Vernacchio, Louis*; Vezina, Richard M.*; Mitchell, Allen A.*; Lesko, Samuel M.*; Plaut, Andrew G.†; Acheson, David W.K.‡
Persistent diarrhea (PD), defined by the World Health Organization as diarrhea lasting 14 days or longer, is an important worldwide pediatric disease (1). In the developing world, most PD is thought to be due to gastrointestinal infections, particularly with bacterial and parasitic pathogens; and it is associated with considerable morbidity and mortality (2,3). Little is known, however, about the incidence and microbiology of persistent diarrhea in US children.
To better define the characteristics of persistent pediatric diarrhea in the United States, we analyzed data from the National Pediatric Diarrhea Surveillance Study, a prospective cohort study of diarrheal illnesses in normal infants and children throughout the United States conducted in 2001-2002 (4).
MATERIALS AND METHODS
The study was performed in the Slone Center Office-based Research (SCOR) Network, a national research network of more than 400 pediatricians and family practitioners. A subset of network physicians representing various regions of the country and community types was recruited by means of a mail invitation. In the course of their routine office practices, these physicians in turn recruited a convenience sample of healthy children between the ages of 6 months and 3 years presenting for well-child visits or other visits unrelated to diarrhea and obtained informed consent from the subjects' parents (or guardians). Subject enrollment began in February 2001 and continued through June 2002. Exclusion criteria included a history of intestinal malabsorption, inflammatory bowel disease and/or cystic fibrosis; antibiotic use at the time of enrollment; lack of home telephone and inability of the parent to speak English.
As soon as possible after enrollment, an initial telephone interview was conducted with each parent to obtain baseline information on demographics and family/household factors, and 2 stool collection kits were sent to the home. The first kit was used to obtain a baseline stool specimen, and the second was for the first diarrhea specimen should diarrhea occur. Parents were asked to report all episodes of diarrhea that occurred during the 6-month follow-up time. Diarrhea was defined as "a change in stool habits with stools that are more frequent than usual or stools that are more watery than usual." For each episode, we conducted a telephone interview with the parent to obtain information on the characteristics of the episode and any changes in recent behaviors, and the parent was instructed to obtain a stool specimen for transport to the investigators. After its receipt, we immediately sent a replacement collection kit to the home for use during a subsequent diarrhea episode. Every 2 months during the study period, we called parents to ascertain any unreported cases of diarrhea; in the alternate months, we sent reminder postcards. If a previously unreported episode was discovered during a routine telephone call, we conducted the diarrhea interview and requested a stool specimen if the diarrhea was ongoing. Data on the incidence, clinical characteristics and microbiology are presented in this article; data regarding risk factors for diarrhea will be presented in another report.
The study was approved by the Boston University School of Medicine Institutional Review Board.
Parents collected stool specimens directly from the subject's diaper or from a clean plastic "hat" placed on the toilet into one Para-Pak Clean Vial and one Para-Pak C&S Vial (Meridian Biosciences, Cincinnati, OH) according to standard written instructions. Specimens were shipped by 2-day air to the microbiology testing laboratory. For polymerase chain reaction (PCR)/reverse transcriptase-PCR, DNA/RNA was extracted from specimens using NucliSens Isolation Kit (bioMerieux, Durham, NC).
All cultured bacteria were identified by standard clinical microbiology laboratory techniques. To culture Aeromonas sp., stool was plated onto blood agar and incubated at 37°C for 48 hours. For Campylobacter jejuni, stool was plated on CAMPY-BAP agar plates and incubated at 42°C for 48 hours in a microaerophilic environment; specimens were also tested for C. jejuni using the ProSpecT Campylobacter Microplate Assay (Alexon-Trend, Inc, Minneapolis, MN). Samples positive by either culture or immunoassay were considered positive. For Clostridium difficile, A and B toxins were detected by C. difficile TOX A/B II assay (Wampole Laboratories, Princeton, NJ). For Clostridium perfringens, toxin was detected by TechLab C. perfringens Enterotoxin test (TechLab, Blacksburg, VA). For Salmonella sp., stool was inoculated into enrichment broth and then plated onto Hektoen enteric agar and incubated at 37°C for 48 hours. For Shigella sp., stool was plated on MacConkey agar and XLD agar and incubated at 37°C for 48 hours. For Vibrio sp., stool was plated onto TCBS agar and incubated at 37°C for 48 hours. For Yersinia enterocolitica, stool was plated on CIN agar (40 g/mL cefsulodin) and incubated at 37°C for 48 hours. For Escherichia coli genotyping, stool was cultured on a MacConkey agar plate at 37°C overnight. After incubation, 10 lactose-positive colonies were pooled in a cryovial containing 800 μL of Luria-Bertani broth and 200 μL of 80% glycerol and frozen at −70°C. This composite was used to perform multiplex PCR testing as previously described (5) for the following genes: eae (enteropathogenic E. coli; EPEC), enteroaggregative (enteroaggregative E. coli; EAEC) and LT-1 and ST-1 enterotoxins (enterotoxigenic E. coli; ETEC). All EPEC were further tested for the presence of the EAF plasmid and the bfp gene to determine whether they were "typical EPEC" (eae, EAF plasmid, and bfp positive) or "atypical EPEC" (eae positive; EAF plasmid and bfp negative) (6). For enterohemorrhagic E. coli (EHEC), stool was first incubated in 5 mL of MacConkey broth overnight at 37°C; and the broth was then tested for the presence of Shiga toxins using Premier EHEC (Meridian Biosciences). Shiga toxin-producing E. coli were then isolated from any positive broths using an immunoblot technique previously described (7).
Stool from the formalin-fixed vial was assayed for Giardia lamblia and Cryptosporidium parvum using the ImmunoCard STAT! Crytosporidium/Giardia Rapid Assay (Meridian Biosciences).
Astrovirus, enteric adenovirus (types 40 and 41), norovirus, rotavirus and sapovirus were detected by PCR using primers obtained from the Centers for Disease Control and Prevention Division of Viral and Rickettsial Diseases. Specimens were also assayed for enteric adenovirus and rotavirus using the Premier Adenoclone Type 40/41 enzyme immune assay and the Premier Rotaclone enzyme immune assay (Meridian Biosciences). Samples that were positive for enteric adenovirus or rotavirus by either PCR or immunoassay were considered positive.
For incidence calculations, we used a general estimating equation approach with negative binomial distribution and clustering by subject; subjects lost to follow-up were assigned an end point that was the time of the last known successful contact. We compared characteristics of persistent versus acute episodes, viral versus nonviral episodes and the proportion of microorganisms found in baseline versus PD stool specimens by χ2 analysis or by Fisher exact test if any expected cell counts were fewer than five. Statistical analyses were performed with SPSS version 11.0 and SAS version 8.02 software packages.
Seventy-two practicing physicians from the Slone Center Office-based Research Network participated in the study and they enrolled 604 children from 34 states. The characteristics of study subjects are shown in Table 1. Follow-up at 2, 4 and 6 months was complete for 93.7%, 90.1% and 84.6% of subjects, respectively. A total of 611 episodes of diarrhea were reported during the study period, of which 50 (8.2%) lasted 14 days or longer. Forty-four subjects reported a single episode of persistent diarrhea during the follow-up period, and 3 subjects reported 2 distinct episodes. The overall incidence of persistent diarrhea was 0.18 episodes per person-year (ppy). The incidence was highest in the winter [December, January, February; 0.22 episodes ppy; 95% confidence interval (CI) 0.13-0.36] and spring (March, April, May; 0.22 episodes ppy; 95% CI 0.14-0.35) and lowest in the summer (June, July, August; 0.12 episodes ppy; 95% CI 0.06-0.23), but the seasonal differences were not statistically significant (Fig. 1).
The median duration of the PD episodes was 22.0 days, with a range of 14 to 64 days and 25th and 75th percentiles of 16.75 and 29.0 days, respectively (Fig. 2). Characteristics of the 50 PD episodes, associated symptoms, medication use and medical care received are shown in Table 2. Compared with acute diarrhea episodes, PD episodes were more likely to be associated with loss of appetite, cold symptoms, fever, vomiting and mucus in the stool; and there was more likely to be a household case of diarrhea preceding the index case. PD episodes were also significantly more likely to be treated with antibiotics or probiotics and to result in an outpatient medical visit.
Baseline stool specimens were obtained for 485 of 604 subjects (80.3%), and stool specimens were obtained during 40 of the 50 persistent diarrhea episodes (80.0%). For baseline specimens, the median time from collection of the specimen to receipt by the laboratory was 2.0 days (25th, 75th percentiles: 1.0, 3.0). For the persistent diarrhea specimens, the median time from the onset of diarrhea to collection of specimen was 2.0 days (25th, 75th percentiles: 0, 5.0) and the median time from collection of specimen to receipt by the laboratory was 1.5 days (25th, 75th percentiles: 1.0, 2.0).
Astrovirus, C. difficile, atypical EPEC, EAEC, norovirus, rotavirus and sapovirus were each detected in at least one persistent diarrhea stool sample. A comparison between the prevalence of each of the microorganisms assayed for in the baseline and diarrhea stool specimens is shown in Table 3. Of the 39 persistent diarrhea episodes for which sufficient stool was available for all microbiologic assays, 23 (59.0%) were negative for all studied pathogens. For the 3 studied pathogens that were associated with PD (norovirus, rotavirus and sapovirus), we calculated the proportion of all the infections identified in the study that lasted 14 days are longer, that is, progressed to PD. For norovirus, 4 of 8 cases became persistent (50.0%); for rotavirus, 4 of 23 cases (17.4%) and for sapovirus, 2 of 13 cases (15.4%).
To determine whether PD cases associated with a viral pathogen (n = 9) differed from nonviral cases (cases with all assays negative or positive for EAEC, atypical EPEC or C. difficile; n = 31), we compared characteristics of the two and found no statistically significant differences in terms of duration of diarrhea, presence of fever, vomiting, abdominal pain, blood in the stool, use of antibiotics in 10 days before onset of diarrhea, proportion with a household contact with diarrhea or proportion of cases resulting in an outpatient medical visit (data not shown). Viral episodes were more likely than nonviral episodes to involve loss of appetite (100.0% vs 72.0%; P = 0.04) and mucus in the stool (55.6% vs 30.0%; P = 0.02).
This study suggests that PD is experienced by approximately 1 in 5 young US children over the course of a year. It seems, however, that PD in the United States is a much more benign illness than it is in the developing world. In a wide variety of developing countries, studies have demonstrated that PD in young children is associated with considerable morbidity and mortality and accounts for approximately half of diarrhea-related deaths (3). There is also an important interplay between PD and malnutrition in the developing world-malnourished children are more likely to experience persistence of diarrhea episodes and PD can cause or exacerbate malnutrition (8). In contrast, PD episodes in our US cohort were quite benign. Only 1 episode in 4 resulted in an outpatient medical visit, and none led to hospitalization. Almost half of the episodes ended between days 14 and 21; although these episodes fit the formal definition of PD, many of them likely represent the tail end of the distribution of normal acute diarrhea in young children and probably have limited health significance. Such prolonged illnesses may, however, have significant social and economic impact on families in terms of lost days of day care, school and work.
Although all of the PD episodes ascertained in our study were followed to resolution and none led to a diagnosis of serious gastrointestinal disease during the study period, we do not have follow-up beyond our 6-month observation to know if any of our subjects subsequently experienced recurrent bouts of PD or manifested serious gastrointestinal disorders. It is likely that an infant or young child presenting with recurrent bouts of PD would undergo a diagnostic evaluation for conditions such as food allergies or intolerances, celiac disease and cystic fibrosis (9); further research to define the risk of serious gastrointestinal disease in a child presenting with PD would be useful.
We extensively assayed stool specimens from the PD episodes for a wide variety of potential bacterial, parasitic and viral pathogens. When compared with the prevalence of various microorganisms in baseline stool specimens, only rotavirus, norovirus and sapovirus were significantly associated with PD. These findings are in contrast to data from the developing world where viral pathogens have generally not been associated with PD in immunocompetent children (10,11); however, older studies may have lacked sensitivity to detect viral infections (12), and more recent studies from Brazil and Bangladesh suggest that rotavirus may be responsible for at least a small proportion of PD in those countries (8,13). An important remaining research question is what factors (eg, infectious load, host immune factors, micronutrient deficiencies etc) may cause some children to develop prolonged diarrhea when infected with viruses, such as rotavirus and the caliciviruses, which are usually associated with acute diarrhea (13,14).
C. difficile, enteroaggregative E. coli and atypical enteropathogenic E. coli were also each encountered in at least one PD stool specimen; but each was found with roughly the same frequency in baseline specimens and PD specimens. This finding suggests that, although these bacteria are often present in the stools of young children, they are probably not associated with PD in this population.
We detected no Aeromonas sp., C. perfringens, C. parvum, enteric adenovirus, enterohemorrhagic E. coli, enterotoxigenic E. coli, G. lamblia, Salmonella sp., Shigella sp., typical enteropathogenic E. coli, Y. enterocolitica or Vibrio sp. in any of our PD stool specimens. It is probably most surprising that we did not identify any cases of giardiasis or cryptosporidiosis in our study, as surveillance data suggest that these conditions occur with a fairly high frequency in young US children and both pathogens are known to cause prolonged diarrhea in some cases (15-17). However, with our sample size of 40 PD specimens assayed, the upper 95% confidence limit for the prevalence of each of the pathogens that we did not detect in any specimens is approximately 7.5% (18).
Despite testing for a wide variety of potential pathogens, we found none in 59.0% of the fully assayed PD specimens. If we exclude specimens where only C. difficile, enteroaggregative E. coli, or atypical enteropathogenic E. coli were found, which our data suggest are not related to PD in this population, 76.9% of the specimens were negative for a likely pathogen. This raises the question of what caused the PD in the assay-negative episodes. Some may be due to pathogens other than those for which we assayed. Another possibility is that they were due to noninfectious causes such as dietary disturbances, food allergies/intolerances or nutritional deficiencies. At least some of the episodes we identified might fall into the category of chronic nonspecific diarrhea of childhood (CNDC), also known as "toddler's diarrhea" or "irritable colon of infancy" (19,20). CNDC is a condition in which persistent diarrhea occurs without impairment of weight gain or growth and without evidence of gastrointestinal infection or other defined disorder, which would be consistent with the benign nature of the episodes found in our study. CNDC is thought to be due to disturbances in diet including an excessively low-fat, high-carbohydrate diet or excessive fluid intake overwhelming the absorptive capacities of the intestinal tract (21-23). Inasmuch as there is no specific diagnostic test for it, CNDC remains a diagnosis of exclusion and our data do not allow us to determine what proportion of our cases may have been due to CNDC.
Although this study provides new data on the occurrence of PD in infants and young children in the US and suggests that it is generally a much more benign disease than its counterpart in the developing world, many important questions about this entity remain. Larger and longer-duration studies will be needed to further define the microbiology of PD (especially in terms of pathogens that occur with low frequency) and to determine what proportion of PD cases ultimately lead to a diagnosis of serious gastrointestinal disease. Such data would be useful in guiding pediatric primary care providers and gastroenterologists in the evaluation of young children with PD and in counseling parents about the causes and likely outcomes of this condition.
The authors thank the following physician-members of the Slone Center Office-based Research Network who enrolled subjects for the study: Charles P. Anderson, MD (San Marcos, TX); Robert F. Anderson, MD (Bettendorf, IA); Russell C. Applegate, MD (Media, PA); David Armsby, MD (Quincy, MA); William Baier, MD (Lockport, NY); Jeffrey H. Baker, MD (State College, PA); Glen S. Bartlett, MD (Hershey, PA); Barbara P. Belcher, MD (San Antonio, TX); Prafulchandra U. Bhatt, MD (Lock Haven, PA); Susan Bollinger, MD (Westminster, MD); Judith A. Brown, MD (Barrington, IL); Alan Chen, MD (Dexter, MO); Amar L. Dave, MD (Ottawa, IL); Philip J. Dawson, MD (Richmond, VA); Jacques Days, MD (Columbia, SC); Archibald W. Dettman, MD (Corunna, MI); Mark DiDea, MD (Orlando, FL); Paul M. Douthitt, MD (Springfield, TN); Arthur S. Dover, MD (Freedom, CA); Patricia Edwards, MD (Concord, NH); Lawrence A. Elfman, MD (Madison, WI); Robert Farron, MD (Far Rockaway, NY); Eileen Fox, MD (Norman, OK); Lawrence M. Galtman, MD (Bethlehem, PA); Michael F. Grossberg, MD (Chambersburg, PA); Daniel E. Halm, MD (Bellevue, NE); Karen M. Hearty, MD (Springfield, MO); Karen C. Hester, MD (Blytheville, AR); C. David Hill, MD (Cullman, AL); David J. Holzsager, MD (Hampton, VA); Michael S. Hortner, MD (Northampton, PA); Hae M. Hwang, MD (Shelby, OH); Mirza G. Jesani, MD (Chicago, IL); Michael Jordan, MD (Newark, NY); F.S. Kapadia, MD (Waterford, MI); Chyung Kim, MD (Riverdale, GA); William E. Kobler, MD (Rockford, IL); Robert Leavitt, MD (Longmeadow, MA); Rhonda Levitt, MD (Seattle, WA); Young K. Lim, MD (Mt. Pleasant, PA); Tich-Hao Machm, MD (Phoenix, AZ); Jane A. Mack, MD (Bridgeport, CT); Carole Mangrem, MD (Clarksdale, MS); Jennifer Margolis, MD (Cincinnati, OH); Forough B. Mokhtari, MD (Indianapolis, IN); David M. Muhs, MD (Jamestown, ND); Sree K. Mulpuru, MD (Fairmont, WV); Edward G. Myers, MD (Warren, OH); Lawrence A. O'Brien, MD (Del Rio, TX); John M. Packard, MD (Guntersville, AL); Ishvarlal U. Patel, MD (New Brunswick, NJ); Jagdish Patel, MD (Bronx, NY); Isabel Rosenbloom, MD (Bridgeton, MO); Barbara H. Rumberger, MD (Marco Island, FL); Zdenka K. Safar, MD (Silver Creek, NY); James Satt, MD (Rocky Ford, CO); Marty Schmidt, MD (Fort Scott, KS); Bruce R. Schober, MD (Westbrook, ME); Suzanne W. Schuessler, MD (Lagrange, GA); Rita M. Seck, DO (Gladwin, MI); David D. Sova, DO (Grand Rapids, MI); Neil J. Stalker, MD (Peru, IN); Alan N. Swartz, MD (Miami, FL); Jeb S. Teichman, MD (Jeffersonville, IN); Jeffrey T. Van Gelderen, MD (Bay City, MI); Louis Vernacchio, MD (Boston, MA); Robert J. Whelpley, MD (Hornell, NY); Scott L. Williamson, MD (Wichita Falls, TX); Gerald G. Woodruff, Jr, MD (Anniston, AL); George Wortley, MD (Big Island, VA); Michael F. Yeiser, MD (Owensboro, KY) and Sabah E. Zara, MD (Trenton, MI).
© 2006 Lippincott Williams & Wilkins, Inc.