Risk of Pelvic Inflammatory Disease in Relation to Chlamydia and Gonorrhea Testing, Repeat Testing, and Positivity: A Population-Based Cohort Study

Abstract

Background

There is uncertainty around whether the risks of pelvic inflammatory disease (PID) differ following Chlamydia trachomatis (chlamydia) and Neisseria gonorrhoeae (gonorrhea) infection. We quantified the risk of PID associated with chlamydia and gonorrhea infection and subsequent repeat infections in a whole-population cohort.

Methods

A cohort of 315123 Western Australian women, born during 1974–1995, was probabilistically linked to chlamydia and gonorrhea testing records and to hospitalizations and emergency department presentations for PID from 2002 to 2013. Time-updated survival analysis was used to investigate the association between chlamydia and gonorrhea testing, and positivity, and risk of PID.

Results

Over 3199135 person-years, 120748 women had pathology test records for both chlamydia and gonorrhea, 10745 chlamydia only, and 653 gonorrhea only. Among those tested, 16778 (12.8%) had ≥1 positive chlamydia test, 3195 (2.6%) ≥1 positive gonorrhea test, and 1874 (1.6%) were positive for both. There were 4819 PID presentations (2222 hospitalizations, 2597 emergency presentations). Adjusting for age, Aboriginality, year of follow-up, health area, and socioeconomic status, compared to women negative for chlamydia and gonorrhea, the relative risk (adjusted incidence rate ratio) of PID was 4.29 (95% confidence interval [CI], 3.66–5.03) in women who were both chlamydia and gonorrhea positive; 4.54 (95% CI, 3.87–5.33) in those only gonorrhea positive; and 1.77 (95% CI, 1.61–1.94) in those only chlamydia positive.

Conclusions

Gonorrhea infection conferred a substantially higher risk than chlamydia of hospitalization or emergency department presentation for PID. The emergence of gonorrhea antimicrobial resistance may have a serious impact on rates of PID and its associated reproductive health sequelae.

pelvic inflammatory disease, chlamydia, gonorrhea, reproductive health

Pelvic inflammatory disease (PID) is a common and potentially serious infection of the female reproductive organs. Approximately 4.1% of sexually active young women in the United States report having been treated for PID in their lifetime [1]. If left untreated, PID can cause scarring and dysfunction of the genital tract, which may go on to cause ectopic pregnancy, chronic pelvic pain, and infertility [2–4]. PID has been commonly associated with the sexually transmitted infections Chlamydia trachomatis (chlamydia) [5–8] and Neisseria gonorrhoeae (gonorrhea) [9, 10]; however, other etiological agents such as Mycoplasma genitalium [10, 11], Trichomonas vaginalis [12], and organisms associated with bacterial vaginosis [13] have been reported in the literature.

Chlamydia and gonorrhea are 2 of the most frequently reported sexually transmitted infections (STIs) globally [14]. Although gonorrhea notification rates are lower than those of chlamydia, gonorrhea disproportionally affects discrete population groups. In the United States in 2015, the rate of gonorrhea in blacks was 9.6 times the rate among whites [15], whereas in Australia in 2015, the gonorrhea notification rate in the Aboriginal and Torres Strait Islander population (hereafter referred to as Aboriginal) was 10 times that of the non-Aboriginal population, increasing to 72 times higher in remote areas [16]. Such disparities may be due to reduced access to health services and differences in testing patterns, as well as in sexual risk behaviors [17, 18]. There is also evidence of an emerging gonorrhea epidemic among heterosexuals in some urban areas of Australia [19], although the reason underlying this increase is unknown.

Studies have suggested that PID following gonorrhea infection may be more clinically severe [10, 20, 21]. With concerns over the emerging antimicrobial resistance of gonorrhea [22] and a potential epidemic in some heterosexual populations [19], the role gonorrhea plays in the development of PID is of increasing importance. Our aim was therefore to compare the risks of PID associated with single and repeated chlamydia and gonorrhea infections in a large cohort of reproductive-aged women.

METHODS

Study Population and Linkage

This study was conducted using population-based record linkage in the Australian state of Western Australia (WA) (population 2.6 million). WA has Australia’s oldest and most comprehensive record linkage system [23]. It maintains probabilistic linkages between core health datasets using personal identifiers such as name, date of birth, address, and sex. Linkage accuracy using this process is high with an error rate estimated at 0.11% [24].

A cohort comprising women of reproductive age residing in WA was constructed using 2 datasets: the WA Birth Registrations Data Collection, which contains a record of all children born and registered in WA from 1974 onward, and the 2014 WA Electoral Roll. Registration on the electoral roll is compulsory in Australia. Eligible women were those with either a birth registration, or a record on the Electoral Roll with year of birth between 1974 and 1995.

The cohort was probabilistically linked to the Hospital Morbidity Data System, the Emergency Department Data Collection, laboratory chlamydia and gonorrhea test records, and death registrations.

Hospitalization and Emergency Department Presentation for Pelvic Inflammatory Disease

The Hospital Morbidity Data System is a statutory data collection including all hospitals in WA (public and private). It comprises inpatient hospital discharge (separation) records dating back to 1970, with every record assigned a principal diagnosis code and up to 20 additional diagnosis fields. The Emergency Department Data Collection contains data on emergency department presentations in WA’s public hospitals as well as private hospitals under contract with the WA government from 2002 onward. Women were classified as having a diagnosis of PID if they had a linked hospital admission or emergency department presentation record where the primary International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM) diagnosis was any of N70–N73, N74.1–N74.8, or A18.1, A51.4, A52.7, A54.2 A56.1, with an additional diagnosis of N74.1–N74.8. If a hospitalization for PID occurred within 7 days of the emergency department presentation, then the diagnosis was classed as a hospitalization and the hospitalization date was selected. In all other cases, the first PID presentation was selected.

Chlamydia and Gonorrhea Testing

Records of all chlamydia and gonorrhea nucleic acid amplification tests (NAATs) conducted between January 2001 and December 2013 at 2 large WA pathology laboratories were also linked to the cohort. The date of specimen collection and test result (positive, negative, or equivocal/undetermined) were supplied to the research team. Multiple tests for the same infection on the same referral date were counted as 1 test and considered positive if any of the results were positive. Tests were then limited to 1 test per infection within a 30-day period.

Analysis

Time-updated survival analysis was used to investigate the association between chlamydia and gonorrhea diagnoses and risk of emergency department presentation or hospitalization for PID. For the main analysis, women were classified into 5 categories:

1. Negative for chlamydia and gonorrhea: women tested negative for both chlamydia and gonorrhea, women tested negative for chlamydia and no linked record of a gonorrhea test, and women tested negative for gonorrhea and no linked record of a chlamydia test.

2. Chlamydia positive: women with at least 1 positive chlamydia test and either no gonorrhea test or a negative gonorrhea test.

3. Gonorrhea positive: women with at least 1 positive gonorrhea test and either no chlamydia test or a negative chlamydia test.

4. Both chlamydia and gonorrhea positive: women with at least 1 positive chlamydia test and at least 1 positive gonorrhea test at some time during the study (not necessarily on the same date).

5. No test record: women with no linked record of chlamydia or gonorrhea test at either of the 2 laboratories.

All women were initially classified as “no test record” and contributed person-time in that category until such time as their first NAAT record. They then moved into another category and contributed person-time to that category depending on the test result. Only tests after the women’s 15th birthday were included [25].

To investigate the association of repeat testing with PID, women were reclassified into 8 categories according to the number of positive and negative tests they had during follow-up (1 negative test, 2 tests all negative, ≥3 tests all negative, 1 chlamydia-positive test, ≥2 chlamydia-positive tests, 1 gonorrhea-positive test, ≥2 gonorrhea-positive tests, and no test record). Women both chlamydia and gonorrhea positive were classified according to how many positive gonorrhea tests they had.

Women were eligible for inclusion in analyses from their 15th birthday or 1 January 2002 (the first date Emergency Department Data Collection records were available), whichever was later. All women were followed until either a diagnosis of PID, death, or 31 December 2013 (the last date of complete records).

The incidence of PID was calculated in the categories outlined above. Poisson regression using generalized estimating equations was used to investigate the association with chlamydia and gonorrhea diagnoses and risk of PID. In addition to chlamydia and gonorrhea diagnoses, age (continuous) and year of follow-up (continuous) were included as time-updated variables. Models were also adjusted for Aboriginality, health area, and socioeconomic status (based on the Index of Relative Socioeconomic Advantage and Disadvantage 2011 [26]). Aboriginality was determined from the Indigenous Status Flag variable [27]. Health area (metropolitan, rural, and remote) was determined through residential postcode information on their electoral roll record or most recent linked record. Women residing outside of WA or with no postcode information were excluded.

Subgroup analysis examined if chlamydia and gonorrhea testing and positivity and risk of PID differed between Aboriginal and non-Aboriginal women, or by: attained age (15–24 vs ≥25 years), socioeconomic group (low vs high), health area (metropolitan vs regional/remote), and year of testing (2002–2007 vs 2008–2013).

All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, North Carolina).

This study was approved by the Government of Western Australia Department of Health HREC (reference number 2012/73) and the Western Australian Aboriginal Health Ethics Committee (reference number 470).

RESULTS

A total of 315123 women born between 1974 and 1995, were included and followed for a total of 3199135 person-years (mean, 10 years; standard deviation [SD], 2.8 years) (Figure 1). The demographic and testing characteristics of the cohort are summarized in Table 1. The mean age at entry was 18 years (SD, 4.3 years), 14521 (4.6%) women were Aboriginal, and 67613 (21.5%) resided in rural or remote areas.

Figure 1.

Formation of the cohort through data linkage. Abbreviations: NAAT, nucleic acid amplification test; WA, Western Australia.

Table 1.

Demographic Characteristics and Sexually Transmitted Infection Testing Patterns of Cohort

Variable		No. (% of Total)
Total		315123 (100)
Year of birth	1974–1979	85172 (27.0)
	1980–1984	72698 (23.1)
	1985–1989	73971 (23.5)
	1990–1995	83282 (26.4)
Aboriginal	No	300602 (95.4)
	Yes	14521 (4.6)
Area of residence	Metropolitan	247510 (78.5)
	Rural	35827 (11.4)
	Remote	31786 (10.1)
Socioeconomic status	Lower	157116 (49.9)
	Higher	158007 (50.1)
STI nucleic acid testing	Chlamydia and gonorrhea	120748 (38.3)
	Chlamydia only	10745 (3.4)
	Gonorrhea only	653 (0.2)
	No test record	182977 (58.1)
		No. (% of Those Tested)
STI test positivity	Chlamydia and gonorrhea	1874 (1.6)
	Chlamydia only	14904 (11.3)
	Gonorrhea only	1321 (1.1)

Abbreviation: STI, sexually transmitted infection.

During follow-up, 38.3% (120748) had at least 1 linked test record for both chlamydia and gonorrhea, 3.4% (10745) had at least 1 linked chlamydia test but no gonorrhea test, and 0.2% (653) had at least 1 gonorrhea test but no chlamydia test. Among those tested, 16778 (12.8%) had at least 1 positive chlamydia test, 3195 (2.6%) had a least 1 positive gonorrhea test, and 1874 (1.6%) were positive for both chlamydia and gonorrhea, of whom 1209 (64.6%) tested positive for both concurrently (on the same day).

A total of 4819 women were diagnosed with PID during our study period (2222 hospitalizations and 2597 emergency department presentations); PID incidence was 1.5 (95% confidence interval [CI], 1.5–1.6) per 1000 person-years. Among these women, 845 (18%) had their first chlamydia and/or gonorrhea test, and 381 (8%) were positive (189 chlamydia, 141 gonorrhea, and 51 both) on the same day as their PID diagnosis.

PID incidence per 1000 person-years was highest in those who were both chlamydia and gonorrhea positive (incidence rate [IR], 24.4; 95% CI, 21.3–27.5) and those only gonorrhea positive (IR, 23.9; 95% CI, 20.6–27.2) followed by those only chlamydia positive (IR, 7.6; 95% CI, 7.0–8.2) and then those negative for chlamydia and gonorrhea (IR, 3.8; 95% CI, 3.7–4.0). The incidence of PID was lowest among women with no record of a chlamydia or gonorrhea test (IR 0.5; 95% CI, .5–.6); Figure 2. Of women only gonorrhea positive and those both chlamydia and gonorrhea positive, the proportion of PID presentations resulting in hospitalization was 85% and 87%, respectively, compared with 43% in women only chlamydia positive (P < .0001). Among women hospitalized for PID, those with positive tests for either chlamydia or gonorrhea spent a median of 2 days in hospital (interquartile range [IQR], 1–3 days) compared with women never tested and women negative for chlamydia and gonorrhea who spent a median of 1 day in hospital (IQR, 0–2 days; P < .0001).

Figure 2.

Crude incidence rate of pelvic inflammatory disease by chlamydia and gonorrhea testing and positivity.

After adjusting for age, Aboriginality, year of follow-up, health area, and socioeconomic status, the risk of PID, when compared to women negative for chlamydia and gonorrhea, was >4 times higher in women who were both chlamydia and gonorrhea positive (adjusted incidence rate ratio [aIRR], 4.29; 95% CI, 3.66–5.03; P < .0001) and those only gonorrhea positive (aIRR, 4.54; 95% CI, 3.87–5.33; P < .0001) (Table 2). Women who were only chlamydia positive had a 1.77 times (95% CI, 1.61–1.94; P < .0001) higher risk of PID compared with women negative for chlamydia and gonorrhea.

Table 2.

Association Between Chlamydia and Gonorrhea Testing and Risk of Pelvic Inflammatory Disease Among Women of Reproductive Age

Test Status	PID	PYFU	Age Adjusted		Fully Adjusteda
			IRR (95% CI)	P Value	IRR (95% CI)	P Value
Negative for chlamydia and gonorrhea	2496	655709	1.00 (ref)		1.00 (ref)
Chlamydia positive	573	75508	1.81 (1.65–1.99)	<.0001	1.77 (1.61–1.94)	<.0001
Gonorrhea positive	201	8422	5.88 (5.05–6.84)	<.0001	4.54 (3.87–5.33)	<.0001
Chlamydia and gonorrhea positive	238	9762	5.76 (5.01–6.62)	<.0001	4.29 (3.66–5.03)	<.0001
No test record	1311	2449733	0.12 (.11–.13)	<.0001	0.12 (.11–.13)	<.0001
Test Status	PID	PYFU	Age Adjusted		Fully Adjusteda
			IRR (95% CI)	P Value	IRR (95% CI)	P Value
1 negative test	1274	403724	1.00 (ref)		1.00 (ref)
2 tests all negative	600	142769	1.34 (1.24–1.51)	<.0001	1.40 (1.27–1.54)	<.0001
≥3 tests all negative	622	109217	1.90 (1.72–2.09)	<.0001	1.93 (1.75–2.14)	<.0001
1 chlamydia-positive testb	487	65606	2.16 (1.95–2.40)	<.0001	2.17 (1.95–2.42)	<.0001
≥2 chlamydia-positive testsb	86	9902	2.49 (2.00–3.11)	<.0001	2.50 (2.01–3.13)	<.0001
1 gonorrhea-positive testc	316	13682	6.74 (5.92–7.68)	<.0001	5.52 (4.78–6.37)	<.0001
≥2 gonorrhea-positive testsc	123	4502	8.12 (6.69–9.86)	<.0001	7.12 (5.75–8.81)	<.0001
No test record	1311	2449733	0.15 (.14–.16)	<.0001	0.14 (.13–0.15)	<.0001

Abbreviations: CI, confidence interval; IRR, incidence rate ratio; PID, pelvic inflammatory disease; PYFU, person-years of follow-up.

^aFully adjusted model included age, Aboriginality, year of follow-up, health area, socioeconomic status.

^bIncluded chlamydia-positive only women who were either negative for gonorrhea or never tested for gonorrhea.

^cIncluded gonorrhea-positive only and both chlamydia- and gonorrhea-positive women.

Table 2 shows there was also an increasing risk of PID with more testing. Compared to those with only 1 negative test, those with 2 negative tests and those with ≥3 negative tests were more likely to have PID (P_trend < .0001). Results also suggest a higher risk of PID in women who had ≥2 positive chlamydia tests compared with 1 positive chlamydia test and women who had ≥2 positive gonorrhea tests compared with 1 positive gonorrhea test, although tests for statistical difference were nonsignificant (P = .26 and P = .06, respectively).

The rate of PID in Aboriginal women was double that of their non-Aboriginal counterparts (aIRR, 2.00; 95% CI, 1.82–2.20; P < .0001). Figure 3 shows that even among those women who were chlamydia and gonorrhea negative, the incidence rates of PID in Aboriginal women were consistently higher compared with non-Aboriginal women. Figure 3 also shows that the relatively higher proportion of PID hospital admissions compared to emergency department presentations in gonorrhea-positive women was observed in both Aboriginal and non-Aboriginal women. Despite differences in PID rates between Aboriginal and non-Aboriginal women, the relative risks of PID among those with positive gonorrhea tests were consistently higher than among those with positive chlamydia tests (compared to women negative for chlamydia and gonorrhea) for both Aboriginal and non-Aboriginal women (Supplementary Table 1).

Figure 3.

Crude incidence rate of pelvic inflammatory disease by chlamydia and gonorrhea testing and positivity, stratified by Aboriginality.

Additional analyses stratified by attained age, socioeconomic group, health area, and time period during which tests were conducted were also generally consistent (Supplementary Table 1).

DISCUSSION

This large population-based study found that, compared with women who had always tested negative, the risk of emergency department presentation or hospital admission for PID was >4 times higher in women who had tested positive for gonorrhea (either alone or with a coexisting chlamydia diagnosis) and 1.8 times higher in women positive only for chlamydia. Repeated infections appeared to further increase the risk of PID, as did a history of repeat but negative tests. The lowest rate of PID was in women who had no test record for chlamydia or gonorrhea.

Several European [5, 7, 28] and 1 Canadian study [6] have used similar methods to estimate the association between chlamydia infection and hospital presentations for PID with relative risk estimates of between 1.3 and 1.7. Our study of Australian women estimated the risk of PID to be around 1.8 times higher comparing chlamydia-only positive women to women negative for both chlamydia and gonorrhea, a comparable yet slightly higher estimate. These earlier studies found repeat chlamydia diagnoses to be associated with an increased risk of PID compared to only 1 positive diagnosis [5, 6, 29].

Our study adds to these earlier findings that only investigated chlamydia infections. We were able to examine the association of gonorrhea with PID and compare the risks to those of chlamydia. We demonstrated a substantially greater risk of PID with gonorrhea. However, the effect was not additive, and women positive for both chlamydia and gonorrhea had a similar-sized risk to those positive only for gonorrhea. Also, our results suggest that gonorrhea-related PID was more clinically severe than that from chlamydia, with a higher proportion of hospitalizations relative to emergency department presentations. To our knowledge, this is the first and largest cohort analysis using both testing and positivity data to quantify and compare the risks of gonorrhea and chlamydia on incidence of PID hospitalization and emergency department presentations. Two previous Australian studies, one using positive STI notifications [20], and a clinic-based study [30], have reported a higher risk of PID with gonorrhea infection. Earlier studies have also suggested that gonococcal PID is more likely to be symptomatic than PID arising from a chlamydia infection [10]. However, it is worth nothing that chlamydia infections are much more common than gonorrhea [14], and therefore the population attributable risk of PID for gonorrhea may still be lower than that for chlamydia.

An increase in the risk of PID was also observed in women with increasing numbers of negative chlamydia and gonorrhea tests, a finding that, to our knowledge, has not previously been reported. STI screening guidelines recommend regular testing of young sexually active women at higher risk of STIs. Recommendations include those aged <30 years and Aboriginal populations in Australia and those aged <25 years in the United States [25, 31]. We have previously demonstrated in this cohort significantly higher testing rates among both younger women and Aboriginal women [32]. As by definition, this population recommended for testing would also be at higher risk of other infections such as Mycoplasma genitalium and trichomoniasis, the higher PID risk observed associated with greater testing may reflect PID caused by a different etiological agent [33]. Therefore, our observation of increased risk with increased negative tests may simply reflect this practice of testing higher-risk women. This is supported by the finding that women with no test records for chlamydia or gonorrhea had the lowest risk of PID and underlies the importance of considering STI testing history in epidemiological studies quantifying risks associated with chlamydia and gonorrhea.

In Australia, gonorrhea disproportionately affects Aboriginal people [16]. We estimated the absolute risk of PID hospitalizations and emergency department presentations in Aboriginal women to be almost double that of non-Aboriginal women; however, the relative risks of PID associated with chlamydia and gonorrhea positivity were similar between Aboriginal and non-Aboriginal women. Possible reasons for the higher rate of PID hospitalizations and emergency department presentations in Aboriginal women could be attributed to poor diagnosis and inadequate treatment of PID in some primary care settings [34] or a higher prevalence of other infections or risk factors associated with PID such as Mycoplasma genitalium [17] or smoking [21] for which we could not account; it may also be that non-Aboriginal women are more likely to present with PID to other healthcare settings.

PID can be difficult to diagnose as symptoms are often subtle and nonspecific [35]. Knowledge of a recent STI diagnosis may lead to detection bias for PID, and this could result in an overestimate of effect sizes. However random misclassification of PID hospitalizations [36] could lead to underestimates. Further, mild to moderate PID is usually managed in the primary care setting [37] and therefore only more serious clinical cases, those with undiagnosed or inadequately treated PID [34], would likely present at the emergency department, with only the most severe hospitalized. Our analyses lacked primary care data; however, when we stratified analyses by factors affecting primary care access such as Aboriginality, socioeconomic status, and region of residence, our results were consistent. Also, some women in our cohort may have moved out of state; however, we know from census data that this was relatively small in WA [38]. Finally, our linked laboratory data did not capture all tests conducted in the state with previous analysis, suggesting coverage of approximately 50% of all chlamydia and 80% of all gonorrhea tests conducted in WA [32]. However, our main comparisons are between women who tested positive for chlamydia or gonorrhea and those who tested negative.

In Australia, gonorrhea antimicrobial resistance has been shown to vary substantially between regions. While national surveillance data show that overall, from 2011 to 2015, gonorrhea antimicrobial resistance (measured as a decreased susceptibility to ceftriaxone (minimum inhibitory concentration = 0.06–0.125 mg/L), has fluctuated between 1.8% and 8.8% [39], in WA over the same period, it has ranged from 0.7% to 3.0%. Furthermore, in remote WA Aboriginal communities, where gonorrhea is endemic, penicillins continue to remain the first-line treatment, as <5% of gonorrhea isolates have been found to be penicillinase producing [40]. Despite this, as gonorrhea infections in Australia are dominated by relatively few strains, there is a significant potential for incursion of antibiotic-resistant gonorrhea into these communities [19]. While chlamydia infections are much more common than gonorrhea [14], our findings suggest that gonorrhea infections confer a substantially higher risk of hospitalization or emergency department presentation for PID. This, combined with concerns around increasing antimicrobial resistance and the potential for rapid spread, make it imperative that individuals in areas with high gonorrhea prevalence are tested, diagnosed, and treated in a timely and effective manner in an effort to reduce both the risk of onward transmission and the risk of PID and its serious health sequelae.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank the staff at the Western Australian Data Linkage Branch, WA Birth Registrations, WA Electoral Roll, WA Department of Health, and the 2 pathology laboratories who provided data for this project.

Chlamydia and Reproductive Health Outcome Investigators. B. Liu, D. Preen, J. Hocking, B. Donovan, C. Roberts, J. Ward, D. Mak, R. Guy, J. Kaldor, S. Pearson, L. Stewart, H. Wand, J. Reekie.

Financial support. This work was supported by the National Health and Medical Research Council (NHMRC) (grant number APP1020628). B. L., J. S. H., B. D., C. R., J. W., R. G., and J. M. K. are supported by NHMRC fellowships.

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.