Overtreatment of gonorrhea and chlamydial infections in 2 inner-city emergency departments

a b s t r a c t

Background: Infections with Neisseria gonorrhoeae (GC) and Chlamydia trachomatis (CT) are the most commonly reported Sexually transmitted diseases in the United States.

Objectives: The primary objective of this study was to estimate the incidence of overtreatment of GC and CT infec- tions in the emergency department (ED). The secondary objective was to determine if there are clinical variables that predict infection with GC and CT. Methods: A retrospective medical record review was performed at 2 inner-city hospitals. Records were obtained from the evaluation of female patients who presented to the ED between January 1, 2012, and December 31, 2012, who were tested for GC and CT infection. A standardized form was used to extract specific information from each medical record.

Results: Data were extracted from 538 medical records. Of the 522 ED visits, 32 (6%) yielded test results positive for either GC or CT, including 3 that were positive for both. Treatment was administered to 101 patients (19%) and declined by an additional 9 (2%). Of those receiving antibiotics, 87 of 101 (0.86; 95% confidence interval, 0.77-0.92) had negative test results. Of those not offered antibiotics, 17 of 412 (0.04; 95% confidence interval, 0.02-0.07) had positive test results. The overtreatment proportion was similar at hospitals (55/66 [0.83] and 32/35 [0.91], respectively). Of clinical variables that were considered, only age less than 19 years was statistically associated with a positive test result for GC and CT.

Conclusion: The rate of overtreatment for GC and CT was 86%. The practice of empirical treatment should be reconsidered.

Introduction

Infections with Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (GC) are the most commonly reported Sexually transmitted diseases in the United States. An estimated 2.8 million cases of CT infection and 718 000 cases of GC infection are reported each year [1]. The city in which this study was conducted, Baltimore, is ranked 22nd in the nation for newly diagnosed cases of gonorrhea, with 97.6 cases per 100 000 res- idents, and 21st in the nation for newly diagnosed cases of chlamydial in- fection, with 455.3 cases per 100000 residents [1]. Despite the prevalence of these infections nationwide, their diagnosis continues to be a challenge in the emergency department (ED). When these infections are suspected,

? We have no conflicts of interest to disclose.

?? CH, TVP, and HM conceived the study, designed the study, and obtained research funding. CH, TVP, HM, and MW supervised the data collection. CH, TVP, HM, and GW un- dertook recruitment of participating centers and patients and managed the data, including

quality control. MW provided statistical advice on study design and analyzed the data. CH, TVP, and HM drafted the manuscript; and all authors contributed substantially to its revi- sion. CH, TVP, and HM take responsibility for the paper as a whole.

* Corresponding author.

E-mail address: mwitt001@umaryland.edu (M.D. Witting).

endocervical specimens are collected and sent for DNA testing; but the re- sults are not available the same day. This delay forces health care pro- viders to make Treatment decisions based on the clinical presentation. When deciding whether to treat a patient with antibiotics, providers must balance the risks associated with untreated disease and the poten- tial consequences of overexposure to antibiotics. Patients with untreated infections are at increased risk of morbidity, specifically infertility, Ectopic pregnancy, urethritis, cervicitis, Pelvic inflammatory disease, and shed- ding of the human immunodeficiency virus [2]. According to 2012 data from the Centers for Disease Control, untreated chlamydial infection progressed to pelvic inflammatory disease in 10% to 15% of women [1]. Because of these possible consequences, emergency care providers often choose to treat patients for GC and CT empirically. This approach carries significant risks, for example, increasing the likelihood of antibiotic resis- tance, Allergic reactions, and increased health care costs [3-5].

The primary objective of this study was to estimate the incidence of overtreatment of GC and CT infections in the ED. Our hypothesis was that more than 50% of patients who are treated empirically for these in- fections will have negative test results. The secondary objective was to determine if there are historic factors or physical findings that are pre- dictive of infection with GC and CT.

http://dx.doi.org/10.1016/j.ajem.2015.06.009

1266 C.E. Holley et al. / American Journal of Emergency Medicine 33 (2015) 1265–1268

Methods

Study design

This retrospective medical record review was designed to determine

Table 1

Characteristics of patients (total = 522)

Variable

Median age (interquartile range) 25 (21-33) Hospital

the rate of overtreatment for GC and CT infection among women pre-	University tertiary care	262 (50%)
senting to EDs with lower abdominal pain or vaginal complaints. The	Community hospital	260 (50%)
study design was approved by the institutional review boards at both hospitals.	History of STD Examination findings Discharge	91 (18%) 161 (31%)
	Cervical motion tenderness	57 (11%)
2.2. Data collection	Adnexal tenderness	89 (17%)

Medical records were obtained from 2 sites–an academic tertiary care center and a private community hospital located within 1 mile of each other in a major US city. Medical records were selected based on the following criteria: female; age between 12 and 85 years; ED presen- tation between January 1 and December 31, 2012; complaint of lower abdominal, pelvic, or vaginal pain; and receiving a pelvic examination, including the collection of an endocervical specimen for laboratory test- ing for GC and CT infections. A random sample was selected from those selected medical records using the randomization function on a com- puterized spreadsheet. Each medical record was then reviewed to see if it met the eligibility criteria for this study: a documented pelvic exam- ination, no recent treatment for gonorrhea or chlamydial infection, doc- umentation of the physician’s decision-making process regarding treatment, and the availability of test results. We did not exclude repeat visits from the same patient.

To standardize data collection, an abstraction form with predefined clinical variables was created. Abstractors were medical students from the institution with which the authors were affiliated who were trained by the study team to collect specific information from the medical re- cords. The data collection process had 2 phases: In the first, to ensure uniformity, data were abstracted from the first 40 medical records by 2 abstractors, one of whom was a co-investigator. Discrepancies were discussed with the abstractors to ensure clear variable definitions. In the second phase, abstractors reviewed medical records independently; a random sample of 10% of these medical records was assigned to 2 re- viewers to allow calculation of Interrater agreement.

2.3. Data analysis

The primary objective of this study was to determine the overtreat- ment proportion, defined as the fraction of patients who received empir- ical treatment but had negative test results. We defined the undertreatment proportion as the proportion of those not prescribed an- tibiotics that had positive test results. The administration of ceftriaxone and azithromycin or prescribing doxycycline was considered empirical treatment for GC and CT. Administration and prescription for metroni- dazole or other antibiotics were not considered as empirical treatment. A normal binomial approximation was used to estimate confidence in- tervals (CIs) around this proportion and other proportion estimates.

The secondary objective was to identify clinical variables that were as- sociated with positive test results for GC and CT infection. The following Clinical predictors were evaluated: age, patient’s report of vaginal dis- charge, discharge noted during physical examination, history of STD, con- dom use, cervical motion tenderness, adnexal tenderness, and positive result for Trichomonas. We dichotomized age at 19 years (<= 18 vs >= 19), the age we perceived to be the end of parental supervision. We statistical- ly compared proportions using ?² or Fisher exact test, as appropriate. We used SAS 9.3 (Cary, NC) for statistical calculations.

Based on an anticipated 12% prevalence of positive test results, as de- termined in a previous study performed in the same city as the current study, we aimed for a sample size of 245 patients per hospital [4]. This number of patients would allow us to calculate the unnecessary treat- ment proportion and to detect differences of 20% between cases and controls (for variables with 20% prevalence in controls).

Results

Of the records reviewed, 262 of 274 at the tertiary hospital and 260 of 264 at the community hospital were eligible for data analysis, yield- ing a total of 522 medical records. Reasons for exclusion were absence of complete pelvic examination documentation (in 6 cases), missing data for the treatment decision (5 cases), missing data or unclear test re- sults (3 cases), treatment on presentation (1 case), and treatment for known partner exposure (1 case). The median age for the study group was 25 years (interquartile range, 21-33), with a range of 13 to 82 years. The patient characteristics are summarized in Table 1. Interrater agreement results were as follows: eligibility, 90% agreement, ? = 0.3; treatment for GC or chlamydia, 83% agreement, ? = 0.7; and test results, 100% agreement, ? = 1.0.

Thirty-two patients (6%) were positive for either GC or Chlamydia, including 3 who were positive for both. Antibiotic treatment was ad- ministered to 101 (19%) and declined by an additional 9 (2%). Of those receiving antibiotics, 87 of 101 (0.86; 95% CI, 0.77-0.92) had neg- ative test results (Table 2). Of those who were not offered antibiotics, 17 of 412 (0.04; 95% CI, 0.02-0.07) had positive test results. The overtreat- ment proportion was similar at the tertiary care hospital and the com- munity hospital (55/66 [0.83] and 32/35 [0.91], respectively).

Table 3 shows the associations between clinical variables and posi- tive test results. Of the 8 clinical variables analyzed in this study, only age less than 19 years was statistically associated with positive GC and CT results.

Discussion

Our results demonstrate significant overtreatment of GC and CT in- fections in inner-city ED populations. Six percent of our patients had positive culture results for GC or CT, but 21% were offered or given anti- biotic therapy. Our overtreatment proportion was 86% vs an undertreatment proportion of 4%. In contrast, in a review of studies pub- lished between 2001 and 2003, Jenkins and colleagues [6] calculated ED overtreatment proportions of 24% to 31% and undertreatment propor- tions of 43% to 84%. They also found that many studies reported lower rates of undertreatment after follow-up visits. Our higher rate of over- treatment and lower rate of undertreatment might reflect physicians’ assumption that our patients are unlikely to receive follow-up care.

Neisseria gonorrhoeae and CT infections are usually considered in pa- tients who present with lower abdominal, pelvic, and genitourinary complaints. One of the dilemmas in relying on the symptomatic

Table 2

Association between treatment decision and test results

Test result	Antibiotics declined	Antibiotics given	No antibiotics offered	Column total
Positive for GC or	1	14	17	32
CT infection
Negative for GC and	8	87	395	490
CT infection
Row total	9	101	412	522

C.E. Holley et al. / American Journal of Emergency Medicine 33 (2015) 1265–1268 1267

Table 3

Association between positive test results for gonorrhea or chlamydial infection and the presence of clinical predictors

Characteristic	Positive test results/those with characteristic (%)	Positive test results/those without characteristic (%)	Prevalence ratio (95% CI)	P
Age b19 y	8/53 (15%)	24/469 (5%)	3.0 (1.4-6.2)	.01
patient reports discharge	17/215 (8%)	9/200 (5%)	1.8 (0.8-3.9)	.2
Discharge seen on examination	14/161 (9%)	14/247 (6%)	1.5 (0.8-3.1)	.2
History of STD	6/91 (7%)	6/57 (11%)	0.6 (0.2-1.8)	.5
Condom use (always)	6/39 (15%)	12/162 (7%)	2.1 (0.8-5.2)	.1
Cervical motion tenderness	1/57 (2%)	22/359 (6%)	0.3 (0.04-2.1)	.3
Adnexal tenderness	3/89 (3%)	19/324 (6%)	0.6 (0.2-1.9)	.4
Trichomonas positive on wet prep	4/38 (11%)	28/475 (6%)	1.8 (0.7-4.8)	.3

presentation is that a great proportion of infected individuals are asymptomatic carriers. Up to 50% of cases of females diagnosed with GC and/or CT are asymptomatic or have only mild symptoms [7]. Five percent to 10% of males infected with GC and up to 90% of males infect- ed with CT are asymptomatic [7].

Currently, there is no protocol that guides ED providers on when to treat patients in whom GC or CT infection is being considered [8,9]. With high co-infection rates and concerns about complications from untreat- ed infections, providers often treat patients empirically for both STDs without having objective data (an approach confirmed by our observa- tions). Since 1937, when antibiotics were introduced, GC has developed multiple resistance mechanisms [7,10]. Neisseria gonorrhoeae‘s resis- tance to sulfonamides was first noted in the 1940s, to penicillins in the 1970s, to tetracyclines in the 1980s, and to fluoroquinolones in 2007 [11]. Surveillance projects in the United States and Great Britain have noted resistance of GC to cefixime, cefpoxodime, ceftriaxone, and azithromycine [12,13]. Within the last 5 years, the Centers for Disease Control and Prevention have changed their recommendations based on resistance patterns [14].

In addition to the large issue of resistance, administering Empirical antibiotics comes with risks and costs. One of these risks is a drug reac- tion. It can be minor (rash, drug fever) or major (eosinophilia, serum- sickness-like reactions, anaphylaxis, acute interstitial nephritis, drug- induced cytopenias). The frequency of skin reactions to cephalosporins has been reported as 1% to 3% [15]. A retrospective study of 350000 re- ports of Adverse drug reactions detected 6 deaths caused by the use of cephalosporins [16].

Moreover, as antibiotic use has increased, so has the incidence of community-acquired Clostridium Difficile (CDI). The overriding risk fac- tor for CDI is exposure to antibiotics [17]. A 2008 case control study found that 54.3% of patients with CDI had taken antibiotics within the preceding 90 days. Cephalosporins, our main antibiotic class for GC, were second only to clindamycin as posing the highest risk for CDI [18]. In our study, clinicians overtreated by a ratio of 21:1. This rate might seem excessive, but it could be reasonable for a patient population that tends to have unreliable access to follow-up appointments. In theory, after receiving positive test results, a health care provider would contact the patient and recommend follow-up treatment. However, this ideal sequence is rarely possible in a socioeconomically disadvantaged popu- lation. Many of the patients seen in the EDs where this study was con- ducted do not provide reliable contact information and do not have a primary care provider. These factors and the local high prevalence of GC and CT infection surely influence medical care providers’ decision- making process regarding empirical treatment. Future research on this topic might include a cost-effectiveness analysis of options such as

delaying therapy or using rapid testing.

The secondary objective of this study was to identify clinical vari- ables that predict infection with GC or CT. From the 8 variables that we analyzed (Table 2), only age less than 19 years proved to be statisti- cally significant. This finding is fairly consistent with surveillance data obtained by the Centers for Disease Control and Prevention from 2009 to 2012, which showed that the numbers of cases of gonorrhea and chlamydial infection were highest among patients aged 15 to 24 years [1]. Our study agrees with previous reports that noted an inability to

construct a prediction model that can be applied to determine which patients should receive empirical treatment against GC and CT infection [7-9].

Previous studies also agree that there is little or no reliable correla- tion among patients regarding the presentation of GC or CT infection and that physical examination findings are not necessarily reliable in identifying patients with these infections [8,19]. In 2007, Hack and Hecht [8] looked into the reasons emergency physicians decide to ad- minister empirical treatment for these infections. They included demo- graphics, historic factors, physical examination findings, and the results of other laboratory tests obtained during the ED visit. Physical examina- tion findings prompted treatment in 68% of cases, and the patient’s his- tory was the basis for 19.5% of those decisions. The authors did not seek clinical indications that correlated with either positive or negative cul- ture results. Berg and associates found that more than half of patients discharged from the ED with the final diagnosis of a urinary tract infec- tion were later found to have an STD [19].

Study limitations

Both of the hospitals that participated in this study treat patients predominantly from urban, lower socioeconomic neighborhoods; therefore, our results cannot be assumed to apply to the general US pop- ulation. Because the study was based on a retrospective review, we en- countered inconsistent reporting and many incomplete medical records. Some potentially important clinical variables, such as patients’ reported use of condoms, were missing from many medical records. In addition, in our ED, pelvic examinations can be performed and speci- mens collected by residents, attendings, nurse practitioners, and physi- cian assistants. The practitioner’s level of training and years of experience could have influenced the recording of clinical findings such as cervical motion tenderness.

Conclusion

The decision to give a patient an antibiotic when gonorrhea or chla- mydial infection is suspected remains a challenge for emergency physi- cians. No clear association of the patient’s history and the physical examination findings with the presence of infection has been suggested by the literature, but they are still the main factors that prompt treatment. In our study, the rate of overtreatment of GC and CT infection was 86%. The only clinical variable that predicted positive test results was being younger than 19 years.

Acknowledgments

The authors thank Zhaoxin Yang, Jessica Medani, and Stephanie Steinweg for abstracting information from patient records for this study. We thank Linda J. Kesselring, MS, ELS, for copyediting the manu- script. This study was supported by a grant from University of Maryland Emergency Network.

1268 C.E. Holley et al. / American Journal of Emergency Medicine 33 (2015) 1265–1268

References

AJEM