Bowel Preparation for Colonoscopy: A Comparative Cost-Effective Analysis of Traditional Per Os Purgatory Prep versus a Novel Method Using High-Volume Colonic Water Irrigation

Data Collection

Computer assisted recursive literature searches of MEDLINE, EMBASE, and Cochrane Databases (October 2001 to October 2016) were performed using predefined search criteria including the terms “colonoscopy” and “bowel preparation”. Abstracts from major gastroenterology meetings were also searched for all relevant abstracts published from 2001 to 2016. Manual searches of the bibliography of selected publications were also performed. For estimates with respect to high- volume rectal irrigation, company data from HyGIeaCare, Inc. were used.

Reference Case Definition

The reference case will be an average 50-year old person undergoing an average-risk screening colonoscopy in the United States.

Bowel Preparation Strategies

Patients assigned to a standard bowel preparation strategy underwent bowel preparation using a 4 L, 2-day split dose of polyethylene glycol with electrolyte solution (PEG-ELS), as recommended by expert consensus guidelines [5]. In the baseline analysis, only a standard PEG-ELS preparation (GoLYTELY, Braintree Laboratories, Braintree, MA) was used, since it is the among most commonly used PEG-ELS for colonoscopy preparation. Sub-group analysis did include other FDA-approved commercial preparation regiments. These included sulfate-free PEG-ELS (NuLYTELY; Trilyte, Braintree Laboratories, Braintree, MA), low-volume PEG-ELS (Suprep or Suclear, Braintree Laboratories, Braintree, MA), and preparations that offer a combination of sodium picosulfate and magnesium citrate (Prepopik, Ferring Pharmaceutical, Parsippany, NJ). Over-the-counter (OTC) preparation regimens, e.g. low-volume PEG (Miralax, Merck, Boston, MA) and magnesium citrate, will not be evaluated in this model as they are not FDA approved for colonoscopy preparation. Additionally, the FDA has issued a black box warning for use of sodium phosphate (in tablet form) given the risk of kidney injury and electrolyte abnormalities, and thus will also not be included in this model. Patients undergoing standard PEG-ELS split-dose preparation will ingest about half of the preparation (approximately 2L) the evening before their scheduled procedure and will wake up at 2 or 3 AM the day of the procedure to ingest the remainder to complete the preparation. Experts recommend that the second dose of the split preparation should ideally begin 4-6 hours before the time of colonoscopy with completion of the second half at least 2 hours before start time [5]. Starting the second half of the split-dosed at 2 or 3 AM the will allow ample time for completion of their preparation and travel to the endoscopy center in order to have their procedure within the optimal window. The alternative to conventional bowel preparation is use of a high-volume rectal irrigation device. These devices are situated at a preparation center typically in close proximity to their endoscopy center. The preparation center is staffed by a trained professional and supervised by a registered nurse. Patients will have the ability to call the center prior to the preparation for education and counselling. The high-volume rectal irrigation system uses a slow stream of water drained by gravity, which is then infused via a soft nozzle into the rectum inducing natural colonic peristalsis. This allows the patient to effectively excrete feculent material while sitting on a sanitized basin, allowing for effective colon prep. The patient will be allowed to ingest clear liquids for up to 2 hours before their schedule procedure time, as recommended by the American Society of Anesthesiology [13].

Decision Analysis Model

In accordance with guidelines of the Panel on the Cost- Effectiveness in Health and Medicine, a model using a decision analysis software (TreeAge Pro 2009; TreeAge Software, Williamstown, MA) was constructed to evaluate two competing bowel preparation regimens (split dose PEG-ELS preparation vs. high-volume rectal water irrigation) using the hypothetical case as described above. A hybrid model of a linear decision tree terminating in a Markov model was also developed (Figure 1).

Click to enlarge

Model Structure

In the Markov model, the natural history of patients undergoing a screening colonoscopy will be modeled using various heath and disease states and compliance to CRC screening, each associated with a different set of costs and utilities (i.e. healthy without history of adenomatous polyps and compliant with CRC screening, noncompliant with CRC screenings, patients with low-risk adenomas, patients high-risk adenomas, patients who experience colonoscopy-related adverse events, patients with symptomatic and asymptomatic CRC (local, regional, and distant disease), and finally, death). At the end of each cycle in the model, which was set at 1 year, each patient was redistributed to different states depending on the estimated transitional probabilities among different health states. Mortality rates were derived from age- specific mortality rates from the United States (US) as well as state specific mortality rates. Each state was allowed to transition to the next state or stay in the same state, however certain transitions were not permitted and death was considered an absorbing state [14, 15]. Upon entry into the model, each patient underwent a screening colonoscopy. A certain portion of patients had an inadequate bowel preparation, characterized as having a Boston Bowel Preparation Score (BBPS) score <5. In such cases, these patients underwent a repeat colonoscopy within 1 year. Those patients who had an adequate preparation, their repeat screening or surveillance colonoscopy was determined in adherence with consensus guidelines [8, 16]. All index procedures and subsequent colonoscopies were performed in an outpatient ambulatory surgery center setting with the assumption that all procedures in both strategies utilized anesthesia support and patients with adverse events required inpatient hospitalization with close observation or surgical intervention.

Clinical Probabilities and Utilities

A clinical probability, including transitional probabilities, between different health states and performance characteristics of patients in each cohort was derived from public information or from expert consensus opinion. Quality adjusted life years (QALY) was estimated by adjusting the life expectancy of each health state by a weight or utility, which reflects patient preferences for that health state. Utility values were obtained from published information [17]. The loss of utility related to surgical resection as well as secondary to procedure- related adverse events (observation and surgical intervention) were obtained through the 2014 Healthcare Cost and Utilization Project (HCUP). Multiple published cost-effectiveness analyses on CRC screening were utilized in order to determine loss of utility associated with colonoscopy preparation (e.g. loss of sleep time, diagnosis and management of CRC, CRC-related mortality, etc.) [18].

Cost Estimates

Costs, as opposed to charges, were considered in this analysis taking into account a third-party payer’s perspective. Both direct and indirect costs were considered and all costs were adjusted to 2016 United States Dollars (USD) [19, 20, 21, 22]. Direct costs were estimated based on the national average reimbursement allowed for each coded procedure by the Centers of Medicare and Medicaid Services during the fiscal year of 2015. Inpatient medical, surgical, and diagnostic services were assigned a current procedural terminology (CPT) or diagnosis- related group (DRG) code to identify resource utilization. This data included costs related to diagnostic and therapeutic colonoscopy, primary resection (endoscopic vs. surgical), and the management of adverse events (conservative vs. surgical) [23, 24]. Outpatient data was based on ambulatory payment classification and CPT codes. Cost of different bowel preparations were obtained from the American Society of Gastrointestinal Endoscope guidelines and the Red Book by Micromedex (Truven Health Analytics, Greenwood Village, CO) [13]. Cost estimates for high-volume rectal irrigation preparation were estimated from current out of pocket costs to patients utilizing this procedure in approved centers across the United States (US). Indirect costs (e.g. cost related to time spent preparing for colonoscopy) were estimated using both the human capital method and available willingness to pay (WTP), which included the average cost of travel to the rectal irrigation center and the endoscopy center [25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35]. Lastly, the cost of time spent by an accompanying adult (presumably a family member) during bowel preparation and for transporting to the patient to and from their procedure was estimated based on published information. Table 1 summarizes the cost variables (both estimated direct and indirect costs).

surgical center (ASC) Cost of preparation Cost of standard preparation with 4L PEG-ELS $25 $25 (20-135) 36-39 Cost of novel high volume rectal Cost of preparation for colonoscopy at a irrigation preparation HyGIeaCare facility Total cost related to a procedure related Cost of complication complication, in this case a perforation, (perforation) including a hospitalization if needed Total cost related to a procedure related Cost of complication (bleeding) complication, in this case a bleeding, including a hospitalization if needed $250 (150-1000) $250 (150-1000) 36-39 $22,650 16,778 36-39 $8,321 $7,579 36-39 Cost of surgery Cost related to surgical resection of an advanced Costs related to colon cancer Stage specific Cost of initial diagnosis and management of CRC Local stage $32,912 $30,297 36, 39 Cost of initial diagnosis and management of CRC Regional stage $55,378 $50,970 36, 39 Cost of initial diagnosis and management of CRC Distant stage $72,314 $66,569 36, 39 Cost of treatment per year for maintenance therapy Local stage $3,535 $2,500 36, 39 Cost of treatment per year for maintenance therapy Regional stage $4,710 $3,212 36, 39 Mortality cost of patients with advanced colon cancer Terminal stages $112,316 $76, 588 36, 39 Indirect costs Indirect costs = Indirect medical cost + Productivity losses for the patient and accompanying High Volume Rectal Irrigation Standard Preparation Loss of Productivity Cost of loss of productivity related to colonoscopy for the patient.

Calculated by total lost time * hourly US average wage ($25.70) 31, 40

Total time spent by patient related to colonoscopy including preparation, and Total Time Spent (Patient) recovery Total time spent by accompanying person Total Time Spent (Accompanying person) related to colonoscopy including preparation, and recovery

50 hours 25 hours 25, 31-35

10 6 hours 25, 31-35

Outcomes Compared and Statistical Methods

The primary outcomes comparing the two strategies were incremental cost-effectiveness ratio (ICER) and net health benefit (NHB). ICER were calculated as the difference in costs divided by the difference in outcomes (life years) between the two competing strategies (ICER = [Cost Strategy I – Cost Strategy II]/[Effectiveness Strategy I – Effectiveness Strategy II]). The ICER is a measure of the added cost for each additional life years gained by Strategy II. The NHB of an alternative option was calculated using the formula: NHB = E – C / WTP, where E represents effectiveness, C represents cost, and WTP is the willingness to pay (i.e. the decision maker’s threshold ICER). Incremental NHB (INHB) will be calculated as the difference between the NHB of each strategy. NHB is often preferred to ICER as a measure of cost-effectiveness because of its direct interpretation as the average health gained per patient who under goes the alternative treatment adjusted for cost and WTP. Also, unlike ICER, the NHB is a monotonic function of both health and cost. Health policymakers should favor a strategy for which the NHB takes the greatest positive value in relation to values of WTP that seem reasonable with respect to known public policy [26]. For analysis of the results of the Monte Carlo analysis, relative risk with 95% confidence intervals and number needed to treat (NNT) were calculated.

Sensitivity Analysis

The robustness of the model was tested by performing a sensitivity analysis using clinical probabilities and cost estimates. Univariate and multivariate sensitivity analysis was performed using clinical variables, cost, and quality of life (QOL) estimates. In a hypothetical cohort of 100,000 patients undergoing a screening colonoscopy, a second-order Monte Carlo simulation was performed to calculate probabilistic sensitivity analysis. Monte Carlo

simulation recalculates a model multiple and incorporates uncertainties into an analysis in keeping with real-life situations [27]. In this method, sampling probability values from probability distributions of variables (rather than from a single range defined by upper and lower bounds) places greater weight on likely combination of parameter values and simulation results quantify the total impact of uncertainty on the model in terms of the confidence that can be placed in the analysis results. Threshold analyses were performed to understand the impact of the variability of key clinical and cost estimates and also the various threshold points, if any, where one strategy dominates the competing strategy.

Assumptions

The following assumptions were made in this model primarily because of limited or lack of published data: 1. The decision models were built on the assumption that CRC screening would negate cancer-related loss in life years and that patients would fully benefit from the restored life expectancy of an average population without CRC. The models did not consider the presence of other competing medical risks in some of the subjects undergoing screening.

2. The baseline model considered only average-risk patients undergoing outpatient screening colonoscopy. 3. The model assumes that once a patient underwent a colonoscopy with a particular regimen that the same regimen will be used for subsequent colonoscopies. 4. The model only took into account colonoscopy for CRC screening. No other methods of CRC screening were evaluated in this model. 5. The model assumes that one adult family member (typically a spouse) was involved in assisting the patient with preparation for colonoscopy and for providing transportation to and from the rectal irrigation center (if needed) and then to the endoscopy center. 6. Recovery time after the procedure, related cost, and disutility after an uncomplicated procedure was assumed to be similar under both strategies.