·25).

Conclusions  Risk factors for cardiovascular disease as well as myocardial infarction and other vascular diseases occur with higher incidence in patients with psoriasis than in the general population. Further work is needed to investigate whether these associations involve causal factors related to psoriasis or its treatment.

Gelfand et al.1 recently reported the results of an observational study which suggested that psoriasis may confer an independent risk of myocardial infarction (MI). Moreover, the same investigators2 and others3,4 have published studies which found a higher prevalence of risk factors for ischaemic heart disease and other atherosclerotic vascular diseases among patients with psoriasis than among the general population. Potential biological mechanisms for these associations have been postulated which involve genetic differences, increased production of inflammatory mediators among patients with psoriasis, other environmental exposures possibly associated with psoriasis, and behavioural factors.2 Because prevalence studies, which are often cross-sectional, do not delineate the time sequence of the occurrence of the illnesses being evaluated, such studies are less useful for identifying possibly causal relationships than incidence studies, which focus on the time of onset of the diseases of interest. Moreover, prevalence estimates confound incidence and duration of illness, which may further obscure aetiological relationships.

In order to begin to understand whether causal relations between psoriasis and risk factors for MI are plausible, we conducted an observational study using the same database studied by Gelfand et al.,1,2 namely the General Practice Research Database (GPRD). We restricted our attention to risk factors for MI that were diagnosed for the first time following a diagnosis of psoriasis, and we evaluated the time course of the incidence of these risk factors in patients with psoriasis compared with the general population. We also studied the incidence of MI and other cardiovascular diseases after a first diagnosis of psoriasis.

Materials and methods

General Practice Research Database

This study was carried out using information derived from the GPRD, a general practice-based data resource from the U.K. The Boston Collaborative Drug Surveillance Program has used the GPRD extensively to conduct epidemiological studies for more than a decade. Since 1987, over six million residents of the U.K. have been enrolled with selected general practitioners (GPs) who have agreed to provide electronically recorded medical information to the GPRD for research purposes with patients’ identities anonymized by code. The GPs received instruction on the standardized recording of medical information which includes patient characteristics, historical information, diagnoses, other findings (smoking status, blood pressure, height and weight), drugs prescribed, referrals to consultants, and hospitalizations. Our experience validating the GPRD has been described in detail.5

Originally, a modification of the Oxford Medical Information System (OXMIS) classification was used to enter medical diagnoses. In the late 1990s, with the availability of updated computer software, this was supplanted by the Read classification.

We carried out this study using GPRD information updated in the fourth quarter of 2005.

Study design

This is a descriptive cohort study in which we estimated cumulative incidences (‘risks’) of various diagnoses over time after the first diagnosis of psoriasis and compared these with risks among age- and sex-matched subjects in the general (nonpsoriasis) population.

Psoriasis and comparison cohorts

We initially identified all patients with a first-time diagnosis of psoriasis after 1 January 1991 using standard OXMIS and Read codes. We restricted the psoriasis cohort to those who had at least 1 year of medical history recorded in the database before their index date (the date of the first-time diagnosis of psoriasis). The index date for each subject in the psoriasis cohort defined the start of follow-up for estimating the cumulative incidences of the outcomes of interest in the psoriasis group.

The comparison cohort comprised randomly selected subjects who were matched in a 5 : 1 ratio to the patients with psoriasis by year of birth, sex, general practice and index date. Subjects in the comparison cohort were required to have at least 1 year of medical history recorded in the database before their index date. The index date for each subject in the comparison cohort defined the start of follow-up for estimating the cumulative incidences of the outcomes of interest in the comparison group.

Risk factors for atherosclerotic cardiovascular disease

We identified the first-time diagnosis of four major risk factors for atherosclerotic cardiovascular disease (diabetes mellitus, obesity, hypertension and hyperlipidaemia) in members of both the psoriasis and comparison cohorts. For the analyses reported here, we confined our attention to patients who had these diagnoses recorded for the first time after their index date. (Patients with each of these diagnoses first recorded before their index date were censored at the start of follow-up). Follow-up time ended when a patient developed an outcome of interest, transferred out of their practice, or died. Patients could, however, continue to contribute person-time to other analyses after they experienced a particular outcome. We noted the frequencies of patients with psoriasis having one, two, three or all four of the risk factors of interest diagnosed for the first time after their index date. Chronological patient profiles were reviewed to confirm that the outcome diagnoses were valid and were incident after the index date.

Drug use among cases and comparison subjects

We noted treatments prescribed to patients with psoriasis after their index date in six categories: (i) retinoids (tretinoin, adapalene, tazarotene, isotretinoin, etretinate); (ii) immunosuppressants (ciclosporin, tacrolimus, leflunomide, pimecrolimus, mycophenolate mofetil, sirolimus, etanercept, infliximab, thalidomide); (iii) drugs with antiproliferative activity that are used as anticancer drugs (methotrexate, azathioprine, fluorouracil, hydroxycarbamide, cyclophosphamide, chlorambucil, mercaptopurine, busulfan, estramustine, melphalan, paclitaxel, etoposide, epirubicin); (iv) systemic corticosteroids (prednisolone, fluticasone, budesonide, methylprednisolone, triamcinolone, hydrocortisone, dexamethasone, prednisone, cortisone, ciclesonide); (v) topical corticosteroids (hydrocortisone, clobetasone, mometasone, dexamethasone, triamcinolone, fluocinolone, prednisolone, alclometasone, flunisolide, halcinonide, fluocortolone, clobetasone); and (vi) ultraviolet (UV) regimens, including PUVA. It should be noted that UV regimens are recorded by GPs in their clinical records rather than in the prescription files and that the recording of such treatments by GPs may be incomplete as they are often administered in specialty clinics or in hospital.

Analysis

We used time-to-event analysis (Kaplan–Meier method) to estimate the cumulative incidences (‘risks’) of each of the outcomes of interest at specific times (1, 3, 5 and 10 years) after the index date. Observation for subjects without a particular outcome was censored at the time of their last recorded follow-up. We tested differences in the full time-to-event distributions between the psoriasis and the comparison cohorts using the log-rank test. We also used Cox regression to estimate for each outcome the hazard ratio (HR) comparing the psoriasis cohort with the comparison group. Due to the matching of the comparison subjects to the patients with psoriasis on age, sex and index date, adjustment for these factors in the regression analysis had no material effect on the results. We therefore report herein only the ‘crude’ HRs (which nevertheless by design take into account age, sex and calendar time).

Statistical calculations were carried out using Stata v. 8.2 (Stata Corp, College Station, TX, U.S.A.) and SAS v. 9.1 (SAS Institute, Inc., Cary, NC, U.S.A.).

Results

We identified 44 164 patients with a first-time diagnosis of psoriasis and at least 1 year of history recorded in the GPRD before their index date. We matched the patients with psoriasis to 219 784 subjects without a recorded diagnosis of psoriasis.

Characteristics of the psoriasis and comparison populations are presented in Table 1. Fifty-two per cent of the patients in the psoriasis group were female. The age distribution of the psoriasis cohort peaked in the 30–39-year-old range, although the distribution was nearly uniform from age 10 to age 69 years. The age and sex distributions in the comparison cohort were closely similar to those in the psoriasis cohort because of the matching. Approximately 95% of the patients with psoriasis received some form of treatment for this disease.

Table 1.   Characteristics of psoriasis and comparison cohorts
 Psoriasis (n = 44 164)
n (%)
Comparison (n = 219 784)
n (%)
  1. NA, not applicable.

Male21 121 (47·8)105 045 (47·8)
Female23 043 (52·2)114 739 (52·2)
Age (years)
 < 101887 (4·3)9418 (4·3)
 10–195058 (11·5)25 248 (11·5)
 20–295848 (13·2)29 198 (13·3)
 30–397079 (16·0)35 363 (16·1)
 40–496415 (14·5)32 021 (14·6)
 50–596648 (15·1)33 193 (15·1)
 60–695740 (13·0)28 607 (13·0)
 70–793938 (8·9)19 520 (8·9)
 80–891389 (3·2)6679 (3·0)
 90+162 (0·4)537 (0·2)
Treatment
 Yes41 790 (94·6)NA
 No2374 (5·4)NA

For each of the outcomes of interest, we will comment on the age and sex distributions of the patients who experienced that outcome in relation to those of the general psoriasis population (see Table 1). We then analyse the cumulative risk for the outcome comparing the psoriasis cohort with the comparator cohort.

Diabetes

Among the psoriasis cohort, 1198 (2·7%) patients had a new diagnosis of diabetes mellitus after their index date. Among the comparison cohort, 4482 (2·0%) had incident diabetes after their corresponding index date. Table 2 provides the distribution of diabetes in the two groups by age and sex. Patients with psoriasis with a diagnosis of diabetes were more likely to be male than the psoriasis population as a whole (Table 1), and they tended to be older; the distribution of these characteristics, however, was similar between patients with diabetes in the psoriasis cohort and patients with diabetes in the comparison cohort.

Table 2.   Incident diabetes cases in the psoriasis and comparison cohorts
 Psoriasis (n = 44 164)
n (%)
Comparison (n = 219 784)
n (%)
Total1198 (2·7)4482 (2·0)
Sex
 Male661 (55·2)2532 (56·5)
 Female537 (44·8)1950 (43·5)
Age (years)
 < 104 (0·3)13 (0·3)
 10–197 (0·6)27 (0·6)
 20–2924 (2·0)91 (2·0)
 30–3986 (7·2)264 (5·9)
 40–49191 (15·9)638 (14·2)
 50–59325 (27·1)1160 (25·9)
 60–69356 (29·7)1357 (30·3)
 70–79168 (14·0)774 (17·3)
 80–8935 (2·9)151 (3·4)
 90+2 (0·2)7 (0·2)

Table 3 provides the estimated cumulative incidences of diabetes at 1, 3, 5 and 10 years after the index date in the two groups, and the cumulative incidence estimates are presented graphically in Figure 1. The cumulative incidence of diabetes is significantly higher in the psoriasis cohort than in the comparison group (P < 0·0001). The HR for incident diabetes after the index date comparing the psoriasis cohort with the comparison group is 1·33 [95% confidence interval (CI) 1·25–1·42].

Table 3.   Estimated cumulative incidence of diabetes at specified times after the index date in the psoriasis and comparison cohorts
 CasesCumulative incidence (per 1000)95% CI (per 1000)
  1. CI, confidence interval.

Psoriasis
 1 year2075·24·5–5·9
 3 years33715·914·6–17·3
 5 years21025·423·6–27·3
 10 years36057·353·5–61·2
Comparison
 1 year6863·43·2–3·7
 3 years116910·910·4–11·4
 5 years88619·018·3–19·7
 10 years136343·942·4–45·5
Figure 1.

 Cumulative incidence of diabetes in the psoriasis and comparison cohorts.

Hypertension

Among the psoriasis cohort, 2765 (6·3%) patients had a new diagnosis of hypertension after their index date. Among the comparison cohort, 12 754 (5·8%) had incident hypertension after their corresponding index date. Table 4 provides the distribution of hypertension in the two groups by age and sex. Although the sex distribution of patients with psoriasis with incident hypertension was similar to that of the whole psoriasis cohort (Table 1), these patients were substantially older than the general population with psoriasis. The age and sex distributions of patients who had incident hypertension were similar in the psoriasis and comparison cohorts.

Table 4.   Incident hypertension cases in the psoriasis and comparison cohorts
 Psoriasis (n = 44 164)
n (%)
Comparison (n = 219 784)
n (%)
Total2765 (6·3)12 754 (5·8)
Sex
 Male1332 (48·2)6147 (48·2)
 Female1433 (51·8)6607 (51·8)
Age (years)
 < 101 (0·0)4 (0·0)
 10–1914 (0·5)70 (0·6)
 20–2959 (2·1)327 (2·6)
 30–39206 (7·5)955 (7·5)
 40–49515 (18·6)2124 (16·7)
 50–59717 (25·9)3340 (26·2)
 60–69724 (26·2)3542 (27·8)
 70–79435 (15·7)2003 (15·7)
 80–8993 (3·4)368 (2·9)
 90+1 (0·0)21 (0·2)

Table 5 provides the estimated cumulative incidences of hypertension at 1, 3, 5 and 10 years after the index date in the two groups, and the cumulative incidence estimates are presented graphically in Figure 2. The cumulative incidence of hypertension is significantly higher in the psoriasis cohort than in the comparison group (P < 0·0001). The HR for incident hypertension after the index date comparing the psoriasis cohort with the comparison group is 1·09 (95% CI 1·05–1·14).

Table 5.   Estimated cumulative incidence of hypertension at specified times after the index date in the psoriasis and comparison cohorts
 CasesCumulative incidence (per 1000)95% CI (per 1000)
  1. CI, confidence interval.

Psoriasis
 1 year50114·012·9–15·3
 3 years79642·239·9–44·5
 5 years52168·265·2–71·4
 10 years732138·5132·7–144·6
Comparison
 1 year221112·111·6–12·6
 3 years344036·135·2–37·1
 5 years244160·459·0–61·7
 10 years3610129·4126·9–132·1
Figure 2.

 Cumulative incidence of hypertension in the psoriasis and comparison cohorts.

Obesity

Among the psoriasis cohort, 2760 (6·3%) patients had a new diagnosis of obesity (defined as body mass index ≥ 30 kg m−2) after their index date. Among the comparison cohort, 11 996 (5·5%) had incident obesity recorded after their corresponding index date. Table 6 provides the distribution of obesity in the two groups by age and sex. Those with obesity are more likely to be female than in the psoriasis population as a whole (Table 1), and they are also older. The same is observed in the comparison cohort.

Table 6.   Incident obesity cases in the psoriasis and comparison cohortsa
 Psoriasis (n = 44 164)
n (%)
Comparison (n = 219 784)
n (%)
  1. aObesity is defined as body mass index ≥ 30 kg m−2.

Total2760 (6·3)11 996 (5·5)
Sex
 Male1183 (42·9)5274 (44·0)
 Female1577 (57·1)6722 (56·0)
Age (years)
 < 1016 (0·6)85 (0·7)
 10–19225 (8·2)903 (7·5)
 20–29342 (12·4)1560 (13·0)
 30–39415 (15·0)2007 (16·7)
 40–49531 (19·2)2307 (19·2)
 50–59561 (20·3)2388 (19·9)
 60–69453 (16·4)1856 (15·5)
 70–79191 (6·9)785 (6·5)
 80–8925 (0·9)103 (0·9)
 90+1 (0·0)2 (0·0)

Table 7 provides the estimated cumulative incidences of obesity at 1, 3, 5 and 10 years of follow-up after the index date, and the cumulative incidence estimates are presented graphically in Figure 3. The cumulative incidence of obesity is significantly higher in the psoriasis cohort than in the comparison group (P < 0·0001). The HR for incident obesity after the index date comparing the psoriasis cohort with the comparison group is 1·18 (95% CI 1·14–1·23).

Table 7.   Estimated cumulative incidence of obesity at specified time after the index date in the psoriasis and comparison cohortsa
 CasesCumulative incidence (per 1000)95% CI (per 1000)
  1. CI, confidence interval.

  2. aObesity is defined as body mass index ≥ 30 kg m−2.

Psoriasis
 1 year52514·813·6–16·1
 3 years77642·139·9–44·5
 5 years51567·764·7–70·9
 10 years745139·0133·2–145·1
Comparison
 1 year219111·811·3–12·3
 3 years333534·633·8–35·6
 5 years229957·055·7–58·3
 10 years3241118·0115·5–120·5
Figure 3.

 Cumulative incidence of obesity in the psoriasis and comparison cohorts (defined as body mass index ≥ 30 kg m−2).

Hyperlipidaemia

Among the psoriasis cohort, 1900 (4·3%) patients had a new diagnosis of hyperlipidaemia after their index date. Among the comparison cohort, 8111 (3·7%) had incident hyperlipidaemia after their corresponding index date. Table 8 provides the distribution of hyperlipidaemia in the two groups by age and sex. Those with hyperlipidaemia are more likely to be male than the general psoriasis population (Table 1), and they are older than the psoriasis cohort as a whole. The same differences are observed in the comparison cohort.

Table 8.   Incident hyperlipidaemia cases in the psoriasis and comparison cohorts
 Psoriasis (n = 44 164)
n (%)
Comparison (n = 219 784)
n (%)
Total1900 (4·3)8111 (3·7)
Sex
 Male978 (51·5)4074 (50·2)
 Female922 (48·5)4037 (49·8)
Age (years)
 < 100 (0·0)3 (0·0)
 10–191 (0·1)13 (0·2)
 20–2935 (1·8)126 (1·6)
 30–39112 (5·9)539 (6·7)
 40–49319 (16·8)1354 (16·7)
 50–59572 (30·1)2337 (28·8)
 60–69580 (30·5)2502 (30·9)
 70–79257 (13·5)1105 (13·6)
 80–8921 (1·1)130 (1·6)
 90+3 (0·2)2 (0·0)

Table 9 provides the estimated cumulative incidences of hyperlipidaemia at 1, 3, 5 and 10 years of follow-up after the index date, and the cumulative incidence estimates are presented graphically in Figure 4. The cumulative incidence of hyperlipidaemia is significantly higher in the psoriasis cohort than in the comparison group (P < 0·0001). The HR for incident hyperlipidaemia after the index date comparing the psoriasis cohort with the comparison group is 1·17 (95% CI 1·11–1·23).

Table 9.   Estimated cumulative incidence of hyperlipidaemia at specified times after the index date in the psoriasis and comparison cohorts
 CasesCumulative incidence (per 1000)95% CI (per 1000)
  1. CI, confidence interval.

Psoriasis
 1 year3057·87·0–8·8
 3 years49523·822·2–25·5
 5 years37741·038·7–43·5
 10 years57091·186·5–96·0
Comparison
 1 year12236·25·9–6·6
 3 years217220·319·7–21·0
 5 years152634·433·5–35·4
 10 years238877·775·7–79·7
Figure 4.

 Cumulative incidence of hyperlipidaemia in the psoriasis and comparison cohorts.

Multiple risk factors

Among the psoriasis cohort, 9022 (20·4%) patients had a single risk factor (diabetes, hypertension, obesity or hyperlipidaemia) recorded for the first time after their index date. An additional 2566 (5·8%) patients had two of these diagnoses first recorded after their index date. A further 717 (1·6%) had three of these diagnoses and 137 (0·3%) had all four first-time diagnoses recorded after they were first diagnosed with psoriasis. In the comparison group, the respective numbers and percentages of subjects with one, two, three and all four risk factors diagnosed for the first time after their index date were 39 811 (18·1%), 10 303 (4·7%), 2586 (1·2%) and 435 (0·2%).

Myocardial infarction and other vascular diseases

In addition to evaluating risk factors for MI, we estimated the cumulative incidence of MI itself and several other forms of vascular disease in the psoriasis and comparison cohorts.

Among the psoriasis cohort, 596 (1·4%) had a first-time diagnosis of MI after their index date. Among the comparison cohort, 2459 (1·1%) had an incident MI after their corresponding index date. The age and sex distributions for cases of incident MI are presented in Table 10. Those with MI were more likely to be male and were older than the psoriasis population as a whole (Table 1), and the same was observed in the comparison cohort.

Table 10.   Incident myocardial infarction cases in the psoriasis and comparison cohorts
 Psoriasis (n = 44 164)
n (%)
Comparison (n = 219 784)
n (%)
Total596 (1·4)2459 (1·1)
Sex
 Male378 (63·4)1596 (64·9)
 Female218 (36·6)863 (35·1)
Age (years)
 < 100 (0·0)0 (0·0)
 10–191 (0·2)3 (0·1)
 20–293 (0·5)4 (0·2)
 30–3921 (3·5)55 (2·2)
 40–4952 (8·7)187 (7·6)
 50–59128 (21·5)472 (19·2)
 60–69176 (29·5)744 (30·3)
 70–79164 (27·5)719 (29·2)
 80–8942 (7·1)262 (10·7)
 90+9 (1·5)13 (0·5)

Table 11 provides the estimated cumulative incidences of MI at 1, 3, 5 and 10 years of follow-up after the index date, and the cumulative incidence estimates are presented graphically in Figure 5. The cumulative incidence of MI is significantly higher in the psoriasis cohort than in the comparison group (P < 0·0001). The HR for incident MI after the index date comparing the psoriasis cohort with the comparison group is 1·21 (95% CI 1·10–1·32).

Table 11.   Estimated cumulative incidence of myocardial infarction at specified times after the index date in the psoriasis and comparison cohorts
 CasesCumulative incidence (per 1000)95% CI (per 1000)
  1. CI, confidence interval.

Psoriasis
 1 year1032·62·1–3·1
 3 years1778·27·3–9·2
 5 years11313·312·0–14·7
 10 years16527·725·2–30·4
Comparison
 1 year4402·22·0–2·4
 3 years7196·76·3–7·1
 5 years45910·910·4–11·5
 10 years67922·621·6–23·7
Figure 5.

 Cumulative incidence of myocardial infarction in the psoriasis and comparison cohorts.

A first-time diagnosis of angina was more frequently recorded in the psoriasis cohort [1006 cases (2·3%)] than in the comparison group [4186 cases (1·9%)]. With the comparison cohort as the reference group, the HR for incident angina after the index date in the psoriasis cohort was 1·20 (95% CI 1·12–1·29).

Incident diagnoses of atherosclerosis were more frequently recorded in the psoriasis cohort [214 cases (0·5%)] than in the comparison group [831 cases (0·4%)]. With the comparison cohort as the reference group, the HR for incident atherosclerosis after the index date in the psoriasis cohort was 1·28 (95% CI 1·10–1·48).

Incident peripheral vascular disease was more frequently diagnosed in the psoriasis cohort [279 cases (0·6%)] than in the comparison group [1071 cases (0·5%)]. The HR for incident peripheral vascular disease after the index date in the psoriasis cohort was 1·29 (95% CI 1·13–1·47).

Incident stroke was more frequent in the psoriasis cohort [388 cases (0·9%)] than in the comparison group [1724 cases (0·8%)]. The HR for incident stroke after the index date in the psoriasis cohort was 1·12 (95% CI 1·00–1·25).

Drugs prescribed after the index date

Among the psoriasis cohort, retinoids were prescribed to 530 (1·3%) patients after their index date; immunosuppressants were prescribed to 847 (1·9%) patients; antiproliferatives were prescribed to 969 (2·2%) patients; systemic corticosteroids were prescribed to 8394 (19·0%) patients; topical steroids were prescribed to 24 217 (54·8%) patients; and 159 (0·4%) were recorded as having received UV-based therapies. These treatments were not all indicated for psoriasis therapy but are included here as they may have mechanisms of action that would increase the likelihood of developing one or more of the risk factors of interest in our study and most are at least sometimes prescribed for psoriasis treatment. Among the comparison cohort, the respective numbers and percentages of subjects who had these treatments recorded after their index date were as follows: retinoids 1208 (0·5%); immunosuppressants 1597 (0·7%); antiproliferatives 1871 (0·9%); systemic corticosteroids 34 435 (15·7%); topical steroids 57 997 (26·4%); and UV-based therapies 25 (0·01%).

Discussion

Using information from the U.K. GPRD, we have found that the risks of incident diabetes (HR 1·33; 95% CI 1·25–1·42), hypertension (HR 1·09; 95% CI 1·05–1·14), obesity (HR 1·18; 95% CI 1·14–1·23) and hyperlipidaemia (HR 1·17; 95% CI 1·11–1·23) are all moderately increased in patients with psoriasis during follow-up after their first-time diagnosis of psoriasis (in relation to a comparison cohort without psoriasis who were matched to the patients with psoriasis on age, sex and calendar time). We further found that the risks of incident MI (HR 1·21; 95% CI 1·10–1·32), angina (HR 1·20; 95% CI 1·12–1·29), atherosclerosis (HR 1·28; 95% CI 1·10–1·48), peripheral vascular disease (HR 1·29; 95% CI 1·13–1·47) and stroke (HR 1·12; 95% CI 1·00–1·25) are all increased to a similar degree in patients after their first-time diagnosis of psoriasis.

Patients with psoriasis with these outcomes have age and sex distributions that differ from those of the general population with psoriasis in ways that are consistent with known risk factors for each outcome. For example, patients with psoriasis with MI are more likely to be male and older than the psoriasis population as a whole, as male sex and older age are known risk factors for MI.

Because of the comprehensive nature of the U.K. National Health Service, it is unlikely that the differences in cumulative incidence we have observed between the psoriasis and comparison cohorts represent ascertainment bias (i.e. a result of possibly differential frequency of medical attention required by patients with psoriasis in relation to the comparison cohort). As the diagnosis of psoriasis preceded the first-time diagnosis of the outcomes we studied, it is possible that psoriasis and/or its treatment may be causes of these illnesses, either directly or indirectly. Indeed, as expected, all of the treatments we evaluated were given to the patients with psoriasis more frequently than to the comparison cohort. Finally, it is also plausible that other, unidentified genetic or environmental factors are common causes of psoriasis and of the outcomes of interest in our study, and that the HRs we have estimated are biased upwards due to residual confounding (by factors other than age and sex).

We consider it unlikely that all of the differences we observed between the psoriasis cohort and the comparison population in the risks of the outcomes of interest were due to a small number of patients with psoriasis having multiple risk factors. In fact, among the 12 442 patients with psoriasis with any risk factors, 11 588 had one or two risk factors while only 854 had three or four risk factors. However, it is impossible to be certain of causal relationships between exposures and outcomes of interest based on a descriptive study such as this.

Previous reports of an association between psoriasis and cardiovascular risk factors2–4 have evaluated the prevalence of each condition using cross-sectional designs. While studies of prevalence are useful for estimating disease burden in a population, such studies have shortcomings with respect to aetiological research – namely, they do not take account of the relative times of onset of associated diseases and may confuse differences in disease duration with aetiological influences. By designing our study to focus on a cohort of patients with first-time diagnoses of psoriasis and incident diagnoses of risk factors for cardiovascular disease (or cardiovascular disease outcomes), we intended to provide more accurate estimates of the relative risk of these outcomes among patients with psoriasis compared with the general population.

A study by Gelfand et al.1 using the GPRD reported an increased risk of MI in patients with mild (unadjusted HR 1·11; 95% CI 1·07–1·17) and severe (unadjusted HR 1·43; 95% CI 1·18–1·72) psoriasis. However, this study included patients with a history of MI. Patients with a history of MI constituted a statistically significantly higher proportion of the cohorts with severe (1·98%) and mild (1·78%) psoriasis than the comparison group (1·4%). Although these authors used multivariable Cox regression in an attempt to adjust their analysis of the relative frequency of MI for several potentially confounding factors, it remains that they did not, in fact, study the incidence of MI (i.e. first-time diagnoses). As a history of MI is the most potent known risk factor for subsequent MI, this is an important difference between their study and ours. Even if their interest had explicitly been directed towards the effect of psoriasis on first or subsequent MI (as opposed to incident MI), it is not clear that confounding due to a history of MI would have been fully corrected in their statistical analysis. A more useful aspect of these authors’ study was that they applied a set of clinical criteria they developed to distinguish between the risk of first or subsequent MI in patients with ‘mild’ vs. ‘severe’ psoriasis.

Our study, like the work of Gelfand and colleagues,1,2 used information from a highly reliable and extensively validated automated medical record database.5 Due to the training given to GPs in recording diagnoses, treatments, referrals and hospitalizations, and the comprehensive nature of the U.K. National Health Service, it is unlikely that a substantial proportion of the diagnoses recorded in the GPRD is inaccurate or that a large number of outcome events in the study population would be missing. Moreover, this data resource has a number of other strengths, such as the long duration of information recorded in the GPRD (which began in the late 1980s); the relative geographic stability of the patient population and assignment of patients to specific GPs as medical system ‘gatekeepers’; and the recording of important diagnoses with dates before the start of follow-up in the GPRD (as historical entries), which can be used to ascertain whether a diagnosis recorded during follow-up is incident or not.

A limitation of our study is that we did not take into account various other known risk factors for cardiovascular disease, such as smoking, for example. Our study was designed as an exploratory (descriptive) analysis, using a more rigorous design than previous work, in order to lay the groundwork for more extensive subsequent work directed more specifically towards investigating whether there may be causal associations between psoriasis and/or its treatment and either risk factors for cardiovascular disease or various cardiovascular diseases themselves. This report should be viewed as a first step in that direction.

In summary, patients with psoriasis are at moderately higher risk than the general population of subsequently developing diabetes, hypertension, obesity, hyperlipidaemia, MI, angina, atherosclerosis, peripheral vascular disease and stroke. Further work is needed to investigate whether these associations involve causal factors related to psoriasis or its treatment.

Acknowledgments

This study was funded by Amgen, Inc.

Ancillary