Last month we explored the application and interpretation of various measures for the signs of ocular disease: the scales we use to rate redness, lid swelling or corneal staining. An analogous methodology is also applied to quantify symptoms of ocular disease such as pain, itch or other forms of discomfort. Because we rely on our patients to report the intensity, character and progression of disease symptoms, it’s crucial that we take great care in how we pose that deceptively simple question, “How are you feeling?”
This month, we delve into the nuances of symptomatic assessment. We’ll discuss both the importance of establishing an approach that allows us to best employ the information our patients provide for diagnosis and treatment assessment, and the role this information plays in developing new therapies. Issues surrounding the optimization of symptom evaluation are critical in developing dry-eye treatments, and they’re also of interest in therapies for allergy, ocular inflammation and postoperative ocular pain.
Dry-eye disease is an ocular condition defined by patients’ symptoms: The name of the condition is itself a symptomatic description. We’re aware that the central confounder in DED is that symptoms and signs are often discordant, and patients often report significant symptomatic disease with little or no ocular surface staining or tear-film dysfunction.
Dry-eye questionnaires such as the National Eye Institute visual function questionnaire, the Women’s Health Study questionnaire and the Dry Eye Epidemiology Project’s questionnaire are tools designed to assess the scope of ocular disease, that were later used to evaluate symptoms. In general, these were limited by either having too many questions or too few dry-eye related queries. Another group of these symptomatic measures, including the Ocular Surface Disease Index, the McMonnies Dry Eye Index and the Dry Eye Questionnaire, focused more on the key issues of DED symptom characteristics, symptom severity and disease impact. The McMonnies and the OSDI were characterized in greater detail and validated for use in clinical trials.
What makes a good questionnaire? The goal is to allow the patient to provide accurate information about his symptoms in a way that precisely and reproducibly reflects the disease state. Most questions are formatted so that the patient responds with a scalar numeric; for example, a question from the OSDI asks:
“Have you experienced the following in the last week: painful or sore eyes (4) all of the time (3) most of the time (2) half of the time (1) some of the time (0) none of the time.”
Structuring questions in this way, where higher numerical scores are associated with increased disease severity, allows questionnaires with 10 to 15 questions like the OSDI to generate a summated score in which ranges of total scores are designated as mild, moderate or severe symptomatic disease.
Several other factors are critical to a reliable, reproducible questionnaire that can provide an accurate DED evaluation. First, the number of questions should be kept to a minimum, allowing patients to maintain a consistent degree of focus in answering each question. Our experience tells us that questionnaires should be no longer than the 12 to 15 questions of the McMonnies or the OSDI; and it may be that four or five well-structured questions are sufficient, or even superior, in terms of the goals stated above. This is particularly true in a clinical trial setting, where a greater number of questions provides more—but not necessarily better—data, and the resulting greater variability in the dataset often confounds the outcome.
Greater numbers of questions in the setting of a clinical trial also provides a greater potential for the Hawthorne Effect, where subjects interject their perceptions regarding right or wrong/better or worse answers in response to a perceived need on the part of those administering the test. This subjective influence can ultimately blur distinctions between treatment and control groups.
A second factor that can have a big impact on symptom questionnaires is the degree to which questions rely on recollection. Questions that ask, “How often in the last week …” or “How many days in the past month …” are dependent upon a subject’s recall and therefore have an inherent potential for inaccuracy. As with longer surveys, those with questions that focus on a historical perspective from the patient’s view will allow for an interpretive response with an increased subjective impact. Recent studies suggest that recall bias is a particularly significant issue in older patient populations, and that recall issues may introduce systematic biases into the responses and, therefore, the symptomatic assessments.
In an ideal world, our symptom questionnaire would have a single question and a correlation with disease severity of 1. As we live in the real world, we strive for a few simple questions and a correlation of 0.7 or better. Further complicating the refinement of symptomatic assessments are the contrasting aspirations of questionnaires designed for the clinic and those optimized for a clinical trial: While an inclusion error is good in the practice of medicine, an exclusion error is better in a drug-development setting.
Several approaches have been used to address the pitfalls of current questionnaires by combining old and new questionnaires or focusing on specific questions with a symptom survey. One study employed the OSDI questionnaire in combination with a simpler, four-question survey of subjects’ current dry-eye symptomatology (Ora Calibra Four-Symptom Scale).
When we talk about disease symptoms, we refer in almost every case to some type of pain or discomfort: dryness; grittiness; burning; itching; and even photophobia. Many of these are associated with conditions other than dry eye, and so should be part of the diagnosis for those conditions as well. Eye pain and photophobia, for example, are two hallmark symptoms of migraine headache.
Pain is often a key sequela to procedures such as cataract or refractive surgery, and in these cases the most reliable measure is obtained with a visual analog scale. Comparisons of numerical scales without verbal descriptors and those with graded descriptors such as mild, moderate or severe show that it is primarily the anchoring descriptors that impact the reliability of the scale.
Use of symptomatic data in clinical trials is always more difficult than endpoints with clearly defined, objective metrics. Despite this, they’re often the most clinically meaningful measures, and so their inclusion is often essential from a regulatory perspective. An approach that is seen with increasing frequency to address this and other trial design issues is the composite endpoint, in which specific treatment goals are combined into one amalgamated efficacy measure.
Often, composite endpoints function to increase the statistical power of a study without increasing the number of subjects required; an example of this is the total nasal symptom score used in trials of allergy therapies. The TNSS combines symptomatic scores for rhinorrhea, nasal congestion, nasal itching and sneezing, and has become the metric for therapeutic assessment of anti-allergics.
Despite these caveats, composite endpoints do have a potential utility, whether for enhancement of our ability to address symptomatic diseases or as a tool in the broader context of clinical therapeutic development. Key to the process of symptom assessment, regardless of the specific measure or metric selected, is the fact that it must be based on a clear and honest conversation between patient and physician.
Dr. Abelson is a clinical professor of ophthalmology at Harvard Medical School. Dr. Hollander is chief medical officer at Ora, and assistant clinical professor of ophthalmology at the Jules Stein Eye Institute at the University of California, Los Angeles. Dr. McLaughlin is a medical writer at Ora Inc. Dr. Abelson may be reached at MarkAbelsonMD@gmail.com.