EPIDEMIC--Paralytic Poisoning and Deaths in Felicidad On July 30, 120 cases of sudden neurologic symptoms and 26 deaths occurred in the village of "San Felipe del Mar" in the country "Felicidad." Paralytic shellfish poisoning, caused by toxic dinoflagellates (the "Red Tide"), was suspected. On request of the Felicidad Ministry of Health through the Pan American Health Organization, two epidemiologists from the Centers for Disease Control (CDC) traveled to Felicidad.  The CDC epidemiologists met with Ministry and local officials, and they and Felicidadian epidemiologists carried out several studies. A team of five local health professionals reviewed hospital records and death certificates and interviewed ill persons (cases) and household controls or their relatives. The interviews were conducted in Spanish, using a two-page typewritten questionnaire.  The CDC investigators had brought a portable computer and printer. These were set up in a motel room, and an Epi Info questionnaire was constructed to enter the data. One investigator read the data aloud while the other typed at the keyboard, and 126 cases were entered in about 6 hours. They then produced a listing of all variables in ANALYSIS, which they checked against the questionnaires to find errors. By the third day of investigation they had produced tables for a descriptive study and a case-control study. The file EPIDEMIC.REC contains the corrected records. The variable names have been changed to English to facilitate understanding by English-speaking readers. You can look at and edit the file using the ENTER program.  The investigators' first step, after "cleaning" the data, was to perform a descriptive study of the cases. The easy way to perform this analysis for all variables is: EPI>SELECT ILL = "Y" (This selects ill persons only) EPI>FREQ * (Displays a frequency distribution for each variable) A useful scheme for analyzing epidemic information is to divide items into Time, Place, and Person, or Time, Geography, and Everything Else. Time is conveniently shown on an epidemic curve, or graph of the course of the epidemic, and you may wish to graph several types of frequencies from the analysis of cases, including time of onset for cases who survived and for those who died. Calculating incubation period requires some mathematical gymnastics because the hours are expressed as 1200, 1830, etc., with the first two digits in hours and the third and fourth in minutes, with a base of 60 per hour. The necessary calculations are shown in the program called TIME.PGM, so  Place was explored initially, and it was found that all the cases in San Felipe del Mar clustered in a few households in the poorest part of town. Subsequent studies in San Felipe del Mar were limited to these households; therefore, Place does not figure prominently in the analysis. Cases were identified and studied in several other surrounding towns, but for the sake of simplicity, our considerations will be limited to the cases in San Felipe del Mar. The investigators also chose to focus their analysis on these cases while they were still in the field.  Person usually includes all the aspects of demography such as age, sex, and race, as well as exposure variables such as food history, medications, other illnesses, and behavioral risk factors. Many of these were explored with the help of a case-control study.  Case-Control Study For the case-control study, a more precise case definition was used, namely "A person in San Felipe del Mar who developed an acute illness on July 30 characterized by headache and at least two of the following sensory symptoms: (numbness of the lips, face, ears, fingers, toes, legs, arms) and at least two of the following motor symptoms (difficulty talking, difficulty walking, difficulty getting up from a lying or sitting position, difficulty breathing, dizziness)."  Programming ANALYSIS to identify the cases and controls by this definition is somewhat complex, and the investigators did not do so in the field. It is sometimes easier to make such decisions manually and enter the result in an extra variable like GROUP that has been included in the questionnaire, or that can be merged with it. As an illustration of techniques, however, we have included a program to identify the cases and controls in the file CASECON.PGM. To see the commands, print or display the file. To use it with the file EPIDEM1.REC, run ANALYSIS, and give the command RUN CASECON.PGM.  CASECON.PGM does the following: *READs the file EPIDEM1.REC *DEFINEs variables for GROUP, NEURO symptoms, and MOTOR symptoms *SETS NEURO and MOTOR counts to zero *Using IF statements, puts the count of NEURO and MOTOR symptoms in these variables *Sets GROUP to blank *IF illness, onset on July 30, headache, two or more NEURO symptoms and two or more MOTOR symptoms are all present, GROUP is set to "CASE" *For noncases who are not ILL, GROUP is set to "CONTROL" *SELECT is used to limit processing to CASES and CONTROLS *SET is used to ignore records with missing values *A page HEADER provides title information *Variables ANYCLAMS, ANYFISH, ANYSHRIMP, and ANYLOBSTER are set to "+" if either the food or soup made from the food (or both) were eaten *A message is displayed on the screen to instruct the user  Alternatives to Programming The investigators avoided many of the programming steps we have included in the .PGM files by using the FREQ commands combined with use of the delete feature in ENTER. To perform an analysis, they deleted unwanted records temporarily and then did the analysis. This can be done by using several copies of the data file with different names. They then combined the results from cases and controls, using STATCALC to do the statistics. Although the method was laborious, it was far less so than hand tabulation and also faster and more accurate. Elegance in programming should always take second place to getting accurate results in a field investigation.  General Results The investigators concluded that consumption of clams and clam soup was responsible for the epidemic and for the 26 deaths. Bioassay and chemical analysis of clams showed high levels of saxitoxin, the chemical basis of the toxicity of shellfish during periods of "red tide." The investigating team made recommendations 1) for the monitoring of clams for saxitoxin and 2) for health education, since (despite the odds ratios) many residents did not realize the danger of eating clams during periods of "red tide." They also recommended a system for warning citizens during danger periods and provided ideas for further research, including exploration of the relationship of clam beds to agricultural runoff and to a nearby shrimp farm.  A preliminary report--containing key tables, hand-drawn maps, the questionnaire, a summary of methods, results of the descriptive and case-control studies, a summary of cases outside San Felipe del Mar, conclusions, and recommendations- -was typed on the portable computer, discussed verbally with CDC headquarters staff, and submitted to the Ministry of Health and to the Pan American Health Organization before the investigators left Felicidad, approximately 2 weeks after they had arrived. Further analysis was conducted after their return to Atlanta. For this purpose, the EXPORT program was used to produce a SAS file for analysis on the CDC mainframe, where logistic regression and other tests could be performed.