|
|
| |||||||||||||||||||||||||||||||||||||||
Background Skin diseases, including adverse reactions to drugs, are thought to be more common among patients with human immunodeficiency virus (HIV) infection and the acquired immunodeficiency syndrome (AIDS) than among other persons. These skin conditions can be disabling or disfiguring and may require discontinuation of essential drugs.
Methods We identified 684 HIV-infected members of a 265,000-member health maintenance organization and reviewed their medical records to determine the frequency of dermatologic diagnoses from April 1, 1988, through January 15, 1991. We compared the rates of visits per year for skin conditions by HIV-infected men, 20 to 49 years of age, with those by non-HIV-infected men. We used an automated prescription data base to quantify exposures to drugs.
Results Of the 684 HIV-infected patients, 540 (79 percent) were given one or more dermatologic diagnoses, for a total of 2281 diagnoses, including 188 cutaneous reactions to drugs. There were 43 hospitalizations for cellulitis (n = 15), cutaneous drug reactions (n = 13), or other skin problems. As compared with non-HIV-infected men, the men with AIDS had visit rates that were at least 5 times higher for 18 of the 20 most common infectious and inflammatory skin conditions and at least 15 times higher for 9 conditions. Drugs with the highest rate of cutaneous reactions (per 1000 courses) included trimethoprim-sulfamethoxazole (149), sulfadiazine (200), trimethoprim-dapsone (156), and aminopenicillins (93). The number of diagnoses of skin conditions increased according to the stage of disease: it was lowest in patients immediately before the documentation of HIV infection and highest in patients with a diagnosis of AIDS.
Conclusions Cutaneous diseases, including drug reactions, are extremely common in patients with HIV infection, and their incidence increases as immune function deteriorates.
Cutaneous drug reactions in HIV-infected persons result in substantial morbidity,3 and there are often no satisfactory alternatives to those drugs. The limited data on adverse cutaneous reactions to drugs in HIV-infected persons come primarily from descriptive case reports and cohort studies of special groups or subjects exposed to selected drugs3,4,5,6,7,8,9,10,11,12,13.
We collected data on skin problems, including those diagnosed as due to drug reactions, for a cohort of 684 HIV-infected patients. We compared the rates of visits for skin conditions by HIV-infected patients with those by other patients and determined the drug-specific incidence, severity, and outcome of drug-associated adverse cutaneous reactions.
Methods
The Harvard Community Health Plan, the largest staff-model health maintenance organization in New England, maintained computerized ambulatory care and pharmacy records for approximately 265,000 members during the study period. These records encompassed essentially all ambulatory diagnoses made, treatments and tests given, and prescriptions dispensed in the ambulatory care departments, and included both coded information and providers' full-text comments.
We reviewed the complete ambulatory care records of all patients who had been given a coded diagnosis of HIV infection or disease, AIDS, or AIDS-related complex or who had undergone HIV testing. To protect patient confidentiality, individual identifiers were deleted from the records we reviewed. By reviewing all ambulatory care records, we determined various demographic and health characteristics, including age, sex, race or ethnic group, risk group for HIV infection, and history. For the 34 months of the study (April 1, 1988, through January 15, 1991), one of us also reviewed all available hospital records (516 of 518 hospitalizations for all reasons) for these patients at the Brigham and Women's Hospital, their principal site of hospitalization. We considered the date of HIV infection to be the date of the first positive test result for HIV (enzyme-linked immunosorbent assay confirmed by Western blotting). AIDS was considered to be present after the first occurrence of a coded diagnosis of AIDS or an illness defining AIDS (according to 1987 CDC criteria)2.
On the basis of each patient's dates of entry and exit from the health plan and dates of diagnosis of HIV infection and AIDS, we calculated the number of person-years of observation for three periods: from the date of enrollment in the plan or April 1, 1988, whichever was later, to the serologic diagnosis of HIV infection; from the serologic diagnosis of HIV infection to the diagnosis of AIDS; and from the diagnosis of AIDS to the end of the study. We considered an incident event to be the first diagnosis of a given condition for an individual patient at any time during the study. Our calculation of the incidence for each of the two later periods of our study (after the diagnosis of HIV infection and after the diagnosis of AIDS) excluded patients given the same diagnosis in an earlier period and truncated the accumulation of all person-time after the first visit for each diagnosis.
We designed coding forms on which to record the most likely diagnosis for all skin findings, diagnostic codes, and symptoms obtained from ambulatory care and hospital records. For each patient, we determined the total number of visits to health care providers and to dermatologists for each skin condition, as well as the date of the first and last encounter for that condition.
We also compared the proportion of HIV-infected men, ranging in age from 20 to 49 years, who made one or more ambulatory care visits for each dermatologic diagnosis during each year of our study (e.g., April 1, 1988, through March 31, 1989) with the standardized proportion of men in the same age group who were enrolled at Harvard Community Health Plan but not known to be HIV-infected who made such visits. To calculate the standardized rates, we first determined the number of these men who had made one or more outpatient visits for each condition under study during fiscal year 1988. From this value, we subtracted the number of such visits attributable to 20- to 49-year-old men known to be HIV-infected. On the basis of the age distribution of HIV-infected men, we calculated age-standardized rates for men who were not infected with HIV14. We then calculated the rate ratios, which compare the visit rate for each skin diagnosis made per year for HIV-infected men with the standardized rate for non-HIV-infected men for each stage of HIV infection and overall15. In calculating the visit rates for HIV-infected patients, we included both the computer-coded diagnoses (which were made for about 75 percent of all outpatient diagnoses) and the diagnoses we made on the basis of our review of the entire ambulatory care record (about 25 percent). Rates for non-HIV-infected persons were calculated on the basis of computer-coded diagnoses only.
Most plan members can obtain a month's supply of a prescription drug for $3 from the Harvard Community Health Plan pharmacies; all such prescriptions are computerized. The Brigham and Women's Hospital also maintains an automated registry of drugs dispensed to inpatients by the pharmacy. From the prescription records for our cohort for the period April 1, 1988, through September 30, 1990, we calculated the number of persons exposed to specific drugs. A drug course was defined as any first-time dispensing of a drug to a patient or the redispensing of that drug to a patient at least 30 days after the most recent prescription of that drug. We calculated the number of drug courses for all drugs suspected of causing at least two skin reactions and for other drugs that were used frequently in our population or had been noted as causes of cutaneous drug reactions in HIV-infected patients3.
For each cutaneous event suspected by us or by the original health care provider to represent an adverse reaction to a drug, we determined the provider's diagnosis and whether the provider attributed the event to a drug. On the basis of established criteria for evaluating potential adverse drug reactions, we assessed the compatibility of the clinical description with the provider's assessment as unable to be calculated due to insufficient data, possible, probable, or extremely probable16. We also graded the severity of the reaction as ungradable because of insufficient data, mild, moderate, severe, life-threatening, or fatal. We defined a moderately severe reaction as one characterized by systemic symptoms other than pruritus or by a generalized or extensive rash. Reactions that included bullous lesions or resulted in hospitalization or prolongation of hospitalization were considered severe.
Results
Study Population
We identified 684 members of Harvard Community Health Plan (0.3 percent) over the age of five years who were HIV-positive at any time between April 1, 1988, and January 15, 1991. For 367 of these patients (54 percent), HIV infection was first documented during the study. Seventy-four persons had AIDS before the study began. AIDS developed in an additional 168 persons during the study.
Table 1 summarizes the characteristics of our cohort. Our record review included 1593 person-years of observation: 445 person-years before the serologic diagnosis of HIV, 799 person-years after the documentation of seroconversion but before the diagnosis of AIDS, and 349 person-years after the diagnosis of AIDS. All drug-specific incidence data were confined to reactions occurring in the 666 patients who were seropositive before October 1, 1990, and during the 1377 person-years of follow-up from April 1, 1988, through September 30, 1990, the period for which prescription data were available.
|
During the study period, 540 of the 684 patients (79 percent) were given one or more skin diagnoses. A total of 2281 separate skin diagnoses were documented. Table 2 provides the rates of first visits at which the most common and more serious skin problems were diagnosed during each of the three stages of disease: before the documentation of HIV, after seroconversion but before the diagnosis of AIDS, and after the diagnosis of AIDS. The average number of new diagnoses, excluding reactions to drugs, was 1.0, 1.2, and 1.9 per person per year, respectively, for the three stages. On average, there were two visits per person for each dermatologic condition. Table 3 lists the overall rate ratios of visits per year for selected skin conditions as well as the rate ratios for HIV-infected men 20 to 49 years of age according to each stage of disease.
|
|
Cutaneous Drug Reactions
From April 1, 1988, through September 30, 1990, we identified a total of 188 cutaneous drug reactions in 125 of 666 patients. Eighty-seven patients (70 percent) had one reaction, 21 had two, 10 had three, 6 had four, and 1 had five. No patient had multiple reactions to the same drug. A history of drug reactions before April 1, 1988, was noted for 119 patients (18 percent), 30 of whom (25 percent) also had a reaction during the study.
Twenty-two reactions (12 percent) were first noted during hospitalization, and 166 (88 percent) were first noted in the outpatient record. We were able to make independent assessments of the type of reaction involved in 69 of 188 eruptions. Morbilliform eruptions predominated (n = 51, 74 percent). Other diagnoses included 12 cases of urticaria, 2 cases of pruritus, and 1 case each of vasculitis, erythema multiforme major, exfoliative erythroderma, and photodermatitis.
Of 103 reactions with sufficient data for an assessment of severity, 5 were rated as life-threatening, 11 as severe, 51 as moderate, and 36 as mild. There were insufficient data to rate the remaining 85 reactions, most of which we presume were mild, given the absence of any record that substantial care was provided. It was clearly stated in the records that treatment with the drug or drugs suspected of causing the reaction was discontinued in 141 cases (75 percent). Altogether, 13 reactions led to hospitalization, which lasted an average of seven days.
Of 29,711 prescriptions dispensed to 634 of 666 patients (95 percent), 19,667 were dispensed to outpatients and 10,044 to inpatients. The provider and record reviewer indicated that a single drug was by far the most likely cause of 161 reactions. Table 4 provides the incidence rates for prescription drugs implicated in at least two reactions.
|
Discussion
Our patients with HIV infection frequently sought care for common infectious and inflammatory skin diseases, and patients with AIDS made 15 times as many visits for these problems as did other members of this health maintenance organization. The large number of visits per affected person during HIV infection suggests the persistence of these problems, a lack of response to treatment, and substantial morbidity. The incidence of newly recorded skin problems increased as the patients' HIV disease progressed. Skin conditions accounted for 8 percent of hospitalizations in our cohort -- a far higher percentage than in the general population17.
Our data may underestimate to some degree the true prevalence of some problems, including chronic infection with herpes simplex virus and eosinophilic pustular folliculitis, because of the use of descriptive or imprecise diagnostic terms in the records (e.g., ulcer, abscess, and folliculitis). Although the majority of patients were evaluated by a dermatologist, many of the diagnoses given did not conform to one of the classic subtypes of eczematous dermatitis. Precise diagnoses of such dermatoses may be difficult in HIV-infected persons. Most infectious skin conditions require drug therapy, often with agents posing a substantial risk of a cutaneous reaction. In contrast to the very high visit rates noted for our cohort for most infectious and inflammatory skin diseases, visit rates for many common skin conditions such as acne and nevi, whose development is unlikely to be related to HIV infection, were comparable in our cohort and controls.
Our data are limited to the ambulatory care medical records of Harvard Community Health Plan and the hospitalization records of plan members at Brigham and Women's Hospital. Some patients also were cared for under study protocols not administered by the plan or were hospitalized elsewhere. Their records were not available to us; however, if such patients had had substantial skin disease, they would have been likely to return to the plan for dermatologic care. Our data probably underestimate the rates of less serious skin conditions in this group of patients, who also had many other medical problems.
Adverse cutaneous drug reactions occur far more often in HIV-infected persons than in the general population18. These reactions usually result in the discontinuation of treatment with the drugs suspected of causing the reactions, making it more difficult to provide optimal drug therapy. As in the general population, a morbilliform rash was the most common type of drug-related eruption, and urticaria the next most common type in HIV-infected patients19. We detected only one case of erythema multiforme major attributed to drugs. Although our data are consistent with the previously suggested increase in the risk of serious skin conditions with HIV infection, given the large number of exposures to drugs that we documented, our data indicate that such reactions are quite infrequent8,20,21,22,23. In calculating the incidence rates for drug reactions, we relied on our computerized prescription records, which include only drugs dispensed by the pharmacies of Harvard Community Health Plan and Brigham and Women's Hospital. The number of exposures to nonprescription and relatively inexpensive prescription drugs was most likely to be underreported.
Our data confirm the high incidence of eruptions caused by trimethoprim-sulfamethoxazole9,10,11,12,24,25. Forty-three percent of the patients with AIDS in our cohort had an adverse cutaneous reaction to this drug combination either before or during the study period. Reactions to aminopenicillins were about as frequent as those to sulfonamides; in patients with AIDS, the incidence of reactions to both these drugs was about 10 times the rate in the general population18. Although in the general population infection with the Epstein-Barr virus or cytomegalovirus is associated with an increased risk of reaction to aminopenicillins, this was not observed in our HIV-infected patients26.
Our finding of comparable incidence rates before serologic documentation of HIV infection and after serologic diagnosis, but before the diagnosis of AIDS, must be interpreted cautiously. Patients may have been infected with HIV for months or years before serologic testing, and the development of cutaneous disease may have prompted HIV testing, with a resultant ascertainment bias for the period before HIV infection was detected. As HIV infection progressed to AIDS, the visit rates increased for more than two thirds of the most common infectious and inflammatory cutaneous disorders and for cutaneous drug reactions. Unfortunately, we lacked access to patients' CD4 counts for all but the last six months of our study. Continued surveillance of this and other cohorts of HIV-infected persons, together with information on CD4 counts and other measures of immune status, may determine which skin problems or combinations of skin problems are the most useful clinical predictors of the presence or progression of HIV infection.
Dr. Coopman was a fellow at the Beth Israel Hospital, Boston, supported in part by the U.S. Pharmacopeial Convention and the Belgian American Educational Foundation.
We are indebted to the Joint Pharmacoepidemiology Research Program of the Harvard Community Health Plan and Brigham and Women's Hospital for use of the Automated Prescription Database.
Source Information
From the Department of Dermatology, Beth Israel Hospital (S.A.C., R.S.S.); Harvard Community Health Plan (R.A.J.); Channing Laboratory Department of Medicine, Brigham and Women's Hospital (R.P.), and Harvard Medical School (S.A.C., R.A.J., R.P., R.S.S.) -- all in Boston.
Address reprint requests to Dr. Stern at the Department of Dermatology, Harvard Medical School, Beth Israel Hospital, 330 Brookline Ave., Boston, MA 02215.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Cutaneous Disease and Drug Reactions in HIV Infection
Johnson J. R., Krop L. C., Johnson D. P., Stern R. S., Platt R., Coopman S.
Extract |
Full Text
N Engl J Med 1993;
329:1582-1583, Nov 18, 1993.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
