Survey of Physician Diagnostic and Treatment Practices for Patients with Acute Diarrhea in Guangdong Province,China

Introduction

S urveillance for laboratory-confirmed infections is an essential tool for monitoring the incidence of, detecting outbreaks of, and characterizing enteric infections that are commonly transmitted by food (Voetsch et al., 2004; Flint et al., 2005; Jain et al., 2009). Laboratory-based surveillance for enteric pathogens relies on many factors, one of the most important of which is having a diarrhea patient's stool specimen cultured for relevant pathogens. Patients with nontyphoidal Salmonella infection usually have diarrhea, and many also have accompanying symptoms such as bloody diarrhea and fever. Although diarrhea usually resolves spontaneously with a few complications, microbiologic testing for bacterial pathogens, such as Salmonella, Shigella, Escherichia coli O157, and Campylobacter, is indicated in patients who have both fever and diarrhea in the United States (Thielman and Guerrant, 2004).

China has an established, internet-based surveillance system that collects data about notifiable diseases (Wang et al., 2008). In this system, case reporting categories for food-related bacterial infections include cholera, dysentery, and “other infectious diarrhea,” and data collected for each of these categories are limited. Most cases reported in the “other infectious diarrhea” category are diagnosed based on symptoms without confirmation of the pathogen and data about pathogen subtype (e.g., Salmonella serotype). In addition, isolates obtained in clinical microbiology hospitals are not routinely sent to public health laboratories for subtyping, as there is no regulation mandating this.

In Guangdong, China, we recently initiated a laboratory-based surveillance system for nontyphoidal Salmonella to measure incidence of laboratory-confirmed infections and to detect dispersed outbreaks. Guangdong is a large province in south China, with an estimated population of more than 100 million permanent residents and migrant workers. Many hospitals in this province have the capacity to culture enteric bacteria from stool, but there are no data about the frequency with which patients with diarrhea actually have stool collected and cultured. Anecdotes suggest that many patients with acute diarrhea are treated with antibiotics and that very few undergo microbiologic testing. Understanding clinical practice patterns can help Guangdong health officials assess the sensitivity of laboratory-based Salmonella surveillance and the need to strengthen antibiotic prescribing guidelines, as has been done with similar surveys conducted in the United States (Hennessy et al., 2004; James et al., 2008). In Guangdong, we surveyed physicians from hospitals participating in a laboratory-based Salmonella surveillance system to understand diagnostic and treatment practices for patients with acute diarrhea.

Methods

In September 2009, a workshop was conducted in Guangdong Province to discuss the plan of initiating a laboratory-based surveillance for nontyphoidal Salmonella infections. Representatives from the infection control department from 22 general hospitals across 14 of 21 prefectures in Guangdong participated in the workshop. The participating hospitals were included in our survey as a convenience sample. As a background, the infection control department in Chinese hospitals serves as a bridge between clinical units in hospitals and local public health agencies. Each infection control department was asked to take 15 questionnaires back to their own hospitals, to administer the questionnaire to at least 15 physicians involved in outpatient care of patients with acute diarrhea, and to return the completed questionnaires to Guangdong Center for Disease Control and Prevention (see Appendix). No further instructions were provided to the infection control department about how to select physicians. Physicians were encouraged, but not required, at the beginning of the surveillance project, to send specimens from patients with fever and diarrhea to clinical laboratories for bacterial culture. Clinical laboratories received supplies and training to ensure that specimens were cultured using standard operating procedures and that Salmonella isolates were forwarded to public health laboratories.

The questionnaire asked physicians to report characteristics about their medical training and practice, about characteristics of the last outpatient with diarrhea whom they had evaluated, and about tests and medications that they had ordered for that patient. As a background, physicians in China can have different levels of training, including bachelor, master, and doctorate. Physicians were included if they routinely evaluate patients in the general medicine outpatient clinic, pediatric outpatient clinic, diarrhea clinic, fever clinic, surgery clinic, or emergency department. All questionnaires were self-completed by physicians. No personal identifiers were recorded. We defined diarrhea as having three or more stools in a 24-hour period that started in the last 7 days; this definition was explicitly written in the text of the questionnaire.

Data were analyzed by using SAS Version 9.1 (SAS Institute, Cary, NC). We used simple proportions to describe the background and common practices of physicians. We calculated the odds ratio (OR) and 95% confidence interval (CI) for factors associated with ordering a stool culture and for prescribing antibiotics. Multivariable analysis was performed by using backward, stepwise logistic regression. Variables included in multivariable analysis included those with p<0.20 in bivariate analysis and those that were plausibly related to a physician's decision to order a stool culture or prescribe antibiotics.

Results

Of 330 questionnaires distributed in the 22 hospitals, 237 (72%) were returned with full responses in November 2009; the response rate by hospital ranged from 27%–93%. Of the 237 physicians, 63 (27%) have master's or Ph.D.-level training. More than half (54%) have been practicing for more than 10 years. The current unit where the responder was working included pediatrics (29%), internal medical (26%), emergency medicine (17%), diarrhea and fever clinics (5%), and others (23%). Ninety-two (39%) reported routinely treating children, and 158 (67%) reported routinely treating adults. One hundred and thirty-four (57%) reported evaluating more than 20 outpatients per day.

A bacterial stool culture was ordered in the past month by 172 (73%) physicians, 145 (84%) of whom reported receiving the results of that culture. For the last outpatient with diarrhea whom they had evaluated, 134 (57%) reported ordering a stool culture, of which 103 (77%) reported receiving the results. The most common reasons for not ordering a stool culture included that it takes too long to receive results, that the patient is not willing to pay for the test, and that the patient's illness was too mild to warrant testing (Table 1). In multivariable analysis, evaluating more than 30 outpatients per month was the only physician characteristic independently associated with ordering a stool culture. Patient characteristics included having fever, having symptoms that had lasted more than 5 days, and taking antibiotics before seeing the physician (Table 2). Working in a pediatric clinic was associated with a lower frequency of stool culturing.

Most physicians prescribed at least one medication for the last patient with diarrhea whom they had evaluated. Of the 237 physicians surveyed, 153 (65%) prescribed antibiotics; 135 (57%), probiotics; 115 (49%), a gastric mucosal protective drug (e.g., H2-blocker, proton pump inhibitor); 117 (49%), oral rehydration salts; 58 (25%), anti-emetic; 72 (30%), antipyretic; and 103 (44%), vitamins. Additionally, 42 (18%) prescribed antibiotics, probiotics, and a gastric mucosal protective drug at the same time; 33 (14%) prescribed both antibiotics and probiotics; 29 (12%) prescribed both probiotics and a gastric mucosal protective drug; and 34 (14%) prescribed both antibiotics and a gastric mucosal protective drug. In multivariable analysis, patient characteristics associated with receiving an antibiotic prescription included having abdominal pain, taking antibiotics before seeing the physician, and being admitted to the hospital as a result of the visit. Physicians working in pediatric units were less likely to prescribe antibiotics (Table 3).

Discussion

In Guangdong, China, physicians working in hospitals that were a part of a Salmonella surveillance project reported high levels of ordering stool cultures and prescribing antibiotics to patients with diarrhea.

Obtaining stool specimens from patients with diarrhea can benefit both patient care and public health practice (Voetsch et al., 2004). In our survey, almost 60% of physicians reported ordering a stool culture for the last outpatient with diarrhea who had been evaluated. In contrast, similar surveys conducted in the United States have found a range from 36% to 48% (Hennessy et al., 2004; James et al., 2008). We were surprised to find such high reported rates in our survey population. Physician anecdotes and clinical microbiology laboratories suggest that only a small percentage of patients evaluated with diarrhea actually have cultures performed in Guangdong, as well as other regions of China.

It is possible that the high rate of stool culturing reported by physicians could be attributable to participation in our surveillance project, although the focus of our surveillance activities was on enhancing clinical microbiology procedures, and no incentives were provided to physicians to collect specimens. In addition to selection bias, other possible explanations include misclassification bias about what stool culture actually is. In China, physicians frequently order a white blood cell count of stool from patients with diarrhea. It is possible that some physicians mistook this test for stool culture. More likely, we suspect that physicians preferred to provide the “right” answer, rather than provide a correct assessment of what they are actually doing. In the analysis, taking antibiotics before seeing the physician was independently associated with ordering a stool culture. Although antibiotic use can reduce the likelihood of a stool culture being positive, physicians may have believed that such patients had more severe illness, thus necessitating further diagnostic tests. Our results also need to be interpreted in the context of a small sample size. It is possible that some situations which would have necessitated stool culture, for example, bloody diarrhea—were not significant, because too few physicians had encountered such patients recently.

Despite these limitations, we believe that our data can assist Guangdong health officials with increasing stool culturing rates in appropriate clinical cases. In turn, this can improve the sensitivity of surveillance for laboratory-confirmed enteric bacterial infections. Reasons that clinicians cited for not collecting a stool culture are opportunities to improve practices in appropriate situations, such as improving laboratory recovery rates and timeliness of reporting. Physicians reported that they did not order stool culture because of the long time needed to obtain results. Further investigation is needed to understand whether this is due to inherent delays associated with culturing bacteria from stool or to problems with recording and reporting. Obtaining the results of stool culture normally takes 2–3 days, whereas physicians in China are expected to diagnose and treat a patient with diarrhea right away. This is reflected in our finding that 23% of physicians responded that they do not order cultures, because the result will not change their treatment. We found that physicians working in pediatric clinics were less likely to order cultures, even though surveillance indicates that enteric pathogens, particularly Salmonella, are a common cause of illness in children (Kosek et al., 2003). Further study may be needed to evaluate what percentage of children in whom a nonbacterial cause of gastroenteritis is clinically diagnosed actually have enteric bacterial infections. This will help determine whether the lower rate of stool culturing found in pediatric clinic physicians is clinically appropriate.

In China, cost, particularly for diagnostic tests, is a major barrier to seeking and obtaining healthcare, although it is possible that ongoing health insurance reforms in China could reduce the financial disincentive to obtaining cultures (Yip and Hsiao, 2008). In the United States, professional societies recommend performing stool culture in patients with suspected enteric bacterial infection, such as those with fever and bloody stool, and algorithms that maximize yield and minimize cost have been validated (Chitkara et al., 1996; DuPont, 1997; Guerrant et al., 2001). Whether such approaches are effective and, from the perspective of payers and patients, cost effective in less economically developed countries is unknown.

We found high rates of medications being prescribed to patients presenting with diarrhea. Our survey did not collect sufficient data about the last patient evaluated with diarrhea to determine whether antibiotic prescribing was clinically indicated. Given that the isolation rates of enteric pathogens from stool submitted to the laboratory in these hospitals were generally less than 5% (data not shown) and that most enteric bacterial infections do not require antibiotic treatment, efforts to promote judicious antibiotic use are needed. Published studies in China have documented high rates of inappropriate antibiotic prescribing for upper respiratory infections, but none have previously evaluated enteric infections (Hui et al., 1997; O'Connor et al., 2001). Probiotics were the second most commonly prescribed agents. Probiotics are live microorganisms that, when consumed, may improve health; a recent Cochrane review has found that, in patients with acute infectious diarrhea, probiotics may reduce the duration and volume of diarrhea (Allen et al., 2010). A concerning finding was the high rate with which medications of dubious clinical value, including vitamins and gastric mucosal protective agents, were prescribed to patients. In China, physicians have strong financial incentives to prescribe medications, regardless of their clinical benefit (Reynolds and McKee, 2009; Yip et al., 2010). Further studies are needed in China to evaluate the appropriateness of medication prescribing, particularly for antibiotics, potentially through prospective enrollment and evaluation of patients presenting to the hospitals with acute diarrhea.

Although our study's conclusions are limited by several issues just noted, we believe that this survey has highlighted important areas for public health evaluation and intervention to improve detection and control of enteric bacterial infections in Guangdong, China.