In 1959, however, the report by Hobbes and colleagues [3] that linked an irreversible retinopathy with chloroquine therapy precipitated a distinct decline in the use of antimalarial agents for rheumatoid arthritis. This disillusionment with the drugs may have reflected the general feeling that they were not as useful as gold salts, as well as the misconception that rheumatoid arthritis was not a disease that led to an increased mortality rate or needed particularly early or aggressive therapy.

The passage of years has put the ocular toxicity of antimalarial agents into perspective. Hydroxychloroquine has caused fewer cases of retinopathy than has chloroquine; indeed, as of 1991, only 18 cases of true retinopathy causing visual loss had been reported [4], mostly in patients using more than 200 mg/d, the standard maintenance dose of the drug.

The paper by Clark and colleagues [5] in this issue of Annals reassesses the use of hydroxychloroquine, 400 mg daily, in rheumatoid arthritis. These investigators did the study to provide a more accurate estimate of the benefit of this antimalarial preparation, recognizing that both the published controlled and uncontrolled trials of hydroxychloroquine, considered in the aggregate, could not establish efficacy or confirm excessive toxicity or lack of efficacy. In addition, the authors have identified major problems common to many of the published studies: for example, multiple methodologic flaws in patient selection, measurement of disease severity and activity, variable doses, unblinded studies, and small sample size. The results presented in this issue are reassuringly affirming: Toxicity was minimal, and no patient required discontinuation of the drug therapy, confirming the published observations that in comparison with other "second-line" or disease-modifying drugs used in rheumatoid arthritis, hydroxychloroquine had a minimal toxicity profile [6]. The treated patients had significant clinical improvement over that of placebo in the broad-based joint score, pain, grip strength, and in both physician and patient global assessments.

This study also illustrates in striking clarity a recurrent finding in all studies of drug intervention in rheumatoid arthritis: Patients receiving placebo often improve initially as much as the test populations. This placebo effect has been seen by all who do drug studies in rheumatoid arthritis. It is related in part to the extra attention given study patients by caregivers, particularly the assessors of results. These data are humbling; we physicians must accept the fact that a kindly presence is often more therapeutic than potent drugs. Interestingly, and emphasized by Clark and colleagues, the profound placebo effect has been seen in other studies of patients with relatively early disease when changes within joints have a greater chance of being reversible.

Good reasons exist to use hydroxychloroquine relatively early in the course of rheumatoid arthritis and to use it as the first drug to add to nonsteroidal anti-inflammatory drugs. An effective argument can be marshaled [7] that irreversible damage to joints in rheumatoid arthritis occurs within the first 12 months of active disease and, therefore, the therapeutic philosophy of prescribing nonsteroidal anti-inflammatory drugs and asking the patient to return if things are getting worse usually leads to a patient who not only feels worse but has developed erosive synovitis.

It is appropriate to begin hydroxychloroquine treatment relatively early after rheumatoid arthritis is definitively diagnosed. How early is "relatively early"? It is any time between 6 weeks and 2 months after beginning a vigorous program of nonsteroidal anti-inflammatory drugs, education, and physical therapy with evidence by history and physical examination that synovitis is both continuous and active.

The price of the drug is not trivial. At approximately $1.00/200 mg, 1 month's course of 400 mg costs approximately $60. However, a particular advantage of hydroxychloroquine is the low cost of monitoring for toxicity. Although an ophthalmologic examination should be done every 6 months, routine hematologic and urine tests are not usually necessary. Like every oral drug that is useful for rheumatoid arthritis, hydroxychloroquine can cause gastrointestinal complaints and skin eruptions. It can exacerbate flares of psoriasis and probably should not be used as early therapy in psoriatic arthritis. The drug can cause rare neuromuscular toxicities, ranging from neuromyopathies to myasthenic reactions, but these are suspected early by history and are idiosyncratic in occurrence. One good reason for starting hydroxychloroquine therapy early in the course of disease is that both patients and physicians must have patience; the drug cannot be considered to have failed until at least 4 months of trial have elapsed. If hydroxychloroquine is not tolerated or appears not to be beneficial, it can be replaced by methotrexate or gold salts within 6 months of the beginning of disease. Using antimalarial agents in combination with two or more cytotoxic, immunosuppressive, or other slow-acting drugs has been advocated, but data showing that these combinations can be effective have yet to be presented in detail sufficient for impartial evaluation. Only in unusual cases should hydroxychloroquine, after failing in use with nonsteroidal anti-inflammatory drugs, be continued after more potent and more toxic drugs are added to the regimen.

Thus, the use of hydroxychloroquine should be considered early in rheumatoid arthritis, be appreciated for its minimal toxicity, and be accepted for its prolonged significant clinical efficacy in fewer than 50% of those in whom it is tried.