Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Chlamydia- engaging the men
- Roger B Karlsson, Daniel Novak (30 April 2005)
-
Chlamydia Postal Testing - another perspective
- Kirsten Kernaghan, Gordon Scott and Katy Carrick-Anderson (3 May 2005)
-
Engaging men in chlamydia screening: Authors' response
- John Macleod, Chris Salisbury, Nicola Low, Anne McCarthy, Jonathan AC Sterne, Aisha Holloway, Rita Patel on behalf of all authors (9 May 2005)
-
How plausible is possible? Quantifying uncertainty by simple sensitivity analysis
- Julius Atashili, Chapel Hill, NC 27599-7435, USA (30 July 2005)
-
Re: How plausible is possible? Quantifying uncertainty by simple sensitivity analysis
- John Macleod, Nicola Low, Matthias Egger, Owen Caul, Jonathan AC Sterne, Alan Herring, Susan Skidmore on behalf of all authors (15 August 2005)
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Roger B Karlsson, MD, PhD, ass. prof. Umeå University 901 85, Daniel Novak
Send response to journal:
|
Chlamydia trachomatis infections being easily treated but mostly asymptomatic have long posed a challenge to primary-and secondary prevention. John Macleod et al share their results from a systemic postal screening for C. trachomatis 1. Although postal screening is a feasible screening method we question the efficacy of current screening methods and emphasise the importance of improving screening by increasing the male participation. Not only postal screening but all chlamydia-screening programs have up till now consistently showed a lower male participation rate than female participation rate. The English Chlamydia Screening Studies (ClaSS) preliminary male participations rates were lower than female participation rates and the results form the first full year of screening in the National Chlamydia Screening Program (NCSP) in England showed that only 8 % of the tests were taken by males 2. Concurrently the Centres for Disease Control and Prevention (CDC) in USA recently recommend including only sexually active females in chlamydia screening. These thoughts can be debated with the fact that a Danish five-year large scale study testing only women in Copenhagen did not reduce the prevalence of Chlamydia infections 3. The reasons for low male participation in clinical chlamydia screening is that young sexually active women have more contacts with health-care professionals, than young men, due to birth control prescriptions and pap-smear test. Men rarely seeking medical health care primarily put the sexual health responsibility upon the female partner. This discrepancy in responsibility is shown in Sweden where 23% of the total number of chlamydia tests were taken by males (majority due to partner tracing) and where the male prevalence during the last five years ranged between 10-12% while the female prevalence ranged between 4-6%. Female screening is more cost effective for the society 4 but we argue the importance of including males in chlamydia screening. Chlamydia being a sexually transmitted infection and males tend to infect females easier than vice versa we find it vital that successful screening includes both sexes equally. In a recent study, where the Internet was used as a screening tool, all 22-year old men in a city in northern Sweden received coded chlamydia home sampling kits 5. Filled urine containers were sent to a laboratory for PCR analysis and participants could later access their results on an Internet page by means of their personal codes. This method proved to be a helpful tool in chlamydia screening, with an almost 40% participation rate by simplifying the testing method, increasing the accessibility, and above all providing an easy non-embarrassing health-care contact for males. Currently we are running a project supported by the Swedish National Institute of Public Health where both males and females have the opportunity to order home a sampling kit and check their test result from an informative chlamydia-Internet page (see www.klamydia.se). The project is popular and shows promising results in equal gender distribution showing that new innovative strategies may be a way to improve chlamydia screening. 1. Macleod J, Salisbury C, Low N, et al. Coverage and uptake of systematic postal screening for genital Chlamydia trachomatis and prevalence of infection in the United Kingdom general population: cross sectional study. Bmj 2005;330(7497):940. 2. LaMontagne DS, Fenton KA, Randall S, Anderson S, Carter P. Establishing the National Chlamydia Screening Programme in England: results from the first full year of screening. Sex Transm Infect 2004;80(5):335-41. 3. Westh H, Kolmos HJ. Large-scale testing of women in Copenhagen has not reduced the prevalence of Chlamydia trachomatis infections. Clin Microbiol Infect 2003;9(7):619-24. 4. Novak DP, Lindholm L, Jonsson M, Karlsson RB. A Swedish cost- effectiveness analysis of community-based Chlamydia trachomatis PCR testing of postal urine specimens obtained at home. Scand J Public Health 2004;32(5):324-32. 5. Novak DP, Edman AC, Jonsson M, Karlsson RB. The internet, a simple and convenient tool in Chlamydia trachomatis screening of young people. Euro Surveill 2003;8(9):171-6. Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Kirsten Kernaghan, Senior Sexual Health Nurse Specialist Lothian NHS Board, Gordon Scott and Katy Carrick-Anderson
Send response to journal:
|
EDITOR - McLeod et al (2005) concluded that though postal testing for Chlamydia was feasible the uptake was modest and the coverage was incomplete. The prevalence of 2.8% is also low when compared to the Chlamydia pilot studies carried out in Portsmouth and the Wirral, where the prevalence was 9% (Pimenta et al, 2003). As part of Healthy Respect, a demonstration project funded by the Scottish Executive, we have been carrying out postal testing for Chlamydia in Lothian since 2001. We have made available kits to those aged <25 within a variety of settings such as retail outlets, sports centres, sexual health clinics and drop-in centres, Further Education Colleges and community pharmacies. Our return rate has been similar to McLeod et al (2005) at 29%. However we have found an overall Chlamydia prevalence of 10.4%, rising to 12.2% in the 16-19 year age group (see table).
Why should two methods of postal testing reveal this difference in prevalence? We believe that postal testing will be most effective where kits are made available in settings used by those at higher risk of infection supported by high quality information and advice from trained professionals. Coverage may remain incomplete as McLeod at al state, but improved access outwith medical settings is a valuable goal on its own, as indicated by recent plans in England to make postal testing available from Community Pharmacies References McLeod, J, Salisbury, C., Low, N. et al (2005) Coverage and uptake of systematic postal screening for genital Chlamydia trachomatis and prevalence of infection in the United Kingdom general population: cross sectional study. BMJ 330:940 Pimenta, J.M, Catchpole, M, Rogers, P.A. et al. (2003) Opportunistic screening for genital chlamydial infection. II: Prevalence among healthcare attenders, outcome, and evaluation of positive cases. Sexually Transmitted Infection 79: 22-7. Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
John Macleod, Senior Lecturer in Primary Care Department of Primary Care and General Practice, University of Birmingham, Birmingham B15 2TT, Chris Salisbury, Nicola Low, Anne McCarthy, Jonathan AC Sterne, Aisha Holloway, Rita Patel on behalf of all authors
Send response to journal:
|
We are glad that our paper has generated debate about strategic approaches to population chlamydia screening amongst BMJ readers. Screening enthusiasts may assume that debate is now redundant and all that is needed is action. [1] The history of screening programmes suggests that such a position may be misguided. [2] Karlsson and Novak suggest that chlamydia screening programmes should try harder to engage men, both for reasons of effectiveness and to undermine notions that sexual health is only the responsibility of women. We agree and have made these points before. [3] That was why our study included an equal number of men and women and why we made equal efforts to engage them in screening. Although our efforts were not as successful as we hoped they would be, comparison of our participation rates amongst men (27% amongst those aged 16-24 contacted) with the 8% of the English National Chlamydia Screening Programme is potentially misleading. The NCSP makes little effort to actively engage men and those few screened are mainly identified through contact tracing. [4] Karlsson and Novak report their experience with an Internet based approach. This sounds promising and may be a useful strategy in situations where a high rate of Internet access can be coupled to a population register – such as amongst college students. However in the UK only around half of households have Internet access and there is currently no means of readily linking information on access to a reliable register. [5] Karlsson and Novak also suggest that young men rarely access primary health care and are therefore unlikely to be reached by opportunistic screening. On the contrary, we have shown that use of primary care by men aged 16-24, though lower than that of women, is still substantial (60% compared to 75% visiting their practice at least once in a year). [6] The opportunities for opportunistic screening of men may thus be higher than Karlsson and Novak assume – unfortunately the English NCSP does not currently take these opportunities. Augmentation of this opportunistic strategy with postal approaches to less frequent service users therefore seems a reasonable strategy, based on the available evidence. Kernaghan and colleagues describe their experience of postal screening based on testing kits distributed in non-health care settings in South East Scotland. As they say, this could be seen as an extension of current plans to pilot pharmacy based testing in England. Innovation is important in approaches to chlamydia screening however there is also need for caution around over-investment in eye-catching policies of unproven effectiveness. [7, 8] Kernaghan and colleagues quote a “return rate” of 29% of packs distributed. This is not equivalent to population coverage of 29% which cannot be estimated from the design described since it is not based on a population register. High rates of infection amongst highly selected groups (in this case people who attended participating facilities, were offered a test pack and who subsequently returned it) are typical [9, 10, 11, 12] and most probably, as these authors state, reflect selection bias. [13] These data simply demonstrate that the approach is feasible, has an uptake similar to a population-based approach and identifies a proportion of people with chlamydia infection. They provide no evidential support for the assertion that “postal testing will be most effective where kits are made available in settings used by those at higher risk of infection supported by high quality information and advice from trained professionals”. We agree with Kernaghan and colleagues that uptake of chlamydia screening may be higher amongst some people who are at lower risk of chlamydia, indeed we reported evidence that this was, to an extent, the case in our study. It would have been surprising if it had not been the case as this variation of the “inverse care law” is typically seen in relation to screening programmes. [14, 15] The phenomenon has important implications for sexual health inequalities as we discuss. However it is wrong to assume that screening only attracts the “worried well”. Qualitative research undertaken as part of the ClaSS study suggested that for some people, the decision to participate in screening is based on a rational and realistic assessment of risk. [16] Presumably this was also the case in Kernaghan and colleagues’ project and is part of the explanation of the high prevalence they found. What both these groups and our own study demonstrate is that innovative approaches to the important objective of including a greater proportion of men in chlamydia screening are feasible. What we now need is robust and rigorous evaluation of competing chlamydia screening models to determine which, if any, are cost effective in preventing morbidity in individuals and in reducing infection in the population. References 1. Coombes R. Doctors demand national screening for chlamydia. BMJ 2004;328:1397. 2. Gray MJA. Evidence-based Healthcare. New York: Churchill and Livingstone, 1997. 3. Macleod J, Davey Smith G. Chlamydia screening can have high take- up rates if right methodology is used. BMJ. 1999 ;319:188-9. 4. LaMontagne DS, Fenton KA, Randall S, Anderson S, Carter P. Establishing the National Chlamydia Screening Programme in England: results from the first full year of screening. Sex Transm Infect 2004;80:335-41 5. http://www.statistics.gov.uk/cci/nugget.asp?id=8 6. Salisbury C, Macleod J, Egger M, McCarthy A, Patel R, Holloway A et al. Consultations amongst young adults in general practice and implications for opportunistic screening for chlamydia: population based study Br J Gen Pract 2005 (in press) 7. Macleod J, Salisbury C, Low N. Screening for chlamydia. The Lancet 2005;365:1539-1540. 8. Low N, Harbord RM, Egger M, Sterne JAC, Herrmann B. Screening for chlamydia. The Lancet 2005;365:1539. 9. Gaydos CA, Howell MR, Pare B, Clark KL, Ellis DA, Hendrix RM, Gaydos JC, McKee KT Jr, Quinn TC Chlamydia trachomatis infections in female military recruits. N Engl J Med 1998;339:739-44. 10. Cecil JA, Howell MR, Tawes JJ, Gaydos JC, McKee KT Jr, Quinn TC, Gaydos CA. Features of Chlamydia trachomatis and Neisseria gonorrhoeae infection in male Army recruits. J Infect Dis 2001;184:1216-9. 11. McKay L, Clery H, Carrick-Anderson K, Hollis S, Scott G. Genital Chlamydia trachomatis infection in a subgroup of young men in the UK. Lancet 2003;361:1792. 12. Pimenta JM, Catchpole M, Rogers PA, et al. Opportunistic screening for genital chlamydial infection. II: Prevalence among health care attenders, outcome and evaluation of positive cases. Sex Transm Infect 2003;79:22–7. 13. Low N, Macleod J, Salisbury C, Egger M. Bias in chlamydia prevalence surveys. Lancet 2003;362:1157-58. 14. Macgregor JE, Fraser ME, Mann EM. The cytopipette in the diagnosis of early cervical carcinoma. Lancet. 1966;1:252-6. 15. Alderson M. An introduction to epidemiology (2nd Edition). London: Macmillan, 1983 p 98. 16. Campbell R, and the ClaSS study Group. Population screening for genital Chlamydia trachomatis: why people do and do not want to take part. MSSVD- IUSTI Europe, ASM, Leeds, May 2003. Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Julius Atashili, Graduate student, Dept of Epidemiology University of North Carolina - Chapel Hill, Chapel Hill, NC 27599-7435, USA
Send response to journal:
|
Sir, MacLeod et al[1] argue that “prevalence estimates might be affected by selection bias” and that the variability in prevalence estimates might result from selection bias given the incomplete study participation observed. However, is it possible that the near 4 fold difference in prevalence reported in this study (2%)[1] and that reported in the other study by Kernaghan et al. (10%)[2], be due to selection bias? In general, there are 2 groups of sources of error when estimating a parameter like prevalence in this case – random error and systematic error otherwise called biases[3]. If researchers generally do well in quantifying the former source of error by computing confidence intervals, the magnitude of the latter remains a matter of more or less qualitative speculation. A simple sensitivity analysis could in this case be used to provide insight on the extent to which the differences in the prevalence could be due to selection bias. In other words, given the observed prevalence and response (uptake) rates, by how much (or less) will the prevalence in the non-studied individuals have to be for the differences in the prevalences to be attributed to selection bias? Using the male data as an example, if we denote: the prevalence in the studied sample, p (=2.8%) the prevalence in the non-studied sample, q (unknown) the proportion sampled or uptake or response rate r (= 34.5%) then the true sample prevalence, t, = p*r + q*(1-r) Though q is unknown we can make a sensible guess at possible ranges of prevalence in this group. This guess could be based on previous studies (as much as possible) or solely on reason (for example, this is very unlikely to be greater than 50%). Bayesian statisticians will attribute a distribution to the values that q can take and calculate ‘posterior’ prevalences with corresponding “credibility intervals” that quantify the error due to selection bias, just as confidence intervals do for random error. Sticking to a simple sensitivity analysis, the values of q can be varied and the impact on t observed. The table shows the result of such an exercise.
This simple sensitivity analysis suggests that the prevalence of chlamydia in the non-studied sample would have to be in the order of 15% for the difference in studies to be the result of selection bias. Strictly speaking, this value has to be considered as relative to the reference study (ie had both studies had the same uptake), rather than absolute as the latter view assumes that the study with 10% prevalence did not have any bias. Such simple sensitivity analysis should be of common use in order for readers to have a quantitative sense of the possible biases that authors' tend to relegate to the discussion section in their papers. There are prospects for these methods to be rendered automated (ie incorporated in statistical softwares) so as to facilitate their implementation[4]. As a side-comment, could the authors please clarify if the prevalence in men, younger than 25, is 5.1%(as suggested in the text) or 4.5% (67/1477) as calculated from values in Table 1?. 1. Macleod J, Salisbury C, Low N, et al. Coverage and uptake of systematic postal screening for genital Chlamydia trachomatis and prevalence of infection in the United Kingdom general population: cross sectional study. BMJ 2005;330: 940-942 2. Kernaghan K, Scott G, Carrick-Anderson K. Chlamydia postal testing another perspective. BMJ rapid response accessed at http://bmj.bmjjournals.com/cgi/eletters/bmj.38413.663137.8Fv1#105672 on July 29, 2005 3.K. Rothman and S. Greenland, Precision and validity in epidemiologic studies. In: K.J. Rothman and S. Greenland, Editors, Modern epidemiology (2nd ed.), Lippincott-Raven, Philadelphia, PA (1998), pp. 115–134. 4. T. Lash and A.K. Fink, Semi-automated sensitivity analysis to assess systematic errors in observational data, Epidemiology 14 (2003), pp. 451–458
Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
John Macleod, Senior Lecturer in Primary Care Department of Primary Care and General Practice, University of Birmingham, Birmingham B15 2TT, Nicola Low, Matthias Egger, Owen Caul, Jonathan AC Sterne, Alan Herring, Susan Skidmore on behalf of all authors
Send response to journal:
|
Julius Atashili discusses the use of sensitivity analysis in the context of our recent paper on Chlamydia screening, [1] he then asks a question about our data. We found a prevalence of around 3% overall in the 35% of our population-based sample who responded to our request for a sample. Atashili computes that the prevalence of Chlamydia in the unstudied proportion of our sample would have to be quite high (around 15%) for the true overall prevalence in our population to be 10% (the overall prevalence reported by Kernaghan and colleagues in another rapid response, based on their Scottish data). [2] We agree with Atashili’s point that sensitivity analysis can often be useful however we are not sure how much it adds in this particular case. The key issue is that the prevalence estimate of Kernaghan and colleagues is unlikely to reflect true population prevalence because it was based on a very selected sub- sample of the population (29% of the people who attended a facility participating in the study and were offered a study pack). We have no idea what proportion of the general population these participants represented, as there was no population sampling-frame. The high prevalence estimate obtained is typical of those seen in highly selected samples and is likely to be a reflection of selection bias. [3] This was the point we made in our response. The point we made in our paper regarding different prevalence estimates from different studies related to comparisons between ClaSS and other general population based studies such as Natsal 2000. [4] Given that these differences are relatively small and that non-participation in both ClaSS and Natsal 2000 was relatively substantial we think it is reasonable to assert that both estimates could derive from a population with the same underlying prevalence. We don’t think that formal sensitivity analysis is necessary to support this argument. Regarding Atashili’s question about our age-specific prevalence estimates we are happy to clarify. The prevalence of Chlamydia amongst men aged 16-24 was 5.1% as reported in the text of our paper. Simply summing the figures in table 1 results in a slightly different estimate, as clustering and sampling probability are not taken into account (as discussed in our Methods section and at www.chlamydia.ac.uk). References 1. Macleod J, Salisbury C, Low N, et al. Coverage and uptake of systematic postal screening for genital Chlamydia trachomatis and prevalence of infection in the United Kingdom general population: cross sectional study. BMJ 2005;330: 940-942. 2. Kernaghan K, Scott G, Carrick-Anderson K. Chlamydia postal testing another perspective. BMJ rapid response accessed at http://bmj.bmjjournals.com/cgi/eletters/bmj.38413.663137.8Fv1#105672 on August 8, 2005. 3. Low N, Macleod J, Salisbury C, Egger M. Bias in chlamydia prevalence surveys. Lancet 2003;362:1157-58. 4. Fenton KA, Korovessis C, Johnson AM, McCadden A, McManus S, Wellings K, et al. Sexual behaviour in Britain: reported sexually transmitted infections and prevalent genital Chlamydia trachomatis infection. Lancet 2001;358: 1851-4. Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||