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

Learn More →

Cervical cancer screening in rural NSW: Health Insurance Commission data compared to self‐report

Cervical cancer screening in rural NSW: Health Insurance Commission data compared to self‐report Abstract There are several sources of data for estimates of community Pap test rates, including self-report, pathology laboratory the Australian Cancer Society recommends that women between the ages of 18 and 70 years who have ever had sexual intercourse and who have an intact uterus, should have a Pap te!jt every two years1 Recent studies inAustralia, however, suggest that many 'at risk' women do not have regular Pap tests.2-h Given the pivotal role of regular cervical cancer screening in reducing mortality from cervical cancer, it is essential that programs aimed at increasing screening rates continue. It is also essential that reliable measures of Pap test rates are available for monitoring the effectiveness of such programs in the community, particularly for 'at risk' groups.4-" There are several possible sources of data for estimates of Pap test rates in the community setting, such as self-report, pathology laboratory records ;and Health Insurance Commission (HIC) 'data.7-21 Table 1 (over page) outlines previous estimates of community cervical cancer screening rates in recent years in Australia and the data source used. Rates can vary considlerably according to the sampling frame and data s o u r ~ e . ~ - ~ ' While self-report rates appear to be higher than HIC or pathology laboratory rates, this comparison is not straightforward. as the samples are not consistent between studies and both rural and urban locatilons have been sampled. Although self-report has often been used to estimate Pap test rate^,^.^^.^^,^^,^^ there are several problems with this data source.11 As with most health screening behaviours, a strong social desirability response bias can be expected.22 That is, people will be inclined to try to report positive rather than negative health behaviours. All self-report is :subject to recall bias,23 so an inaccurate response may be a function of failing to remember rather than deliberate deception. Any estimate is only as good as the sample obtained, so sa.mple selection must be VOL. o reduce the risk of cervical cancer, rigorous and consent rates to participation high if the estimate obtained is to be generalised to the population of interest.24 However, self-report can be particularly useful for providing details of the context of cervical cancer screening, such as who provided the Pap test, the reason for attending for the test, the test result and any follow-up of results. Table 1 shows that the self-reported Pap test rates in the community setting for the three studies noted were very similar (75% in three years for two studies,",20 and 78% in two years for the other).14 Pathology laboratory records have also previously been used as a data source for Pap test rates.R.10-12.17.'8.21 can be considered They the 'gold standard' for deciding whether or not a Pap test has been performed, as all Pap tests must be processed through a pathology laboratory.1'.'2.25 Where data is available online, pathology record searches can be an effective way of validating self-report However, there are also problems rates. I with pathology record searches. Pathology laboratory record searches can be inefficient and time-consuming if laboratory records are not computerised. Patient informed consent needs to be obtained before such a search can be made, if validation of self-report is the aim. It may be very difficult to define the denominator to use when estimating community rates from pathology laboratory records. The service areas for pathology laboratories often do not correspond to accepted boundaries such as postcode, local government area or even state. There may be many pathology laboratories which service one area, or many areas serviced by one large l a b ~ r a t o r y . ' ~ . ~ ~ . ~ ' In Australia, Pap test information can also be obtained from the HIC. This information is collected through the database attached to the national health system, during processing of provider service claims. There are records and Health Insurance Commission (HIC) data. Estimates of screening rates can vary considerably according to the sampling frame and data source. This study aimed to compare the self-reported estimates of cervical cancer screening with HIC estimates for women in rural NSW towns. Self-report of a Pap test in the past two years from 2,498 women in 19 rural towns of NSW was compared to HICprovided Pap test rates. Self-report levels were adjusted for non-HIC providers and HIC levels included data from the Victorian Cytology Register. Self-report estimates were significantly higher than HIC estimates in 18 of the 19 towns, with discrepancies ranging from 13% to 29%. HIC-recorded providers accounted for between 65% and 100% of Pap tests per town, according to self-report. The highest Pap test rate by self-report was 70.1%, the highest by HIC was 49.2%. The lowest Pap test rate by self-report was 45.2%, the lowest by HIC was 26.1%. There was significant variation in Pap test rates between towns for adjusted self-report estimates, but not for the crude self-report estimates. Researchers should always be aware of both the possible variations according to data source and the inherent biases for whichever data source is used. An extra caution is given to consider the public/ private provider profile when exploring possible geographical differences in Pap test rates. (AustNZJPublicHealth1998;22: 307-312) Correspondence to: The Secretary, NSW Cancer Council Cancer Education Research Program, Locked Bag 10. Wallsend, NSW 2287. 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Hancock et al. several advantages to the HIC as a data source for community Pap test rates.2R HIC data can be collected on all community members, according to postcode, so sampling problems can be reduced. Consumer address checks are routinely carried out during the periodic Medicare Card Replacement Program (about every 2.5 years). A recent estimate of agreement of 8587% was given, when comparing HIC addresses with Telecom White Pages details. (Geoff Gillett, HIC, personal communication). Data can be collated according to both patient and provider, allowing examination of both patient and provider behaviours. HIC data can be considered more reliable and less biased than It data collected in self-report surveys.28 has been estimated that HIC can account for 90% of all Pap tests conducted in NSW.9,28 HIC data is less expensive to collect than survey data. However, there are several problems associated with HIC data.27-29 Procedures performed by non-HIC providers will not be included in the HIC database,2xe.g. Pap tests performed by community nurses. This should not be a major problem for cross-sectional studies, as only a small number of smears fall into this category. However, problems can occur for longitudinal studies if a significant shift in the public/private mix occurs between data collection frames or if screening rates are compared between regions with significantly different public/private mix.29 The reliability and validity of HIC data will be affected by programming and coding errors, and by change over time as people submit outstanding claims (there is a 6-12 month time lag between attendance and submission of claims to HIC). The first error must be assumed to be minimal,2xand the second can be ameliorated by requesting data supply at 12 months following the period of interest to the study. Given the several sources of Pap test rate data and their varying advantages, there have been several previous studies which have compared rates between different sources or assessed the reliability of different data sources.",20,2s However, there is no published study which has compared rates from different sources for a spe- Table 1 : Previous estimates of community cervical cancer screening rates in Australia. Study Guest, Mitchell, Plant, 19907 Hirst, Mitchell, Medley, 19908 Shelley, Irwig, Sirnpson, Macaskill, 19909 Year Sample location, n, age Aboriginal women in rural Victoria n=l70 Rural Victoria n=234.000. 15+ vears NSW n=170,178 20-69 years Measure Self-report Aboriginal Health Service records Victorian Cytology (Gvnaecoloaical) Service HIC 10% sample Rate and interval 31% in 7 years 58 per day 20-29 years, 52% in 3years 30-39 years, 70% 40-49 years, 68% 50-59 years, 52% 60-69 vears. 29% all >15 years, 24-30% in 1 year 15-24 years, 30-36% 55+ years. 7-10% self-report, 75% in 3 years path lab, 62% in 3 years 3.3-4.8% in 12 weeks 59% in 3 years 78% in 2 years 59% in 2 years 53.6% in 1 year Young, Trevan, 199010 North Coast, NSW all >15 10,670 15-24=1,838 55+=3,600 HIC + pathology laboratory report Bowman, Redman, Dickinson, Gibberd, Sanson-Fisher, 199111 Mitchell, Hirst, Cockburn, Reading, Staples, Medley, 1991 Shelley, Irwig, Sirnpson, Macaskill, 199113 Stathers, Gibson, 1991l4 Urban NSW self-report with pathology n=l57 (self-report) laboratory validation n = l l 1 (path lab), 18-70 years Rural Victoria n=10,620, 40-69 years NSW and ACT n=155,281, 25-69 years Victorian Cytology Service HIC 10% sample self-report self-report Fitzroy Valley Pap Smear Register Not stated Sutherland Shire, NSW n=l,302, 20+years Cockburn, White, Hirst, Hill, 199215 Not stated Ballarat, Victoria n=347, 40-70 years Mak, Straton, 199316 1990 Aboriginal women in Fitzroy Valley, WA n=507, 15-69 years Western Australia n=l5,767 15+ years 12 NSW towns n=848-13,485, 18-69 years NSW and ACT n=35,054 (metro) n=18,062 (non-metro) 20-69 years Straton, Holman, Edwards, 199317 1992 Byles, Sanson-Fisher, Redman, Dickinson, Halpin, 199418 Mills, Simpson, Shelley, Turnbull, 199419 1989 1990 pathology laboratories HIC and pathology laboratory records 10% sample HIC 303 per 1000 woman years 54-76% in 3 years Metro, 40.6% in 15 months Non-metro. 39% in 15 months Bowman, Sanson-Fisher, Boyle, Pope, Redman, 199520 Byles, Sanson-Fisher, 199621 Not stated Urban NSW n=5,706, 16-70 years 1992 12 NSW towns n=873-13,725, 18-69 years self-report with HIC validation 75% in 3 years self-report 26% in 1 year HIC 18% in 1 year 55-80% in 3 years HIC and pathology laboratory records AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 22 NO. 3 Cervical cancer screening in rural NSW cifically rural population, where screening rates may be expected to be lower than an urban population.30 The present study aimed to compare the self-reported rates of cervical cancer screening with HIC screening rates, for women in rural NSW towns. Method Self-report Design years ago and more than two years ago. Where procedure numbers had changed within the two-year time frame, equivalent procedure numbers were used. Additionally, we had some anecdotal evidence from providers that a substantial proportion of Pap tests in some towns were being sent to the Victorian Cytology Service for examination and were therefore not being recorded by HIC. To overcome this, a request was made to Victorian Cytology Service to supply the same data as requested from HIC, for all postcodes. The data was collected as baseline for a large scale community action project, the CancerAction in Rural Towns (CART) p r ~ j e c t . ~ ’ were seTwenty towns in rural NSW (population 5,001to 15,000) lected for inclusion in the CART project, using postcode to define each town. Data from 1 of the CART towns were utilised for this 9 study. One town was excluded as census population figures for that postcode were not available due to aggregation of the postcode with another during the study time frame. Data collections for the CART project were conducted in the first 10 towns from June to September, 1993,and in the nlext 10 from January to March, 1994. Procedure and sampling Statistical analyses Self-report The percentage of women reporting having had a Pap test in the past two years was calculated for each town. This percentage was then multiplied by the percentage of women who reported that the provider of their last Pap test was an HIC-registered provider (a general practitioner or other specialist doctor, outside a hospital), to gain an adjusted figure for Pap test rate per town. No adjustments for hysterectomy or sexual activity were made as it was assumed that the population and sample would have similar rates for these attributes. Chi-square analyses were used to explore whether self-reported and adjusted Pap test rates were significantly different between towns.32Ninety-five per cent confidence intervals are reported for the adjusted self-report Pap test rates3* A simple random sampling approach was used. One thousand households from each town were selected for phone contact using random telephone number lists generated from Telecom White Pages. An information letter and consent form were sent to each household address. Consenting households were then phoned and a computer-assisted telephone interview conducted with the ‘nextbirthday’ person aged 1 to 70 years. All women within the first 8 200 participants sampled in each town were included in this study. Measures HIC Several manipulations of the HIC-supplied Pap test counts needed to be performed before they could be compared to adjusted selfreport screening rates. First, counts from the Victorian Cytology Service were added into the HIC tallies. Data from the Victorian Cytology Service was requested in the same format as the HIC data, to simplify this process. Second, given that the self-report survey measured only whether each woman had been screened at least once in the past two years, an estimate of the number of women having had at least one Pap test in the past two years needed to also be calculated from the HIC-supplied data. It was not appropriate to simply tally the number of women recorded by HIC as having had a Pap test in the eight quarters over the two years previous to the self-report survey. If this had been done, women who had repeat Pap tests within these two years would have been double counted. HIC counts were provided as a tally for each quarter, for three groups: women who had a previous Pap test less than one year ago; 1-2years ago; or more than two years ago. So, some notion of whether women in each quarter may have had a repeat Pap test could be gained. Four equations were used to calculate the HIC count estimate for each town. In the following equations, Q1 represents the tally of women who had a Pap test in the quarter farthest from the survey period and Q8 represents the tally of women tested in the quarter closest to the survey period, etc. For women who had a previous Pap test more than two years ago: Tally 1 = Ql+Q2+Q3+Q4+Qs+Qs+Q7+Q8 That is, all women screened in Quarters 1 to 8 whose previous Pap test was more than two years ago were included in the HIC count estimate. The questionnaire schedule included several Pap test history questions. Standard demographic questions were asked. The Pap test history questions ‘were generated with reference to past studies6,7J1J4,15.20 underwent an iterative critical review process with and clinicians and experts in this field. The Pap test items were: “Do you know what a Pap smear is?” (If the answer was no, or the explanation given was inadequate, then ‘Pap smear’ was explained, using a standard definition.) “Have you had a Pap smear in the past two years?” Then, for those reporting having a Pap test: “Where did you go to have your last Pap smear?” (Coded using the options: GP, specialist doctor, family planning clinic, women’s health nurse, hospital, other). HIC The HIC was asked for data on cervical cancer screening rates for each postcode which included the two years covered by the survey. The item numbers related to screening cervical smears were 73053 and 73054 (Cervical Smears: Routine cytological examination of smears from the cervix for detection of pre-cancerous or cancerous changes in women with no symptoms, signs or recent history suggestive of cervical neoplasia and smears repeated due to an unsatisfactory routine smear - each examination). For each quarter, a tally was requested of the number, by postcode, of women (18-70 years) who had either of these procedures, divided into those whose previous similar procedure was less than one year ago, 1-2 VOL. 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Hancock et al. For women who had a previous Pap test 1-2 years ago: Tally 2 = Q 1 +Q2+Q3+7/8Q4+5/8Q5+3/8Q6+118Q7 That is, for women whose previous Pap test was 1-2 years ago, the proportion of each quarter tally included in the overall tally was assumed to decrease from 1 to 118 for Quarters 1-7, but women screened in the quarter closest to the survey were not included. For women who had a previous Pap test less than one year ago: Tally 3 = 7/8Q1+5/8Q2+3/8Q3+1/8Q4 That is, the proportion of the tallies for Quarters 1-4 included in the overall tally was assumed to decrease from 718 to118, while women screened in Quarters 5 to 8 whose previous Pap test was less than 1 year ago were not included. To gain the final total estimate for each town: HIC Count Estimate = Tally 1 + Tally 2 + Tally 3 There were two assumptions made to generate these equations. First, that a proportion of those women whose previous Pap test was less than two years ago would have had a repeat Pap test in the eight quarters of interest. Second, that this proportion would decrease as the quarter in which they had their latest Pap test came closer to the survey period. The rate at which inclusion was declined across quarters was somewhat conservative (114 per quarter). While it cannot be assured that these equations provide an exact estimate of the number of women having had at least one Pap test in the past two years for HIC, they could be expected to be closer to the true estimate than a direct tally across quarters. In particular, the information on recency of previous Pap test adds some strength to the assumptions and calculations made and therefore the final estimates.32 Once final HIC counts had been estimated, they were converted to percentages, using 1991 census population counts for women 1870 years of age for each postcode as the denominator. ‘Unadjusted’ HIC proportions are included in Table 2 as a comparison. Statistical tests were not performed comparing HIC estimates between towns or HIC and self-report estimates, given that HIC rates are population rates and very small differences would be found to be significant. However, 95% confidence intervals are reported for the self-report Pap test rates, as a measure of agreement between the two estimates.32 Results Sample for self-report A total of 2.498 women from the 19 rural towns consented to take part in the self-report study. The sample size per town ranged from 110 to 156. Consent rate varied across towns from 64% to 90%. Census population count varied from 1,642 to 6,746 across towns. Statistical tests were not performed comparing 1991 census data and the study sample, since the large size of the census data would ensure that very small differences between the two data sets would be found to be significant. Generally, across postcodes, the study sample had a lower proportion of the younger (18-34 years) and a higher proportion of the older age groups (55 -70 years) when compared to the I99 1 census populations. However, differences were generally less than 10% between data sources. Providers of Pap tests In all towns, general practitioners were the main providers of Pap tests, providing 5 8 5 9 2 . 5 % of Pap tests. Providers tracked by the HIC accounted for 65- 100%of Pap tests per town, according to self-report The main other provider of Pap tests were women’s health nurses, accounting for 23.9% of Pap tests in one town. Self-report versus HIC rates Table 2 and Figure 1 present the data on self-report versus HIC Pap test rates. HIC rates were outside the confidence intervals of the adjusted self-report values for all towns except Town 4. For Table 2: Pap test rates in two years in 19 rural NSW towns: HIC compared to self-report. Town code 01 03 04’ 05 06 07 HIC % screened in past 2 years (unadjusted) (N) 67.2 (6746) 79.2 (3474) . . 61.6 (1932) 45.1 (2967) 39.4 (2435) 68.3 (2498) 51.8 (2517) 55.7 (2055) 69.7 (4139) . , HIC Estimated % screened at least once in past 2 years 49.2 48.6 40.3’ 31.4 26.1 42.1 35.2 37.2 43.0 44.3 40.6 41.4 47.6 30.0 35.2 37.9 Self-report % Pap test in past 2 years (n) 70.9 (110) 78.1 (128) . . 69.6 (135) 66.7 (120) 65.9 (1381 74.0 (123) 77.3 (141) 75.6 (1 19) 74.7 (146) . . 68.6 (156) 71.9 (139) 64.7 (1 19) 76.1 (134) 71.4 (1191 70.8 (130) 66.4 (137) Self-report % Pap test by HIC provider in past 2 years (95% C ) I 66.4 (57.6-75.2) 61.6 (53.3-70.1) 45.2 (36.8-53.6)’ 59.5 (50.7-68.3) 46.6 (38.3-54.9) 66.8 (58.5-75.1) 61.0 (53.0-69.1) 63.7 (55.1-72.3) 69.2 (61.7-76.7) 65.4 (57.9-72.9) 60.4 (52.3-68.5) 58.8 (50.0-67.6) 70.1 (62.4-77.9) 52.9 (43.9-61.9) 53.9 (45.3-62.5) 63.5 (55.4-71.6) 51 .O (2796) 64.0 (4264) 69.0 (2471) 63.6 (2399) 43.2 11702) 55.7 (2397) 57.6 (1642) 61.8 (53.5-70.1) AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 22 NO. 3 Cervical cancer screening in rural NSW 'Q L__ O I 10 1' I2 11 I 4 16 16 47 II 19 20 I Town Code Figure 1: Self-report and HIC Pap test data across the 20 towns in the survey. towns where there was a difference, self-report Pap test rates ranged from 13% (Town 3) to 29% (Town 18) higher than HIC rates. The highest Pap test rate by self-report was 70.1 %, the highest by HIC was 49.2%. The lowest Pap test rate by self-report was 45.2%, the lowest by HIC was 26.1%. The mean Pap test rate across towns was 60.8% by self-report and 39.5% by HIC. Chi-square tests revealed no significant variation in Pap test rates between towns for unadjusted self-report rates ( ~ ~ ' 2 3 . 8df=18, , p=O. 160), but significant variation did occur for adjusted selfreport rates ( ~ ~ ' 4 2 . 8df=18, p=O.OOl). , Discussion This study compared self-reported rates of cervical cancer screening with HIC screening rates for women in 19 rural NSW towns. Self-report estimates were significantly higher than HIC estimates in 18 of the 19 towns. with discrepancies between the two estimates ranging from 13% to 29%. Estimates varied from 45.2% to 70.1 % by self-report and from 26.1 % to 49.2% by HIC, according to town. The mean percentage of women screened across towns was 60.8% by self-report and 39.5% by HIC. The results are interesting for several reasons. First, this sample represented a significant proportion of women in rural NSW. Second, this study compares self-report and HIC screening rates for the same rural population. Third, the study reports rates on a town basis, rather than as an aggregate. This data provides two estimates from different data sources with which to explore how rural Pap test rates compare to urban rates. If we consider self-report first, a study which assessed rates of screening in an urban population in 1987 found that 75% of women reported being screened in the past three years, a figure which is only a little higher than the 71 % in two years (unadjusted) for the present study." Another study in an urban population reported a rate of 78% in two years.lJ So, for self-report estimates, it appears that the mean Pap test rate may be similar to past urban rates. However, comparison with these past studies cannot be made directly. For the 1987 study, hysterectomy rates were considered, whereas for the current study, these were ignored on the assumption that rates would be similar for HIC and self-report samples. If we assume that the hysterectomy rate is 15%,28then the new rate for this rural sample would be reduced to 60%, which is much lower than either VOL. previous urban rate. However, the Bowman et al. study used a threeyear timeframe rather than two years, which renders comparisons very dubious and the Stathers and Gibson study did not state whether hysterectomy rate was c ~ n s i d e r e d . ~ ~ it' ~ If . was not, then their urban rate remains very similar to the current rural rate. This level of agreement, however, obviously varies for individual towns. HIC estimates have been most commonly used in past studies examining Pap test rates.9.10.13-'8-2' While the problems associated with HIC estimates are acknowledged, they are still often considered a more reliable, if not more valid, measure of changes in Pap test rates than self-report and are easier to obtain than pathology laboratory data. However, past urban HIC estimates are not commonly reported. In some cases, HIC estimates have been obtained from 10% samples across states and urban-rural rates have not been rep~rted.~,'~ However, Bowman et al. found that urban HIC rates were 18% per year,20 while Mills et al. report an HIC urban rate of 41 % in 15 months.I9 Unfortunately, neither estimate is readily comparable to the current data as the time frames used are different. It should be noted that Mills et al. did not find their 'metropolitan' rate to be significantly more than their 'non-metropolitan' rate. This study's main aim was to compare self-report and HIC screening rates for the same rural population, which had not been reported before. Self-report estimates were significantly higher than HIC estimates in 18 of the 19 towns and the mean percentage of women screened across towns was higher by self-report than by HIC. Both estimates were adjusted to be as close as possible to the other, by considering screening by non-HIC providers for self-report, possible double-counting of women by HIC and services to these postcodes by theVictorian Cytology Service.2x.2y Additionally, the problem of outstanding HIC claims was reduced by requesting data supply at 12 months following the period of interest to the study. There are several possible explanations for the differences between the two Pap test rate estimates. First, when the self-report samples were compared to census demographic data for the same postcode regions, the study samples generally had a higher proportion of older women (55-70 years) and fewer younger women than the HIC denominators (1991 census data). Consequently, the age distribution could have affected agreement between data sources. However, past research has shown that younger women commonly screen at a higher rate than older women,12,1s thus this sampling bias should potentially have worked to reduce self-report levels from those expected for a more representative sample and move the estimate closer to the HIC estimate. Despite this potential effect, the HIC rate was lower than the self-report rate for 18 of the 19 towns. Second, an inaccurate denominator may have been used for HIC data: 1991 census population for postcode. There are two possible (opposing) problems with this denominator: The denominator may have been too high as the HIC-recorded population (Medicare recipients only) is not necessarily the same as the census population (all residents of Australia). The HIC estimates may have been reduced and the discrepancy between the two data sources increased. HIC-supplied denominators were not used as they have previously been found to be quite ~ n r e l i a b l e . ' ~ ~ ~ ~ ~ ~ The denominator used may have been too low, as although HIC data was collated from 1991 to 1994, census figures for 1991 only were used. It would be expected that the populations for towns would have risen from 1991 to 1994. In this case, HIC 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Hancock et al. levels would have been increased. This is a dilemma which cannot be solved here and census figures have commonly be used as the denominator for HIC Pap test data.i0~’8~2i~28 Third, self-report has acknowledged problems which work to increase rates estimated by this means: social desirability response biasz3 and recall bias,24as previously discussed. Neither of the estimates obtained in this study could be considered a ‘gold standard’ measure of Pap test rates, due to the inherent biases of each source. Probably the ‘true’ rate lies somewhere between these two estimates and it would be problematic to attempt to decide which of the estimates derived here was the ‘truer’ measure. The decision about the type of outcome measure to be used within individual studies will depend on factors such as the study question, design and resources. For example, if contextual issues are important, such as attitudes about Pap tests or reason for attendance, then self-report may be a good choice. However, if a large population needs to be measured and the resources are limited, then HIC rates can provide an effective gauge of changes in cervical cancer screening behaviour.y.’0.i8.2h While the continuing implementation of reliable registries will render these issues less important, researchers should always be aware of both the possible variations in estimates according to data source and the inherent biases for whichever source is used. This study explored Pap test rate estimates by two different data sources on a town basis, rather than as an aggregate. Chi-square tests revealed that rates between towns did not differ significantly overall for the unadjusted self-report rates, but did vary for the adjusted self-report rates. While this finding is relatively easily explained by the differing publdprivate provider profile within each town, it has some implications for studies which attempt to examine differences in HIC Pap test rates across different geographic areas. In particular, it appears that public/private mix should be included as a confounding variable in any such examination. An apparent geographical variation may in fact be due to differences in the public/private mix rather than a true difference. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Cervical cancer screening in rural NSW: Health Insurance Commission data compared to self‐report

Download PDF
6 pages

Loading next page...
 
/lp/wiley/cervical-cancer-screening-in-rural-nsw-health-insurance-commission-XGb41T12GK

References (33)

Publisher
Wiley
Copyright
Copyright © 1998 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.1998.tb01382.x
Publisher site
See Article on Publisher Site

Abstract

Abstract There are several sources of data for estimates of community Pap test rates, including self-report, pathology laboratory the Australian Cancer Society recommends that women between the ages of 18 and 70 years who have ever had sexual intercourse and who have an intact uterus, should have a Pap te!jt every two years1 Recent studies inAustralia, however, suggest that many 'at risk' women do not have regular Pap tests.2-h Given the pivotal role of regular cervical cancer screening in reducing mortality from cervical cancer, it is essential that programs aimed at increasing screening rates continue. It is also essential that reliable measures of Pap test rates are available for monitoring the effectiveness of such programs in the community, particularly for 'at risk' groups.4-" There are several possible sources of data for estimates of Pap test rates in the community setting, such as self-report, pathology laboratory records ;and Health Insurance Commission (HIC) 'data.7-21 Table 1 (over page) outlines previous estimates of community cervical cancer screening rates in recent years in Australia and the data source used. Rates can vary considlerably according to the sampling frame and data s o u r ~ e . ~ - ~ ' While self-report rates appear to be higher than HIC or pathology laboratory rates, this comparison is not straightforward. as the samples are not consistent between studies and both rural and urban locatilons have been sampled. Although self-report has often been used to estimate Pap test rate^,^.^^.^^,^^,^^ there are several problems with this data source.11 As with most health screening behaviours, a strong social desirability response bias can be expected.22 That is, people will be inclined to try to report positive rather than negative health behaviours. All self-report is :subject to recall bias,23 so an inaccurate response may be a function of failing to remember rather than deliberate deception. Any estimate is only as good as the sample obtained, so sa.mple selection must be VOL. o reduce the risk of cervical cancer, rigorous and consent rates to participation high if the estimate obtained is to be generalised to the population of interest.24 However, self-report can be particularly useful for providing details of the context of cervical cancer screening, such as who provided the Pap test, the reason for attending for the test, the test result and any follow-up of results. Table 1 shows that the self-reported Pap test rates in the community setting for the three studies noted were very similar (75% in three years for two studies,",20 and 78% in two years for the other).14 Pathology laboratory records have also previously been used as a data source for Pap test rates.R.10-12.17.'8.21 can be considered They the 'gold standard' for deciding whether or not a Pap test has been performed, as all Pap tests must be processed through a pathology laboratory.1'.'2.25 Where data is available online, pathology record searches can be an effective way of validating self-report However, there are also problems rates. I with pathology record searches. Pathology laboratory record searches can be inefficient and time-consuming if laboratory records are not computerised. Patient informed consent needs to be obtained before such a search can be made, if validation of self-report is the aim. It may be very difficult to define the denominator to use when estimating community rates from pathology laboratory records. The service areas for pathology laboratories often do not correspond to accepted boundaries such as postcode, local government area or even state. There may be many pathology laboratories which service one area, or many areas serviced by one large l a b ~ r a t o r y . ' ~ . ~ ~ . ~ ' In Australia, Pap test information can also be obtained from the HIC. This information is collected through the database attached to the national health system, during processing of provider service claims. There are records and Health Insurance Commission (HIC) data. Estimates of screening rates can vary considerably according to the sampling frame and data source. This study aimed to compare the self-reported estimates of cervical cancer screening with HIC estimates for women in rural NSW towns. Self-report of a Pap test in the past two years from 2,498 women in 19 rural towns of NSW was compared to HICprovided Pap test rates. Self-report levels were adjusted for non-HIC providers and HIC levels included data from the Victorian Cytology Register. Self-report estimates were significantly higher than HIC estimates in 18 of the 19 towns, with discrepancies ranging from 13% to 29%. HIC-recorded providers accounted for between 65% and 100% of Pap tests per town, according to self-report. The highest Pap test rate by self-report was 70.1%, the highest by HIC was 49.2%. The lowest Pap test rate by self-report was 45.2%, the lowest by HIC was 26.1%. There was significant variation in Pap test rates between towns for adjusted self-report estimates, but not for the crude self-report estimates. Researchers should always be aware of both the possible variations according to data source and the inherent biases for whichever data source is used. An extra caution is given to consider the public/ private provider profile when exploring possible geographical differences in Pap test rates. (AustNZJPublicHealth1998;22: 307-312) Correspondence to: The Secretary, NSW Cancer Council Cancer Education Research Program, Locked Bag 10. Wallsend, NSW 2287. 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Hancock et al. several advantages to the HIC as a data source for community Pap test rates.2R HIC data can be collected on all community members, according to postcode, so sampling problems can be reduced. Consumer address checks are routinely carried out during the periodic Medicare Card Replacement Program (about every 2.5 years). A recent estimate of agreement of 8587% was given, when comparing HIC addresses with Telecom White Pages details. (Geoff Gillett, HIC, personal communication). Data can be collated according to both patient and provider, allowing examination of both patient and provider behaviours. HIC data can be considered more reliable and less biased than It data collected in self-report surveys.28 has been estimated that HIC can account for 90% of all Pap tests conducted in NSW.9,28 HIC data is less expensive to collect than survey data. However, there are several problems associated with HIC data.27-29 Procedures performed by non-HIC providers will not be included in the HIC database,2xe.g. Pap tests performed by community nurses. This should not be a major problem for cross-sectional studies, as only a small number of smears fall into this category. However, problems can occur for longitudinal studies if a significant shift in the public/private mix occurs between data collection frames or if screening rates are compared between regions with significantly different public/private mix.29 The reliability and validity of HIC data will be affected by programming and coding errors, and by change over time as people submit outstanding claims (there is a 6-12 month time lag between attendance and submission of claims to HIC). The first error must be assumed to be minimal,2xand the second can be ameliorated by requesting data supply at 12 months following the period of interest to the study. Given the several sources of Pap test rate data and their varying advantages, there have been several previous studies which have compared rates between different sources or assessed the reliability of different data sources.",20,2s However, there is no published study which has compared rates from different sources for a spe- Table 1 : Previous estimates of community cervical cancer screening rates in Australia. Study Guest, Mitchell, Plant, 19907 Hirst, Mitchell, Medley, 19908 Shelley, Irwig, Sirnpson, Macaskill, 19909 Year Sample location, n, age Aboriginal women in rural Victoria n=l70 Rural Victoria n=234.000. 15+ vears NSW n=170,178 20-69 years Measure Self-report Aboriginal Health Service records Victorian Cytology (Gvnaecoloaical) Service HIC 10% sample Rate and interval 31% in 7 years 58 per day 20-29 years, 52% in 3years 30-39 years, 70% 40-49 years, 68% 50-59 years, 52% 60-69 vears. 29% all >15 years, 24-30% in 1 year 15-24 years, 30-36% 55+ years. 7-10% self-report, 75% in 3 years path lab, 62% in 3 years 3.3-4.8% in 12 weeks 59% in 3 years 78% in 2 years 59% in 2 years 53.6% in 1 year Young, Trevan, 199010 North Coast, NSW all >15 10,670 15-24=1,838 55+=3,600 HIC + pathology laboratory report Bowman, Redman, Dickinson, Gibberd, Sanson-Fisher, 199111 Mitchell, Hirst, Cockburn, Reading, Staples, Medley, 1991 Shelley, Irwig, Sirnpson, Macaskill, 199113 Stathers, Gibson, 1991l4 Urban NSW self-report with pathology n=l57 (self-report) laboratory validation n = l l 1 (path lab), 18-70 years Rural Victoria n=10,620, 40-69 years NSW and ACT n=155,281, 25-69 years Victorian Cytology Service HIC 10% sample self-report self-report Fitzroy Valley Pap Smear Register Not stated Sutherland Shire, NSW n=l,302, 20+years Cockburn, White, Hirst, Hill, 199215 Not stated Ballarat, Victoria n=347, 40-70 years Mak, Straton, 199316 1990 Aboriginal women in Fitzroy Valley, WA n=507, 15-69 years Western Australia n=l5,767 15+ years 12 NSW towns n=848-13,485, 18-69 years NSW and ACT n=35,054 (metro) n=18,062 (non-metro) 20-69 years Straton, Holman, Edwards, 199317 1992 Byles, Sanson-Fisher, Redman, Dickinson, Halpin, 199418 Mills, Simpson, Shelley, Turnbull, 199419 1989 1990 pathology laboratories HIC and pathology laboratory records 10% sample HIC 303 per 1000 woman years 54-76% in 3 years Metro, 40.6% in 15 months Non-metro. 39% in 15 months Bowman, Sanson-Fisher, Boyle, Pope, Redman, 199520 Byles, Sanson-Fisher, 199621 Not stated Urban NSW n=5,706, 16-70 years 1992 12 NSW towns n=873-13,725, 18-69 years self-report with HIC validation 75% in 3 years self-report 26% in 1 year HIC 18% in 1 year 55-80% in 3 years HIC and pathology laboratory records AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 22 NO. 3 Cervical cancer screening in rural NSW cifically rural population, where screening rates may be expected to be lower than an urban population.30 The present study aimed to compare the self-reported rates of cervical cancer screening with HIC screening rates, for women in rural NSW towns. Method Self-report Design years ago and more than two years ago. Where procedure numbers had changed within the two-year time frame, equivalent procedure numbers were used. Additionally, we had some anecdotal evidence from providers that a substantial proportion of Pap tests in some towns were being sent to the Victorian Cytology Service for examination and were therefore not being recorded by HIC. To overcome this, a request was made to Victorian Cytology Service to supply the same data as requested from HIC, for all postcodes. The data was collected as baseline for a large scale community action project, the CancerAction in Rural Towns (CART) p r ~ j e c t . ~ ’ were seTwenty towns in rural NSW (population 5,001to 15,000) lected for inclusion in the CART project, using postcode to define each town. Data from 1 of the CART towns were utilised for this 9 study. One town was excluded as census population figures for that postcode were not available due to aggregation of the postcode with another during the study time frame. Data collections for the CART project were conducted in the first 10 towns from June to September, 1993,and in the nlext 10 from January to March, 1994. Procedure and sampling Statistical analyses Self-report The percentage of women reporting having had a Pap test in the past two years was calculated for each town. This percentage was then multiplied by the percentage of women who reported that the provider of their last Pap test was an HIC-registered provider (a general practitioner or other specialist doctor, outside a hospital), to gain an adjusted figure for Pap test rate per town. No adjustments for hysterectomy or sexual activity were made as it was assumed that the population and sample would have similar rates for these attributes. Chi-square analyses were used to explore whether self-reported and adjusted Pap test rates were significantly different between towns.32Ninety-five per cent confidence intervals are reported for the adjusted self-report Pap test rates3* A simple random sampling approach was used. One thousand households from each town were selected for phone contact using random telephone number lists generated from Telecom White Pages. An information letter and consent form were sent to each household address. Consenting households were then phoned and a computer-assisted telephone interview conducted with the ‘nextbirthday’ person aged 1 to 70 years. All women within the first 8 200 participants sampled in each town were included in this study. Measures HIC Several manipulations of the HIC-supplied Pap test counts needed to be performed before they could be compared to adjusted selfreport screening rates. First, counts from the Victorian Cytology Service were added into the HIC tallies. Data from the Victorian Cytology Service was requested in the same format as the HIC data, to simplify this process. Second, given that the self-report survey measured only whether each woman had been screened at least once in the past two years, an estimate of the number of women having had at least one Pap test in the past two years needed to also be calculated from the HIC-supplied data. It was not appropriate to simply tally the number of women recorded by HIC as having had a Pap test in the eight quarters over the two years previous to the self-report survey. If this had been done, women who had repeat Pap tests within these two years would have been double counted. HIC counts were provided as a tally for each quarter, for three groups: women who had a previous Pap test less than one year ago; 1-2years ago; or more than two years ago. So, some notion of whether women in each quarter may have had a repeat Pap test could be gained. Four equations were used to calculate the HIC count estimate for each town. In the following equations, Q1 represents the tally of women who had a Pap test in the quarter farthest from the survey period and Q8 represents the tally of women tested in the quarter closest to the survey period, etc. For women who had a previous Pap test more than two years ago: Tally 1 = Ql+Q2+Q3+Q4+Qs+Qs+Q7+Q8 That is, all women screened in Quarters 1 to 8 whose previous Pap test was more than two years ago were included in the HIC count estimate. The questionnaire schedule included several Pap test history questions. Standard demographic questions were asked. The Pap test history questions ‘were generated with reference to past studies6,7J1J4,15.20 underwent an iterative critical review process with and clinicians and experts in this field. The Pap test items were: “Do you know what a Pap smear is?” (If the answer was no, or the explanation given was inadequate, then ‘Pap smear’ was explained, using a standard definition.) “Have you had a Pap smear in the past two years?” Then, for those reporting having a Pap test: “Where did you go to have your last Pap smear?” (Coded using the options: GP, specialist doctor, family planning clinic, women’s health nurse, hospital, other). HIC The HIC was asked for data on cervical cancer screening rates for each postcode which included the two years covered by the survey. The item numbers related to screening cervical smears were 73053 and 73054 (Cervical Smears: Routine cytological examination of smears from the cervix for detection of pre-cancerous or cancerous changes in women with no symptoms, signs or recent history suggestive of cervical neoplasia and smears repeated due to an unsatisfactory routine smear - each examination). For each quarter, a tally was requested of the number, by postcode, of women (18-70 years) who had either of these procedures, divided into those whose previous similar procedure was less than one year ago, 1-2 VOL. 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Hancock et al. For women who had a previous Pap test 1-2 years ago: Tally 2 = Q 1 +Q2+Q3+7/8Q4+5/8Q5+3/8Q6+118Q7 That is, for women whose previous Pap test was 1-2 years ago, the proportion of each quarter tally included in the overall tally was assumed to decrease from 1 to 118 for Quarters 1-7, but women screened in the quarter closest to the survey were not included. For women who had a previous Pap test less than one year ago: Tally 3 = 7/8Q1+5/8Q2+3/8Q3+1/8Q4 That is, the proportion of the tallies for Quarters 1-4 included in the overall tally was assumed to decrease from 718 to118, while women screened in Quarters 5 to 8 whose previous Pap test was less than 1 year ago were not included. To gain the final total estimate for each town: HIC Count Estimate = Tally 1 + Tally 2 + Tally 3 There were two assumptions made to generate these equations. First, that a proportion of those women whose previous Pap test was less than two years ago would have had a repeat Pap test in the eight quarters of interest. Second, that this proportion would decrease as the quarter in which they had their latest Pap test came closer to the survey period. The rate at which inclusion was declined across quarters was somewhat conservative (114 per quarter). While it cannot be assured that these equations provide an exact estimate of the number of women having had at least one Pap test in the past two years for HIC, they could be expected to be closer to the true estimate than a direct tally across quarters. In particular, the information on recency of previous Pap test adds some strength to the assumptions and calculations made and therefore the final estimates.32 Once final HIC counts had been estimated, they were converted to percentages, using 1991 census population counts for women 1870 years of age for each postcode as the denominator. ‘Unadjusted’ HIC proportions are included in Table 2 as a comparison. Statistical tests were not performed comparing HIC estimates between towns or HIC and self-report estimates, given that HIC rates are population rates and very small differences would be found to be significant. However, 95% confidence intervals are reported for the self-report Pap test rates, as a measure of agreement between the two estimates.32 Results Sample for self-report A total of 2.498 women from the 19 rural towns consented to take part in the self-report study. The sample size per town ranged from 110 to 156. Consent rate varied across towns from 64% to 90%. Census population count varied from 1,642 to 6,746 across towns. Statistical tests were not performed comparing 1991 census data and the study sample, since the large size of the census data would ensure that very small differences between the two data sets would be found to be significant. Generally, across postcodes, the study sample had a lower proportion of the younger (18-34 years) and a higher proportion of the older age groups (55 -70 years) when compared to the I99 1 census populations. However, differences were generally less than 10% between data sources. Providers of Pap tests In all towns, general practitioners were the main providers of Pap tests, providing 5 8 5 9 2 . 5 % of Pap tests. Providers tracked by the HIC accounted for 65- 100%of Pap tests per town, according to self-report The main other provider of Pap tests were women’s health nurses, accounting for 23.9% of Pap tests in one town. Self-report versus HIC rates Table 2 and Figure 1 present the data on self-report versus HIC Pap test rates. HIC rates were outside the confidence intervals of the adjusted self-report values for all towns except Town 4. For Table 2: Pap test rates in two years in 19 rural NSW towns: HIC compared to self-report. Town code 01 03 04’ 05 06 07 HIC % screened in past 2 years (unadjusted) (N) 67.2 (6746) 79.2 (3474) . . 61.6 (1932) 45.1 (2967) 39.4 (2435) 68.3 (2498) 51.8 (2517) 55.7 (2055) 69.7 (4139) . , HIC Estimated % screened at least once in past 2 years 49.2 48.6 40.3’ 31.4 26.1 42.1 35.2 37.2 43.0 44.3 40.6 41.4 47.6 30.0 35.2 37.9 Self-report % Pap test in past 2 years (n) 70.9 (110) 78.1 (128) . . 69.6 (135) 66.7 (120) 65.9 (1381 74.0 (123) 77.3 (141) 75.6 (1 19) 74.7 (146) . . 68.6 (156) 71.9 (139) 64.7 (1 19) 76.1 (134) 71.4 (1191 70.8 (130) 66.4 (137) Self-report % Pap test by HIC provider in past 2 years (95% C ) I 66.4 (57.6-75.2) 61.6 (53.3-70.1) 45.2 (36.8-53.6)’ 59.5 (50.7-68.3) 46.6 (38.3-54.9) 66.8 (58.5-75.1) 61.0 (53.0-69.1) 63.7 (55.1-72.3) 69.2 (61.7-76.7) 65.4 (57.9-72.9) 60.4 (52.3-68.5) 58.8 (50.0-67.6) 70.1 (62.4-77.9) 52.9 (43.9-61.9) 53.9 (45.3-62.5) 63.5 (55.4-71.6) 51 .O (2796) 64.0 (4264) 69.0 (2471) 63.6 (2399) 43.2 11702) 55.7 (2397) 57.6 (1642) 61.8 (53.5-70.1) AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 22 NO. 3 Cervical cancer screening in rural NSW 'Q L__ O I 10 1' I2 11 I 4 16 16 47 II 19 20 I Town Code Figure 1: Self-report and HIC Pap test data across the 20 towns in the survey. towns where there was a difference, self-report Pap test rates ranged from 13% (Town 3) to 29% (Town 18) higher than HIC rates. The highest Pap test rate by self-report was 70.1 %, the highest by HIC was 49.2%. The lowest Pap test rate by self-report was 45.2%, the lowest by HIC was 26.1%. The mean Pap test rate across towns was 60.8% by self-report and 39.5% by HIC. Chi-square tests revealed no significant variation in Pap test rates between towns for unadjusted self-report rates ( ~ ~ ' 2 3 . 8df=18, , p=O. 160), but significant variation did occur for adjusted selfreport rates ( ~ ~ ' 4 2 . 8df=18, p=O.OOl). , Discussion This study compared self-reported rates of cervical cancer screening with HIC screening rates for women in 19 rural NSW towns. Self-report estimates were significantly higher than HIC estimates in 18 of the 19 towns. with discrepancies between the two estimates ranging from 13% to 29%. Estimates varied from 45.2% to 70.1 % by self-report and from 26.1 % to 49.2% by HIC, according to town. The mean percentage of women screened across towns was 60.8% by self-report and 39.5% by HIC. The results are interesting for several reasons. First, this sample represented a significant proportion of women in rural NSW. Second, this study compares self-report and HIC screening rates for the same rural population. Third, the study reports rates on a town basis, rather than as an aggregate. This data provides two estimates from different data sources with which to explore how rural Pap test rates compare to urban rates. If we consider self-report first, a study which assessed rates of screening in an urban population in 1987 found that 75% of women reported being screened in the past three years, a figure which is only a little higher than the 71 % in two years (unadjusted) for the present study." Another study in an urban population reported a rate of 78% in two years.lJ So, for self-report estimates, it appears that the mean Pap test rate may be similar to past urban rates. However, comparison with these past studies cannot be made directly. For the 1987 study, hysterectomy rates were considered, whereas for the current study, these were ignored on the assumption that rates would be similar for HIC and self-report samples. If we assume that the hysterectomy rate is 15%,28then the new rate for this rural sample would be reduced to 60%, which is much lower than either VOL. previous urban rate. However, the Bowman et al. study used a threeyear timeframe rather than two years, which renders comparisons very dubious and the Stathers and Gibson study did not state whether hysterectomy rate was c ~ n s i d e r e d . ~ ~ it' ~ If . was not, then their urban rate remains very similar to the current rural rate. This level of agreement, however, obviously varies for individual towns. HIC estimates have been most commonly used in past studies examining Pap test rates.9.10.13-'8-2' While the problems associated with HIC estimates are acknowledged, they are still often considered a more reliable, if not more valid, measure of changes in Pap test rates than self-report and are easier to obtain than pathology laboratory data. However, past urban HIC estimates are not commonly reported. In some cases, HIC estimates have been obtained from 10% samples across states and urban-rural rates have not been rep~rted.~,'~ However, Bowman et al. found that urban HIC rates were 18% per year,20 while Mills et al. report an HIC urban rate of 41 % in 15 months.I9 Unfortunately, neither estimate is readily comparable to the current data as the time frames used are different. It should be noted that Mills et al. did not find their 'metropolitan' rate to be significantly more than their 'non-metropolitan' rate. This study's main aim was to compare self-report and HIC screening rates for the same rural population, which had not been reported before. Self-report estimates were significantly higher than HIC estimates in 18 of the 19 towns and the mean percentage of women screened across towns was higher by self-report than by HIC. Both estimates were adjusted to be as close as possible to the other, by considering screening by non-HIC providers for self-report, possible double-counting of women by HIC and services to these postcodes by theVictorian Cytology Service.2x.2y Additionally, the problem of outstanding HIC claims was reduced by requesting data supply at 12 months following the period of interest to the study. There are several possible explanations for the differences between the two Pap test rate estimates. First, when the self-report samples were compared to census demographic data for the same postcode regions, the study samples generally had a higher proportion of older women (55-70 years) and fewer younger women than the HIC denominators (1991 census data). Consequently, the age distribution could have affected agreement between data sources. However, past research has shown that younger women commonly screen at a higher rate than older women,12,1s thus this sampling bias should potentially have worked to reduce self-report levels from those expected for a more representative sample and move the estimate closer to the HIC estimate. Despite this potential effect, the HIC rate was lower than the self-report rate for 18 of the 19 towns. Second, an inaccurate denominator may have been used for HIC data: 1991 census population for postcode. There are two possible (opposing) problems with this denominator: The denominator may have been too high as the HIC-recorded population (Medicare recipients only) is not necessarily the same as the census population (all residents of Australia). The HIC estimates may have been reduced and the discrepancy between the two data sources increased. HIC-supplied denominators were not used as they have previously been found to be quite ~ n r e l i a b l e . ' ~ ~ ~ ~ ~ ~ The denominator used may have been too low, as although HIC data was collated from 1991 to 1994, census figures for 1991 only were used. It would be expected that the populations for towns would have risen from 1991 to 1994. In this case, HIC 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Hancock et al. levels would have been increased. This is a dilemma which cannot be solved here and census figures have commonly be used as the denominator for HIC Pap test data.i0~’8~2i~28 Third, self-report has acknowledged problems which work to increase rates estimated by this means: social desirability response biasz3 and recall bias,24as previously discussed. Neither of the estimates obtained in this study could be considered a ‘gold standard’ measure of Pap test rates, due to the inherent biases of each source. Probably the ‘true’ rate lies somewhere between these two estimates and it would be problematic to attempt to decide which of the estimates derived here was the ‘truer’ measure. The decision about the type of outcome measure to be used within individual studies will depend on factors such as the study question, design and resources. For example, if contextual issues are important, such as attitudes about Pap tests or reason for attendance, then self-report may be a good choice. However, if a large population needs to be measured and the resources are limited, then HIC rates can provide an effective gauge of changes in cervical cancer screening behaviour.y.’0.i8.2h While the continuing implementation of reliable registries will render these issues less important, researchers should always be aware of both the possible variations in estimates according to data source and the inherent biases for whichever source is used. This study explored Pap test rate estimates by two different data sources on a town basis, rather than as an aggregate. Chi-square tests revealed that rates between towns did not differ significantly overall for the unadjusted self-report rates, but did vary for the adjusted self-report rates. While this finding is relatively easily explained by the differing publdprivate provider profile within each town, it has some implications for studies which attempt to examine differences in HIC Pap test rates across different geographic areas. In particular, it appears that public/private mix should be included as a confounding variable in any such examination. An apparent geographical variation may in fact be due to differences in the public/private mix rather than a true difference.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Jun 1, 1998

There are no references for this article.