Table 1.
Select model parameters: base case value and plausible range
| Variables | Base case | Range for Sensitivity Analysis | Probabilistic Sensitivity Analysis* | Reference | |
|---|---|---|---|---|---|
| Distribution | Range (95%CI) | ||||
| Clinical | |||||
| Screening, % | |||||
| Serum pepsinogen test to detect atrophy† | [18, 46] | ||||
| Sensitivity | 71 | 58–82 | Beta | 59–82 | |
| Specificity | 98 | 97–99 | Beta | 97–99 | |
| H. pylori test | [47, 48] | ||||
| Sensitivity | 85 | 68–99 | Normal | 70–98 | |
| Specificity | 79 | 63–95 | Normal | 63–94 | |
| Endoscopy to detect dysplasia and asymptomatic cancer | [49, 50] | ||||
| Sensitivity | 81 | 78–84 | Beta | 78–84 | |
| Specificity | 100 | 95–100 | Beta | 99–100 | |
| Endoscopic screening complications requiring surgery, proportion | [51] | ||||
| Bleeding | 0.0011 | 0.000–0.003 | Beta | 0.000–0.002 | |
| Perforation | 0.0009 | 0.000–0.002 | Beta | 0.000–0.002 | |
| Endoscopic mucosal resection treatment | |||||
| Proportion eligible | |||||
| Dysplasia | 1.0 | 0.9–1.0 | ‡ | -- | |
| Preclinical local cancer | 0.65 | 0.55–0.70 | ‡ | -- | [24] |
| Proportion with complete resections | |||||
| Dysplasia | 0.87 | 0.81–0.91 | Beta | 0.82–0.91 | [52] |
| Preclinical local cancer | 0.94 | 0.91–0.97 | Beta | 0.92–0.97 | [53] |
| Proportion of incomplete resections requiring surgery | 0.36 | 0.31–0.42 | Beta | 0.31–0.41 | [54, 55] |
| Complications | |||||
| Bleeding | 0.014 | 0.005–0.033 | Beta | 0.003–0.027 | [53] |
| Perforation | 0.003 | 0.000–0.016 | Beta | 0.000–0.009 | [53] |
| Surgical mortality risk§ | 0.005 | 0.0002–0.015 | Beta | 0.004–0.006 | [56] |
| Outcomes | |||||
| H. pylori treatment effectiveness | [20] | ||||
| Progression of gastritis to atrophy, relative risk | 0.2 | 0.0–0.8 | ‡ | --- | [65] |
| EMR treatment effectiveness, proportion | |||||
| Dysplasia | [52] | ||||
| Remnant lesion | 0.026 | 0.008–0.059 | Beta | 0.005–0.048 | |
| Preclinical local cancer | [53, 57] | ||||
| EMR | |||||
| Local recurrence | 0.012 | 0.001–0.041 | Beta | 0.001–0.028 | |
| Metachronous cancer | 0.058 | 0.028–0.104 | Beta | 0.024–0.092 | |
| Surgery | |||||
| Local recurrence | 0.011 | 0.003–0.027 | Beta | 0.002–0.021 | |
| Metachronous cancer | 0.011 | 0.003–0.027 | Beta | 0.002–0.021 | |
| Cancer stage-specific annual mortality rate|| | 0.21–0.98 | --- | --- | [24] | |
| Quality of life | |||||
| Age-related quality weight, utility | 0.782–0.928 | --- | --- | [58] | |
| Utility reduction, duration | |||||
| Endoscopy or EMR | − 1 day | --- | --- | ||
| Gastrectomy | − 2 weeks | --- | --- | ||
| Symptomatic cancer quality weight, utility | 0.49 | 0.17–0.79 | Normal | 0.17–0.78 | [59] |
| Costs | |||||
| Direct, 2012$ | [60, 61] | ||||
| Physician visit (CPT 99213) | 70 | --- | --- | ||
| Serum pepsinogen I, II test (CPT 83519) | 40 | --- | --- | ||
| H. pylori test (CPT 86677) | 20 | --- | --- | ||
| H. pylori triple therapy | 120 | --- | --- | ||
| Endoscopy (CPT 43230, 88305; APC 0141) | 980 | --- | --- | ||
| EMR (CPT 43239, 43244, 43251, 88305; APC 0141) | 1500 | --- | --- | ||
| Endoscopic complication requiring surgery (CPT 43501, 99222, 99232, 99238, 88305; DRG 155) | 8510 | --- | --- | ||
| Surgery | |||||
| Dysplasia (CPT 43610, 99222, 99232, 99238, 88305; DRG 155) | 8150 | --- | --- | ||
| Preclinical local cancer (CPT 43611, 99222, 99232, 99238, 88305; DRG 154) | 23,570 | --- | --- | ||
| Cancer treatment by stage, per year¶ | [62] | ||||
| Initial year | |||||
| Local | 51,410 | Normal | 46,950–55,930 | ||
| Regional | 74,720 | Normal | 69,350–80,070 | ||
| Distant | 69,160 | Normal | 60,480–77,670 | ||
| Final year | |||||
| Local | 59,340 | Normal | 54,970–64,090 | ||
| Regional | 68,320 | Normal | 64,780–72,310 | ||
| Distant | 102,740 | Normal | 97,560–107,940 | ||
| Indirect | |||||
| Median hourly wage, 2012$ | 16.83 | Normal | 9.40–25.10 | [63] | |
| Lost time, hours | |||||
| Screening attendance | 2 | --- | --- | ||
| Endoscopy or EMR | 8 | --- | --- | ||
| Gastrectomy | 80 | --- | --- | ||
| Cancer treatment, per year | [64] | ||||
| Initial year | 350 | --- | --- | ||
| Final year | 520 | --- | --- | ||
CPT = current procedural terminology; APC = ambulatory payment classifications; DRG = diagnosis-related group; EMR = endoscopic mucosal resection.
For probabilistic sensitivity analyses, key model parameters were varied as indicated. We used beta distributions for parameters for which count data were available. For all others, we used normal distributions (bounded by 0 and 1) based on reported 95% confidence intervals. For costs, we assumed that Medicare reimbursement rates were fixed, but varied stage- and phase-specific cancer treatment costs based on published 95% confidence intervals.
Based on the following cutoff values: pepsinogen I levels ≤70μg/l and pepsinogen I/II ratio ≤ 3.0.
Parameter value was based on assumption and was not varied in probabilistic sensitivity analysis.
We assumed individuals were healthy or had mild systemic disease and surgery was emergent with intermediate procedure risk.
Annual mortality rate varies by time since diagnosis. After 10 years, we assumed individuals faced negligible risk of dying from intestinal-type NCGA (and did not incur any additional costs associated with cancer treatment).
Stage-specific (local, regional, distant). Annual cost for continuing phase (years in between first and last year) was $2540. We assumed that preclinical (asymptomatic) regional and distant cancers detected via screening would not experience any gains in survival, but would incur additional costs, approximated by continuing phase costs until they would otherwise become symptomatic.