Article Text

Download PDFPDF

Original article
A cross-sectional study showing differences in the clinical diagnosis of pelvic inflammatory disease according to the experience of clinicians: implications for training and audit
  1. Georgina C Morris1,2,
  2. Catherine M W Stewart1,
  3. Sarah A Schoeman1,
  4. Janet D Wilson1
  1. 1Department of Genitourinary Medicine, Leeds Teaching Hospitals NHS Trust, Leeds, UK
  2. 2Centre for Immunology and Infection, Hull-York Medical School, University of York, York, UK
  1. Correspondence to Dr Janet Wilson, Consultant in Genitourinary Medicine, Department of Genitourinary Medicine, Leeds General Infirmary, Leeds LS1 3EX, UK; janet_d.wilson{at}leedsth.nhs.uk

Abstract

Objectives Pelvic inflammatory disease (PID) generates diagnostic difficulty even for experienced doctors. Junior doctors and nurses also assess women with symptoms suggestive of PID. We aimed to determine if and how PID diagnoses vary between clinicians with different experience levels.

Methods Cross-sectional study conducted in UK sexual health clinic, nested within a Chlamydia trachomatis (CT), and Neisseria gonorrhoea diagnostic test accuracy study. Proportions and characteristics of women diagnosed clinically with PID by clinicians with varying experience were compared. Outcomes included demographics, presenting symptoms and signs and CT, and CT and/or gonococcal (GC) (CT/GC) positivity.

Results In 3804 women assessed by 36 clinicians, rates of PID, CT and GC were 4.4%, 10.5%, and 2.5%, with no differences between experienced and inexperienced clinicians (p=0.84, p=0.13 and p=0.07, respectively). 63.7% of PID diagnosed by experienced clinicians met Centers for Disease Control and Prevention (CDC) key clinical criteria versus 41.2% by inexperienced; experienced versus inexperienced OR 2.51; 95% CI 1.16 to 5.40). Proportions of CT (CT/GC)-positive PID increased with experience (5.9% (11.8%) to 31.9% (34.1%)); experienced versus inexperienced (OR 3.90; 95% CI 1.12 to 13.5). Percentages of women with CT (CT/GC) who were diagnosed with PID also rose with experience (2.2% (3.9%) to 14.2% (13.7%)), but CT prevalence in PID cases diagnosed by inexperienced clinicians (8.8%) was no greater than in all women they assessed (9.0%), suggesting poorer discriminative skills.

Conclusions Clinical diagnostic acumen for PID improves with experience. Inexperienced clinicians should focus on the presence of lower abdominal pain with pelvic tenderness and consider additional supportive symptoms, to improve specificity of their diagnoses.

Trial registration number: ISRCTN 42867448.

  • PELVIC INFLAMMATORY DISEASE
  • DIAGNOSIS
  • EDUCATION
View Full Text

Statistics from Altmetric.com

Introduction

Pelvic inflammatory disease (PID) is a spectrum of inflammatory disorders of the upper female genital tract, arising mainly in sexually active young women from ascending infections.1 About 25% of UK cases are attributed to Chlamydia trachomatis (CT) and/or Neisseria gonorrhoeae (NG).2 ,3 Other causes include Mycoplasma genitalium4 and vaginal organisms (anaerobes, Gardnerella vaginalis, streptococci).5 Complications include infertility, ectopic pregnancy and chronic pelvic pain.6

Clinical features include recent onset lower abdominal pain, deep dyspareunia, abnormal vaginal bleeding, mucopurulent discharge and bimanual examination tenderness (cervical motion, uterine and/or adnexal). Symptoms and signs can be mild or absent making PID difficult to diagnose accurately.2 Distinction must be made from other conditions, for example, ectopic pregnancy, appendicitis, and irritable bowel syndrome. Compared with laparoscopy, clinical diagnosis of salpingitis has a positive predictive value (PPV) of only 65%.7 ,8 The presence of bacterial vaginosis (BV), and inflammatory cells on microscopy are additionally supportive but lack specificity.79

The syndromic nature of PID, and lack of definitive criteria, generates the potential for diagnostic variation between clinicians. An Australian sexual health clinic found significant differences in proportions of women diagnosed with PID between senior doctors concluding that even experienced clinicians were missing cases of PID.10 Their study did not include junior trainee doctors, but they assess significant numbers of symptomatic women in settings such as A&E, gynaecology, general practice and sexual health. Also, sexual health nurses are increasingly managing symptomatic as well as asymptomatic patients. If diagnosing PID is difficult for experienced clinicians it may be even harder for those with less experience. National guidelines recommend a low threshold for diagnosis with empirical antibiotics at first presentation, as delaying treatment can increase infertility.2 ,11 ,12 Thus, less experienced clinicians may compensate by having a higher rate of PID diagnosis than those with more experience.

Our aim was to compare the rates of PID diagnoses, demographic factors, presenting symptoms and signs, and proportions with CT and CT and/or gonococcal (GC) (CT/GC) positivity. in women diagnosed with PID, between clinicians according to their levels of clinical experience.

Methods

We conducted a cross-sectional study to investigate the proportions and characteristics of women diagnosed with PID by clinicians with varying levels of experience. Data were obtained from records of a CT and NG diagnostic test accuracy study conducted at Leeds Centre for Sexual Health (LCSH) between March 2009 and January 2010, in which this study was nested. The full methods are described elsewhere.13 ,14 The range of clinicians involved included Band 5 nurses (B5) who are junior grade nurses and Band 6 nurses (B6) who are clinical nurse specialists, Foundation Year and Core Medical Trainee doctors who are postgraduate interns (FY/CMT), General Practitioners (GP), Specialty doctors (SAS), Specialty Trainee Registrars years 3–6 (STR) who are resident doctors and consultants who are attending doctors. For the analyses, experienced clinicians were defined as having at least 1 year of full-time equivalent (FTE) experience of assessing symptomatic women in sexual health clinics at the start of the study. This included all GPs, SAS doctors, STRs and consultants. B6 nurses had been assessing symptomatic women for less than 12 months, and FY/CMT doctors worked for a maximum of 4 months FTE in the clinic, so were classed as inexperienced clinicians. At registration, all women completed triage forms allowing them to be categorised as asymptomatic or symptomatic. All grades except B5 nurses assessed symptomatic women. They were seen in the order of registration by the next available clinician.

Women aged 16 years and over requesting testing for sexually transmitted infections (STI), who gave consent, were recruited. Data were extracted directly from clinical case notes of study participants and entered contemporaneously onto case report forms and the study database. Data entry was crosschecked for errors. Details of age, history of STIs, being in contact with an STI, symptoms of vaginal discharge, abnormal bleeding, deep dyspareunia, lower abdominal pain, and bimanual pelvic examination tenderness were collected. Anything not explicitly documented in the history or examination within the case notes was considered not to be present. Clinical diagnosis and treatment at presentation were recorded.

Women had self-taken vulvovaginal swabs (VVS) for NG/CT NAAT analysed by Aptima Combo-2 (AC2), clinician-taken urethral and endocervical samples for NG culture, and endocervical samples for NG/CT NAAT. Women with symptoms had microscopy for NG, Trichomonas vaginalis, BV and yeasts. Bimanual pelvic examination was performed in symptomatic women at the discretion of the assessing clinician.

LCSH guidelines for diagnosis and management of PID recommend that clinicians have a low threshold for diagnosing and empirically treating PID at first presentation in women with (1) lower abdominal pain (particularly with symptoms/signs of mucopurulent discharge, deep dyspareunia, abnormal bleeding), (2) cervical motion, uterine or adnexal tenderness on bimanual examination and (3) a negative pregnancy test.15 Table 1 shows how LCSH guidelines distinguish between key and additional supportive criteria in diagnosing PID, akin to Centers for Disease Control and Prevention (CDC) guidelines.11 ,16 This distinction was not explicit in the contemporaneous British Association for Sexual Health and HIV (BASHH) guidelines.17 Women diagnosed with, and treated for, PID at first presentation, as recorded in clinical case notes, were defined as PID cases. They were defined as CT infected if the endocervical swab or VVS was positive by AC2 and confirmed by Aptima CT,14 and as NG infected if positive on culture with biochemical confirmation for N. gonorrhoeae, or on endocervical or VVS AC2 confirmed by Aptima GC.13

Table 1

Diagnostic criteria for PID from National Guidelines and Leeds Centre for Sexual Health

We calculated rates of PID diagnoses, demographic factors and presenting symptoms and signs in women assessed by different grades of clinician. Symptoms and signs were analysed for concordance with LCSH PID diagnostic criteria. We compared the overall rates of CT and CT/GC, proportions of PID cases that were CT, and CT/GC positive, and proportions of women with CT and CT/GC infections diagnosed with PID. Comparisons were made between all experienced and all inexperienced clinicians, and between consultants and inexperienced clinicians, that is, the most versus the least experienced, assuming consultants to be the ‘gold standard’. ORs and 95% CIs were calculated using Pearson χ2 or 2-tailed Fisher's exact (FE) tests as appropriate, using SPSS Statistics V.19.0 (IBM).

Results

Summary data of whole cross-sectional study

Three thousand nine hundred and seventy-three women were included in the CT/GC diagnostic study; 169 were diagnosed with PID. B5 nurses, who managed 169 asymptomatic women, diagnosed no cases of PID. These clinicians and their patients were therefore excluded from the cross-sectional study leaving 3804 women assessed by 36 clinicians (7 B6 nurses, 10 FY/CMT doctors, 2 GPs, 2 SAS doctors, 9 STRs, 6 Consultants). The overall rate of PID was 4.4%. The demographics of women with PID are summarised in table 2. There were no differences in age, rates of previous STIs, previous PID, or being a contact of CT or GC between women diagnosed by experienced or inexperienced clinicians, but 23.5% of women diagnosed with PID by B6 nurses reported being a CT contact compared with 2.1% of those diagnosed by consultants (OR 14.15 (1.45 to 137.9)).

Table 2

Summary of key demographics of the PID cases (n=169)

Presenting symptoms and signs of PID cases

Table 3 shows the rates of presenting symptoms and signs and concordance with LCSH PID diagnostic criteria. There were no differences in proportions of women in the cross-sectional study who met the key criteria of lower abdominal pain and pelvic tenderness between clinician groups, but the proportion of such women diagnosed with PID increased with experience. The proportion of women with lower abdominal pain, and other cumulative supportive symptoms, diagnosed with PID rose with increasing clinician experience. The presence of three or more symptoms was documented in a lower proportion of all women assessed by consultants than other clinicians, (consultants 6.3%, all experienced 10.2%, inexperienced 12.4%; consultants vs inexperienced OR 0.48 (0.34 to 0.68)), yet consultants diagnosed a greater percentage of these women as having PID (consultants 45.8%, all experienced 18.1%, inexperienced 11.2%; consultants vs inexperienced OR 6.67 (2.78 to 16.31)). Women diagnosed with PID by consultants were 4 times more likely to meet key criteria than those diagnosed by inexperienced clinicians. By contrast, PID diagnoses based on pelvic tenderness in the absence of lower abdominal pain were 5 times more common by inexperienced clinicians compared with consultants (OR 5.08 (1.60 to 16.67)).

Table 3

(a) Relationship between lower abdominal pain and pelvic tenderness (ie, key LCSH PID diagnostic criteria) and PID diagnoses in all women assessed (whole cross-sectional study group) (n=3804); (b) Presenting symptoms and signs in PID cases (n=169) and (c) concordance with LCSH PID diagnostic criteria

Rates of CT and GC infections

Rates of CT and CT/GC infection detected overall and in women diagnosed with PID are displayed in table 4. The overall prevalence of CT was 10.5%, GC 2.5%, CT/GC 11.8%. There were no differences in CT, GC, or CT/GC prevalence between clinician groups. Infection rates of PID cases were CT 23.7%, GC 8.3%, CT/GC 27.8%. Rates of CT (and CT/GC)-positive PID rose with increasing experience: 5.9% (11.8%) B6 nurses; 11.8% (11.8%) FY1/CMT; 26.3% (28.9%) GP/SAS; 24.0% (30.0%) STR; 31.9% (34.1%) Consultants. Experienced clinicians had CT positive PID rates over 3 times those of inexperienced clinicians.

Table 4

Rates of CT and CT and/or GC (CT/GC) infection detected in (a) all women assessed (whole cross-sectional study group) (n=3804) and (b) the PID cases. (n=169) (c) Rates of PID diagnoses among women subsequently testing positive for CT and CT /GC (n=3804)

As a cause of PID, higher rates of CT infection would be expected in PID cases compared with the overall prevalence diagnosed by that clinician group. Consultants (OR 3.64 (1.82 to 7.24)), and all experienced clinicians (OR 3.10 (2.05 to 4.69)), had threefold higher rates of CT in women they diagnosed with PID. However, for all inexperienced clinicians there was no difference (OR 0.98 (0.29 to 3.28)); for FY/CMT, 11.8% of PID cases had CT compared with 8.1% in all the women they saw; for B6 nurses rates were 5.9% and 10.2%, respectively.

The percentages of all women found to have CT (and CT/GC) who had been diagnosed with PID by the different grades of clinician rose with increasing experience: B6 nurses 2.2% (3.9%); FY1/CMT 7.7% (7.1%); GP/SAS 8.9% (9.7%); STR 11.0% (12.4%) consultants 14.2% (13.7%).

Relationship of infections to key PID diagnostic criteria

Table 3 shows that women diagnosed with PID who met the key clinical criteria were 1.5 times as likely to be CT/GC positive than those with lower abdominal pain without pelvic tenderness (OR 1.79 (0.46 to 6.95), and over twice as likely as those with pelvic tenderness without lower abdominal pain (OR 2.85 (1.15 to 7.05). No CT or GC infections were detected in PID cases with neither pain nor tenderness.

Among women diagnosed with PID who reported lower abdominal pain, the likelihood of testing positive for CT (and CT/GC) rose with clinician experience with 34.2% (36.8.%) diagnosed by consultants compared with 29.6% (35.7%) for all experienced clinicians, 25.0% (25.0%) for FY/CMT and 0% (14.0%) for B6 nurses. Of women diagnosed with PID who had pain but no tenderness, those who were CT/GC positive all had at least two additional supportive symptoms. Of PID cases without pain but with tenderness testing CT positive (none were GC positive), 83.3% diagnosed by experienced clinicians had at least two symptoms, including deep dyspareunia, whereas, the single CT positive case diagnosed by an inexperienced clinician was a CT contact complaining only of discharge. Of women with two or more symptoms diagnosed with PID, 31.0% diagnosed by consultants had CT compared with 26.7% for all experienced and 5.0% for all inexperienced clinicians.

Estimates of underdiagnosis and overdiagnosis and treatment of PID

There were no differences in rates of diagnoses of PID between clinician groups, but experienced clinicians had CT positive PID rates 3 times those of inexperienced, suggesting inexperienced clinicians missed some true cases of PID and overdiagnosed some women who did not have PID. Assuming consultants to be ‘gold standard’, they diagnosed PID in 89.7% of women they assessed meeting key criteria compared with 70.0% for inexperienced clinicians, suggesting inexperienced clinicians may be missing 20% of cases of PID. Likewise, consultants diagnosed PID in 14.2% of women seen by them who had CT compared with 4.2% by inexperienced clinicians suggesting they may have missed 70% of CT positive PID cases. Of women subsequently testing CT positive, 100% of those meeting key clinical criteria seen by consultants had been treated for PID compared with 66.7% seen by inexperienced clinicians. For each CT positive PID case, consultants treated an additional 2 women for PID compared with an additional 10 women treated by inexperienced clinicians. This figure rose to an additional 15 women in those with tenderness without lower abdominal pain.

Discussion

This is the first study to compare PID diagnoses between experienced and inexperienced clinicians, and it has demonstrated important differences. Of women diagnosed with PID by inexperienced clinicians, only 41.2% met the LCSH key clinical criteria compared with 63.7% for experienced clinicians and 74.5% for consultants. Also, more women diagnosed with PID by nurses reported being a CT contact. Taken in isolation, these could suggest that inexperienced clinicians had a lower threshold for diagnosing PID. However, the lack of difference in PID diagnostic rates between grades of clinician, together with the pattern of variation in the clinical characteristics and infection rates of PID cases does not support this, being instead more in keeping with improving diagnostic acumen with experience.

The data reveal weaknesses in the diagnostic approach of inexperienced clinicians, with less ability to discern significant symptoms and signs and assimilate these to formulate a diagnosis. For example, the presence of three or more symptoms in the whole study group was inversely associated with clinician experience, whereas the converse was seen in PID cases, suggesting inexperienced clinicians may be uncritical of reported symptoms rather than judging their significance. Inexperienced clinicians were more likely to identify pelvic tenderness without lower abdominal pain implying they may be misinterpreting ‘normal’ discomfort during pelvic examination as tenderness and overdiagnosing PID in some women. Of greater concern was the lower proportion of women who had CT who were diagnosed with PID by inexperienced compared with experienced clinicians suggesting that inexperienced clinicians are missing true cases of CT positive PID.

Based on the proportion of cases with confirmed CT/GC infections, diagnostic specificity for PID increased with experience with the largest difference seen between consultants and B6 nurses. Proportionately more women diagnosed with PID were CT positive than all women assessed; the difference increasing with experience. However, there was no difference for all inexperienced clinicians and for B6 nurses CT prevalence in PID cases was lower than in all women they assessed suggesting poorer discriminative skills.

Our findings are supported by educational theory for skill acquisition. The Dreyfus and Dreyfus Model describes developmental stages from novice, progressing through advanced beginner, competent, proficient, expert, then master.18 Exposure to clinical scenarios enables clinicians to build on their knowledge with learning gained through practical experience, creating a library of illness scripts. Novice clinicians rely on hypothetico-deductive reasoning solving problems by applying knowledge of the relationships between signs, symptoms and pathophysiology of disease. With experience, clinicians employ more pattern recognition, reserving analytical approaches for complex cases.19 ,20

Strengths of our study are that it was conducted in a clinic setting but as part of a large diagnostic accuracy study. Also, PID cases were identified from clinical notes, not from diagnostic codes. Although obtained from one clinic, which may limit the generalisability of findings, our data are broadly in keeping with those of others. Overall, the proportion of women diagnosed with PID (4.4%) was higher than the mean rate (1.7%) reported in a study of experienced sexual health doctors. However, it was comparable with their higher rate diagnosing doctors (1.8–5.7%) who were thought to detect greater numbers of true PID cases.10 Proportions of PID cases in our study that were positive for CT (23.7%), and CT/GC (27.3%) were comparable to rates reported in a UK audit of PID in sexual health clinics (24.2% and 25.3%, respectively).3

The number of PID cases at 169 was relatively small but there were sufficient diagnoses to enable significant differences in clinical presentations, and CT/GC rates, between grades to be identified. However, it is possible that the power to detect all true differences has been limited. Additionally, the calculation of multiple ORs may have led us to detect some differences due to chance. Repeating our study in different clinic settings with a larger sample would strengthen our findings. Further limitations include the assumption that all women had a similar chance of having PID. Women were not randomly allocated to the different grades of clinician but were seen in order of registration by the next available clinician. It is possible that this introduced some bias, but there were no differences in rates of CT/GC and meeting key PID criteria between the clinician groups. Also, using CT /GC positivity to indicate the specificity of PID diagnoses is not considered a ‘gold standard’. These infections are common causes of PID but other pathogens also play a role.4 ,5 Last, our study focused on clinician groups with varying levels of experience, but wide variation can exist even between similarly experienced individuals.10

This study highlights a number of implications for training. Recently, the focus of postgraduate education has shifted to competency-based outcomes.21 ,22 Our study shows competence is not an all-or-none phenomenon, but a continuum. The importance of the content and context of prior learning and personal attitudes renders the pattern and speed of clinical skill acquisition unique to each individual.20 ,23 PID is a complex syndrome, presenting a challenge even to experienced clinicians.10 Thus, even after appropriate training, clinicians may struggle with the diagnosis and should be appropriately supported by senior colleagues. At the time of our study, B6 nurses and FY/CMT doctors were trained in-house, under the supervision of senior clinicians with ascertainment of competence for autonomous practice largely determined by their supervisor. Since this study, additional training is given on: the importance of fulfilling key criteria; being cautious of diagnosing PID without lower abdominal pain; cumulative clinical features adding specificity; and performing and interpreting bimanual pelvic examinations. Courses designed to train and assess competency in management of symptomatic women, including pelvic bimanual examinations, (eg, BASHH STIF Level 2 Competency courses) may help create more clearly defined objective standards for independent practice.24

It is not surprising that inexperienced clinicians struggle to formulate an accurate diagnosis of PID, given its complex syndromic nature. It is apparent from our data that experienced clinicians are better at obtaining and assimilating information from the history and examination to reach the most likely diagnosis. Audit of individuals’ clinical practice is becoming a requirement for appraisal and revalidation. We suggest that analyses of PID rates and proportions with confirmed CT/GC, compared with the overall clinic rates, may be useful for evaluating quality assurance, and could be used for longitudinal assessment of individual clinicians’ developmental progression over time.

Key Messages

  • Women diagnosed with pelvic inflammatory disease (PID) by experienced clinicians were more likely to have lower abdominal pain with pelvic tenderness and additional supportive symptoms than those diagnosed by inexperienced clinicians. Inexperienced clinicians should be cautious of diagnosing PID in the absence of lower abdominal pain.

  • Performance and interpretation of bimanual pelvic examinations should be a focus of training, including evaluation of their usefulness in women without lower abdominal pain.

  • Specificity of PID diagnoses based on the proportions with confirmed Chlamydia trachomatis (CT) and/or Neisseria gonorrhoea infections increased with clinician experience.

  • Analyses of rates of PID, with CT and/or gonorrhoea, compared with overall clinic rates could be used for individuals’ developmental progression and quality assurance.

References

View Abstract

Footnotes

  • Handling editor Jackie A Cassell

  • Contributors JDW conceived the study and wrote the protocol. GCM, CMWS, SAS, and JDW recruited participants. CMWS, SAS and JDW, produced the database. GCM and JDW analysed the data. All authors contributed to writing the paper. JDW and GCM are the guarantors for the study. All authors had full access to all of the data in the study and can take responsibility for the integrity of the data and the accuracy of the data analysis.

  • Funding Funding, in the form of the extra diagnostic reagents and equipment needed for the CT/GC diagnostic test study, was provided by Gen-Probe. The funders had no role in the initiation or design of the study, collection of samples, analysis, interpretation of data, writing of the paper, or the submission for publication. The study and researchers are independent of the funders, Gen-Probe.

  • Competing interests None.

  • Ethics approval Leeds (East) research ethics committee granted ethical approval for the study (REC reference 09/H1306/4). All participants gave informed consent before taking part in the study.

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

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.