OBJECTIVES: The OCER literature that evaluates comparative effectiveness of alternative medical intervention using existing databases has seen explosive growth in the use of PSM in recent years. However, different PSM algorithms (e.g., one-to-one, one-to-many, radius matching etc.) yield different estimates of the treatment effect. Moreover, matching-induced attrition in the original sample may change the target population for which the average treatment effect (ATE) was intended. This paper, using a real-world example, proposes a range of ATEs that results from different PSM algorithm instead of a single ATE that is typically reported in the literature. METHODS: Data for the empirical example come from healthcare claims and clinical information from electronic health records in a large U.S. hospital on patients that underwent hysterectomy with robotic (N=315) and abdominal (N=265) approaches. The ATE of interest is the predicted difference in 6-week all-cause costs starting from index surgery date. Baseline characteristics including age, race, insurance type, BMI, comorbidities and other clinical characteristics adjusted in each of the 8 PSM models. 95% confidence interval (CI) for the range of ATE is obtained through 500 bootstrap repetitions. Generalized linear modeling approach was used to model skewed costs. RESULTS: Following eight PSM algorithms were implemented: one-to-one, one-to-many (with replacement), 3 radius matching with caliper=0.01, 0.001 and 0.0001, kernel matching, and local linear regression matching. Robotic approach was found costlier with its costs varying between $16,713 and $17,522, while abdominal costs varied between $13,326 and $14,615. The minimum predicted cost difference was $2,812 (95%CI: $1,186; $4,437) while the maximum difference was $3,892 (95% CI: -$13, $7,796). As expected, the attrition in the matched sample size was highest in radius matching with caliper=0.0001. CONCLUSIONS: The range for ATEs instead of a single ATE from a specific PSM algorithm provides a better understanding of minimum and maximum possible benefit of a medical intervention.