Numerous highly efficacious treatment modalities exist in dermatology, yet patients may be highly wary of their possible adverse events, even when those risks are rare.1,2 Such fears can lead to poor medication adherence and treatment refusal. A key determinant in successful patient-provider care is to effectively communicate risk. The communication of risk is hampered by the lack of any common currency for comparing risks. The development of a standardized unit of risk could help facilitate risk comparisons, allowing physicians and patients to put risk levels into better perspective.
One easily relatable event is the risk of injury in an automobile crash. Driving, whether to the dermatology clinic for a monitoring visit or to the supermarket for weekly groceries, is associated with risk of injury and death. The risk of automobile-related injury warranting a visit to the emergency department could provide a comparator that physicians can use to give patients a more objective sense of treatment risks or to introduce the justification of a monitoring visit. The objective of this study was to develop a standard risk unit based on the lifetime risk (LTR) of automobile injury and to compare this unit of risk to various risks of dermatologic treatments.
We first identified common risks in dermatology that would be illustrative and then identified keywords. PubMed searches for articles indexed for MEDLINE from November 1996 to February 2017 were performed combining the following terms: (relative risk, odds ratio, lifetime risk) and (isotretinoin, IBD; melanoma, SCC, transplantation; indoor tanning, BCC, SCC; transplant and SCC; biologics and tuberculosis; hydroxychloroquine retinal toxicity; psoriasis and psoriatic arthritis). An additional search was performed in June 2018 including the term blindness and injectable fillers. Our search combined these terms in numerous ways. Results were focused on meta-analyses and observational studies.
The references of relevant studies were included. Articles not focused on meta-analyses but rather on observational studies were individually analyzed for quality and bias using the 9-point Newcastle-Ottawa Scale, with a score of 7 or more as a cutoff for inclusion.
Determination of Risk Comparators
Data from the 2016 National Safety Council’s Injury Facts report were searched for nonmedical-related risk comparators, such as the risk of death by dog attack, by lightning, and by fire or smoke.3 Data from the 2015 US Department of Transportation Traffic Safety Facts were searched for relatable risk comparators, such as the LTR of automobile death and injury.4
Automobile injury was defined as an injury warranting a visit to the emergency department.5 Automobile was defined as a road vehicle with 4 wheels and powered by an internal combustion engine or electric motor.6 This definition excluded light trucks, large trucks, and motorcycles.
Lifetime risk was used as the comparative measure. Lifetime risk is a type of absolute risk that depicts the probability that a specific disease or event will occur in an individual’s lifespan. The LTRof developing a disease or adverse event due to a dermatologic therapy or interventionwas denoted as LTRadverse event and calculated by the following equation7,8:
In this equation, LTRgeneral population is the LTR of developing the disease or adverse event without being subject to the therapy or intervention, and RRintervention is the relative risk (RR) from previously published RR data (relating to the development of the disease in question or an adverse event of the intervention). The use of equation (1) holds true only when the absolute risk of developing the disease or adverse event (LTRgeneral population) is low.7 Although the calculation of an LTR using a constant lifetime RR may require major approximations, studies evaluating the variation of RR over time are sparse.7,9 The Newcastle-Ottawa Scale was used to control such variance; only high-quality, nonrandomized studies were included. Although the use of residual LTR would be preferable, as LTR depends on age, such epidemiological data do not exist for complex diseases.
When not available, the LTRgeneral population was calculated from the rate of disease (cases per 100,000 individuals per year) multiplied by the average lifespan of an American (78.8 years)10:
When an odds ratio (OR) was presented, its conversion to RR followed11:
In this equation, RC is the absolute risk in the unexposed group. If the prevalence of the disease was considered low, the rare disease assumption was implemented as the following11,12:
The use of this approximation overestimates the LTR of an event. From a patient perspective, this approach is conservative. If prior LTR values were available, such as the LTR of automobile injury, automobile death, or other intervention, they were used without the need for calculation.
The LTRs of all adverse events were normalized to a unit comparator, using the LTR of an automobile injury as reference point, denoted as 1 risk unit (RU):
This equation allows for quick comparison of the magnitude of LTRs between events. Events with an RU less than 1 are less likely to occur than the risk of automobile injury; events with an RU greater than 1 are more likely than the risk of automobile injury. All RR, LTR, and unit comparators were presented as a single pooled estimate of their respective upper-limit CIs. The use of the upper-limit CI conservatively overestimates the LTR of an event.