• Systematic review;
  • vaginal vault smear


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

Background  Vaginal vault smears are used to detect persisting neoplasia of the lower genital tract after hysterectomy. Recent data suggest both widespread use and uncertain evidence of their effectiveness.

Objectives  To identify and synthesise evidence on the use and effectiveness of vaginal vault smears and to assess the quality.

Search strategy  ‘vault smear’ OR ‘vaginal vault smear’ OR ‘cervical vault smear’ OR (‘Hysterectomy’) AND (‘Follow up’ OR ‘Smear’).

Selection criteria  Primary research, women who had a hysterectomy and were followed up by vault cytology.

Data collection and analysis  Systematic search (eight electronic databases), supplemented by contact with experts and review of bibliographies. Two independent reviewers determined eligibility/validity and extracted data concerning test performance characteristics. Quality was assessed according to the established criteria.

Results  Of 441 unique references, only 19 were suitable. Quality of studies varied considerably and few were of ‘high’ methodological quality. Studies were geographically diverse, and were published over more than 40 years in 16 journals. From the higher scoring papers, there were 11 659 hysterectomies [6546, benign; 76, cervical intraepithelial neoplasia (CIN) I/CIN II; 5037, CIN III]. Proportions of abnormal vault smears and abnormal biopsies during follow up increased with worsening histology at hysterectomy (P < 0.0001 and P= 0.0001). There was only one report of vaginal cancer subsequent to hysterectomy for CIN and insufficient data to allow for reliable meta-analysis.

Conclusions  Vault smears cause anxiety, consume resources and their value is largely unproven. Inconsistency of study design and limited methodological quality means that the value of vault smears could not be established. High-quality research is required to ensure that the guidelines are evidence based.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

It has been suggested that too many vaginal vault smears are being undertaken both in the USA1 and the UK.2 The value of this test, in the follow up of women after hysterectomy, has not been established and guidelines are based on consensus opinion.

By the age of 65 years, the proportion of women having a hysterectomy reaches a third in the USA3 and 20% in the UK (60 000 procedures annually).4 Subtotal surgery (sparing the cervix) is undertaken in less than 3% of the hysterectomies performed in the UK.5 Total hysterectomy includes removal of the cervix uteri leaving the vagina as a blind ending pouch; since the cervix has been removed, there is no possibility of the development of a primary cervical cancer and thus no indication for routine cervical screening. Papanicolaou (Pap) smear tests6 of the vaginal vault are a means of detecting recurrent invasive or preinvasive disease of the lower female genital tract in women who do not have a cervix uteri.7

Benign indications (e.g. excessive bleeding or fibroid disease) account for more than 90% of hysterectomies in the UK.5 The proportion of hysterectomies performed for cancer or precancerous lesions has been reported to range from 6 to 10%.8 After surgery, current policy in the UK is to follow up with vaginal vault smears those women who have had a hysterectomy for high-grade preinvasive disease of the cervix.9 However, recent work suggests that practice varies among clinicians, and many additional women may receive follow-up smears.2

The purpose of undertaking vault smears on asymptomatic women who had no abnormal cervical pathology at hysterectomy is to screen for vaginal intraepithelial neoplasia (VAIN) and prevent vaginal cancer. However, VAIN is 150 times less common than cervical intraepithelial neoplasia (CIN), and vaginal cancer is one of the rarest gynaecological malignancies (0.7 per 100 000 women in UK).10 Recognised risk factors for VAIN include: CIN, immunosupression, genital warts/human papilloma virus infection, radiation therapy and smoking.10 Besides VAIN, the only group of women appearing to be at increased risk of primary vaginal cancer are those whose mothers took diethylstilbestrol during pregnancy.11

Opinion regarding the role of the vault smear has changed over time. In the 1950s, there was enthusiasm for follow up after hysterectomies showing the presence of carcinoma in situ.12,13 Since the 1990s, this strategy has been questioned, and it has been suggested that too many vault smears may be being undertaken.1,3 In 2001, it was estimated that approximately 11 million vault smears per annum were being performed ‘unnecessarily’, of the 12.5 million women who had a hysterectomy and were continuing to have Pap smears, in the USA; the study concluded that only 7–15% of women should require vault smears after hysterectomy (or cervical smears after subtotal/supracervical hysterectomy). If more than 80% of vault smears are unnecessary, this represents a huge waste of resources and may be the cause of unwarranted anxiety and inconvenience for women.14

In 1996, Pearce et al.15 reported the results of screening a large cohort of women who had previously undergone hysterectomy for benign indications. The positive predictive value of the vault smear, as a means of screening women who have undergone hysterectomy for benign reasons is low16, and 10 years ago it was recommended that ‘the use of the Pap smear after hysterectomy, for benign disease, should become a thing of the past’.17

This study aimed to establish the evidence base for the use of vaginal vault smears subsequent to hysterectomy for benign or precancerous conditions.


This systematic review16 aimed to identify all the studies that either evaluated the vault smear test, or which reported the follow up of a series of patients treated by hysterectomy for reasons other than malignancy, and contained data to enable the value of vault smears to be estimated.

Searches were performed on the following electronic databases: Medline (from 1966), Embase (from 1980), Cumulative Index to Nursing and Allied Health Literature (CINHAL) (from 1982), CancerLit (from 1960) NHS Centre for Reviews and Dissemination, Database of Abstracts of Reviews of Effectiveness, Turning Research into Practice (from 1986), Cochrane Collaboration Database, and Web of Science. The search terms comprised ‘vault smear’ OR ‘vaginal vault smear’ OR ‘cervical vault smear’ OR (‘Hysterectomy’ AND (‘Follow up’ OR ‘Smear’)). The specific search strategies (i.e. text words or index terms) were varied according to the search engine (Appendix S1). Citations were downloaded into Reference Manager to facilitate identification of duplicate entries and for ease of handling. Electronic searching was supplemented by asking authors of papers relevant to the subject of this review to assist with the identification of papers that predated electronic search facilities or for relevant unpublished data.

Two authors independently scanned the titles and, where available, the abstracts of all articles identified by the electronic searches, excluding those that had no relevance. Complete copies of all the remaining references, and those where a decision could not be made on the basis of the title or abstract alone, were requested, and two authors independently reviewed these to identify all eligible publications. Any disagreements were resolved by discussion with a third reviewer.

Papers were ‘eligible’ for inclusion if they reported on a population of women who had undergone a hysterectomy and at least some of the population had vault smear tests. Case reports or expert opinion were excluded. Where the study population comprised a cohort of women with abnormal smears and where the number of women having had a hysterectomy was not stated (i.e. the rate of abnormal smears subsequent to hysterectomy could not be established), the publication was also excluded. All the papers selected as eligible for inclusion had their bibliographies reviewed to identify any further papers of relevance to the review. Eligible papers were considered in full by two independent reviewers using a standardised pro-forma (Appendix S2) to determine the relevance of the papers to the aims of this review. Two reviewers undertook independent data abstraction, and any discrepancies in data abstraction were discussed and consensus reached.

Included papers were scored, by two independent reviewers, for methodological quality. The scoring system was modified from the validated NHS Critical Appraisal Skills Programme (CASP) tool for assessing a diagnostic test.18 (Appendix S3) Again, any disagreement between the reviewers was resolved by consensus. The quality score (QS)19 was made up of points being awarded for a ‘Yes’ response to different aspects of the methodology, resulting in a possible score of between 0 and 10 for each. Papers having a QS greater than six were deemed to be of a ‘good’ methodological quality, however, even validated quality scoring systems can be criticised,20 so the data from poorer quality papers were retained for comparison, and three reviewers considered the methodological aspects of all studies at length.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

The literature searches

The systematic review identified 526 citations in total: 453 from electronic databases, seven from experts in the field and 66 from the bibliographies of included papers. There were 441 unique references once duplicates were excluded (371 from electronic sources, four from experts, 66 from bibliographies). Of these, 319 were excluded on the basis of title and abstract alone; a further 102 were excluded at eligibility assessment. Thus, 20 research publications were eligible for inclusion. Of these, two reported the same piece of research and were published by the same group in the same year;21,22 the more comprehensive was included.22 This gave a total of 19 studies:13 from electronic searches, one from experts and five from bibliographies. Table 1 summarises the source of publications.

Table 1.  Evaluation of references by source
Source of papersTotal number of referencesNumber of unique references, n (%)Excluded papersIncluded referencesSensitivity of sourcePrecision of source
  1. NHSCRD, NHS Centre for Reviews and Dissemination; TRIP, Turning Research into Practice; CINHAL, Cumulative Index of Nursing and Allied Health Literature; WOS, Web of Science.

Medline147147 (33.3)13890.060.47
Cancerlit6441 (9.3)3920.050.11
EmBase3111 (2.5)920.180.11
NHSCRD117117 (26.5)117000
Cochrane4435 (7.9)35000
TRIP4818 (4.1)18000
CINHAL22 (0.5)2000
WOS20 (0.0)0000
Bibliographies6666 (15.0)6150.080.26
Experts74 (0.9)310.250.05

All the identified studies used a form of cohort design and one had a control group23. The included papers were published between 1963 and 2000 in 16 different journals, with 11 published prior to 1990 (58%) (Table 2).

Table 2.  Results table summarising all the papers finally included in the review
Study (author; region; method; year)Study populationResults and authors conclusionsReviewers comments (QS = 0–10)
Number of womenInclusion details
  1. CEIN, ceruical glandular intraepithelial neoplasia; CIS, carcinoma in situ; LCEIN, lowgrade ceruical glandular intraepithelial neoplasia; N, number of women in the original study who were followed up after hysterectomy by at least one vault smear test; n1, number of abnormal vault smears in N; n2, number of cases of biopsy-proven vaginal dysplasia; n3, number of cases of invasive cancer; PPV, positive predictive value; VAIN, vaginal intraepithelial neoplasia.

Berget et al.;24 County of Maribo, Denmark; Prospective population cohort study; 1972243 had a hysterectomy subsequent to a cervical abnormality: 8, slight/moderate atypia; 206, CIS; 29, stage IA.Followed up from 2–30 months by vault smear plus clinical examination and questionnaire. Excludes 13 cases of pre-existing invasive disease237 followed up from 2 to 30 months by vault smear plus clinical examination and questionnaire. n1 = 3: 4 actual abnormal smears but 1 awaiting data so excluded. n2 = 3: carcinoma in situ 2/206, stage 1a 1/29. 16/237 women regretted hysterectomy. Therapeutic conisation may be more appropriate for young women with stage 0.Follow up was short term only thus limiting usefulness of data. Incomplete information regarding proportion of patient followed up. Possible that study underestimates true event rate. QS = 7.5.
N= 237 had vault smears
Boyes and Worth;25 British Columbia, Canada; Cohort study; 1970Population offered free cytology service.70% of population had at least one smear during study period: 1–5 years, 1837; 6–15 years, 993; >16 years, 19n2 = 24: 12, recurrent; 12, new cases. 3/24, invasive carcinoma; 19/24, vaginal CIS; 2/24, occult invasive disease. 14 within 5 years, 22 within 10 years, all within 15 years. n2 = 4: detected 1–5 years, n3 = 1: detected at 20 months. Hysterectomy does not mean the end of follow up as there is a significant risk of the development of subsequent disease in the vaginal vault. Patients require careful and prolonged cytological follow up for the rest of their lives.An unknown number had a hysterectomy. Large cohort study. No definition of how recurrent or new disease is distinguished. Authors’ conclusions were influenced by higher rates of subsequent disease observed in subgroups with microinvasion and occult disease. QS = 8.
N= 2849 had vault smears.
N= 159 followed up after microinvasive disease detected
Fawdry;26 UK, SE Scotland; retrospective cohort from cytology records; 19841062 women had hysterectomies for confirmed carcinoma in situ or severe dysplasia.1035, CIS; 27, severe; dysplasia. Follow up: 253, regular vault smears; 557, had some vault smearsn1 = 64: but 62 were only ‘grade 2 smears’, 1 = unknown, 1 = presented symptomatically. n2 = 12: presented at 1 = 8 months, 8 = <12 months, 2 = 1–2 years. n3 = 1: with improved management it is believed that early ‘recurrences’ will become less evident. However, at least two smears in the 2 years after hysterectomy seems essential. Subsequently, for most, the very low detection rate does not justify an annual hospital visit with its attendant anxieties. Such women should attend for annual cytology at any convenient clinic.Comprehensive study, focusing on the follow up of CIS. Losses to follow up may underestimate the true event rate. QS = 7.5.
N= 810 had vault smears.
Gemmell et al.;22 UK, Tayside region of Scotland; retrospective review (cohort); 1990341 had a hysterectomy for CIN III.All asked to attend for regular vault smears for at least 10 yearsn1 = 9; n2 = 3: at 9 and 46 months and 16 years. Incidence of VAIN following hysterectomy for CIN III was 0.91% … 75% of abnormal vault smears reverted to normal without any active intervention … Little evidence to justify screening women who have undergone hysterectomy for CIN III any more frequently than the general population. Follow up of women after hysterectomy for CIN III should involve 6 monthly smears for the first year with a third smear at 2 years. If these smears are normal, the woman could revert to the national screening programme.Well-designed study with sufficient follow up. Losses to follow up have been excluded from the denominator and may result in an overestimate of the incidence of disease. QS = 9.
N= 219 had vault smears.
Halberg et al.;27 Denmark; cohort (unclear if retrospective or prospective); 196910 200 women undergoing a gynaecological examination. n1 = 4: ‘atypical cells’ at 2, 5, 5, 6 years. n3 = 1: invasive carcinoma with metastases at 18 months posthysterectomy. Need improved measures of diagnosis.Study population, duration and frequency of follow up incompletely described. Outcome of patients with atypia not documented. QS = 5.
N= 49 had a hysterectomy for carcinoma in situ and were followed up by vault cytology.
Hellberg and Nilsson;28 Sweden; retrospective review of records; 198790 000 fertile women served by a Swedish hospital.Followed up with colposcopy and cytology twice a year for 2 years then annually. Average duration of follow up = 10.3 years.n2 = 4: 2 ‘within 1 year’, 2 at 1–2 years. 1, CIN I; 2, CIN II; 1, CIN III. n3 = 0: the cure rate for hysterectomy is 99% (152/154). The results of conisation are comparable with those for hysterectomy. Disease may recur after as long as 15–20 years, and patients therefore need to be regularly followed up probably for a minimum of 10 years.Study aimed to compare outcomes from conisation, cryosurgery and hysterectomy. Outcomes by initial histological diagnosis and completeness of follow up not documented for subgroup treated by hysterectomy. Conclusions relate more to follow up after conisation or destructive treatments of CIN, rather than after hysterectomy–data does not support the authors’ conclusions. QS = 3.5.
N= 154 Had CIN II or worse at hysterectomy and had vault smears.
Kalogirou et al.;29 Greece; cohort study; 1997993 women undergoing hysterectomy for CIN III.All had CIN and completed 10 years follow up: 6 monthly for first year, annually thereafter.n1 = 210; n2 = 41: VAIN. Mean age at developing VAIN is 57 (35–75 years). <1 year since hysterectomy = 6; 1–2 years = 12; 2–5 years = 20; >5 years =A 3 (up to 10 years). Stage at detection: VAIN I = 4, VAIN II = 10, VAIN III = 27. In 21 (51%) cases, the grade corresponded to original cervical lesion. Incidence of VAIN in women who had completed 10 years follow up was 5.1%. Vault cytology provides a suitable and acceptable test for detection … and is therefore indicated in follow up after hysterectomy for CIN. Cost must be justified in terms of benefit … remember that they (vault smears) still cause considerable stress and anxiety to women … smears should be performed every 6 months for the first 2 years then annually to 5 years. Follow up should include colposcopic review on each examination … The highest incidence of VAIN is found in the first 2 years after hysterectomy; after 2 years, the incidence similar to that of general prehysterectomy population–0.7/1000. Patients with history of CIN before hysterectomy represent a higher risk group.Appropriately designed study with good follow up. Limited information about characteristics of population. Lack of clarity as to whether entire study population was CIN III. QS = 7.5.
N= 793 had vault smears.
Kirkup et al.;30 UK, Sheffield; retrospective cohort; 1979N= 112 women had hysterectomy for CIN III and all were followed up with vault smearsFollowed up for a minimum of four and where the first two postoperative smears were normal.n1 = 0: recurrence of new lesions no higher than that in the general population. Our findings suggest that residual CIN after treatment should be detected by the first two postoperative smears. New lesions of CIN may occur but not more commonly than in the general populations and that the sudden occurrence of a possible ‘second type’ carcinoma … is rare.A highly selected population. Characteristics of population and number of excluded cases not provided. QS = 5.5.
Kurian and al-Nafussi;31 UK, Edinburgh; retrospective cohort of computerised records; 1999N= 34 had a hysterectomy for glandular neoplasia of cervix and were followed up by cytology.Women identified from pathology department files, with glandular dysplasia, in situ and invasive disease and treated by hysterectomy.n1 and n2 = unknown. No recurrent abnormal glandular cells on follow up for patients with LCGIN or microinvasive disease. CGIN is a precursor of invasive adenocarcinoma. Important to ensure adequate free margins for women with glandular precursor lesions not treated by hysterectomy.Study included 80 women treated by diathermy loop excision. Unable to disentangle those women who had a hysterectomy from the rest of the cohort. Time to identification of CIN not stated. QS = 5.
Liukko et al.;32 Finland, Turku; retrospective cohort, review of case notes; 1978245 women with CIS treated at a University Hospital.Followed up for 5–10 years.n2 = 9: two dysplasia, four CIS, three invasive on biopsy. Recurrences all occurred between 3 months and 4 years. Recurrence may also be a new epithelial lesion on the basis of a multifocal mechanism of production. Cytologic and colposcopic follow up of these patients is thus necessary.Lack of clarity concerning subset of study population who were treated by hysterectomy. QS = 3.5.
N= 160 had vault smears.
McIndoe and Green;33 New Zealand, Aukland; retrospective cohort, case notes review; 1969539 women with cervical CIS.Followed up 1–17 years (6.5 mean)n1 = 9; n2 = 4: at 0, 6, 7, 7 years. Our conservative attitude to diagnosis and treatment of CIS is reinforced by the rarity of invasive vaginal cancer in these cases. No more … than periodic clinical, cytological and colposcopic examination is necessary.There is a lack of clarity with respect to study population and timing of abnormal smear results. QS = 3.5.
N= 175 had vault smears.
McIndoe et al.;34 New Zealand, Aukland; retrospective cohort, case notes review; 1984N= 250 women with CIS had a hysterectomy and all were followed up with vault smears.Follow up for 5–28 years at 3, 6 months and then annually.n1 = 33: at 2 years (13.2%); n3 = 8: cases of invasive disease, one at 1 year, two at 1–5 years, one at 5–10 years and four at over 10 years. CIS cervix has significant invasive potential. Importance of observing patients for a long period is apparent. Patients with continuing abnormal cytology … are 24.8 times more likely to develop invasive carcinoma.Watch and wait policy and lack of intervention when abnormal smears occurred may have led to a higher than expected rate of invasive disease. Same series as 1969 paper. QS = 5.5.
Michalkiewicz et al.;35 Poznan, Poland; retrospective cohort, case notes review; 1963701 Women with epithelia dysplasia.Follow up 3 monthly for 3 years then 6 monthly for 5 years. Dysplasia, 12; carcinoma in situ, 160.n2 = 2 recurrences. 2/160 = 1.25% after carcinoma in situ (11 and 22 months). The most frequent cause of recurrence seemed to be insufficiently radical treatment … The rarity of recurrence does not contraindicate conservative methods of treatment … Of the utmost importance is thorough and long-term follow up.Not possible to determine grade of dysplasia. QS = 6.
N= 172 had vault smears.
Miller et al.;36 USA; retrospective cohort, review of records; 19873008 black women seen in a cancer screening clinic.Followed up by vault smears annually. Time from hysterectomy ranged from 1 year to >30 years, 78% >5 years previously.n2 = 1: vaginal dysplasia. (0.13%); n3 = 0: the absence of carcinoma in the posthysterectomy group suggests that although an annual physical examination is indicated, the time interval between performances of the Pap test may be lengthened. If one asymptomatic cancer had been detected, the $10 000 cost would have been justified.Incomplete information on follow up. Possible that all subsequent cases not traced or included. QS = 5.
N= 775 hysterectomy for ‘benign’ reasons and had vault smears.
Pearce et al.;15 New Orleans, USA; retrospective cohort; 19966265 women who had vaginal smears at a large inner city charity hospital.83% Black, 9% White, 5% Hispanic, 3% other race. The number of women with benign indications for hysterectomy was estimated from a sample of 150 women.n1 = 79; n2 = 5; n3 = 0. Positive predictive value of pap smear for detecting vaginal cancer, 0%. For detecting VAIN, 6.3%. On average, 19 years from hysterectomy to first abnormal smear. Because of the low prevalence of disease and poor PPV of the test, periodic, routine screening by vaginal Pap smears is probably not necessary for women who had hysterectomy for benign disease. Vault smears should be considered only for women with a history of cancer of the genital tract or CIN III because they have increased risk of disease.Hysterectomy number is an estimate based on review of only 150 records. Exact follow up and length of follow up not clear. No information on time from hysterectomy to each event. QS = 7.
N= 5682 had a hysterectomy for ‘benign disease’ and had vault smears. 99%, low socioeconomic class.
Piscitelli et al.;37 North Carolina, USA; retrospective cohort, case notes review; 1995N= 697 hysterectomy for benign disease and had vault smears. 630, had normal cervical smears prehysterectomy; 63, abnormal smears prehysterectomy.Followed up by vault smears for an average of 13.7 years. 1266 vault smears total (average 1.8 per patient).n1 = 33: slight atypia, two; mild dysplasia, eight; moderate, two; severe, 1. 25 were found at the initial examination (0.25–36 years). Eight were found at a subsequent smear (2.25–45 years). n2 = 2: the extremely low incidence of vaginal carcinoma combined with the lack of evidence supporting the effectiveness of screening patients after hysterectomy suggests that less screening may be more desirable.One of the few papers where the stated aim was to determine the relevance of vault smear tests. Incomplete description of the frequency and reasons for vaults smears. QS = 6.
Videlefsky et al.;38 Atlanta, USA; retrospective cohort; 19972066 identified as eligible, random selection of N= 220. With a hysterectomy for benign conditions and vault smears. 164, no prior cytological abnormalities; 56, prior abnormality: 12 human papilloma virus-induced changes/mild dysplasia, 44 moderate/severe dysplasia.All had one or more vault smears. 1211 smears (range 1–21). Mean interval = 19.5 months. Mean time from hysterectomy to first smear was 13.2months (range 0–155). Average duration of follow up 89 (3–175) months.n1 = 7: four, no intervention; three treated. Long term, six had normal vault smears and one had atypical squamous cells of undetermined significance. five were from the 164 with previously normal histology. 2 from the 56.with previously abnormal histology. n2 = 2; n3 = 0. Most routine vaginal cuff cytology screening tests need not be performed in women who have had a hysterectomy for benign uterine conditions … There are currently no known scientific benefits from routine vaginal cuff smear screening, and there can be possible risks associated with performing unnecessary procedures.The use of routinely collected computerised records at one centre and the absence of active follow up of study participants indicates the potential for under ascertainment of subsequent events. QS = 7.5.
Wiener et al.;39 South Glamorgan, UK; retrospective cohort of women identified from, ‘The Cardiff Cytology Study’; 1992N= 195 women hysterectomy for preinvasive disease of cervix (CIN or adenocarcinoma in situ). All had vault smears.143, >10 years follow up; 95, 15 years follow up; 43, >20 years. A total of over 2800 woman followed up with vault smears.n1 = 5: in women who had CIN III originally, at 4, 4 and 20 months, 12 and 16 years; n2 = 3: at 20 months, 12 and 16 years; n3 = 1: at 16 years having been lost to follow up. Cytological screening of all the women who had a hysterectomy, with a history of CIN is indicated for the first 2 years after hysterectomy. Thereafter, the estimated incidence of 0.7 per 1000 women years is higher than the general population, but it is not a sufficient reason to screen more frequently.Selection of and distribution of disease in the cohort is not adequately described. Events only occurred in women with CIN III at hysterectomy, but the proportion of the cohort with CIN III is not known. Apparently, no review of eligibility of cohort. At least one subtotal hysterectomy is included, which may increase the likelihood of observing abnormal smears. QS = 4.
Williams et al.;23 USA; retrospective case–control study; 2000Nine HIV positive and 43 HIV negative women hysterectomised (controls).Controls were matched for diagnosis date, race and age. Only five HIV positive patients complied with any follow up.n1 = 3: abnormal smears at an average of 12 months (6–24 months). No recurrences in HIV negative controls. Compliance with gynaecologic follow up is very poor in this patient population … we continue to recommend treating HIV positive women in the same manner as their noninfected counterparts.Incomplete information on the follow up and characteristics of control group. Event rates in HIV positive women are unlikely to be generalisable. QS = 5.5.
N= 4 with cervical carcinoma in situ at hysterectomy and had vault smears.

Three of the papers contained information about women who had more than one histological diagnosis23,24,28 Five papers followed up women after hysterectomy for benign disease.15,23,36–38 There were no papers concentrating exclusively on women who had CIN I or CIN II, the majority of papers considered follow up after carcinoma in situ or CIN III.22,24–27,29,30–35,39

Data extraction

Table 2 summarises the abstracted data from all 19 included papers. Cytological nomenclature has changed over time, and reports from early studies have been ‘translated’ into the currently accepted terminology to allow for comparison.40,41 The authors’ conclusions about the study, if related to the role of the vaginal vault smear, are reported along with our comments on the significance of the results. QS ranged from 3.5 to 9, indicating wide variability in the methodological quality of the publications; only nine received a QS of ≥6.

The 19 studies were geographically diverse with six from North America, five from the UK, five from mainland Europe and two from New Zealand. One was a prospective cohort study, the rest were retrospective, and they were published in 16 different journals. For each paper, the number of women having a hysterectomy and the number subsequently followed up by means of a vault smear (N) is given (range 4–5682). All the available outcome data were abstracted including: (1) number of abnormal vault smears, (2) number of abnormal vaginal vault biopsies (i.e. histological confirmation of the presence of VAIN or other premalignant change but not frank malignancy) and (3) number of vaginal cancers identified subsequent to hysterectomy. Table 3 summarises the numbers of ‘events’ documented in only the papers of better methodological quality (QS ≥ 6).

Table 3.  Numbers of abnormal vault smears, vaginal biopsies and invasive vaginal cancers per study, grouped by histology at time of hysterectomy (QS ≥ 6)
OutcomeAuthorScoreVault smear, NEvents, nBenign, n/NCIN I and II, n/NCIN III, n/N
Vault smears
n1 = Abnormal vault smearsBerget et al.247.52374 0/82/206
Fawdry267.581064 64/810
Gemmell et al.2292199 9/219
Kalogirou et al.297.5793210 210/793
Pearce et al.15756827979/5682 
Piscitelli et al.3766973333/697 
Videlefsky et al.387.522075/1642/56 
Total n1 = 406/8658 (4.7%)8658406117/6543 (1.8%)2/64 (3.1%)285/2 028 (14.1%)
Vaginal biopsies
n2 = biopsy-proven dysplasiaBerget et al.247.52142 0/82/206
Boyes and Worth;258284924 24/2 849
Fawdry267.581012 12/810
Gemmell et al.2292193 3/219
Kalogirou et al.297.579341 41/793
Michalkiewicz et al.356.51722 0/122/160
Pearce et al.157568255/5682 
Piscitelli et al.37669722/697 
Videlefsky et al.387.522021/1641/56 
Total n2 = 93/11 656 (0.8%)11 656938/6543 (0.12%)1/76 (1.3%)84/5037 (1.7%)
Invasive vaginal cancers
n3 = invasive carcinomaBoyes and Worth25828490 0/2849
Fawdry267.58101 1/810
Pearce et al.157568200/5682 
Videlefsky et al.387.522000/1640/56 
Total n3 = 1/9561 (0.01%) 956110%0%1/3569 (0.03%)

Narrative review

The papers considered women with a variety of diagnoses at surgery, ranging from benign through to microinvasive cancer, although the majority considered women with CIN. The papers were divided into two main categories, women who had benign disease at hysterectomy and those who had CIN; these mainly comprised CIN III, a smaller number of papers also included cases of CIN I and II.

Follow up after hysterectomy for benign histology

Five studies were identified, of which, three were of better methodological quality.15,37,38 Only one provided data suitable for life table analysis; this was a retrospective case note review and included women who had abnormal cytology before surgery but a benign diagnosis at the time of surgery.38 All concluded that there is little benefit in follow up by vault smear screening: ‘The target condition is not common … less screening may be more desirable’,37‘there are currently no known scientific benefits from routine screening … and there can be possible risks’,38‘because of the low prevalence … and poor positive predictive value of the test routine screening … is probably not necessary’.15

Follow up after hysterectomy for CIN III

This diverse group of 15 papers only included six of better methodological quality.22,24–26,29,35 Three papers included cases of CIN I and CIN II, in addition to CIN III;24,28,29 one paper also included benign cases, but comprised a population of women who were HIV positive—a population not comparable with the general population and therefore excluded from further analysis.23

The six best quality papers were published over more than a 40-year span, and their recommendations changed over time; in the earliest paper, the conclusion was ‘of the utmost importance is thorough and long-term follow up’, while the most recent stated ‘The highest incidence of VAIN is found in the first 2 years after hysterectomy, after that incidence falls to that of the general prehysterectomy population.’29


Outcomes after hysterectomy
  • • 
    Subsequent to benign indications (n= 6543), 1.8% (117) of women had an abnormal smear, 0.12% (eight) had an abnormal biopsy and no cancers were identified.
  • • 
    Subsequent to CIN I or II: 3.1% had an abnormal vault smear, 1.3% an abnormal biopsy and no cancers were detected.
  • • 
    Subsequent to CIN III, 14.1% of women had an abnormal smear, 1.7% an abnormal biopsy and one vaginal cancer (0.03%) was detected. This was the only report of invasive vaginal cancer, subsequent to hysterectomy for CIN III.

Abnormal subsequent events [i.e. abnormal smear (χ2= 522.6, 2 df, P < 0.0001) and abnormal biopsy (χ2= 86.1, 2 df, P= 0.0001)] were positively associated with the histology at hysterectomy (benign, CIN I/II, CIN III), i.e. more adverse outcomes with increased severity of disease.

Follow up after hysterectomy for benign histology

Table 4a contains details of all studies reporting follow up after hysterectomy for benign reasons with follow-up information irrespective of methodological quality. Only one paper37 (QS = 6) included sufficient censoring and event data to enable estimation of a survival distribution.

Table 4a.  Benign histology at hysterectomy—summary of all events with time including denominator population
Duration of follow upN= follow up by vault smearsRecorded denominator population/eventsRecurrence rate (%)Cumulative rate events/denominator population 
  1. N, number of women in the original study who were followed up after hysterectomy by at least one vault smear test; n1, number of abnormal vault smears in N; n2, number of cases of biopsy-proven vaginal dysplasia.

TotalsN= 7318n1 = 117n2 = 9n1 =1.6%n2 = 0.12%n1 = 1.6%n2 = 0.12%Miller et al.;35 Pearce et al.;15 Pisctilli et al.;37 Videlefsky et al.38
Time unknownN= 6457568271561.40.179/5682 (1.4%)8/7154 (0.1%) 
<1 yearN= 861861 0.2 2/861 (0.2%) Williams et al.23 excluded as population too different from general population. There were no cancers (n3), in this population.
1–2 858 0.5 6/861 
2–4.9 8541640.50.610/8611/164 (0.6%)
5–9.9 850 0.7 16/861 
10–14.9 844 0.7 24/861 
15–19.9 838 1.3 35/861 
20–24.9 668 0 35/861 
25–29.9 668 0.1 36/861 
30–34.9 667 0.1 37/861 
35–39.9 666 0 37/861 
40–44.9 666 0 N/A37/861
45–49.9 666 0.1 N/A38/861
Follow up after CIN III

Events occurring during follow up after hysterectomy for CIN III are summarised in Table 4b, which includes all papers from the review, irrespective of QS.

Table 4b.   CIN III histology at hysterectomy, summary of all events with time
Duration of follow upN2 = follow up by vault smears (n= events)CIN IIIevents only 
  1. N, number of women in the original study who were followed up after hysterectomy by at least one vault smear test; n1, number of abnormal vault smears in N; n2, number of cases of biopsy-proven vaginal dysplasia; n3, number of cases of invasive cancer.

TotalsN= 5822n1 = 329 (5.7%)n2 = 108 (1.9%)n3 = 10 (0.2%)Berget et al.;24 Boyes and Worth;25 Fawdry;26 Gemmell et al.;22 Halberg et al;26 Hellberg and Nilson;28 Kalogirou et al.;29 Kirkup et al.;29 Liukko et al.;31 McIndoe et al.;33 Michalkiewicz et al.;35 Wiener et al.39 Williams et al.23 excluded as in Table 4a. Denominator data impossible to calculate from aggregate data thus columns omitted for clarity.
Time unknown 28780
<1 year 36831
1–2 3
2–4.9 1
5–9.9 3121
10–14.9 132
15–19.9 123


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

Summary of the evidence

Vaginal cancer is very rare. This project aggregated data from studies following up over 6600 women after hysterectomy for benign indications, CIN I and CIN II; despite up to 50 years follow up, no cases of subsequent vaginal cancer were identified. Only one case of vaginal cancer was observed, 3 years after hysterectomy, in the cohort of 3569 women having hysterectomy with reported evidence of CIN III. Unfortunately, all the studies following up women after hysterectomy with CIN III had significant methodological or reporting flaws. These results are compatible with the very low background incidence of primary vaginal cancer (approximately seven per million women per annum).42

There is little good-quality evidence concerning the role of vaginal vault smears, and there are insufficient data to enable the calculation of robust aggregated assessments of the sensitivity or specificity of the test. Most of the observed events occurred within the first 2 years of follow up; 36 of the 42 documented abnormal vault smears occurred within 2 years of hysterectomy.

Of the six best quality studies, two considered data from women who had a hysterectomy for benign indications or CIN I/II, and four studies considered women who had CIN III. The authors’ conclusions following hysterectomy for benign conditions were largely opposed to the use of the vault smear test for screening, whereas recommendations following CIN III ranged from advising annual smears to a more intensive regimen for 2 years followed by reversion to routine smears, with the frequency recommended by the national screening programme (Table 2).


UK guidelines9 recommend that vault smears should only be undertaken where there is reasonable suspicion that their pre-existing cervical pathology has not been fully treated (i.e. for women on routine recall for at least 10 years prior to hysterectomy and no CIN in the sample at hysterectomy, no vault cytology is required; for women with less than 10 years routine recall and no CIN at hysterectomy, a sample should be taken 6 months after surgery and there should be no further cytology if it is negative; for women with completely excised CIN, a sample should be taken from the vault at 6 and 18 months after surgery, there should be no further cytological follow up if both are negative). These guidelines are not based on gold standard evidence: they were derived from expert opinion. Thus, there is a need for good-quality research to establish the evidence for continued surveillance and reassess the appropriateness of current guidelines. This systematic review has aimed to address this need.


Pooled estimates of sensitivity and specificity for the vault smear test for each histological subgroup were anticipated outcomes of this review, but the variability in reference populations, incompleteness of data on censoring or confounding and large losses to follow up meant the data were insufficient for meta-analysis.

Searches of electronic databases and bibliographies of included papers provided the richest source of papers for inclusion, Medline provided almost half of the identified papers with bibliographies providing studies predating electronic search engines. Papers were published in 16 different journals. The total number of women followed up by vault smears in all the studies was 13 338 (range 4–5682; mean = 744; median = 220).

Quality scores

Quality assessment scales for identifying trials of genuinely high quality are problematic,20 however, this limitation was considered from the outset and all data retained for consideration. All the three reviewers agreed with the study ‘quality’ based on the modified CASP score and so no deviation from this was deemed necessary. The mean QS was 5.5 (range 3.5–9, SD = 2.02), with only six papers having QS greater than six and thus being deemed of ‘good’ methodological quality. There was no correlation between year of publication of the study and the QS for that study (Spearman correlation coefficient = 0.133, P= 0.588). The low QS reflect the fact that these data have been abstracted predominantly from studies that were not designed to answer the question addressed by this review.

Conclusions and implications

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

This comprehensive, systematic review has collated and assessed the available literature relating to the appropriateness of follow up subsequent to hysterectomy for indications other than cancer. The better quality studies had a combined study population of 11 656 hysterectomies, of which 6543 were for benign disease, 76 for CIN I/II and 5037 for CIN III. Nevertheless, incompleteness of follow up and recording prohibited meta-analysis of these data, and it was not, therefore, possible to provide robust estimates of the value of the vaginal vault smear test in the follow up of women who have had a hysterectomy for reasons other than malignancy.

Based on the available evidence, the current UK guidelines9 appear reasonable. However, this study confirms the need for definitive research to determine the appropriate duration and frequency of vaginal vault smears after hysterectomy for reasons other than cancer. Women’s views concerning the appropriateness of follow up and a better understanding of the anxiety related to vault smears are also required if appropriate; acceptable and evidence-based guidelines are to be developed.

As it is not practical to undertake a prospective randomised controlled trial (because vaginal cancer is rare and such a trial would have considerable ethical implications), epidemiological techniques continue to offer the most practical approach to answering some of these questions. A large, prospective audit of a cohort of women undergoing hysterectomy for reasons other than cancer could establish the frequency and characteristics of those who later develop VAIN and/or vaginal cancer.

Any future decisions concerning the use of the vaginal vault smear must consider the potential harms of screening, including possible overdiagnosis (false positives) and overtreatment, as well as the potential benefits. Women’s views concerning the appropriateness and duration of follow up need to be established. It is known that in cancer screening programmes, false positive results cause a high level of anxiety.14

This review has raised a number of important issues. Existing published research inadequately defines the clinical, financial and personal consequences of performing vault smears. The vaginal vault smear may be regarded as a ‘low priority area’ when compared with the large number of smears undertaken within cervical screening programmes. However, the public health services are required to justify all expenditure and further research in this subject is long overdue.

Finally, we conclude that there is currently no evidence to suggest that there is any demonstrable benefit in screening after hysterectomy for benign disease and, there is no evidence to support changing current guidelines for screening after CIN I/II. Screening after a diagnosis of cancer is outside the scope of this review. Screening after CIN III to 5 years is proposed by some authors, however, the data from this review indicate that 86% (36 of 42) abnormal smears occurred within 2 years, and only one case of vaginal cancer was identified in all the reported series. The value of the vaginal vault smear test as a screening tool after hysterectomy for reasons other than cancer is not supported by the existing literature.

Data access and responsibility

Helen Stokes-Lampard had full access to all the data and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Contribution to authorship

All the authors have approved the final manuscript. S.W., C.W. and S.K. conceived the study, H.S.L., S.W., C.W. and S.K. developed the methodology. H.S.L. and S.W. conducted the searches; H.S.L., S.W. and A.R. obtained the data, appraised it and undertook the literature review. R.H., H.S.L., S.W. and A.R. undertook the analysis. H.S.L. drafted the manuscript with all the authors contributing to its critical revision. S.W. supervised H.S.L. throughout this study.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

Charlotte Mann assisted with electronic searches, obtaining material and reviewing papers, Linda Grovesnor assisted with drafting the paper.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

S.W. was funded by a UK, Department of Health National Primary Care Career Scientist Award; H.S.L. and A.R. were funded by UK, Department of Health Researcher Development Awards. These funding sources had no other role in this study.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information
  • 1
    Saraiya M, Lee N, Blackman D, Smith M-J, Morrow B, McKenna MA. Self-reported papanicolaou smears and hysterectomies among women in the United States. Obstet Gynaecol 2001;98:26978.
  • 2
    Stokes-Lampard H, Wilson S, Allan T, Waddell C, Kehoe S. Vaginal vault smears—‘know more—do less’ a questionnaire survey of primary healthcare practitioners. Cytopathology 2005;16:18.
  • 3
    Eaker ED, Wierkant RA, Mas, Konitzer KA, Remington PL. Cervical cancer screening among women with and without hysterectomies. Obstet Gynaecol 1998;91:5515.
  • 4
    Hospital Episode Statistics. 2003 []. Accessed 21 April 2005.
  • 5
    Maresh MJA, Metcalfe MA, McPherson K, Overton C, Hall V, Hargreaves J, et al. The VALUE national hysterectomy study: description of the patients and their surgery. BJOG 2002;109:30212.
  • 6
    Papanicolaou G, Traut HF. The diagnosis of uterine cancer by the vaginal smear. New York, NY: Commonwealth, 1943.
  • 7
    Fetters MD, Fischer G, Reed BD. Effectiveness of vaginal Papanicolaou smear screening after total hysterectomy for benign disease. JAMA 1996;275:9407.
  • 8
    Vessy M, Villard-Mackintosh L, McPherson K, Coulter A, Yeates D. The Epidemiology of hysterectomy: findings in a large cohort study. Br J Obstet Gynaecol 1992;99:4027.
  • 9
    Luesley D, Leeson S. Colposcopy and programme management. Sheffield, UK: NHSCSP publication; 2004. p. 3738.
  • 10
    Dodge JA, Eltabbakh GH, Mount SL, Walker RP, Morgan A. Clinical features and risk of recurrence among patients with vaginal intraepithelial neoplasia. Gynaecol Oncol 2001;83:3639.
  • 11
    Herbst AL, Ulfelder H, Poskanzer DC. Adenocarcinoma of the vagina: association of maternal stilbestrol therapy and tumour appearance in young women. New Engl J Med 1971;284:878.
  • 12
    Fenell RH. Carcinoma-in-situ of the uterine cervix: a report of 118 cases. Cancer 1956;9:37484.
  • 13
    Skales KJ, Hartmann WL. Problems in carcinoma in situ of the vaginal following primary treatment of cervical neoplasia. Trans Pac Coast Obstet Gynecol Soc 1959;27:52.
  • 14
    Wilkinson C, Jones JM, McBride J. Anxiety caused by abnormal results of cervical smear test: a controlled trial. BMJ 1990;300:440.
  • 15
    Pearce KF, Hope K, Haefner HK, Sarwar SF, Nolan TE. Cytopathological findings on vaginal Papanicolaou smears after hysterectomy for benign gynaecologic disease. N Engl J Med 1996;335:155962.
  • 16
    KhanK, ReitG, GlanvilleJ, SowdenH, KleijnenJ, editors. Undertaking systematic reviews of research on effectiveness: CRD guidelines for those carrying out or commissioning reviews. York: NHS Centre for Reviews and Dissemination, University of York; 1996. CRD Report No. 4.
  • 17
    Noller KL. Screening for vaginal cancer. N Engl J Med 1996;335:15991600.
  • 18
    The Critical Appraisal Skills Programme (CASP) and Evidence Based Practice. Milton Keynes Primary Care NHS Trust. 2003. []. Accessed 4 April 2005.
  • 19
    Moher D, Cook DJ, Jadad AR, Tugwell P, Moher M, Jones A, et al. Assessing the quality of reports of randomized trials: implications for the conduct of meta-analyses. HTA 1999;3:198.
  • 20
    Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring the quality of clinical trials for meta-analysis. JAMA 1999;282:105460.
  • 21
    Gemmell J, Holmes DM, Duncan ID. How frequently need vaginal smears be taken after hysterectomy for cervical intraepithelial neoplasia? Br J Obstet Gynaecol 1990;45:4845.
  • 22
    Gemmell J, Holmes DM, Duncan ID. How frequently need vaginal smears be taken after hysterectomy for cervical intraepithelial neoplasia? Br J Obstet Gynaecol 1990;97:5861.
  • 23
    Williams FS, Roure RM, Till M, Vogler M, Del Priore G. Treatment of cervical carcinoma in situ in HIV positive women. Int J Gynaecol Obstet 2000;71:1359.
  • 24
    Berget A, Gyldensted M, Skaarup P, Szczepanski K. Clinical consequences in patients with suspect cell findings on initial vaginal cytology. Danish Med Bull 1972;19:1317.
  • 25
    Boyes DA, Worth AJ. The results of treatment of 4389 cases of pre-clinical cervical squamous carcinoma. J Obstet Gynaecol Br Commonw 1970;77:76980.
  • 26
    Fawdry RD. Carcinoma-in-situ of the cervix: is post-hysterectomy cytology worthwhile? Br J Obstet Gynaecol 1984;91:6772.
  • 27
    Halberg J, Schou P, Weberg E. Frequency and prognosis in a 5-year material of carcinoma in situ of the uterine cervix. Acta Obstet Gynecol Scand 1969;48(Suppl 3):99101.
  • 28
    Hellberg D, Nilsson S. 20-year experience of follow-up of the abnormal smear with colposcopy and histology and treatment by conization or cryosurgery. Gynecol Oncol 1990;38:1669.
  • 29
    Kalogirou D, Antoniou G, Karakitsos P, Botsis D, Papadimitriou A, Giannikos L. Vaginal intraepithelial neoplasia (VAIN) following hysterectomy in patients treated for carcinoma in situ of the cervix. Eur J Gynaecol Oncol 1997;18:18891.
  • 30
    Kirkup W, Singer A, Hill AS. Follow-up of women treated for cervical pre-cancer: an argument for a more rational approach. Lancet 1979;2:224.
  • 31
    Kurian K, Al-Nafussi A. Relation of cervical glandular intraepithelial neoplasia to microinvasive and invasive adenocarcinoma of the uterine cervix: a study of 121 cases. J Clin Path 1999;52:1127.
  • 32
    Liukko P, Pinnonen R, Gronroos M. Carcinoma in situ cervicis uteri: diagnosis, treatment and prognosis. Int J Gynaecol Obstet 1978;15:4946.
  • 33
    McIndoe WA, Green MD. Vaginal carcinoma in situ following hysterectomy. Acta Cytologica 1969;13:15862.
  • 34
    McIndoe WA, Mclean MR, Jones RW, Mullins PR. The invasive potential of carcinoma in situ of the cervix. Obstet Gynecol 1984;64:4518.
  • 35
    Michalkiewicz W, Prxybora LA, Simm S, Wolna M. Recurrence and Therapeutic problems in cervical dysplasia and in situ cancer. Cancer 1963;16:1212.
  • 36
    Miller J, Chambers D, Miller J. The need for Pap tests after hysterectomy for benign disease: results of a study of black patients. Postgrad Med 1987;82:2003.
  • 37
    Piscitelli JT, Bastia LA, Wiles A, Sime DL. Cytologic screening after hysterectomy for benign disease. Am J Obstet Gynecol 1995;173:42432.
  • 38
    Videlefsky A, Grossl N, Denniston M, Sehgal R, Lane JM, Goodenough G. Routine vaginal cuff smear testing in post hysterectomy patients with benign uterine conditions: when is it indicated? J Am Board Fam Med Prac 2000;13:2338.
  • 39
    Wiener JJ, Sweetnam PM, Jones JM. Long term follow up of women after hysterectomy with a history of pre invasive cancer of the cervix. Br J Obstet Gynaecol 1992;99:90710.
  • 40
    Anderson M, Jordan J, Morse A, Sharp F. A Text and Atlas of Integrated Colposcopy. New York, USA: Chapman & Hall Medical Publishing; 1992. p. 19.
  • 41
    Anderson M, Jordan J, Morse A, Sharp F. Integrated Colposcopy. 2nd edn. New York, USA: Chapman & Hall Medical Publishing; 1996. p. 26.
  • 42
    National Statistics Series MB1 no.33, Cancer Registrations 2002, England. []. Accessed 23 August 2005.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Conclusions and implications
  7. Acknowledgements
  8. Funding
  9. References
  10. Supporting Information

Appendix S1. The electronic search strategies. Appendix S2. Vault smear study: eligibility & validity form. Appendix S3. Assessment tool (modified CASP) for assessment of cohort studies eligible for inclusion in the vaginal vault smears systematic review.

BJO_1099_sm.doc57KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.