Table 2.

Types of sensitivity analyses described and their advantages and disadvantages

Type of sensitivity analysis	Disadvantages	Advantages	Example
Semiparametric	Requires arbitrary models for unmeasured variables. Requires positing plausible values for the unintelligible coefficients of the above model.	Mathematical convenience.	Franks et al. [26] – antihypertensives and diastolic blood pressure.
Nonparametric	Typically, none, although some methods such as instances of the E-value might use implausible assumptions [31].	Requires positing plausible values for intelligible scientific quantities (e.g., spontaneous probability of Chagas seroreversion).	Nifurtimox on Chagas disease, discussed in this manuscript.
Negative controls	Does not conclusively guarantee that associations are causal.	Requires positing an outcome with a null treatment effect, which is often feasible.	Dickerman et al. [44] – COVID vaccines
Identification bounds	Bounds are often too wide to be informative.	Operates with few or no assumptions.	Bhattacharya et al. [38] – right heart catheterization and 30-day mortality