|
|
| |||||||||||||||||||||||||||||||
Background In the United States many correctional facilities now operate at far over capacity, with the potential for living conditions that permit outbreaks of respiratory infections. We investigated an outbreak that was identified in an overcrowded Houston jail after two inmates died of pneumococcal sepsis on the same day. Outbreaks of pneumococcal disease have been rare in the era of antibiotics.
Methods We assessed risk factors for pneumococcal disease in both a case-control and a cohort study. Ventilation was evaluated by measuring carbon dioxide levels and air flow to the living areas of the jail. The extent of asymptomatic infection was determined by culturing pharyngeal specimens from a random sample of inmates. Type-specific immunity was determined with an enzyme immunoassay.
Results Over a four-week period, 46 inmates had either acute pneumonia or invasive pneumococcal disease due to Streptococcus pneumoniae serotype 12F. The jail's capacity had been set at 3500 inmates, but it housed 6700 at the time of the outbreak; the inmates had a median living area of only 34 ft2 (3.2 m2) (interquartile range, 28 to 56 ft2 [2.6 to 5.2 m2]) per person. There were significantly fewer cases of disease among inmates with 80 ft2 (7.4 m2) per person or more (P = 0.030). Carbon dioxide levels ranged from 1100 to 2500 ppm (acceptable, <1000), and the ventilation system delivered a median of only 6.1 ft3 of outside air per minute per person (interquartile range, 4.4 to 8.5 ft3; recommended,
Conclusions Severe overcrowding, inadequate ventilation, and altered host susceptibility all contributed to this outbreak of pneumococcal disease in a large urban jail. .
20 ft3). The attack rate was highest among inmates in cells with the highest carbon dioxide levels and the lowest volume of outside air delivered by the ventilation system (relative risk, 1.94; 95 percent confidence interval, 1.08 to 3.48). Of underlying medical conditions, intravenous drug use was most strongly associated with disease (odds ratio, 4.50). The epidemic strain (serotype 12F) was cultured from 7 percent of the asymptomatic inmates. Of 11 case patients tested with the enzyme immunoassay, 9 (82 percent) lacked preexisting immunity to this strain.
We describe our investigation of an unusual epidemic of pneumococcal disease at a crowded urban jail. The outbreak was first recognized when fatal pneumococcal sepsis developed in two inmates on the same day. Within a four-week period, pneumonia or invasive pneumococcal disease was diagnosed in 46 inmates. All isolates from culture-confirmed cases had the same serotype (Streptococcus pneumoniae type 12F). Outbreaks of serotype-specific pneumococcal disease have been rare in the era of antibiotics.
In this investigation we were able to examine the relation of crowding, ventilation, and other factors to the risk of epidemic pneumococcal disease in an institutional setting. The outbreak has important implications for health care and standards for air quality in correctional facilities.
Methods
Characteristics of the Jail
In contrast to prisons, which are designed for long-term incarceration, jails are designed for pre-sentencing detention averaging less than two months. The jail building involved in this outbreak had the highest inmate population of any jail in the United States. The 13-story building had a designated capacity of 3500 inmates but was housing 6700 when the outbreak occurred. Inmates were housed on 10 floors (floors 3 to 12), with each floor divided into 20 to 22 secured areas called cell blocks. Each of the 212 cell blocks in the jail was designed to hold 8 to 24 inmates in a floor space of 1200 to 1700 ft2 (110 to 160 m2) (ceiling height, 12 ft [3.6 m]).
Each cell block consisted of a day room and adjoining sleeping quarters. There were three types of cell blocks, distinguished by their sleeping quarters: single cells (style A), four-person cells (style B), or an open dormitory (style C). Because of severe overcrowding, as many as half the inmates slept on mattresses on the floor throughout the cell blocks, including the day rooms. Except in 13 high-security cell blocks that housed a total of only 119 inmates, living was communal, and doors between sleeping areas and day rooms were never secured.
On floors 3 to 11, inmates were confined to their cell blocks except to attend court or activities (such as those in a recreational area, classroom, or library). Each inmate was limited to two hours of activities a week. The 12th floor housed inmates who had work assignments outside their cell blocks. Each day, 500 to 700 inmates from throughout the jail were moved through underground tunnels to crowded holding cells for trial in an adjacent court building.
A clinic and inpatient infirmary served inmates 24 hours a day. Facilities for chest radiology were available at the clinic. Patients who needed hospitalization were transferred to the county hospital. The jail had a staff of 950 members, who worked in offices on floor 2 or below it or as guards on floors 3 to 12.
Case Definition and Case Finding
For this investigation, a case of pneumococcal disease was defined as an invasive infection confirmed by culture of a specimen from a normally sterile site (blood, pleural fluid, or spinal fluid) or as acute pneumonia confirmed radiographically, and detected between September 6 and October 2, 1989 (outbreak period), in any inmate. Prospective surveillance was instituted on September 11. The clinical and radiologic records of the jail clinic were reviewed retrospectively to define the beginning of the outbreak. The medical records of inmates hospitalized for pneumonia were reviewed. Paired serum samples were obtained two to three weeks apart from case patients who remained incarcerated during this period.
Investigation of the Outbreak
Attack rates of disease were compared according to the inmates' floor, type of cell block, and degree of crowding. The inmate census of each cell block was updated daily, and conditions of crowding were measured by calculating the floor space per inmate from architectural records.
The first 25 case patients who were hospitalized or identified by prospective surveillance were enrolled in a case-control study. For each case patient, three inmates who lived in the same cell block were randomly selected as controls. A standardized questionnaire regarding underlying diseases and activities was administered. Preexisting pneumococcal antibodies were measured in acute-phase serum from a sample of culture-positive case patients and their matched controls. IgG antibodies to S. pneumoniae type 12F capsular polysaccharide were measured by enzyme immunoassay after antibody to cell-wall polysaccharide was removed by absorption, as previously described9.
To evaluate the ventilation system, carbon dioxide levels in a random sample of 31 cell blocks were measured with detector tubes (Dragerwerk, Lubeck, Germany), and a systematic study of the air flow to 49 cell blocks was undertaken10,11. Air flow in the direct center of each duct was measured through drill holes by use of a thermal anemometer calibrated with a manometer.
To determine the extent of asymptomatic carriage of S. pneumoniae, pharyngeal cultures were obtained from a random sample of inmates before any intervention in the outbreak and from inmates present at the end of the outbreak who had been incarcerated since the epidemic began. Specimens were also obtained from security guards and newly admitted inmates. Persons who had received antibiotics during the preceding two weeks were excluded. Dacron swabs of the posterior pharynx and tonsils were inoculated onto trypticase soy agar with 5 percent sheep's blood and gentamicin and incubated overnight at 35 °C in carbon dioxide. Pneumococci were identified according to standard methods, and their serotypes were established by precipitin and capsular-swelling tests12.
To determine whether respiratory viruses were cofactors in pneumococcal disease, swabs of nasopharyngeal secretions obtained from patients treated after September 11 were inoculated into primary rhesus kidney cells, human-lung-fibroblast cell lines WI-38 and MCR-5, and human foreskin tissue; the cultures were monitored for cytopathic effects for 21 days. Paired serum samples from case patients were tested for IgG antibodies to respiratory syncytial virus, adenovirus, and parainfluenza viruses 1 and 3 by indirect enzyme immunoassay; for IgM antibodies to enteroviruses by an adaptation of a previously described enzyme immunoassay13; and for antibodies to mycoplasma species by complement fixation14.
Vaccination and Prophylactic Antibiotics
Shortly after beginning the investigation, we recommended that all staff members and inmates receive the 23-valent pneumococcal vaccine, which includes serotype 12F15. To provide additional protection until immunity could be elicited, inmates with risk factors for infection with the human immunodeficiency virus (HIV) or pneumococcal disease15 were given oral penicillin or erythromycin for one week.
Statistical Analysis
Analysis was performed with the Epi-Info program (version 5.01)16. P values were calculated with the chi-square test with Yates' correction or with a two-tailed Fisher's exact test (if an expected cell value was less than 5). Matched odds ratios were calculated with the Mantel-Haenszel chi-square test. Continuous variables were compared by the Wilcoxon two-sample test. Unconditional logistic regression was performed with the Logistic program (version 3.02)17.
Results
Descriptive Epidemiology of the Outbreak
Between September 6 and October 2, 1989, 46 inmates met the case definition for pneumococcal disease, including 12 with documented bacteremia (2 of whom had meningitis) (Figure 1). No cases were detected between August 1 and September 5, and no pneumococcal disease developed in any staff member. The demographic characteristics and underlying medical conditions of the 46 case patients are shown in Table 1. Twenty-five of these patients (54 percent) were hospitalized, and the remainder were treated empirically for pneumococcal disease in the jail infirmary. Two patients, both asplenic, died. S. pneumoniae was isolated from 35 percent of the patients whose blood was cultured and 40 percent of those whose sputum was cultured. All isolates from 17 culture-positive patients were serotype 12F and were susceptible to penicillin. Patients who had pneumonia and negative cultures (or for whom no cultures were performed) had clinical findings similar to those who had positive cultures and no other cause of their infections. Only two patients (both with a history of alcohol abuse) reported having received pneumococcal vaccine before their incarceration.
|
|
Of 25 patients enrolled in the case-control study, 16 had culture-confirmed disease and 20 were hospitalized. The 25 case patients were similar to their 75 matched controls in mean age (30 vs. 29 years), sex (92 percent of each group were men), race, and duration of incarceration (median, 49 vs. 44 days). Participation in activities outside cell blocks (in the recreational area, classroom, or library or at court) were not associated with disease. A history of intravenous drug use was significantly associated with disease (48 percent of case patients vs. 20 percent of controls; matched odds ratio, 4.50; 95 percent confidence interval, 1.52 to 13.32), and other underlying conditions that tended to be associated with disease included chronic alcohol use (42 percent vs. 28 percent; odds ratio, 1.82; 95 percent confidence interval, 0.65 to 5.05), asplenia (12 percent vs. 1 percent; odds ratio, 9.00; 95 percent confidence interval, 0.94 to 86.53), and cirrhosis (9 percent vs. 0 percent; odds ratio undefined; P = 0.051). Current smoking was not associated with disease (80 percent vs. 71 percent). Acute-phase serum samples were available for 11 of the patients with positive cultures (including 8 with bacteremia) and 21 matched controls. Of these 11 patients, 9 (82 percent) had no detectable antibodies to S. pneumoniae serotype 12F, as compared with 12 (57 percent) of 21 controls (P not significant).
Cohort Study
Cases of pneumococcal disease occurred among inmates on all floors of the jail except floor 12, with the highest attack rates on two floors where the first cases were detected (floors 7 and 8). Inmates assigned to private rooms (inmates needing high-security confinement) or the 12th floor (inmates working outside their cell block) had no disease and lived in much less crowded conditions than inmates in shared quarters on floors 3 to 11 (Table 2).
|
7.4 m2) per person (8 percent of the inmate population) had significantly fewer cases of disease than all other inmates (attack rate, 0 vs. 4.7 per 1000; P = 0.030). Among inmates housed in shared quarters on floors 3 to 11, the median living area available for each inmate was 33 ft2 (3.1 m2) (interquartile range, 27 to 45 ft2 [2.5 to 4.2 m2]). Crowding was most severe in style B and C cell blocks (with four-person and dormitory sleeping quarters) (Table 2). Although style C cell blocks were the most crowded, the attack rate of disease was significantly higher among inmates housed in style B cell blocks (Table 2). The association of disease with style B cell blocks remained significant after underlying conditions, floor (floors 7 and 8 vs. other floors), and level of crowding were controlled for by logistic regression (odds ratio, 2.02; 95 percent confidence interval, 1.07 to 3.82).
Environmental Studies
The jail was ventilated by a constant-volume recirculated-air system for central heating, ventilating, and air conditioning that was operating at maximal capacity. Approximately 20 percent of the total volume of air delivered was outdoor air and 80 percent was recirculated air. Air-handling units on the 13th floor were connected with small recirculation systems serving each cell block. The recirculation systems were designed to work most efficiently when the doors between day rooms and sleeping quarters were closed; however, because of the severe overcrowding, these doors routinely remained open.
Carbon dioxide levels in 31 cell blocks ranged from 1100 to 2500 ppm (acceptable level, <1000),18 with the highest levels recorded in style B cell blocks (Table 3). Temperature and humidity levels were also slightly higher in style B cell blocks than style A and C cell blocks (mean temperature, 80 vs. 79 °F; P = 0.044; humidity, 54 percent vs. 51 percent; P = 0.028).
|
In the 49 cell blocks where air flow was measured directly, there was an overall reduction of 27 percent in the expected volume of air delivered. The median volume of outside air delivered to each cell block was 199 ft3 per minute (interquartile range, 154 to 272 ft3) and corresponded to a median of 6.1 ft3 per minute per person (interquartile range, 4.4 to 8.5 ft3), depending on the level of crowding. Style B cell blocks had the worst combination of crowding and poor ventilation, with a significantly lower volume of outside air per inmate than was delivered to either of the two other types of cell block (Table 3). There were no significant differences between the floors in the level of ventilation.
Other Studies
During the peak of the outbreak, 11 (7 percent) of 152 inmates were found to be carriers of the epidemic serotype 12F, as compared with 1 (1 percent) of 105 other inmates whose cultures were examined after the interventions (Table 4). Nasopharyngeal secretions were obtained from 11 case patients for viral culture. A virus was isolated from only one (9 percent) of these patients (rhinovirus species). Of 21 case patients for whom paired serum samples were available, only 4 had evidence of recent viral infection (2 patients had increases in titers of IgG antibodies to adenovirus, and 2 had high titers of IgM antibodies to enterovirus); none had evidence of mycoplasma infection.
|
Between September 13 and September 21, 1989, 79 percent of inmates accepted pneumococcal vaccine, and 41 percent received antibiotic therapy for underlying conditions. In 1990 and 1991, crowding was reduced by opening two new jails and by distributing inmates more evenly between jail and prison facilities. Other measures to control respiratory infections in the jail included improving ventilation, prohibiting smoking, and establishing an isolation ward for inmates with respiratory disease and an infection-control program to detect clusters of cases of communicable diseases. No further cases of pneumococcal disease acquired within the institution have been confirmed since October 2, 1989.
Discussion
In the pre-antibiotic era, epidemics of pneumococcal disease were well documented in communities19 and in other settings where people lived in crowded conditions, such as psychiatric hospitals, military training centers, and correctional facilities20,21,22,23. With improved living conditions and the availability of antibiotics, outbreaks of this disease have been rare. Although S. pneumoniae remains the leading cause of community-acquired pneumonia, almost all cases are sporadic.
The outbreak described here emerged rapidly, with disease occurring throughout the jail and evidence of widespread asymptomatic carriage of the 12F serotype. The carriage rate of 7 percent (95 percent confidence interval, 4 percent to 13 percent) was comparable to rates documented during serotype-specific outbreaks in the pre-antibiotic era20,21. Currently, serotype 12F isolates account for only 1 to 2 percent of clinical isolates submitted to the Centers for Disease Control and Prevention (CDC)24,25 and are very rarely detected in asymptomatic persons26,27.
Our investigation suggested that a combination of factors was responsible for the outbreak, including host susceptibility, crowding, and ventilation that was inadequate for the number of inmates in the building. We found no evidence that coinfection with a respiratory virus or mycoplasma contributed. S. pneumoniae is generally transmitted from person to person by the respiratory route28; the association of disease with crowding and poor ventilation in the outbreak described here supports this route of transmission. The movement of large numbers of inmates to court allowed contact between persons from different areas of the jail.
Inmates of U.S. jails generally come from a lower socioeconomic background than others in the population, have limited access to primary health care, and are at high risk for underlying conditions known to predispose people to pneumococcal disease1,15,29. The association of disease with intravenous drug use in this study may have been related to other underlying conditions or to HIV infection. However, HIV infection alone was unlikely to have been a major factor in this outbreak since 84 percent of the case patients who were tested were seronegative for HIV. Inmates may have been immunologically susceptible to the epidemic serotype of S. pneumoniae. Over 80 percent of case patients who were tested had no measurable preexisting antibodies to this strain.
The living conditions in this jail exemplify a national problem in correctional facilities1,29. Although overcrowding is widespread in jails and prisons, detailed descriptions of the space available per inmate are lacking. We documented the fact that over 98 percent of inmates lived in shared quarters with a median living space per inmate of less than 6 ft (1.8 m) by 6 ft.
Adequacy of ventilation is generally measured by determining the volume of outside air per occupant and is inversely proportional to the number of occupants in a building11,18. In the jail, extremely high carbon dioxide levels served as a marker for insufficient ventilation, and a low supply of outside air per person was documented. Cell blocks with the worst combination of crowding and poor ventilation had the highest rates of disease. In 1981, the minimal standard for the outdoor-air supply was raised from 5 to 10 ft3 per minute per person30. The jail's ventilation system, built in 1979, met the latter standard for the population capacity. However, after reports of symptoms attributed to the quality of indoor air (sick building syndrome), the guidelines of the American Society of Heating, Refrigeration, and Air Conditioning Engineers (ASHRAE) were revised in 1989: the minimal standard for outdoor-air supply is now 20 ft3 per minute per person in office buildings and correctional facilities and 60 ft3 in smoking areas18. (Smoking was permitted throughout the jail.) These new standards recommended by ASHRAE far exceeded the capability of the ventilation system in the jail, given the degree of overcrowding.
Despite acceptance of ASHRAE standards, the guidelines are largely empirical and are based in part on studies of symptoms reported by workers in office buildings31,32,33,34. It is difficult to correlate symptoms with the level of ventilation in buildings where occupants are distributed uniformly. In one of the few controlled studies, there was no correlation between the prevalence of symptoms and the outdoor-air supply when the supply volume was varied between 20 and 50 ft3 per minute per person35. Similar studies have not been reported in which investigators assessed lower volumes of air, like those observed in our study.
In the jail building, the variation in the degree of crowding permitted the outside-air supply per person to be correlated with disease rates. Our findings emphasize the importance of designing ventilation systems for correctional facilities that anticipate the possibility of severe overcrowding. The current ASHRAE standard assumes that inmates will have no less than 50 ft2 (4.6 m2) of space,18 although we documented that some inmates had as little as 20 ft2 (1.9 m2) per person. The recent increase in multidrug-resistant Mycobacterium tuberculosis and the well-documented association of the transmission of tuberculosis with incarceration2,3,4,5,36 also indicate the need to reassess standards for indoor-air quality and other preventive measures in correctional facilities.
It is not clear whether vaccination and prophylaxis with antibiotics helped end the epidemic, since it had already peaked by the time these measures were initiated. However, the efficacy of pneumococcal vaccine has been demonstrated in numerous studies,15,37,38 and medical screening during incarceration represents an opportunity for immunization.
In a preliminary announcement of this outbreak, the CDC requested that any clusters of cases of pneumococcal infection be reported39. Although no additional outbreaks have been identified in correctional facilities, several epidemics in other settings have been investigated. These include large outbreaks in military training camps40,41 and clusters in a variety of institutional settings (homeless shelters, nursing homes, and day care facilities)42,43,44. Until 1989, reports of serotype-specific outbreaks in the antibiotic era had been limited primarily to small nosocomial clusters45,46,47. The recent reemergence of outbreaks highlights the epidemic potential of S. pneumoniae. This potential, together with the development of multidrug-resistant strains,48 raises questions about whether isolation in a respiratory ward should be recommended in certain circumstances and increases the need to vaccinate institutionalized persons at risk for pneumococcal disease15,49.
We are indebted to numerous persons who greatly contributed to this investigation: at the Texas Department of Health -- Dr. Diane Simpson and Marcia Roberts; Harris County Health Department -- Dr. Thomas Hyslop, Mark Canfield, Linda Forys, Lucinda Kilborn, Catherine Cooksley, Pat Buzbee, Anna Poulin, Sally Stalter, David M. Varela, and Rita Martin; Houston City Health Department -- Dr. John E. Arradondo, Janeene Pappas, Dr. Kathleen H. Sullivan, Ann Schachtel, Virginia L. Flannery, Ann Doggett, Robert Nackman, Ernest Hatton, Harriett Hadnott, Regina Henderson, Natalee L. Shelton, Anna Maller, and Christine Redford; Harris County Hospital District -- Lois J. Moore, Alicia Reyes, and Yolonda Gloria; Harris County Jail -- Johnny Klevenhagen, Dr. Mark Kellar, Michael Quinn, Don McWilliams, Candy Henderson, Lanny Chopin, Dr. David Marrack, Dr. Zae Zeon, Cheryl A. Warren, Suzanne Mark, Sally Storms, George Adams, Corporals Allen and Anderson, Martin L. Elder, Deborah Sorola, George Lee, Susan Cathey, and Michele Evans; Ben Taub Hospital -- Dr. Steven Greenberg, Uma Rajguri, Elizabeth Jones, and M.R. Bullard; Respiratory Diseases Branch, CDC -- Dr. Benjamin Schwartz, Brian D. Plikaytis, and Sheila Collin; and the Respiratory and Enterovirus Branch, CDC -- Dr. Mark A. Pallansch and Dr. Dean D. Erdman.
Source Information
From the Respiratory Diseases Branch, Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta (C.W.H., M.R.R., T.D.M., J.A.E., R.R.F., R.F.B.); the Veterans Affairs Medical Center and Baylor College of Medicine (E.A.D., D.M.M.) and the Harris County Health Department (J.C.B.), Houston; and the Texas Department of Health, Austin (K.A.H.). A preliminary report of the outbreak appeared elsewhere. (MMWR Morb Mortal Wkly Rep 1989;38:733-4.)
Address reprint requests to Dr. Breiman at the Childhood and Respiratory Diseases Branch (Mailstop C09), DBMD, NCID, Centers for Disease Control and Prevention, Atlanta, GA 30333.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
