Tuddenham, Susan MD, MPH; Ghanem, Khalil G. MD, PhD
Sexually transmitted infections (STIs) represent an important cause of morbidity. However, STI clinics, particularly in poor and underserved areas, face increasing resource constraints, with declining budgets in the face of significant demand. Given these limitations, it is important to reassess how to direct and manage clinical care to optimize disease detection and treatment.
Molecular techniques that enable the detection of gonorrhea, chlamydia, and trichomonas from genital and extragenital sites may offer opportunities to detect disease in time- and cost-efficient ways.1 The performance characteristics of self-collected specimens have been found to be equivalent to clinician-obtained specimens.2 The Centers for Disease Control and Prevention recommends urine as the specimen of choice for the molecular diagnosis of urethral gonorrhea and chlamydia in men.3 Certain STIs may still require a physical examination for diagnosis. However, it is possible that groups of men who are deemed through initial screening to be at low risk for STIs, which require a physical examination to diagnose may be able to proceed directly to accelerated or “express” STI testing. In “express testing,” these men would bypass the routine STI clinic process, which involves a lengthy interview and physical examination, and would simply submit genital and extragenital swabs for molecular testing, as well as blood samples for HIV and syphilis testing. Because patients going through express testing would not have to undergo a face-to-face interview and physical examination, this might enable the clinic to serve more patients in a limited amount of time.
This type of express testing is already becoming a reality: several studies have documented the roll out of express testing algorithms for men and women in STI clinics. In one study by Shamos et al.,4 a Denver STI clinic offered express testing without physical examination to patients who were deemed low risk (ie, excluding symptomatic patients, contacts to patients with STIs, men who have sex with men, intravenous drug users, commercial sex workers, and those with multiple partners or very high risk sexual behaviors) through a brief triage questionnaire. Express visits resulted in a 39% time saving for men and 56% for women. Express visit clients had lower rates of gonorrhea, chlamydia, syphilis, and HIV (9.8% vs. 26.0% in standard visit clients).The researchers were unable to assess the number of missed STIs in the express testing group, although a subgroup analysis implied that these rates might be low. A study done in the Netherlands showed similar results: the prevalence of STIs in the express testing group was 7.6% versus 18.1% in the standard group. Using the express testing strategy, they were able to increase the number of patients screened by 9%.5 In Australia, a new testing “Xpress” service at a large sexual health clinic was associated with improved patient journey and clinic capacity.6 Most clients were highly satisfied with all aspects of care in the Xpress option and reported that they would both return to use it again and would recommend it to friends.7 In New York City, an express testing option increased patient volumes, increased detection of chlamydia and gonorrhea, and even decreased time to treatment.8
These studies demonstrate that rolling out express testing in STI clinics is feasible and may be associated with improved clinic capacity and high client satisfaction. Because clinics around the country and the world are faced with whether to move forward with express testing options, it is important to define the contribution of the physical examination in detecting clinically meaningful diagnoses that may not be identified by current laboratory methods and thus better understand what diagnoses might be missed in patients undergoing express testing. It is equally important to identify characteristics that can stratify men into appropriate risk categories for STIs, which might be missed without physical examination, to understand which groups of men might be eligible for express testing.
The analysis below examines a cohort of men attending 2 STI clinics in Baltimore, Maryland. Our goals were 2-fold: (1) to define the contribution of the physical examination in detecting clinically meaningful diagnoses that may not be identified by current laboratory methods (ie, to define which diagnoses would be missed without a physical examination as well as to identify missed opportunities for same-day treatment) and (2) to identify characteristics that can stratify men into the appropriate risk category.
MATERIALS AND METHODS
This is a retrospective study from a large electronic database of all male visits to 2 urban STI clinics in Baltimore between 1990 and 2001. Data collected after 2001 contained less extensive information on symptoms and physical findings. Because there is no reason to think that signs and symptoms of STI infections at these clinics in Baltimore have changed from 2001 to the present, data after 2001 were excluded. This analysis was granted approval by the institutional review boards of the Johns Hopkins Medical Institutions and the Baltimore City Health Department.
The standardized male clinical assessment at the Baltimore STI clinics includes a structured interview on current symptoms, STI history and behavioral risk factors, a physical examination, laboratory testing, clinician impressions, treatment, and referrals. Patients are asked about the reason for their visit (eg, routine checkup, contact with an infected partner, current symptoms [which included discharge, dysuria, genital lesion, genital itching, rash, and irritation/odor]), antibiotic use, illicit drug use, number of sexual partners, STI history, sexual orientation, and sexual risk behaviors. A directed physical examination on each patient includes evaluation of the eyes, oropharynx, skin, abdomen, inguinal nodes, penis, testicles, and, when indicated, rectum. This examination documents lymphadenopathy, rash, discharge, ulcers, vesicles, warts, and other lesions. All findings on the history, physical examination, and outside referrals are documented on the encounter form and captured in the electronic clinical database.
Definitions and Data Analyses
Our first goal was to define the probability of missed diagnoses occurring in men attending the STI clinics if physical examinations had not been performed. We defined a “missed diagnosis” as any clinical diagnosis that would have been missed if a physical examination were not performed and for which a diagnostic test is unavailable. We assumed that diagnostic tests for gonorrhea, chlamydia, and trichomoniasis could be easily performed on self-collected specimens, thereby bypassing the need for a physical examination. Although serological tests are available to detect syphilis and herpes simplex virus (HSV) infections, serological tests available at the time might have been negative in up to 30% of persons with primary syphilis, in up to 25% of persons with tertiary syphilis, and in persons with early primary HSV infections. We therefore assumed that the presence of an ulcer that was diagnosed as primary HSV would have been missed with express testing and that some ulcers diagnosed as primary syphilis would also have been missed. There were no diagnoses of tertiary syphilis in our database. The findings of balanitis, scabies, pthirus pubis, penile warts, anogenital lesions, molluscum, scrotal masses, epididymitis, and disseminated gonococcal infection require a physical examination to be diagnosed. We therefore posited that all these findings would be also missed.
At our STI clinics, all findings on the history, physical examination, and outside referrals are documented on the encounter form and captured in the electronic clinical database. Primary syphilis was diagnosed clinically, and dark-field microscopy was performed for any consistent lesions. Blood samples were also sent for rapid plasma reagin and florescent treponemal absorption testing. Syphilis staging followed Centers for Disease Control and Prevention criteria. Herpes simplex virus cultures were performed on all anogenital ulcers in patients without an HSV history. Gonorrhea was diagnosed by culture. Disseminated gonococcal infection was diagnosed using cultures and clinical findings. Nongonococcal urethritis (NGU) was diagnosed by Gram stain because testing for Chlamydia trachomatis was not performed. HIV serological testing was performed using a screening enzyme immunoassay, followed by a Western Blot. Of note, patients with early HIV could be missed by serological testing that did not include HIV RNA testing. Routine testing for Trichomonas vaginalis was not performed.
For determining men’s risk of missed diagnosis based on initial screening, men were retrospectively stratified into 3 separate groups based on the self-reported reason for their visit. Group 1 included men who were asymptomatic presenting for general checkups (ASYM), group 2 included all men who came in complaining of symptoms (SYM; symptoms could include discharge, dysuria, genital lesion, genital itching, rash, and irritation/odor), and group 3 included asymptomatic men who presented as known contacts of STI-infected partners (CON; these included partners diagnosed as having gonorrhea, chlamydia, syphilis, trichomonas, or pelvic inflammatory disease). Only men coming in for their first clinic visit were included in this analysis.
Proportions were compared using the χ2 test. P values less than 0.05 were assumed to represent statistical significance. We used the k statistic to measure agreement between self-reports and objective findings. k values less than 0.20, less than 0.6, and greater than 0.6 were assumed to represent low, moderate, and high agreement, respectively. We used logistic regression to assess which risk factors were most predictive of missed diagnoses. Ninety-five percent confidence limits are presented for all point estimates. Data analyses were performed using STATA v. 11.0. For the multivariable models, we performed a stepwise analysis. We ran separate multivariable models for drug use, clinical factors, and, for SYM patients, symptoms. For the predictive model, we included the parameters from each category that remained significant in the multivariable model. Because all kinds of drug use were significant and we anticipated significant collinearity between these, we chose one, intravenous drug use (IVDU), rather than including all 4 types of drug use in the final model.
Only the first visit to the STI clinic was used for each patient. Of 140,052 patient visit records, 58,073 met this entry criterion. The other 81,979 records were excluded from the study because they represented multiple visits. A total of 29,172 men were asymptomatic (ASYM), 23,971 were symptomatic (SYM), and 4929 were contacts (CON). Of note, the mean age was 30 years, and most patients in all groups were African American. The demographic and behavioral characteristics of each group and those excluded from the study are summarized in Table 1. The proportion of men with significant findings on physical examination, including penile discharge, penile ulcer, penile rash, rectal abnormalities, and oropharyngeal abnormalities, was generally higher in the SYM and CON groups than in the ASYM group.
Potential missed diagnoses were represented in Table 1 and include balanitis, scabies, pthirus pubis, epididymitis, scrotal mass, penile wart, herpes, molluscum, and primary syphilis with negative serology results. These conditions all require a physical examination for identification, and therefore, the values listed for them represent the patients in each group whose disease would be missed if a physical examination were not done. In general, the proportions of patients who were found to have these conditions were higher in the SYM group than in the ASYM and CON groups. However, in no group was the proportion of missed diagnoses higher than 2.5%. The numbers of patients with disseminated gonococcal infection or primary syphilis, which was negative on serology and would therefore have been missed without a physical examination, was also low in all groups.
Concordance between reported symptoms and findings on physical examination was high for the reported symptom of discharge (κ = 0.6). Concordance between the other reported symptoms (dysuria, irritation/odor, lesion, genital itching, and rash) and physical examination findings was poor (κ < 0.2). We also stratified our findings by circumcision status (based on clinician assessment), HIV status, sexual orientation, and illicit drug use to determine whether these factors affected reporting of symptoms or prevalence of diseases. In general, patients who were uncircumcised were more likely to report being asymptomatic when in fact they were found on examination to have penile ulcers (2.37% vs. 0.35%, P < 0.001) or penile rashes (1.46% vs. 0.15%, P < 0.001). Patients who were uncircumcised and reported no symptoms were also more likely than those who were circumcised and those who were asymptomatic to be diagnosed as having primary syphilis (1.18% vs. 0.14%, P < 0.001). The proportions overall, however, were low (<2.5%). Patients who were HIV positive and claimed to be asymptomatic were more likely than those who were HIV negative to have penile ulcers (1.15% vs. 0.48%, P = 0.006) but were not more likely to have penile rashes detected on physical examination. Importantly, HIV-positive men who claimed to be asymptomatic were not more likely than HIV-negative asymptomatic patients to be diagnosed as having primary syphilis. In fact, the only disease they were more likely to be diagnosed with was balanitis (0.62% vs. 0.26%, P = 0.04), but the probabilities were small (<0.7%). Very few patients admitted to using illicit drugs, so differences, if any, were small (<2%) and not clinically meaningful.
To determine whether certain symptoms among the SYM group could predict potential missed diagnoses if a physical examination were not performed, we analyzed the proportion of men with each potential missed diagnosis stratified by 6 symptoms: discharge, dysuria, lesion, rash, genital itch, and irritation/odor (Table 2). Compared with men not reporting “lesions,” there was a higher proportion of men who reported lesions diagnosed as having penile warts, herpes, and primary syphilis. Similarly, a larger proportion of men who reported a “rash” were diagnosed as having balanitis, scabies, and herpes, and a larger proportion of men who reported a “genital itch” were diagnosed as having scabies and pthirus pubis. No symptom increased the probability of detecting epididymitis, scrotal masses, molluscum, and disseminated gonococcal infections.
We constructed multivariable models to identify factors that would predict a missed diagnosis if no physical examination were performed (Table 3). Younger age (≤22 years) and illicit drug use were predictive of missed diagnoses in all 3 groups. Being uncircumcised increased the risk of a missed diagnosis for the ASYM and SYM groups but not for the CON group. History of STI increased the risk of a missed diagnosis for the ASYM and CON groups, but not for the SYM group. Among the SYM group, self-reported discharge and dysuria significantly predicted a decreased probability of a missed diagnosis.
We constructed a hypothetical population of 1000 men for each of our 3 groups, ASYM, CON, and SYM (Table 4), to quantify the number of potential missed diagnoses should a physical examination not be performed. Numbers were small in all groups. The overall proportion of patients who would have had any missed diagnosis (including scabies, pthirus pubis, epididymitis, scrotal mass, penile wart, herpes, molluscum, or primary syphilis with negative serologies) if they had not had a physical examination was relatively low for patients in the ASYM group (2.65%) and CON group (4.54%). However, it was higher in the SYM group (10.40%).
Patients who go through express testing also may miss out on the opportunity for same-day treatment of STIs of public health significance, including primary or secondary syphilis, HSV, and urethritis (either gonorrhea [GC] or NGU). Numbers for HSV and syphilis, as previously discussed and shown in Tables 1 and 4, were low in all groups. Assessing NGU and GC was more difficult. Only patients who had a discharge on physical examination received a Gram stain, so more than 80% of asymptomatic patients did not have a Gram stain. Table 1 shows that 24.2% of SYM patients were diagnosed as having GC urethritis on Gram stain and 26.0% were diagnosed as having NGU. Also, 0.55% of ASYM patients would be diagnosed as having GC on Gram stain and 6.1% were diagnosed as having NGU.
In a separate analysis (not included in the table), we found that 6.0% of ASYM patients had a penile discharge on examination. Of these ASYM patients, 77.3% with discharge received a Gram stain. On Gram stain, 6.3% of these patients had GC and 55.8% of these patients had NGU. Based on these numbers, 0.29% of ASYM patients were found to have GC and 2.6% were found to have NGU. For SYM patients, 67.4% had a penile discharge on examination. Of these patients, 72.8% received a Gram stain and 47.9% were positive for GC, whereas 38.1% were positive for NGU. Based on these Gram stain numbers, 23.5% of SYM patients were found to have GC and 18.7% were found to have NGU. Among CON patients, 29.7% had a discharge on examination and 76.1% of these had a Gram stain. Of these patients, 4.6% had GC and 56.1% had NGU on Gram stain. Based on these Gram stain numbers, 1.0% of CON patients had GC and 12.7% had NGU. Based on this calculation, in Table 4, we showed that for a hypothetical population of 1000 patients, 187 SYM men and 26 ASYM men would be diagnosed as having NGU and 23 SYM men and 2.9 ASYM men would be diagnosed as having GC.
Ideally, all patients seeking STI care would undergo a physical examination in addition to appropriate laboratory studies. However, declining resources in the face of increasing demand mean that STI clinics must consider adopting new strategies to maximize efficiency without compromising quality of care. One such approach, which is being implemented in a number of STI clinics, is to make use of molecular tests to establish express testing options that bypass physical examinations and face-to-face interviews in patients at low risk for having an STI that requires physical examination for diagnosis. In this study, we attempted to define the contribution of the physical examination to identifying STIs in men in the era of molecular tests and identify clinical predictors that would aid in risk stratification.
Our study suggests that physical examination is important for identifying some STIs in men, including balanitis, scabies, pthirus pubis, epididymitis, scrotal mass, penile wart, herpes, and molluscum, as well as a disseminated gonococcal infection and the small proportion of primary syphilis cases, which are negative on serological blood tests. However, when all possible missed diagnoses were combined, the proportion of men who were asymptomatic (the ASYM group) who would have had a missed diagnosis without a physical examination was relatively low (only 2.6%). The proportion of missed diagnoses in the CON group was slightly higher (4.5%). There was a higher proportion of missed diagnoses in the SYM group (10.4%).
Contacts of patients with STIs necessarily require same-day treatment and therefore would not be candidates for accelerated testing. However, our results do suggest that asymptomatic men (ASYM) could be safely shunted into express testing after an initial screening interview with a resultant relatively low number of missed diagnoses. For disseminated gonococcal infections, missed diagnoses never exceeded 0.15% in any of the 3 groups. Missed primary syphilis diagnoses were also low, not exceeding 0.24% in SYM and only 0.03% and 0.02% in ASYM and CON groups, respectively. Scrotal masses and epididymitis were also uncommon. In a hypothetical population of 1000 men for each of our 3 groups, ASYM, CON, and SYM (Table 4), it is clear that the numbers of missed diagnoses are small across all groups and particularly low in ASYM men, again suggesting that the probability of a serious missed diagnosis in most ASYM men is quite low and may be clinically acceptable.
We performed a similar assessment in women attending our STI clinics.9 In that study, we found that if self-collected vaginal specimens and serologies for syphilis and HIV were obtained from all patients, and no speculum and bimanual examinations were performed, clinically relevant diagnoses (ie, 21% of primary syphilis diagnoses missed by serology, a missed pelvic inflammatory disease diagnosis, and lesions detected by examination, which lead to referral) would be missed in 2.3% of ASYM, 9.3% of SYM, and 3.3% of CON patients. These findings are quite similar to our findings in men.
As discussed in our study, there was certainly a higher proportion of missed diagnoses in the SYM group of men (10.4%) as compared with the CON (4.5%) and ASYM groups (2.6%). Therefore, it would seem that symptomatic men coming to STI clinics should always undergo a full physical examination. However, this finding also raises an important question: Among the SYM group, are there specific symptoms that could predict a potential missed diagnosis should the physical examination not be performed? Could men who did not endorse these particular symptoms therefore be safely shunted to express testing? To answer these questions, we analyzed the proportion of people with each missed diagnosis stratified by 6 self-reported symptoms: discharge, dysuria, rash, lesion, genital itch, and irritation/odor. We constructed multivariable models to help predict the independent influence of each variable. We found that 3 self-reported symptoms (rash, lesion, and genital itch) increased the probability of detecting a potential missed diagnosis, whereas 2 others, dysuria and discharge, decreased the probability. If all SYM patients not reporting a rash, lesion, or genital itch were assigned the express testing route, the probability of a missed diagnosis would decrease from 10.4% to 3.7%. This implies that although men reporting rash, lesion, and genital itch should always have a physical examination, it is possible that other symptomatic men could safely undergo expedited STI testing.
Of note, with the exception of the reported symptom of discharge, there was poor concordance between reported symptoms and findings on physical examination. However, men’s report of any symptom was correlated with having an STI, and the reported symptoms of rash, lesion, and genital itch were particularly correlated with having a disease process that would have been missed without a physical examination. This suggests that although men cannot necessarily identify their exact pathology, they are able to articulate symptoms that suggest that they may be at higher risk for some STI, which could be identified on physical examination. For selected physical findings and diagnoses, there was a trend toward uncircumcised men, HIV-positive men, cocaine users, IVDU, and cannabis users being less reliable in reporting than the rest of the population. However, numbers were generally small.
Patients who do not undergo physical examination may miss out on the opportunity for same-day treatment of STIs of public health importance, including syphilis, HSV, GC, and NGU. Paneth-Pollak et al.8 have shown that express testing options may lead to decreased time to treatment. In addition, delay in treatment may become much less of an issue as rapid point-of-care tests for syphilis, GC, chlamydia, and trichomonas become available and can be implemented in STI clinics. Nonetheless, these point-of-care tests are not yet widely available, and delay in treatment does remain a consideration. The proportions of patients who were found to have syphilis and HSV on examination and therefore missed the opportunity for same-day treatment were low across all groups, suggesting that lack of same-day treatment of these STIs may not provide a significant barrier to express testing in either ASYM or SYM groups. The issue of urethritis is more complex. Studies such as those by Paneth-Pollak8 and Orellana et al.10 suggest that Gram stain can be significantly less sensitive than nucleic acid amplification test, particularly among asymptomatic (or pauci symptomatic) men. Therefore, it is possible that express testing in asymptomatic men may not significantly impact the treatment for index patients and their partners. This is supported by our results, which showed that only a small proportion of ASYM men (6.0%) were found to have penile discharge on examination. Most of these men with penile discharge received a Gram stain. Based on these Gram stain results, approximately 0.29% of ASYM patients would have had a missed opportunity for same-day treatment of GC and 2.6% would have had a missed opportunity for same-day treatment of NGU. These numbers are relatively low, suggesting that although missed opportunity for same-day treatment is a consideration for ASYM patients, it is may still acceptable for them to go through express testing. Of note, not all NGU is caused by GC, CT, or trichomonas. A significant proportion may be caused by mycoplasma and other organisms. Unless molecular tests become widely available for all organisms that cause NGU, some patients not undergoing physical examination may miss out on treatment entirely.
The Gram stain among symptomatic men, however, can certainly impact same-day treatment. More than half (67.4%) of SYM patients had discharge on examination, and of those who received a Gram stain, 47.9% (or 23.5% of all SYM patients) had GC and 38.1% (or 18.7% of all SYM patients) had NGU. This represents a significant proportion of patients in the SYM category in whom the opportunity to have same-day treatment would have been missed. Neither can specific symptoms be used to reliably distinguish patients who will have urethritis. The proportions of patients who will have GC are certainly highest for those complaining of discharge and dysuria; however, a high percentage of patients (22.3%–29.9%) complaining of rash, lesion, genital itch, or irritation/odor will still be diagnosed as having NGU on gram stain (Table 2). This may mean that until point-of-care tests become available, STI clinics may still choose to perform physical examinations on all symptomatic men.
Our study has several limitations. This was a retrospective analysis. As such, misclassification of missed diagnoses could not be ruled out. Any misclassification, however, was likely nondifferential across the 3 risk groups. Physical examinations were conducted by more than 30 providers during the inclusion period. Although this reflects the reality of clinical care in most settings, provider-specific differences cannot be ruled out. This variability was likely small because providers in our clinics undergo the same training and the medical directors at each clinic reviewed 100% of all charts in real time. Any discrepancy or inconsistency was immediately addressed. We did not routinely test for C. trachomatis and T. vaginalis in men at the time (but molecular tests for both infections are now routinely available), and we did not perform routine serological testing for HSV infection. In-depth data for this study were only collected from 1990 to 2001. Data collected after 2001 contained less extensive information on symptoms and physical findings. There is no reason to think that signs and symptoms of STI infections at these clinics in Baltimore have changed from 2001 to the present. Using these data, it should be possible for us to extrapolate what might be useful in the new, current era of molecular tests. Finally, our article did not address the issue of extragenital gonorrhea and Chlamydia because we have relatively limited data on these infections in our patient population. There have been publications that have validated self-collected rectal specimens for nucleic acid amplification tests. For example, Moncada et al.11 showed that sensitivities for self-collected and clinician collected rectal swabs for GC and CT in MSM were comparable. Van der Helm et al.12 showed similar results for men and women. Data on self-collected pharyngeal swabs for GC testing are limited; however, it is conceivable that express testing could include an option for self-collected rectal and/or pharyngeal swabs.
No one testing strategy will fit every STI clinic. However, our study suggests that it may be possible for STI clinics to adopt new testing algorithms for men based on appropriate risk stratification. In particular, asymptomatic men could safely undergo expedited STI testing without a lengthy clinical interview and physical examination. Symptomatic men with certain specific complaints may also be considered for expedited STI testing, although the missed opportunity for same-day treatment of NGU and GC may be an issue of public health significance in these patients. New express testing options can allow clinics to enhance efficiency and potentially decrease costs.