Risk and Protective Factors for Cholera Deaths during an Urban Outbreak—Lusaka, Zambia, 2017–2018

INTRODUCTION

Cholera is endemic in more than 40 countries globally, with an estimated 1.3–4.0 million cholera cases and 21,000–143,000 cholera deaths each year, most of which occur in developing countries.¹ The first cholera outbreak recorded in Zambia occurred in 1977 with 178 cases; cholera has been recorded nearly every year since then.² Zambia experienced outbreaks of more than 10,000 cases in 1991, 1992, 1999, and 2004.³ Environmental conditions conducive to cholera spread, such as lack of access to safe water and improved sanitation, are common in Zambia, particularly in remote regions and in high-density, peri-urban neighborhoods of Lusaka, the capital city.^4,5

Cholera is caused by consuming food or water contaminated by toxigenic Vibrio cholerae serogroup O1 or O139. Most people exposed to these pathogens do not develop symptoms. However, approximately 25% will experience the acute onset of watery diarrhea,⁶ which in about one-third of symptomatic cases will be profuse and accompanied by vomiting. Without treatment, severe dehydration will ensue with case fatality rates (CFRs) that can reach up to 50%. With appropriate oral and intravenous rehydration, the CFR can be less than 1%.⁷

A cholera outbreak, centered predominantly in Lusaka, was declared on October 6, 2017. Lusaka district has an estimated population of 1.7 million, with a population density of 4,853 people per square kilometer as of 2010.⁸ Within Lusaka Province, it had been estimated that 24% live in poverty and 66% lack basic and safely managed sanitation.^8,9 Although cases were eventually recorded in seven of 10 provinces in Zambia, most cases outside Lusaka had epidemiological links with cases in Lusaka. By December 11, 20 of 661 reported cases had died (CFR 3%), prompting the CDC and the Zambian Ministry of Health (MoH) through the Zambia National Public Health Institute to conduct a case–control study to identify factors associated with mortality and determine the interventions that could best prevent future cholera-related deaths.

MATERIALS AND METHODS

We conducted a retrospective case–control study from January to March 2018. Decedents were patients who died from cholera (defined as severe dehydration or death from acute watery diarrhea, regardless of age) in Lusaka district from October 1, 2017 to January 15, 2018. The deaths occurred in either a cholera treatment center (CTC) or the community. A community decedent was defined as a person who died in the community or who died with suspected cholera within 45 minutes of arrival at a CTC. Survivors were cholera patients admitted to a CTC in Lusaka district during the same period who were discharged alive. Decedents and survivors were identified through a convenience sampling of those included on the MoH cholera line list. In addition, some decedents were identified before their inclusion on the line list through multiple sources, including review of patient records at CTCs and information from community health workers. We attempted to include all cholera deaths occurring during the specified time period. We assigned two survivors to each decedent, matched by age-group (< 2, 2–17, ≥ 18 years) and date of onset (±1 day). If more than two survivors met the matching criteria, two survivors were randomly selected from eligible controls. If fewer than two survivors met the matching criteria, the matching criterion for date of onset was extended to ± 5 days. We excluded decedents and survivors if they did not reside in Lusaka district, if the patient or family member did not consent for interview, if the patient originally designated as deceased was found alive, or if the patient or a family member was unable to be located.

Interviews were conducted in the home of the patient or respondent. For survivors who were unavailable for interview and decedents, we attempted to interview close family members who had knowledge of the patient’s history and clinical course. We obtained written consent from all participants aged ≥ 18 years before the interview. Parents or guardians of participants aged < 18 years were asked for consent, and assent was obtained from all adolescents aged 13–17 years. Questionnaires were written in English and distributed electronically to tablets used by enumerators. Questions were translated into standardized Nyanja or Bemba by enumerators during the interview to ensure understanding.

Information collected included demographics, education level, socioeconomic indicators, comorbidities, symptoms and clinical course of illness, care-seeking behaviors, cholera treatment at home and at any health facilities, knowledge about cholera, cholera prevention activities, and water, sanitation, and hygiene characteristics and behaviors at the household.

In addition to information collected during the case–control study, we reviewed available medical records of cholera decedents from the CTCs, abstracting data on illness onset, symptoms, length of stay, comorbidities, and treatment received. Medical records of all decedents were requested from each CTC from the beginning of the outbreak until the time of data collection (January 9–26, 2018).

The Zambian MoH Research Ethics Committee approved the study protocol, and the project was reviewed in accordance with CDC human research protection procedures and was determined to be a non-research outbreak response activity.

RESULTS

During this 6-month-long cholera outbreak, 5,905 suspected cases and 114 deaths were reported. Most cases (5,414 [91.7%]) and deaths (98 [86.0%]) were reported by Lusaka district (Figure 1). A total of 38 decedents and 76 survivors (matched 1:2) were included in the study. For the decedents, proxies interviewed were parents (34%), spouses (18%), children (13%), siblings (11%), in-laws (11%), aunts or uncles (8%), and grandmothers (5%). Most (62%) survivor respondents were also proxies. Proxies interviewed for survivors were parents (29%), spouses (8%), children (7%), siblings (8%), other friend or family member (4%), in-laws (3%), aunts or uncles (2%), and grandmothers (1%).

Epidemiologic curve showing suspected cases of cholera during the outbreak by week of illness onset—Lusaka district, Zambia, 2017–2018.

Citation: The American Journal of Tropical Medicine and Hygiene 102, 3; 10.4269/ajtmh.19-0678

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Epidemiologic curve showing suspected cases of cholera during the outbreak by week of illness onset—Lusaka district, Zambia, 2017–2018.

Citation: The American Journal of Tropical Medicine and Hygiene 102, 3; 10.4269/ajtmh.19-0678

• Download Figure
• Download figure as PowerPoint slide

Epidemiologic curve showing suspected cases of cholera during the outbreak by week of illness onset—Lusaka district, Zambia, 2017–2018.

Citation: The American Journal of Tropical Medicine and Hygiene 102, 3; 10.4269/ajtmh.19-0678

• Download Figure
• Download figure as PowerPoint slide

DISCUSSION

During the early weeks of the cholera outbreak in December 2017, cholera mortality approached 3%, with one-quarter of deaths occurring in the community. The risk factors for cholera mortality identified in this study were as follows: age > 55 years and not having completed primary school. Receiving ORS immediately on presentation to a CTC was protective. The relatively small sample size of our study may have limited our ability to detect other factors previously associated with cholera mortality in other places and settings.

Early access to rehydration (such as ORS and IVF) is the cornerstone of clinical management for cholera. The use of ORS has been associated with decreased mortality from diarrhea in infants and children aged < 5 years,^10,11 and increasing the availability of ORS in the community has been shown to decrease mortality from cholera.^12–15 This study supports these previous findings, as receiving ORS immediately at the healthcare facility was identified as a protective factor. Unlike previous studies,^12,13 this study did not demonstrate an association between receiving ORS at home and mortality. The short distance necessary to travel to a CTC (mean: 2 km) in this peri-urban outbreak may explain why home ORS treatment was not shown to provide the protection noted in previous studies.^12,13 Home ORS may be most critical to bridge the gap when patients travel longer distances to receive additional health care.

Infection with cholera is more common in communities with a lower socioeconomic status (SES), likely due in part to limited access to safe water and sanitation. However, there are limited published data related to SES and cholera mortality among those infected. A previous study in Cameroon has highlighted an association between lower SES and cholera mortality.¹² Our study did not find a statistically significant association between proxies for lower SES (such as use of pit latrines and household assets) and cholera mortality. In a previous study, household ownership of a mobile phone, which may represent a higher relative SES, was found to be associated with decreased cholera severity.¹⁶ Although our data did not demonstrate a statistically significant association between smart phone ownership and decreased cholera mortality, mobile phone ownership may improve a patient’s ability to communicate and seek timely health care, which merits further exploration. Lower SES has also been associated with fewer years of education.¹⁷ However, this is the first study to identify lower education as a risk factor for cholera mortality.^{12,13,18–22} The increased odds of cholera mortality in those with lower education levels, in the context of an affected population with potentially similar SES, emphasizes the importance of education as a key contributor to one’s health and well-being. A lower education level can inhibit one’s ability to protect oneself from diseases, such as cholera. Limited literacy can also affect a person’s ability to receive health messaging, which may impact survival.

In our study, a higher odds of mortality was found among adults with increased age (> 55 years), which was similar to findings from previous studies in Cameroon, Haiti, and Guinea Bissau.^12,19,21,22 Previous studies have shown mixed results regarding the relationship between cholera mortality and gender. Although several studies have identified male gender as a risk factor for mortality,^13,19 no association was seen in this study, which is consistent with other published reports.^12,20–22

Appropriate clinical management is essential for decreasing cholera mortality. Severe cholera leads to rapid dehydration due to large volumes of stool output (as high as one liter per hour, in adults), which requires close monitoring and aggressive rehydration with ORS and IVF. With proper rehydration, the CFR for cholera should be less than 1%.^6,7 In the chart review, we were unable to assess the adequacy of rehydration due to limited documentation. However, the reported duration between admission and death for most decedents who received care at a CTC was greater than 12 hours, which should be adequate time for appropriate fluid resuscitation in patients without complications.²⁰ The case–control investigation and the chart review demonstrated frequent use of antibiotics, which are recommended for severe cholera and selected high-risk patients. Unlike a previous study,²⁰ neither home nor hospital-administered antibiotics demonstrated statistically significant protection from death in this setting.

The chart review also raised concerns regarding the appropriate use of RDT and the impact RDTs may have on clinical management. Three of 13 patients with documentation initially tested negative by RDT. One of these patients was then transferred out of the CTC, despite meeting the cholera case definition, to the female ward where she died a few hours later. The other two patients initially testing negative were retested and found to be positive by RDT after death (or on documentation certifying death). Because of limited documentation, it is unclear if inaccurate RDT results altered clinical decision-making. Current recommendations by the Global Task Force on Cholera Control advise that RDTs are intended for surveillance purposes in peripheral healthcare facilities and have limited utility for individual diagnosis as the results would have no influence on the immediate management of the patient with suspected cholera, which should be guided by clinical assessment of dehydration and fluid losses and on cholera treatment guidelines.²³ Rapid diagnostic test results are not reliable enough to influence individual treatment decisions of suspected cholera cases during a cholera outbreak.

The increasing availability of oral cholera vaccines (OCVs) has provided an additional tool for cholera control and prevention. A single-dose OCV campaign targeting > 500,000 people was conducted in Lusaka in April 2016. A second dose was offered to persons at risk in December 2016.^24,25 The average two-dose effectiveness of the killed OCV has been demonstrated to be 76% at 18 months (weighted mean duration) in a systematic review.²⁶ Similarly, in Haiti, two-dose effectiveness has been reported as 76% at 4 years.²⁷ Decreased efficacy has been documented in patients aged less than 5 years and beyond the second year after administration.^26–28 With this knowledge, we asked patients and their proxies if the patient received zero, one, or two doses of the OCV. According to respondents, only four (4%) patients had previously received two doses of OCV, and a minority of decedents (13%) and survivors (14%) had previously received one or more doses of OCV. No statistically significant difference was noted. However, these results should be interpreted cautiously as vaccination was not validated by vaccination records. Recall bias and responses from proxies may have led to inaccuracies in reporting. Future studies should consider investigating the relationship between OCV and cholera mortality more comprehensively.

This investigation has several limitations. First, in the case–control study, we attempted to include all cholera deaths occurring during the study time period and to select a representative sample of matched controls. However, we encountered challenges in locating the selected decedents and survivors because of a number of factors. For example, some patients (or patients’ families) provided incorrect information at the CTC, including false names and incorrect or incomplete addresses. We received reports that some had voluntarily relocated to another community (or had been forced to move by a landlord) because of stigma from the patient’s illness with cholera. In addition, there were errors in the line list, including patient ages, names, addresses, and outcomes, that made it difficult to trace each patient. We collaborated with community health workers to assist our efforts, but we were unable to locate all the selected decedents and survivors. A second limitation is that although the questionnaires were administered within 4 months of the patient’s illness (median: 30 days; range 5–99 days), recall bias by patients or their proxies may have been introduced, given the time duration between illness and interview. Another limitation, response bias, is introduced when interviewing proxy respondents. To attenuate this risk, we attempted to interview proxies that were primary caregivers and those who provided support during the illness. However, as many studies have described,²⁹ proxy respondents may introduce biases and inaccuracies, especially as the relationship distance widens. For example, we suspect that inaccurate recollections and responses from proxies may have contributed to underreporting of diarrhea in decedents. Fourth, because this study was limited to a relatively small sample size of 38 decedents and 76 survivors (or less), we could not further examine the joint contribution of factors to cholera mortality in a multivariable analysis. Last, there is a risk of misclassification of illness. Not all participants had confirmatory testing for cholera by stool culture. Although all study participants reportedly met the case definition for suspected cholera as evidenced by their inclusion on the line list, some patients may have experienced an illness unrelated to cholera.

Our study highlights the increased risk for cholera mortality in adults aged > 55 years and among patients with lower education levels. It also underscores the importance of early rehydration with ORS. These findings can help guide messaging during future outbreaks, emphasizing the need to reach all community members, including those with advanced age and limited education. Communication strategies used during an outbreak response should consider the literacy level (including health literacy) of those most at risk for death. Messages should be simple and should include spoken messages in the primary language of those at risk. Messages should also be shared via platforms accessible to and used by older adults. Traditional cholera prevention and treatment messages remain relevant to prevent deaths during outbreaks. These include underscoring the importance of drinking ORS at home for mild illness and that all suspected cholera patients should seek health care immediately if symptoms worsen or dehydration occurs. Because rehydration remains the cornerstone of cholera treatment, public health interventions should include increasing availability of ORS in the community and facilitating ease of access to treatment at healthcare facilities with a focus on early rehydration with ORS and IVF to decrease cholera mortality.