Medications are the principal tools doctors use to maintain health, reverse illness, and extend patients’ survival, hopefully with good quality of life. Yet medications can also cause serious illness and fail to have the desired effect if they are used improperly. Additionally, medications can be extraordinarily expensive, and the cost to individual patients, to hospitals, and to our health system can become almost prohibitive. So the proper use of medications is critically important.

Before any medication is ordered in a hospital or prescribed for an outpatient, the prescriber needs to consider the (1) efficacy, (2) safety, and (3) cost of the medication, in that order of importance. Without efficacy for the condition being treated, no medication should be given. “It’s not likely to be harmful” is no justification for trying something without demonstrated efficacy for the patient’s problem, unless the intervention is in the setting of a clinical trial or the patient is informed of off-label use without evidence of benefit.

Evidence for Efficacy

The quality of medical studies supporting the use of medications varies widely. In recent years, the quality of data has been graded by the groups reviewing the literature and making recommendations, such as the Chest guidelines for anticoagulation (Table 11-1). These grading systems consider the methodologies of the studies as well as the strength of the results.

Among the difficult issues with clinical trials is whether they can be extrapolated for all drugs in the same class. In general, extrapolation across a class is somewhat hazardous, as drug formulation, absorption, duration of effect, and sometimes drug interactions differ among drugs in the same class. Even with HMG CoA reductase inhibitors, whose effects on LDL cholesterol are mostly affected by drug potency and can often be equated through adjustment of dose, the efficacy related to clinical outcomes and adverse effects may vary. Thus what is true for one drug in a certain class may not be true for other drugs in the same class.

Another issue regarding the validity of clinical trials is the use of “surrogate markers” in place of “hard clinical end points.” An example is a reduction of HIV RNA levels as a surrogate for medication efficacy instead of prolonged survival in patients with AIDS. Some surrogate markers have been demonstrated through rigorous clinical studies to be closely associated with hard clinical end points, providing assurance that they can be trusted as substitutes. Other surrogate markers have less data to justify their use as substitutes. A recent study points out the hazard of surrogate markers: a study of interleukin-2 therapy in patients with HIV infection showed a substantial and sustained elevation of CD4+ cell count over a period of 7–8 years average follow-up, but no improvement in survival or the incidence of opportunistic infections.

Another common outcome strategy in clinical trials is “composite end points,” combining as an “event” any one of several conditions, such as cardiac death, nonfatal myocardial infarction, and admission to a hospital for unstable angina. Obviously, all of these conditions are defensible as outcomes in patients with coronary artery disease, but they are decreasingly reliable as “hard clinical end points” for an intervention intended to influence the course of coronary artery disease. Especially when one of the three conditions contributing to the composite end point is the result of variable clinician judgment (eg, when to admit a patient for unstable angina), the reliability of the composite end point decreases.

Safety

Throughout all phases of drug development before drug approval (phases I, II, and III), safety is assessed, but at best these studies involve only a few thousand study subjects for the vast majority of drugs. With this number of patients, only side effects of moderate frequency (around 1–10 per thousand) will be identified. More rare (and often more serious) side effects may only become recognized with much more extensive use, involving tens of thousands of people. The experience with drugs such as troglitazone emphasizes the importance of postmarketing reporting of toxicities associated with newly approved medications to MedWatch and/or to the manufacturer.

There is a risk of toxicity with virtually all medications, so there must be a consideration of risk and benefit before starting or continuing medications. In many cases, the toxicity emerges without warning (“idiosyncratic”), such as rashes in response to sulfa drugs. These “adverse drug events” are usually unpredictable and are not considered “medication errors.” In other cases, the possible toxicities of medications can be identified and treated before they become clinically dangerous (eg, hypokalemia with loop diuretics or hyperkalemia with ACE inhibitors). These adverse drug events are not medication errors either, unless the patient is not monitored appropriately with occasional serum potassium measures.

Cost

The cost of medical care seems to steadily rise. The contribution of medication cost to overall healthcare expenses more than doubled from 4.7% in 1982 to 10.5% in 2002. Interestingly, while drug costs continue going up, the rate of increase in the cost of prescription drugs has decreased over the past 2 years, increasing only 4% from 2006 to 2007. Individuals sometimes find that they are unable to afford their medications, and as a result these patients often go without them. This “economic noncompliance” increases during difficult economic periods or when people have fixed incomes and must choose between paying for these medications or their food or mortgage. If hospitals and health systems could pay less for their medications, they would have more funds available for capital improvements or expanded personnel services.

Clinical trials have increasingly been including assessment of the quality of life saved, not just the survival rate. The measure of quality-adjusted life years (QALYs) is a standard and internationally recognized method to assess the relative benefit of medical interventions. It combines duration of survival and the quality of life during each year of life. Although one treatment might help someone live longer, it might also have serious side effects (eg, it might make them feel sick or put them at risk of other illnesses). Another treatment might not extend survival but it may improve quality of life (eg, by reducing pain). The quality of life rating can range from 0 (worst possible health) to 1 (best possible health). Having the QALY measurement allows one to consider cost effectiveness—that is, how much the drug or treatment costs per QALY. This is the cost of providing a year of the best quality of life available, which could be one person receiving one QALY, but is more likely to be a number of people receiving a portion of a QALY—for example, four people receiving 0.25 QALY. In this example, cost effectiveness is expressed as dollars per QALY.

Cost effectiveness analysis is another increasingly popular approach. This is another increasingly popular approach to assess the impact of intervention that may have financial benefit. For example, aspirin’s cost is much lower than the cost of caring for the heart attacks it prevents. Sometimes the benefit is secondary or indirect. For example, acetylcholinesterase inhibitors are reported to cause a temporary delay in the cognitive decline of patients with dementia. If this delay in cognitive decline can prevent a patient from requiring institutionalization or full-time care at home for a period of months or years, the costs of such care may be much more than the cost of the medication.

Policy makers including governmental bodies, payers, and influential foundations are interested in maximizing cost effectiveness. They are convinced, with some justification, that many practices and interventions might well be replaced with less costly approaches, without diminishing the quality of the care and the benefit our patients derive.

Patient Preferences and Values

With rare exception, prescribers have a number of possible medications for managing diseases, and each may cause likely responses (good or bad) in addition to the intended response. In all cases, the patient’s inclination to accept the proposed therapy should be considered.

The very choice of initiating medication treatment or not should be weighed. It is always an option in medicine to do nothing (offer no treatment), and sometimes no treatment is the best option. For example, in a patient with an acute inferior wall myocardial infarction who develops a Mobitz I block (Wenkebach), the occasional missed beat is of no clinical consequence, creates no risk for the patient, and almost always resolves without intervention. Treating such a problem with atropine or a pacemaker would be a mistake, introducing some risk of toxicity or complication for no clinical benefit, so no treatment is the best approach for such patients.

The prescriber should also consider coexisting medical conditions that might likewise benefit from the same therapy, as this may magnify the benefit of the medication without adding additional risk of toxicity. For example, in a patient with hypertension who also suffers from frequent migraine headaches, a beta blocker or verapamil might be favored over other medications because they may reduce the frequency and/or severity of the migraine episodes at the same time the blood pressure is being reduced.