A fundamental step in improving medication safety is to establish guidelines and protocols for drugs that are frequently overlooked, are difficult or dangerous to use, or are of high cost. Written or electronic tools to prevent prophylactic therapies from being overlooked (or overused) are wise. For example, patients at high risk for developing gastric ulceration should receive prophylaxis, but it is clearly not necessary that every patient in the ICU receives gastrointestinal (GI) bleeding prophylaxis. Hence, guidelines to help physicians decide who are appropriate candidates, will match risk to treatment. Because of the costs, ease of therapy, and effectiveness, histamine blockers or proton pump inhibitors represent good therapeutic options when indicated (see Chapter 39). Another case in point is deep venous thrombosis (DVT) prophylaxis. Without prevention, DVT is so frequent in the critically ill that it makes sense to use preventative therapy in almost all patients but it can be overlooked. Unfractionated heparin (UFH) and low-molecular-weight heparin (LMWH) are generally safe and so inexpensive that they should be the agents of choice unless contraindications to anticoagulation are compelling (e.g., active hemorrhage, recent high-risk trauma or surgery, potential neuraxial bleeding, significant thrombocytopenia). For patients unable to receive anticoagulants, a combination of graded compression stockings and intermittent pneumatic devices is a less satisfactory nonpharmacological alternative. From an economic standpoint, the annual costs of prophylaxis for an entire ICU may be dwarfed by the price to treat just one case of thromboembolism or massive GI bleeding. The importance of DVT prophylaxis has been magnified, now that funding and regulatory agencies are holding hospitals responsible for un-prevented thromboembolism. A related example is implementation of a treatment protocol for UFH dosing. Without such guidance, therapeutic failure is commonplace and complications frequent (see Chapter 23).

To prevent excessive or inadequate treatment, it is also an excellent idea to develop guidelines for dosing mediations to objective endpoints. Using a validated pain scale to guide opioid dosing can achieve better analgesia with fewer side effects. Use of sedation dosing tools (e.g., Richmond Agitation Sedation Scale [RASS]) with mandated drug interruptions has been shown to reduce total doses of administered drugs and shorten the length of mechanical ventilation and ICU stay while lowering costs (see Chapter 17). Although the best target value for glucose can be debated, validated glucose control protocols are also very sensible to minimize risks of hypoglycemia while improving glycemic control. For some medications that are used infrequently, providing a checklist or protocol to maximize the chance a patient can benefit from and safely receive the drug makes sense. Examples include drotrecogin alfa activated for severe sepsis, tissue plasminogen activator for ischemic stroke, and nitric oxide for hypoxemic respiratory failure.

For the most complicated or dangerous drugs, it even makes sense to restrict prescribing to physicians with special training or qualifications. Cancer chemotherapy is a prime example. Another case in point is infectious disease consultation for decisions regarding use of side effect prone or expensive antimicrobial therapies (e.g., ganciclovir, liposomal amphotericin B, voriconazole). Another instance would be restricting prescribing of tissue plasminogen activator and glycoprotein IIb/IIIa inhibitors to qualified cardiologists. A final example would be to restrict drotrecogin alfa activated prescribing to critical care physicians experienced in treating severe sepsis. However, it is a very bad idea to design impediments to timely use of therapies merely for cost control. Doing so might reduce acquisition costs for that drug but are not likely to save any money as outcomes worsen and stays lengthen.

Another step toward optimizing medication use is to eliminate duplicative or overlapping therapies. It is common to see patients’ prescribed suboptimal doses of two or more narcotics and a similar number of benzodiazepines for pain and sedation, respectively. It is also reasonably common to see a patient with asthma or chronic obstructive lung disease to have inhaled corticosteroids wastefully coadministered on top of high dose of oral or parenteral corticosteroids. Antibiotic therapy is frequently
duplicated. Examples include concomitant use of a third generation cephalosporin and extended spectrum penicillin; two simultaneous quinolones; coadministration of clindamycin and metronidazole; or perhaps simultaneous treatment with oral vancomycin and metronidazole. In each case, a much better strategy is to reduce the number of drugs and dose each to optimal effect. Parsimony reduces costs and risks of adverse effects and drug interactions. In addition, if something does go wrong, it is much easier to identify the culprit when there are fewer medicines.

It is always a good idea to ask if one drug can be used to accomplish two purposes. For example, in a patient with suspected pneumonia and a possible urinary tract infection, is there one antibiotic, or combination, that will effectively treat both? Another example of this principle would be selecting a benzodiazepine or propofol for sedation over another drug class in a patient who has had a seizure. Choosing the benzodiazepine or propofol provides a “free” anticonvulsant. Likewise, choosing a nonsteroidal anti-inflammatory drug like ibuprofen to control fever also provides free nonnarcotic analgesia that is much more effective than acetaminophen.

In most cases, more than one drug alternative exists, and often there are differences in safety between choices. When two drugs are equally effective, choosing the safer alternative makes sense. A prime example is using fluconazole or voriconazole in place of amphotericin B to reduce the risk of kidney injury. Another situation would be use of a fluroquinolone in place of an aminoglycoside-ampicillin regimen for a hospitalized patient with urinary tract infection, to avoid renal injury. Sometimes, the safer alternative is more expensive but now and then not. For each patient, the clinician must decide if the safety advantages between two equally effective options justify cost differences.

When two courses of therapy are equally safe and effective, cost should be considered. For example, Escherichia coli bacteriuria could be treated with generic enteral amoxicillin for pennies or with a proprietary intravenous (IV) extended spectrum penicillin for hundreds of dollars. Generic equivalents are almost always less expensive, and the good business practice of competitive bidding further reduces costs. Sometimes, the choice is between two expensive therapies, as is the case with nitric oxide and nebulized prostacyclin. Neither compound has proven outcome benefits but both lower pulmonary artery pressure and may at least temporarily improve oxygenation in life-threatening hypoxemia. A protocol detailing who might receive these treatments and substitution of inhaled prostacyclin for nitric oxide can save a hospital hundreds of thousands of dollars annually with no decrement in quality.

Establishing an automatic substitution program in which the least expensive therapeutically equivalent compound is substituted for a brand name medication also saves money. Excellent areas in which to realize these savings are with antibiotics (e.g., quinolones, advanced-generation cephalosporins), gastric acid-modifying drugs (histamine blockers and proton pump inhibitors), and sedatives (e.g., brands of propofol). The process of therapeutic substitution requires a proactive pharmacy committee and consensus, though not universal agreement, of local experts that the substitutions are reasonably “equivalent.” Reducing the number of like medications stocked by the pharmacy can also produce benefits. Pharmacy size is reduced, fewer personnel are necessary to track and manage inventory, and waste is reduced as fewer expired drugs are discarded. In the case of restriction or substitution, however, a multidisciplinary pharmacy committee must remain open to well-reasoned arguments for formulary additions or exceptions and a mechanism must exist for waiver of formulary rules under emergency circumstances.

There are numerous other ways to cut costs while maintaining or improving safety and efficacy. Surprisingly, the cost of a course of therapy often depends more on the route and frequency of administration than it does on the drug acquisition cost. Because patient charges for preparation of a dose of any IV medication average $20 to $40, it is apparent that charges for a drug purchased for $1/dose that must be given six times daily will exceed those for a drug that costs $100/dose given but once daily. In most cases, a very cost-effective measure is to minimize the number of times a day a drug is given, even if that requires using a more expensive drug. Prime examples are the substitution of once daily tiotropium for chronic
obstructive pulmonary disease treatment instead of four times daily ipratropium; use of ceftriaxone or cefepime once daily instead of cefotaxime thrice daily; and once daily LMWH instead of UFH every 8 h. Reducing the number of scheduled administrations each day has also been shown to be associated with fewer missed doses.

The route of therapy can have profound impact on costs. In general, the cost of an equivalent dose of an oral medicine is one tenth to one hundredth that of the same drug given intravenously. This vast discrepancy exists because IV preparations are usually more expensive to purchase, some drug is wasted, and there are substantial labor costs associated with stocking, retrieving, mixing, transporting, and administering an IV preparation. Essentially, all patients eating or tolerating tube feeding can receive enteral medications. In fact, many medications (including benzodiazepines, gastric acid-suppressing drugs, narcotics, and some antibiotics) have equal bioavailability when given orally and intravenously. Hence, almost any time an IV preparation can be changed to an oral route, substantial savings can be achieved. Fluconazole, quinolone antibiotics, and parenteral nutrition are great examples.

Continuous infusion is the most costly method of administration, because a dedicated line, infusion pump, and specialized cassettes and tubing, all of which are expensive, are required. Also, each infusion site increases the risk of infection, and the mere presence of an IV catheter in a patient with fever is likely to prompt an expensive evaluation and empiric antibiotic therapy. Furthermore, if a central venous catheter must be inserted for access, the danger of infection persists and the risks of arterial puncture and pneumothorax are incurred. Sometimes, even switching from continuous infusion to intermittent IV dosing is cost effective. A constant infusion of a short-acting agent usually requires a dedicated line and a pump for precise control. By contrast, intermittent dosing of a longer acting agent can free up an IV line for administration of other required medications and in the process may avoid inserting another catheter. Examples of this principle would be substitution of intermittent IV or even enteral metoprolol for continuous esmolol and intermittent IV lorazepam for continuous infusion midazolam.

The belief that giving medications by continuous infusion automatically confers accurate control over drug effects is fallacious even when the practice achieves precise regulation of plasma drug levels. Exact titration of a plasma drug level is rarely necessary or achievable, and drug levels often do not correlate with effects. Critically ill patients commonly have such altered pharmacodynamics that “short-acting drugs” have prolonged actions. Accumulation of medications given by continuous infusion is frequent enough to be considered routine (lidocaine, fentanyl, theophylline, and midazolam are prime examples). In addition, continuous infusion may obscure signs that the drug is no longer necessary. A prime example is the continuous infusion of a sedative or neuromuscular blocker in which recognition of the ability to do without sedation or paralysis is delayed by the therapy itself.

A “hidden” cost of drug use is monitoring drug levels or indices of organ toxicity (e.g., creatinine, liver function tests). Although aminoglycosides and vancomycin are inexpensive to purchase and are typically dosed infrequently, their cost can be enormous: patients have peak and trough serum levels tested on multiple occasions (at about $100/determination) along with frequent creatinine determinations, not to mention the costs associated with renal failure if it develops. (Moreover, it is far from clear that knowing the blood levels of these compounds improves efficacy, and perhaps not even safety.) Another example is the use of warfarin versus LMWH for DVT prophylaxis. At 1 cent/tablet, compared to about $15/injection, warfarin seems to be the clear choice. However, the costs of prothrombin and hemoglobin determinations typically performed on patients treated with warfarin, but not done in patients treated with LMWH rapidly outstrip the savings for the drug itself. The same situation exists for UFH versus LMWH where multiple venapunctures and the costs of activated partial thromboplastin time (aPTT) determinations are avoided by using LMWH.

It makes no sense to provide one drug that negates or counteracts the effect of another, yet it happens frequently. Common examples of this problem include simultaneous use of a histamine blocker or proton pump inhibitor and sucralfate (sucralfate requires gastric acid for mucosal binding), or coadministration of oral fluroquinolones and aluminummagnesium containing antacids (antacids chelate quinolones). Another situation which leads to antagonistic therapy is forgetting to stop certain chronic outpatient medications. One example
would be continuing a regimen of outpatient antihypertensive drugs for a patient in vasopressor dependent septic shock.

One of the most important areas for safety improvement, which also often reduces drug costs, is careful attention to dosing as organ function changes. Whole books have been written on the dose adjustment of drugs cleared by the kidney, but one basic principle is clearly important. Bedside calculation of creatinine clearance using the Cockcroft-Gault equation, which typically serves as the basis for dose adjustments, is not valid until renal function plateaus. The practical implication is that calculated creatinine clearance lags behind actual declines in renal function, usually by 1 to 3 days. Hence, a defensible strategy for adjusting renally cleared drugs is to assume a glomerular filtration rate (GFR) of zero as soon as patients develop significant oliguria (urine output < 0.5 mL / kg / h). Once the creatinine stabilizes then recalculate the GFR. Just as there are times when doses must be reduced, occasionally drug requirements go up. For instance, a higher dose of renally cleared medications may be needed in situations where GFR is increased (e.g., pregnancy and fully resuscitated major burns).