Drug Use in Humans

The evolution of drug resistance is facilitated by a number of factors, including increasing use of antibiotics and antimalarials; insufficient controls on drug prescribing; inadequate compliance with treatment regimens; poor dosing; lack of infection control; increasing frequency and speed of travel, which lead to the rapid spread of resistant organisms; and insufficient incentives for patients, physicians, or even governments to care about increasing resistance. It is important to distinguish between risk factors for the emergence of resistance (de novo resistance) and those for the spread of resistance (primary resistance).

The molecular basis of resistance may give a clue to the likelihood of resistance emerging. If a single DNA base pair mutation leads to the development of resistance, then its selection is likely to be widespread, especially if the biological fitness cost of the mutation is low. De novo or acquired resistance results in the appearance of a resistant strain in a single patient. Subsequent transmission of such resistant strains from an infectious case to other persons leads to disease that is drug resistant from the outset, a phenomenon known as primary resistance (IUATLD 1998). Independent, cumulative events result in multidrug-resistant bacteria or tuberculosis (MDR-TB). Both the creation and the transmission of drug resistance contribute to its prevalence in a given population. This mechanism also holds true in the case of antimalarials; that is, resistance develops when malaria parasites encounter drug concentrations that are strong enough to eradicate the susceptible parasite population, but they fail to inhibit the multiplication of naturally occurring resistant strains. Commonly used antimalarial drugs are not mutagenic.

In the case of tuberculosis, spontaneous mutations leading to drug resistance occur rarely in Mycobacterium tuberculosis, and multidrug regimens can prevent the emergence of clinical drug resistance (Cohn, Middlebrook, and Russell 1959). Resistance is thus an avertable phenomenon resulting from inadequate treatment, which, in turn, is often the result of an irregular drug supply, prescription of inappropriate regimens, or poor adherence resulting from a lack of supervision. In the case of malaria, the widespread misuse of chloroquine as prophylaxis is believed to be an important factor in the emergence and spread of resistance to this drug.

Despite conventional wisdom, the highest rates of antibiotic resistance in the pneumococcus bacterium globally are not for penicillins or macrolides, which usually require multiple DNA mutations or the import of foreign genes, respectively, but for sulfamethoxazole-trimethoprim, which can be selected from among a population of susceptible pneumococci by a single base change in the dihydrofolate reductase gene (Adrian and Klugman 1997). The direct selection of resistance following exposure of children carrying pneumococci has been shown in a prospective study in Malawi to occur in 42 percent of children exposed to sulfadoxine-pyremethamine for a week and in 38 percent of children a month after exposure to drug treatment for malaria (Feikin and others 2000).

Evolutionary biology suggests that drug selection pressure is an important factor in the emergence and spread of drug resistance. Although the relationship between antimicrobial use and drug resistance (in the pneumococcus, for example) is well established in developed countries (Bronzwaer and others 2002), direct evidence to support this hypothesis is less forthcoming in developing countries because of a lack of data on antibiotic use. Resistance to antimicrobials is less likely to arise in the poorest developing countries simply because of the lower levels of antibiotic use associated with poorer socioeconomic status. For instance, India—a large country with scant control over antibiotic prescribing—has very low rates of resistance among systemic isolates of pneumococci, at least in rural areas (INCLEN 1999). These low rates exist despite wide antibiotic availability, probably because extreme poverty limits the duration of antibiotic exposure for the treatment of acute pneumococcal infections. Rising incomes and increased affordability of antibiotics will likely change this low incidence of resistance; the same may be true of quinolones, which are widely available at relatively affordable prices, even in semirural and rural populations. This trend may be responsible for the emergence of nalidixic acid resistance to Shigella in Bangladesh and fluoroquinolone resistance to Salmonella typhi in India.

Recent evidence suggests that shorter courses of antibiotics may select for less resistance in the pneumococcus compared with longer courses (when patients comply with those courses) (Schrag and others 2001). Very low levels of resistance have also been found in isolated rural African communities (Mthwalo and others 1998). This observation, however, should not lead to complacency. Increased access to antibiotics in developing countries, without controls on over-the-counter use, has led to some of the highest rates of resistance in the world, as was seen with penicillin resistance in Vietnam. Relatively wealthy countries such as the Republic of Korea and Japan also have lax control and even greater access to funds to purchase antibiotics (Song and others 1999). Patterns of resistance differ by antimicrobial class, and resistance to several classes has been linked to particular patterns of use in developing countries. Macrolide use in children in China may be preferred to the use of beta-lactams, which are known to be associated in rare instances with serious anaphylactic reactions, and in Beijing and Shanghai, the highest global rates of macrolide resistance are encountered in nasopharyngeal isolates from children (Wang and others 1998; Yang, Zhang, and McGee 2001). Tetracycline use remains widespread in developing countries, and poor African countries, such as the Central African Republic, may have higher rates of resistance to tetracycline than to beta-lactams or macrolides (Rowe and others 2000).

The relationship between compliance and resistance emergence in the treatment of acute and largely self-limiting infections is less robust than in the case of chronic infections such as tuberculosis (TB). It is likely that resistance selection occurs more readily in the commensal flora (for example, the pneumococcal flora of the nasopharynx) than among the organisms causing the acute infection. Thus, shorter courses (and reduced compliance) may reduce the selection of resistance in commensal flora. In contrast, in TB, selection takes place in the infecting pathogen, and poor compliance is associated with the selection of resistant strains.

Antibiotic Use in Animals

Many developed countries use antibiotics for veterinary uses, both for improving feed efficiency and rate of weight gain (subtherapeutic use) and for disease prevention and treatment (therapeutic use) (Levy 1992). Although the extent of antibiotic use in animals in developing countries is unknown, one study from Kenya reported that tetracyclines, sulfonamides, and aminoglycosides were the most commonly used antimicrobials for veterinary purposes (Mitema and others 2001). Over 90 percent of the antibiotics used were for therapeutic purposes, and there was no evidence of use for growth promotion.

There is strong evidence that the use of antibiotics in farm animals promotes the development of drug-resistant bacteria in animals (Aarestrup and others 2001). Because routes for the movement of these resistant bacteria to humans are available, there is sufficient circumstantial evidence that drug resistance in bacteria associated with food animals can influence the level of resistance in bacteria that cause human diseases (Wegener and others 1999). Furthermore, mathematical models indicate that the effect of subtherapeutic use on resistance in humans is greatest when resistance levels are undetectable (Smith and others 2002). The appearance of drug-resistant strains of Enterococcus faecium in broiler meat products at retail outlets declined after the ban of antimicrobial growth promoters in Denmark (Emborg and others 2003). Salmonella has been recovered from chicken (35 percent), turkey (24 percent), and pork (16 percent) samples obtained from area supermarkets in Washington, D.C. (White and others 2001). There is evidence that dissemination of tetracycline-resistance-encoding plasmids between aquaculture and humans has already occurred in Europe (Rhodes and others 2000). The global nature of this problem became apparent in 2001, when authorities in some European countries found residues of chloramphenicol in tiger shrimp imported from China, Indonesia, and Vietnam (Holmstrom and others 2003).

Transmission of Resistant Pathogens

Once resistance has emerged in a population, it can spread both geographically and between age groups. Unsafe drinking water, unsanitary conditions, and poor infection control in hospitals are risk factors for the transmission of all infections, including resistant ones. The transmission of resistant strains from children to adults has been suggested by anecdotal reports as far back as the 1980s (Klugman and others 1986). That association is strongly supported by the role of conjugate pneumococcal vaccine in reducing antimicrobial resistance among adult pneumococcal bacteremic isolates in the United States (Whitney and others 2003). The association of HIV infection with pediatric serotypes and antimicrobial resistance in pneumococci suggests the potential utility of this approach in reducing the burden of antimicrobial resistance in pneumococci in developing countries where the burden of disease is overwhelmingly associated with HIV infection in both children (Madhi and others 2000) and adults (Jones and others 1998).

Disease Burden

Although no estimates of disease burden are currently available that are specific to drug resistance, the contribution of drug resistance to the burden of infectious diseases is believed to be large. Resistance has emerged in malaria, HIV, TB, and other bacterial infections that together constitute a significant proportion of the burden of disease in developing countries. An indication of the extent of the problem is provided by the burden of diseases for which drug resistance is a problem (table 55.1), as well as by the levels of drug resistance among these pathogens (table 55.2 and figures 55.1 and 55.2).

Table 55.1

Estimated Burden of Disease in Disability-Adjusted Life Years, by Cause and Gender, 2001.

Table 55.2

Prevalence of S. pneumoniae Not Susceptible to Three or More Drug Classes, Alexander Project 1998–2000.

Figure 55.1

Chloroquine Treatment Failure in Africa

Figure 55.2

Prevalence of MDR-TB among New TB Cases, 1994–2002

Drug Treatment Strategies

The appropriate choice of drug treatment is an important step in delaying the evolution of drug resistance. Drug combinations that include drugs with different targets were first used in the treatment of tuberculosis and have now become routine practice in the treatment of cancer and HIV/AIDS. Combinations of artemisinin and its derivatives with other antimalarials, notably mefloquine, have accelerated recoveries, increased cure rates, and reduced transmissibility. In the refugee camps on the western border of Thailand, where most of the recent studies with artemisinin combinations have been conducted, the use of combinations delayed the development of resistance and reduced the incidence of disease (Nosten and others 2000). The rationale behind drug combinations is that, if resistance results from spontaneous genetic mutations, the chance that a parasite will emerge that is simultaneously resistant to two drugs with unrelated modes of action (that is, drug targets) is the mathematical product of the individual parasite mutation frequencies multiplied by the total number of parasites exposed to the drugs (White 1998, 1999). Combinations, therefore, reduce enormously the probability that a resistant mutant will arise. Sequential deployment of the drugs is much less effective, because it does not exploit the multiplicative reduction in selection risk.

In the context of antibiotics, combinations have typically been used to broaden the spectrum of antimicrobial coverage rather than to reduce the likelihood of the emergence of resistance. With the development of new penicillins and cephalosporins with broader spectra of activity a decade ago, most serious infections have been treated with monotherapy. The use of combination therapy to preserve new classes of antibiotics from the emergence of resistance at a societal level may be rational, but it has not been implemented because of cost concerns and the potential for enhanced toxicity associated with the use of more agents than necessary to effect a cure in an individual patient.

Other strategies include periodic withdrawal of a drug or rotation between different drugs. These strategies depend on the extent of the fitness cost of resistance ¹ and the extent of multidrug resistance, which may vary with the specific combination of pathogen and drug. Withdrawal or drastic decline in antimicrobial use is occasionally but not always accompanied by the replacement of resistant strains with sensitive ones. The effects of antimicrobial removal have best been assessed for antibacterial drugs in northern Europe, where drug use is tightly regulated and susceptibility patterns are closely monitored (Seppala and others 1997). A recent study demonstrated that after chloroquine was replaced by sulfadoxine-pyrimethamine in Malawi because of a loss of effectiveness attributable to resistance, chloroquine-susceptible Plasmodium falciparum strains appear to have returned (Kublin and others 2003). Though the results of this study offer promise for stopping or reversing resistance trends, if antimicrobials are more sparingly and less indiscriminately applied, little is known about the rate at which the resistance to chloroquine may reemerge with widespread use of this drug.

Rotating between two or more antibiotics has been proposed to address the problem of nosocomial drug resistance in the United States (Bergstrom, Lipsitch, and McGowan 2000; McGowan 1986), even though there is not much supporting empirical evidence to date. In one hospital-based study, switching from gentamicin to other aminoglycosides reduced resistance to gentamicin. However, when gentamicin was reintroduced, resistance developed rapidly (Gerding 2000). Modeling studies have indicated that a superior strategy may be to increase antimicrobial heterogeneity so that different patients are treated with drugs to which mechanisms of resistance are independent (Bonhoeffer, Lipsitch, and Levin 1997; Laxminarayan and Weitzman 2002). Although this may be difficult to implement in many developing countries, the approach incorporates an evolutionary perspective that may help deal with drug resistance.

Reducing Selection Pressure

Inappropriate antimicrobial use constitutes selective pressure without a corresponding benefit to individual or public health. (Eliminating all antibiotic use could, of course, eliminate the problem of drug resistance, but this strategy is clearly undesirable.) This multifaceted problem arises from behaviors of prescribers (not always physicians), dispensers (not always pharmacists), and consumers (not always infected). An important factor in overprescription is the issue of externalities; physicians, patients, and pharmacists have few incentives to consider the effects of their prescriptions or drug use on overall levels of resistance and the burden imposed on the rest of society. Physicians, both in private practice and in hospital settings, may also derive income from drugs sold and may, therefore, prescribe antibiotics more frequently than is desirable. In China, for instance, many hospitals rely on selling drugs for the bulk of their revenue (Hu and others 2001). Patient pressure demanding a prescription is known to influence prescribing in developed countries but could be less important in developing countries.

Interventions at the Provider Level

In this section, we discuss interventions directed at health care providers and local retail pharmacies, such as education and professional accountability.

Interventions at the Patient Level

Improving communication between patients and providers could improve adherence to prescribed antimalarial regimens (Agyepong and others 2002). Compliance can also be positively affected by packaging. Blister packages, when combined with proper instruction about drug use, have been shown to produce modest increases in antimalarial compliance, particularly for long-term regimens, such as with primaquine (Qingjun and others 1998). Blister packages are also time-savers for primary health care workers, potentially allowing them more time to advise patients on drug use. However, the introduction of blister packages must be accompanied by clear directions to avoid injury following ingestion of blister packages.

Reducing patient self-medication may also desirable, although the effect on drug resistance remains to be precisely quantified. Enforcement of prescription-only regulations for most antimicrobials reduces self-medication in developed countries (Carey and Cryan 2003; Goff, Koff, and Geiling 2002; Pechere 2001) and may be desirable in developing countries as well. However, such a strategy may be difficult to implement in developing countries. There have been only rare reports of reduction in antibiotic use following blanket enforcement of prescription supply legislation in areas where antibiotics are freely available (Bavestrello, Cabello, and Casanova 2002). Opponents to enforcement demand a heavy financial and political investment, implying that a black market for medicines could emerge, particularly if the demand for these drugs is not lowered (Bhutta and Balchin 1996; le Grand, Hogerzeil, and Haaijer-Ruskamp 1999). The sale of antimicrobials is a lucrative business, even when illegal, because of high demand.

Educational interventions directed at consumers have been proposed to reduce self-medication and increase compliance, but evidence for the effectiveness of this strategy remains inconclusive, largely because so few studies have been conducted. Gonzalez Ochoa and others (1996) demonstrated that, although refresher training for health personnel managing acute respiratory infections reduced antibiotic prescribing by 9 to 19 percent in two intervention areas in Havana, no benefit was seen from community education programs when used alone or in conjunction with prescriber training. In contrast, Denis (1998) was able to show that a poster had 5 percent effectiveness alone and 20 percent effectiveness when used with a video to promote appropriate use of quinine and tetracycline regimens for malaria among Cambodian villagers. The best method to deliver information about the consequences of antibiotic resistance remains to be identified and will need to be modified to suit different cultures (Marin and others 1995).

In the case of tuberculosis, first-line directly observed therapy (DOT) remains one of the most cost-effective of all public health strategies (WHO 1994). The discovery that DOT can be administered successfully as a short course (DOTS) has been pivotal to successful implementation. Relatively simple, the DOTS approach can improve patient compliance, cure the vast majority of new TB patients, and prevent transmission of the disease and the emergence of MDR-TB (Balasubramanian, Oommen, and Samuel 2000; Dye and others 2002). Unfortunately, many countries have been slow to adopt and implement DOTS programs correctly, and only a minority of TB patients worldwide are managed according to this protocol (Dye and others 2002; Pungrassami and others 2002). Because many patients either are treated outside the DOTS regimen or do not adhere to the long-term chemotherapy necessary to eradicate the causative organism, MDR-TB is likely to emerge and treatment costs are likely to escalate to as high as 1,000 times the cost of conventional treatment of drug-sensitive infection.

A DOTS program requires good laboratory support for case identification and is highly employee intense, requiring health workers to observe the ingestion of every dose of antimicrobials over several months. However, the potential economic gains are immense because of the high costs of allowing the continued spread of the disease and of managing resistant patients and because of the benefit of reduced hospital admissions (Dye and others 2003). Several studies have investigated means for reducing the cost of conventional DOTS programs without compromising effectiveness. Lwilla and others (2003) demonstrated that both institutional and community-based DOTS programs are effective, permitting cost-effective implementation in remote areas. Furthermore, although DOTS involves supervised drug dosing, highly trained health workers are not essential. Studies in South Africa, Haiti, and Thailand have found that DOTS was effective when drug administration was supervised by appropriately trained volunteers, including storekeepers and former TB patients (Barker, Millard, and Nthangeni 2002; Pungrassami and Chongsuvivatwong 2002).

Although DOTS was not designed to cure patients with MDR-TB, it succeeds in 50 percent of cases (Espinal and others 2000). Controversy has emerged about the best approach to MDR-TB in resource-constrained settings. Although some experts assert that standard TB control prevents the emergence of MDR-TB in a cost-effective way (Chaulet, Raviglione, and Bustreo 1996) and that expensive treatment of MDR-TB would divert scarce resources from struggling DOTS programs, others argue that it is unethical to abandon MDR-TB patients and maintain that, if untreated, MDR-TB strains will become dominant, undermining tuberculosis control in future generations (Farmer, Becerra, and Kim 1999). These arguments are of particular consequence for programs in poor countries.

Application of Antibiotics for Nonhumans

The use of antimicrobial agents in agriculture and aquaculture can contribute to the spread of antimicrobial resistance in humans, although the extent of this contribution has not been precisely quantified. It is believed that agricultural antibiotic use hastens the emergence of resistant pathogens in humans, and antibiotic use in humans contributes to the spread of resistance once it has emerged (Smith and others 2002). There are no reliable estimates of the extent to which antibiotics are being used for such nonhuman purposes even in developed countries. In developing countries, one would expect use to increase with rising incomes and greater industrialization of agriculture and food production.

WHO has recommended that antimicrobials normally prescribed for humans should no longer be used to promote growth in animals (WHO 2000). Some countries in Europe, including Denmark and Sweden, have already phased out such use, and under the current plan, the European Union will ban the use of antibiotics for growth promotion by 2006. Evidence of the effect of a ban on antibiotic use in swine production is available from Denmark. Although the ban significantly lowered the use of antibiotics in growth promotion and raised the cost of swine production by less than 1 percent, the resulting higher incidence of disease among swine increased the use of antibiotics used for veterinary therapeutic purposes (Hayes and Jensen 2003). A similar increase was noted in Sweden, where antibiotic use in animals was banned in 1986; however, this problem was temporary, and in the longer term, livestock producers were able to move to an effective production system with lower antibiotic use (Wierup 2001).

Containing the Spread of Resistant Micro-organisms

Although selection is a necessary component of the resistance archetype, the dissemination of resistant organisms may have a far greater effect on the current situation (Zaidi and others 2003). Interventions that block this dissemination have the added benefit of improving health by interrupting disease transmission and reducing the need for antibiotics in the first place. A case in point is the reduction in drug-resistant Streptococcus pneumoniae infections in the United States following the introduction of a multivalent vaccine that protected against the serovars with which resistance is commonly associated (Whitney and others 2003). By contrast, a drop in antibiotic use had no detectable effect on resistance within a short period (Arason and others 2002). Because these types of interventions are easier to evaluate, and in many cases are cheaper to implement (Coast and others 2002), they are likely to be of great value to public health in developing countries.

Interventions that interfere with the spread of many infectious diseases will have a parallel effect on the dissemination of resistant micro-organisms. Because the potential health benefits are obvious, these types of interventions are likely to be sustained (Wilson and Chandler 1993). A simple, cost-effective example is hand washing, which could reduce diarrhea by 47 percent while also having beneficial effects on acute respiratory tract and other community-acquired infections (Curtis and Cairncross 2003) (see chapters 19 and 35). Similarly, insecticide-impregnated bednets are another important intervention for the control of malaria (see chapter 21).

The emergence, persistence, and intra- and interhospital spread of multidrug-resistant organisms have all been facilitated by inadequate infection control practices. Furthermore, the emergence and spread of drug-resistant nosocomial pathogens from hospitals to the community are also a concern, and a history of hospitalization has been identified as a significant risk factor for the acquisition of a resistant infection in family members (OR 4.5, p = 0.007) (Zaidi and others 2003). Unfortunately, we lack good clinical trials that compare the different approaches to infection control programs and their ability to control antimicrobial resistance in hospitals and other health care facilities (Duse and Smego 1999). It seems reasonable to assume, however, that if the overall frequency of nosocomial infections is decreased in a health care facility, then the need for antimicrobial agents may be reduced. Furthermore, well-structured and rational infection control strategies that balance resources with the magnitude of the local problem must surely play an important role in decreasing morbidity, mortality, and costs (direct and indirect) to the patient, his or her family, the hospital, and the health care sector in general. Hospital-acquired infections rank among the most important causes of death, either directly or indirectly, in the developing world (Duse and Smego 1999; Ponce-de-Leon and Rangel-Frausto 1993).

Global Coordination

Antimicrobial resistance is a global challenge that requires urgent global action, not just across national boundaries but also across the whole range of sectors involved. International travel and trade, particularly of food products, have facilitated the rapid globalization of resistance, and actions undertaken by any single nation have consequences for other countries. Drug resistance may threaten health gains made in other spheres of public health. Thus, coinfection with antimicrobial-resistant pathogens and HIV can lead to more rapid disease progression and enhanced dissemination of resistant pathogens. The emergence of antimicrobial resistance is considered a major threat to the future security and political stability of some regions (CIA 1999). Figure 55.3 shows the likely geography of the emergence, spread, and evolution of chloroquine resistance.

Figure 55.3

Global Spread of Chloroquine-Resistant Strains of P. falciparum

Concerted international action is needed to contain antimicrobial resistance. Failure of individual countries to act to contain resistance could lead to both national and international consequences. For instance, the use of artemisinin as monotherapy in any single country could potentially lead to the rapid evolution of resistant strains, which could threaten the use of this valuable drug in any other part of the world.

There is a considerable need for greater international collaboration on surveillance of antimicrobial resistance, both for routine surveillance and as an early warning system for unusual resistance events. Although existing laws at the international level require reporting of some infectious diseases, they do not include any systematic reporting of antimicrobial resistance. Certain multiresistant pathogens, such as methicillin-resistant S. aureus, are now notifiable at the national level in some countries, but the global nature of the resistance problem means that national legal measures alone are not enough. At the same time, the creation of new international duties would be undermined if they were not incorporated into national law (Fidler 1998).

To support such surveillance, the World Health Organization makes software (WHONET) available to enable clinical laboratories to enter their drug susceptibility test results into databases that can be analyzed for local management of resistance. Those results can also be merged, creating, thus far, approximately 80 national surveillance databases. Statutory notification about pathogens with new resistance phenotypes is under active discussion in several countries. Interpretation of existing surveillance data is hampered by the multiplicity of methods used to measure resistance and by the difficulties of assessing the quality of the data. Enhanced laboratory capacity is needed in many countries to provide effective diagnostic services and resistance surveillance. Multicountry external quality assurance schemes already exist but need to be extended to cover more resource-poor nations. WHO has begun establishing international surveillance standards by issuing antimicrobial resistance surveillance standards (WHO 2002a), guidelines for the management of drug-resistant tuberculosis (WHO 1997), and protocols for detection of antimicrobial drug resistance (WHO 1996). Monitoring the use of antimicrobial agents is an important aspect of surveillance and needs to be strengthened internationally.

Coordinated surveillance for drug resistance to TB may be a useful prototype to follow. By 2003, a global network of 20 supranational reference laboratories and five regional subnetworks, known as the World Health Organization–International Union against Tuberculosis and Lung Disease (WHO–IUATLD) Global Project, was reporting data on representative cases of TB from 62 countries. The network has monitored participants' laboratory testing methods and developed uniform definitions, such as establishing clinical categories for new cases and previously treated cases. The WHO–IUATLD Global Project, which covers approximately 33 percent of the world's population and 28 percent of its reported cases of TB, and other surveillance efforts reveal great variations from place to place in the prevalence of resistance to antituberculosis drugs.

The WHO Global Strategy for Containment of Antimicrobial Resistance (WHO 2001) recommends more than 60 interventions that can slow the emergence and reduce the spread of resistance in diverse settings, and it provides a blueprint for global coordinated action. A recent report by the Institute of Medicine calls for such coordinated action at a global level to subsidize artemisinin-based combination treatments for malaria for deployment in developing countries to ensure that combination treatments are available at the same or lower price than monotherapies. The underlying rationale is that this strategy would both save lives in the short term and delay the emergence of resistance in the long term, thereby benefiting all malarious countries (Arrow, Panosian, and Gelband 2004).

Encouraging the Development of New Drugs

The development of new products has not kept pace with the problem of addressing microbial resistance to drugs that are used to treat infections in human and veterinary medicine. In the context of drug resistance, it is not sufficient to design policies that will encourage only the development of new drugs. It is also important that such policies give pharmaceutical firms (a) an incentive to invest in new classes of drugs, rather than products that belong to existing drug classes and that are, therefore, more likely to be rendered obsolete by resistance, and (b) an incentive to care about drug resistance when making decisions about how to market and sell their product. (See chapter 6 for a discussion of product development not specific to drug resistance.)

Current public and private investment in drugs, vaccines, and other products to control major infectious diseases in developing countries has recently been less than 2 percent of total health research expenditures throughout the world. Of the 1,393 new chemical entities registered by Western health authorities between 1975 and 1999, only 13 (less than 1 percent) were intended for the treatment of tropical diseases, and half of those came from veterinary research.

Evidence suggests that in recent years, a steadily decreasing proportion of pharmaceutical company profits has been invested in antimicrobial research and development (R&D). One reason given for this decline is increased development costs, which result from more complex clinical trials, longer development times, and the relative attractiveness of returns from investing in drugs for chronic diseases, which must be taken continuously rather than just for the duration of an infection. Efforts to reduce inappropriate antibiotic use may paradoxically also reduce incentives for pharmaceutical firms to invest in developing new products (Monnet and Sorensen 2001; Philipson and Mechoulan 2003). The related issue, of using shorter treatment courses for many infections, may also affect R&D incentives (Pichichero and Cohen 1997).

Encouraging research into antimicrobial agents that will be used primarily in low-resource countries poses particular challenges, given the need for drug companies to make a profit. Various incentives to the industry, including both push and pull mechanisms, have been discussed (Kettler 2000). Push mechanisms consist of incentives to offset R&D costs, such as research grants, tax credits, public investment in applied research, cost-sharing between companies, and establishment of local development facilities. Pull mechanisms are designed to create a market, thereby improving the likelihood of a return on investments. They could include an international purchase fund for a new antimicrobial that meets specific criteria (United Kingdom Government 2001), tax credits on sales, and favorable patent policies, such as extension of patent terms or market exclusivity on new products. Time-limited exclusivities on new, clinically useful formulations might stimulate the additional pharmaceutical and clinical studies that are needed to support licensure. Better patent protection for new antimicrobial agents in areas of the world where patent laws are not enforced today would also help.

Patent policies to encourage investment in developing new antibiotics should also take into consideration the effect of those policies on incentives that would encourage pharmaceutical firms to conserve the effectiveness of their products. If effective patent lengths are too short, pharmaceutical companies will be less likely to care about growing drug resistance to their product and will be more interested in maximizing sales of their products during the period of patent production. Extending patent lengths may not solve the problem, however. If different firms make closely related antibiotics that have linked mechanisms of action, no single firm may have an incentive to care about drug resistance (Laxminarayan 2002). Increasing patent breadth, one solution to this problem, would have the added advantage of creating an incentive for firms to invest in developing new classes of drugs rather than introducing drugs that are closely related to existing products in modes of action.

Integrated Management of Childhood Illness

The program on the integrated management of infant and childhood illness (IMCI) was initiated by WHO, the United Nations Children's Fund, and other technical partners. The program, which provides a framework for stepwise assessment and management of sick children, has been successful in averting unnecessary antibiotic and antimalarial use and in reducing the cost of medications (Gove 1997). An underlying objective of IMCI is to detect, where present, conditions other than those responsible for the primary complaint. The result is that rational prescribing may actually increase, but because IMCI guidelines reduce irrational prescribing, the net effect is a reduction in drug use, including antimicrobials (Oluwole, Mason, and Costello 2000). Drug costs associated with treating children in Kenya were found to be reduced from an average of 44 cents per patient (U.S. currency) to between 16 and 39 cents per patient when IMCI guidelines were followed (Boulanger, Lee, and Odhacha 1999). A study that compared prescriptions arising from standard and IMCI-guided consultation with a health worker in Kaduna state, Nigeria, found a reduction in polypharmacy when IMCI guidelines were used, from a median of five drugs down to two drugs, and an 80 percent reduction in the cost of all medicines (Wammanda, Ejembi, and Iorliam 2003). This reduction represented savings of 93.4 and 68.6 percent for antibacterials and antimalarials, respectively, although part of the savings could be attributed to the substitution of tablets for more expensive syrup formulations.

The contribution of IMCI to promoting judicious drug use is worth noting. A study that compared four districts of Kenya—two districts that had IMCI programs (Morogoro and Rufiji) and two that did not (Ulanga and Kilombero)—found that in 73 percent (95 percent CI; 65–80) of consultations studied in the IMCI intervention districts, a child needing an oral antibiotic or antimalarial was prescribed correctly, compared with 35 percent (95 percent CI; 25–45) in the control districts (Armstrong Schellenberg and others 2004). Also, in the IMCI districts, 86 percent (95 percent CI; 80–92) of children who did not need antibiotics left the facility without an antibiotic, compared with 57 percent (95 percent CI; 48–66) in the control district.

Directly Observed Therapy

Strict compliance and cure verification make it possible to conduct directly observed therapy using shorter-term regimens (the DOTS program). Outpatient treatment of a single TB case in Pakistan, with complete compliance, was recently estimated at US$164 and increased to US$310 when an institution-based DOTS program was applied (Khan and others 2002). In Beijing, the cost of saving one disability-adjusted life year (DALY) is 10 times higher when a DOTS TB program is not used (Xu, Jin, and Zhang 2000). Recent expansion of DOTS care in India has resulted in savings of about 0.2 million lives and US$400 million in indirect costs (Khatri and Frieden 2002). Progress with DOTS in India was initially slow, but following better implementation, recent findings suggest that huge successes have been made in the past five years. The successes can largely be attributed to improved management of the program, area-specific appraisals, infrastructure, personnel, and technical support as well as continuous supervision (Khatri and Frieden 2002). This model is one that may apply in other developing countries.

In New York city, individualized chemotherapy based on drug susceptibility testing was nearly as effective in new patients with MDR-TB as in those with drug-susceptible TB (Telzak and others 1995), and the number of MDR-TB cases decreased by more than 90 percent within the decade (Fujiwara and others 1997). The relatively large number of MDR-TB cases in high-income countries made second-line drugs more available and affordable, and experience in the management of MDR-TB grew considerably over the past decade (Mukherjee and others 2004). Yet questions remained about which interventions had led to the city's success. In 1999, WHO created a working group on DOTS-plus to assess the feasibility and cost-effectiveness of treating MDR-TB in low- and middle-income countries (Espinal and others 1999). DOTS-plus has already negotiated a 90 percent price reduction from the pharmaceutical industry (Gupta and others 2001), and experience from Peru suggests feasibility and a mean cost of US$211 per DALY gained (Suarez and others 2002).

Recommended Interventions

Sets of interventions specific to the diseases discussed in this chapter are described below.

Research Agenda

A description of a research agenda to explain the full range of issues related to drug resistance is outside the scope of this chapter and has been accomplished by other groups. Here, we restrict our focus to the following five priorities:

• accounting for attributable morbidity and mortality and the economic burden of drug resistance in the developing world

• measuring the cost-effectiveness of interventions to improve prescribing and patient compliance

• researching incentives for firms to invest both in developing new drugs and in maintaining the effectiveness of existing drugs

• identifying socioeconomic, demographic, and cultural factors that determine antibiotic use and misuse and projecting how antibiotic use will change in future years

• designing international coordinating mechanisms for surveillance to report resistance outbreaks and coordinate strategies for appropriate drug use, recognizing the global nature of drug resistance.