There are seven indications for colposcopy by abnormal cervical cytology established in the 2006 and updated 2012 Consensus Guidelines [4, 22]. Colposcopy is not restricted to abnormal cytology alone and may be used for common indications (Table 15.1).

Diethylstilbestrol (DES) was used from about 1940 to 1970 to prevent adverse pregnancy outcomes in women with previous history of miscarriage. In 1971, Herbst published a report linking in utero DES exposure to vaginal clear cell adenocarcinoma, which prompted the FDA to place a black box warning label later that year [23]. Today, women with in utero DES exposure are approximately 40+ years or older.

While there are no absolute contraindications to performing colposcopy, it may be limited in the setting of active pelvic infections (i.e., cervico-vaginitis) such as the classic strawberry spots representing dilated capillary loops in Trichomoniasis or thick vaginal discharge of Candidiasis. Late term pregnancy may yield little treatment information in the absence of gross cervical cancer. Severe vulvo-vaginal atrophy may also yield poor diagnostic information.

The main objective of colposcopy is to prevent cervical cancer by finding cancer precursors using directed biopsies. To accomplish this task, colposcopy requires several essential steps that include applying appropriate indications, maximizing visualization, forming an accurate impression, performing directed biopsies, and establishing appropriate follow-up.

While there is some question to its overall effectiveness and accuracy (see section “Effectiveness”), colposcopy performance is dependent upon provider experience. The continued decline in smoking rates in the USA and introduction of the HPV vaccine is anticipated to erode cervical cancer incidence. Most recently, screening intervals have been extended from traditional annual Pap smears to once every 3–5 years in most women. While this is sure to dwindle the number of colposcopies performed today, it places greater emphasis on establishing the correct diagnosis by colposcopy.

Colposcopy is an office-based procedure, so there are no credentialing requirements by hospitals, insurance companies, or malpractice carriers. The American College of Obstetrics and Gynecology (ACOG) suggests performing 30–50 colposcopies under supervision with an additional 80–100 exams to achieve proficiency [24]. The ASCCP estimates that a provider performs 25–100 colposcopies with a minimum of 10 high grade lesions to become competent. In a 30-question survey sent to 485 Ob/Gyn and Family Practice residency program directors, most programs did not meet the minimum number of procedures for its residents [25]. In a study at a university-based Ob/Gyn residency program, increasing level of training correlated with improvement in accurate impression. Overall accuracy was only 32 %, however, and nurse practitioners outperformed resident physicians [26].

As the number of colposcopy procedures diminishes, it is uncertain whether colposcopy will be practiced widely or become concentrated among providers with keen interest and expertise. For those motivated to perform colposcopy, one might consider obtaining additional teaching at postgraduate curricula offered through various organizations of the ASCCP, ACOG, and American Academy of Family Physicians.

Colposcopy is considered the reference standard for which cervical cytology is measured. Likewise, histology measures the performance of colposcopy as a diagnostic tool. In 1988, a meta-analysis by Mitchell estimated colposcopy performance at a mean-weighted sensitivity of 96 % and specificity of 48 % comparing all abnormal findings to normal tissue. When comparing only high grade results to low grade abnormalities, sensitivity remained high at 85 % with specificity of 69 % [27].

More recently, it has become recognized that studies often overestimate colposcopy performance due to inherent biases of study design. The methods used to obtain tissue histology are variable (e.g., conization versus colposcopic-directed biopsies) and influence the rate of CIN diagnoses. In many studies, verification bias is introduced when only a select group of women at risk for high grade dysplasia undergo biopsy or treatment. Therefore, calculating the true values for sensitivity and specificity is difficult.

Among experienced colposcopists, there seems to be good interobserver agreement for normal epithelium, high grade dysplasia, and cancer. There is more variability for lesser grade lesions of CIN-1-2 [28]. This was confirmed during the ASC-US and LSIL Triage Study (ALTS) trial, when three experienced reviewers examined digitized images with poor-to-fair agreement on impression [29]. Even for high grade lesions, colposcopy may miss over half of CIN-3 initially [30].

A meta-analysis of 32 studies including 7,873 biopsies examined the diagnostic performance of colposcopic-directed biopsies to that of excisional cervical conization or hysterectomy. Limiting the analysis to four pooled studies with lower rates of positive biopsies provided a more accurate reflection of colposcopic performance to a sensitivity of 81.4 % and specificity of 63.3 % [31]. However, Stoler reviewed the findings of the FUTURE I, II, and III trials in 2011 revealing that LEEP conization diagnosed twice as many CIN-2-3 disease versus same day colposcopic-directed biopsies [32].

It is estimated that over half of missed high grade lesions are due to sampling error [33]. Intuitively, performing more biopsies as opposed to a single biopsy decreases the likelihood of missing a high grade lesion. In the ALTS trial, where only women with ASC-US or LSIL cytologic abnormalities were enrolled, there were 408 women with CIN-3 or cancer of which 69.9 % had a colposcopic-directed biopsy of CIN-2 or greater. Sensitivity increased for procedures with two or greater biopsies [34].

Whether it is better to take multiple biopsies from the most abnormal area, biopsy every abnormal finding, or consider “random” biopsies is uncertain [13]. Belinson and colleagues reported the results of their cross-sectional study Shanxi Province Cervical Cancer Screening Study (SPOCCS I) in 2001 (Table 15.2). Among 1,997 women enrolled, the cervix was divided into four quadrants. “Random” biopsies were performed at the squamo-columnar junction (SCJ) if there was no identifiable lesion, and ECC was performed. Among 86 women with CIN-2 or greater, 16 had normal findings in all four quadrants (12 CIN-2, 4 CIN-3) [35]. In their follow-up study SPOCCS II, they analyzed 364 patients with CIN-2 or greater and a satisfactory colposcopy. The diagnosis of CIN-2 or greater was made by colposcopic-directed biopsy in only 208 (57.1 %). The remaining cases were diagnosed by random biopsy in 136 (37.4 %) and ECC in 20 (5.5 %). The yield for random biopsy was greater at 17.6 % when cytology was high grade versus only 2.8 % when cytology was low grade [36].

The utility of ECC has been debated especially in women with low grade abnormalities. In the ALTS trial, the ECC was positive in only 1 % (10/1,119) of women with a negative colposcopy and biopsy. High grade abnormalities of CIN-2 or greater were found by ECC in 3.7 % of 1,119 exams versus 21.7 % by biopsy. Omitting ECC would risk missing 1.1 % (7/653) of women with CIN-2 or greater. In women <40 years old, only 2 % (7/312) were diagnosed with CIN-2 or greater with low grade biopsy versus 13 % (3/23) women >40 years old [37].

The disparity between colposcopic impression and histology may be attributed to the provider’s experience in forming an accurate impression and/or performing an adequate biopsy. This does not fully explain the poor performance among experienced colposcopists. It is likely that there are some high grade lesions that do not present with classic findings. Jeronimo and colleagues proposed that different subtypes of HPV may present with different aceto-white epithelium (AWE). In the ALTS trial, HPV-16 was associated with AWE regardless of whether the lesion was low or high grade. AWE did not seem to be associated with more HPV types or increased number of lesions, suggesting that other subtypes of HPV were less likely to appear [38].

Among 144 cases of CIN-2-3 in the SPOCCS I study, 111 were diagnosed by colposcopic-directed biopsy with impression of CIN-1 or greater. The average epithelial thickness was 321 μm. Among the remaining 33 cases diagnosed by random biopsy due to normal appearance, the average epithelial thickness was a mere 184 μm [39]. Given that some high grade lesions may not demonstrate typical characteristics, it may be reasonable to perform random biopsies and ECC especially in nonpregnant women with high grade cervical cytology [40].

Prior to the colposcopy procedure, the provider should review the indications for colposcopy to ensure proper guidelines are met. If the Pap smear was performed at another facility, then it is beneficial to request the cytology slides for the pathologist reading the biopsies. Most pathologists will want to corroborate the biopsy results with the cytology. If the patient has any signs or symptoms of cervico-vaginitis, it is beneficial to postpone the visit until this has resolved. Any vaginal creams or douching should be stopped at least 1 day in advance. Relative contraindications to colposcopy include bleeding disorders and anticoagulation therapy so strategies to control bleeding (e.g., discontinuation of anticoagulation) should be implemented prior to the procedure. Colposcopy may be performed in menses if there is light flow.

It is helpful to obtain pertinent patient medical information including last menstrual period, pregnancy and delivery history, and any difficulties with pelvic exams. Prior Pap smear or colposcopy results, previous ablation or conization procedures, and HPV immunization status are important. Modifiable risk factors including previous sexually transmitted diseases, number of sexual partners and practices, and smoking provide opportunity for positive messaging. HIV status, steroid use, uncontrolled diabetes, and organ transplant history provide additional risk factor information. Most importantly, compliance and social situation should be taken into account especially when establishing a follow-up plan.

Check a spot urine pregnancy test in all premenopausal women unless pregnancy status can be confidently assured. The provider should obtain informed consent. The colposcope and any video equipment and monitors should be turned on to avoid fumbling with plugs and cables during the procedure. Clean all eyepieces, handles, and dials between exams.