Workstream 1 studies 1.i and 1.ii () address objective 1 of the programme: to confirm which patients with psoriasis are at highest risk of developing additional long-term conditions.
Workstream 1: relationship of studies 1.i and 1.ii to other IMPACT programme workstreams.
Publications relating to this section and workstream are listed in Publications and cited throughout this section.
Prevalence and incidence of psoriasis: a systematic review
The worldwide epidemiology of psoriasis is poorly understood. A systematic review that provided a detailed critique of the existing literature on the worldwide incidence and prevalence of psoriasis was undertaken as part of workstream 1, comparing studies in relation to geography, age and gender. See Parisi et al.1
The results from the systematic review confirmed that psoriasis is a common disease, and is less common in children and more common in adults. Prevalence rates showed a worldwide geographic variation that probably reflects the fact that psoriasis is a complex disease influenced by both genetic and environmental factors. The incidence of psoriasis in the UK is estimated to be 140 per 100,000 person-years.35 Studies on the prevalence of psoriasis in the UK highlight that the disease is uncommon before the age of 9 years (0.55%)36 and has an estimated prevalence of between 1.30% [95% confidence interval (CI) 1.21% to 1.39%]37 and 2.60% (95% CI 2.47% to 2.78%)38 in adults and between 1.48% (95% CI 1.20% to 1.80%) and 1.87% (95% CI 1.89% to 1.91%)6,36,39 taking into account all ages.
Furthermore, some studies have found an increasing trend in the prevalence of psoriasis with age;36,39 however, there is no agreement about whether or not the prevalence differed between men and women.36,39 The systematic review served to inform the recently published World Health Organization’s Global Report On Psoriasis40 and a subsequent call for further epidemiological research on the disease.
Systematic review of published economic evaluations
Although the economic systematic review and economic model of costs were both conceived as part of workstream 1, they were developed in parallel with other workstreams. In particular, the economic model drew on the findings of these to inform its structure and to populate it with data. The methods and results of the systematic review are reported here and those of the economic model are reported in Valuing the interventions with a stated preference survey.
A systematic review of published economic evaluations was conducted to understand what was known about the cost-effectiveness of psoriasis management.41 The review’s aims were to identify full economic evaluations that compared the costs and health benefits of alternative interventions, summarise what was known about the relative cost-effectiveness of different interventions and assess the quality of the evidence, uncertainties and evidence gaps. The review highlighted inconsistencies between analyses and uncertainties where models have so far struggled to accurately characterise the disease. The review informed the need to develop a new model structure and the direction such a model should take.
The databases EMBASE, MEDLINE and NHS Economic Evaluation Database (NHS EED) were searched for full economic evaluations in January 2012, and updated with a search of NHS EED in April 2014. Predefined inclusion and exclusion criteria were used to screen abstracts and titles and select papers for review. Studies were included in the review if they compared services or treatments for psoriasis in adult patients, measured health outcomes and costs, and used either primary data or synthesised data in a full economic evaluation. The search identified 1355 articles. Three reviewers independently performed the primary and secondary screening of identified papers, cross-checking their results with each other. The full paper was obtained and reviewed for titles and abstracts when one or more reviewer was uncertain. Any discrepancies or uncertainties about the inclusion of papers in the secondary screening were resolved by discussion between the reviewers, with reviewer four acting as arbiter for any remaining uncertainty or disagreement. Two reviewers extracted data from the studies included for full review. Predefined data extraction forms and quality assessment forms were used (see Appendix 1). A total of 37 papers met the inclusion criteria and reported 71 treatment comparisons. The treatments evaluated in the 71 comparisons were systemic (n = 45), topical (n = 22), phototherapies (n = 14) and combinations (n = 4). The 37 economic papers reviewed mainly evaluated individual therapeutic agents rather than packages of care. Typically, these evaluations did not consider the context in which the treatment was delivered. Four articles and seven comparisons directly addressed the organisation and delivery of care. These included a programme to support patients’ self-management at home, online care management and home-based phototherapy. Four of the papers explored differing time and convenience demands on patients.
The review indicated that most of the economic evaluations were modelling studies, synthesising data from several sources. In summary, the systematic review identified a number of key areas of uncertainty in the existing psoriasis economic literature, which future economic analyses should seek to improve on. A limited effectiveness evidence base is over-represented in the available economic evidence and, overall, there was a lack of high-quality head-to-head clinical comparisons of different interventions. Repeated use of previous model structures was commonplace. When different sources of evidence or models have been used, uncertainty persists owing to diversity in setting, perspective and study design.
Many of the studies were limited in terms of reporting the methods used. In addition, the short-term follow-ups used in clinical trials of new interventions meant that 27 out of 37 economic evaluations were restricted to a time horizon of ≤ 1 year, despite the chronic nature of psoriasis. The review found that parameter uncertainty was not typically incorporated into analyses to a suitable degree.
Although the current evidence base is inconclusive about the relative cost-effectiveness of individual treatments, it contributes valuable data and methods to inform complex decisions and develop robust evaluation methods.
Risk of major cardiovascular events in patients with psoriasis: a population cohort study using the Clinical Practice Research Datalink
The severity of psoriasis can range from limited disease involving small body surface area to extensive skin involvement. As described in Synopsis, people affected by the disease often have an impaired quality of life. People with psoriasis may have other comorbid conditions such as obesity, hypertension, hyperlipidaemia and diabetes mellitus, which are associated with an increased risk of CVD.
In the last decade, a number of studies have suggested an association between psoriasis and CVD. It has been argued that increased systemic inflammation in those with psoriasis exacerbates other chronic inflammatory diseases including atherosclerosis, which could lead to MI or stroke (see Synopsis).
However, any purported association between psoriasis and CVD is complex for several reasons: those with psoriasis are more likely to engage in unhealthy lifestyle habits (increased likelihood of smoking, low levels of physical activity and obesity),28 have higher prevalence of CVD risk factors (e.g. diabetes mellitus, hypertension and hyperlipidaemia)23 and therapies that may raise (e.g. ciclosporin)28 or lower (e.g. methotrexate)29 the CVD risk. Each of these are aspects that may confound the association between the two conditions.30
After controlling for several major CVD risk factors, a number of studies have suggested an increased risk of fatal and non-fatal CVD events in patients with psoriasis.18,42–51 By contrast, other studies have concluded that psoriasis is not an independent risk factor for CVD.52–55 In a recent systematic review, Samarasekera et al.56 concluded that a possible association between severe psoriasis and CVD may exist; however, the authors warn that the existing studies were limited by failing to adequately adjust for important risk factors.
Inflammatory arthritis, a common comorbidity in patients with psoriasis and a recognised risk factor for CVD,57–59 has rarely been considered as a possible confounder. It is also important to note that, in many studies using electronic medical record databases, severe psoriasis is typically defined by exposure to systemic or biologic therapies that may also be used to treat inflammatory arthritis. This raises the possibility of misclassification of severe psoriasis when not taking account of the presence of inflammatory arthritis. Furthermore, little consideration has been given in earlier studies to the time-varying nature of the development of risk factors, or the severity of psoriasis.30
A large population-based cohort study was undertaken with the aim of investigating whether or not psoriasis is an independent risk factor for major CV events [including MI, acute coronary syndrome (ACS), unstable angina and stroke] when taking into account relevant risk factors for CVD. See Parisi et al.30
Methods
Study design
Using a primary care database from the UK [specifically the Clinical Practice Research Datalink (CPRD)], a population-based cohort study was conducted that included people with and people without psoriasis. The CPRD comprises the entire medical history (demographics, treatments, clinical events, test results and referrals to hospitals) of patients registered in a general practice in the UK. The database is broadly representative of the UK population in terms of age and gender. The protocol of this study was approved by the CPRD’s Independent Scientific Advisory Committee (protocol reference number 11_134A). The study is reported according to Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.60
At the time data were extracted (September 2012), data were available for 652 practices and > 12 million patients.
Study population
An inception cohort of adult (aged ≥ 20 years) patients with psoriasis and a matched comparison group of up to five people without psoriasis were identified between 1994 and 2009. Patients with and patients without psoriasis were included if they had no history of CVD or diabetes mellitus before index date (first diagnosis of psoriasis or corresponding consulting date for comparison patients) and were registered for ≥ 2 years in the general practice before entry into the study cohort.
Comparison patients were individuals who had never received a diagnosis for psoriasis. For each patient with psoriasis up to five comparison patients were matched on age, gender and general practice. Furthermore, to ensure that comparison patients visited the practice at approximately the same time as when patients with psoriasis received their first diagnosis of psoriasis, comparison patients were also matched on index date (psoriasis diagnosis date) in a 6-month window. All patients were followed up from their respective index date (first diagnosis of psoriasis) or consulting date (for the comparison cohort) and ended at the earliest date of the occurrence of a major CV event (MI, ACS, unstable angina or stroke), transfer out of the practice, death date or end of follow-up (31 December 2011).
Definition of exposure
Patients with a first diagnosis of psoriasis between 1994 and 2009 and received a recognised treatment for psoriasis {(emollients, topical treatment, phototherapy, systemic therapy or biologics [National Institute for Health and Care Excellence (NICE) clinical guideline CG153]} were included in the cohort.61 Patients were classified as having severe psoriasis if they had received a systemic treatment (acitretin, etretinate, ciclosporin, hydroxycarbamide, methotrexate, fumaric acid), phototherapy or a biologic therapy (etanercept, adalimumab, infliximab, ustekinumab, efalizumab); if they had received none of these they were classified as having mild psoriasis.30
Outcome of interest
The outcome of interest was a combined CV end point, including the first event of fatal or non-fatal MI, ACS, unstable angina or stroke. In the main analysis, the outcome of interest was identified using data from CPRD by using the Read code classification, which is a hierarchical coding system used to record diagnoses in primary care.62 To assess the robustness of the main findings to potential outcome misclassification, in a sensitivity analysis, the combined CV end point was identified using primary care data (CPRD) and the national mortality records [from the Office for National Statistics (ONS)] of those patients in practices providing linked data between CPRD and ONS. In this situation, both Read codes and the International Classification of Diseases codes were used.
Covariates
To assess whether or not there is a relationship between psoriasis and risk of major CV events, a statistical model was built including multiple risk factors (other than psoriasis). This approach will help us to understand whether or not any observed association between psoriasis and major CV events could be accounted for by other concomitant diseases/risk factors that are known to be linked to CVD. Risk factors included in the model were age, gender, depression and calendar year calculated at baseline and having or developing severe psoriasis or inflammatory arthritis (which included both diagnostic codes for PsA or rheumatoid arthritis). Diabetes mellitus, chronic kidney disease, hypertension, hyperlipidaemia, atrial fibrillation (AF), transient ischaemic attack, congestive heart failure, thromboembolism, valvular heart disease and smoking status were modelled as time-varying risk factors; therefore, an individual’s classified status could change during the study period. Psoriasis was defined as being severe from the date of first exposure to phototherapy, systemic or biologic treatment. It was considered as severe from that point onward. Smoking status was classified as ‘current’, ‘former’, ‘never smoker’ or ‘unknown smoking status’. This enabled switching from one smoking class to another during follow-up.30 Additional risk factors such as a high body mass index (BMI) score and a high score on the Index of Multiple Deprivation (IMD),63 which is a measure of socioeconomic status, were included only in sensitivity analyses owing to the high proportion of missing data.
All code lists used for the exposure and outcome are available for download [URL: www.clinicalcodes.org (accessed 15 May 2019)].64
Statistical analysis
Medians and interquartile ranges were used to summarise continuous variables; proportions were used to summarise discrete covariates. The combined CV end-point events and incidence rates per 1000 person-years with 95% CIs were calculated for patients with and patients without psoriasis, and by disease severity. The estimate of the age- and gender-adjusted hazard ratios (HRs) and 95% CIs for each variable were made using Cox proportional hazard regression. Cox regression with a shared frailty model was used to estimate the fully adjusted HRs and 95% CIs. The assumption of proportionality for each variable and the model overall was tested by using Schoenfeld residuals.
Using the initial cohort identified from CPRD, multiple sensitivity analyses were performed. These included assessing whether or not a more complex or parsimonious multivariate model (by including different sets of risk factors) would change the conclusions. BMI and IMD scores contained a high proportion of missing data; therefore, they were included in the model only in sensitivity analyses. Smoking status had a small proportion of missing values; therefore, it was included in the main analysis and a category was introduced to account for missing values. Accordingly, sensitivity analyses explored the impact of using multiple imputation to estimate missing BMI and IMD scores so that they could be included in the model.
Additional sensitivity analyses were (1) including only those patients with at least one GP visit per year, (2) including only those patients with ≥ 6 months of follow-up, (3) testing for an interaction between the presence of psoriasis or severe psoriasis with age and (4) additional adjustment for patients exposed to methotrexate, ciclosporin or oral retinoids.
Finally, a subgroup analysis was performed by linking data from CPRD to the mortality data contained in the ONS. The nested cohort included patients identified from CPRD practices linked to the ONS and who had an index date (first diagnosis of psoriasis or corresponding consulting date) between 1998 and 2009; they were followed up until 2011. The outcome was the first fatal or non-fatal major CV event (MI, ACS, unstable angina or stroke) recorded in CPRD or ONS.
All statistical analyses were performed using Stata® 12 (StataCorp LP, College Station, TX, USA).
Results
The number of patients meeting all inclusion criteria who formed the final cohort was 48,523 patients with psoriasis and 208,187 comparison patients. Overall, there was a higher proportion of females (56.40%) and the median age at index date was 47 years (see table 1 of Parisi et al.30).
At baseline, patients with psoriasis had a higher prevalence of the majority of risk factors (inflammatory arthritis, hypertension, hyperlipidaemia, depression, current or ex-smoker, overweight or obese) than the comparison group (see original table 1). At the end of follow-up, patients with psoriasis had a higher prevalence of all time-varying risk factors except for AF, transient ischaemic attack and congestive heart failure (see original table 2). Inflammatory arthritis was present in 2.39% of patients with psoriasis and 0.98% of the comparison patients at baseline. It was also present in 4.69% of the patients with psoriasis and 1.38% of the comparison patients by the end of follow-up. In addition, at baseline, 1.03% of patients with psoriasis had been treated with phototherapy or systemic or biologic therapies. This increased to 4.29% of patients by the end of follow-up. Just over 50% (50.62%) of those treated with systemic or biologic therapies had also been diagnosed with inflammatory arthritis by the end of the follow-up period.30
The most commonly used systemic treatment used in patients with psoriasis was methotrexate. Original table 3 shows the distribution of phototherapy, systemic therapy or biologic received by patients with psoriasis.
Patients included in the study were followed up for a median time of 5.2 years. Over this period, 1257 (2.59%) patients with psoriasis had a major CV event, compared with 4784 (2.30%) comparison patients (see original table 4 of the published paper). The unadjusted incidence rate of a major CV event per 1000 person-years was higher in the psoriasis group than in the comparison group [4.13 per 1000 person-years, 95% CI 3.91 to 4.36, and 3.87 per 1000 person-years, 95% CI 3.76 to 3.98, respectively] (see original table 4).
There were time-varying effects related to risk of outcome for hypertension, transient ischaemic attack, AF and gender (revealed in Schoenfeld residuals). Allowing each of these variables to have different effects for the first 3 years of follow-up and the later follow-up did, however, remove the non-proportionality (p = 0.12).
In the age- and gender-adjusted analysis, all the variables tested were significantly associated with the risk of major CV events. In particular, the HRs of major CV events due to psoriasis and severe psoriasis were 1.10 (95% CI 1.04. to 1.17) and 1.40 (95% CI 1.07 to 1.84), respectively. However, both HRs were attenuated and became non-significant in the fully adjusted model (HR 1.02, 95% CI 0.95 to 1.08, and HR 1.28, 95% CI 0.96 to 1.69, respectively) (original table 5 of the published paper).
In addition, in the multivariate analysis, each of the following risk factors was significantly related to the risk of major CV events (original table 5): inflammatory arthritis (HR 1.36, 95% CI 1.18 to 1.58), diabetes mellitus (HR 1.18, 95% CI 1.06 to 1.31), chronic kidney disease (HR 1.18, 95% CI 1.07 to 1.31), hypertension (HR 1.37, 95% CI 1.29 to 1.45), transient ischaemic attack (HR 2.74, 95% CI 2.41 to 3.12), AF (HR 1.54, 95% CI 1.36 to 1.73), valvular heart disease (HR 1.23, 95% CI 1.05 to 1.44), thromboembolism (HR 1.32, 95% CI 1.17 to 1.49), chronic heart failure (HR 1.57, 95% CI 1.39 to 1.78), depression (HR 1.16, 95% CI 1.01 to 1.34), current smoker (HR 2.18, 95% CI 2.03 to 2.33), age (HR 1.07, 95% CI 1.07 to 1.07) and male gender (HR 1.83, 95% CI 1.69 to 1.98). Hyperlipidaemia was not significantly related to the risk of major CV events (HR 1.04, 95% CI 0.96 to 1.11).
Interaction terms between psoriasis or severe psoriasis and age were tested; however, they were not included in the multivariate model because they were non-significant (p = 0.40 and p = 0.25, respectively).
The sensitivity analyses did not change the main findings. The HRs of major CV events due to psoriasis or severe psoriasis when taking into account the different sets of risk factors can be found in original table 6 of the published paper. In particular, the HR of major CV events due to severe psoriasis changed from 1.28 (95% CI 0.96 to 1.69) to 1.46 (95% CI 1.11 to 1.92) when inflammatory arthritis was not included in the model.
When using multiple imputation and adding BMI and IMD scores to the fully adjusted model, the HRs of major CV events were > 1 but still non-significant (see original table 6).
Likewise, the results obtained by analysing data from patients with at least one GP visit per year, patients with ≥ 6 months’ follow-up or results that took into account patients exposed to methotrexate, ciclosporin or oral retinoids gave the same results as the main findings (see original table 6).
Finally, the subgroup analysis that analysed data from CPRD linked to the ONS mortality data and patient-level socioeconomic status (IMD) yielded similar results to the main findings. Here, the fully adjusted HRs of major CV event due to psoriasis and severe psoriasis were 1.02 (95% CI 0.93 to 1.11) and 1.10 (95% CI 0.72 to 1.68), respectively (see original table 6).
Discussion
This study investigates whether or not psoriasis is independently associated with a higher risk of CVD. The results confirm that patients with psoriasis have more prevalent comorbid conditions associated with CVD. However, taking into account other established risk factors for CVD, psoriasis itself was not found to be directly associated with short- to medium-term (3–5 years) risk of major CV events. These results highlight that individuals who have psoriasis co-occurring with inflammatory arthritis have a 36% higher risk of a major CV event than those who do not.
Similar to other studies,18,23,46 our research highlights that patients with psoriasis have a higher prevalence of CVD risk factors. Our findings are comparable to those reported by Brauchli et al.52 and Wakkee et al.,55 neither of which found an overall higher risk of MI associated with psoriasis. However, our research is more robust because the sample size was larger, 55 we used more stringent inclusion criteria (e.g. psoriasis cases were identified on the bases of diagnosis and treatment received)52 and in our analysis we accounted for important confounders including inflammatory arthritis.52,55
Our finding that psoriasis is not an independent risk factor for a major CVD event was in contrast to other influential studies that reported that it was an independent risk factor.42,44,65
In a study that utilised the Health Improvement Network database, Ogdie et al.65 reported an increased risk of major adverse CV events in patients with either mild or severe psoriasis. It is possible that the cohort of patients with psoriasis used in that study had a longer disease duration than patients identified in our cohort. Different study designs were used (prevalent vs. incident cohort). An important issue is raised by the use of a prevalent cohort, namely that it is associated with the problem of left censoring. For example, those with the most severe psoriasis or CVD could have died prior to cohort entry, meaning that these studies can address only the question of what happens to individuals with psoriasis who have already survived.30
Another study that reported an increased and significant risk of coronary heart disease in patients with severe psoriasis was conducted by Dregan et al.,44 using the CPRD; however, it is possible that they misclassified severe psoriasis on the basis of treatments received without taking account of the possibility of comorbid inflammatory arthritis.30
Using a Danish nationwide database, Ahlehoff et al.42 reported increased risks of CVD in patients with severe psoriasis with and without PsA. However, in that study, patients with severe psoriasis were classified using hospitalisation for psoriasis or PsA, which could be subject to surveillance bias.30,42 Furthermore, comorbidities were assigned by linking with prescribed treatments instead of diagnostic codes, which could lead to potential misclassification.
There are a number of additional differences between our study and earlier studies. We examined a wider range of risk factors in the analyses41,43,64 and modelled these to account for development of new risk factors over time. We examined psoriasis severity as a time-varying covariate, so that patients with severe psoriasis become ‘at risk’ only once they started phototherapy, systemic therapy or biologic therapies (rather than simply whether or not they had ‘ever’ been exposed to systemic treatment. That approach would have meant that an individual would have been classified as having severe psoriasis for the whole observation period).
Some strengths and limitations of the study can be listed. Selection bias was minimised, as was information bias and detection bias by identifying patients from the same database. We selected patients with and patients without psoriasis from the same general practice and during the same time-window.
Furthermore, our findings were consistent after multiple sensitivity and subgroup analyses. Several additional strengths can be identified in our study. (1) Important confounders were taken into account to investigate the association between psoriasis and major CV events. These included both traditional and non-traditional CVD risk factors, in particular inflammatory arthritis. (2) We present a large population-based study representative of the UK. (3) Only those with at least a diagnostic code of psoriasis plus a treatment for psoriasis were included in the cohort to minimise the risk of disease misclassification. (4) More advanced methodology was employed, such as the use of the shared frailty model. These methods take better account of the matched nature of the data, and the use of time-varying covariates.30
Some limitations also need to be taken into account. Given that this was an observational study, the risk of residual confounding was considered. The CPRD is a primary care database and therefore diagnoses of psoriasis were not necessarily confirmed by dermatologists; phototherapy and systemic and biologic therapies were used as a surrogate to classify disease severity rather than standard clinical assessments, such as the Psoriasis Area and Severity Index (PASI) or the body surface area affected by psoriasis. The size of the group of individuals with or developing severe psoriasis may be underpowered to investigate that end point of interest. The duration of follow-up was > 5 years (on average) for those with psoriasis and > 3 years for those who developed severe psoriasis during the observation period. It may be the case that chronic inflammation takes longer to produce adverse CV outcomes and therefore studies involving longer follow-up periods are recommended.30
Due to the small number of patients exposed to biologics it was not possible to discern whether or not biologic therapy reduces the risk of major CV events. The specific aim of the current study was to investigate whether or not psoriasis was independently associated with the risk of major CV events. Thus, to minimise the risk of bias, patients from our cohort who had a history of either CVD or diabetes mellitus were excluded.30
Key conclusions
Adults with psoriasis are more likely to have prevalent conditions associated with CVD.
However, psoriasis alone is not independently associated with the short- to medium-term (i.e. 3–5 years) risk of major CV events, after adjusting for important CVD risk factors.
Despite this, the co-occurrence of inflammatory arthritis and psoriasis is an independent risk factor for major CV events.
Implications
Screening psoriasis patients for inflammatory arthritis is important, as is screening to prevent development of CVD risk factors in people with psoriasis.
Patients with inflammatory arthritis are at an increased risk of CVD; this may be an additional reason to minimise a patient’s cumulative inflammatory burden.
