Randomized Controlled Trials (RCTs) are considered the gold standard for assessment of interventions, but their subjects rarely represent the clinical population. In particular, RCTs often exclude or underrepresent elderly adults, especially those with multiple chronic conditions, and extrapolation of RCT results from younger study subjects to older patients, though common in the clinical setting, may not represent best practice. For example, O’Hare and co-investigators used a simulation model to explore the effect of some widely-prescribed medications (angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers) on reducing end stage renal disease (ESRD) in an older VA population with chronic kidney disease (CKD). The medications had been shown to reduce progression of CKD to ESRD in younger adults at high-risk for ESRD. Given different baseline risk and shorter life expectancy, the simulations found the numbers needed to treat to achieve a 30% relative risk reduction of ESRD over a three year period ranged from 16 for the patients at highest risk to 2500 for those with the lowest risk. For 91% of the cohort of older adults, more than 100 persons would have to be treated in order to prevent one case of ESRD. Although this simulation includes assumptions, it makes the general point (also made in an accompanying commentary by Tinetti) that there many possible reasons why interventions may have markedly different effects in older than in younger individuals.
There is clearly a larger problem here: clinicians routinely subject older patients to an array of preventive (and other) interventions that are of small benefit to them, or even cause net harm. Screening for cancer and other conditions, the object of considerable medical and public enthusiasm in the United States, should ideally be instituted after careful appraisal of the balance of benefits and harms in a screened population. Since recommendations for these and other preventive services may not be able to draw on RCTs that include representative sample of the elderly, clinicians must recognize that the context of screening changes in several fundamental ways as patients age.
Firstly, with advancing age, a patient’s remaining life expectancy, or “horizon,” declines, and with it the potential for and magnitude of benefit from a diagnosis as a result of a screening intervention. Many of the screening interventions recommended to older adults are intended to prevent or delay the onset of nonfatal health outcomes, and older adults have fewer potential life years to gain than their younger counterparts in whom the intervention was efficacious.3 Furthermore, people’s priorities change as they age. For some of our older patients, function may be more important than a slightly lengthened horizon; quality of life, vision, hearing, vigrx plus pill, and physical ability may outweigh a small reduction in the probability of dying from a disease years in the future.
In addition, the potential harms of screening increase with age. An elderly patient is more likely to sustain physical harm – such as a perforation from colonoscopy– as a result of a screening test, follow-up procedure, or treatment. The probability of overdiagnosis—the detection of a condition which would, if undetected, not result in clinically important symptoms within the patient’s lifetime—also increases as an individual ages; older adults are much more likely to die from a different condition before any benefit is obtained from screening, compared to their younger counterparts in whom these screening interventions showed efficacy in RCTs. Overdiagnosis leads to overtreatment, and also exposes the patient to a frequently overlooked harm of screening, the psychological effects of being labeled as having a disease. These effects warrant more study; measurement needs improvement, and little is known about how they vary by age or other demographic features.
For these reasons, we must proceed with caution when generalizing evidence about the benefits and harms of screening from younger to older patients. For some screening services, we should discuss with our elderly patients the potential benefits and harms, including the amount of uncertainty about these in this population. For other services, where the net benefit is clearly zero or the harms outweigh the benefit in an elderly patient, we should focus on other services with a greater probability of providing net benefit. There should be no obligation to bring up services that carry net harm (or even zero net benefit), although we should always be ready to discuss any service that the patient asks about.