Which tests are the most useful for diagnosing PID?

Carol E. Blenning, John Muench, Dolores Zegar Judkins

Research output: Contribution to journalReview articlepeer-review

14 Scopus citations


Our search for articles that examined patient- and primary care-oriented PID diagnostic tests resulted in 2 systematic reviews, no randomized controlled trials, 4 data analyses, and 5 cohort studies, all of which were fair- to good-quality. Systematic reviews don't show consistent results: One systematic review of 12 fair- to good-quality studies, based in Europe and the US, included urban populations treated in Ob/Gyn departments, emergency rooms, and sexually transmitted disease clinics. This review supports a thorough evaluation when more severe disease is suspected and the use of sensitive diagnostic tests for suspected mild disease-eg, CRP (74%-93% sensitivity) and ESR (64%-81% sensitivity for value >20 or 15 mm/h). Another systematic review of 19 fair- to good-quality cohort studies found a sensitivity of only 64% for laparoscopy, 50% to 87% for endometrial biopsy, and up to 80% for microbiological tests. Results were not consistent for the reported sensitivity of WBC, ESR, or CRP. Multivariate analyses of Swedish data come to different conclusions: We identified no randomized controlled trials that addressed the diagnosis of PID. Two multivariate analyses of the same Swedish data from the 1960s came to different conclusions. The Lund analysis includes data collected between 1960 and 1969 at Lund University Hospital in Sweden on women with suspected PID, with about 625 cases included for these analyses. Simms et al found insufficient evidence from these data for any existing diagnostic criteria. Looking at the same data, Hagdu et al proposed the use of a clinical criteria model including low abdominal pain and 2 or more of the following other criteria: vaginal discharge, temperature greater than 38°C, vomiting, irregular menses, urologic or proctitis symptoms, pelvic tenderness, adnexal mass or swelling, and ESR ≥15. This model had a reported sensitivity of 87%, specificity of 52.5%, and false-positive and false-negative rates of 21.2% and 33.3%, respectively. Looking at adnexal tenderness aids sensitivity of other tests: Cross-sectional analysis of a multisite US randomized treatment trial supported using adnexal tenderness as a minimum clinical criterion to increase sensitivity. Further analysis of that trial suggests that some asymptomatic women are at equivalent risk of developing sequelae compared with symptomatic women diagnosed with PID. Those asymptomatic women who met diagnostic criteria with a positive endometrial biopsy were more likely to have pelvic tenderness than asymptomatic women who were not diagnosed. Symptoms >1 week and elevated WBC also helpful: Two small, fair-quality cohort studies (N=61 and 176, respectively) investigated the use of clinical diagnostic criteria for PID. The smaller study compared clinical criteria to several reference standards (laparoscopy, histology, microbiological markers, and transvaginal ultrasound) and found clinical criteria, specifically adnexal tenderness, most sensitive (87%), and laparoscopy most specific (100%). In the second study, the authors evaluated 176 consecutive admissions for clinically diagnosed PID, 76% of which were laparoscopically confirmed. Reviewing clinical indicators, they found that a combination of adnexal tenderness, symptoms for <1 week, and elevated WBC was the most sensitive set of predictors (sensitivity 86.6%, specificity 45.7%) with positive predictive value of 0.84 and negative predictive value of 0.52. Useful lab indicators: C-reactive protein, serum CA-125 Three small cohort studies (N=50-152) of fair-quality evaluated various laboratory indicators in the diagnosis of PID. Each used a different reference standard: clinical criteria, laparoscopy, and endometrial biopsy, respectively. One study found CRP >10 to be 93% sensitive and 83% specific in a cohort of women admitted to the emergency department with an acute gynecological disorder. This population had a high baseline incidence of PID, pregnancy, and intrauterine device use. A study of serum CA-125 levels showed a predictive value of 97% for values >16 U/mL in diagnosing salpingitis. This test might therefore be useful in confirming peritoneal involvement when PID is suspected clinically. Another study developed a model using vaginal WBC (the single most sensitive factor at 78%), serum WBC (the single most specific factor at 88%), CRP, and ESR. The model was 100% sensitive if the diagnosis only required 1 positive test, although the specificity was only 18%. The positive predictive value was 65%. If all 4 were positive, specificity was 95%, with 29% sensitivity, a positive predictive value of 90%, and a negative predictive value of 47%. Prevalence was 60% in the group studied. Recommendations from others: The CDC recommends empiric treatment of women with lower abdominal or pelvic pain who are at risk for sexually transmitted diseases with uterine, adnexal, or cervical motion tenderness and no other identifiable cause. Clinical Evidence found no RCTs that compared empiric treatment of suspected PID with waiting for microbiological test results for guidance. The Agency for Healthcare Research and Quality recommends requiring the presence of lower abdominal, adnexal and cervical tenderness, without alternative diagnosis, for the diagnosis of PID. Temperature >101°F, cervical or vaginal discharge, elevated ESR, and positive gonococcal or chlamydia cultures all increase specificity of diagnosis. The United Kingdom's national guideline recommends maintaining a low threshold for empirical treatment, citing a lack of definitive diagnostic criteria and potential for sequelae, but does recommend testing for gonorrhea and chlamydia.

Original languageEnglish (US)
Pages (from-to)216-220
Number of pages5
JournalJournal of Family Practice
Issue number3
StatePublished - Mar 2007

ASJC Scopus subject areas

  • General Medicine


Dive into the research topics of 'Which tests are the most useful for diagnosing PID?'. Together they form a unique fingerprint.

Cite this