Background
Psoriatic disease, encompassing skin psoriasis and psoriatic arthritis, is a common condition affecting 2-3% of the Western population associated with reduced quality of life and increased healthcare costs. To improve patients’ lives and the stewardship of societal resources, a nuanced understanding of the associations between patient characteristics and health outcomes are needed for optimal clinical and societal decision making. While randomised clinical trials play a central role in evidence-based medicine, they have limitations relating to patient selection, follow-up duration, and idealised clinical conditions. In recent years, the role of real-world evidence in health research has grown, including in dermatology, due in part to its ability to describe and compare patients in routine clinical care. This thesis contains four analyses using Swedish real-world data describing associations between patient characteristics and outcomes in psoriatic disease.
Methods
Data from several administrative population registers in Sweden and the clinical registry PsoReg were collected for those with psoriatic disease in routine clinical care and linked at the patient level. The data include diagnoses, pharmacy dispensed medications, mortality, socioeconomic information, quality of life, and clinical severity. A matched non-psoriatic control group was also extracted.
The first analysis assessed the association between disease severity (measured by the Psoriasis Area and Severity Index [PASI]) and health-related quality of life (measured the EuroQol 5-dimension instrument [EQ-5D]) in 2,674 patients with longitudinal follow-up in PsoReg, a fixed effects regression model was deployed adjusting for confounding factors including certain types of unobserved confounding.
Second, the independent associations of skin psoriasis and somatic comorbidity with incident psychiatric illness were described across 93,239 psoriasis patients and 1.4 million controls. Their individual contributions and synergistic interaction were assessed using a time-to-event model. Diagnosis codes were used to construct the composite psychiatric illness outcome consisting of depression, anxiety, and suicidality. Diagnosis codes were also used to create the Elixhauser and Charlson comorbidity indexes to measure somatic comorbidity.
The third analysis assessed age-related inequities in biologic prescriptions in 1,465 patients enrolled in PsoReg using a time-to-event analysis controlling for decision-making variables including disease severity. The analysis also described the non-parametric relationship between age and incident biologic prescriptions using a kernel regression.
Finally, the fourth analysis describes rates of non-persistence in individuals with psoriatic arthritis treated with adalimumab, ustekinumab, and secukinumab using a time-to-event model. A total of 4,649 drug exposure period across 3,918 patients were assessed. Non-persistence was defined as a treatment switch or discontinuation, the latter defined as a failure to refill an existing prescription within two times the days of drug supplied.
Results
The analysis found a statistically significant, negative association between PASI and the EQ-5D ( =-0.0186, 95% confidence interval [CI]: -0.0360, -0.0201) which was non-linear ( =0.0003, 95% confidence interval [CI]: 0.0001, 0.0004). Incremental PASI improvements for those with less skin involvement were associated with larger increases in quality of life than in those with more skin involvement.
Skin psoriasis was found to be independently associated with the onset of psychiatric illness (hazard ratio [HR]=1.32, 95% CI: 1.27-1.36) as was somatic comorbidity (HR=2.09, 95%CI: 2.06-2.13). However, the results were compatible with a lack of synergistic effect between skin psoriasis and somatic comorbidity on psychiatric illness (HR=0.93; 95% CI: 0.81-1.04).
Older psoriasis patients appeared less likely to initiate biologic therapies than their younger counterparts after controlling for disease severity and comorbidity (HR=0.97, 95% CI: 0.95-0.99).
Individuals with psoriatic arthritis treated with ustekinumab were found to have lower rates of non-persistence compared to adalimumab (HR=0.56, 95% CI: 0.49-0.64). Secukinumab and adalimumab appeared to have similar rates of non-persistence (HR=1.01, 95% CI: 0.88-1.15), although the results differed in biologic-naïve and biologic-experienced subgroups. The definition of discontinuation was sensitive and had material effects on the results.
Conclusions
Patient characteristics appear to play an important role in a variety of health outcomes in psoriatic disease, demonstrated across four real-world settings. The utilisation of data from routine clinical care enabled the investigation of research questions that are not suitable for clinical trial contexts, providing relevance for patients in everyday clinical practice. The study designs and methodologies applied in this thesis are associated with a variety of strengths and limitations, many of which are closely linked to the observational nature of the data, which have material importance for the interpretation of the results and implications for healthcare and policy. Understanding the associations between patient characteristics and subsequent outcomes is an important element in the delivery of holistic, personalised healthcare.