Module 3 – Screening

Introduction

Screening is the identification and treatment of asymptomatic persons who have already developed subclinical (unrecognized) disease by use of tests, examinations, or procedures. Crucial to any screening program are the accuracy of the screening test and the effectiveness of early disease detection.

Test Accuracy

Sensitivity and specificity are intrinsic properties of a screening test that help determine its accuracy. Test sensitivity indicates the proportion of persons with a disease who correctly test positive when screened. A test with low sensitivity generates many false negative results. That is, persons with subclinical disease may be missed and medical treatment will be delayed. Test specificity indicates the proportion of persons without a disease who correctly test negative when screened. A test with low specificity generates many false positive results. That is, healthy persons will be told that they may have a disease. This leads to additional testing and procedures that incur unnecessary risk, distress and expense.

The positive predictive value [PPV] of a test is the proportion of patients with positive test results that truly have disease. Unlike sensitivity and specificity, PPV is not an intrinsic property of a test. It is dependent upon the prevalence of the disease in the screened population. As disease prevalence in a population increases, the PPV of a screening test in that population will also increase. Likewise, as disease prevalence decreases, the PPV of a screening test will also decrease. For example, consider a screening test with 90% sensitivity and 90% specificity. If the population being screened has a disease prevalence of 1%, then the PPV of a positive test result is 8.3%. This means that 92% of people who test positive do not have the disease. If the same test is used in a screening population with disease prevalence of 0.1%, then the PPV of a positive test result is 0.9%. This means that 99% of people who test positive do not have the disease. (More on these concepts can be found in chapter 7 of Glaser’s book, High-Yield Biostatistics. They will also be discussed in lecture)

For a screening program to be effective, a clinical intervention that can prevent or delay progression of disease must be available. Furthermore, early detection and treatment must offer some benefit (such as improved survival rate) over conventional diagnosis and treatment when the disease is clinically evident. In evaluating the effectiveness of a screening program, one must consider several types of bias that could affect the reported results. Lead-time bias may result in the appearance of improved survival with screening. Survival will appear to be improved even if screening only identifies cancer earlier. It can lengthen the period between diagnosis and death, without truly prolonging life expectancy. Length bias may result in the identification of a disproportionate number of slowly progressive cancers but miss aggressive cases that are present only briefly. If slowly progressive cases do not generally become clinically significant diseases, then the screening program may not provide a benefit.

When considering the benefits of screening, one must ask about the effect of a program on an entire population (population benefit). If the disease in question is of low prevalence or morbidity, then even a highly sensitive screening test will have little impact on the population as a whole. Likewise, screening with even modestly sensitive tests may have dramatic impact on the total population if the disease is of high prevalence or morbidity.

Finally, the effectiveness of any a screening program should also take into consideration the potential adverse effects of screening. There may be physical harm from the test itself, from misdiagnosis, and from treatment for non-aggressive disease. Cost-effectiveness must also be considered. From a population standpoint, the benefits must be weighed against the costs of implementation (especially since those resources could have been used for other interventions). ‘Years of life saved’ and ‘quality-adjusted life years’ are some of the measures used in a cost-benefit analysis.

Quality of the Evidence about the Effectiveness of Screening

Randomized controlled trials (and well constructed meta-analyses of such trials) provide the highest level of evidence. The random assignment of volunteers should result in the equal distribution of potential confounding variables (known or unknown) among the intervention and control groups. These are prospective studies and thus require large sample sizes and years of observation.

A cohort study provides the next best level of evidence. These observational trials examine the outcome of individuals who received an intervention and those who did not. They are subject to multiple confounding variables, which must be accounted for. These studies also require large sample size and years of observation.

A case-control study is retrospective. Persons known to have disease are compared to a control group without disease. Data about exposure to the intervention is sought retrospectively. It is more difficult to control for confounding variables in these studies. Furthermore, case control studies may be subject to both observer bias and recall bias.

1. Define screening. Understand the role it should play in primary care practice.
    
2. Define and be able to utilize the following terms when reading a study about a screening test: sensitivity, specificity, positive predictive value, cutoff points.
    
3. Recognize the characteristics of diseases that are suitable for screening and the characteristics of a good screening test.
    
4. Understand the most recent US Preventive Services Task Force (USPSTF) and American Cancer Society (ACS) recommendations for breast, colon, prostate and cervical cancer screening.
    
5. Understand the most recent USPSTF recommendations on screening for lipid disorders, osteoporosis, hypertension, obesity, diabetes and coronary artery disease.
    

Instructions for the Small Group Forum

 Note: bring your laptop computer to the first small group session in this module. You will need to review certain websites and begin a literature search. A librarian will join your group to discuss appropriate techniques for searching the literature related to your presentation.

1. Chapter 7: “Statistics in Medical Decision Making” in Glaser, A. High Yield Biostatistics, 3 rd Edition. Lippincott Williams and Wilkins. 2005
    
2. Current Methods of the U.S. Preventive Services Task Force: A Review of the Process http://www.ahrq.gov/clinic/ajpmsuppl/harris1.htm (In particular, pay attention to the sections on ‘Scope and Selection of Topics’, ‘Review of the Evidence’, and ‘Translating Evidence into Recommendations’.
    
3. USPSTF (summary and clinical considerations), NCI and ACS recommendations for breast cancer screening*
    
4. USPSTF (summary and clinical considerations), NCI and ACS recommendations for colon cancer screening*
    
5. USPSTF (summary and clinical considerations), NCI and ACS recommendations for cervical cancer screening*
    
6. USPSTF (summary and clinical considerations), NCI and ACS recommendations for prostate cancer screening*
    
7. USPSTF recommendations for hypertension screening (summary and clinical considerations)*
    
8. USPSTF recommendations for lipid disorders screening (summary and clinical considerations)*

* The USPSTF recommendations can be found from their home page ( http://www.ahrq.gov/clinic/uspstfix.htm); click on the clinical category and find the individual reports. The NCI recommendations can be found from their PDQ website (http://www.cancer.gov/cancertopics/pdq); go to screening/detection summaries. The ACS recommendations can be found from their prevention and early detection website (http://www.cancer.org/docroot/PED/ped_0.asp); go to early detection and follow the links to their guidelines.