Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Predicting nonattendance for colposcopy clinic follow‐up after referral for an abnormal Pap smear

Predicting nonattendance for colposcopy clinic follow‐up after referral for an abnormal Pap smear Predicting nonattendance for colposcopy clinic follow-up after referral for an abnormal Pap smear Anne M. Kavanagh National Catrefor Epidtmiology and Population Health, Australian National University, Canberra Judy M. Simpson Department of Public Health and Community Medicine, University of Sydney Abstract: This is a retrospective cohort study of women who were recommended for further assessment or treatment after their first visit to a gynaecologist for an abnormal Pap smear. The sample included women who first attended a private outpatient colposcopy service in Canberra between 1January 1989 and 30 April 1990.Only women who had never previously seen a gynaecologist for an abnormal Pap smear were included in the sample. The nonattendance rate was 2.2 women per 100 women-months ( n = 493). Cox proportional hazards modelling was used to examine the relationships between sociodemographic and clinic variables and nonattendance. After adjusting for age and the degree of abnormality on presenting smear, women without private health insurance and women who had had treatment were less likely to continue attending. The current focus on idenufying barriers to screening services needs to be broadened to consider attendance at all points along the screening pathway. (Aust N Z JPublic Health 1996; 20:2 6 7 1 ) who did not use mainstream services. Examples included services for Aboriginal women, women from non-English-speaking backgrounds and women living in rural Australia.' Similarly, health promotion strategies have traditionally emphasised recruiting for cervical screening women who are identified as 'poor attenders'. Hirst et al. evaluated the effectiveness of different campaign strategies designed to recruit women in rural Victoria to cervical screening.4 However, a successful cervical cancer prevention program also depends upon the adequate assess VOL. ESEARCH on cervical cancer prevention in Australia has concentrated on identifymg hich groups of women do not attend screening services. In conjunction with the national evaluation of cervical cancer screening undertaken by the Cervical Cancer Screening and Evaluation Steering Committee in 1989-90, many states developed and evaluated screening services that focused on women R, Correspondence to Dr Anne M. Kavanagh, Visiting Fellow, Center for Cancer Prevention, Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston MA 02115, USA. Fax 1 617 432 2435. AUSTRALIAN AND N W ZEALAND JOURNAL O WBUC HEALTH 1996 E F 20 NO. 3 COLPOSCOPY FOLLOW-UP ment, treatment and follow-up of women who have cervical abnormalities. Women who have previously had abnormal cervical cytology are at greater risk of a further abnormality and of developing cervical cancer than other women.3 When a woman has an abnormal Pap smear she may be referred for further assessment with colposcopy andl, if necessary, biopsy. The appropriate follow-up of women who have abnormal cervical cytology is widely debated. However, it is generally accepted that women with cytological evidence of CIN I1 or 111 should have a colposcopy and biopsy. If the lesion is confirmed, ablative or excisional treatment is rec~mmended.~ appropriate care for The women who have minor lesions such as CIN I or human papilloma virus (HPV) is more controversial. Recent Australian guidelines recommend that women with CIN I on a Pap smear have colposcopic assessment and biopsy. If CIN I is confirmed, two approaches to the continuing care of the wdman are proposed: she may have immediate treatment or be followed with six-monthly Pap smears until the lesion progresses or regresses. The guidelines also recommend that women with HPV on their Pap smears have repeat cytology 6 and 12 months after the initial result, and, if they still have evidence of HPV 12 months later, they should be referred for a colpo~copy.~ Australian guidelines contrast with 'The overseas approaches. For example, in Canada, referral only of women with cytological evidence of CIN I1 or 111 is re~omrnended.~ Recent Australian studies have shown that the assessment and treatment of women who have abnormal Pap smears is According to protocols developed by experts, most women with CIN I1 or 111 receive appropriate assessment and treatment. However, adherence to the recommended follow-up colposcopy is poor.6 Follow-up colposcopy is necessary because a significant proportion of women will have residual or recurrent disease after treatment. A survey of gynaecologists showed that there is considerable variation in recommendations for follow-up after treatment of CIN.' The gynaecological care of women with lowgrade abnormalities is more varied than the care of women with high grade abnormalities.'** Effective ciare of women with cytological abnormalities requires both a consistent clinical approach and the attendance of women with abnormalities at colposcopy services. Do women with Pap smear abnormalities attend the appointments recommended by their clinicians?What sociodemographic and clinical variables are associated with attendance? This study examines the predictors of nonattendance for ongoing care at a private outpatient colposcopy assessment and treatment service in Canberra. Methods An audit was made of the case-note records of 502 women attending a private outpatient colposcopy service in Canberra. (There is no public outpatient colposcopy service in Canberra.) The clinic has a CO, laser on site. Four gynaecologists with many years' experience in colposcopy and the diagnosis and treatment of cervical abnormalities work from the clinic. Only women referred for the first time with an abnormal Pap smear between 1 January 1989 and 30 April 1990 were eligible for the study. Most services were provided on an outpatient basis. Only 16 per cent of women had treatment in hospital. The clinic sample represents an estimated 27 percent of all women from the Australian Capital Territory (ACT) having treatment for an abnormal Pap smear for the first time. (The denominator for this calculation was obtained from a 10 per cent sample of people enrolled in Medicare and includes women who claimed for colposcopy between 1 January 1989 and 30 April 1990, who had not claimed for colposcopy in the previous two years.) The following data were abstracted from the case-notes: age (< 25, 25-34, 35+ years), Pap smear history (normal; abnormal smears including inflammation, HPV, atypical and CIN; never had a Pap smear), self-reported Pap smear frequency (at least biennially; less than biennially), marital status (ever married or defacto, single), private health insurance status at the time of the first clinic visit (public health insurance only; private health insurance), parity (nulliparous; parous) and degree of abnormality on presenting smear (minor abnormalities including women with smears showing CIN I or less; major abnormalities including CIN I1 and CIN 111). Even though some women had had an abnormal Pap smear prior to the smear result that led to their referral to the clinic, no women in the sample had previously seen a gynaecologist about an abnormal Pap smear. The details of all tests, treatment procedures and hospital admissions were recorded, from the date of first attendance at the clinic until 16 August 1991, the date the case-note audit was commenced. Follow-up status was classified into five categories on the date the women last consulted their gynaecologist: 1. discharged during the study period; 2. still attending on 16 August 1991 when the case-note audit was complete; 3. moved or changed gynaecologists during the course of treatment and follow-up; 4. did not attend an appointment recommended by the gynaecologist; and 5. unknown. The length of time in the study was defined as the number of days between the first clinic visit and the date the woman last consulted her gynaecologist. If a clinic appointment was recommended later than 16 August 1991, women were classified as still attending. For the unknown category, the case-notes did not reveal whether women had been discharged, had moved or did not attend. Women were classified as not attending if they did not attend during the period recommended by the gynaecologist (for example, six months), provided the specified period had elapsed before 16 August 1991. Several proportions were calculated. First, the proportions of women who did not attend after one and two assessment visits were calculated. Second, we calculated .the proportions of women who did not attend for one or more follow-up visits after they had ablative or excisional treatment. For these cal267 AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 195'6 va. 20 NO. 3 F KAVANAGH AND SIMPSON Table 1: Follow-up status a n d days in study (n=502) Frequency Days in study 25th-75th Median percentile Table 2: Sociodemographic characteristics and clinical f details o sample (n= 493) Not attending Follow-up status Discharged Still attending Moved or changed gynaecalogists Nonattendance Unknown (%I Variable Presenting smear Minor abnormality Major abnormality Unknown Age Y O Y O 141 to 442 369 to 580 13 to 173 41 to 229 123 to 414 < 25 25 to 34 35 + Heolfh insurance stafus Public insurance only Private insurance Unknown culations the denominator was all women who were at risk of not attending at the subsequent visit. For example, for the proportion of women who did not attend after the first assessment visit, the denominator included all women who were not classified into a n y of the follow-up groups (other than those recorded as ‘did not attend’) on the date of their first assessment visit. That is, the denominator comprised all women who were at risk of being nonattenders for the second visit. Unfortunately, no variables indicating socioecw nomic status could be collected in this study. However, income has a strong relationship with private health insurance status. Although fewer younger people have private health insurance, the relationship between income and private health insurance is strong across ail age groups. Young single-parent families are least likely to have private health insurance, whereas families where the major contributor is between 35 and 54 have the highest levels of insurance? Using Cox proportional hazards modelling, we examined how sociodemographic and clinical variables were related to nonattendance. Only women who, after their first visit to their gynaecologists, were recommended for further gynaecological assessment or treatment were included in this analysis. (Nine women were excluded because they were discharged after one visit.) Cox proportional hazards modelling is a multivariate analytical technique appropriate for follow-up data with a dichotomous outcome and censoring. It provides an estimate of relative risk called the hazard ratio.l0 For this analysis, the outcome of interest was nonattendance. Women in the other follow-up status groups were censored on the date they received their follow-up status code. This means that women who were not classified as nonattenders, contributed to the analysis until they were discharged, had moved or changed gynaecologists, or their follow-up status became unknown; those who were still attending at the end of the case-note audit contributed to the analysis for the period that they attended the clinic. The results met the assumption of proportional hazards. Graphical techniques were used to test this assumption. Treatment was entered into the models as a timedependent variable. A woman was assigned to the treatment category when she had ablative or excisional treatment to her cervix. Until she had treatment, she was in the ‘no treatment’ category. Initial analyses of the association between nonattendance and each of the sociodemographic and clini268 Marital sfafus Ever married Single Unknown Parify Nulliparous Porous Unknown Previous Pap smear hisfory 255 Normal Abnormal 170 Never 32 Unknown 36 Self-reporfed Pap smear frequency At least biennial Less than biennial Unknown Treatment N o treatment Treatment cal variables used Cox regression. These provided estimates of unadjusted relative risks. Potential p r o b lems of collinearity due to strong associations between the explanatory variables were assessed using cross-tabulations and the chi-square test. A multivariable hazards model was developed by adding variables in order of statistical significance. The same final model was produced by progressively omitting the least significant variable from the full model with all the variables. The likelihood ratio test was used to assess the statistical significance of variables with more than two categories and the Wald statistic was used to assess the significance of dichotomous variables. These analyses used SPSS and Results Fifty per cent of women in the study, regardless of their follow-up status, had a period of at least eight months between their first and last clinic visit (minimum 1 day, maximum 883 days). Most women (54 per cent) were discharged during the study period. The median duration in the study for each follow-up status category is shown in Table 1. Women who did not attend and women who had moved had the shortest durations in the study. VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO. 3 COLPOSCOPY FOLLOW-UP Table 3: Initial and multivariable Cox proportional hazards models and nonattendance Adjusted hazards ratio Variable Presenting smear Minor abnormality Major abnormolity Age Hazards ratio (n=438) 1.o 0.69 CI 1.o 0.78 1 .o 0.62 0.51 1 .o 0.47 0.44 to 1.06 woman had treatment, the proportion of those who did not attend for subsequent visits after their second and third visits was 17.7 and 23.5 per cent respectively. Table 2 details the frequencies of variables and the number and proportion of women with each characteristic who did not attend. Percentages are based on different sample sizes because of missing values. < 25 25 to 34 35 + Heolfh insurance stafus Public insurance only Private insurance .o .o 0.37 to 0.99 to 1.18 0.31 to 0.77 Marital status Ever married Single Parity Nulliparous Porous 1 .o 1.06 1 .o 0.97 Previous Pap smear history Normal 1 0.72 Abnormal 1.71 Never .o Self-reported Pop srneor frequency At least biennial1 Less than biennial Treatment Treatment N o treatment Notes: 1.o 1.17 1 .o 2.27 1 .o 2.41 1.23 to 4.71 (a) Null model: devionce=980,436 df; final model: devionce=954,431 df; A? = 26, 5 df (b) CI =95% confidence interval Twenty per cent of women did not continue to attend the clinic as recommended by their gynaecologists. This is a nonattendance rate of 2.2 women per 100 women-months ( n = 493). Some of the women of unknown follow-up status may have been nonattenders. If these were reclassified as nonattenders, 29 per cent of the sample was lost to follow-up and the nonattendance rate was 3.1 per 100 women-months. Only two women who had a major abnormality and evidence of persistent disease did not have appropriate treatment. Of the 102 women who were classified as nonattenders, 49 had a normal colposcopy and Pap smear on their last visit. After one clinic visit, 1.4 per cent of women (485) did not attend the subsequent visit. Of the women who had treatment, 6.1 per cent (395) did not attend for the next follow-up visit. Reclassifymg as nonattenders those women who were of unknown follow-up status on their assessment or treatment visit changed the proportion not attending after the first assessment and treatment visits to 1.9 and 6.3 per cent respectively. For those who did not have treatment, the proportion of those who did not attend after one follow-up visit was 21.7 per cent (297); if those women in the unknown category were reclassified as nonattenders, the proportion was 30.9 per cent (307). Irrespective of whether a Predictors o nonutkndunm f The unadjusted and adjusted hazard ratios are shown in Table 3. Only insurance status, treatment and age were significant predictors of nonattendance in the initial analyses. M r t l status and aia self-reported Pap smear frequency were not significant predictors of nonattendance nor confounders of any other variables and were therefore dropped from the final model. Previous Pap smear history was strongly associated with age (x'= 30, 4 df,P < O.OOOOl), because 78 per cent of women who had never had a previous Pap smear were under 25, while none of them was over 35. The age distribution was similar for women who had had previously normal and women who had had previously abnormal smears, however. Previous Pap smear history was also very strongly associated with the type of presenting smear (x2 = 21.7, 2 df, P < 0.0001). Forty-six per cent of women with minor abnormalities had had previously abnormal smears whereas only 24 per cent of women with major abnormalities had had previously abnormal smears. It may be that many of the women with minor abnormalities had had persistent minor abnormalities on smear prior to referral. Previous Pap smear history was not a significant predictor of nonattendance in the multivariable model nor did it confound the effect of any other variables. Parity was strongly associated with age (x2= 142, 2 df, P < O.OOOOl), with older women being more likely to have children than younger women. Age was a confounder of parity; after adjustment for age, women with children were more likely not to attend, but parity remained nonsignificant. Moreover, coefficients of the other variables changed substantially when data for the 25 women with parity missing were excluded, so parity was omitted from the model. There were also strong associations between presenting smear and treatment (x2= 26.2, 1 df, P < 0.00001);age and health insurance status (x2= 26, 1 df,P < 0.00001); and age and treatment 24, 2 df, P = 0.00001). Ninety-two per cent of women with major abnormalities had treatment, compared with 74 per cent of women with minor abnormalities. Older women were more likely to have private health insurance. Of women younger than 35, 87 per cent had treatment, while only 68 per cent of women older than 35 had treatment. In the multivariable model only health insurance status ( P = 0.002) and treatment status ( P =0.01) were significant predictors of nonattendance. After adjustment for treatment status, the effect of the presenting smear became stronger, but was still not (xp= AUSTRALIAN AND N W ZEALAND JOURNAL O PUBLIC HEALTH 1996 v a 20 NO. 3 E F KAVANAGH AND SIMPSON statistically significant (P=0.09).Although age was initially a significant predictor of nonattendance ( P =0.02), its effect was reduced (P=O . l l ) , particularly for the oldest women, after adjustment for insurance and treatment status. The effects of insurance, treatment and presenting smear were not affected by the exclusion of age from the model. Discussion A significant proportion of women in this sample did not complete the follow-up recommended by their clinicians. After adjustment for age and the type of presenting smear, attendance was associated with treatment and health insurance status. Women with private health insurance were 50 per cent less likely to stop attending than women who did not have private health insurance. Women who had had treatment were two and a half times more likely to discontinue attending than women who had not had treatment. Most women who had treatment attended at least one follow-up visit. The new Australian guidelines recommend at least one colposcopy and Pap smear following treatment. It is suggested that a woman has a Pap smear 12 months after treatment. A colposcopy may also be performed 12 months after treatment.4 However, this study was conducted before such guidelines were in place. This study can investigate only the predictors of nonattendance for women who attend at least one clinic appointment. Also, some women may have attended other gynaecologists. However, as ,there are no public hospital clinics in Canberra, there is no financial reason to change gynaecologists. If some women attended other clinics, this study may have overestimated the proportion of nonattenders. Some women in the unknown follow-up status group are probably nonattenders. Censoring their observations at the time they received their unknown follow-up code would only bias the estimates obtained if the censoring were associated with the explanatory variable and nonattendance. If censoring were related to nonattendance, estimates of the hazard ratio could lack precision but would be unbiased.I0 Screening history data An effective cervical cancer screening program requires attention to all steps along the screening pathway, a concept used to describe: screening of an identified population at designated intervals; the recruitment of women; provision of appropriate services for taking and processing Pap smears; ensuring the adequate follow-up of women who have abnormal Pap smears and continuing evaluation of the screening program; policy support; and coordination to ensure communication between all steps. In Australia we d o not know which groups of women are at risk of not completing the screening pathway. This is partly because there is no population database from which to examine this issue. If screening and clinical data were linked, we could examine at which points along the screening pathway groups of women were more at risk of being lost to follow-up. The report of the Steering Group on Quality Assurance in Screening for the Prevention of Cancer of the Cervix recommends that cervical cytology registries carry details of women’s treatment as well as their colposcopy, cytology and histology finding^.'^ If cervical cytology registries carried such details we could, using the methods of this study, examine which groups of women are at risk of not completing the pathway at various points, using a population database. However, such data would be difficult to collect. Whether it is possible for cytology registries to gather data on the follow-up and treatment received by women who have abnormal Pap smears is currently being debated. Barriers to participation? Nonattendance may be explained by structural or cultural barriers. Structural barriers include p r o b lems of access such as cost or travel. Cultural barriers include the beliefs and attitudes of the woman and of her health care practitioner towards her abnormal Pap smear. Cost may partly explain the findings of higher risks of nonattendance among women without private health insurance. Medicare pays a rebate for private outpatient services, but women pay the difference between what the specialist charges and the Medicare rebate. Therefore, in the outpatient setting a woman’s private health insurance status does not affect the amount she pays. As many services are provided at one visit, women often face significant out-of-pocket expenses. For example, the estimated average cost to women of a visit with laser treatment, consultation and local anaesthetic fees was $99.18 in the financial year 1991-92. (This estimate is based on fees charged and benefits paid for the item combination 104, 35614, 35539 and 18200 during the financial year 1991-92 using a 10 per cent sample of the Medicare data for the whole of Australia. The estimate is relevant to services provided in a private outpatient service.) The Canberra clinic has a policy of directly billing women with health care cards and other women whom they perceive to be economically disadvantaged, but these women may comprise only a small proportion of women who find cost a barrier to access. Women who are admitted to hospital as private patients also face significant out-of-pocket expenses. To test whether cost is an important explanation for nonattendance in this group of women, one would need to examine whether they were at greater risk of nonattendance prior to their admission to hospital. If cost is an important explanation for nonattendance for women in this study, the associations between private health insurance status and attendance may be particularly strong in the ACT because Canberra does not have any public hospital clinics. In other Australian cities women have the option of attending public hospital clinics, at no financial cost. Lack of transport and child care may be barriers to participation in follow-up for some women. VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO 3 COLPOSCOPY FOLLOW-UP American women without private health insurance who had abnormal Pap smears were more likely to attend for follow-up if they were sent bus tickets to get to the clinic. In contrast, a health education interventioin did not increase the attendance of uninsured women.'* Reminder letters, which have been shown to increase participation in cervical screening, might also increase attendance if they were sent to women when the next appointment is due.15 Many gynaecologists do not have adequate systems to send reminder letters to women about follow-up after treatment: Why were women who had had treatment less likely to attend? A recent qualitative study, based on interviews with women w t Pap smear abnormaliih ties, found that women often experienced their risk of cervical cancer as reduced or negligible after treatment of their cervical abnormality.16\Further emphasis may be needed on the importance of ongoing care after treatment. Acknowledgments A RADGAC grant from the Department of Human Services and Health provided financial support for this project. We wish to thank Drs Armellin, Chiragakis, Cutter and Mutton from the Canberra Laser Clinic for their cooperation in this research project. We are grateful to Robyn Attewell, Dorothy Broom, David Legge, Heather Mitchell, Gigi Santow and the anonymous reviewers for their comments on earlier di-afts of this paper. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Predicting nonattendance for colposcopy clinic follow‐up after referral for an abnormal Pap smear

Download PDF
6 pages

Loading next page...
 
/lp/wiley/predicting-nonattendance-for-colposcopy-clinic-follow-up-after-hdz3gJFXBh

References (13)

Publisher
Wiley
Copyright
Copyright © 1996 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.1996.tb01027.x
Publisher site
See Article on Publisher Site

Abstract

Predicting nonattendance for colposcopy clinic follow-up after referral for an abnormal Pap smear Anne M. Kavanagh National Catrefor Epidtmiology and Population Health, Australian National University, Canberra Judy M. Simpson Department of Public Health and Community Medicine, University of Sydney Abstract: This is a retrospective cohort study of women who were recommended for further assessment or treatment after their first visit to a gynaecologist for an abnormal Pap smear. The sample included women who first attended a private outpatient colposcopy service in Canberra between 1January 1989 and 30 April 1990.Only women who had never previously seen a gynaecologist for an abnormal Pap smear were included in the sample. The nonattendance rate was 2.2 women per 100 women-months ( n = 493). Cox proportional hazards modelling was used to examine the relationships between sociodemographic and clinic variables and nonattendance. After adjusting for age and the degree of abnormality on presenting smear, women without private health insurance and women who had had treatment were less likely to continue attending. The current focus on idenufying barriers to screening services needs to be broadened to consider attendance at all points along the screening pathway. (Aust N Z JPublic Health 1996; 20:2 6 7 1 ) who did not use mainstream services. Examples included services for Aboriginal women, women from non-English-speaking backgrounds and women living in rural Australia.' Similarly, health promotion strategies have traditionally emphasised recruiting for cervical screening women who are identified as 'poor attenders'. Hirst et al. evaluated the effectiveness of different campaign strategies designed to recruit women in rural Victoria to cervical screening.4 However, a successful cervical cancer prevention program also depends upon the adequate assess VOL. ESEARCH on cervical cancer prevention in Australia has concentrated on identifymg hich groups of women do not attend screening services. In conjunction with the national evaluation of cervical cancer screening undertaken by the Cervical Cancer Screening and Evaluation Steering Committee in 1989-90, many states developed and evaluated screening services that focused on women R, Correspondence to Dr Anne M. Kavanagh, Visiting Fellow, Center for Cancer Prevention, Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston MA 02115, USA. Fax 1 617 432 2435. AUSTRALIAN AND N W ZEALAND JOURNAL O WBUC HEALTH 1996 E F 20 NO. 3 COLPOSCOPY FOLLOW-UP ment, treatment and follow-up of women who have cervical abnormalities. Women who have previously had abnormal cervical cytology are at greater risk of a further abnormality and of developing cervical cancer than other women.3 When a woman has an abnormal Pap smear she may be referred for further assessment with colposcopy andl, if necessary, biopsy. The appropriate follow-up of women who have abnormal cervical cytology is widely debated. However, it is generally accepted that women with cytological evidence of CIN I1 or 111 should have a colposcopy and biopsy. If the lesion is confirmed, ablative or excisional treatment is rec~mmended.~ appropriate care for The women who have minor lesions such as CIN I or human papilloma virus (HPV) is more controversial. Recent Australian guidelines recommend that women with CIN I on a Pap smear have colposcopic assessment and biopsy. If CIN I is confirmed, two approaches to the continuing care of the wdman are proposed: she may have immediate treatment or be followed with six-monthly Pap smears until the lesion progresses or regresses. The guidelines also recommend that women with HPV on their Pap smears have repeat cytology 6 and 12 months after the initial result, and, if they still have evidence of HPV 12 months later, they should be referred for a colpo~copy.~ Australian guidelines contrast with 'The overseas approaches. For example, in Canada, referral only of women with cytological evidence of CIN I1 or 111 is re~omrnended.~ Recent Australian studies have shown that the assessment and treatment of women who have abnormal Pap smears is According to protocols developed by experts, most women with CIN I1 or 111 receive appropriate assessment and treatment. However, adherence to the recommended follow-up colposcopy is poor.6 Follow-up colposcopy is necessary because a significant proportion of women will have residual or recurrent disease after treatment. A survey of gynaecologists showed that there is considerable variation in recommendations for follow-up after treatment of CIN.' The gynaecological care of women with lowgrade abnormalities is more varied than the care of women with high grade abnormalities.'** Effective ciare of women with cytological abnormalities requires both a consistent clinical approach and the attendance of women with abnormalities at colposcopy services. Do women with Pap smear abnormalities attend the appointments recommended by their clinicians?What sociodemographic and clinical variables are associated with attendance? This study examines the predictors of nonattendance for ongoing care at a private outpatient colposcopy assessment and treatment service in Canberra. Methods An audit was made of the case-note records of 502 women attending a private outpatient colposcopy service in Canberra. (There is no public outpatient colposcopy service in Canberra.) The clinic has a CO, laser on site. Four gynaecologists with many years' experience in colposcopy and the diagnosis and treatment of cervical abnormalities work from the clinic. Only women referred for the first time with an abnormal Pap smear between 1 January 1989 and 30 April 1990 were eligible for the study. Most services were provided on an outpatient basis. Only 16 per cent of women had treatment in hospital. The clinic sample represents an estimated 27 percent of all women from the Australian Capital Territory (ACT) having treatment for an abnormal Pap smear for the first time. (The denominator for this calculation was obtained from a 10 per cent sample of people enrolled in Medicare and includes women who claimed for colposcopy between 1 January 1989 and 30 April 1990, who had not claimed for colposcopy in the previous two years.) The following data were abstracted from the case-notes: age (< 25, 25-34, 35+ years), Pap smear history (normal; abnormal smears including inflammation, HPV, atypical and CIN; never had a Pap smear), self-reported Pap smear frequency (at least biennially; less than biennially), marital status (ever married or defacto, single), private health insurance status at the time of the first clinic visit (public health insurance only; private health insurance), parity (nulliparous; parous) and degree of abnormality on presenting smear (minor abnormalities including women with smears showing CIN I or less; major abnormalities including CIN I1 and CIN 111). Even though some women had had an abnormal Pap smear prior to the smear result that led to their referral to the clinic, no women in the sample had previously seen a gynaecologist about an abnormal Pap smear. The details of all tests, treatment procedures and hospital admissions were recorded, from the date of first attendance at the clinic until 16 August 1991, the date the case-note audit was commenced. Follow-up status was classified into five categories on the date the women last consulted their gynaecologist: 1. discharged during the study period; 2. still attending on 16 August 1991 when the case-note audit was complete; 3. moved or changed gynaecologists during the course of treatment and follow-up; 4. did not attend an appointment recommended by the gynaecologist; and 5. unknown. The length of time in the study was defined as the number of days between the first clinic visit and the date the woman last consulted her gynaecologist. If a clinic appointment was recommended later than 16 August 1991, women were classified as still attending. For the unknown category, the case-notes did not reveal whether women had been discharged, had moved or did not attend. Women were classified as not attending if they did not attend during the period recommended by the gynaecologist (for example, six months), provided the specified period had elapsed before 16 August 1991. Several proportions were calculated. First, the proportions of women who did not attend after one and two assessment visits were calculated. Second, we calculated .the proportions of women who did not attend for one or more follow-up visits after they had ablative or excisional treatment. For these cal267 AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 195'6 va. 20 NO. 3 F KAVANAGH AND SIMPSON Table 1: Follow-up status a n d days in study (n=502) Frequency Days in study 25th-75th Median percentile Table 2: Sociodemographic characteristics and clinical f details o sample (n= 493) Not attending Follow-up status Discharged Still attending Moved or changed gynaecalogists Nonattendance Unknown (%I Variable Presenting smear Minor abnormality Major abnormality Unknown Age Y O Y O 141 to 442 369 to 580 13 to 173 41 to 229 123 to 414 < 25 25 to 34 35 + Heolfh insurance stafus Public insurance only Private insurance Unknown culations the denominator was all women who were at risk of not attending at the subsequent visit. For example, for the proportion of women who did not attend after the first assessment visit, the denominator included all women who were not classified into a n y of the follow-up groups (other than those recorded as ‘did not attend’) on the date of their first assessment visit. That is, the denominator comprised all women who were at risk of being nonattenders for the second visit. Unfortunately, no variables indicating socioecw nomic status could be collected in this study. However, income has a strong relationship with private health insurance status. Although fewer younger people have private health insurance, the relationship between income and private health insurance is strong across ail age groups. Young single-parent families are least likely to have private health insurance, whereas families where the major contributor is between 35 and 54 have the highest levels of insurance? Using Cox proportional hazards modelling, we examined how sociodemographic and clinical variables were related to nonattendance. Only women who, after their first visit to their gynaecologists, were recommended for further gynaecological assessment or treatment were included in this analysis. (Nine women were excluded because they were discharged after one visit.) Cox proportional hazards modelling is a multivariate analytical technique appropriate for follow-up data with a dichotomous outcome and censoring. It provides an estimate of relative risk called the hazard ratio.l0 For this analysis, the outcome of interest was nonattendance. Women in the other follow-up status groups were censored on the date they received their follow-up status code. This means that women who were not classified as nonattenders, contributed to the analysis until they were discharged, had moved or changed gynaecologists, or their follow-up status became unknown; those who were still attending at the end of the case-note audit contributed to the analysis for the period that they attended the clinic. The results met the assumption of proportional hazards. Graphical techniques were used to test this assumption. Treatment was entered into the models as a timedependent variable. A woman was assigned to the treatment category when she had ablative or excisional treatment to her cervix. Until she had treatment, she was in the ‘no treatment’ category. Initial analyses of the association between nonattendance and each of the sociodemographic and clini268 Marital sfafus Ever married Single Unknown Parify Nulliparous Porous Unknown Previous Pap smear hisfory 255 Normal Abnormal 170 Never 32 Unknown 36 Self-reporfed Pap smear frequency At least biennial Less than biennial Unknown Treatment N o treatment Treatment cal variables used Cox regression. These provided estimates of unadjusted relative risks. Potential p r o b lems of collinearity due to strong associations between the explanatory variables were assessed using cross-tabulations and the chi-square test. A multivariable hazards model was developed by adding variables in order of statistical significance. The same final model was produced by progressively omitting the least significant variable from the full model with all the variables. The likelihood ratio test was used to assess the statistical significance of variables with more than two categories and the Wald statistic was used to assess the significance of dichotomous variables. These analyses used SPSS and Results Fifty per cent of women in the study, regardless of their follow-up status, had a period of at least eight months between their first and last clinic visit (minimum 1 day, maximum 883 days). Most women (54 per cent) were discharged during the study period. The median duration in the study for each follow-up status category is shown in Table 1. Women who did not attend and women who had moved had the shortest durations in the study. VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO. 3 COLPOSCOPY FOLLOW-UP Table 3: Initial and multivariable Cox proportional hazards models and nonattendance Adjusted hazards ratio Variable Presenting smear Minor abnormality Major abnormolity Age Hazards ratio (n=438) 1.o 0.69 CI 1.o 0.78 1 .o 0.62 0.51 1 .o 0.47 0.44 to 1.06 woman had treatment, the proportion of those who did not attend for subsequent visits after their second and third visits was 17.7 and 23.5 per cent respectively. Table 2 details the frequencies of variables and the number and proportion of women with each characteristic who did not attend. Percentages are based on different sample sizes because of missing values. < 25 25 to 34 35 + Heolfh insurance stafus Public insurance only Private insurance .o .o 0.37 to 0.99 to 1.18 0.31 to 0.77 Marital status Ever married Single Parity Nulliparous Porous 1 .o 1.06 1 .o 0.97 Previous Pap smear history Normal 1 0.72 Abnormal 1.71 Never .o Self-reported Pop srneor frequency At least biennial1 Less than biennial Treatment Treatment N o treatment Notes: 1.o 1.17 1 .o 2.27 1 .o 2.41 1.23 to 4.71 (a) Null model: devionce=980,436 df; final model: devionce=954,431 df; A? = 26, 5 df (b) CI =95% confidence interval Twenty per cent of women did not continue to attend the clinic as recommended by their gynaecologists. This is a nonattendance rate of 2.2 women per 100 women-months ( n = 493). Some of the women of unknown follow-up status may have been nonattenders. If these were reclassified as nonattenders, 29 per cent of the sample was lost to follow-up and the nonattendance rate was 3.1 per 100 women-months. Only two women who had a major abnormality and evidence of persistent disease did not have appropriate treatment. Of the 102 women who were classified as nonattenders, 49 had a normal colposcopy and Pap smear on their last visit. After one clinic visit, 1.4 per cent of women (485) did not attend the subsequent visit. Of the women who had treatment, 6.1 per cent (395) did not attend for the next follow-up visit. Reclassifymg as nonattenders those women who were of unknown follow-up status on their assessment or treatment visit changed the proportion not attending after the first assessment and treatment visits to 1.9 and 6.3 per cent respectively. For those who did not have treatment, the proportion of those who did not attend after one follow-up visit was 21.7 per cent (297); if those women in the unknown category were reclassified as nonattenders, the proportion was 30.9 per cent (307). Irrespective of whether a Predictors o nonutkndunm f The unadjusted and adjusted hazard ratios are shown in Table 3. Only insurance status, treatment and age were significant predictors of nonattendance in the initial analyses. M r t l status and aia self-reported Pap smear frequency were not significant predictors of nonattendance nor confounders of any other variables and were therefore dropped from the final model. Previous Pap smear history was strongly associated with age (x'= 30, 4 df,P < O.OOOOl), because 78 per cent of women who had never had a previous Pap smear were under 25, while none of them was over 35. The age distribution was similar for women who had had previously normal and women who had had previously abnormal smears, however. Previous Pap smear history was also very strongly associated with the type of presenting smear (x2 = 21.7, 2 df, P < 0.0001). Forty-six per cent of women with minor abnormalities had had previously abnormal smears whereas only 24 per cent of women with major abnormalities had had previously abnormal smears. It may be that many of the women with minor abnormalities had had persistent minor abnormalities on smear prior to referral. Previous Pap smear history was not a significant predictor of nonattendance in the multivariable model nor did it confound the effect of any other variables. Parity was strongly associated with age (x2= 142, 2 df, P < O.OOOOl), with older women being more likely to have children than younger women. Age was a confounder of parity; after adjustment for age, women with children were more likely not to attend, but parity remained nonsignificant. Moreover, coefficients of the other variables changed substantially when data for the 25 women with parity missing were excluded, so parity was omitted from the model. There were also strong associations between presenting smear and treatment (x2= 26.2, 1 df, P < 0.00001);age and health insurance status (x2= 26, 1 df,P < 0.00001); and age and treatment 24, 2 df, P = 0.00001). Ninety-two per cent of women with major abnormalities had treatment, compared with 74 per cent of women with minor abnormalities. Older women were more likely to have private health insurance. Of women younger than 35, 87 per cent had treatment, while only 68 per cent of women older than 35 had treatment. In the multivariable model only health insurance status ( P = 0.002) and treatment status ( P =0.01) were significant predictors of nonattendance. After adjustment for treatment status, the effect of the presenting smear became stronger, but was still not (xp= AUSTRALIAN AND N W ZEALAND JOURNAL O PUBLIC HEALTH 1996 v a 20 NO. 3 E F KAVANAGH AND SIMPSON statistically significant (P=0.09).Although age was initially a significant predictor of nonattendance ( P =0.02), its effect was reduced (P=O . l l ) , particularly for the oldest women, after adjustment for insurance and treatment status. The effects of insurance, treatment and presenting smear were not affected by the exclusion of age from the model. Discussion A significant proportion of women in this sample did not complete the follow-up recommended by their clinicians. After adjustment for age and the type of presenting smear, attendance was associated with treatment and health insurance status. Women with private health insurance were 50 per cent less likely to stop attending than women who did not have private health insurance. Women who had had treatment were two and a half times more likely to discontinue attending than women who had not had treatment. Most women who had treatment attended at least one follow-up visit. The new Australian guidelines recommend at least one colposcopy and Pap smear following treatment. It is suggested that a woman has a Pap smear 12 months after treatment. A colposcopy may also be performed 12 months after treatment.4 However, this study was conducted before such guidelines were in place. This study can investigate only the predictors of nonattendance for women who attend at least one clinic appointment. Also, some women may have attended other gynaecologists. However, as ,there are no public hospital clinics in Canberra, there is no financial reason to change gynaecologists. If some women attended other clinics, this study may have overestimated the proportion of nonattenders. Some women in the unknown follow-up status group are probably nonattenders. Censoring their observations at the time they received their unknown follow-up code would only bias the estimates obtained if the censoring were associated with the explanatory variable and nonattendance. If censoring were related to nonattendance, estimates of the hazard ratio could lack precision but would be unbiased.I0 Screening history data An effective cervical cancer screening program requires attention to all steps along the screening pathway, a concept used to describe: screening of an identified population at designated intervals; the recruitment of women; provision of appropriate services for taking and processing Pap smears; ensuring the adequate follow-up of women who have abnormal Pap smears and continuing evaluation of the screening program; policy support; and coordination to ensure communication between all steps. In Australia we d o not know which groups of women are at risk of not completing the screening pathway. This is partly because there is no population database from which to examine this issue. If screening and clinical data were linked, we could examine at which points along the screening pathway groups of women were more at risk of being lost to follow-up. The report of the Steering Group on Quality Assurance in Screening for the Prevention of Cancer of the Cervix recommends that cervical cytology registries carry details of women’s treatment as well as their colposcopy, cytology and histology finding^.'^ If cervical cytology registries carried such details we could, using the methods of this study, examine which groups of women are at risk of not completing the pathway at various points, using a population database. However, such data would be difficult to collect. Whether it is possible for cytology registries to gather data on the follow-up and treatment received by women who have abnormal Pap smears is currently being debated. Barriers to participation? Nonattendance may be explained by structural or cultural barriers. Structural barriers include p r o b lems of access such as cost or travel. Cultural barriers include the beliefs and attitudes of the woman and of her health care practitioner towards her abnormal Pap smear. Cost may partly explain the findings of higher risks of nonattendance among women without private health insurance. Medicare pays a rebate for private outpatient services, but women pay the difference between what the specialist charges and the Medicare rebate. Therefore, in the outpatient setting a woman’s private health insurance status does not affect the amount she pays. As many services are provided at one visit, women often face significant out-of-pocket expenses. For example, the estimated average cost to women of a visit with laser treatment, consultation and local anaesthetic fees was $99.18 in the financial year 1991-92. (This estimate is based on fees charged and benefits paid for the item combination 104, 35614, 35539 and 18200 during the financial year 1991-92 using a 10 per cent sample of the Medicare data for the whole of Australia. The estimate is relevant to services provided in a private outpatient service.) The Canberra clinic has a policy of directly billing women with health care cards and other women whom they perceive to be economically disadvantaged, but these women may comprise only a small proportion of women who find cost a barrier to access. Women who are admitted to hospital as private patients also face significant out-of-pocket expenses. To test whether cost is an important explanation for nonattendance in this group of women, one would need to examine whether they were at greater risk of nonattendance prior to their admission to hospital. If cost is an important explanation for nonattendance for women in this study, the associations between private health insurance status and attendance may be particularly strong in the ACT because Canberra does not have any public hospital clinics. In other Australian cities women have the option of attending public hospital clinics, at no financial cost. Lack of transport and child care may be barriers to participation in follow-up for some women. VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO 3 COLPOSCOPY FOLLOW-UP American women without private health insurance who had abnormal Pap smears were more likely to attend for follow-up if they were sent bus tickets to get to the clinic. In contrast, a health education interventioin did not increase the attendance of uninsured women.'* Reminder letters, which have been shown to increase participation in cervical screening, might also increase attendance if they were sent to women when the next appointment is due.15 Many gynaecologists do not have adequate systems to send reminder letters to women about follow-up after treatment: Why were women who had had treatment less likely to attend? A recent qualitative study, based on interviews with women w t Pap smear abnormaliih ties, found that women often experienced their risk of cervical cancer as reduced or negligible after treatment of their cervical abnormality.16\Further emphasis may be needed on the importance of ongoing care after treatment. Acknowledgments A RADGAC grant from the Department of Human Services and Health provided financial support for this project. We wish to thank Drs Armellin, Chiragakis, Cutter and Mutton from the Canberra Laser Clinic for their cooperation in this research project. We are grateful to Robyn Attewell, Dorothy Broom, David Legge, Heather Mitchell, Gigi Santow and the anonymous reviewers for their comments on earlier di-afts of this paper.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Jun 1, 1996

There are no references for this article.