Data on the quality and methods of studies reporting healthcare costs of post-prostate biopsy sepsis

This data article presents the supplementary material for the review paper “Healthcare Costs of Post-Prostate Biopsy Sepsis” (Gross et al., 2019). A general overview is provided of 18 papers, including the details about year and journal of publication, country of dataset, data population characteristics, cost basis, and potential for bias evaluation. Quality assessment and the risk of bias of the 18 papers are detailed and summarized.


Data
The data in this article consists of additional and expanded tables and figures provided in a systematic review [1] of the literature including 18 research papers [2e19], as well as the methodological and bias evaluation of these papers.
The data includes additional variables and characteristics included as a supplement to the systematic review [1]. Methodological and bias evaluation was conducted by two separate evaluators and are thoroughly described here in more detail. Table 1 is an overview of the included datasets, comprised of author year, journal of publication, cases of sepsis and overall biopsies included in the dataset, the source of the data, and median age of men at biopsy. Table 2 is a breakdown of the datasets including country where data was gathered, publication year, and whether the dataset was a single institution, multi-institution, state-wide, or national dataset. Table 3 is a detailed description of how cost within their specific cohort was determined for each dataset. Table 4 is the assessment of bias using the Newcastle-Ottawa Quality Assessment Scale. Fig. 1 is the number of datasets at low, medium, and high risk of bias.

Experimental design, materials, and methods
The review was performed following the instructions set forth by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [20]. All 874 articles were reviewed at the abstract level. Articles that were reviewed in full included abstracts that commented on post-prostate biopsy sepsis cost. Articles that were reviewed in full were systematically evaluated by two reviewers (MG and MA) to assess for eligibility of inclusion/exclusion criteria. Articles were included if they had individual or system-wide cost or burden of post-prostate biopsy hospital admission. Extraction of data then took place. A modified STROBE criterion was used to evaluate dataset quality metrics [21].
Specifications Table   Subject  Medicine  Specific subject area  Urology  Type of data  Tables, figure  How data were acquired  Review and analysis of the relevant literature searched through Ovid MEDLINE,  Value of the data The data serves as a way to provide greater insight into the various datasets examining the cost of post-prostate biopsy sepsis.
The data assists readers in understanding the review article [1] about the costs of post-prostate biopsy sepsis. The data, along with the accompanying research article [1], provides an example of how to assess the quality and risk of bias of the included papers that can be used in other cost reviews.
The data provides greater detail in how sepsis and cost were derived in each dataset and how the risk of bias of each dataset was evaluated.
Healthcare costs of hospitalization for infection following prostate biopsy was defined as the primary outcome. Standard aspects of reviewing and extracting data, as described by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [20], were adopted for this review. The extracted aspects included the following:   General information, dataset characteristics, selected data populations, source of data, exposure with primary outcome of sepsis, and data extracted from the 18 selected papers are outlined in Table 1.  In the table we describe authors, year of publication, country of dataset, journal of publication, overall population examined, number of sepsis cases reported, specifics of data location and design, and selected demographics when provided in the respective papers. Table 2 includes counts of the countries of dataset, years of publication, and methodology performed by each paper. Healthcare costs were analyzed for 10 datasets and directly and indirectly evaluated in the context of expenditures for an episode of sepsis related to prostate biopsy. Each dataset had its own method of determining costs, which is described in Table 3, along with author, year of publication, average cost of sepsis, and the inflation-adjusted cost [22]. All costs were adjusted for inflation to the May 2018 urban and inpatient hospital service consumer price indices [22]. This method is modeled after the same approach used by the Agency for Healthcare Research and Quality [23]. If a paper did not specify what dollar year their costs were originally derived from, the year of publication was used. For the international cohorts, all currency amounts were compared to the U.S. dollar using the historical exchange rate from the federal reserve before adjusting for inflation as described above [24].
We then assessed the quality of each paper according to the Newcastle-Ottawa Quality Assessment Scale for cohort analysis [25]. This was considered the most appropriate evaluation of bias as it: 1. Is widely used and accepted for quality assessment 2. Demonstrates both inter-and intra-rater reliability 3. Demonstrates criterion and construct validity 4. Demonstrates objectivity, as questions are well defined and easy to understand 5. Is described as a tool that evaluates papers that are included in this type of analysis Five questions were used from the Newcastle-Ottawa Quality Assessment Scale, given the observational nature of the included papers and lack of control groups. Questions were evaluated by two independent reviewers, with conflicts being resolved with further analysis of the papers. Questions are answered with a yes (1) when information was directly available in the text of the papers or no (0) when information in the text was either directly contradictory to the question, not sufficient or specific enough to answer properly, or not available to analyze. Detailed analysis of the risk of bias assessment, along with more detail of answers to individual questions, are reported in Table 4 [1].
For the selection category, representativeness of the exposed cohort was used to evaluate bias. A yes (1) in this category means that the exposed cohort either (a) truly or (b) somewhat represented the average man of average age undergoing prostate biopsy with no increased risk of sepsis or hospital admission due to other co-morbidities in the community. For our analysis, we defined truly or somewhat representative of the average man if patients were all chosen within a specified timeframe in multiple institutions. A no (0) means that the selected group of users were from a specific cohort (i.e. nurses, volunteers, all from a single institution) or there was no description of the derivation of the cohort.
For the exposure category, ascertainment of exposure was used to evaluate bias. A yes (1) in this category means that the exposed group was found by either (a) secure records (e.g. surgical records, national database, billing codes) or (b) structured interview (e.g. medical records). A no (0) in this category means that the exposed was ascertained by either (a) a written self-report from the patient or (b) there was no description of how patients who had a prostate biopsy was chosen.
For the outcome category, multiple questions were used to assess bias. The first question includes the assessment of outcome. A yes (1) for this question means that the outcome was assessed by either (a) independent blind assessment (i.e. an assessment from a physician) or (b) record linkage (i.e. medical records, billing codes, databases). A no (0) for this question means that the outcome in questions was either self-reported or had no description of how the outcomes were chosen. The second question is whether the exposure and outcome had the same assessment or if they differed. A yes (1) means that the exposure and outcome had the same means of assessment, where a no (0) means that the way that each was selected differed. The last question in this category is whether there was adequate follow-up of patients. A yes (1) means that follow-up was complete for all subjects with an adequate amount of time given for follow-up, here defined as at least 1-month post-biopsy. A no (0) constitutes that follow-up was not complete for all subjects, patients were lost to follow-up, patients were evaluated for outcomes before 1-month and not evaluated again, or there was no comment on how follow-up was defined.
After completing the risk of bias of each dataset, we found that among 18 datasets included, 10 were of low risk, 6 were of medium risk, and 2 were of high risk (Fig. 1).