OBJECTIVE: Rural women have increased rates of cervical neoplasia, but colposcopic services are limited in rural clinics. The purpose of this study was to estimate the efficacy of telecolposcopy for women in rural health care sites.
METHODS: Women with an indication for colposcopy were examined by local colposcopists at one of two rural clinics. Images of the colposcopic examination were transmitted to a tertiary care center for interpretation by an expert colposcopist. Another colposcopist (site expert) in attendance at the rural site also examined the same subjects, but did not share findings with the other colposcopists. Colposcopists independently determined the adequacy of the examination, colposcopic impression, biopsy intent and site, and management. Agreement between colposcopic impressions and cervical histology were assessed by using percent agreement, Cohen's κ statistic, and McNemar's test with Bonferroni's adjustment.
RESULTS: Teleconsultation was required for 36.2% of colposcopic examinations. A significantly lower percentage of satisfactory colposcopic examinations was noted by the distant colposcopists (60.0%) compared with the other colposcopists (P < .001). Colposcopic impression agreement with histology varied minimally, 59.7% (κ = 0.31) for local colposcopists, 52.7% (κ = 0.22) for site experts, 55.7% (κ = 0.27) for distant experts who concurrently viewed the examination, and 49.7% (κ = 0.16) for distant experts who viewed the examination on videotape at a later time.
CONCLUSION: Teleconsultation was used for a substantial number of examinations. Diagnostic accuracy was maintained, but determination of colposcopic examination adequacy may be impaired by telecolposcopy. Telecolposcopy may help reduce barriers to medical access for women in rural areas.
Telecolposcopy does not compromise diagnostic accuracy and is beneficial for teleconsultation, but judgment of colposcopic examination adequacy may be impaired.
Gynecologic Cancer Prevention Center, Departments of Family Medicine, and Obstetrics and Gynecology, and Office of Biostatistics and Bioinformatics, Medical College of Georgia, Augusta, Georgia; Tri‐County Health Clinic, Warrenton, Georgia; and Ware County Health Department, Waycross, Georgia.
Address reprint requests to: Daron G. Ferris, MD, Medical College of Georgia, 1423 Harper Street, HH‐100, Augusta, GA 30912; E‐mail: email@example.com.
Supported by a grant (R01 HS08814) from the Agency for Health Care Policy and Research, and the National Cancer Institute, National Institutes of Health, Bethesda, Maryland.
We thank Max Stachura, MD, and Ann Brown, MHSA, of the Medical College of Georgia Telemedicine Center for administrative and technical assistance; and Eileen Dickman, PhD, MBA, and Roberta Shiver for data management services.
Received June 7, 2001. Received in revised form September 17, 2001. Accepted September 24, 2001.
Although cervical cancer can affect any woman, certain segments of the population are at increased risk. Proportionally, a greater percentage are poor and of minority races.1–6 Moreover, rural women, regardless of socioeconomic or racial status, have the greatest incidence of carcinoma in situ when compared with national data.7 Consequently, any effort to reduce cervical cancer mortality must include targeting poor, black, and rural women.
Prevention of cervical cancer depends on widespread screening, accurate diagnosis of precursor lesions, and effective treatment. Inadequate patient compliance after an abnormal Papanicolaou smear may occur because some women may be reluctant to have further evaluation by colposcopy.6,8–10 This noncompliance is more evident among women referred long distances.8 Furthermore, there appears to be a deficit of colposcopic services, especially for women located in rural sites.11 Limited access to care, excessive health care cost, or lack of insurance coverage, and transportation barriers may further impede proper evaluation and management of these rural women.
The use of telecommunications and computer technology to deliver medical services to patients located at remote clinical sites has been termed telemedicine.12 Telemedicine may provide a flexible, technological response to health care delivery problems, including restricted access to care, geographic maldistribution of medical resources, isolation of rural physicians, and increasing cost of general health care. Technologic advances of fiberoptic and satellite transmission, image compression, simultaneous annotation, efficient computers, and reduced technology costs have allowed a proliferation of telemedicine systems. Readily available, technology‐enhanced informed consultation from a central site may enrich the quality of rural health care. Referral delays and patient follow‐up noncompliance are minimized, whereas continuity of care is maintained with primary care providers. Furthermore, the medical experts can supervise nonexpert clinicians and provide consultation services.
Telecolposcopy may be able to address simultaneously the triad of quality, access, and cost in health care. Telemedicine systems are being established quickly around the country. Knowledge of patient‐provider health care system interactions and efficacy should ideally precede the rapid expansion of technology in medicine. Telemedicine has been applied to various disciplines of medicine, but gynecologic applications are limited.13–22 Telecolposcopy has not been explored in a comprehensive manner. The purpose of this study was to estimate the efficacy or operative quality of telemedicine colposcopy for women in rural health care sites.
MATERIALS AND METHODS
Women 18 years of age or older presenting consecutively for colposcopic evaluation after detection of an abnormal Papanicolaou smear or genital condyloma, abnormal appearing cervix, or other condition conducive to colposcopic examination were asked to participate. Women were enrolled at one of two rural sites, the Ware County Health Department, Waycross, and the Tri‐County Health Clinic, Warrenton, Georgia. All subjects read and signed the institutional review board‐approved informed consent document. Exclusion criteria included severe cervicitis, menses, pregnancy, and patient's refusal to participate. Transient telemedicine technical difficulties (experienced for only one colposcopic session) also excluded participation by some eligible subjects. Clinician participants included a local colposcopist (family physician or nurse practitioner), a site expert (expert colposcopist at the rural site), and two distant experts (expert colposcopists at the central telemedicine center). The same three expert colposcopists equally rotated positions as site or distant experts for each session. No preclinical telecolposcopy image interpretation training was conducted.
The rural clinics represented two operational sites in the Medical College of Georgia Telemedicine Network. The network hub is located at the Medical College of Georgia Telemedicine Center in Augusta. The Medical College of Georgia Telemedicine Network uses an interactive color video and audio telecommunication system integrated with biomedical telemetry to consult with clinicians located in remote rural areas of the state. The bandwidth available for rapid motion images was 1.54 Mb/s. Video and audio signals were compressed and decompressed using a Video Telecom Mediamax CODEC. The audio CODEC provided two‐way simultaneous room audio with a band pass of 200 Hz to 3.3 kHz. All visits were recorded using a videotape system. The clinician and subject viewed one another on a 20‐inch Sony Trinitron color monitor. Each site had two such monitors, one to display the rural site image being transmitted and the other to view the image of the colposcopy expert at the telemedicine center. A Chinon CX 100 color charge coupled device camera was used for video conferencing and interviewing. A small charge coupled device camera was attached to the colposcope (DF Vasconcellos, Sao Paulo, Brazil) to transmit colposcopic images at varied levels of magnification.
The telecolposcopy trial was a prospective, blinded study design. After volunteering and signing the informed consent, subjects provided pertinent demographic information. Subjects were then introduced to the distant colposcopy expert via the telemedicine linkage. As per Medical College of Georgia Telemedicine policy, this initial encounter, the colposcopic examination, and postassessment discussion were recorded on VHS tapes at both the central and rural sites. After the site expert visualized the cervix with a vaginal speculum, a Papanicolaou smear was obtained if 3 months had transpired since the last collection. Otherwise, after a brief inspection of the cervix using the colposcope, 5% acetic acid was applied gently to the cervix and vaginal fornicies. Two cervigrams were taken for interpretation by cervicography evaluators (NTL Worldwide, Fenton, MO).23,24
A comprehensive colposcopic examination was then conducted except for the application of dilute Lugol's iodine solution and collection of cervical histologic specimens, which would have altered the cervical appearance before the local colposcopist's examination. The distant colposcopic expert and local colposcopist were blinded to this initial cross‐validation examination. The site expert recorded the findings of their examination including examination adequacy, colposcopic impression, biopsy intent and location, and anticipated management. Thereafter, the telemedicine colposcope camera was activated, and the local colposcopist entered the room to complete a colposcopic examination, including collection of cervical biopsies and endocervical curettage (ECC), if necessary. The distant expert primarily observed this examination without initiating communication with the local colposcopist, but the distant colposcopist was available for consultative purposes if the local colposcopist considered it necessary. The reasons for consultation were not predetermined. These two colposcopists then independently recorded the findings of their examination. All clinical observations were blinded between these two colposcopists except for certain items discussed when consultation was necessary. If the distant expert was not available in the central telemedicine center at the time of the examination, a videorecording of the local colposcopist's examination was observed at a later time. A third colposcopic expert, not present in the clinic or telemedicine center during the original examination, examined the videotape at a later date. The two distant colposcopists were blinded to all other colposcopist's observations.
Percent agreement including 95% confidence intervals for each pair of colposcopic impressions was calculated as a descriptive measure. Agreement of responses by the three colposcopists was evaluated for each pair of colposcopists using Cohen's κ statistic for dichotomous response categories (examination adequacy, decision to biopsy, and ECC necessity), and Bowker's test of symmetry for responses with more than two categories (cervical diagnosis). Examination adequacy (satisfactory or unsatisfactory) and management were compared using McNemar's test. The accuracy of colposcopists' diagnoses was determined by comparison with the histologic results (normal, cervical intraepithelial neoplasia [CIN] 1–3, or cancer). Four types of accuracy agreement were determined: 1) telemedicine accuracy — distant expert's impression versus site expert's impression; 2) clinician accuracy — local colposcopist's impression versus site expert's impression; 3) practical accuracy — local colposcopist's impression versus distant expert's impression; and 4) delayed accuracy — immediate distant expert's impression versus delayed second distant expert's impression. Each colposcopist's rating of examination adequacy, biopsy intent, decision to do ECC, and agreement with biopsy diagnosis were tabulated as percentages of the total number of examinations. Pairwise comparisons of these percentages among the local colposcopist, site expert, and two distant experts were made for patients evaluated by all four colposcopists, using McNemar's test with Bonferroni's adjustment to preserve the overall 95% confidence level. Comparisons of percentages in each response category for each pair of colposcopists were performed using McNemar's test. Agreement was compared among telemedicine, clinician, and practical modalities using McNemar's test with Bonferroni's adjustment for multiple comparisons to preserve the overall 95% confidence level.
A total of 264 subjects were enrolled in the trial and, depending upon the specific measure, matched responses ranged from 262 for ECC intent to 175 for colposcopic impression agreement with histology. The mean age of the subjects was 31.7 years, and the range was 18–78. Mean gravidity, parity, and abortive status were 2.4, 2.1, and 0.3, respectively. The subject's cytologic and histologic results are listed in Table 1.
Colposcopic adequacy and management parameters for each type of colposcopist were determined (Table 2). All four colposcopists reported colposcopic adequacy for 245 subjects. The distant experts who viewed primarily live examinations reported a significantly lower percentage of satisfactory examinations (60.0%) compared with the other colposcopists (P < .001, McNemar's test with Bonferroni's adjustment). None of the other colposcopists differed significantly in the percentage of satisfactory examinations reported. With respect to sampling, biopsy intent and ECC indication were reported by all four colposcopists for 238 and 239 subjects, respectively. The site experts recommended a significantly lower percentage of cervical biopsy (57.6%) and ECC (59.0%) compared with the other colposcopists (P < .001, McNemar's test with Bonferroni's adjustment). None of the other colposcopists differed significantly in the percentage of examinations for which biopsy and ECC were recommended. Local colposcopists consulted with the distant experts during 36.2% of colposcopic examinations.
We then determined colposcopists' responses to cervical biopsy or ECC necessity. There were 56 cases in which the local colposcopists obtained a biopsy, but the site expert indicated that no biopsy was necessary. The histologic results for these controversial biopsies were normal 51.9%, 38.9% for CIN 1, 5.6% for CIN 2, and 3.7% for CIN 3. There were no discordant biopsy intent cases if the experts determined a biopsy was necessary but the local colposcopist did not.
The evaluation of telecolposcopy must consider the accuracy or exclusive effects of the telecommunications system (telemedicine accuracy), the accuracy of the local clinician (clinician accuracy), overall accuracy of the telecolposcopy system as it exists (practical accuracy), and the temporal effects of viewing (delayed accuracy) (Table 3). With respect to colposcopic examination adequacy, clinician agreement (82.2%, κ = 0.51) was significantly better than that of telemedicine and practical agreement. Agreement of the need for cervical biopsy (84.1%, κ = 0.62; 86.9%, κ = 0.65) was significantly greater among the practical arrangement (P < .05, McNemar's test with Bonferroni's adjustment) than for telemedicine or clinician. Agreement on need for ECC differed significantly for each of the arrangements (P < .05, McNemar's test with Bonferroni's adjustment), with telemedicine agreement (69.1%, κ = 0.34) less than clinician agreement (74.5%, κ = 0.45), and less than practical agreement (86.9%, κ = 0.65). Colposcopic diagnosis agreement was significantly lower from the telemedicine perspective (52.0%) than for the other two arrangements (clinician agreement 65.6%; practical agreement 61.7%).
We also compared colposcopic impressions with cervical biopsy results for each specific type of colposcopist (Table 4). Diagnostic agreement was best for the local colposcopists (59.7%). κ values for colposcopic/histologic agreement ranged from 0.31 to 0.16.
The goal of telemedicine is to enable quality health care delivery to populations without readily available access to certain medical expertise. The efficacy of telecolposcopy is based on its diagnostic accuracy and subsequent patient management decisions. Telecolposcopy efficacy is contingent upon sufficient and accurate visual image transfer and communications to derive satisfactory outcomes. Efficacy is realized if telecolposcopy delivers care equivalent to in‐person traditional colposcopy. This study examined telecolposcopy compared with traditional colposcopy. We determined that the telecolposcopy diagnostic accuracy of distant experts was similar to that of onsite experts. Our cross validation study design was able to verify that colposcopic diagnoses are not impaired by the telecolposcopy process. Colposcopic agreement with histology was lowest for the distant expert who viewed a videotape version of the original transmission at a later time. Image resolution degradation by transfer to videotape may likely explain this difference. Synchronous viewing is preferred, but distant experts may not always be available for teleconsultation, as was found in a minority of cases. Colposcopic diagnoses are derived by considering magnified, transilluminated epithelial features.25 These colposcopic signs include blood vessel caliber, configuration and spacing, epithelial opacity, color and integrity, lesion shape and contour, and reaction to various transient contrast solutions. Telediagnosis is influenced by image focus, resolution, color, contrast, light intensity, and motion. The telemedicine system used in this trial provided sufficient technical qualities to maintain local telediagnostic accuracy from a distant site. Because transmission lines and hardware are improving constantly, telecolposcopy diagnosis accuracy should be maintained in the future.
Patient management decisions are based on cytology results, colposcopic appraisal, and colposcopically directed histology. Examination adequacy determines the necessity for endocervical sampling and further diagnostic evaluation by cervical conization. Our study demonstrated that the distant expert determination of examination adequacy may be impaired by telecolposcopy. Detection of the entire squamocolumnar junction and transformation zone can be a difficult psychomotor exercise. Colposcopists may have only a fleeting visualization of the squamocolumnar junction. A distant expert is unable to align or maneuver the cervix, and consequently, may not anticipate a quick view. If the distant expert has any uncertainty about the adequacy of the colposcopic examination, it would be prudent to assume that it is unsatisfactory to protect the patient's welfare. Therefore, any visual uncertainty may prompt a precautionary management decision provided the telemedicine expert's opinion supercedes that of the local colposcopist's. This study's finding may denote some overutilization of conization for women examined by telecolposcopy if adequacy discordance is not resolved by referral or acceptance of the local colposcopist's satisfactory finding. Further study of this important issue is warranted.
The decision to biopsy is of paramount importance for patient management. Local colposcopists and experts usually agreed on biopsy intent. Yet, there were a substantial number of cases in which the site experts disagreed with the local colposcopists and considered biopsy to be unnecessary. In half (52%) of these cases, biopsy was truly not required. In 39% of other cases, the biopsy indicated CIN 1. Most CIN 1 resolves spontaneously without treatment. Differentiating immature squamous metaplasia from CIN 1 can be challenging during colposcopy. The task is obviously not any easier when the patient is viewed by telecolposcopy. The expert likely considered the majority of these women to not have a significant cancer precursor, of which they did not. Moreover, because the site expert was unable to use Lugol's iodine solution and conduct an aggressive evaluation of the endocervical canal to prevent obscuring the local colposcopist's examination, the local colposcopist may have had a slight diagnostic advantage. Our concern is for the five (9%) cases (three cases for CIN 2, and two cases for CIN 3) in which high‐grade disease was diagnosed strictly by the local colposcopist's biopsy when the site expert considered it to be unnecessary. Whether these cases were in fact CIN 2/3 or histologic overcall is not known. Regardless, local clinician discretion and a lower biopsy intent threshold are imperative to maintain a high standard of care.
It is naive to think that availability invariably yields use. However, our study determined that teleconsultation was used for a substantial number of colposcopic examinations. The fact that our local colposcopists were more skilled than expected likely minimized the number of consultations which otherwise may equate to a patient referral. Furthermore, many of these women or a population with a greater percentage of abnormalities may have prompted an increase in the consultation rate. We are unsure of what percentage of this consultative interaction occurred by convenience or necessity. As such, the precise number of women who otherwise would have been referred to the colposcopy expert is uncertain. Yet, teleconsultation can improve patient care even for patients who would not normally be referred to a specialist.
In retrospect, our study encountered several unanticipated limitations. Our rural clinicians were more skilled than originally expected. Consequently, the attributes of telecolposcopy may be more profound and extensive if a greater dichotomy of colposcopic experience were to exist. In contrast, our experts had varied colposcopic skill. The greater the overall expertise, the more beneficial telecolposcopy will be for distant health care providers and their patients. Telecolposcopy utility will then vary accordingly. Our analysis of the necessity for histologic sampling was likely influenced positively by communication between the local colposcopists and distant experts (practical model). In comparison, colposcopic diagnosis and satisfactory examination agreement were much less for this same model in part because of an attempt to not share these findings between colposcopists. Furthermore, because different pathologists interpreted the specimens, instead of a centralized pathology group able to provide a consensus diagnosis, the agreement of colposcopic impressions and histologic results may have demonstrated a greater variability than that seen in other trials.
Domestic interstate telecolposcopy is limited in scope by licensure and reimbursement obstacles in addition to general unavailability. Federal legislation designed to open the access of telemedicine to women confronted with geographic barriers to health care is needed desperately. As technology improves and hardware costs decrease, telecolposcopy use may expand in the United States. Alternatively, international telecolposcopy is free to grow without regulatory and financial burden. In fact, need may be greater for certain areas of the world with a paucity of skilled colposcopists. Otherwise, telecolposcopy may facilitate distant preceptorship assistance for novice colposcopists, remedial training for colposcopists in need of an update or more advanced exposure, and colposcopy quality control for clinical trials.26 Further research of telecolposcopy with special respect to cost‐effectiveness/cost‐benefit analyses, various telecolposcopy systems, and their psychological impact on women may help advance this important health care option.
1. Carey P, Gjerdinger DK. Follow-up of abnormal Papanicolaou smears among women of different races. J Fam Pract 1993;37:583–7.
2. Baquet CR, Horm JW, Gibbs T, Greenwald P. Socioeconomic factors and cancer incidence among blacks and whites. J Natl Cancer Inst 1991;83:551–7.
3. Freeman HP. Cancer in the socioeconomically disadvantaged. CA Cancer J Clin 1989;39:266–88.
4. Friedell GH, Tucker TC, McManmon E. Incidence of dysplasia and carcinoma of the uterine cervix in an Appalachian population. J Natl Cancer Inst 1992;84:1030–2.
5. Centers for Disease Control. Breast and cervical cancer surveillance. United States, 1973–1987. MMWR 1992; 41:9.
6. Ackerman SP, Brackbill RM, Bewerse BA, Cheal NE, Sanderson LM. Cancer screening behaviors among U.S. women: Breast cancer, 1987–1989, and cervical cancer 1988–1989. MMWR 1992;41:17–34.
7. Centers for Disease Control. Cervical cancer control —Rhode Island. MMWR 1989;38:659–62.
8. Laedtke TW, Dignan M. Compliance with therapy for cervical dysplasia among women of low socioeconomic status. So Med J 1992;85:5–8.
9. Martin PL. How preventable is invasive cervical cancer? A community study of preventable factors. Am J Obstet Gynecol 1972;113:541–8.
10. Ferris DG. Colposcopy. In: Ferris DG, ed. Primary care: Clinics in office practice. Office procedures. Philadelphia, PA: WB Saunders; 1997, 241–67.
11. Gordon P, Hatch K. Survey of colposcopy practices by obstetrician/gynecologists. J Reprod Med 1992;37:861–3.
12. Bashshur RL. Telemedicine and medical care. In: Bashshur RL, Armstrong PA, Youssef ZI, eds. Telemedicine: Explorations in the use of telecommunications in health care. Springfield, IL: Charles C Thomas; 1975.
13. Andrus WS, Bird KT. Teleradiology: Evolution through bias to reality. Chest 1972;62:655–7.
14. Preston J, Brown FW, Hartley B. Using telemedicine to improve health care in distant areas. Hosp Comm Psych 1992;43:25–32.
15. Wittson CL, Benschotre R. Two-way television: Helping the medical center reach out. Am J Psychiat 1972;129:136–9.
16. Armstrong PA, Youssef ZI, Bashshur RL. Telemedicine in the United States: A summary of operational programs. In: Bashshur RL, Armstrong PA, Youssef ZI, eds. Telemedicine: Explorations in the use of telecommunications in health care. Springfield, IL: Charles C Thomas, 1975: 309–46.
17. Batnitzky S, Rosenthal SJ, Siegel EL, Wetzel LH, Murphey MD, Cox GG, et al. Teleradiology: An assessment. Radiology 1990;177:11–7.
18. Franken EA, Driscoll CE, Berbaum KS, Smith WL, Sato Y, Steinkraus LW, et al. Teleradiology for a family practice center. JAMA 1989;261:3014–5.
19. Hudson HE, Parker EB. Medical communication in Alaska by satellite. N Engl J Med 1973;289:1351–6.
20. Sobczyk WL, Solinger RE, Rees AH, Elbl F. Transtelephonic echocardiography: Successful use in a tertiary pediatric referral center. J Pediatr 1993;122:584–8.
21. Weinstein RS, Bloom KJ, Rozek LS. Telepathology: Long-distance diagnosis. Am J Clin Pathol 1989;91:539–42.
22. Harper DM, Moncour MM, Harper WH, Burke GC, Rasmussen CA, Mumford MC. The technical performance and clinical feasibility of telecolposcopy. J Fam Pract 2000;49:623–7.
23. Ferris DG. Cervicography — An adjunct to Papanicolaou screening. Am Fam Phys 1994;50:363–70.
24. Ferris DG, Payne P, Frisch LE, diPaola FM, Petry LJ. Cervicography: Adjunctive cervical cancer screening by primary care clinicians. J Fam Pract 1993;37:158–64.
25. Ferris DG, Greenberg MD. Reid's colposcopic index. J Fam Pract 1994;39:160–9.
26. Ferris DG, Cox JT, Burke L, Litaker MS, Harper DM, Campion MJ, et al. Colposcopy quality control: Establishing colposcopy criterion standards for the National Cancer Institute ALTS trial using cervigrams. J Lower Genital Tract Dis 1998;4:195–203.