FPIN Journal Club

DIAGNOSTIC STUDY WORKSHEET

1. What question did the study attempt to answer? (for the NEXT study)

Patients – pregnant women carrying singletons

Intervention – cell free DNA screening (maternal blood testing)

Comparison – first-trimester screening

Outcome – rate of detection of trisomy 21

Did the study address an appropriate and clearly focused question Yes No

2. Determining Relevance:

a. Is the diagnostic test feasible and common to your practice? Yes No

This was a question that evolved almost as fast as we could publish. When we first read about this test (February 2014), no one in the room was familiar with cell-free DNA testing; when we read the second article about it (April 2015) it was increasingly common practice.

b. Is the proportion of patients with the target illness comparable

to the patient group seen by family physicians? Yes No

c. Did the authors study a clinically meaningful Yes No

and/or a patient oriented outcome?

3. Determining Validity:

a. What test is being evaluated

Cell free DNA testing on maternal serum

b. What is the reference standard with which the test being evaluated is compared?

They compared it to first trimester screening (nuchal transluceny ultrasound + pregnancy-associated plasma protein A and HCG levels done at 10-14 weeks gestational age); the gold standard for diagnosis for both tests was genetic testing

c. The test and reference standard are measured

independently (blind) of each other? Yes No Unclear

d. Did the patient sample include an appropriate

spectrum of patients to whom the diagnostic test will be

applied in clinical practice? Yes No Unclear

e. Patients for testing are selected either as a consecutive

series or randomly, from a clearly defined study Yes No Unclear

f. Results are reported for all patients that are entered

into the study Yes No Unclear

About 7% lost to follow up; another 9% were excluded for having twins, not having one or both samples, or other reasons

4. What are the results?

a. What is the estimated sensitivity of the test being evaluated? (state 95% CI)

Sensitivity = proportion of results in patients with the disease that are correctly identified by the new test

(We will calculate these below)

CF DNA: 38 cases of suspected trisomy 21, 38 confirmed, sensitivity=100%

95%CI: 90.7-100

First trimester screening: 30 cases identified, 38 confirmed cases, sensitivity=78.9%

95% CI: 62.7-90.4

b. What is the estimated specificity of the test being evaluated (state 95% CI)

Specificity = proportion of results in patients without the disease that are correctly identified by the new test

CF DNA:

99.9% (95%CI: 99.9–100)

First trimester screening:

94.6% (95% CI 94.2–94.9)

c. What are the likelihood ratios for the test being evaluated?

LR+ = sens / (1-spec)
- LR+ > 10 indicates a large change in likelihood, < 2 indicates no change in likelihood

CF DNA: LR+: 1700 (rounded)

First trimester screening LR+: 14.6

LR− = (1-sens) / spec
- LR− < 0.1 indicates a large change in likelihood, > 0.5 indicates no change in likelihood

CF DNA: LR-: 0

CF DNA: LR-: 0.22

5. Applying the evidence

a. Will the results help me in caring for my patients? Yes No

b. If the findings are valid and relevant, will this change

your current practice? Yes No

c. Is the change in practice something that can be done in

a medical care setting of a family physician? Yes No

d. Can the results be implemented? Yes No

e. Are there any barrier to immediate implementation? Yes No

f. How was this study funded?

The CF DNA company + the Perinatal Quality Foundation

2x2 table

2x2 tables, and sensitivity and specificity, are the hallmark of epidemiology and biostats—but they are actually pretty rare in the medical literature. This is in large part because they only arise from diagnostic cohort studies—and most of the literature we read is randomized controlled trials and meta-analyses.

Today we have a diagnostic cohort study, and we can make a 2x2 table and calculate sensitivity and specificity.

Sensitivity and specificity are called ‘test characteristics’, and as the name suggests, they are inherent to the test. Testing more people will not yield different results, and they are not dependent on the prevalence of disease in the population. Positive and negative predictive value, on the other hand, do change as the prevalence of disease changes.

Back to the calculations:

This online calculator can help with some of the math:

http://ktclearinghouse.ca/cebm/practise/ca/calculators/statscalc

Or you can just draw everything out and practice calculating sensitivity and specificity by hand (it will be on the boards). Use the data in table 2 to draw 2 2x2 tables, one for standard screening and one for CF DNA.

OUTCOME OF INTEREST (trisomy)

+ -

38 / 9
0 / 15,794

CF DNA

OUTCOME OF INTEREST (trisomy)

+ -

30 / 854
8 / 14,949

Standard screening

Before you calculate anything, just eyeballing these tables, you can see where the interesting bit is going to be: that upper right corner, the false positives.

Calculate sensitivity: (true positive)/(true positive + false negative)

CFDNA: 38/(38+0)=100

Standard screening: 30/(30+8)=78.9. Just as it says in the article! Love it when that happens.

Calculate specificity: (true negative)/(true negative + false positive)

CF DNA: 15,794/(15,794+9)= 99.9

Standard screening: 14,949/(14,949+854)=94.6

Hey! We got that one right, too.