SESSION IV. STRATEGIES TO INCREASE EARLY CLINICAL RECOGNITION
Applying Genetics and Public Health Strategies to Primary Immunodeficiency Diseases
November 8-9, 2001 ~ Atlanta, Georgia
Prepared by: Office of Genomics and Disease Prevention, Centers for Disease Control and Prevention Department of Health and Human Services
Speakers discussed the development and assessment of early recognition systems and initial laboratory tests, considered which PI diseases could benefit from and should be targeted for early recognition, and discussed the components of early recognition instruments.
The conventional approach to assessment of the value of early identification through screening is the criterion-based approach. This approach evaluates a screening test based on criteria such as public health importance, suitability of the test, effectiveness of treatment, and cost of case finding. Each criteria is assessed separately, with no algorithm to weight the relative importance of different criteria. In contrast, the prevention effectiveness approach uses economic evaluation and decision-analytic techniques to assess the overall balance of benefits, harms, and costs of prevention interventions. This approach uses a single metric that can be used to rank different interventions in terms of cost-effectiveness or cost-benefit. The difference between the two approaches can be illustrated by applying them to a case study of screening strategies for two hypothetical disorders:
Applying the criterion-based approach, disorder A appears to be the more important public health priority: it is more prevalent and causes more deaths. However, disorder B has a more accurate screening test, and its treatment is more efficacious; the proportion reduction in mortality is 80 percent for B and only 50 percent for A. When economic burden is expressed as cost per person screened, testing for A is twice as expensive as for B, $10 vs. $5 per person. When expressed as cost per case detected, however, the balance shifts: $125,000 to detect a case of A compared with $500,000 to detect a case of B. In this example, the criterion-based approach can yield conflicting conclusions, depending on which criteria are given greatest weight. Cost per case detected does not measure cost-effectiveness.
A limitation to this type of analysis is that it only compares two interventions; it says nothing about whether either, or both, should be funded. One alternative is a cost-utility analysis, which converts outcomes to quality-adjusted life-years (QALYs). QALYs combine two components, additional years of life and improvement in life quality resulting from reduced morbidity or disability. A commonly used but arbitrary cost-effectiveness threshold is $50,000 per QALY saved. Another approach is a cost-benefit analysis, which converts health outcomes into a dollar value based on how much society values or is willing to pay for the outcome. This conversion of morbidity and mortality into dollars requires methods for placing a value on various states of being. The most commonly accepted value for willingness to pay to avoid a premature death in the United States is in the range of $3 million to $7 million per death. Beyond this simple case study, a real-world economic analysis would require consideration of additional health effects (e.g., morbidity, disability, adverse effects of false-positive results) and costs (e.g., data management, sample collection, averted costs of care).
The prevention effectiveness approach has several advantages. It lacks the subjectivity of the criterion-based methodology, it clarifies sources of uncertainty, and it can demonstrate the cost-effectiveness of early identification despite very prevalence and high treatment costs.
Early clinical recognition is in the hands of primary care practitioners. All patients with PI diseases need someone on the front lines to recognize the possibility of an immunologic problem and the need for appropriate evaluation. The role of the primary care practitioner is to provide the first line of care in the form of early clinical recognition and appropriate initial evaluation. Whatever system is used to identify patients with PI diseases must therefore accommodate the needs and fit into the clinical practices of primary care physicians (pediatricians, family practitioners, internal medicine practitioners). Physicians cannot afford to overstress the system by identifying too many potential cases, but they also do not want to miss cases.
A barrier to implementing symptom-based screening is awareness of PI diseases among primary care practitioners. Increased awareness begins with an understanding of what is abnormal in the patient population. Infectious diseases represent a large part of any pediatric practice. In the flood of normal childhood infections, how does a primary care practitioner identify children who fall outside the normal range? Practitioners are interested in helping, but they need a tool that is right for them. Any early recognition instrument must place PI diseases on physicians’ radar screens, increase their awareness of the frequency, duration, and types of disorders, and enhance their appreciation of the value of a genetic family history in early recognition.
One approach is to concentrate on categories and clues. Categories of PI diseases include: local barrier deficiencies, complement deficiencies, neutrophil deficiencies, cell-mediated deficiencies, and humoral deficiencies. Clues to the presence of PI diseases can be revealed through a family or patient history. For example, clues to complement deficiencies include encapsulated organisms and recurrent gram-negative organisms; clues to neutrophil defects are abscesses and recurrent staphylococcal infections; clues to cell-mediated deficiencies are intracellular organisms and eczema; clues to humoral deficiencies are noninvasive mucosal infections and increased frequency and duration of infections.
Once a concern is identified, the primary care practitioner can play an important role in initial evaluation of the patient with a possible PI disease. Although an immediate referral to an appropriate subspecialist is a reasonable approach, the work-up can often be initiated by the pediatrician or internist. Many managed-care organizations now mandate this approach, which with guidance (e.g., algorithms, use of categories, and clues) can be efficient and cost effective. Proposed methods should be simple and not burdensome. The instrument should include recommended/available initial steps for work-up (to meet the demands of managed care) and ensure access to expert consultants.
Additional barriers to early recognition of PI diseases are the use of multiple providers, different settings of care (office, emergency room, clinics), and the unavailability or incompleteness of the medical record. Increased use of centralized databases (e.g., office, clinic, hospital, managed care, pharmacies, related care facilities) with built-in red flags, increased awareness of treatment options, and continued advancement of the molecular basis of primary immunodeficiency can accelerate the prompt detection and treatment of PI diseases and prevent permanent damage or life-threatening complications.
Early diagnosis is clearly beneficial in the majority of the life-threatening immunodeficiencies. Fischer et al. noted that 2-year survival after bone marrow transplantation for SCID was 70 percent in patients who received transplants before 6 months of age and 45 percent in those who received transplants after 6 months of age (Lancet 1990;336:850-854). Specialists who provide care for patients with antibody deficiencies, CGD, or hyper-IgE syndrome are aware that patients who have had pneumonias multiple times before diagnosis are likely to develop progressive, debilitating pulmonary disease. XLA and CVID are often associated with chronic pulmonary disease. In a study examining 22 patients with hypogammaglobulinemia who had been followed for a mean of 10 years, the predictor of poor pulmonary function was the presence of pulmonary disease at the initial visit (Sweinberg, et al. J Allergy Clin Immunol 1991;88:96-104).
Few data shed light on whether primary immunodeficiency is being diagnosed in a timely fashion. Nor do physicians agree about what constitutes a timely diagnosis. For most physicians, a diagnosis that is made before long-lasting sequelae have developed would be considered a timely diagnosis. For most patients and their families, a diagnosis that is made months or years after the onset of symptoms would not constitute a timely diagnosis. The issue also can be examined from a cost-benefit perspective. Screening for PI is expensive. Therefore, we need to find ways to identify patients who should be screened and to focus on the screening tests that are the least expensive and the most informative.
To examine these issues, Dr. Conley and colleagues reviewed the records of 60 patients with sporadic XLA who were referred for genetic analysis. To reflect current practices in North America, the study was limited to patients born in the United States or Canada in whom XLA was diagnosed after January 1, 1990. Approximately 20 percent of the patients were recognized to have immunodeficiency at 12 months of age or less. All of these patients were hospitalized at the time of diagnosis, and most presented dramatically with overwhelming infection. For many of these patients, the overwhelming infection was the first sign of their disease.
XLA was diagnosed in approximately half of the patients at 13 to 40 months of age. In this group of patients, three boys were recognized to have immunodeficiency because of chronic otitis media or outpatient pneumonia. Slightly more than half of these toddler patients were recognized to have XLA during or after their first hospitalization for infection, and the remaining patients were hospitalized two, three, or four times before immunodeficiency was diagnosed.
The remaining 30 percent of the patients, who were recognized to have immunodeficiency at >40 months of age, were not necessarily sicker than the patients whose illness was diagnosed when they were infants or toddlers. Half were recognized to have immunodeficiency during or after their first hospitalization for infection. Almost all of the patients who were more than 12 months old at diagnosis had a history of chronic otitis media or sinusitis.
Because patients with XLA have barely detectable or absent lymph nodes and tonsils, physicians who considered the diagnosis of XLA in patients with chronic otitis media could theoretically screen patients by physical exam and make the diagnosis before a significant infection.
In summary, although most patients with XLA have a timely diagnosis from a physician’s point of view, the diagnosis is often markedly delayed from the patient’s point of view because the disease is diagnosed during or after the first hospitalization for infection,. A clue from the physical exam (absence of tonsils) could help identify patients who are appropriate for screening for immunodeficiency, but this clue does not appear to be commonly used.
The disproportionate number of white patients diagnosed with and treated for PI diseases has led researchers to question why these disorders are not recognized with the same frequency among minority and economically disadvantaged persons. A study at Mount Sinai School of Medicine is investigating the causes of and solutions to missed diagnoses of PI diseases. Mount Sinai serves an area that is predominantly Hispanic (50 percent) and black (30 percent), but the patient population in the hospital’s immunology clinic is predominantly white (92 percent). The theory behind the study is that PI diseases are under-diagnosed in persons from ethnic minorities and economically disadvantaged communities. Possible reasons for the disparity are: lack of regular contact with a primary physician; receipt of medical care in emergency departments and clinics where physicians rotate schedules; and treatment for obvious symptoms, such as ear or sinus infections, without testing for possible underlying causes of recurring illnesses.
STRIDE (Study Targeting Recognition of Immune Deficiency and Its Evaluation) is based on the hypothesis that persons with PI diseases in large patient populations might be identified on the basis of characteristic disease manifestations and ICD codes. A goal is to develop standards to help health care providers identify persons with PI diseases and to equalize the diagnostic work-up available to different racial and ethnic groups.
One phase of the project involved compiling a disease-code scoring system for recognizing immunodeficiencies. By examining ICD codes commonly related to PI diseases, Dr. Cunningham-Rundles and colleagues developed profiles of the combinations of diagnostic codes that indicate probable PI. They reviewed 5 years of inpatient and outpatient billing and discharge data using the code-based scoring system to identify patients with histories of recurrent infections or other findings who might merit investigation. Findings showed that 0.3 percent of encounters in the outpatient department had scores suggestive of PI. The cases were heavily weighted in 1- to 10-year-olds,
with most coming from the pediatric emergency department, general pediatrics clinic, and Eye, Nose, and Throat (ENT) clinic. Most patients lived in the neighborhood, and the racial breakdown reflected the demographics of the area. For the inpatient department, 0.1 percent of patients (337) had scores suggestive of PI. Of these, about half were non-white, and most lived in the neighborhood.
Investigators are also working with community clinics and hospitals in New York City to identify patients whose records show more than one of the relevant disease codes and to offer the patients clinical testing for PI disease. The study team is developing educational materials for physicians serving minority populations to increase awareness and improve diagnoses.
Dr. Cunningham-Rundles concluded that many patients with PI diseases are never referred for work-up. Reasons are complex and include lack of a medical home and use of multiple providers. A scoring system based on the diseases of a large patient population is able to identify patients with an immune defect. The evaluation of this system showed that: (1) inpatient ICD coding is more reliable than outpatient coding, (2) a scoring system will not identify patients who have not been followed long enough to acquire complications, (3) a scoring system will be contaminated by conditions such as autoimmunity and organ failure, and (4) codes that designate persons with immunodeficiency are used indiscriminately.
Dr. Hong began by stating the need to quantify the extent of the problem. Are PI diseases under-diagnosed or misdiagnosed, and, if so, what is the magnitude of the problem? He suggested that disease registries are the most informative source for this information.
Once a problem of sufficient magnitude is determined, the next step is to devise a strategy that will be effective in the primary-care setting. In contrast to subspecialty clinicians, primary-care physicians deal with countless different issues and lack the luxury of confining their diagnostic approaches to a circumscribed group of disorders. Nonetheless, the approaches to disease diagnosis and management are generally designed by subspecialists and reflect a narrow viewpoint. A specialist who sees a patient has the benefit of prescreening. If the patient does not fall within the specialist’s area of expertise, the patient is referred. Textbook approaches to PI disease can suffer from this parochialism and thus may not be useful to busy clinicians in their offices or at the bedside.
PI diseases can be likened to needles in the haystack of the symptoms and signs of the thousands of patients seen by the average primary care practitioner. These needles would be more readily revealed if most of the haystack could be removed. This requires somehow setting apart the suspects from the rest of the patients so they form a smaller group. A helpful aide to primary care physicians would be a series of red lights to trigger alarms or actions at the appropriate time, i.e., a way to indicate when a patient is unlike 99 percent of others in the office. For a system of alerts to be useful to busy clinicians, it must be sufficiently sensitive (few or no missed cases), specific, and easily retrieved so as not to overburden the physician or the system. Clinicians must also be directly involved in the development and testing of the system.
Multiple approaches for early recognition exist (e.g., 10 warning signs; Conley approach), but none have been validated.
|Consider a diagnosis of:||In the presence of:|
|Systemic illness or untoward result after viral infection or live virus vaccine; opportunistic infection; persistent lymphopenia; etc.|
|Recurrent infection with extracellular pyogens; etc.|
|Symptoms of both from above; features of WAS; etc|
The approach then takes the clinician through a decision tree based on laboratory findings, culminating with either referral or reassurance. These and other approaches will be discussed in the breakout session on early recognition strategies.
Discussion–A brief discussion yielded these additional points:
- Increased use of antibiotics may be masking the presentation of PI disorders and contributing to late diagnoses.
- Two early recognition approaches have been discussed: (1) screening of large populations and (2) increasing provider awareness. Perhaps a two-pronged approach is indicated.
- Data are needed on the incidence of PI diseases. Rather than continuing to base decisions on anecdotal information, the group should agree on a data collection approach and initiate it right away.
- Screening of cord blood is a promising approach. One possibility is to collaborate with the Red Cross, which is banking cord blood as transplant material.