Ethical considerations

This observational study was approved by two Human Research Ethics Committees (St Vincent’s Hospital Melbourne Human Research Ethics and Bellberry Human Research Ethics Committee). Permission for use of the non-PSP PBS 10% sample was sought from Services Australia’s External Request Evaluation Committee (EREC MI6714). Patients prospectively enrolled were provided with a written Informed Consent Form (ICF) containing adequate information to allow them to make an informed decision to participate in the study and were given ample opportunity to discuss the study with site staff and have any questions adequately answered. The ICF was signed before any study procedures were performed. Authors who were also Principal Investigators for the study may have had access to only their own patients’ data during data collection, however for the manuscript development all authors were only provided de-identified and/or aggregated data.

Study design and endpoints

This was an observational cohort study with the primary objective of estimating the potential effect of the PSP on HUMIRA® (adalimumab) persistence in adult (≥ 18 years of age) Australian patients initiating for the first time adalimumab, as defined by the time to discontinuation. The primary endpoint was to compare the time to discontinuation between a prospective cohort of PSP patients, and either a prospective non-PSP cohort or a retrospective non-PSP cohort established from the PBS 10% sample.

The secondary objectives were to estimate the potential impact of the PSP on patient-reported outcomes measures (PROMs), including: health-related quality of life using the EQ-5D-5L (rheumatology and gastroenterology indications) or Dermatology Life Quality Index (DLQI; HS only); attitudes to medicines in general using the Beliefs about Medicines Questionnaire (BMQ):General (all indications); effect of health problems on ability to work and perform regular activities using the Work Productivity and Activity Impairment (WPAI)-General Health (GH) V2.0 (all indications); global assessment using the Patient Global Assessment of Disease Activity (PGDA; RA and PsA); evaluation of disease activity using the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI; AS only). The DLQI was used as an alternative to the EQ-5D-5L for HS, as the EQ-5D-5L is not considered a validated tool with adequate disease activity discrimination for dermatological conditions. The PGDA was used for RA and PsA only since this measure has been most widely used in these patient groups. The secondary endpoints included change from baseline in PROMs scores, and the proportion of patients with a reduction in PGDA scores of 25% and 50% from baseline, up to 12 months after initiation of adalimumab in the prospective PSP versus non-PSP cohort.

Additional secondary endpoints included assessing the impact of PSP enrolment on rates of adalimumab persistence up to 12 months after initiation and analyses by disease indications or pooled indications where the sample size allowed it, with the aim of describing any potential differences by specific disease or therapeutic area.

Supplementary analyses included time to adalimumab discontinuation compared between the prospectively enrolled patients who reported having received self-injection training and those who reported no self-injection technique training.

Subjects–Prospective cohorts

Subjects were eligible to be enrolled in the prospective cohorts of this study if they: were Australian adult (≥ 18 years of age); had a diagnosis of RA, AS, PsA, CD (moderate to severe refractory CD or fistulizing CD), UC, or HS by their treating specialist physician; were prescribed PBS-reimbursed adalimumab for the first time (initiation course) for their diagnosed condition; had not been administered the first dose at the time of study enrolment.

Patients decided whether to enrol in the PSP after a joint clinical/patient decision had been made to initiate adalimumab, and prior to being approached for potential recruitment to the study. Following screening and informed consent, patients who had opted to enrol in the PSP were included in the study’s prospective PSP cohort, while those who did not were included in the prospective non-PSP cohort. Enrolled patients were prospectively recruited from 12 urban Australian sites (7 public and 5 private), comprising specialist rheumatology (5 sites), gastroenterology (3 sites) and dermatology (4 sites) sites between May 2018 to September 2019. Enrolled patients were followed-up for a maximum of 12 months after administration of the first dose of adalimumab. Recruitment processes included study advertisement emails sent to patients enrolled in the PSP, printed study brochures/flyers provided to potentially eligible patients through HCPs at specialist practices (rheumatologists, gastroenterologists, and dermatologists) or at investigative sites, and a study registration of interest website.

Following enrolment and baseline visit procedures, patients were asked to complete baseline PROMs prior to their first dose of adalimumab. Additional PROMs were completed via an online database at timepoints according to indication.

Subjects–Retrospective cohort

A retrospective cohort comprising Australian adult (aged ≥ 18 years) patients who initiated adalimumab prior to the implementation of the PSP in 2009 was also established as a non-PSP control cohort and consisted of historical patient data extracted from the Australian Department of Human Services PBS 10% sample data, which is a nationally representative longitudinal systematic random sample of 10% of the Medicare-eligible Australian population and all of their reimbursed dispensations [24–27].

Patients were included in the retrospective non-PSP cohort if they had been dispensed an initial prescription for adalimumab (initiation course) during the sample selection window for a diagnosis of RA, PsA, or CD, based on PBS item codes. UC and HS indications were not included in the retrospective analyses as adalimumab was not approved and reimbursed by the PBS for these indications prior to the introduction of the first adalimumab PSP. Patients must not have previously used adalimumab and patients with multiple diagnoses were excluded.

Data sampling windows (including 12 months adalimumab sample selection window followed by up to 12 months follow up) were: April 2007 until March 2009 for AS; January 2007 until December 2008 for PsA and RA, and January 2009 until December 2010 for CD (including refractory CD only as fistulising CD was not PBS reimbursed until April 2011). AbbVie PSP program data showed that there were only 6 Crohn’s disease patients enrolled in the PSP during 2009, indicating that the 2009–2010 PBS 10% sample dataset represents a valid ‘non-PSP’ like dataset for this indication.

Data were extracted and analysed by Prospection Pty Ltd, a health analytics company approved to undertake analytics on data extracted from the Services Australia 10% PBS dataset under a licence agreement with Services Australia. This study and publication of subsequent results were approved by the Australian Government Department of Human Services External Request Evaluation Committee and complied with relevant data protection and privacy regulations.

Patient support program

The Australian HUMIRA® PSP is sponsored by AbbVie (the Australian company, AbbVie Pty Ltd) and is available to Australian patients prescribed HUMIRA® (adalimumab). It was first launched in 2009 and offerings of the program have changed over time. At the time of this study, the PSP included a range of general components such as a Welcome kit (sharps bin, patient booklet, cold pack, step by step injection guides, QR codes to injection videos, fridge magnet injection reminder); and access to a website where patients can learn about HUMIRA® (adalimumab), disease state and the PBS, access safety information, request a nurse injection training session, set up reminders (treatment, appointments and prescription collection), learn how to travel with HUMIRA® (adalimumab), order replacement sharps bin, order travel solutions (containers to maintain product temperature whilst travelling), request a health coach (dietician/nutritionist or exercise physiologist), information on mental health resources, sharps disposal information, patient organisations information and additional resources. Patients enrolled into the PSP decide which of the services available they want to use.

Variables–Prospective cohorts

In the two prospective cohorts, discontinuation was defined as the interruption of continual adalimumab therapy for at least 3 months between doses. If the patient was switched onto an alternate bDMARD, then discontinuation was confirmed without the need for further follow-up. Periods of less than 3 months between doses were defined as temporary dose delays, not meeting the criteria for discontinuation. The 3-month interruption interval defining discontinuation was chosen as it represented the minimum period when it was likely that a patient could be determined to have discontinued, taking into account that patients ordinarily have to pick up their scripts monthly from the pharmacy and their treating physician would normally review their need for scripts every 3 to 6 months. This period would also allow for potential delays in picking up scripts or temporary interruptions. If a potential adalimumab discontinuation occurred between 9 and 12 months after the first dose, patients were followed up by a telephone call from the investigative site to confirm whether the patient eventually met the definition of adalimumab discontinuation. bDMARD line of therapy was derived from the CRF (Case Report Form) page where patients were asked to select prior b/tsDMARDs (biologic/targeted synthetic DMARDs) received. If the patients did not select prior b/tsDMARD use, they were categorised as line 1, and if patients did indicate other b/tsDMARD(s) prior to adalimumab, they were categorised as line 2+. Additional source data included information on adalimumab dose escalations and co-prescribed medications entered directly by patients into the Patient Medication Diary via the ePRO database throughout follow-up. The main reason for discontinuation was determined by the patient’s treating clinician and entered into the CRF by study site staff. The clinician reported discontinuation options in the CRF included: due to adverse event, due to lack of therapeutic efficacy, due to loss of efficacy, due to process failure, due to patient/HCP/payer preference decisions unrelated to efficacy or adverse events, lost to follow up. Patient-reported reasons for discontinuation were collected, but were not used for sub-analyses, rather to assist investigators to seek confirmation of the main reason for any adalimumab discontinuations. Where the date of first injection was missing, or partial, for a patient, and there was post-baseline data to indicate that the patient did start adalimumab, the baseline assessment date was imputed as adalimumab start date (the latest date was imputed when assessments were completed over a number of days). Time to drug discontinuation for adalimumab was calculated based on the start and stop dates that were patient-reported in the Patient Medication Diary. The CRF also recorded whether patients had received training from specialised nurses for self-injection technique for adalimumab, based on patient self-report at the 3-month timepoint. PROMs were entered into an electronic Patient-Reported Outcomes (ePRO) database. Within the database there were no restricted windows for completion of the baseline questionnaires; if any baseline data were entered by a patient in the ePRO the day after first adalimumab dose or later, the patient remained in the study on the rationale that the patient would be expected to have less severe disease intensity measures at the delayed baseline than if baseline PROMs were measured prior to the first dose. Any impact on endpoints would therefore be expected to be conservative and potentially reduce the estimate of PSP impact. Data from questionnaires to be completed at weekly intervals were collected within 3 days of the expected completion date of the questionnaire. Data from questionnaires to be completed at monthly or 3 monthly intervals were collected within two weeks of the expected completion date of the questionnaire.

Variables–Retrospective non-PSP PBS 10% cohort

In the retrospective non-PSP PBS 10% cohort, discontinuation was defined as the date of the last prescription for adalimumab plus the length of time the filled prescription is known to last (28 days), where there is no subsequent prescription of adalimumab in the following 3-month period. Subjects without an observed discontinuation event were censored at the end of the study window. Gender was available in the dataset. Age was calculated based on date of birth and index date for each patient. Line of therapy was defined based on prior use of bDMARDs. If patients had no dispensing for bDMARDs prior to adalimumab, they were categorised as line 1 and if patients had a dispensing for another bDMARD prior to adalimumab, they were categorised as line 2+. Concomitant treatments with cDMARD (conventional DMARD) or corticosteroids were analysed 3, 6 and 12 month post index date.

Statistical methods–Sample size

The sample size calculation was based on the cross-indication analysis of the primary endpoint (12-month adalimumab time to discontinuation) with significance set at 0.05 and the hypothesis test 2-sided. The study aimed to recruit 420 patients (PSP + non-PSP cohorts) initiated on adalimumab for the first time after meeting PBS-reimbursement criteria, in an approximate 2:1 ratio between the PSP and non-PSP cohorts (280:140), estimated to provide a power of 81% with 165 discontinuations and 83% with 180 drug discontinuations, respectively, to detect a hazard ratio of 0.645. The actual sample size of the prospective cohorts was lower than planned (Full analysis Set [FAS]: N = 170 PSP cohort, N = 75 non-PSP cohort). The retrospective non-PSP cohort FAS was N = 297.

Statistical methods–Prospective comparisons

The general analytical approach for all endpoints was descriptive in nature. All descriptive and outcome data was summarised and analysed by PSP status and Total (PSP and prospective Non-PSP combined), with indication subgrouping when appropriate, based on sample size in each category. Pooled indications were defined as Rheumatology (RA, PsA and AS) and Gastroenterology (CD and UC). Two datasets were established. The FAS consisted of all patients who were recruited into the study, who had a valid adalimumab start date and/or at least one post-baseline follow-up for continuation of adalimumab. Where the date of first injection was missing, or partial, for a patient, and there was post-baseline data to indicate that the patient did start adalimumab, the latest baseline assessment date was imputed as the adalimumab start date. Within the FAS, grouping into PSP/non-PSP was based on the PSP enrolment intention recorded at baseline. The “Primary Analysis Set” consisted of all FAS patients who had valid values for all covariates included in the propensity score model used to weight the data within the endpoint analyses.

Survival analyses were performed to assess the primary endpoint, with an ‘event’ defined as a drug discontinuation. Time to adalimumab discontinuation was compared between the PSP and prospective non-PSP cohort by using an inverse probability of treatment weighting (IPTW) approach. A propensity model (logistic regression) was fitted for all patients included in the analysis using the PROC PSMATCH procedure to estimate the propensity score (PS) of participating in the PSP for each patient, incorporating the following baseline covariates: age, sex, disease duration (derived from year of clinical diagnosis, round up to the nearest year), disease indication, health-related quality of life at adalimumab initiation, concomitant treatment with non-biological DMARDs (Yes|No), highest education level (graduate or above|other). bDMARD line of therapy was not included in the final propensity model to optimize number of patients with no missing data that could be included in the model, given that the propensity score models including this variable were found to make negligible contribution to the overall model and subsequent analyses. Based on the estimated PS, stabilised IPTW, the inverse probability of participating in the PSP (PSP or non-PSP), was estimated for each patient by including the PSWEIGHT statement in the PSMATCH procedure. A weighted Cox regression, weighted by the stabilised IPTW, was used to compare the time to adalimumab discontinuation between PSP and non-PSP. The Cox model estimated the hazard-ratio (HR) comparing the hazard (risk) of drug discontinuation in the PSP cohort to the hazard (risk) in the non-PSP cohort. A sensitivity analysis was undertaken using propensity score trimming, where stabilised IPTWs below the 5th and above the 95th percentile were excluded. A second sensitivity analysis was undertaken whereby the weighting was omitted from the Cox regression, allowing the analysis to include all patients in the FAS. Changes from baseline for PROM scores were analysed over time with a mixed model for repeated measurements (MMRM). MMRM model was also weighted by IPTW. Time to adalimumab discontinuation was also compared between the prospectively enrolled patients who reported having received self-injection training and those who reported no self-Injection training, using an unweighted Cox Regression. The Cox Regression model was implemented to calculate the adjusted hazard-ratio (HR), to compare the hazard (risk) of drug discontinuation in the Self-Injection Training cohort to the hazard (risk) in the No Self-Injection Training cohort. All prospective cohort analyses were performed using SAS® Version 9.4 (SAS Institute, Cary, North Carolina, USA).

Statistical methods–Prospective PSP versus retrospective non-PSP analyses

A similar IPTW approach, as applied for the prospective analyses, was also used for the comparisons against the retrospective PBS 10% non-PSP cohort. The independent covariates used as predictor variables in the propensity score model for these analyses were (based on available covariates in the PBS 10% dataset): age, sex, concurrent treatment, bDMARD line of treatment. The FAS comprised the full analysis set for the PSP prospective cohort for RA, AS, PsA and CD, and all available data from the PBS 10% dataset within the specified sampling window and meeting the inclusion criteria. Comparisons between the PSP and non-PSP PBS 10% cohort were conducted using a “Primary Analysis Set”, including all FAS patients who had valid values for all covariates included in the propensity score model. Two sensitivity analyses were performed, the first one using a trimmed analysis set similarly to the prospective cohorts’ analyses, and one using a dataset which included an additional year of follow-up time for the non-PSP PBS 10% sample. This dataset allows the exploration of temporal patterns of drug discontinuation and drug persistence. This sensitivity analysis was deemed not to compromise the non-PSP definition of the cohort, as it was likely that there was less than 10% PSP ‘contamination’ based on PSP enrolment data during those years. All analyses involving the retrospective cohort were performed using Prospection’s PharmDash software or R, as appropriate.

Participants

Patient flow is summarized in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Patient flow.

Patient flow for prospective PSP versus prospective non-PSP analyses (1A) or prospective PSP versus retrospective non-PSP analyses (1B).

https://doi.org/10.1371/journal.pone.0300624.g001

A total of 280 patients were recruited across the two prospective cohorts (193 in the PSP cohort and 87 in the non-PSP cohort) over 17 months from May 2018 to September 2019 and were included in the FAS for the prospective analyses (Fig 1A). A total of 159 patients from the prospective PSP and 297 from the non-PSP PBS 10% sample were included in the FAS for the analyses against the retrospective cohort (Fig 1B). The Primary Analysis Set used for the final weighted and adjusted model included 162 PSP patients and 65 Non-PSP patients for the prospective analyses (1A), while it included 151 patients in the prospective PSP cohort and 297 in the retrospective PBS 10% non-PSP cohort for the analyses against the retrospective cohort (Fig 1B). Unless differently specified, only the outcomes from the Primary Analysis Set are described here.

Most of the patients in any cohort were prescribed adalimumab for a rheumatology condition (172/227 [76%] total rheumatology patients for prospective analysis; 336/448 [75%] total rheumatology patients for retrospective analysis). Within rheumatology indications, most of the patients had a diagnosis of RA (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patient disposition, demographics and baseline characteristics (Primary Analysis Set).

https://doi.org/10.1371/journal.pone.0300624.t001

Across all indications, the distribution of patient characteristics was similar in the PSP and Non-PSP groups, except for slightly numerically lower percentages in the non-PSP prospective group of patients born in Australia (PSP: 82%; non-PSP: 75%) and having obtained a university degree (PSP: 52%; non-PSP 42%) compared to the PSP group, and a lower numerical percentage in the non-PSP PBS 10% of first line adalimumab users (PSP: 91%; non-PSP PBS 10%: 74%) compared to the PSP group (Table 1). Between 57% and 68% of the patients were female in the cohorts (Table 1). Overall, the majority of patients were first line bDMARD users (91% of patients in the PSP prospective cohorts used in the prospective and retrospective analyses; 84% in the prospective non-PSP cohort; and 74% in the non-PSP PBS 10% cohort) (Table 1). Similar trends were observed in the Rheumatology pooled patients, while within the prospective RA subgroup, the proportion of patients treated with adalimumab as a first-line bDMARD therapy was notably lower in the Non-PSP group (77%) compared with the prospective cross-indication Non-PSP group (84%) and the prospective PSP group (92%).

Self-injection training status (confirmation of trained or untrained) was reported by 180 patients across all indications (129 [80%] PSP patients and 51 [78%] Non-PSP patients) (Table 1A). Across all indications, the proportion of patients who reported having received self-injection training was higher in the PSP group (91 [71%] patients) than the Non-PSP group (24 [47%] patients) (Table 1A).

Time to discontinuation

Kaplan-Meir curves for adalimumab time to discontinuation and associated log rank score p value are presented in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Adalimumab time to discontinuation Kaplan–Meier curves and 12-month persistence rates across all indications.

Primary Analysis Sets, analyses adjusted by the inverse probability of treatment weighting (IPTW). The log rank score P value for the persistence survival curve shown. 2A) Prospective PSP cohort versus prospective non-PSP cohort. 2B) Prospective PSP cohort versus retrospective PBS 10% non-PSP cohort.

https://doi.org/10.1371/journal.pone.0300624.g002

In the final weighted and adjusted model for the comparison between the two prospective PSP and non-PSP cohorts, participation in the PSP was not significantly associated with adalimumab discontinuation, as determined by log-rank test (Fig 2A) and Cox Regression (Fig 3). Weighted Kaplan-Meir estimates of adalimumab 12-month persistence rates were 0.79 (95% CI 0.72, 0.84) in the overall prospective population, 0.78 (95% CI: 0.69, 0.84) in the prospective PSP cohort and 0.82 (95% CI: 0.67, 0.90) in the prospective non-PSP cohort (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Hazard-Ratio (HR) for adalimumab discontinuation across all pooled indications.

A Cox regression, weighted by the inverse probability of treatment weighting (IPTW), was used to calculate the hazard-ratio (HR) comparing the hazard (risk) of drug discontinuation in the PSP cohort to the hazard (risk) in the non-PSP cohorts (prospective and retrospective) across all indications.

https://doi.org/10.1371/journal.pone.0300624.g003

S1 Fig presents time to discontinuation Kaplan Meier curves in the RA and Rheumatology prospective subgroups. Both subgroup analyses demonstrated no significant difference in the survival functions of adalimumab time to discontinuation between the PSP and Non-PSP prospective cohorts. The hazard ratio findings are also presented in S3 Fig. The sample size was deemed insufficient to run indication-specific analyses for AS, PsA, CD, UC and HS or pooled Gastroenterology indications. A sensitivity analysis was also undertaken using propensity score trimming whereby stabilised IPTWs below the 5th and above the 95th percentile were excluded, and the weighted Cox regression was repeated using the trimmed population. The results, presented in S1 Table, were similar to the non-trimmed analysis. A second sensitivity analysis was undertaken whereby the weighting was omitted from the Cox regression, allowing the analysis to include all patients in the FAS. The results of this sensitivity analysis, presented in S1 Table, were also consistent with the main analysis.

As shown in Table 2, overall the most common clinician-reported reason for discontinuation across all indications was “Lack of therapeutic efficacy” (28 [51%] patients), followed equally by “Loss of Efficacy” (11 [20%] patients) and “Adverse Event” (11 [20%] patients). Adverse events were reported being a reason for discontinuation in 27% of Non-PSP patients (n = 4) and 17.5% of PSP patients (n = 7). Discontinuations for reasons unrelated to efficacy or adverse events were only reported by 2 (3.5%) discontinuing patients overall. Reasons for discontinuation in the RA and Rheumatology subgroups were similar to the cross-indication group, with “Lack of therapeutic efficacy” most frequently reported (Rheumatology: 27 [54.0%] patients; RA: 12 [50.0%] patients) followed by “Loss of Efficacy” (Rheumatology: 11 [22.0%] patients; RA: 5 [20.8%] patients) and “Adverse Event” (Rheumatology: 9 (18.0%) patients; RA: 5 [20%] patients).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Primary reason for discontinuation (clinician reported) in the prospective cohorts (Primary Analysis Set).

https://doi.org/10.1371/journal.pone.0300624.t002

In the final weighted and adjusted model for the comparison between the prospective PSP cohort and retrospective PBS 10% non-PSP cohort, patients enrolled in the PSP were less likely to discontinue compared to the non-PSP PBS 10% sample, as determined by log-rank test (p = 0.0007 (Fig 2B) and cox regression (HR [95% CI] = 0.44 [0.28 to 0.68], p<0.001) (Fig 3). Similar results were observed in the IPTW trimmed sensitivity analysis (HR [95% CI] = 0.51 [0.32 to 0.81], p<0.004; S1 Table) and in the sensitivity analysis using the broader IPTW adjusted PBS sample (HR [95% CI] = 0.44 [0.28 to 0.69], p<0.001, S1 Table). The persistence rate at 12 months in the IPTW adjusted model was 81% (95% CI 76%, 90%) in the PSP group and 61% (95% CI 56%, 67%) in the non-PSP PBS 10% sample (Fig 2B). As shown by the Cox regression analysis of the time to discontinuation (Fig 4), male patients were less likely to discontinue compared to female patients (HR 0.66, 95% CI 0.47 to 0.94, p = 0.0019), while patients over 65 years of age were also more likely to discontinue treatment compared to younger patients aged 18 to 34 years (HR 1.90, 95% CI 1.07, 3.35, p = 0.0028).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Hazard-Ratio (HR) for adalimumab discontinuation across all pooled indications using the Primary Analysis Sets from the prospective PSP and retrospective PBS 10% non-PSP cohorts.

A Cox regression, weighted by the inverse probability of treatment weighting (IPTW), was used to calculate, across all pooled indications, the hazard-ratio (HR) comparing the hazard (risk) of adalimumab discontinuation in (i) the PSP cohort versus the PBS 10% non-PSP cohort; (ii) Patients in different age brackets versus aged 18–34; (iii) Male versus female patients; (iv) Later lines of bDMARD treatment versus first line bDMARD users; (v) patients on concurrent treatment versus no concurrent treatment.

https://doi.org/10.1371/journal.pone.0300624.g004

S2 Fig presents time to discontinuation Kaplan Meier curves in the RA and Rheumatology subgroups. Both subgroup analyses demonstrated significant difference in the survival functions of adalimumab time to discontinuation between the PSP and Non-PSP groups (Rheumatology Log-rank p = 0.001; RA Log-rank p = 0.008). The hazard ratio findings are also presented in S3 Fig (RA HR [95% CI] = 0.43 [0.23, 0.81]; Rheumatology HR [95% CI] = 0.53 [0.34, 0.84]. The sample size was deemed insufficient to run indication-specific analyses for AS, PsA, or CD.

The median time until first discontinuation could not be derived for the PSP or Non-PSP cohorts as the proportion of patients who reported discontinuation was less than 50%.

Other analyses

The secondary objective of the study was to assess the potential impact of the PSP on PROMs, using the secondary endpoints of changes in scores over time for EQ-5D-5L, DLQI, BMQ:General, WPAI:General Health, PGDA, and BASDAI. This analysis was only possible in the prospectively enrolled cohorts as the PBS 10% database does not include such measures. The MMRM analyses across all indications combined found no effect of enrolment in the PSP on changes over time in any PROMs (S2 Table). MMRM analyses for DLQI scores could not be performed due to insufficient patient numbers. Independently of PSP status, all PROMs except BMQ Overuse, BMQ Harm (RA subgroup only), and WPAI Percent work time missed due to health, demonstrated statistically significant improvements in scores over time after adalimumab initiation, across all indications combined, as well as in the Rheumatology and RA subgroups (S3 Table).

Time to adalimumab discontinuation was also compared between the overall prospectively enrolled patients (FAS) who reported having received self-injection training and those who reported no-self-injection training using an unweighted Cox Regression. Across all indications, adalimumab discontinuation dates were reported by 23 (20%) patients who received self-injection training, compared with 11 (17%) patients who did not receive self-injection training. Self-injection training status was not significantly associated with adalimumab discontinuation, as determined by both log-rank test (p-value = 0.5577) and Cox regression (HR (95% CI): 1.239 (0.604, 2.542), p-value: 0.5587) (S4 Table).