Suspected deaths due to dengue with onset of illness in 2007 were identified from three sources: 1) the passive dengue surveillance system (PDSS) maintained by the Puerto Rico Department of Health (PRDH) and Centers for Disease Control and Prevention (CDC) Dengue Branch, 2) death certificates filed at the Demographic Registry of Puerto Rico, and 3) hospital-based infection control nurse dengue surveillance system (ICNDSS) as previously described [4], [6]. Complete medical records from all hospitalizations, emergency room and clinic visits for suspected dengue deaths were obtained and reviewed by physician investigators using a standardized instrument to collect demographic, clinical, and laboratory data. This project underwent institutional review at the CDC and it was determined not to be subject to formal institutional review board review requirements as defined by US regulations (i.e., Title 45 Code of Federal Regulations Part 46).

Suspected dengue cases are reported to PDSS by health care providers who submit a serum specimen for diagnostic testing accompanied by a Dengue Case Investigation Report (DCIR). For fatal cases, autopsy tissue is occasionally sent with the serum specimen and DCIR. The DCIR includes patient demographic, clinical, travel, vaccination, and disease outcome data including whether the illness resulted in hospitalization, death or both. DCIR data are entered into an electronic data base. Deaths reported to PDSS are immediately confirmed by calling the reporting hospital.

Hospitalized suspected dengue cases are reported to the ICNDSS by nurse epidemiologists or infection control nurses. Because of a steady decline in participation, data from this reporting system are only used to augment PDSS data. When an ICNDSS case report indicated a patient death these cases were confirmed and investigated.

Death certificates with “dengue” included as a cause or contributing factor in the death were obtained on a monthly basis. The PDSS database was queried to determine if a diagnostic specimen was received for the case.

All serum specimens were tested by DENV serotype-specific, real-time, reverse transcriptase polymerase chain reaction (RT-PCR) [12], [13]. Specimens were also tested for anti-DENV IgM with an IgM antibody-capture enzyme-linked immunosorbent assay (MAC-ELISA) [14]. Specimens with borderline results were retested against a standard negative serum. A quantitative IgG ELISA was performed to detect anti-DENV IgG in all specimens [15]. As human West Nile virus (WNV) infections were identified for the first time in Puerto Rico in early 2007 (i.e., among healthy blood donors), [16] all fatal cases were tested by anti-WNV MAC-ELISA and if positive, WNV specific RT-PCR [17] and plaque reduction neutralization test (PRNT90) assays were performed to differentiate between DENV and WNV infections [18].

All serum specimens with sufficient volume remaining after dengue diagnostic testing was completed were also screened for IgM antibodies to Leptospira at the CDC Bacterial Zoonoses Branch, Zoonoses and Select Agent Laboratory using the rapid dipstick ELISA ImmunoDOT kit (GenBio, Inc., San Diego, CA). Specimens with positive or borderline ELISA results were further tested using the microscopic agglutination test (MAT) using 20 Leptospira reference antigens representing 17 serogroups [19].

Autopsy tissue was sent to CDC Infectious Diseases Pathology Branch for identification of DENV antigen by immunohistochemical (IHC) microscopy and DENV-specific RT-PCR. If microscopic examination of the tissue or the clinical history was suggestive of leptospirosis (e.g., interstitial nephritis, pulmonary hemorrhage), IHC microscopy was conducted using 16 polyclonal anti-Leptospira spp. antibodies. Depending on the clinical presentation and histopathology of the tissue specimens submitted, additional testing with histochemical stains, immunohistochemistry, and/or additional molecular assays was performed to determine the infecting pathogen.

A suspected dengue case is a dengue-like, acute febrile illness in a person with that clinical diagnosis and a specimen submitted for dengue diagnostic testing. A laboratory-positive dengue case is a suspected dengue case with any of the following: (1) detection of DENV RNA in serum, cerebrospinal fluid, or tissue by RT-PCR; (2) identification of DENV antigen in tissue by IHC assay; (3) IgM anti-DENV seroconversion or demonstration of a fourfold or greater increase in anti-DENV IgG titers in paired serum specimens; or (4) positive anti-DENV IgM in a single serum specimen. A laboratory-negative dengue case is a suspected dengue case for which no anti-DENV IgM is detected in a serum specimen collected >5 days after fever onset (i.e., a negative convalescent phase specimen), and no anti-DENV IgM, DENV RNA or DENV antigen is detected from serum collected ≤5 days after onset (i.e. a negative acute phase specimen) or tissue (if submitted). Laboratory-negative dengue cases are hence considered to have the diagnosis of dengue ‘ruled out’. A laboratory-indeterminate dengue case is a suspected dengue case that subsequently had no DENV RNA or anti-DENV IgM detected in specimen collected ≤5 days after fever onset, and no convalescent specimen submitted for diagnostic testing.

A laboratory-confirmed leptospirosis case is a suspected dengue case with any of the following laboratory results: 1) isolation of Leptospira from a clinical specimen, 2) fourfold or greater increase in MAT titer between acute- and convalescent-phase serum specimens studied at the same laboratory, 3) demonstration of Leptospira in tissue by immunohistochemistry, or 4) MAT titer ≥800 in a serum specimen. A presumptive leptospirosis case is a suspected dengue case with any of the following laboratory results: 1) presence of MAT titer >200 but <800 in a serum specimen, or 2) demonstration of Leptospira in a clinical specimen by darkfield microscopy.

A terminal event is defined as the first event in a patient's clinical course that resulted in the need for cardiopulmonary resuscitation (e.g., a hypoxic seizure, an intracranial bleed).

A hemorrhagic manifestation is defined by the presence of any of the following: petechiae, purpura, ecchymosis, epistaxis, gingival bleeding, hematuria, menorrhagia, hemoptysis, hematemesis, melena, or an intracranial bleed.

Dengue fever (DF) is any suspected dengue case that meets the 1997 World Health Organization (WHO) case definition [20], which include acute onset of fever plus two or more of the following sign or symptoms: headache, retro-orbital pain, myalgia, arthralgia, rash, hemorrhagic manifestation, and leucopenia. Leucopenia was defined by white cell count <5.0×10⁹/L.

Dengue hemorrhagic fever (DHF) is any suspected dengue case that meets the following WHO criteria [20]: (1) fever or recent history of fever, (2) any hemorrhagic manifestation, (3) platelet count of ≤100,000/mm³, and (4) evidence of increased vascular permeability and plasma leakage which includes: (a) hemoconcentration with a hematocrit ≥20% above the U.S. population mean for age and sex, [21], [22] (b) ≥20% decline in hematocrit following volume-replacement treatment compared to baseline, (c) presence of pleural effusion or ascites detected by any imaging method, or (d) a serum protein or albumin <2.5 percentile for age and sex [23]–[25].

Dengue shock syndrome (DSS) is any case that meets the four criteria for DHF and has evidence of circulatory failure manifested by (1) rapid and weak pulse and narrow pulse pressure (≤20 mmHg [2.7 kPa]) or (2) hypotension for age and cold, clammy skin and restlessness.

A primary DENV infection (first DENV infection) is a laboratory-positive case in which the anti-DENV IgG titer was <1∶160 in an acute serum specimen collected ≤5 days after the onset of symptoms [14].

A secondary DENV infection (≥second DENV infection) is a laboratory-positive case in which the anti-DENV IgG titer was ≥1∶160 in an acute serum specimen [14].

Of the 11 laboratory-positive case-patients who died, eight were DENV RT-PCR positive in tissue, serum or both, and three were anti-DENV IgM positive in a single serum specimen (Table 1). Of the five DENV RT-PCR positive in serum, three were DENV-3, one was DENV-2, and one was DENV-1. Among the six case-patients with an acute serum specimen, four had secondary infections and two had primary infections. In the four laboratory-positive case-patients with tissue specimens, DENV was identified by IHC or RT-PCR in lung, liver and kidney specimens, and the most common histopathologic findings were intraalveolar edema and hemorrhage; congestion in the spleen, liver, and/or kidney; and fatty metamorphosis of the liver.

The median age of laboratory-positive case-patients was 26 years (range: 5 months to 78 years). Five were aged <15 years, four were 20–45 years, and two were >70 years. Seven were male. Five of six adults had at least one co-morbidity: two had asthma; one had an autoimmune hypothyroid disease; one had Type 2 diabetes mellitus (DM II) and hypertension; and one had DM II, chronic anemia, congestive heart failure, chronic obstructive pulmonary disease, and hypertension. In addition, four adults were overweight (i.e., body mass index [BMI] of 25.0–29.9), and one adult and one child were obese (i.e., BMI ≥30.0 or a BMI for age >95%).

Seven of the 11 case-patients sought medical care at least once prior to first hospital admission or presenting dead on arrival (DOA) to an emergency department (ED); three were seen by a clinician more than once but only 1 was diagnosed with a “dengue-like-syndrome”. Instead, the most common diagnoses given at these outpatient visits were upper respiratory infection with pharyngitis and/or cough, followed by acute gastroenteritis and viral syndrome. None of the seven case-patients had specimens submitted for dengue diagnostic testing until hospitalization (median 5 days post fever onset; range: 3–9 days) even though they saw clinicians early in the clinical course (median 2 days post fever onset; range 1–5 days). In addition, three case-patients had one or more warning signs for severe dengue at the time of fever defervescence, including persistent vomiting, severe abdominal pain, and narrow pulse pressure, and were sent home. A fourth case-patient, diagnosed with an upper respiratory tract infection as an outpatient, had a seizure at home the day after first being seen as an outpatient and died on the way to the hospital.

Upon final presentation to an ED, the 11 laboratory-positive case-patients had been sick a median of 4 days (range: 1–7 days). Two case-patients presented DOA and four were afebrile, three of which, had signs of shock. A seventh case-patient became afebrile while in ED and developed tachycardia, delayed capillary refill, and a narrow pulse pressure. Six case-patients had warning signs for severe dengue upon arrival to the ED including persistent vomiting (5/9) and abdominal pain (4/9). However, five of the nine case-patients were given a low (least severe) or intermediate severity triage score.

Nine case-patients were admitted to a hospital after a mean ED stay of 15 hours (median 12 hours, range: 3–48 hours) (Table 2). Initial complete blood count done in ED found that six case-patients had platelet counts <100,000/mm³ (median 78,000; range 8,000–410,000), five were leukopenic, two were hemoconcentrated (hematocrit 20% above mean for age/sex) and two had a hematocrit <2.5% for age and sex. Six of the nine case-patients met criteria for DF and three met criteria for DHF/DSS. Five case-patients had “dengue” listed in the admission differential diagnosis.

During their hospital stay, several case-patients developed warning signs for severe dengue including persistent vomiting (1/9), abdominal pain (1/9), restlessness (4/9), and mental status changes (4/9). In six cases, warning signs were not recognized as such as there were no new orders or change in the clinical management.

In the end, six of the 11 laboratory-confirmed fatal cases met criteria for DHF or DSS as determined throughout their clinical course and at autopsy (Table 3). Ten of the 11 case-patients had at least 1 hemorrhagic manifestation and nine case-patients had evidence of plasma leakage. Nine of 11 had thrombocytopenia documented.

Of the nine laboratory-positive case-patients admitted to a hospital, only three had capillary refill time assessed in the ED or at admission. Vital signs were measured at a median of every 3 hours in the ED but most case-patients had less frequent measurements after admission (Table 2). Two case-patients were admitted from the ED to the ICU and had vital signs measured every 1 or 2 hours.

Eight of nine hospitalized case-patients died during hospitalization, and the other case-patient was found dead at home within 18 hours of hospital discharge (Table 3). Three of the eight case-patients who died in the hospital had no recorded blood pressure measurements during the 8 hours before their terminal event. Five case-patients had their terminal event on the inpatient ward and another case-patient was transferred to ICU and had a terminal event within minutes of the transfer. Of those who died in hospital, six of eight case-patients died during a weekend (between 1701 on Friday and 0759 Monday), and five had a terminal event between 2300 and 0759.

All nine hospitalized case-patients received intravenous fluids, most commonly 0.9% normal saline (Table 2). Four received 0.45% normal saline while in shock. Four received intravenous albumin (5% or 25% solution) during their hospitalization but only one received colloids prior to the terminal event. Three of the nine hospitalized case-patients had signs of fluid overload prior to death including periorbital edema, dyspnea, and abdominal distention documented in their medical record.

There was frequent use of corticosteroids in laboratory-positive case-patients (Table 2). Five of the nine hospitalized case-patients received intravenous methylprednisolone as inpatients and one received dexamethasone in the ED before hospital admission. One case-patient who was DOA received oral prednisone during an outpatient visit.

Hematocrit levels were assessed every 17 hours on average (range 10–42 hours) for the six hospitalized case-patients with clinically significant hemorrhage (Table 2 and 3). Three of these six case-patients were given packed red blood cells; two in response to frank blood per nasogastric tube or rectum and one during the final code. One of these case-patients also received fresh frozen plasma in response to clinically significant bleeding. No case-patients were given whole blood. Two of the case-patients who received packed red blood cells also received platelets; one in response to clinically significant bleeding and one during the final code. Four additional case-patients had platelets ordered but they did not receive them prior to death.

Many of the nine hospitalized case-patients developed complications including metabolic acidosis (6/9), prolonged shock (6/9), acute respiratory failure (6/9), fluid overload (3/9), and seizures (3/9) (Table 2). Only 1 of the 11 case-patients had evidence of a secondary bacterial infection even though most patients had blood and urine cultures taken. Dengue, DHF, DSS, or status-post dengue syndrome was listed as the cause of death or a contributing factor in only five case-patients. The remaining six death certificates listed causes or contributing factors including hypovolemic shock, hypotension, metabolic acidosis, septicemia unspecified, bronchopneumonia unspecified, viral infection unspecified, brain death, and ischemic cerebral infarct.