Article Text


Original article
Rates of pelvic inflammatory disease and ectopic pregnancy in Australia, 2009–2014: ecological analysis of hospital data
  1. Jane L Goller1,
  2. Alysha M De Livera1,
  3. Rebecca, J Guy2,
  4. Nicola Low3,
  5. Basil Donovan2,
  6. Matthew Law2,
  7. John M Kaldor2,
  8. Christopher K Fairley4,
  9. Jane S Hocking1
  1. 1 Centre for Epidemiology and Biostatistics, Melbourne School of Population & Global Health, University of Melbourne, Parkville, Victoria, Australia
  2. 2 The Kirby Institute, University of New South Wales, Sydney, New South Wales, Australia
  3. 3 Institute of Social and Preventive Medicine, University of Bern, Bern, Switzerland
  4. 4 Central Clinical School, Monash University and Melbourne Sexual Health Centre, Carlton, Victoria, Australia
  1. Correspondence to Jane L Goller, Centre for Epidemiology and Biostatistics, Melbourne School of Population & Global Health, University of Melbourne, Parkville, Victoria 3010, Australia; jane.goller{at}


Objective To analyse yearly rates of pelvic inflammatory disease (PID) and ectopic pregnancy (EP) diagnosed in hospital settings in Australia from 2009 to 2014.

Methods We calculated yearly PID and EP diagnosis rates in three states (Victoria, New South Wales, Queensland) for women aged 15–44 years using hospital admissions and emergency department (ED) attendance data, with population and live birth denominators. We stratified PID diagnoses as chlamydial-related or gonorrhoeal-related (Chlamydia trachomatis (CT)-related or Neisseria gonorrhoeae (NG)-related), acute, unspecified and chronic, and analysed variations by year, age and residential area using Poisson regression models.

Results For PID, the rate of all admissions in 2014 was 63.3 per 100 000 women (95% CI 60.8 to 65.9) and of all presentations in EDs was 97.0 per 100 000 women (95% CI 93.9 to 100.2). Comparing 2014 with 2009, the rate of all PID admissions did not change, but the rate of all presentations in EDs increased (adjusted incidence rate ratio (aIRR) 1.34, 95% CI 1.24 to 1.45), and for admissions by PID category was higher for CT-related or NG-related PID (aIRR 1.73, 95% CI 1.31 to 2.28) and unspecified PID (aIRR 1.09, 95% CI 1.00 to 1.19), and lower for chronic PID (aIRR 0.84, 95% CI 0.74 to 0.95). For EP, in 2014 the rate of all admissions was 17.4 (95% CI 16.9 to 17.9) per 1000 live births and of all ED presentations was 15.6 (95% CI 15.1 to 16.1). Comparing 2014 with 2009, the rates of all EP admissions (aIRR 1.06, 95% CI 1.04 to 1.08) and rates in EDs (aIRR 1.24, 95% CI 1.18 to 1.31) were higher.

Conclusions PID and EP remain important causes of hospital admissions for female STI-associated complications. Hospital EDs care for more PID cases than inpatient departments, particularly for young women. Updated primary care data are needed to better understand PID epidemiology and healthcare usage.

  • women
  • pelvic inflammatory disease
  • chlamydia infection
  • gonorrhoea

Statistics from


Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) can ascend to the upper genital tract and have serious health consequences for women, including pelvic inflammatory disease (PID), which can lead to ectopic pregnancy (EP), chronic pelvic pain or infertility,1 2 which may be unrecognised until affected women try to conceive. These sequelae account for substantial healthcare costs,3 and their prevention is an important reason for STI control policies.4

In many countries including Australia, PID and EP are not statutorily notifiable, but data sets about diagnoses in hospitals provide information about rates over time. The gonorrhoea epidemics in many industrialised countries during the 1960s and 1970s were associated with increasing PID incidence followed by increasing EP incidence.5 From the 1980s to 2010, declining PID rates in hospital admissions and general practice were reported in several countries including Australia,5–14 with some reports suggesting declines were influenced by STI control.12 However, in the 1980s and 1990s, falls in PID and STIs were also attributed to sexual behaviour changes in response to the HIV epidemic.6 Stable or declining EP rates have been reported during the 1990s and 2000s in some countries,7–9 but increasing rates have also been reported.10

In 2007 in Australia, the hospital admission rate for PID was 89 per 100 000 population and EP was 16 per 1000 live births among women aged 15–39 years.8 Since then, chlamydia and gonorrhoea diagnosis patterns have changed. Among women, age-standardised chlamydia diagnosis rates increased from 2007 to 2011, were stable to 2015, then increased in 2016.15 Although gonorrhoea occurs predominantly among men who have sex with men, notification rates in women more than doubled from 2007 to 2016, raising concerns about potential reproductive tract complications.15 The primary objective of this study was to analyse yearly rates of PID and EP diagnosed in Australian hospital settings. The secondary objective was to examine associations between PID or EP diagnosis and characteristics of residential area.


We undertook an ecological study using data on the number of hospital admissions and emergency department (ED) attendances for PID and EP in the three most populous Australian states: New South Wales, Victoria and Queensland.

We obtained data from state Departments of Health from separate hospital admissions and ED attendances registers (online supplementary table 1). We received non-identifiable, line-listed records for female patients aged 15–44 years old during 2009–2014. Hospital admission data sets included data from all public (government-funded) and private hospitals. ED data sets included data for presentations to public hospitals with a designated ED. Data reporting from EDs was voluntary and clinicians assigned diagnosis codes, making ED data more variable than admissions data. ED attendances can result in discharge or hospital admission, the latter also counted in admissions data sets, but we could not merge these data sets owing to de-identified records and different data systems.

Supplementary file 1

Each patient record included a principal diagnosis for the main reason for care and ‘other’ diagnoses made, each coded using the International Classification of Diseases 10th Revision, Australian Modification (ICD)-10-AM, or for some EDs the ICD-9 or Systematized Nomenclature of Medicine. Data items included year, age group, residential postcode and principal diagnosis code on which a PID or EP diagnosis was assigned (table 1, online supplementary table 2). Like other Australian studies,7 8 we excluded records with an ‘other’ PID or EP diagnosis because they might represent pre-existing conditions. We categorised PID admissions further as chlamydial-related or gonorrhoeal-related (CT-related or NG-related PID), acute PID, PID unspecified or chronic PID. As this analysis focused on trends, we only included ED records from hospitals contributing data in all years and if annual presentation numbers varied by <50%.

Table 1

Number of hospital admissions and ED presentations for PID and EP, and age breakdown of the population and live birth denominators, by year, 2009–2014

Denominator data were obtained from the Australian Bureau of Statistics (online supplementary table 1). At postcode level these included estimated female residential population by year, age, remoteness and Index of Relative Socioeconomic Disadvantage (IRSD).16 Like another Australian study,8 we obtained the number of live births by maternal age and year for our EP denominator. The three states comprised 1738 postcodes. Postcodes were excluded (n=18) if the population was zero (eg, company postcodes) or offshore island/s; or recoded to a neighbouring postcode (n=42) if IRSD was unavailable or the population for some age groups was zero (eg, remote postcodes). The remaining 1678 postcodes were categorised for remoteness (metropolitan, inner regional, outer regional and/or remote using standard definitions) and deciles of increasing socioeconomic disadvantage based on the IRSD.

We prepared three data sets. The all-admissions and all-ED data sets included all hospital admissions or ED presentations with a principal PID or EP diagnosis, and population by postcode, year and age group. The non-admitted-ED data set was a subset comprising PID or EP episodes discharged from EDs without admission.

Statistical analysis

We analysed data sets separately using Stata V.14. We calculated yearly PID and EP rates per 100 000 women using population denominators and EP rates per 1000 live births. We examined variation in rates by year, age group, remoteness and socioeconomic disadvantage of postcode using univariable and multivariable Poisson regression models with clustered sandwich estimator to account for intragroup correlation.17 We included year as a categorical variable to see whether rates differed from the reference category of 2009. We report incidence rate ratios with 95% CI. Where necessary, we used zero-inflated Poisson (ZIP) regression to account for large numbers of postcodes with no cases, and compared the fit with ordinary Poisson models using the Vuong test.18 Using likelihood ratio tests, we investigated interactions between residential area and age group and reported them if statistically and clinically meaningful. We conducted a subgroup analysis of admission rates by PID category. Two sensitivity analyses were undertaken to examine the robustness of our results. The first used linear splines to explore the rate of change in overall rates per 2-year period. The second repeated our analysis of population rates, omitting postcodes recoded to neighbouring postcodes.


From 2009 to 2014 we recorded a total of 14 271 admissions and 20 522 ED presentations with a principal PID diagnosis, and 23 579 admissions and 19 382 ED presentations with a principal EP diagnosis, across 1678 postcodes (table 1, online supplementary table 2). The median population of women aged 15–44 years old in study postcodes in 2009 was 946 (IQR 212–3254). Two-fifths (42%) of postcodes were metropolitan (representing 76% of the population), 29% in outer regional/remote areas (7% of the population) and 41% of the population lived in more disadvantaged (five most disadvantaged IRSD deciles) areas.

Pelvic inflammatory disease

Two-thirds (65.8%) of PID hospital admissions were unspecified PID, and the remainder were chronic PID (22.6%), acute PID (6.3%), CT-related PID (5.3%) and NG-related PID (0.1%). Most (93.7%) PID in EDs were unspecified PID, and the remainder were acute PID (6.0%), chronic PID (0.1%) and CT-related PID (0.2%). One-third (32%) of PID in EDs resulted in hospital admission (table 1).

Figure 1A and online supplementary table 3 show the annual PID rates per 100 000 women. The overall PID admission rate increased from 60.6 in 2009 to 69.5 in 2011, then decreased to 63.3 in 2014. Between 2009 and 2014, the overall PID rate in EDs increased from 72.6 to 97.0. In multivariable analysis (table 2) comparing 2014 with 2009, the rate of all PID admissions did not change (aIRR 1.05, 95% CI 0.98 to 1.12), but within PID categories (table 3) were higher for CT-related or NG-related PID (aIRR 1.73, 95% CI 1.31 to 2.28) and unspecified PID (aIRR 1.09, 95% CI 1.00 to 1.19), similar for acute PID and lower for chronic PID (aIRR 0.84, 95% CI 0.74 to 0.95). PID admission rates were higher for women aged 15–24 than those aged 35–44 years (aIRR 1.09, 95% CI 1.04 to 1.14), including CT-related or NG-related PID (aIRR 11.68, 95% CI 8.60 to 15.85) and unspecified PID. Chronic PID admission rates were highest for women aged 35–44 years. In EDs, overall PID rates were higher in 2014 than in 2009 (aIRR 1.34, 95% CI 1.24 to 1.45), including PID managed without admission, and were more than twice as high for women aged 15–34 than those aged 35–44 years. Higher PID rates were observed in more disadvantaged and non-metropolitan (regional or remote) than metropolitan areas.

Figure 1

Overall and age-specific admission, emergency department (ED) presentation and non-admitted ED presentation rates of pelvic inflammatory disease and ectopic pregnancy, 2009–2014. (A) Pelvic inflammatory disease, (B) ectopic pregnancy presentation rates per 100 000 women and (C) ectopic pregnancy rates per 1000 live births.

Table 2

Factors associated with population rates of (A) PID, (B) ectopic pregnancy and (C) ectopic pregnancy rates per 1000 live births, 2009–2014

Table 3

Factors associated with PID admissions, by PID category, 2009–2014

Ectopic pregnancy

The most frequent EP diagnoses codes were O00.1 tubal pregnancy (70% of admissions) and O00.9 EP unspecified (83% of ED), and 78% of EPs in EDs resulted in hospital admission. Yearly population rates of EP are shown in figure 1B and online supplementary table 3 and EP rates among live births in figure 1C.

Population rates

The overall EP hospital admission rate per 100 000 women in 2014 was 107.8 and did not differ from 2009 (table 2). Between 2009 and 2014, EP rates in EDs increased overall (84.1–96.7) and for women discharged without admission (18.4–22.5).

In multivariable analysis (table 2), comparing 2014 with 2009, EP rates in hospital admissions did not change but were higher in EDs (aIRR 1.12, 95% CI 1.05 to 1.20). Admission and ED rates were highest for women aged 25–34 compared with those aged 35–44 years and in more disadvantaged and non-metropolitan areas. The rates of EP discharged from EDs without admission were higher in 2014 than in 2009 (aIRR 1.17, 95% CI 1.00 to 1.37) and lowest in the outer regional/remote areas (aIRR 0.69, 95% CI 0.49 to 0.96).

Live birth rates

Overall EP rates per 1000 live births in 2014 were 17.4 in admissions, 15.6 in EDs and 3.6 for EDs but not admitted. In multivariable analysis, EP admission rates (table 2) were higher in 2014 than in 2009 (aIRR 1.06, 95% CI 1.04 to 1.08) and highest for women aged 35–44 years. In EDs, EP rates were highest for women aged 15–24 years and in 2014 compared with 2009 (aIRR 1.24, 95% CI 1.18 to 1.3).

Sensitivity analyses

Linear splines showed the rate of change for population rates did not alter during the study, and the rate of change for ED-EP live birth rates during 2011–2012 was higher than for 2009–2010 (online supplementary table 4). Omission of postcodes recoded to neighbouring postcodes showed negligible change to results (data available on request).


This ecological study found that for women of reproductive age, overall PID admission rates were similar between 2009 and 2014. Within PID categories, admission rates increased for CT-related or NG-related PID and unspecified PID, but declined for chronic PID. PID rates in EDs increased and were 2.7 times higher among women aged 15–24 than those aged 35–44 years. Age variability in overall PID admission rates was less pronounced. EP rates among live births were higher in 2014 compared with 2009 in admissions and EDs.

Strengths and weaknesses

Our study had two main strengths. First, inclusion of ED data provided new information about PID and EP diagnoses in Australia, and like other studies7 8 included public and private hospital admissions for a complete picture of PID and EP admissions. PID rates for women admitted from EDs showed a similar pattern between the admissions and ED data sets. Second, undertaking our analysis at the postcode level allowed exploration of the relationship between area characteristics and population rates. Large numbers of postcodes with no diagnoses were accommodated by the multivariable ZIP analysis for which our findings were consistent with ordinary Poisson models.

An important study limitation is that our PID rates included only women managed in hospitals. Australian guidelines recommend inpatient management for severe PID and outpatient management for mild-moderate PID.19 While admissions data can tell us about severe PID, most mild-moderate PID are managed in primary care,20 general practice being Australia’s mainstream primary care setting. Primary care data are needed for a more complete picture of PID, but are not routinely available. Second, because ED data provision is voluntary, we consider our admission data more reliable than ED data. We minimised variability in ED rates by limiting our analysis to EDs contributing data for all study years and with high completeness. Third, clinical PID diagnosis has low sensitivity and specificity compared with laparoscopic visualisation.2 Absolute diagnosis rates might be inaccurate, but if diagnostic practices were unchanged these trends should be reliable. Uterine, cervical motion and adnexal tenderness in sexually active women with pelvic pain are the mainstay of PID diagnosis,2 and until non-invasive biomarkers for upper genital tract inflammation are widely available2 large-scale diagnostic changes that affect estimated rates are unanticipated. However, current policies promote opportunistic chlamydia testing4 and could contribute to identifying more STI-associated PID cases. Fourth, our birth denominator did not include all conceptions and EP rates could be influenced by other pregnancy outcomes (eg, stillbirths, abortion) over time. We could not address this issue because data about all conceptions are not routinely available, but live birth denominators have been accepted previously.8 21 Finally, being an ecological study, we cannot make causal inferences about factors that might influence rates over time. We show yearly age-adjusted rates, and our area measures allowed comparisons between more or less affluent or urban and non-urban areas.

Comparison with other studies

We found admission rates in 2009 per 100 000 of 61 for PID and 110 for EP among women of reproductive age. An earlier Australian study (2001–2008) reported annual infertility admission rates for same-aged women of around 400 per 100 000.22 Our overall PID admission rates were similar between 2009 and 2014, which is broadly consistent with a commissioned review presenting hospital discharge rates for inflammatory diseases of female pelvic organs (including any-cause PID) during 1990–2014 across Europe, America and Australia, showing declining country-specific rates to around 2007, which then appeared to plateau in several countries including Australia.23 This is the first Australian study to assess PID rates using routinely collected ED data. Our findings of increasing PID rates contrast with a study in the USA that found falling PID rates (2002–2009) among adolescents attending EDs.14 For EP, stable or declining admission rates using live birth denominators have been reported in Australia and internationally until the 2000s, with increases in some groups.7 8 21 We found EP admission rates among live births in 2014 were 8% higher than for 2009. We are unaware of other studies measuring EP trends in EDs.

Interpretation of the findings

This study provides some evidence that declines in PID observed in Australia7 8 and elsewhere8 might have ceased or even reversed. Several factors might contribute to this. First, STI epidemiology and sexual behaviour might be changing, and PID is most common among young sexually active women.2 Population-based data show increasing numbers of lifetime sex partners for young Australians, potentially increasing STI risk,24 and surveillance data show increasing chlamydia and gonorrhoea rates among women that appear to reflect increased testing and transmission.15 Other Australian data show higher risks for PID hospitalisation following gonorrhoea or chlamydia compared with no infection.25 Although we found increasing CT-related PID rates and to a lesser extent NG, this might reflect increased testing or that clinicians are more likely to diagnose PID for women with lower abdominal symptoms and a positive test. Second, increased screening and treatment of diagnosed infections renders more women susceptible to reinfection. Chlamydia reinfection substantially increases PID risk1 2 23 and is common; a repeat chlamydia diagnosis rate of 22% in the year after treatment has been reported in Australia.26 Third, most PID cases were unspecified so other causes should be considered. Mycoplasma genitalium has been detected in 2.4% of Australian women attending primary care27 and is receiving attention worldwide as a PID pathogen with worrying levels of antimicrobial resistance.2 23 Bacterial vaginosis (BV) has been diagnosed in up to 12% of Australian women, and BV-associated microbes are often found in the upper genital tract of women with PID.2 23 PID can also develop after uterine instrumentation, although this risk is greatest if an STI is present.28 The extent these factors contribute to our findings is unknown and further research about PID causes is needed.

Australian data for 1998–2003 have shown around 59 000 PID general practice encounters annually.20 Factors restricting primary care access could also contribute to increased ED rates. Timely access to Australian general practice is a concern particularly in non-metropolitan areas where there are ongoing workforce shortages,29 which could contribute to higher STI rates,15 thereby increasing PID risk. Further, out-of-pocket expenses30 in primary care might prompt women to attend EDs instead for mild-moderate PID not requiring hospital admission. During the study period, the average out-of-pocket costs increased by 41%.31

Increasing EP rates could reflect increased risk of extrauterine conception or increased detection of extrauterine pregnancy. Risk factors for extrauterine conception include smoking, postinfection tubal damage (particularly chlamydia), assisted reproductive technologies and older maternal age.32 The extent these risks impact on EP rates is unknown, although smoking rates in pregnancy have declined and maternal age has increased in Australia, where EP-related mortality is rare.33 34 Factors that increase EP diagnosis include more sensitive β-HCG tests to detect EP that might previously have resulted in undiagnosed tubal abortion, high-resolution transvaginal ultrasound, early pregnancy units and close monitoring of assisted reproduction outcomes. Our finding that a fifth of EPs in EDs were managed without admission is consistent with increased use of non-surgical (methotrexate) or conservative (wait and see) management34 in some areas.

Implications for research, practice and policy

Prevention of PID and its associated complications is a key goal of STI control, yet trends in these conditions are generally not routinely monitored. The challenges in measuring PID and EP rates in this and other studies highlight the need for improved data sources and surveillance systems (reflecting hospital and primary care) that facilitate comparable measures over time. Australian policy identifies the need for interventions in primary care to enhance STI management, particularly partner notification and retesting.4 23 Further analyses of hospital and primary care data can support evaluation of enhanced STI management impacts. Research is also needed to better understand the role of other infections in PID and to develop non-invasive and objective methods that can improve PID diagnosis in any setting.

In conclusion, we found increasing rates of PID and EP diagnosis in ED and EP hospital admissions. These results could represent changing sexual practices, increasing STI transmissions and reinfections, changing healthcare usage or increased EP detection from improved diagnosis. Without primary care data knowledge of PID epidemiology and healthcare use in Australia is incomplete. PID and EP remain important causes of hospital admission for STI-associated complications. EDs provide care for many additional PID cases, particularly for young women, warranting a strengthened focus on understanding the drivers of these rates and on reducing the risks of these sequelae.

Key messages

  • Pelvic inflammatory disease (PID) and ectopic pregnancy remain important causes of hospital admissions for STI-associated complications among women of reproductive age.

  • Emergency departments (EDs) care for many more PID cases, particularly for young women.

  • PID rates in EDs were substantially higher for younger than older women.

  • PID hospital admission rates varied little by age.

  • Updated primary care data are needed to better understand PID epidemiology and healthcare usage, particularly given different patterns between hospital admissions and ED attendances.


Australian Chlamydia Control Effectiveness Pilot (ACCEPt) investigators and project team;

Victorian Government, Department of Health and Human Services; NSW Government, NSW Ministry of Health;

Queensland Government, Department of Health; Government of South Australia, SA Health.


View Abstract


  • Handling editor Gwenda Hughes

  • Contributors JLG collected, cleaned and analysed the data, contributed to the analysis plan, and drafted and revised the manuscript. AMDL provided statistical advice, supervised the analysis, and contributed to the manuscript draft and revisions. JSH, RG, NL, CKF, BD, ML and JMK designed the Australian Chlamydia Control Effectiveness Pilot study, contributed to the analysis plan, interpretation of results, and contributed to the manuscript draft and revisions. All authors approved the final submitted version of the manuscript.

  • Funding These data are being analysed as part of the Australian Chlamydia Control Effectiveness Pilot (ACCEPt) study funded by the Australian Government Department of Health and the National Health and Medical Research Council. JLG is supported by an Australian Government Research Training Program Scholarship at the University of Melbourne.

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval The study was approved by the Royal Australian College of General Practitioners National Research and Evaluation Ethics Committee (NREEC09.019).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Correction notice This paper has been amended since it was published Online First. The authors have noticed that in table 3 one line of results is missing for the year 2013. The table has since been updated to include this data.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.