Studies Identified for Inclusion
The PRISMA flow diagram for the search results is shown in Fig. 1. The search strategy identified 5,021 publications in total. Removal of duplicates resulted in 4714 publications to be screened. Review of titles and abstracts (‘TiAb’ screening) by two independent reviewers resulted in the exclusion of 4127 publications, with 587 publications remaining for the full-text review to assess for inclusion based on the inclusion and exclusion criteria. Four-hundred fifty-eight publications were excluded because they failed to meet one or more of the inclusion criteria, met one or more of the exclusion criteria, or were identified as a duplicate study during the full-text review. Moreover, the full text could not be retrieved for seven studies. One additional publication was identified through a search of the references cited in the included studies. Ultimately, 123 publications were included in the final analysis.
Details of Studies Included
Table 2 provides an overview of the characteristics of the studies included in this analysis. Studies related to PSO comprised the majority of the studies. One hundred-four studies (84.6%) covered PSO, 14 studies (11.4%) covered AD, and 5 studies (4.1%) covered CU.
Type of Intervention
Seventy-two studies (58.5%) were for interventions that included biologics, whereas 41 studies (33.3%) did not report whether biologics were included or not. Among the 104 studies that covered PSO, 65 studies (62.5%) reported that they included biologics. Moreover, among the 5 CU studies included, 4 studies (80.0%) reported that they included biologics. However, only 3 out of 14 studies (21.4%) that covered AD reported that they included biologics.
Eighty-nine studies (72.4%) were published more recently, from 2018 to 2020. In particular, studies related to AD were published more recently (from 2018 to 2020), including 13 out of 14 of the studies included (92.9%). A similar trend was observed for both studies that included biologic interventions and studies that did not, with 73.6% and 70.0% of studies being published from 2018 to 2020, respectively. Only 2 out of 5 studies (40.0%) relating to CU, however, were published from 2018 to 2020.
Type of Study Design
Sixty-five studies (52.8%) were retrospective analyses, 40 studies (32.5%) were model-based studies not conducted alongside a clinical trial, 10 studies (8.1%) were cross-sectional analyses, and 8 studies (6.5%) were conducted alongside a clinical trial. Cross-sectional analyses were more common among studies covering AD (35.7% vs. 4.8% and 0.0% for PSO and CU, respectively). Moreover, model-based studies not conducted alongside a clinical trial were somewhat more common when biologics were included as an intervention (50.0% vs. 40.0% and 0.0% for studies with nonbiologics and those in which the intervention was not reported/unknown, respectively).
Type of Economic Evaluation
About half of the studies included were full economic evaluations, and about half were other cost studies. Specifically, 67 studies (54.5%) were cost studies, 25 studies (20.3%) were CUAs, 22 studies (17.9%) were CEAs, 6 studies (4.9%) were budget impact analyses, 1 study (0.8%) was a cost–benefit analysis, and 1 study (0.8%) was a cost-consequence analysis. One study included both a CEA and CUA. For PSO and CU, 42.3% and 40.0%, respectively, of the studies included were CEA and/or CUAs. However, for studies covering AD, the majority were cost studies, and only 2 out of 14 studies (14.3%) were CUAs. Among the studies reviewed for which biologics were included as an intervention, 40 out of 72 studies (55.6%) were a CEA or CUA, compared to 8 out of 10 studies (80.0%) for those that did not include biologics.
Forty-six studies (37.4%) were conducted based on the payer/third-party/healthcare system perspective, 13 studies (10.6%) were conducted based on the societal perspective, 2 studies (1.6%) were conducted based on both the societal and payer perspective, and 7 studies (5.7%) were based on another perspective. However, 55 of the studies (44.7%) identified did not report the study perspective. Studies of PSO were somewhat more commonly conducted from the payer/third-party/healthcare system perspective, whereas studies of AD and CU were more commonly conducted from the societal perspective compared to PSO studies.
Region/Country of Evaluation
Sixty-two studies (50.4%) were conducted for North America or South America, with 57 studies (46.3%) conducted for the US. Europe was the next most common region, with 45 studies (36.6%) conducted for a European country, including 8 studies (6.5%) conducted for the United Kingdom (UK), 8 studies (6.5%) for Italy, and 7 studies (5.7%) for Germany. Among studies conducted in the Middle East/Asia/Oceania regions, Japan was the most common country for conducting studies, with 7 studies (5.7%) conducted for Japan. North/South America and Europe were the most commonly included regions/countries for PSO and AD, with over 75% of the studies conducted for those regions. However, only 2 out of 5 studies (40.0%) of CU were conducted for those regions. Studies conducted in the US more commonly reported that they included a biologic treatment.
Study Funding Source
One hundred three studies (83.7%) were sponsored by a company in the healthcare industry, most commonly a pharmaceutical company. Five studies (4.1%) received funding from a non-industry sponsor such as a government agency, 4 studies (3.3%) received no funding, and 11 studies (8.9%) did not report whether they received any funding or not. Sources of funding were relatively similar across conditions and types of intervention.
Details of Cost Information Included
Cost Information Included
Table 3 provides an overview of the cost information included in the economic evaluations and cost studies of this analysis. Seventy-four studies (60.2%) examined costs only, 41 studies (33.3%) examined costs and resource use, 7 studies (5.7%) examined costs, resource use, and the budget impact, and 1 study (0.8%) examined the budget impact only. While 67 out of 104 studies (64.4%) of PSO and 3 out of 5 studies of CU (60.0%) considered costs only, the majority of studies conducted for AD (64.3%) examined both costs and resource use. Among studies that included biologics, 58 out of 72 studies (80.6%) only examined costs. Similarly, among the studies that included nonbiologic interventions, 7 out of 10 studies (70.0%) examined costs only. However, only 9 out of 41 studies (22.0%) for which the intervention (biologics or nonbiologics) was not reported or was unknown included only costs.
Costs Elements Reported
Nearly all studies (97.6%) included direct costs, 22 studies (17.9%) included indirect costs, 19 studies (15.4%) included total costs, 9 studies (7.3%) included adverse event costs, and 2 studies (1.6%) included health state costs. The inclusion of indirect costs was somewhat more common among studies of AD and CU (28.6% vs. 17.9% overall and 60.0% vs. 17.9% overall, respectively). No major differences were observed between types of intervention in the cost elements reported.
Sources of Costs
A claims database was the source of costs for 46 studies (37.4%), 39 studies (31.7%) sourced costs from formulary/government listings, 26 studies (21.1%) sourced costs from an online database, and 8 studies (6.5%) sourced costs from a survey. A claims database or a survey were the most common sources used for studies related to AD (10 out of 14 studies). Formulary/government listings and an online database were more commonly used as sources of cost information for studies related to PSO compared to studies of AD and CU. Among the studies that included biologics, those that sourced costs from a formulary/government listings were somewhat more common (40.3% vs. 31.7% overall). Among studies that did not include biologics, those that sourced costs from formulary/government listings were also somewhat more common (50.0% vs. 31.7% overall).
Sixty-five studies (52.8%) were based on USDs, despite the fact that 57 studies (46.3%) were conducted for the US. Thirty-seven studies (30.1%) were based on euros (EUR), and 8 studies (6.5%) were conducted based on British pounds, which is consistent with the fact that 45 studies (36.6%) were conducted for a European country. No major differences in the currencies used were observed between conditions or types of intervention, although studies of CU more commonly used a currency other than USD or EUR.
Cost Year Considered
Forty-three studies (35.0%) included costs for a year during 2016–2018 and 32 studies (26.0%) included costs for a year during 2013–2015. Over one-fourth of the studies (27.6%), however, did not report the cost year for the analysis. Nearly all of the studies of CU included costs for a year during 2013–2015, but this may be due to that fact that most of the CU studies included were published between 2016 and 2018, whereas studies of PSO and AD were more commonly published between 2018 and 2020.
Key Cost Drivers and Their Data Sources
Table 4 provides a summary of the key cost drivers—i.e., aspects that trigger higher costs—and the sources of the data. Sixty-seven studies (54.5%) did not report the key cost drivers clearly. Thirty-two studies (26.0%) reported that drug costs were a key cost driver. Outpatient costs were the second most common key cost driver, with 11 out of 123 studies (8.9%) having outpatient costs as a key cost driver. Pharmacy costs were the next most common key cost driver, with 9 out of 123 studies (7.3%) having pharmacy costs as a key cost driver. Drug costs, outpatient costs, and/or pharmacy costs were also the key cost drivers for studies whereby the source of the cost information was claims data, formulary/government listing(s), or online database(s). However, for studies where the source of the cost information was survey data, productivity loss was a key cost driver for 3 out of 11 studies (27.3%), and drug costs were a key cost driver for only 2 out of 11 studies (18.2%). For the 4 studies whereby the source of the cost information was medical records, the key cost driver was not reported.
Quality of Evidence
The quality of the 123 studies included was evaluated using the Drummond checklist. A summary of the evaluation of the studies is shown in Table 5. Moreover, detailed results by condition and by publication are provided in Tables S3 and S4 of the Supplementary Material. Reporting was generally good, with only 8 out of 36 of the checklist items being evaluated as “No” for 50% or more of the studies. In terms of the data collection process, 76.4% of the studies included did not report costs based on productivity changes (item 14), and 74.0% and 58.5% did not report the quantity of resources separately from their unit costs, respectively (item 16 and item 17). Details concerning price adjustments for inflation or currency conversion (item 19) were reported by only 40.7% of the studies included. For the analysis, 74.0% of the studies included did not clearly state the discount rate used (item 23), and 50.4% did not describe the approach used for the sensitivity analysis if applicable (item 27). Moreover, 52.8% of the studies included did not report an incremental analysis (item 31). However, all of the studies stated the research question, answered the study question, and based their conclusions on the data (items 1, 33, and 34, respectively, in Table 5). The quality of the evaluations and studies included for PSO, AD, and CU was similar (see Table S3 in the Supplementary Material).