Retinal Physician Article Submission Guidelines-Prescribing for Astigmatism and Presbyopia

CLASSIFIEDS

Pre-owned equipment, practices for sale, open positions, helpful practice management resources and more!

Click here to view the latest classifieds from Contact Lens Spectrum.

Article Date: 9/1/2008

Print Friendly Page
Contact Lens Study Design and Analysis
CONTACT LENS STUDIES

Contact Lens Study Design and Analysis

Knowledge of study methods and designs can help you analyze the latest contact lens research.

By Eric R. Ritchey, OD, MS


Dr. Ritchey is a 2001 graduate of The Ohio State University College of Optometry and completed a two-year Cornea and Contact Lens Fellowship in 2003. He is currently a senior research associate at The Ohio State University College of Optometry and practices in the Cornea and Contact Lens Service.

With the development of new contact lens products, it's easy for busy practitioners to become overwhelmed by the amount of new information available. As quickly as each new contact lens reaches the market, studies appear that examine the performance characteristics for each new product.

Given this abundance of information, savvy contact lens practitioners need to be able to quickly access the most recent research available and to assess the quality of that information so that they can provide patients with the best contact lens products.

Don't Ignore the Study Methods

Given the time demands placed on busy practitioners, it's easy to progress directly to the conclusions when given a study about a contact lens product. While the temptation to go directly to the findings is powerful, you can obtain more information by thinking about the methods a researcher used to obtain his conclusions. When reviewing an article, the most important question you should ask is, "What is the central question the study tries to answer?" This seems simple enough, and if you want to obtain maximum benefit from an article, it's worth the time to seriously consider the main study question. Specifically, you should ask whether the study answers a question that is relevant to the proposed topic and does it provide new information on the subject?

After you have identified the study question, you should closely look at the study methods to determine whether the researchers could answer the questions they pose in the study. A key to determining whether a study can effectively answer the intended questions is to critically examine the summary statistics of the study sample. The features of the study sample are often described in the study methods or in the results as summary statistics.

Certain parameters of the summary statistics are of universal interest to readers such as the number of subjects enrolled, the number of subjects completing the study, the male-to-female subject ratio, the average and range of subject age with standard deviation, and percentage of subjects identified with a specific ethnic group. This information provides the most basic picture of who the subjects are in a particular study and how generalizable the study is to the population at large.

Identifying the type of subjects enrolled in a study is important because one goal for a contact lens study is to have a study population that is as diverse as possible while remaining appropriate to answer the study question. This allows the study to provide information that is generalizable to all contact lens patients who may utilize a particular product. This is critical in contact lens research because these products are designed to be marketed in a global marketplace.

For example, if you are looking at a study examining a contact lens material designed to treat dry eye, the study population may be heavily populated by female subjects for a number of reasons, such as the prevalence of dry eye or the prevalence of contact lens wearers older than age 40 in females versus males. However, if the study is too heavily inclusive of one subject demographic, the generalizability of the study's conclusions becomes narrowed to that sample. By looking at the subject demographics, you can determine whether a study provides information that can change the way you treat your patients.

Next, it's important to look at the other study parameters that are key to the conclusions proposed by the authors. Studies examining the performance of a contact lens will often provide statistics on the average subject manifest refractive error, the average best-corrected high- and low-contrast visual acuity or the average corneal keratometry values. Studies examining contact lens materials and solutions may focus on grading type and quantity of corneal staining or may evaluate the subjective responses of patients using questionnaire data.

Once you identify the study parameters central to the study question, examine the methods used to collect and analyze the data. Are the researchers using objective methods of data collection, such as cycloplegic auto-refraction for refractive error, or are they using subjective techniques such as manifest refraction? Are they taking single measurements or repeated measurements over time? Identifying the type of data and how it was acquired will provide an idea of the study design and of the statistical tests the researchers should use in data analysis.

Examining Study Design

The study design is a key piece of information when determining whether a study can answer its proposed question(s). Study designs fall under four general categories:

  1. Randomized clinical trial
  2. Cohort design
  3. Cross-sectional design
  4. Case-control design

Randomized clinical trials, cohort studies and case-control studies are commonly used in vision research. The conclusions an investigator can make in a study are directly related to the study design.

The gold standard of clinical research is the double-masked, randomized, controlled clinical trial, as it is considered experimental. In this design, the study subjects and the study examiners do not know which patients have received the investigational treatment and which have received the control, or "sham," treatment. By preventing both examiner and patient from having any knowledge of the respective treatment group, you eliminate subject and investigator bias in favor of the investigational treatment versus the control. Randomized clinical trials are commonly used to assess the safety and effectiveness of ophthalmic pharmaceuticals or in contact lens material or solution studies in which researchers randomize study subjects into an experimental solution treatment or a control product.

A classic example of a double-blind, randomized clinical trial in the ophthalmic literature is the Age-Related Eye Disease Study (AREDS), which examined the use of nutritional supplements and the progression of age-related macular degeneration and cataract. AREDS enrolled thousands of patients who were randomized into treatment or placebo groups and examined by masked examiners over time.

In contact lens research, there are instances in which it's not possible to use a double-blind design. An example of a randomized clinical trial in which the subject was not masked to the assigned treatment group in contact lens research is The Contact Lens and Myopia Progression (CLAMP) Study. The goal of the CLAMP study was to determine the effects of GP contact lenses and soft contact lenses on myopia progression in children. The CLAMP study enrolled children between ages 8 and 11 who were randomized into a GP or a soft contact lens treatment group. Because the study required the subjects to wear a contact lens, it was not possible to mask the patients to their assigned treatment groups. The CLAMP study was single-blind, with a masked clinical examiner performing the testing procedures. To ensure that the examiner remained masked, the subjects did not wear contact lenses in the presence of the masked examiner and were instructed not to discuss their contact lens treatment assignment. By masking the examiner to the treatment group, the study could avoid investigator bias and still answer the research question.

The other three study designs are observational designs in which an investigator traditionally does not apply a treatment. Cohort studies feature a longitudinal design in which the investigator recruits a group of subjects who have a specific condition and then follows the subjects over time. An example of a cohort design in optometry is The Collaborative Longitudinal Evaluation of Keratoconus (CLEK) study, a multi-center study previously funded by the National Eye Institute, which enrolled more than 1,200 patients who have keratoconus and followed them over eight years with a study goal of observing the progression of the condition.

Cross-sectional design studies are similar to cohort design studies, with the primary difference being that the subjects are measured at only one point in time. Because the subjects are not followed over time, the cross-sectional design does not provide information on disease progression or the development of new cases of the condition among at-risk subjects, known as disease incidence. Cross-sectional designs can determine the number of subjects who have the condition at a given point in time, known as disease prevalence.

The case-control design is another study design used in contact lens research. Case-control studies examine two groups: one with a treatment or risk factor of interest and another without the treatment or risk at one specific point in time. This design is often implemented in retrospective studies where researchers review cases in which patients develop an adverse event, such as contact lens-related microbial keratitis, and compare them to controls with matched parameters (sex, age, etc.) to identify risk factors associated with the development of the condition in question.

It should be noted that there are times when investigators wish to examine a research question, but there is not enough previously published background research available for the researchers to design a full-scale clinical trial. This often occurs when researchers need data to determine the sample size required for the proposed question. Before committing time and resources to a study that may not have enough subjects to answer the question, researchers often decide to perform a pilot study. Pilot studies are an inexpensive way for researchers to quickly gather information required to determine whether a larger study is feasible. Pilot studies typically feature a small number of subjects observed over a short period of time. In one example in the contact lens literature, Berntsen et al (2006) published a study looking at the effect of Corneal Refractive Therapy (CRT) on refractive error-specific quality of life determined using a questionnaire developed by the National Eye Institute (NEI). Because there was minimal research published on the use of the questionnaire with CRT patients, Berntsen et al used a small sample of subjects to determine the number of subjects that would be required to perform a large-scale clinical trial. Pilot studies can provide interesting information, but you must remember the goal of the study and that the limited number of subjects can limit its statistical power and the generalizability of the results.

Do Statistics Really Matter?

It's very easy for people to become a little cynical regarding statistics. Some people believe that statistics can be manipulated to provide a desired result. The truth is that statistical tests are tools and, as with any tool, the statistical tests selected for a study should match the job required. In general, statistical tests can be broken down into two main categories: parametric and non-parametric statistics. Parametric statistics are commonly utilized in large clinical trials and include tests such as the student t-test and analysis of variance (ANOVA). In studies in which the sample size is limited, such as pilot studies, or when a study deals with data values that are not a continuous measure, such as grading scales for corneal staining, non-parametric statistics are typically used. Most parametric statistical tests have a non-parametric equivalent. Table 1 shows a summary of common parametric tests, their non-parametric equivalents and their use.

Parametric tests are typically used when the study data is normally distributed. An example of normally distributed data would be the average height of males in a population. If you took a random sample of males, some would be extremely short or tall, while most would be around the average height. When graphed, the data would be symmetrically distributed around the average height, creating a classic bell-shaped curve distribution. These data would be normally distributed. In general, larger sample sizes are more likely to be normally distributed. Therefore, when analyzing the data presented in a study, you should make sure to look at the total number of subjects used in the data analysis. If the sample size for a study is large, more than likely the data will be normally distributed and parametric tests are appropriate. If the sample size is small, non-parametric tests may be more appropriate for data analysis. Researchers can use non-parametric tests for data that meet the requirements of parametric tests; however, because non-parametric tests have less statistical power than their parametric equivalents, most researchers choose to use the parametric tests in these situations.

The References Are Important

Today's research is built upon previously published research. By looking at the references cited in an article, you can find further information that can shape your practice. Furthermore, the techniques used in a study have often been established in previous research, and examining these papers will help you decide whether the research presented answers the research question.

Fortunately, tools are available to search for studies that may be of interest regarding contact lenses. Perhaps the most powerful such tool available today is the Internet database PubMed (www.pubmed.gov). PubMed is a database of biomedical journal citations and abstracts that is available free to the public from The National Library of Medicine. Administered by the National Institutes of Health, PubMed accesses MedLine, the National Library of Medicine's biomedicine bibliographic database, which catalogues most major biomedical and life sciences journals from the 1950s to present day. There you can look up a variety of cornea and contact lens topics from journals that are of interest to contact lens practitioners such as Ophthalmology, Cornea, Investigative Ophthalmology and Vision Science, Contact Lens and Anterior Eye, Ophthalmic and Physiological Optics, Eye and Contact Lens (CLAO journal) and Optometry and Vision Science. PubMed allows you to view abstracts for most articles and has free online access to some journals. Using PubMed, you can easily compile a list of articles about a contact lens topic and use these resources to help shape your practice.

The Most Important Question

The last question you should ask yourself is, "Is the focus of the study clinically significant to my practice?" The only person who can decide this is you. Hopefully, with the steps listed above, you can quickly assess whether the next article that crosses your desk will change the way you practice and affect people's lives. CLS

To obtain references for this article, please visit http://www.clspectrum.com/references.asp and click on document #154.



Contact Lens Spectrum, Issue: September 2008

Table of Contents Archives



AWS-#2