Chlamydia and pelvic inflammatory disease: Population Excess Fraction based on prospective, retrospective and routine data

Malcolm J Price; AE Ades; Kate Soldan; Nicky J Welton; John Macleod; Ian Simms; Daniela DeAngelis; Katherine ME Turner; Paddy J Horner

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Price MJ, Ades AE, Soldan K, et al. The natural history of Chlamydia trachomatis infection in women: a multi-parameter evidence synthesis. Southampton (UK): NIHR Journals Library; 2016 Mar. (Health Technology Assessment, No. 20.22.)

The natural history of Chlamydia trachomatis infection in women: a multi-parameter evidence synthesis.

Show details

Contents

< Prev Next >

Chapter 7Chlamydia and pelvic inflammatory disease: Population Excess Fraction based on prospective, retrospective and routine data

Objectives

To:

generate and compare estimates of the incidence of PID
generate and compare plausible estimates of the proportion of PID attributable to CT (the PEF)
produce coherent estimates of: CT incidence, PID incidence, the CT-to-PID progression rate, and the PEF.

Introduction

The previous chapter produced estimates of the risk of clinical PID (as defined in Chapter 2) caused by an episode of CT, based on prospective, and, particularly, randomised, data. We now consider the relation between CT and PID from a broader perspective: not only the risk of PID following CT, but also how much PID is attributable to CT. Viewed in isolation, the two questions are technically independent of each other. However, suppose we had independent estimates of (1) CT incidence; (2) PID incidence; (3) the CT-to-PID progression risk; and (4) the proportion of PID attributable to CT, the PEF. It would then be possible to generate a prediction for each of these quantities from the other three. Therefore, estimates of the two quantities are not independent if we examine the available data sources more broadly.

In fact, the relation between untreated CT and PID has been studied in a wide variety of ways, drawing on prospective, retrospective and routine data. Following a MPES approach, we want to know whether the different methods give consistent answers.

Given that previous chapters have dealt with CT incidence and the CT-to-PID progression risk, in order to provide a complete account of CT and PID, we must answer two remaining questions: first, what is the incidence of all-cause PID; and second, how much PID is caused by CT. We begin by describing evidence sources and statistical methodology for each topic. Two sources of evidence of PID incidence are considered: a direct estimate from the control arm of the POPI trial¹⁹ and an estimate derived from routine data. We then consider five separate methods of estimating the PEF in turn. For each estimate we first review the evidence sources available, describing the relationships between the different elements. We then set out the statistical models used to obtain estimates from the data. In the results we assess the consistency of evidence on PID incidence, and compare the various estimates of PEF. In discussion we review some of the key assumptions made and provide a rationale for our choice of PEF estimate to take forward to later chapters. We also present a coherent set of estimates of CT incidence, prevalence and duration, CT-to-PID progression rate, PID incidence and the PEF.

We note here that there is a wider literature on the relation between CT and PID which we have not reviewed here, although it is, on the face of it, highly relevant. In particular, we have not included estimates of the CT-to-PID progression risk based on register studies³⁶^,²¹⁵ because women with CT in these studies are treated, and are therefore not able to inform the CT-to-PID progression risk in untreated infection. Further, although these studies have been used for this purpose in the past,²^,¹⁵ the CDC Task Force⁹⁸ also concluded that such studies could not help determine the relationship between CT and PID. A second set of estimates not reviewed here are those of van Valkengoed et al.,³⁹ who used a complex procedure to estimate the prospective risk of PID, EP and TFI, following CT, from retrospective and routine data sources. These risk estimates are extraordinarily low, and a methodological critique is given in Appendix 9.

Methods: review of evidence sources and statistical estimation

Estimates of the incidence of all-cause pelvic inflammatory disease

This section describes two independent sources of PID incidence estimates, and a third estimate based on pooling them.

Estimates based on the control arm of the Prevention of Pelvic Infection trial

A comparatively direct estimate of $λ_{a}^{A L L P I D}$ can be derived from the control arm of the POPI trial,¹⁹ if we assume that the trial sample is approximately representative of the general female population of the same age, approximately age 16–24 years. In the unscreened arm, 23 (r^POPI) cases of clinical PID were reported in a sample of 1186 (n^POPI) women, aged 16–27 years, followed up for a period of 1 year. Following comments on the definition and ascertainment of PID in Chapters 2 and 6, we assume that all symptomatic PID meeting the ‘probable/definite’ criteria will be ascertained, including those normally undiagnosed. The POPI estimates are therefore ‘grossed up’ to account for the underascertainment of asymptomatic (silent) PID in prospective studies. The POPI¹⁹ control arm estimates $λ_{a}^{A L L P I D}$ . ψ^Sym, where ψ^Sym is the proportion of PID that is symptomatic.

\begin{matrix} r^{P O P I} ~ B i n (p^{P O P I}, n^{P O P I}) \\ p^{P O P I} = 1 - \exp (- λ^{P O P I}) \\ λ^{P O P I} = λ^{A l l P I D} . ψ^{S y m} \\ r^{S y m} ~ B i n (ψ^{S y m}, n^{S y m}) \end{matrix}

(22)

The probability that incident PID is symptomatic ψ^Sym is informed by the 32 out of 36 cases who reported symptoms in the Wolner-Hanssen study.¹⁰³

Incidence of all-cause pelvic inflammatory disease based on routine data

A second, more indirect, estimate of incidence can be derived from PID data routinely collected in England. These data represent only diagnosed ‘probable/definite’ PID, and so must be grossed up by an estimate of the proportion of PID that is diagnosed in a general population context (see below).

There are three sources of data on PID in England: HES,¹¹⁷ GPRD⁹⁰ and KC-60 (the routine returns from STI clinics¹⁹⁸) (Table 18). The three sources pick up cases from different care pathways. HES¹¹⁷ reports the number of cases of PID diagnosed in hospital (codes N70-N74). GPRD⁹⁰ reports all diagnosis made in an approximately 8% sample of patients registered with general practices across the country. These data do not include PID as a separate diagnosis, so we take our estimate from a recent paper that used the recorded clinical descriptions of symptoms to estimate population incidence of definite and probable, and of possible, PID diagnosed in this setting.⁹⁰ Finally, PID diagnosed in STI clinics is reported as ‘complicated chlamydia’. Data shown in Table 18 is for 2002, the final year before the introduction of the National Chlamydia Screening Programme.

TABLE 18

Number of incident cases of PID in England, 2002

To use these data to derive age group-specific estimates of the annual incidence of diagnosed PID, we need to make assumptions about the degree of overlap between the three sources. Consultation with experts suggests that referrals of PID cases to and from departments of GUM (STI clinics) were rare prior to 2008. Referrals from GP practices to hospital are likely to have been more common for severe cases. A primary care guideline for STI management published in 2006²²³ recommended referral to a GUM clinic for mild/moderate cases of PID, and immediate commencement of treatment was advised if an urgent appointment was not available. Prior to 2008, access to GUM services within 48 hours was poor.¹⁰⁵ It was recommended that severe cases should be referred to gynaecology for admission to hospital. We therefore assume that the total of the STI, GPRD and HES data, within each age group, represents an upper bound for the number of PID cases diagnosed in England each year. A minimum was formed by adding the number of GUM cases to either the GPRD or the HES cases, whichever was largest.

Routine data on diagnosed PID, given information on age-specific population sizes N_a, provides information on $λ_{a}^{D i a g P I D}$ . An estimate of total (all cause) rate of PID can then be constructed by using information on the proportion diagnosed, ψ^Diag:

λ_{a}^{Α L L P I D} = \frac{λ_{a}^{D i a g P I D}}{ψ^{D i a g}}

(23)

Binomially distributed data based on Table 18 are used to inform minimum and maximum limits for the annual probability of (diagnosed) PID, and these are used to define the upper and lower limits for $λ_{a}^{D i a g P I D}$ . The WinBUGS ‘cut function’ was used to ensure that this uniform distribution was not updated (see Chapter 3).

\begin{array}{l} r_{a}^{min} ~ B i n o m i a l (p_{a}^{\min}, N_{a}), r_{a}^{\max} ~ B i n o m i a l (p_{a}^{\max}, N_{a}) \\ p_{a} ~ U n i f o r m (p_{a}^{\min}, p_{a}^{\max}) \\ λ_{a}^{D i a g P I D} = - log (1 - p_{a}) \\ λ_{a}^{A L L P I D} = λ_{a}^{D i a g P I D} / ψ^{D i a g} \\ r^{D i a g} ~ B i n (ψ^{D i a g}, n^{D i a g}) \end{array}

(24)

where N_a is the number of women in each age group in England in 2002 from Census data.

The degree of underdiagnosis was based on a single study of 36 infertile women, only 11 of whom (30.6%) reported that they had been diagnosed with PID.¹⁰³ This study reported data with a binomial likelihood.

Synthesis of pelvic inflammatory disease incidence estimates

The two estimates of PID incidence were compared. A DAG (Figure 14) shows how the two estimates were combined. The WinBUGS code and data sets are set out in Appendix 10.

FIGURE 14

Directed acyclic graph showing combination of two sources of evidence on PID incidence.

Population Excess Fraction: how much PID is caused by chlamydia?

In this section we examine estimates of the PEF based on microbiological studies, estimates based on retrospective case–control studies, and estimates derived from the ratio of CT-related PID incidence and all-cause PID incidence, and finally estimates based on randomised trials. DAGs for each of the estimates are shown in Figure 15.

FIGURE 15

Directed acyclic graph showing methods of estimating Population Excess Fraction. Symbols: PEF-1: π_a, CT prevalence in age band a; OR from case–control study. PEF-2 OR* adjusted OR; ψ^Erf correct for under-ascertainment of CT infection (more...)

Uncontrolled studies of the microbial aetiology of PID

Microbiological analysis of samples from the genital tract of women with PID provides direct information on potentially causative organisms. Based on 19 microbiological studies conducted between 1977 and 1992, in a number of countries, Paavonen et al.⁸² reported that CT was involved in 30% of PID cases. It has been argued²⁰ that this must be an underestimate of the current role of CT, as gonorrhoea was a common cause of PID during the period when many of these studies were undertaken, and it is generally agreed that gonorrhoea is a far less common aetiology at present, particularly in Europe.

Simms and Stephenson⁶² provide a summary of studies in which patients with laparoscopically proven PID, in other words salpingitis (see Chapter 2), recruited in various clinical settings, were examined for evidence of current CT infection. The proportion with evidence of current CT in upper genital tract samples varied from 12% to 65%, reflecting considerable variation over time and between countries. The largest UK study,²²⁴ conducted between 1989 and 1993, reported 39%. In another UK study,³⁷ conducted from 2000 to 2002, 42 of 140 (30%) salpingitis cases had evidence of exposure to CT.³⁷

It is tempting to consider these studies as producing a direct estimate of the proportion of PID that is attributable to CT.¹ But this may be misleading for several reasons, the most obvious of which is that a control group is required consisting of women without PID. The proportion of control samples in which we would expect to find evidence of CT is, of course, quite small. The general population prevalence of CT is in the order of only 3% (see Chapter 5).

A more serious weakness is that the results obtained depend markedly on the sites from which samples are taken. In particular, the presence of CT in the upper genital tract, which is more likely to be causally related to PID, may not be well predicted by its presence in the lower genital tract.²²⁵^,²²⁶ These issues were clarified by a recent study in Erfurt, Germany,²²⁵ looking at 363 women with laparoscopically confirmed PID. CT was found in the genital tract of 103 (28.4%), and in 55 (15.2%) it could be isolated from the cervix. In 23 (6.3%), CT was isolated from both the cervix and the fallopian tubes, whereas in 47 (12.9%) CT was isolated in the fallopian tubes only.

If we take infection in the fallopian tubes as the causal factor in subsequent reproductive damage, this implies that studies based on cervical isolates may be underestimating the role of CT in PID-related reproductive damage by a factor of 103/(103–47) = 1.84. The Erfurt study²²⁵ found CT in the fallopian tubes of just under 20% of cases of laparoscopically confirmed PID. Few studies of PID routinely sample both the cervix and fallopian tubes for CT, so direct comparisons are difficult. However, high titre antibody can be used as a surrogate for upper genital tract infection.⁸⁵ Taylor-Robinson et al.⁸⁸ also observed that infection at the cervix probably underestimates the role of CT in PID-related reproductive damage by a similar factor 16/(16–6), or 1.6. This was based on the observation that of the 22 women diagnosed with acute salpingitis on laparoscopy: 10 had CT detected at the cervix and an additional six had high-titre serum CT IgG antibody.

These considerations help to quantify the likely degree of underestimation of the OR, and hence the PEF, from case–control studies based on lower genital tract samples, and this can be used to adjust estimates from those studies.

Population Excess Fraction based on case–control studies

We noted in Chapter 3 that the presence of positive confounding in estimates of the RR of the disease (PID) given the exposure (CT), if based on observational data, would tend to result in overestimation of the PEF. This can be mitigated to some extent by choosing study designs in which the risk of confounding is minimised. For example, it is preferable to use a marker of current CT infection rather than a marker of previous CT infection. We have therefore used data only from case–control studies that use current infection as a marker of exposure. (We set aside, here, the further difficulties with the serological markers used in the published literature, namely their poor sensitivity and specificity: this is discussed at greater length in Chapter 11.)

Identification of case–control studies on CT prevalence in women with PID and control subjects

We searched for studies using current CT infection as a marker of CT exposure. Strictly speaking, in a synthesis focusing on prevalence and sequelae of CT in the UK, only contemporary UK data should be used. We have, nevertheless, used data from Europe on the basis that the epidemiology of STIs is generally similar in Western European countries. Studies from North America have been excluded because gonorrhoea has tended to have a more important role in the aetiology of PID in North America than the UK.²²⁷^–²²⁹ We also excluded studies published before the 1990s. Our literature identification process (see Chapter 3) identified three studies,³¹^,³⁷^,²³⁰ shown in Table 19.

TABLE 19

Odds ratios from retrospective studies

Estimates of the Population Excess Fraction from case–control studies, PEF-1 and PEF-2

A pooled OR was estimated from these studies (see below), and used to derive an estimate of the PEF, using equation 1. Although the same OR is assumed to apply to women of all ages, the formula will generate age-specific estimates of PEF, because age-specific estimates of CT prevalence from Chapter 5 are entered, as follows:

{P E F}_{a}^{(1)} = γ_{a}^{(1)} = \frac{π_{a} . (R R - 1)}{π_{a} . (R R - 1) + 1}

(25)

The six data points are used to estimate four parameters: three study-specific ‘baselines’, the log odds in the control groups µ_s, with s indexing study, and one ‘FE’ log OR β. Using a standard logistic regression model, with 0 for controls, 1 for PID cases:

\begin{array}{l} logit (λ_{s, 0}) = µ_{s} \\ logit (λ_{s, 1}) = µ_{s} + β \end{array}

(26)

The OR can be recovered via: OR = exp(β), and used as a RR in (see equation 25) (see Chapter 3) to generate a set of estimates, PEF-1.

A second set of estimates (PEF-2) were generated by adjusting the OR from the retrospective studies for under-ascertainment of CT in the PID cases, using the 1.84 estimate from the Erfurt study.²²⁵ The uncertainty in the adjustment is taken into account by setting: OR^Adj = OR/ψ^Erf, where ψ^Erf ∼ Beta(56,47).

Estimates of Population Excess Fraction based on CT-related and all-cause PID incidence, PEF-3

Adams et al.¹ obtained estimates of all-cause PID incidence in 16- to 44-year-olds from a GP-based study.²³¹ Based on a retrospective study,³⁷ they reckoned that a maximum of 30% of PID could be attributed to CT, and used this to identify a range of estimates of CT-related PID incidence. They then multiplied a number of CT-to-PID progression rates into estimates of CT incidence, to identify progression rates that were consistent with their results on CT-related PID. We mention this study,²³¹ not to review the findings in detail but simply to draw attention to the general method, which we use below in an inverted form to estimate the PEF, although with different data inputs.

Here, the age-specific PEF is taken to be the ratio of the incidence of CT-related PID to the incidence of all-cause PID, where the incidence of CT-related PID is the product of CT incidence (see Chapter 5) and the CT-to-PID progression rate (see Chapter 6):

{P E F}_{a}^{(3)} = γ_{a}^{(3)} = \frac{λ_{a}^{C T} θ^{C T \to P I D}}{λ_{a}^{A L L P I D}}

(27)

To implement this, we pool two independent sources for estimating all-cause PID incidence (see below), and estimates of CT incidence and CT-to-PID progression rate from Chapter 6.

For these last two parameters, we use a multivariate normal approximation to the posterior distribution estimated in the incidence, prevalence, duration synthesis (see Chapter 5), and this is entered as data in the calculations. Specifically, incidence and duration were normal on the log scale, the proportion symptomatic was normal on the natural scale, and prevalence was normal on a logit scale. Similarly, a normal likelihood for CT-to-PID progression risk is derived from the posterior analysis in Chapter 6.

Estimates of Population Excess Fraction based on screening trials, PEF-4 and PEF-5

The Scholes trial¹³ compared PID risk 1 year later in women randomised to screening and treatment and women randomised to no screening. The observed RR was 0.44 (95% CI 0.20 to 0.88). This represents a RRR of 0.56, implying that 56% of PID is due to CT in the study population, and this can be interpreted as an estimate of the PEF (see Chapter 3). One of the early criticisms of the trial¹⁸ was that this was an unrealistically optimistic result because it conflicted with the microbiological studies cited above, apparently showing that 30% of PID was CT related. Gottlieb et al.,²²⁶ in an important review, also draw attention to this apparent discrepancy, but points out that a similar RR was observed in the Ostergaard trial.¹²

An approximate PEF can be derived from the RR of PID in the POPI trial, using equation 3.¹² There were 23 PIDs in the untreated group of 1186, and 15 in the treated group of 1191. An estimate of the PEF is therefore:

\begin{array}{l} p^{C t r l} ~ B e t a (23,1163), p^{T r t} ~ B e t a (15,1176) \\ P E F^{(4)} = γ_{a}^{(4)} = \frac{p^{C t r l} - p^{T r t}}{p^{C t r l}} \end{array}

(28)

However, this estimate is likely to be biased because 29 out of 67 (43%) untreated controls, who were initially CT+ undertook to be tested independently. We can assume, therefore, that the excess risk of PID in the untreated controls, p^Ctrl – p^Trt has been underestimated by a factor of 1 – p^IT, where p^IT is the probability that a CT+ control is independently tested and treated. Accordingly, we can form an adjusted estimate:

\begin{array}{l} p^{I T} ~ B e t a (29,38) \\ p_{*}^{C t r l} = (\frac{p^{C t r l} - p^{T r t}}{1 - p^{I T}}) + p^{T r T} \\ P E F^{(5)} = γ_{a}^{(5)} = \frac{p_{*}^{C t r l} - p^{T r t}}{p_{*}^{C t r l}} \end{array}

(29)

This should be regarded as a ‘maximum’ correction, as it assumes that all the independent testing took place at the outset.

Results

Consistency of evidence on pelvic inflammatory disease incidence

Table 20 provides a series of age-specific estimates for the incidence of (all cause) PID, from the different data sources. The ‘direct’ estimate from the POPI trial,¹⁹ 2.0% per year – when adjusted for under-ascertainment due to asymptomatic presentation – represents a 2.4% per year incidence. This is consistent with the estimate derived from routine data, 3.0% per year for that age range, following adjustment for the underdiagnosis inherent in routine PID statistics. Figure 16 shows the posterior distributions of PID incidence from the two sources, and the pooled estimate, and establishes the consistency of the estimates. The consistency of registry data with the POPI study¹⁹ can be studied only for the 16- to 24-year-old age group.

TABLE 20

Results of synthesis of evidence on the incidence of all-cause PID in England

FIGURE 16

Evidence consistency plot for the incidence of all-cause PID in England in women aged 16–24 years.

Note that when combining the direct and indirect evidence sources, the incidence information back-propagates to ‘update’ the estimate of the proportions of PID that is undiagnosed. This was 31% in the original study but is estimated to be 36% following synthesis with the POPI¹⁹ data.

The final pooled PID incidence estimates are 2.5% per year (95% CrI 1.8% to 3.4%) in women aged 16–24 years, and 1.6% (95% CrI 1.1% to 2.2%) in women aged 25–44 years. Taking into account the age structure of the female population, we can estimate that 62.9% (95% CrI 57.8% to 67.4%) of PID episodes occur in women over 24 years.

Estimates of the Population Excess Fraction

The synthesis of the three retrospective studies³¹^,³⁷^,²³⁰ of CT in women with PID and control subjects (see Table 19) produces a posterior mean OR of 9.2 (95% CrI 4.4 to 18.1). The model fitted well (residual deviance of 5, compared with six data points). The OR adjusted by the under-ascertainment factor from the Erfurt study²²⁵ is 17.1 (95% CrI 7.9 to 34).

The several estimates of the PEF are shown in Table 21. The adjustment for under-detection of CT infection in the case–control studies, based on the Erfurt study,²²⁵ almost doubles the overall PEF (women aged 16–44 years) from 14.5% to 24.6%. The dramatic fall in PEF with age is no more than a reflection of our assumption that the OR is not related to age, whereas the incidence and prevalence of PID varies, as is evident from the routine data (see Table 18).

TABLE 21

Alternative estimates of the PEF, the percentage of PID caused by CT

The PEF-3 estimates (column 3) based on the ratio of CT-related PID to all-cause PID also show a marked falling off with age. It is reassuring that the adjusted estimates based on case–control studies are close to estimates derived in a very different way from the progression rate, CT incidence and all-cause PID incidence.

The direct estimate of the PEF based on the POPI trial,¹⁹ PEF-4, is 35.7%, but if this is corrected for independent testing in the deferred treatment (control) arm, this becomes 49.4%. The CrIs on both estimates are wide, and span zero because the difference in the crude PID rates between the two trial arms itself is too uncertain to rule out no difference with 95% confidence.

Discussion

The chapter establishes that routine data on PID incidence collected from GP, hospital and STI clinic sources, taking account of the likely overlap between these sources and the extent of undiagnosed PID, is consistent with an estimate of PID derived from the POPI¹⁹ trial.

Second, setting aside the actual estimates obtained, we have established for the first time a coherent framework in which CT incidence and prevalence, PID incidence, CT-to-PID progression rate, and the PEF can all be considered together. The coherent set of estimates obtained, for the first time in any country or region, reflect what we feel is the totality of evidence on all these parameters, subject, of course, to our interpretation of the evidence sources.

Although one might wish for greater precision in each one of the data inputs that have been examined here, we would suggest that the analysis lends a degree of robustness to our interpretation of all of the data sources, but particularly to our quantitative information on the causal connection between incident CT and PID. This estimate of the causal connection between CT and PID is pivotal, as the further downstream sequelae of CT, namely EP and infertility, are viewed as arising from CT-related PID, and not from CT in the absence of PID.

The general consistency between the various estimates of PEF, based on fundamentally different sources, is reassuring, although the CrIs are wide. The two direct estimates based on the POPI trial,¹⁹ PEF-4 and PEF-5, probably represent lower and upper limits, and they are close to estimates from previous trials. Interestingly, of the 26 PID patients in the POPI trial¹⁹ who were tested, 16 (61.5%) were CT+ at the time the PID was diagnosed. Of the 16 who were CT– at the start of the trial, 10 (62.5%) were CT+ at diagnosis. These further observations from the POPI trial¹⁹ suggest that that the corrected estimate may be closer to the truth.

Although the failure to find inconsistency raises the credibility of the analyses, confidence in the estimates has to be tempered by their relatively low precision, which makes inconsistency difficult to demonstrate. Moreover, each of the analyses are subject to a number of caveats.

Limitations of the analysis

Overlap between sources of routine data on incident pelvic inflammatory disease

One important limitation is the poor information currently available on the degree of overlap between the three routine data sources contributing information on incident PID (see Table 18). Our assumption was that the true number of incident cases was from a uniform distribution bounded by a ‘minimum’ (GUM plus the largest of GPRD and HES) and a ‘maximum’ in which there was no overlap. Although this seems a reasonable way to use the data available, the ‘gap’ between the minimum and maximum estimates was 30–37% of the lower estimate, depending on age. A recent paper on PID management based on GPRD noted that only 56% of the 3669 cases examined were managed entirely within the practice. If the age-specific GPRD figures are divided by 0.56, the number obtained lies between the minimum and maximum, which provides some support for the procedure we have followed.⁹⁷

Obtaining a more precise estimate of the number of incident cases, by collecting data on referrals between GPRD, hospital and GUM clinics, could improve the accuracy of estimates considerably. This would be a useful avenue for further research (see Chapter 12).

Proportion of pelvic inflammatory disease patients diagnosed and ascertained

Our estimate of the proportion of PID episodes diagnosed is based on a single study looking at infertile women with evidence of past PID.¹⁰³ This study¹⁰³ has informed two key parameters which have been used to ‘gross up’ each of two estimates of PID incidence. First, we grossed up the estimate seen in the POPI trial,¹⁹ to account for the proportion of ‘silent’ PID that would not be observed even in a prospective study. Second, we use it to gross up evidence from routine data, to account for both silent PID and for symptomatic PID that does not lead to medical treatment or which is not diagnosed. This single, small study,¹⁰³ carried out in another country 30 years ago, is being asked to carry quite a considerable weight and this must be recognised as a limitation of our analyses.

Retrospective data and confounding

As noted earlier in this chapter and elsewhere, the use of equation 25 to estimate the PEF from retrospective data is inherently likely to overestimate it as a result of positive confounding with other exposures, both other STIs that are known to be causal agents for PID, and non-STI infections that can be transmitted as a result of sexual intercourse. It is therefore possible that the PEF-1 estimate, and particularly the PEF-2 estimate, which is adjusted for probable under-ascertainment of CT infection in women with PID, represent overestimates. On the other hand, this adjustment may not fully account for under-ascertainment if a proportion of women with CT-related PID may clear their infection before the diagnosis of PID is made.⁸⁸^,²²⁶ In addition, estimates of PEF are time- and place-dependent: the extent to which we can generalise from these results to the UK situation is very hard to assess. Therefore, although the PEF-1 and PEF-2 estimates based on retrospective data are of interest, we conclude that they are biased to an extent that is had to quantify, and they play no further role in this report.

Impact of age on the relationship between Chlamydia trachomatis and pelvic inflammatory disease

The PEF estimates show a clearly declining trend with age. In the retrospective estimates, this follows as a consequence of assuming a constant OR and applying this to a prevalence that declines steeply with age. In the case of PEF-3, the decline with age is partly due to the decline in CT incidence with age (exactly mirroring the decline in prevalence with age), the age profile of PID in the routine data, and the assumption that the proportion of PID that is undiagnosed is constant over age.

Although the decrease in PEF with age is to some extent a result of our assumption that neither the probability that PID is diagnosed nor the risk of PID following CT are age-dependent, the degree of variation with age, a factor of 4–6 between age 16–19 years and age 35–44 years, is so extreme that we would have to propose quite extreme trends in one or both of those quantities to reverse it. The proposal that PEF decreases with age is, apparently, not one that has been made before, but it is consistent with the observation that estimates based on the trial evidence tend to be higher because the trials are heavily weighted towards younger women.

Age dependency in PEF, if confirmed (see Chapter 2 for discussion of alternative explanations), would have significant impact on the public health importance of CT, as the majority of EP and TFI occurs in older women. Although it is certainly possible that CT infections in younger women have a key role in reproductive health problems that emerge many years later, these results focus attention on the distinctly different age profiles of CT and PID. There are insufficient data in either the prospective or retrospective studies to address this question more fully. Further work to examine these issues directly is proposed in our research recommendations.

Post-hoc element in choice of estimates

We record here that, when the work in this chapter was first carried out, we included only the unadjusted (PEF-1), and we pooled that information with the evidence sources contributing to estimate PEF-3. At that time we believed that the estimates based on retrospective studies must be upper bounds, and we did not examine the trial-based estimates. After we had completed the work on TFI (see Chapters 10 and 11), we became aware that the PEF-1 estimates were too low to be consistent with the results of those chapters. We then reconsidered the relationship between CT and PID, and re-examined the trial-based estimates of PEF, particularly in the light of the comments in Gottlieb et al.²²⁶ At the same time, results from the recent Erfurt study²²⁵ made it clear that the retrospective estimates were not necessarily upper bounds. There is, therefore, a post-hoc element to the reasoning in this chapter. The ‘hypothesis-generating’ aspect to our findings on PEF are reflected in the research recommendations. Nevertheless, that fact that two estimates of PID incidence are consistent with each other, and with an account of the PEF that is consistent with other independent sources of data, to some extent mitigates the post-hoc process in our investigation.

Coherent estimates of CT and PID incidence, CT-to-PID risk, and Population Excess Fraction

There is an option to strengthen our estimate of the PEF by pooling PEF-3 with either the adjusted retrospective evidence PEF-2, or the evidence directly derived from the POPI trial,¹⁹ PEF-4 or PEF-5, or both. We have chosen not to do this, on the basis that the PEF-2 estimate is vulnerable to many biases, and relates to a different time and place, as discussed above. The estimate of PEF from the POPI trial¹⁹ is not statistically independent of the estimate of the CT-to-PID risk estimate, and is also subject to uncertainties described above. However, we regard the similarity of these independently derived estimates as an important validation.

In taking forward the PEF-3 estimate $γ_{a}^{C T \leftarrow P I D}$ , which is calculated from the CT incidence estimates in Chapter 5, the CT-to-PID progression risk in Chapter 6, and all-cause PID incidence in this chapter, we have generated a fully coherent set of estimates for CT incidence, prevalence, duration and the proportion of infection that is symptomatic; PID incidence, the proportion of PID that is symptomatic and the proportion that is diagnosed, the CT-to-PID progression risk, and the PEF. The summary posterior means and CrIs can be found in Tables 13 (full synthesis model), 17 (trials only), 20 (synthesis model) and 21 (column 3).

These estimates are carried forward to the analyses in the subsequent chapters.

Summary of assumptions and findings

Summary of assumptions

The probability that PID is diagnosed is independent of age.
The risk of PID due to CT is independent of age.
The probability of diagnosis is the same in CT- and non-CT-related PID.
In prospective studies, including the screening trials, all PID – except for asymptomatic PID – is diagnosed.
In routinely collected data on PID, a lower proportion of all PID is diagnosed than in prospective studies.

Summary of findings

Two estimates of PID incidence in the UK were consistent: one derived from the control arm of the POPI trial,¹⁹ and the other from routine statistics combined with evidence on the proportion of PID that is diagnosed.
A pooled estimate of PID incidence in age 16- to 24-year-olds is 2.5% per year (95% CrI 1.8 to 3.4), and 1.8% per year (95% CrI 1.3% to 2.5%) in 16- to 44-year-olds.
36% (95% CrI 26% to 48%) of PID that causes reproductive damage is diagnosed; 12.6% (95% CrI 4 to 25) is asymptomatic.
Estimates of the PEF were compared: (1) an estimate based on retrospective studies adjusted for under-ascertainment; (2) an estimate based on CT incidence, CT-to-PID progression rate, and PID incidence; and (3) estimates based on the POPI¹⁹ trial. There was no evidence of inconsistency although Crls were wide.
We estimate that 62.9% of PID episodes (95% CrI 57.8% to 67.4%) occur in women aged over 24 years.
The PEF declines sharply with age.
Our estimate of PEF, derived from CT and PID incidence data and the CT-to-PID progression risk is 19.7% (95% CrI 5.9% to 38.1%) in women aged 16–44 years, and 35.3% (95% CrI 10.5% to 68.5%) in women aged 16–24 years.

The analyses thus far in the report establish a set of coherent estimates for the UK of CT incidence, CT prevalence, CT duration, probability that an incident CT infection is symptomatic, PID incidence, proportion of PID that is diagnosed, proportion of PID that is symptomatic, the CT-to-PID progression, and the PEF of PID due to CT.

Copyright © Queen’s Printer and Controller of HMSO 2016. This work was produced by Price et al. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Included under terms of UK Non-commercial Government License.

Bookshelf ID: NBK350646

Contents