<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.1 20151215//EN" "https://jats.nlm.nih.gov/publishing/1.1/JATS-journalpublishing1.dtd">
<article xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" article-type="other" dtd-version="1.0" specific-use="sps-1.6" xml:lang="en">
  <front>
    <journal-meta>
      <journal-id journal-id-type="nlm-ta">J Bras Pneumol</journal-id>
      <journal-id journal-id-type="publisher-id">jbpneu</journal-id>
      <journal-title-group>
        <journal-title>Jornal Brasileiro de Pneumologia</journal-title>
        <abbrev-journal-title abbrev-type="publisher">J. bras. pneumol.</abbrev-journal-title>
      </journal-title-group>
      <issn pub-type="ppub">1806-3713</issn>
      <issn pub-type="epub">1806-3756</issn>
      <publisher>
        <publisher-name>Sociedade Brasileira de Pneumologia e Tisiologia</publisher-name>
      </publisher>
    </journal-meta>
    <article-meta>
      <article-id pub-id-type="publisher-id" specific-use="scielo-v3">nHVQM5B33TpNSSJSGDCvm5J</article-id>
      <article-id pub-id-type="publisher-id" specific-use="scielo-v2">S1806-37132018000100004</article-id>
      <article-id pub-id-type="doi">10.1590/S1806-37562018000000017</article-id>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>CONTINUING EDUCATION: SCIENTIFIC METHODOLOGY</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Understanding diagnostic tests. Part 3.</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0001-6548-1384</contrib-id>
          <name>
            <surname>Ferreira</surname>
            <given-names>Juliana Carvalho</given-names>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff2">
            <sup>2</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0001-5742-2157</contrib-id>
          <name>
            <surname>Patino</surname>
            <given-names>Cecilia Maria</given-names>
          </name>
          <xref ref-type="aff" rid="aff1">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff3">
            <sup>3</sup>
          </xref>
        </contrib>
      </contrib-group>
      <aff id="aff1">
        <label>1</label>
        <institution content-type="original">. Methods in Epidemiologic, Clinical, and Operations Research-MECOR-program, American Thoracic Society/Asociaci&#243;n Latinoamericana del T&#243;rax, Montevideo, Uruguay.</institution>
        <institution content-type="orgdiv1">Methods in Epidemiologic, Clinical, and Operations Research-MECOR-program</institution>
        <institution content-type="orgname">American Thoracic Society/Asociaci&#243;n Latinoamericana del T&#243;rax</institution>
        <addr-line>
          <named-content content-type="city">Montevideo</named-content>
        </addr-line>
        <country country="UY">Uruguay</country>
      </aff>
      <aff id="aff2">
        <label>2</label>
        <institution content-type="original">. Divis&#227;o de Pneumologia, Instituto do Cora&#231;&#227;o, Hospital das Cl&#237;nicas, Faculdade de Medicina, Universidade de S&#227;o Paulo, S&#227;o Paulo (SP) Brasil.</institution>
        <institution content-type="normalized">Universidade de S&#227;o Paulo</institution>
        <institution content-type="orgdiv2">Hospital das Cl&#237;nicas</institution>
        <institution content-type="orgdiv1">Faculdade de Medicina</institution>
        <institution content-type="orgname">Universidade de S&#227;o Paulo</institution>
        <addr-line>
          <named-content content-type="city">S&#227;o Paulo</named-content>
          <named-content content-type="state">SP</named-content>
        </addr-line>
        <country country="BR">Brazil</country>
      </aff>
      <aff id="aff3">
        <label>3</label>
        <institution content-type="original">. Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles (CA) USA.</institution>
        <institution content-type="normalized">University of Southern California</institution>
        <institution content-type="orgdiv2">Department of Preventive Medicine</institution>
        <institution content-type="orgdiv1">Keck School of Medicine</institution>
        <institution content-type="orgname">University of Southern California</institution>
        <addr-line>
          <named-content content-type="city">Los Angeles</named-content>
          <named-content content-type="state">CA</named-content>
        </addr-line>
        <country country="US">USA</country>
      </aff>
      <pub-date pub-type="epub-ppub">
        <season>Jan-Feb</season>
        <year>2018</year>
      </pub-date>
      <volume>44</volume>
      <issue>01</issue>
      <fpage>04</fpage>
      <lpage>04</lpage>
      <permissions>
        <license license-type="open-access" xlink:href="https://creativecommons.org/licenses/by-nc/4.0/" xml:lang="en">
          <license-p>This is an open-access article distributed under the terms of the Creative Commons Attribution License</license-p>
        </license>
      </permissions>
      <counts>
        <fig-count count="2"/>
        <table-count count="0"/>
        <equation-count count="0"/>
        <ref-count count="3"/>
        <page-count count="1"/>
      </counts>
    </article-meta>
  </front>
  <body>
    <p>In the previous articles from this series<xref ref-type="bibr" rid="B1"><sup>1</sup></xref><sup>,</sup><xref ref-type="bibr" rid="B2"><sup>2</sup></xref> we discussed important characteristics used in order to evaluate diagnostic tests: sensitivity, specificity, positive predictive value, and negative predictive value. In this final part, we discuss positive likelihood ratio (LR+), negative likelihood ratio (LR&#8722;), and ROC curves.</p>
    <sec>
      <title>LIKELIHOOD RATIOS</title>
      <p>LRs combine sensitivity and specificity to quantify how helpful a new diagnostic test is in changing (increasing or decreasing) the probability of having a disease compared with the prevalence of that disease (pretest probability) in the population studied. The LR+ of a test is the probability of a positive result in patients with the disease divided by the probability of a positive result in patients without the disease, whereas LR&#8722; is the probability of a negative result in patients with the disease divided by the probability of a negative result in patients without the disease. LR+ ranges from 1 to infinity, and an LR+ of 1 indicates that the probability of a positive test result is the same for patients with and without the disease; therefore, the test is useless. An LR+ greater than 1 supports the presence of the disease, and the greater LR+ is, the more a positive test result increases the probability of the disease when compared with the pretest probability. LR&#8722; ranges from 1 to 0, and the closer the LR is to 0, the lower the probability of the disease is if the test result is negative.</p>
    </sec>
    <sec>
      <title>ROC CURVES</title>
      <p>We use ROC curves to make a global assessment of the value of a diagnostic test by calculating the area under the curve (AUC). The values of the AUC can vary from 0 to 1.0, and values over 0.8 indicate that the diagnostic test has very good accuracy. The ROC curve plots sensitivity (true positives) against &#8220;1 &#8722; specificity&#8221; (false negatives) for all the possible cut-off values of the new test (<xref ref-type="fig" rid="f1">Figure 1</xref>). As we have previously discussed, there is always a trade-off between sensitivity and specificity when we define a cut-off value for quantitative test results. If a new test were perfect, there would be a complete separation of values between patients with and without the disease, the cut-off value would be the lowest value among patients with disease, and the AUC would be 1. However, since there are no perfect tests, there will always be some false positive or some false negative results. The more accurate a test is, the greater the AUC is, which is the probability that a random person with the disease has a higher value of the measurement than a random person without the disease.<xref ref-type="bibr" rid="B3"><sup>3</sup></xref>
			</p>
      <p>
        <fig id="f1">
          <label>Figure 1</label>
          <caption>
            <title>ROC curve plotting sensitivity vs. &#8220;1 &#8722; specificity&#8221; for two different tests. Both tests have good accuracy; however, test 1 (closed circles) has an area under the curve (AUC) of 0.946 and test 2 has an AUC of 0.832 (open circles), meaning that test 1 has overall better accuracy to discriminate between patients with and without the disease. This figure was created with fictitious data.</title>
          </caption>
          <graphic xlink:href="1806-3756-jbpneu-44-01-0004-Cvm5J-gf01.jpg"/></fig>
      </p>
    </sec>
    <sec>
      <title>MAKING SENSE OF DIAGNOSTIC TEST PERFORMANCE CHARACTERISTICS </title>
      <p>If you are wondering which of the parameters described is more useful to evaluate a diagnostic test-sensitivity, specificity, LRs, or ROC curve-the answer is: it depends! Each parameter describes a specific characteristic of the test, and depending on how you will use the test, one or another may be more useful. Now that you understand these concepts, interpreting a test result will be much more than just looking at the result.</p>
    </sec>
  </body>
  <back>
    <ref-list>
      <title>References</title>
      <ref id="B1">
        <label>1</label>
        <mixed-citation>1 Ferreira JC, Patino CM. Understanding diagnostic tests. Part 1. J Bras Pneumol. 2017;43(5):330. <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.1590/s1806-37562017000000330">https://doi.org/10.1590/s1806-37562017000000330</ext-link>
				</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Ferreira</surname>
              <given-names>JC</given-names>
            </name>
            <name>
              <surname>Patino</surname>
              <given-names>CM</given-names>
            </name>
          </person-group>
          <article-title>Understanding diagnostic tests Part 1</article-title>
          <source>J Bras Pneumol</source>
          <year>2017</year>
          <volume>43</volume>
          <issue>5</issue>
          <fpage>330</fpage>
          <lpage>330</lpage>
          <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.1590/s1806-37562017000000330">https://doi.org/10.1590/s1806-37562017000000330</ext-link>
        </element-citation>
      </ref>
      <ref id="B2">
        <label>2</label>
        <mixed-citation>2 Patino CM, Ferreira JC. Understanding diagnostic tests. Part 2. J Bras Pneumol. 2017;43(6):408. <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.1590/s1806-37562017000000424">https://doi.org/10.1590/s1806-37562017000000424</ext-link>
				</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Patino</surname>
              <given-names>CM</given-names>
            </name>
            <name>
              <surname>Ferreira</surname>
              <given-names>JC</given-names>
            </name>
          </person-group>
          <article-title>Understanding diagnostic tests Part 2</article-title>
          <source>J Bras Pneumol</source>
          <year>2017</year>
          <volume>43</volume>
          <issue>6</issue>
          <fpage>408</fpage>
          <lpage>408</lpage>
          <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.1590/s1806-37562017000000424">https://doi.org/10.1590/s1806-37562017000000424</ext-link>
        </element-citation>
      </ref>
      <ref id="B3">
        <label>3</label>
        <mixed-citation>3 Altman DG, Bland JM. Diagnostic tests 3: receiver operating characteristic plots. BMJ. 1994;309(6948):188. <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.1136/bmj.309.6948.188">https://doi.org/10.1136/bmj.309.6948.188</ext-link>
				</mixed-citation>
        <element-citation publication-type="journal">
          <person-group person-group-type="author">
            <name>
              <surname>Altman</surname>
              <given-names>DG</given-names>
            </name>
            <name>
              <surname>Bland</surname>
              <given-names>JM</given-names>
            </name>
          </person-group>
          <article-title>Diagnostic tests 3 receiver operating characteristic plots</article-title>
          <source>BMJ</source>
          <year>1994</year>
          <volume>309</volume>
          <issue>6948</issue>
          <fpage>188</fpage>
          <lpage>188</lpage>
          <ext-link ext-link-type="uri" xlink:href="https://doi.org/10.1136/bmj.309.6948.188">https://doi.org/10.1136/bmj.309.6948.188</ext-link>
        </element-citation>
      </ref>
    </ref-list>
  </back>
  <sub-article article-type="translation" id="s1" xml:lang="pt">
    <front-stub>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>EDUCA&#199;&#195;O CONTINUADA: METODOLOGIA CIENT&#205;FICA</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Entendendo os testes diagn&#243;sticos. Parte 3.</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0001-6548-1384</contrib-id>
          <name>
            <surname>Ferreira</surname>
            <given-names>Juliana Carvalho</given-names>
          </name>
          <xref ref-type="aff" rid="aff01">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff02">
            <sup>2</sup>
          </xref>
        </contrib>
        <contrib contrib-type="author">
          <contrib-id contrib-id-type="orcid">0000-0001-5742-2157</contrib-id>
          <name>
            <surname>Patino</surname>
            <given-names>Cecilia Maria</given-names>
          </name>
          <xref ref-type="aff" rid="aff01">
            <sup>1</sup>
          </xref>
          <xref ref-type="aff" rid="aff03">
            <sup>3</sup>
          </xref>
        </contrib>
      </contrib-group>
      <aff id="aff01">
        <label>1</label>
        <institution content-type="original">. Methods in Epidemiologic, Clinical, and Operations Research-MECOR-program, American Thoracic Society/Asociaci&#243;n Latinoamericana del T&#243;rax, Montevideo, Uruguay. </institution>
      </aff>
      <aff id="aff02">
        <label>2</label>
        <institution content-type="original">. Divis&#227;o de Pneumologia, Instituto do Cora&#231;&#227;o, Hospital das Cl&#237;nicas, Faculdade de Medicina, Universidade de S&#227;o Paulo, S&#227;o Paulo (SP) Brasil. </institution>
      </aff>
      <aff id="aff03">
        <label>3</label>
        <institution content-type="original">. Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles (CA) USA.</institution>
      </aff>
    </front-stub>
    <body>
      <p>Nos artigos anteriores desta s&#233;rie,<xref ref-type="bibr" rid="B1"><sup>1</sup></xref><sup>,</sup><xref ref-type="bibr" rid="B2"><sup>2</sup></xref> discutimos caracter&#237;sticas importantes usadas para avaliar os testes diagn&#243;sticos: sensibilidade, especificidade, valor preditivo positivo e valor preditivo negativo. Nesta &#250;ltima parte, discutiremos raz&#227;o de verossimilhan&#231;a positiva (RV+), raz&#227;o de verossimilhan&#231;a negativa (RV&#8722;) e curvas ROC. </p>
      <sec>
        <title>RAZ&#213;ES DE VEROSSIMILHAN&#199;A</title>
        <p>As RV combinam sensibilidade e especificidade para quantificar o qu&#227;o &#250;til um novo teste diagn&#243;stico &#233; para mudar (aumentar ou diminuir) a probabilidade de ter uma doen&#231;a em compara&#231;&#227;o com a preval&#234;ncia dessa doen&#231;a (probabilidade pr&#233;-teste) na popula&#231;&#227;o estudada. A RV+ &#233; a probabilidade de um resultado positivo em pacientes com a doen&#231;a dividida pela probabilidade de um resultado positivo em pacientes sem a doen&#231;a, ao passo que a RV&#8722; &#233; a probabilidade de um resultado negativo em pacientes com a doen&#231;a dividida pela probabilidade de um resultado negativo em pacientes sem a doen&#231;a. A RV+ varia de 1 a infinito, e uma RV+ igual a 1 indica que a probabilidade de resultado positivo do teste &#233; a mesma para pacientes com e sem a doen&#231;a; portanto, o teste &#233; in&#250;til. Uma RV+ maior que 1 corrobora a presen&#231;a da doen&#231;a; quanto maior a RV+, maior ser&#225; a probabilidade de que o resultado positivo do teste aumente a probabilidade de doen&#231;a se o resultado do teste for positivo. A RV&#8722; varia de 1 a 0, e quanto mais pr&#243;xima de 0 a RV for, menor ser&#225; a probabilidade de doen&#231;a na presen&#231;a de resultado negativo do teste. </p>
      </sec>
      <sec>
        <title>CURVAS ROC</title>
        <p>Usamos curvas ROC para fazer uma avalia&#231;&#227;o global do valor de um teste diagn&#243;stico por meio do c&#225;lculo da &#225;rea sob a curva (ASC). Os valores da ASC podem variar de 0 a 1,0; valores &gt; 0,8 indicam que a precis&#227;o do teste diagn&#243;stico &#233; muito boa. A curva ROC &#233; uma representa&#231;&#227;o gr&#225;fica da sensibilidade (verdadeiro-positivos) contra &#8220;1 &#8722; especificidade&#8221; (falso-negativos) para todos os valores de corte poss&#237;veis de um novo teste (<xref ref-type="fig" rid="f01">Figura 1</xref>). Como discutimos na parte 1, h&#225; sempre uma troca entre sensibilidade e especificidade quando definimos um valor de corte para resultados de testes quantitativos. Se um novo teste fosse perfeito, haveria uma separa&#231;&#227;o completa de valores entre pacientes com e sem a doen&#231;a, o valor de corte seria o menor valor em pacientes com a doen&#231;a e a ASC seria = 1. No entanto, como n&#227;o h&#225; testes perfeitos, sempre haver&#225; resultados falso-positivos ou falso-negativos. Quanto mais preciso for um teste, maior ser&#225; a ASC, que &#233; a probabilidade de que o valor da medida seja maior em uma pessoa qualquer com a doen&#231;a do que em uma pessoa qualquer sem a doen&#231;a.<xref ref-type="bibr" rid="B3"><sup>3</sup></xref>
				</p>
        <p>
          <fig id="f01">
            <label>Figura 1</label>
            <caption>
              <title>Curva ROC da sensibilidade contra &#8220;1 &#8722; especificidade&#8221; de dois testes diferentes. Ambos t&#234;m boa precis&#227;o; entretanto, o teste 1 (c&#237;rculos fechados) apresenta &#225;rea sob a curva (ASC) = 0,946 e o teste 2 apresenta ASC = 0,832 (c&#237;rculos abertos), o que significa que o teste 1 &#233; de modo geral mais preciso para discriminar entre pacientes com e sem doen&#231;a. Esta figura foi criada com dados fict&#237;cios. </title>
            </caption>
            <graphic xlink:href="1806-3756-jbpneu-44-01-0004-Cvm5J-gf01-pt.jpg"/></fig>
        </p>
      </sec>
      <sec>
        <title>ENTENDENDO AS CARACTER&#205;STICAS DO DESEMPENHO DOS TESTES DIAGN&#211;STICOS</title>
        <p>Se voc&#234; est&#225; se perguntando qual dos par&#226;metros descritos &#233; mais &#250;til para avaliar um teste diagn&#243;stico - sensibilidade, especificidade, RVs ou curva ROC - a resposta &#233;: depende! Cada par&#226;metro descreve uma caracter&#237;stica espec&#237;fica do teste, e a utilidade de cada par&#226;metro depender&#225; de como voc&#234; usar&#225; o teste. Agora que voc&#234; entende esses conceitos, interpretar o resultado de um teste ser&#225; muito mais do que apenas ver o resultado. </p>
      </sec>
    </body>
  </sub-article>
</article>
