Original Research

Low-Risk Human Papillomavirus Testing andOther Nonrecommended HumanPapillomavirus Testing Practices Among U.S.Health Care ProvidersJennifer Wai-Yin Lee, RN, BS, Zahava Berkowitz, MSPH, MSc, and Mona Saraiya, MD, MPH

OBJECTIVE: To assess self-reported human papillomavi-rus (HPV) DNA testing practices by health care providersand clinics, including nonrecommended practices such aslow-risk HPV testing, HPV cotesting in women youngerthan age 30 years, and HPV reflex testing for high-gradeabnormal Pap test results.

METHODS: We analyzed responses to a cross-sectionalsurvey of a nationally representative sample of Papani-colaou test providers administered in conjunction withthe 2006 National Ambulatory Medical Care Survey andNational Hospital Ambulatory Medical Care Survey. Dataanalysis was performed on responses from 376 office-based health care providers and 216 outpatient clinics.

RESULTS: Overall, 75.5% (95% confidence interval [CI]68.7–81.2%) of health care providers and 77.2% (95% CI60.3–88.3%) of clinics reported ever using the HPV DNAtest. Of health care providers who used HPV testing, 28.5%

(95% CI 21.6–36.6%) used both high-risk and low-risk HPVtests. Most health care providers (59.6%, 95% CI 48.5–69.7%) and clinics (66.0%, 95% CI 48.0–80.3%) used HPVcotesting in women younger than age 30 years. A highpercentage of health care providers and clinics performedreflex HPV testing after Pap test results of atypical squa-mous cells, cannot exclude high-grade squamous intraepi-thelial lesions (71.4%, 95% CI 63.5–78.3% and 62.8%, 95%CI 49.0–74.9%, respectively) and high-grade squamous in-traepithelial lesions (50.7%, 95% CI 42.4–58.9% and 49.0%,95% CI 33.1–65.2%, respectively), results for which HPVtesting is not recommended.

CONCLUSION: Many health care providers reportedinappropriate uses of HPV testing, which may lead tounnecessary follow-up and increased medical costs with-out added benefits. Interventions such as eliminating thelow-risk HPV test from the U.S. market and educatinghealth care providers and patients on appropriate indi-cations for HPV testing are needed to discourage healthcare providers from such practices.(Obstet Gynecol 2011;118:4–13)DOI: 10.1097/AOG.0b013e3182210034

LEVEL OF EVIDENCE: III

Persistent high-risk human papillomavirus (HPV)infections are recognized as the cause of cervical

cancer.1 A high-risk HPV DNA test to detect infectionwith oncogenic or high-risk HPV types has beenavailable since the early 2000s and is most commonlyused to screen or manage women with abnormal Paptest results.2,3 Recent surveys report that approxi-mately one-third of health care providers use HPVtests as part of routine cervical cancer screening.4,5 Asa screening test, the high-risk HPV test can be used asa cotest along with cervical cytology screening inwomen aged 30 years old and older. The mostcommon use of HPV tests is as a management test,which involves HPV DNA testing after an atypical

See related editorial on page 1.

From the Division of Cancer Prevention and Control, Centers for DiseaseControl and Prevention, Atlanta, Georgia; The CDC Experience AppliedEpidemiology Fellow, Scientific Education and Professional Development Pro-gram Office, and the National Center for Chronic Disease Prevention andHealth Promotion, Centers for Disease Control and Prevention, Atlanta,Georgia.

Jennifer Wai-Yin Lee completed this project during a 1-year fellowship withThe CDC Experience, a public/private partnership supported by a grant tothe CDC foundation from the External Medical Affairs, Pfizer Inc.

The findings and conclusions of this study are those of the authors and do notnecessarily represent the official position of the Centers for Disease Control andPrevention.

Corresponding author: Mona Saraiya, MD, MPH, Division of Cancer Preven-tion and Control, Centers for Disease Control and Prevention, 4770 BufordHighway, Mailstop K-55, Atlanta, GA 30341; e-mail: MSaraiya@cdc.gov.

Financial DisclosureThe authors did not report any potential conflicts of interest.

© 2011 by The American College of Obstetricians and Gynecologists. Publishedby Lippincott Williams & Wilkins.ISSN: 0029-7844/11

4 VOL. 118, NO. 1, JULY 2011 OBSTETRICS & GYNECOLOGY

squamous cells of undetermined significance (ASC-US) Pap test result, a practice also known as reflextesting.4,5 A low-risk HPV DNA test, which detectsfive nononcogenic HPV types including HPV 6 andHPV 11, the most common types found in genitalwarts, is commercially available although there are noclinical indications nor recommendations by any or-ganizations for its use.2,3,6

Few studies have examined health care provideruse of HPV DNA tests, but inconsistencies have beenshown between recommended uses of HPV DNAtests and health care provider-reported practices.4,7

Although the benefits of HPV testing include greatersensitivity in detecting precancerous lesions as well asa high negative predictive value when used with thePap test, use of nonrecommended HPV testing poten-tially increases health care costs and unnecessarywork-up of patients who test positive without addingclinical value. In an effort to reduce nonrecom-mended HPV DNA testing, a statement on recom-mended uses of HPV DNA tests was published in2009, which was endorsed by various professional orga-nizations, including the American Cancer Society andAmerican Society for Colposcopy and Cervical Pathol-ogy, and included an outline of when high-risk HPVDNA testing is generally not recommended (Box 1).The purpose of our study is to assess reported HPVtesting practices among U.S. health care providers whoperform Pap tests—including low-risk HPV testing,which has not been examined previously—using a na-tionally representative survey of office-based health careproviders and hospital outpatient clinics.

MATERIALS AND METHODSWe used data from the 2006 Cervical Cancer Screen-ing Supplement, a self-administered survey commis-sioned by the Centers for Disease Control and Pre-vention, Division of Cancer Prevention and Control.The data were collected as a supplement to theCenters for Disease Control and Prevention, NationalCenter for Health Statistics’ National AmbulatoryMedical Care Survey, and National Hospital Ambu-latory Medical Care Survey to obtain information onphysicians’ practices and beliefs related to cervicalcancer screening. The National Ambulatory MedicalCare Survey collects visits, practice, and health careprovider-level data annually from office-based physi-cians and community health centers. The samplingframe for the National Ambulatory Medical CareSurvey is derived from American Medical Associa-tion and American Osteopathic Association lists of‘‘office-based physicians” providing patient care. Ex-cluded are federally employed physicians or those

specializing in anesthesiology, radiology, or pathology,The sampling frame for the National Ambulatory Med-ical Care Survey community health center health careproviders is developed using data from the HealthResources and Services Administration Bureau of Pri-mary Health Care Uniform Data System and the IndianHealth Service who received a list of physicians andmidlevel health care providers available during a pre-determined 1-week reporting period. In 2006, 3,500physicians and 104 community health centers weresampled. National Center for Health Statistics’ NationalAmbulatory Medical Care Survey collects informationon visits and clinic data from general and short-stayhospitals with emergency or outpatient departments inall 50 states and the District of Columbia excludingFederal hospitals. In 2006, 480 hospitals were sampled.Additional information about both surveys can be foundat www.cdc.gov/nchs/ahcd.htm.

Health care providers eligible for the CervicalCancer Screening Supplement among sampled Na-tional Center for Health Statistics’ National Ambula-

Box 1. Situations in Which HumanPapillomavirus DNA Testing Is NotRecommended*

Testing for Low-Risk HPV Types†

Routine cervical cancer screening in women youngerthan 30 years†

Initial triage of ASC-H, LSIL,‡ HSIL, or AGC§ or AIS†

Adolescents (age 20 years and younger) with any abnor-mal cytologic resultRoutine screening with HPV testing and cervical cytol-ogy more often than every 3 years for women 30 yearsand older whose tests were negative at last screenRepeat high-risk HPV DNA testing in less than 12months in women whose HPV tests were positiveRoutine sexually transmitted disease screeningTo determine who should get the HPV vaccineRoutine sexual assault workup

HPV, human papillomavirus; ASC-H, atypical squamouscells, cannot rule out HSIL; LSIL, low-grade squamousintraepithelial lesion; HSIL, high-grade squamousintraepithelial lesion; AGC, atypical glandular cells; AIS,adenocarcinoma in situ.* Based on guidelines from the American Cancer Society, theAmerican College of Obstetrics and Gynecology, and theAmerican Society for Colposcopy and Cervical Pathology.3,27–30

† Assessed in this study.‡ HPV testing can be used for the triage of LSIL inpostmenopausal women.§ HPV testing should not be used for initial triage of AGCbut can be performed at the time of colposcopy to directmanagement postcolposcopy.

VOL. 118, NO. 1, JULY 2011 Lee et al HPV Testing Practices Among U.S. Providers 5

tory Medical Care Survey health care providers wererestricted to the following specialties: general or familypractice, internal medicine, or obstetrics and gynecol-ogy; midlevel health care providers in communityhealth centers were also eligible. Among National Hos-pital Ambulatory Medical Care Survey clinics, onlythose specializing in general medicine or obstetrics andgynecology were eligible; emergency departments werenot eligible. Health care providers from both surveyswere further restricted to only those performing cervicalcancer screening. Eligible health care providers andclinics had the option of completing the Cervical CancerScreening Supplement survey as a paper survey oronline. A total of 387 National Ambulatory MedicalCare Survey health care providers (response rate 61.1%)and 216 National Hospital Ambulatory Medical CareSurvey clinics (response rate 84.7%) responded. ElevenNational Ambulatory Medical Care Survey respondentswere not in eligible specialties and were excluded fromour analysis. The Research Ethics Review Board of theCenters for Disease Control and Prevention’s NationalCenter for Health Statistics approved the National Am-bulatory Medical Care Survey and National HospitalAmbulatory Medical Care Survey protocols.

The 2006 Cervical Cancer Screening Supplementcontained nine multipart questions that examinedcervical cancer screening practices, including screen-ing methods; whether the practice performed colpos-copy; HPV DNA test use, including types of HPVDNA tests used, HPV testing as a screening test, andHPV testing for management of abnormal Pap testresults; and screening intervals. Health care providerdemographics and practice characteristics collectedfrom the National Ambulatory Medical Care Surveyand used in our study included the health careprovider’s age, sex, specialty, board certification sta-tus, practice size, practice type, geographic region,and participation in the National Breast and CervicalCancer Early Detection Program. The variables fromthe National Hospital Ambulatory Medical Care Sur-vey used in our analysis included clinic characteristicssuch as clinic specialty and geographic region but nothealth care provider-specific demographics, whichare not collected.

Four items assessed HPV testing practices, begin-ning with whether the practice ever ordered or collectedthe HPV DNA test. Health care providers who re-sponded yes were asked to select the different types ofHPV DNA tests ordered or collected in their practicefrom the following choices: “high-risk HPV DNA test,”“low-risk HPV DNA test,” “not aware there was ahigh-risk or low-risk HPV DNA test,” or “unknown.”Responses were categorized first into responses that

indicated the total amount of low-risk HPV use, but inbivariate analyses, responses were categorized into mu-tually exclusive practices of high-risk HPV DNA test useonly, both high-risk and low-risk HPV DNA test use,low-risk HPV DNA test use only, and not aware orunknown. HPV testing for management of an abnormalPap test result was assessed in two items: “If the patient’sscreening Papanicolaou test result is borderline or ab-normal, does your practice routinely order an HPVDNA test to be performed on that sample (commonlycalled reflex HPV DNA testing)?” and “Does yourpractice routinely recall patients to come back for asecond sample for an HPV DNA test if their screeningPapanicolaou test is abnormal or borderline (recalltesting)?” Health care providers who routinely per-formed either reflex or recall testing were asked forwhich types of abnormal Pap test results would the HPVDNA test be ordered: ASC-US, atypical squamous cells,cannot exclude high-grade squamous intraepithelial le-sions (ASC-H), low-grade squamous intraepithelial le-sions (LSIL), high-grade squamous intraepithelial lesions(HSIL), or atypical glandular cells. Responses for reflexand recall HPV testing for management of abnormalPap test results were combined and are referred to asreflex testing in the remainder of the article. HPVcotesting was assessed by a two-part item: “Does yourpractice routinely order or collect an HPV DNA test atthe same time as the Papanicolaou test as part of routinecervical cancer screening (sometimes called adjunctHPV testing or cotesting)?” Health care providers whoresponded yes were asked, “For which patients doesyour practice routinely order or collect an HPV DNAtest along with the Papanicolaou test?” Response op-tions were nonexclusive and included: “women youngerthan 30 years old,” “women 30 years old and older,”“women who request the test for cervical cancer screen-ing,” “women who request the test to check their HPVinfection status,” or “other—specify [fill-in].”

We used guidelines from American Cancer Soci-ety, the American College of Obstetricians and Gy-necologists, and the American Society for Colpos-copy and Cervical Pathology as the basis forinappropriate HPV testing practices, as illustrated inBox 1.2,3,8 Only the high-risk HPV test, which tests foroncogenic HPV types, is recommended for clinicaluse. For our analysis, we defined the following testingpractices, which were consistently not recommended(or actively discouraged) in any guidelines, as nonrec-ommended HPV testing: low-risk HPV testing; HPVcotesting in women younger than age 30 years; HPVtesting after Pap test results of ASC-H; and HPV testingafter Pap test results of HSIL. Although we asked aboutHPV testing after other Pap test results such as atypical

6 Lee et al HPV Testing Practices Among U.S. Providers OBSTETRICS & GYNECOLOGY

glandular cells and LSIL, we did not include them asinappropriate practices because there are special situa-tions in which HPV testing is considered acceptablewithin the guidelines.

We used �2 statistics to compare various HPVtesting practices by health care provider or clinicspecialty for each sample of health care providerswho responded to the Cervical Cancer ScreeningSupplement. Sampling weights provided by the Na-tional Center for Health Statistics were used to adjustfor nonresponse and to provide nationally represen-tative patterns of practice. We considered an estimateunstable if the relative standard error, calculated as(standard error/estimated percentage)�100, wasmore than 30% and, thus, should be interpretedcautiously. All statistical analyses were done usingSAS 9.2 with SUDAAN 10.

RESULTSThe distribution of demographic and practice charac-teristics of both health care providers and clinics isshown in Table 1. Among National AmbulatoryMedical Care Survey respondents, most were physi-cians (96.7%) and were more likely to be male,board-certified, practice in a solo or a single specialtypractice, and work in a private setting. The majorityof National Hospital Ambulatory Medical Care Sur-vey clinics were general medicine clinics (76.0%).

Overall, 75.5% (95% confidence interval [CI]68.7–81.2%) of health care providers and 77.2% (95%CI 60.3–88.3%) of clinics reported ever ordering orcollecting the HPV DNA test. Obstetricians–gyne-cologists were significantly more likely to report usingthe HPV DNA test than family physicians and inter-nists (98.9% compared with 76.8% and 44.7%, respec-tively) (P�.001). Of health care providers who col-lected the HPV test, 46.6% (95% CI 38.2–55.2%) usedit with the Pap test for routine cervical cancer screen-ing (HPV cotesting), whereas 88.5% (95% CI 83.3–92.3%) used it for managing an abnormal Pap testresult (reflex or recall testing). Similarly, clinics weremore likely to perform reflex or recall HPV testingthan HPV cotesting (Table 2). There were differencesin the use of HPV cotesting by geographic region ofthe country with the highest reported in the Northeastregion (72.6%, 95% CI 59.1–82.9%) and the lowest inthe Midwest (35.9%, 95% CI 23.1–51.1%).

Overall, 31.4% of office health care providers and25% of hospital clinics reported using low-risk HPVtesting with variation by specialty regardless ofwhether it was in an outpatient setting (obstetricians–gynecologists: 41.5%; midlevel: 36.7%; family physi-cians: 25.6%; internal medicine: 23.9% [the estimate

Table 1. Office-Based Health Care Provider andClinic Characteristics

Characteristic

NAMCSOffice-BasedHealth CareProviders*(n�376)†

(95% CI)

NHAMCSClinics*

(n�216)†

(95% CI)

Age (y)‡

Younger than 35 8.0 (4.8–12.8) N/A35–44 33.4 (26.7–40.9)45–54 31.0 (25.0–37.6)55 or older 25.9 (20.3–32.4)

Sex‡

Male 58.6 (51.7–65.2) N/AFemale 41.3 (34.7–48.3)

Board-certified§

Yes 78.0 (72.7–82.6) N/ANo 18.5 (13.9–24.0)

Health care provider orclinic specialty

General practice orfamily practice

46.0 (40.6–51.5) N/A

Internal medicine 23.9 (18.5–30.3)Obstetrics–

gynecology26.8 (22.7–31.2) 24.0 (17.4–32.1)

Midlevel healthcare provider

3.3 (2.4–4.7)

General medicine N/A 76.0 (67.9–82.6)Practice type‡

Single specialty 74.9 (67.1–81.4) N/AMultispecialty 24.8 (18.4–32.5)

Practice setting‡

Private practice 91.7 (88.6–94.0) N/ACommunity health

center8.3 (6.0–11.5)

Number of physiciansin practice‡

1 31.2 (24.1–39.2) N/A2–5 45.2 (37.6–53.0)More than 5 23.6 (17.3–31.4)

Geographic regionNortheast 16.9 (12.7–22.1) 27.0 (17.0–40.1)Midwest 24.2 (18.8–30.5) 28.6 (17.5–43.0)South 37.1 (31.2–43.3) 33.0 (20.5–48.4)West 21.9 (16.4–28.6) 11.5 (5.4–22.6)

Participate in the Breastand CervicalCancer EarlyDetection Program

Yes 21.1 (16.3–26.9) 27.3 (19.1–37.4)No 54.6 (47.3–61.7) 42.4 (30.7–55.1)Do not know 23.7 (18.2–30.2) 29.6 (20.2–41.3)

NAMCS, National Ambulatory Medical Care Survey; NHAMCS,National Hospital Ambulatory Medical Care Survey; CI,confidence interval; N/A, not applicable.

* Estimates are weighted to population size.† Missing data not shown if less than 2%‡ Data collected in NAMCS only.§ Limited to physicians only.

VOL. 118, NO. 1, JULY 2011 Lee et al HPV Testing Practices Among U.S. Providers 7

Table 2. Cervical Cancer Screening and Human Papillomavirus Testing Practices Reported by Office-Based Health Care Provider Specialty or Clinic Type

Health Care Provider Office Specialty*† Clinic Specialty*

Family PracticeInternal

MedicineObstetrics–Gynecology Midlevel

GeneralMedicine

Obstetrics–Gynecology

Practice n�170 n�57 n�90 n�59 n�108 n�108

Screening methodConventional Pap

test only20.1 (13.4–29.0) 36.2 (20.0–56.3) 0.9 (0.1–6.3)‡ 16.2 (7.0–33.0) 7.7 (2.7–20.1)‡ 12.8 (4.9–29.3)‡

Liquid-basedcytology only

56.6 (46.2–66.4) 40.7 (24.0–59.9) 75.3 (64.1–83.9) 43.8 (24.0–65.7) 69.4 (55.5–80.5) 68.7 (53.1–80.9)

Both conventionalPap test andliquid-basedcytology

21.1 (13.7–31.0) 16.6 (7.0–34.3)‡ 23.8 (15.3–34.9) 40.1 (21.8–61.5) 21.2 (12.1–34.3) 18.6 (10.7–30.4)

Perform colposcopy 21.4 (14.9–29.7) 3.8 (1.6–8.5)‡ 95.7 (85.9–98.8) 51.0 (34.2–67.6) 25.2 (15.4–38.5) 76.8 (62.8–86.7)n�131 n�34 n�89 n�49 n�79 n�94

Ever order or collectthe HPV test

76.8 (67.9–83.8) 44.7 (26.7–64.1) 98.9 (93.0–99.8) 89.1 (76.0–95.5) 75.3 (53.9–88.9) 83.0 (67.3–92.0)

Routinely performsreflex or recallHPV testing§

86.7 (76.9–92.7) 73.8 (45.2–90.6) 96.5 (90.8–98.7) 92.5 (70.3–98.5) 88.1 (75.6–94.6) 93.5 (86.0–97.1)

Routinely performsHPV cotesting§

43.1 (33.1–53.8) 36.7 (14.5–66.6) 55.2 (42.6–67.1) 46.0 (25.3–68.2) 43.4 (28.8–59.2) 63.9 (49.0–76.6)

Performs high-riskHPV testing§

84.6 (74.5–91.2) 76.0 (47.7–91.6) 91.4 (82.4–96.1) 80.3 (58.2–92.3) 78.8 (59.8–90.3) 87.7 (74.0–94.7)

Performs low-riskHPV testing§

25.6 (16.4–37.6) —‡ 41.5 (30.6–53.3) 36.7 (19.4–58.3) 17.7 (9.9–29.8) 45.8 (29.9–62.7)

n�117 —¶ n�85 n�46 n�69 n�88

Perform HPV reflex orrecall testingfor the afterPap test results�

ASC-US 82.7 (70.8–90.4) — 95.5 (87.9–98.4) 88.0 (62.0–97.1) 87.4 (70.5–95.3) 97.3 (90.2–99.3)ASC-H 76.1 (64.3–84.9) — 69.6 (58.5–78.8) 71.6 (47.8–87.4) 61.1 (43.9–76.0) 67.5 (52.0–79.9)LSIL 56.4 (43.9–68.2) — 50.3 (38.7–61.9) 52.3 (33.0–70.9) 56.0 (35.8–74.5) 57.2 (40.8–72.2)HSIL 51.6 (40.2–62.9) — 47.4 (35.9–59.2) 51.6 (32.2–70.5) 51.9 (31.8–71.5) 41.1 (25.5–58.7)AGC 40.7 (29.9–52.4) — 34.9 (24.6–46.8) 43.2 (21.8–67.4) 28.2 (15.5–45.7) 35.2 (20.6–53.2)

n�54 n�48 —¶ n�36 n�52

Routinely perform HPVcotesting forthese women#

Women youngerthan 30 y

59.7 (43.9–73.8) — 53.9 (37.4–69.6) — 70.1 (46.6–86.3) 57.9 (35.7–77.3)

Women 30 y andolder

63.3 (48.1–76.3) — 83.2 (68.5–91.9) — 65.8 (43.0–83.1) 77.7 (57.2–90.1)

Women whorequest test forcervical cancerscreening

50.7 (36.1–65.1) — 31.8 (19.1–48.0) — 45.7 (23.4–69.8) 47.5 (24.5–71.5)

Women whorequest test tocheck theirHPV infectionstatus

31.8 (19.8–46.9) — 38.6 (24.2–55.3) — 48.7 (28.5–69.3) 52.5 (28.9–75.1)

CI, confidence interval; HPV, human papillomavirus; ASC-US, atypical squamous cells of undetermined significance; ASC-H, atypicalsquamous cells, cannot rule out HSIL; LSIL, low-grade squamous intraepithelial lesion; HSIL, high-grade squamous intraepitheliallesion; AGC, atypical glandular cells.

Data are % (95% confidence interval).* Estimates weighted to population size.† Percentages were calculated from whole population of health care providers including those with missing responses.‡ Relative standard error greater than 30%, reflecting an unstable estimate.§ Of health care providers who collected the HPV test.� Of health care providers or clinics who routinely perform reflex or recall HPV testing.¶ Data not shown as a result of small sample size (n�30).# Of health care providers or clinics who routinely perform HPV cotesting.

8 Lee et al HPV Testing Practices Among U.S. Providers OBSTETRICS & GYNECOLOGY

for internal medicine should be interpreted withcaution because its relative standard error is greaterthan 0.3]) or in clinics (obstetricians–gynecologists:45.8%; general medicine: 17.7%). Using mutuallyexclusive categories, reported high-risk and low-riskHPV testing is shown in Figure 1. The majority ofhealth care providers (56.4%, 95% CI 47.8–64.6%)reported using high-risk HPV tests only, 28.5% (95%CI 21.6–36.6) of health care providers reported usingboth high-risk and low-risk HPV tests in their prac-tice, and 12.9% (95% CI 8.5–12.1%) did not know orwere unaware there was a high-risk or low-risk HPVtest. Approximately 2% reported performing low-riskHPV only (data not shown). Obstetricians–gynecolo-gists (39.0%) had the highest reported percentage ofboth high-risk and low-risk HPV testing followed bymidlevel health care providers (34.8%), family physi-cians (23.7%), and internists (16.6%), although nosignificant differences among specialties were de-tected. Among clinics, 56.1% (95% CI 43.5–68%)reported using high-risk HPV tests only, and 25%(95% CI 17.1–35%) used both high-risk and low-risk

HPV tests. An estimated 18.7% of clinics were un-aware there was a high-risk or low-risk HPV test. Thepattern of testing in general medicine clinics wassimilar to that of family practitioners.

The majority of family physicians, obstetricians–gynecologists, and hospital clinics that routinely per-formed HPV cotesting reported testing women 30 yearsand older, the only age group for which cotesting isrecommended (Table 2). Cotesting in women youngerthan 30 years old was also reported by 59.6% (95% CI48.5–69.7%) of all health care providers and 66.0%(95% CI 48.0–80.3%) of clinics. Cotesting after pa-tient request was also reported; overall, 43.8% (95%CI 33.4–54.7%) of health care providers performedcotesting in women who requested the test for cervicalcancer screening and 40.8% (95% CI 30.7–51.7%) forwomen who requested the test to check their HPVinfection status.

Health care providers’ and clinics’ use of HPVtesting after an abnormal Pap test result is shown inTable 2. Among those that routinely performed reflexHPV testing, most reported testing after an ASC-US Pap

0

10

20

30

40

50

60

70

80

90

Family practice

Internal medicine*

Obstetrics and

gynecology

Mid-level

Medicine

Wei

ghte

d (%

)

Office specialty†

Obstetrics and

gynecology

High-risk HPV onlyHigh-risk and low-risk HPV

Clinic specialty‡

Fig. 1. High-risk and low-risk human papillomavirus (HPV) testing reported by health care providers who use HPV DNA testsby office-based health care provider or clinic specialty. The percentages shown here are mutually exclusive combinationsof low-risk and high-risk HPV testing. For a report of overall low-risk HPV testing, see text. Not shown in the figure arelow-risk HPV only (2.2% for health care providers) and being unaware or unknown whether high-risk or low-risk test wasused (12.9% and 18.7% for health care providers and clinics, respectively). Error bars indicate 95% confidence intervals.*Relative standard error greater than 30% for “both high-risk and low-risk HPV” use in internal medicine, reflecting anunstable estimate. †Office specialty includes physicians who work in office settings and physicians and midlevel health careproviders who work in community health centers. ‡Clinic specialty includes specialty of the hospital outpatient department.Lee. HPV Testing Practices Among U.S. Providers. Obstet Gynecol 2011.

VOL. 118, NO. 1, JULY 2011 Lee et al HPV Testing Practices Among U.S. Providers 9

test result, a use that is consistent with guideline recom-mendations. A high percentage of health care providersand clinics performed reflex HPV testing after Pap testresults of ASC-H (71.4%, 95% CI 63.5–78.3% and62.8%, 95% CI 49.0–74.9%, respectively) or HSIL(50.7%, 95% CI 42.4–58.9% and 49.0%, 95% CI 33.1–65.2%, respectively), results for which HPV testing is notrecommended. For an LSIL Pap test result, in whichreflex HPV testing is an acceptable option for post-menopausal women, 54.2% (95% CI 45.7–62.5%) ofhealth care providers and 56.4% (95% CI 40.6–70.9%)of clinics used reflex HPV testing. Reflex HPV testingafter an atypical glandular cell Pap test result wasreported by 37.5% (95% CI 28.9–47.0%) of health careproviders and 30.1% (95% CI 19.6–43.1%) of clinics.Reflex HPV testing in women younger than 30 yearsold was reported by 56.1% (95% CI 47.8–64.1%) ofhealth care providers and 54.5% (95% CI 42.6–65.9%)of clinics, whereas 53.7% (95% CI 45.2–62.0%) of healthcare providers and 44.8% (95% CI 33.5–56.6%) of clinicsreported using reflex HPV testing in women 30 yearsand older.

DISCUSSIONIn this national survey of Pap test providers, we foundHPV DNA test use to be widespread among variousspecialties that offer routine cervical cancer screening.Although many health care providers reported guide-line-consistent HPV testing, a large number also re-ported inappropriate HPV testing such as low-risk HPVtest use by nearly one-third of health care providers andHPV cotesting in women younger than age 30 years by60% of health care providers. These findings suggest thatalthough HPV testing has become a routine part ofcervical cancer screening, many health care providersare not following guideline recommendations, whichmay increase medical costs and lead to unnecessaryfollow-up for women with positive test results.

Our findings of low-risk HPV testing by 31% ofhealth care providers and 25% of clinics that performHPV testing, a practice that has not been assessedpreviously, is surprising. The low-risk HPV DNA testscreens for infection with nononcogenic HPV typesand thus serves no purpose in the context of cervicalcancer screening.2,3 In addition, although low-riskHPV types such as HPV 6 and HPV 11 cause 90% ofgenital warts, low-risk HPV testing should not beperformed for genital warts screening or managementbecause most HPV infections are asymptomatic andwill clear on their own. Further information on themanagement of genital warts is available from theCDC at http://www.cdc.gov/std/treatment/2010/genital-warts.htm.9,10 A positive low-risk HPV test

may prompt further follow-up testing or proceduresby health care providers, outcomes that we wereunable to assess in this survey. Although a positivelow-risk HPV test may not be clinically significant,whether there is disclosure to patients of test resultsand the psychosocial effect of diagnosis with low-riskHPV are unclear.

The continued use of low-risk HPV testing byhealth care providers in the United States may be drivenby a combination of financial gain, test marketing, andhealth care provider confusion on the difference be-tween the low-risk and high-risk tests.4,7,11,12 Eliminatingthe availability of the low-risk HPV test, which has noclinical indications, should be considered regardless ofthe test’s current level of penetration because suchtesting adds cost to the medical system without addingvalue to patient care. When HPV testing first becameavailable, the early generations of Digene’s HybridCapture HPV DNA tests included both high-risk andlow-risk HPV types.11 Since then, HPV DNA teststhat screen only for high-risk types have becomeavailable, yet the combined low-risk and high-riskHPV tests remains in active use.13 It is unknown howlaboratories influence the amount of low-risk HPVtesting that occurs, but a study in 2007 found 45% ofpathology laboratories reported offering the low-riskHPV test with the widespread availability attributedmainly to reimbursement and physician demand.13

We examined laboratory requisition forms from sev-eral U.S. laboratories, some of which had the com-bined high-risk and low-risk HPV test available as acheckbox option, illustrating the ease with whichhealth care providers can perform low-risk testing aspart of routine cervical cancer screening.

When ordered together, the low-risk HPV DNAtest is billed at the rate of the high-risk HPV DNA testusing the same Current Procedural Terminologycode, which doubles the cost of HPV testing. Becausemost health care providers (91%) in our study were inprivate practice, liberal test reimbursement by privateinsurance may be an important factor behind thispractice.

More than half of health care providers and clinicsthat perform HPV cotesting reported doing so inwomen younger than 30 years old, in which a positiveHPV test more likely signifies a transient HPV infectionthat will resolve spontaneously without needing furtherintervention.9,10 Because of the high prevalence of HPVinfection in women in their 20s,14 a large number ofwomen in this age group would subsequently be re-ferred to colposcopy and follow-up testing after HPVcotesting. Compared with a study of health care provid-ers conducted in 2004,4 our study found the percentage

10 Lee et al HPV Testing Practices Among U.S. Providers OBSTETRICS & GYNECOLOGY

of health care providers who reported cotesting womenyounger than age 30 years old has doubled. Avoidingunnecessary workup or treatment of transient HPVinfections in younger women is an important goal incervical cancer screening because studies have foundassociations between certain procedures used to treatcervical dysplasia and adverse birth outcomes.15,16

Another common reason reported by health careproviders for HPV cotesting was a request frompatients to check HPV infection status. In a study ofdirect-to-consumer advertising and its effect on HPVtesting, direct-to-consumer advertising was associatedwith significant increases in HPV test use by healthcare providers.17 Because HPV is the most commonsexually transmitted infection in the United States andmost HPV infections are asymptomatic and transient,screening for HPV infection status outside the contextof cervical cancer screening should not be per-formed.18 Several studies found women who receivepositive HPV test results can experience negativepsychosocial effects and health care providers shouldconsider discussing with patients whether HPV test-ing is clinically recommended for them if patientsrequest to be tested.19–21

As a management test, most health care providersused reflex HPV testing after an ASC-US Pap testresult, a recommended use in patients aged 21 yearsand older. However, health care providers and clinicsalso reported using HPV testing after higher-gradePap test result abnormalities such as ASC-H andHSIL. We were unable to assess for nuances in HPVtesting recommendations for certain Pap test resultabnormalities in our study such as newer guidelinesthat recommend against reflex HPV testing for pa-tients younger than age 21 years2; however, excludingthose special circumstances, there are limited data onthe use of HPV testing for higher grade abnormalities.Additionally, guidelines recommend that patientswith such Pap test results be referred to immediatecolposcopy (or excisional treatment for HSIL) as thenext management step instead.2

Our findings of nonrecommended HPV DNA test-ing among Pap test providers indicate a need for devel-oping and testing interventions, especially ones tieddirectly to reimbursement, to encourage guideline-con-sistent test use. The continued use of low-risk HPVtesting by health care providers warrants special consid-eration. Based on our knowledge of the natural historyof HPV and the association of high-risk types withcervical cancer, there is simply no role for low-risk HPVtesting in cervical cancer screening nor any other clinicalscenarios to justify its use. Although health care provid-ers have now had the option of performing only high-

risk HPV testing for a number of years, our findingsindicate that stronger measures such as engaging labo-ratories to stop the availability of the low-risk HPV testand limiting reimbursement from insurance companiesto high-risk testing only may be needed. Health careproviders also have an ethical obligation to informpatients, who are ultimately responsible for the costeither directly or indirectly, when ordering this test thathas no clinical value. Finally, providing evidence thatlow-risk HPV tests can result in unnecessary and evenharmful follow-up procedures and tests may be thearmamentarium needed to take the low-risk test off themarket.

Convincing health care providers and patients thatmore testing does not equal better medical care is achallenging task with medicolegal concerns and patientdemand cited as some of the barriers.22,23 Formativeresearch is needed to clarify the issues surroundinghealth care providers’ use of HPV tests and the factorsinfluencing guideline adherence. Educational materialsfor the general public on HPV and HPV DNA tests canhelp patients understand when testing may or may notbe needed.24 Health care provider education on recom-mended uses of HPV testing should reinforce the ben-efits of testing as well as potential harms of overtestingsuch as unnecessary follow-up or overtreatment of tran-sient HPV infections. Changing laboratory requisitionforms to clearly reflect guideline-consistent HPV testingand point-of-care decision support algorithms designedto generate patient-specific recommendations in clinicalcare have been effective in improving health care pro-vider performance in other areas and should be furtherinvestigated for this purpose.25,26

Our study had several limitations. The smallsample size limited our ability to describe certainHPV testing practices for internists and midlevelhealth care providers and caused some estimates to beunstable. It also limited our power to detect differ-ences among specialties and perform multivariateanalysis of demographic and practice characteristicsassociated with nonrecommended HPV testing. Also,information on the insurance distribution of the pa-tient population in a practice was unavailable on anindividual health care provider level and we wereunable to assess for correlation between testing prac-tices and the patient population served. With addi-tional years of data and a larger sample, furtheranalysis may identify associated health care provideror practice factors. Although our respondents weresimilar to nonrespondents in certain demographiccharacteristics such as geographic region and boardcertification, nonrespondents were more likely to be

VOL. 118, NO. 1, JULY 2011 Lee et al HPV Testing Practices Among U.S. Providers 11

older males, and our results may not be representativeof that group.

In our survey, health care providers were askedto identify patients for whom they would performHPV testing based on age group, an abnormal Paptest result, or other short descriptive characteristic.The questions did not take into account other factorssuch as patient sexual history or previous Pap testresults, which may influence health care providers’actual practices. Finally, our data were self-reportedand not validated by data of actual laboratory testsordered by health care providers.

In conclusion, many Pap test providers are usingHPV DNA tests in ways inconsistent with guidelinerecommendations, most notably, the use of low-riskHPV testing. Our findings indicate the need forwide-reaching interventions such as limiting test re-imbursement to facilitate recommended uses of HPVDNA tests and eliminate low-risk HPV testing. Edu-cational materials for patients and health care provid-ers to address patient demand for HPV testing mayalso help to increase guideline consistent HPV testuse. Further research is needed to identify interven-tions that effectively promote health care provideradherence to guideline recommendations, especiallyfor newer technologies.

REFERENCES1. Bosch FX, Lorincz A, Munoz N, Meijer CJ, Shah KV. The

causal relation between human papillomavirus and cervicalcancer. J Clin Pathol 2002;55:244–65.

2. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, WilkinsonEJ, Solomon D; 2006 ASCCP-Sponsored Consensus Confer-ence. 2006 consensus guidelines for the management ofwomen with abnormal cervical screening tests. J Low GenitTract Dis 2007;11:201–22.

3. Saslow D, Runowicz CD, Solomon D, Moscicki AB, SmithRA, Eyre HJ, et al. American Cancer Society guideline for theearly detection of cervical neoplasia and cancer. CA CancerJ Clin 2002;52:342–62.

4. Irwin K, Montano D, Kasprzyk D, Carlin L, Freeman C,Barnes R, et al. Cervical cancer screening, abnormal cytologymanagement, and counseling practices in the United States.Obstet Gynecol 2006;108:397–409.

5. Saraiya M, Berkowitz Z, Yabroff KR, Wideroff L, Kobrin S,Benard V. Cervical cancer screening with both human papil-lomavirus and Papanicolaou testing vs Papanicolaou testingalone: what screening intervals are physicians recommending?Arch Intern Med 2010;170:977–85.

6. Clinical management guidelines for obstetrician-gynecologists.Cervical cytology screening (replaces committee opinion 152,March 1995). ACOG Practice Bulletin No. 45. AmericanCollege of Obstetricians and Gynecologists. Obstet Gynecol2003;102:417–27.

7. Jain N, Irwin K, Carlin L, Freeman C, Montano D, KasprzykD. Use of DNA tests for human Papillomavirus infection by USclinicians, 2004. J Infect Dis 2007;196:76–81.

8. Cervical cytology screening. ACOG Practice Bulletin No. 45.American College of Obstsetricians and Gynecologists. ObstetGynecol 2003;102:417–27.

9. Ho GY, Bierman R, Beardsley L, Chang CJ, Burk RD. Naturalhistory of cervicovaginal papillomavirus infection in youngwomen. N Engl J Med 1998;338:423–8.

10. Moscicki AB, Shiboski S, Broering J, Powell K, Clayton L, JayN, et al. The natural history of human papillomavirus infectionas measured by repeated DNA testing in adolescent and youngwomen. J Pediatr 1998;132:277–84.

11. Moriarty AT. A rock and a hard place: HPV testing andfinancial gain. Diagn Cytopathol 2007;35:463–4.

12. Austin RM, Benstein B, Bentz J, Bigner S, Freund GG, RoccoGL, et al. Market survey predictions on the future of US Paptesting. Cytojournal 2009;6:17.

13. Moriarty AT, Schwartz MR, Eversole G, Means M, Clayton A,Souers R, et al. Human papillomavirus testing and reportingrates: practices of participants in the College of AmericanPathologists Interlaboratory Comparison Program in Gyneco-logic Cytology in 2006. Arch Pathol Lab Med 2008;132:1290–4.

14. Dunne EF, Unger ER, Sternberg M, McQuillan G, Swan DC,Patel SS, et al. Prevalence of HPV infection among females inthe United States. JAMA 2007;297:813–9.

15. Arbyn M, Kyrgiou M, Simoens C, Raifu AO, Koliopoulos G,Martin-Hirsch P, et al. Perinatal mortality and other severeadverse pregnancy outcomes associated with treatment ofcervical intraepithelial neoplasia: meta-analysis. BMJ 2008;337:a1284.

16. Kyrgiou M, Koliopoulos G, Martin-Hirsch P, Arbyn M, Pren-diville W, Paraskevaidis E. Obstetric outcomes after conserva-tive treatment for intraepithelial or early invasive cervicallesions: systematic review and meta-analysis. Lancet 2006;367:489–98.

17. Price RA, Frank RG, Cleary PD, Goldie SJ. Effects of direct-to-consumer advertising and clinical guidelines on appropriateuse of human papillomavirus DNA tests. Med Care 2011;49:132–8.

18. Cox JT, Moriarty AT, Castle PE. Commentary on statementon human papillomavirus test utilization. Arch Pathol LabMed 2009;133:1192–4.

19. Kahn JA, Slap GB, Bernstein DI, Kollar LM, Tissot AM,Hillard PA, et al. Psychological, behavioral, and interpersonalimpact of human papillomavirus and Pap test results. J Wom-ens Health (Larchmt) 2005;14:650–9.

20. Maissi E, Marteau TM, Hankins M, Moss S, Legood R, GrayA. Psychological impact of human papillomavirus testing inwomen with borderline or mildly dyskaryotic cervical smeartest results: cross sectional questionnaire study. BMJ 2004;328:1293.

21. McCaffery K, Waller J, Forrest S, Cadman L, Szarewski A,Wardle J. Testing positive for human papillomavirus in routinecervical screening: examination of psychosocial impact. BJOG2004;111:1437–43.

22. Cooper CP, Saraiya M, McLean TA, Hannan J, Liesmann JM,Rose SW, et al. Report from the CDC. Pap test intervals usedby physicians serving low-income women through theNational Breast and Cervical Cancer Early Detection Program.J Womens Health (Larchmt) 2005;14:670–8.

23. Palfrey S. Daring to practice low-cost medicine in a high-techera. N Engl J Med 2011;364:e21.

24. Papa D, Moore Simas TA, Reynolds M, Melnitsky H. Assess-ing the role of education in women’s knowledge and accep-tance of adjunct high-risk human Papillomavirus testing for

12 Lee et al HPV Testing Practices Among U.S. Providers OBSTETRICS & GYNECOLOGY

cervical cancer screening. J Low Genit Tract Dis 2009;13:66–71.

25. Kahan NR, Waitman DA, Vardy DA. Curtailing laboratorytest ordering in a managed care setting through redesign of acomputerized order form. Am J Manag Care 2009;15:173–6.

26. Garg AX, Adhikari NK, McDonald H, Rosas-Arellano MP,Devereaux PJ, Beyene J, et al. Effects of computerized clinicaldecision support systems on practitioner performance andpatient outcomes: a systematic review. JAMA 2005;293:1223–38.

27. Centers for Disease Control and Prevention (CDC). FDAlicensure of bivalent human papillomavirus vaccine (HPV2,Cervarix) for use in females and updated HPV vaccination

recommendations from the Advisory Committee on Immuni-zation Practices (ACIP). MMWR Morb Mortal Wkly Rep2010;59:626–9.

28. Cervical cytology screening. ACOG Practice Bulletin No. 109.American College of Obstetricians and Gynecologists. ObstetGynecol 2009;114:1409–20.

29. Solomon D, Papillo J, Davis Davey D; Cytopathology Educa-tion and Technology Consoritum. Statement on HPV DNAtest utilization. J Low Genit Tract Dis 2009;13:135–6.

30. American Society for Colposcopy and Cervical Pathology. HPVgenotyping clinical update. http://www.asccp.org/ConsensusGuidelines/HPVGenotypingClinicalUpdate/tabid/5963/Default.aspx. Accessed March 25, 2011.

VOL. 118, NO. 1, JULY 2011 Lee et al HPV Testing Practices Among U.S. Providers 13