Previous Page
Table 4 shows detection rates for both tests, by phase and for the study as a whole,
together with their predictive values. We used HGSIL or worse on colposcopy or biopsy to
define true disease status.
Table 4: Detection rates and predictive values for VIA and Pap
smear
| Test |
Parameter (95% CI) |
| Phase I |
Phase II |
Total |
| VIA |
| PPV |
25·9%
(23·3-28·7) |
18·6%
(16·1-21·4) |
22·7%
(20·8-24·6) |
| NPV |
73·3%
(69·3-77·0) |
96·3%
(95·1-97·2) |
89·5%
(88·0-90·9) |
| Detection rate* |
31/1000 |
72/1000 |
39/1000 |
| Pap smear |
| PPV |
43·9%
(39·9-48·0) |
33·3%
(27·7-39·3) |
40·7%
(37·4-44·0) |
| NPV |
85·4%
(82·9-87·6) |
93·9%
(92·7-94·9) |
91·1%
(89·9-92·1) |
| Detection rate** |
32/1000 |
42/1000 |
34/1000 |
PPV = positive predictive
value
NPV = negative predictive value |
*Among the 8731 phase I and 2182 phase II women with an informative VIA
result.
**Among the 8348 phase I and 2144 phase II women with an informative Pap smear result. |
Table 5 shows screening test results according to reference test outcomes for phase II
data for both thresholds that we used to define disease. 2130 (97·6%) of 2182 women with
a VIA result, and 2092 (97·5%) of 2144 women with an adequate Pap smear had both a
screening-test result and a reference-test result in phase II.
Table 5: Phase II screening-test results by reference standard
outcomes
| Test |
T+,D+ |
T+,D- |
T-,D+ |
T-,D- |
| VIA (n=2130) |
| Low threshold |
316 |
533 |
182 |
1099 |
| High threshold |
158 |
691 |
48 |
1233 |
| Pap smear (n=2092) |
| Low threshold |
144 |
123 |
343 |
1482 |
| High threshold |
89 |
178 |
112 |
1713 |
T+ = test positive
T- = test negative
D+ = disease according to reference test
D- = no disease according to reference test
Low threshold > LGSIL
on reference test
High threshold > HGSIL on reference test. |
Table 6 lists the test
qualities for phase II derived from the data in Table 5. Phase I sensitivity and
specificity values are not shown because, despite the use of a randomised-selection
protocol in phase I to identify a proportion (10%) of test-negatives to receive the
reference test, upon which statistical adjustments for verification bias were to have been
made, the sample of VIA test-negative women who actually underwent colposcopy was
non-random. Consequently, adjustment for verification bias by use of phase I data was
inappropriate, and the unadjusted test-quality results were biased. Verification bias
(also called workup bias) occurs when the chances of being referred for the reference test
are different for those who test positive on screening and for those who test negative.20
Such bias inflates estimates of sensitivity and falsely reduces specificity.21
Table 6: Phase II sensitivity and specificity of each test
| Test |
Sensitivity
(95% CI) |
Specificity
(95% CI) |
| VIA (n=2130) |
| High threshold* |
76·7%
(70·3-82·3) |
64·1%
(61·9-66·2) |
| Low threshold** |
63·5%
(59·1-67·7) |
67·3%
(65·0-69·6) |
| Pap smear (n=2092) |
| High threshold* |
44·3%
(37·3-51·4) |
90·6%
(89·2-91·9) |
| Low threshold** |
29·6%
(25·6-33·8) |
92·3%
(90·9-93·6) |
* > HGSIL on the reference test.
** > LGSIL on the
reference test. |
In more-developed countries, Pap smears have formed the basis of cervical-cancer
screening and detection programmes for many years. National cytology-based screening
programmes have contributed substantially to the marked decline in deaths from cervical
cancer in these countries.22-24 Yet, in many less-developed countries, the
technical complexity of cytological testing, and the infrastructure required to implement
it, precludes the effectiveness of national Pap-smear programmes.2,6,19
Several studies have shown the potential value of VIA (cervicoscopy) as a screening
approach in less-developed countries,9-11,14,25-27 but none established
precisely the quality of the VIA test under clinical conditions likely to occur in
less-developed countries--screening done by non-physicians in a basic health facility.
In such a setting, our results showed that VIA can effectively identify most cases of
cervical pre-cancer and cancer. In phase I, the detection rates for VIA and the Pap smear
were equal. In phase II, however, the detection rate for VIA was higher than that for
cytology. The detection rate for both tests increased in phase II since 98% of
VIA-positive and 99% of Pap-smear-positive women received the reference standard in that
phase compared with only 60% (VIA) and 49% (Pap smear) of women in phase I.
The test-positive rate for VIA doubled between phase I and phase II of the study. The
priority for the study clinicians was detection of cervical cancer rather than avoidance
of over-referral, particularly in phase II, because that protocol called for all women to
receive the follow-up test irrespective of their screening result. Given that the
nurse-midwives knew that all of their VIA assessment would be confirmed or rejected that
same day by a physician who did colposcopy, the nurse-midwives may have defined borderline
cases as abnormal in phase II to ensure that they missed as few cases as possible.
The positive predictive value of VIA was lower than that of the Pap smear in both study
phases, partly because the specificity of cytology was consistently higher than that of
VIA. We could not obtain an accurate estimate of the prevalence of disease from phase I
but, given that rates of sexually transmitted infection were different in the two study
phases, the incidence of cervical lesions may have differed between phases. Differences in
positive predictive value between phases may thus be due to prevalence differences, but
are also in part an artifact of biases introduced in phase I that affected our estimates
of sensitivity, specificity, and the prevalence of disease. As expected, definition of
disease by use of lower threshold (LGSIL or worse) increased the positive predictive value
of both tests.
Given that the appearance of high-grade lesions was used to dictate clinical treatment
decisions, the high negative predictive value shown by both tests warrants particular
mention. In places characterised by similar incidence of cervical disease, including many
countries in Africa, Asia, and Latin America, the use of VIA as a primary screening test
means that women assessed as test-negative would be reassured that most probably they do
not have HGSIL or cancer.
Symptoms of sexually transmitted infection were more common among women in phase II of
the study, which suggests that the women tested in that phase were a slightly different
reference population from that of phase I. Nevertheless, the presence of sexually
transmitted infections did not affect screening capabilities between phases in different
ways, and the full spectrum of disease was observed in both phases. Therefore, we believe
that the sensitivity and specificity results of phase II show the potential for VIA
screening from a group of trained nurse-midwives.
Given our study conditions, our results are likely to represent the lower end of the
range for sensitivity and specificity of VIA. Higher test qualities for both tests are
likely to be observed under better service-delivery conditions (eg, better lighting,
examination tables, speculae) and with more standardised VIA training than was available
in our study. Additional improvements in VIA specificity could result from repeat VIA
testing in women likely to return to the same site for healthcare, or by use of VIA for
triage of those who should undergo further tests when such follow-up is feasible.10,26,27
Our results are consistent with recent studies that have shown that VIA is more
sensitive but usually less specific than cytology.19,28 The finding that a
visual inspection test can identify a greater proportion of diseased cases under certain
clinical circumstances than the Pap smear is not unexpected, since measurements of
sensitivity and specificity shown by the Pap smear in various studies are not consistently
high and have ranged from 20%-85%.20,29
When attempting to diagnose whether a cervix is healthy, or diseased and in need of
treatment, the clinician must "find the lesion". In our study, in more than 75%
of the cases in which a lesion was found on colposcopy or biopsy, the lesion was also
found with simple VIA. We agree with Ottaviano and La Torre9 that
"colposcopic magnification is not essential . . . [to identify] the cervix at
risk". Our data show that VIA could be an acceptable means of screening for cervical
cancer, especially in poorly-resourced locations. However, the acceptance of VIA depends
on the attitudes of policymakers and opinion leaders in various countries. The observed
high number of false-positive results for VIA may lead to high rates of referral, and may
increase rates of treatment, with the associated potential for increased patients'
discomfort and increased numbers of side-effects. Nonetheless, our study shows that VIA
can identify most true cases of cervical pre-cancer and cancer. Where large-scale
Pap-smear screening is not now available and is not likely to be available consistently in
the future, VIA could be a readily available, potentially sustainable means of testing
that, when coupled effectively with treatment, could reduce the burden of disease in
populations in which the incidence of cervical cancer is high. Even where cytology
services are well established, VIA might be a cost-effective method of rapidly
differentiating between a potentially diseased cervix and a healthy one.
| Writing committee |
L Gaffikin, P D Blumenthal, J McGrath, Z M Chirenje |
| Study design |
L Gaffikin, Z M Chirenje, P D Blumenthal, H Sanghvi, T Chipato |
| Training interventions |
P D Blumenthal, H Sanghvi, R Makunike, Z M Chirenje, J Kasule |
| Fieldwork |
Z M Chirenje, R Makunike, T Chipato, J Kasule, E Ngwalle |
| Study monitoring |
Z M Chirenje, T Chipato, R Makunike, L Gaffikin, H Sanghvi, P D Blumenthal |
| Data management |
S Rusakaniko. J McGrath |
| Statistical analysis |
J McGrath, L Gaffikin, P D Blumenthal, S Rusakaniko |
Acknowledgments
The University of Zimbabwe (Department of Obstetrics and Gynaecology) oversaw the
fieldwork. The Ministry of Health, Zimbabwe, gave logistical support, and the Zimbabwe
National Family Planning Council (ZNFPC) allowed its colposcopy clinic to be the referral
site for diagnostic testing. We thank the women who took part in the study from the City
of Harare and City of Chitungwiza Department of Health facilities; the Minister of Health
(T Stamps), A Zinanga, R Munochivyi, L Mbengeranwa, M Simoyi, and C Madziwa for their
support and assistance, and the research nurse-midwives V Akino, F Chanakira, L Chidede, M
M Kamba, P Kunaka, M Kurangwa, M Mlingo, M Mpofu, and E Mutukwa. JHPIEGO staff members
provided technical support throughout the study. We also thank Noel McIntosh for project
ideas; Susan J Griffey Brechin, Natalie Maier, and Chris Davis for administrative and
production support; the Commonwealth Regional Health Community Secretatiat (CRHCS) and
representatives of member states for original study-design input; Richard Szumel, who was
the US-based pathologist, and Stephen Walter for help with the revised protocol and
analysis plan.
Technical assistance and financial support came from the JHPIEGO Corporation and an
affiliate of The Johns Hopkins University through funds from the US Agency for
International Development (USAID Office of Population, Center for Population, Health and
Nutrition, Global Bureau, under the terms of Cooperative Agreement Number
CCP-3069-A-00-3020-00).
- Sherris JD, Wells ES, Tsu VD, Bishop A. Cervical cancer in developing countries: a
situation analysis. Seattle: PATH, 1993.
- Lunt R. Worldwide early detection of cervical cancer. Obstet Gynecol 1984; 63:
708-13.
- Machoki JMN. Rogo KO. Knowledge and attitudinal study of Kenyan women in relation to
cervical carcinoma. Int J Gynaecol Obstet 1990; 34: 55-59.
- Kasule J. The pattern of gynaecological malignancy in Zimbabwe. East Afr Med J
1989; 66: 393-99.
- Du Toit JP. A cost-effective but safe protocol for the staging of invasive cervical
carcinoma in a third world country. Int J Gynaecol Obstet 1988; 26: 261-64.
- Sankaranarayanan R, Pisani P. Prevention measures in the third world: are they
practical? In: Franco E, Monsonego J, eds. New developments in cervical cancer screening
and prevention. Oxford: Blackwell Science, 1997: 70-83.
- Gaffikin L, Blumenthal PD, Davis C, Brechin SJG, eds. Alternatives for cervical cancer
screening and treatment in low-resource settings: workshop proceedings; May 21-22, 1997,
Baltimore, Maryland. Baltimore: JHPIEGO Corporation, 1997.
- Stjernswärd J, Eddy D, Luthra U, Stanley K. Plotting a new course for cervical cancer
screening in developing countries. World Health Forum 1987; 8: 42-45.
- Ottaviano M, La Torre P. Examination of the cervix with the naked eye using acetic acid
test. Am J Obstet Gynaecol 1982; 143: 139-42.
- Van Le L, Broekhuizen FF, Janzer-Steele R, Behare M, Samter T. Acetic acid visualization
of the cervix to detect cervical dysplasia. Obstet Gynecol 1993; 81: 293-95.
- Megevand E, Denny L, Dehaeck K, Soeters R, Bloch B. Acetic acid visualization of the
cervix: an alternative to cytologic screening. Obstet Gynaecol 1996; 88:
383-86.
- Wertlake PT, Francus K, Newkirk GR, Parham GP. Effectiveness of the Papanicolaou smear
and speculoscopy as compared with the Papanicolaou smear alone: a community-based clinical
trial. Obstet Gynecol 1997; 90: 421-27.
- Jonsson M, Karlsson R, Evander M, Gustavsson Å, Rylander E, Wadell G. Acetowhitening of
the cervix and vulva as a predictor of subclinical human papillomavirus infection:
sensitivity and specificity in a population-based study. Obstet Gynecol 1997; 90:
744-47.
- Cecchini S, Bonardi R, Mazzotta A, Grazzini G, Iossa A, Ciatto S. Testing cervicography
and cervicoscopy as screening tests for cervical cancer. Tumori 1993; 79:
22-25.
- Wright TC, Moscarelli RD, Dole P, Ellerbrock TV, Chiasson MA, Vandevanter N.
Significance of mild cytologic atypia in women infected with human immunodeficiency virus.
Obstet Gynecol 1996; 87: 515-19.
- Mitchell MF, Schottenfeld D, Tortolero-Luna G, Cantor SB, Richards-Kortum R. Colposcopy
for the diagnosis of squamous intraepithelial lesions: a meta-analysis. Obstet Gynecol 1998;
91: 626-31.
- Blumenthal P, Sanghvi H. Atlas for unaided visual inspection of the cervix. Baltimore
(Maryland) and Harare (Zimbabwe): JHPIEGO Corporation and University of Zimbabwe Medical
School, 1997.
- Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject
to selection bias. Biometrics 1983; 39: 207-15.
- Sankaranarayanan R, Syamalakumari B, Wesley R, et al. Visual inspection as a screening
test for cervical cancer control in developing countries. In: Franco E, Monsonego J, eds.
New developments in cervical cancer screening and prevention. Oxford: Blackwell Science,
1997: 411-21.
- Fahey MT, Irwig L, Macaskill P. Meta-analysis of Pap test accuracy. Am J Epidemiol
1995; 141: 680-89.
- Choi BCK. Sensitivity and specificity of a single diagnostic test in the presence of
workup bias. J Clin Epidemiol 1992; 45: 581-86.
- US Preventive Services Task Force. Screening for cervical cancer. Am Fam Physician
1990; 41: 853-57.
- de Wolf CJM. Organization and results of cervical cancer screening in Europe over the
past 20 years. In: Franco E, Monsonego J, eds. New developments in cervical cancer
screening and prevention. Oxford: Blackwell Science, 1997: 209-19.
- Nieminen P, Kallio M, Hakama M. The effect of mass screening on incidence and mortality
of squamous and adenocarcinoma of cervix uteri. Obstet Gynecol 1995; 85:
1017-21.
- Abrams J. A preliminary study of the Gynoscope: an adjunct to cytologic screening of the
cervix. Am J Gynecol Health 1990; 4: 37-43.
- Frisch LE, Milner FH, Ferris DG. Naked-eye inspection of the cervix after acetic acid
application may improve the predictive value of negative cytologic screening. J Fam
Pract 1994; 39: 457-60.
- Slawson DC, Bennett JH, Herman JM. Are Papanicolaou smears enough? Acetic acid washes of
the cervix as adjunctive therapy: a HARNET study: Harrisburg Area Research Network. J
Fam Pract 1992; 35: 271-77.
- Sankaranarayanan R, Shyamalakumary B, Wesley R, Sreedevi Amma N, Parkin DM, Krishnan
Nair M. Visual inspection with acetic acid in the early detection of cervical cancer and
precursors. Int J Cancer 1999; 80: 161-63.
- Wilkinson EJ. Pap smears and screening for cervical neoplasia. Clin Obstet Gynecol 1990;
33: 817-25.
Previous Page
|
