Conversation with Monica Dashen and Rachel Green: Haitian Aid/Hurricane Matthew Aftermath Survey Project¹

Safe water

An estimated 750,000 people require safe water for drinking, cooking and personal hygiene. UNICEF has delivered safe water to 281,000 people more than one time. And 38,923 hygiene kits have been distributed since the beginning of the response.

Food

An estimated 1.4 million people are in need of urgent food aid because there was a widespread loss of crops, livestock and seeds. In response, the World Food Programme has delivered 14,096 metric tons of food. To offset the inevitable acute malnutrition, UNICEF has provided 23,193 children under age five with micronutrients.

Shelter/Personal Belongings

An estimated 1,579,900 persons have lost or damaged shelter and non-food times. As of December 21, 2016, and since the beginning of the response 37,233 blankets, 18,166 jerry cans, 29,913 kitchen kits, and 92,030 tarpaulins have been distributed.

The long distances and poor road infrastructure can limit the storage, division, and delivery of aid. For example, Hurricane Matthew swept away the only bridge leading into the south department from Port-au-Prince, thereby making it difficult for trucks filled with aid to enter into the south quickly right after the crisis. In preparation of Matthew’s wrath, USAID and their partners pre-positioned supplies in the most affected departments (or states), thereby making the distribution of aid more efficient

The damage was certainly greater after the 2010 earthquake than after Hurricane Matthew. The earthquake of January 2010 destroyed homes, schools, government buildings, and roads around Port-au-Prince.1212 In a nationally representative phone survey, two groups of SwB researchers found that 49% of the households experienced changes of members and about 18% reported that at least one child left the household or died. Employment rates were still affected five months later.1313,1414

The assessment of damage after a crisis is not new. The Salvation Army officials, for example, are assessing school damage after Matthew in the Grand’ Anse department. What is new is the more general assessment of damage that is not just specialized to a particular program. Like Kim et al. (2014)1515 and Orelien et al. (2013),1616 the present work looks at the type of home damage – total or partial and where people are currently residing as well as household membership change and items lost. The present work compliments that of OCHA’s updates on home displacement (the number of people living in shelters) by providing more detail about the type of home damage and loss.

Monica Dashen

I first thought of the aid distribution survey idea after talking to an UN official in Ethiopia. Right now, the northern and eastern areas of Ethiopia are experiencing a long term drought and aid is delivered to the schools. Gary Shapiro, Rachel Green and I helped this UN official set up a pilot survey where school officials were asked to report the number of student absences and aid deliveries on a monthly basis. I was reminded of this aid project after Hurricane Matthew struck Haiti and thought of an earlier SwB project on the Haitian Earthquake. I spoke to Fritz Scheuren about the scope and implementation of this project and am grateful to the other earthquake authors who took the time to share their materials, insights and contacts in Haiti. Likewise, I am appreciative of the American Statistical Association’s financial support.

Chart 1.

Source: https://en.wikipedia.org/wiki/Effects_of_Hurricane_Matthew_in_Haiti.

Rachel Green

I first heard about this project in an email from Monica, which was something like “Hey, I’m about to go to Haiti for a SwB project, can you help out with some survey design?” (Monica and I met when we worked on a SwB project together involving the Humanitarian ID software.)

When I first saw the survey aims it seemed a little bit optimistic to collect a range of information on many sensitive topics while maintaining short interview times. There is something of a challenge in designing survey questions for an environment that can seem far from where you live. For example when looking at the questions on awareness of cholera treatments and symptoms, I wondered how I and others would have done at identifying cholera.

Chatr 1 shows the total rainfall projected across Haiti from NOAA. It was projected that the greatest rainfall totals from Hurricane Matthew would be localized in the southeastern Haiti.1717

Monica Dashen

I had some interesting discussions with the interviewers in the field about the complications of delivering aid. The team and I agreed that these people were on the border line of survival prior to Matthew. These people grew their own food (e.g., plantains and corn) and rarely bought food. Matthew swept their crops away and sometimes their food stashes. The team and I agreed that handouts were a short term solution. Seeds and animals are a long term solution. Frankly, I wonder whether the survey participants (of November 2016) are still alive.

Rachel Green

One of the challenges in this project was in deciding how to ask about sensitive topics in a way that provided people an opportunity to share their stories without adding to (or causing) distress for participants. Balanced with this we needed data that was useful for discussions on health, aid, and the impact of Matthew. After doing some of the data analysis, I appreciate even more the people who translated the results into English.

Fig. 2.

The team from left to right – Chauncy, Franzlande, Ginny, Emilio, Mackenson and Antoine.

Questionnaires

Two questionnaires were designed. The first questionnaire, i.e., the Household Questionnaire, pertained to people who resided in their homes. People were asked about (1) aid received after Matthew, (2) damaged homes and lost possessions after Matthew, (3) job status before and after Matthew, (4) family members who moved in after Matthew, (5) danger signs of diseases, (6) clinic accessibility before and after Matthew, (7) injuries after Matthew and (8) change in dietary habits before and after Matthew.

The second questionnaire, i.e., the Health Worker Questionnaire (HW), pertained to those people who worked in a clinic or hospital. Health workers were asked about (1) supplies, (2) patient diseases, (3) injuries, (4) patient accessibility, and (5) staffing and hours.

Both the Health Worker and Household questionnaires originally written in English were translated into Creole and back into English for review. The same Household questionnaire was used in the face-to-face and phone survey. In contrast, the Health Worker questionnaire was only used in the face-to-face interviews.

Training

Interviewers underwent a fairly standard training program where, topics such as questionnaire conventions and interview protocols were covered. All questions were asked and recorded in Creole. Consent was obtained verbally after an explanation of the survey goals and use of the data. A screener was set up so that only people 15 years or older were interviewed.

Data Collection

The data was collected in two modes (1) face-to-face interviews and (2) phone surveys. Each mode is discussed in more detail below.

Face-to-face interview data collection in the south department was November 6–8, 2016. Data collection was carried out over three days. The data collection was preceded by four hours of training the day before. The team visited five communes: (1) Camp-Perrin, (2) Torbeck, (3) Chantal, (4) Port Salut, and (5) Arniquet. Two to six sub communes were visited per commune including the center. (The selection of the communes and sub communes was based on road accessibility and lodging.) To travel to those “out of the way” sub communes, an official from the Civil Protection Service familiar with the back roads assisted the team.

Upon arrival at a commune or sub commune, the team coordinated household interview routes at the truck. Afterwards, each interviewer headed out to adjacent houses by foot. A supervisor monitored the team. No two interviewers went to the same house or in the same direction. Interviewers were instructed to interview (1) any household member 15 years old or older and (2) only one member per household. Upon completion of each household interview, the interviewer walked to the adjacent home and conducted the next interview. Anti-bacterial bars of soap were provided as incentives. A debriefing was offered at the end of the day where interviewers discussed problems.

Repeated face-to-face interviews (re-interviews) were conducted before leaving each commune or sub-commune. A total of eight repeat interviews were conducted. Comparisons between the repeat and original interviews did not reveal many discrepancies. There were no respondent refusals.

As a part of this face-to-face portion of the survey, four health workers were interviewed about the aftermath of Matthew. The skilled healthcare workers resided in each commune center and either worked at a clinic, dispensary, or hospital. (The team did not find unskilled health workers to interview.)

Phone interview data collection for all ten departments was November 9–12, 2016. Data collection was carried out over three and half days. The interviewers worked from a randomly generated set of phone numbers. (A list of working phone numbers provided by the phone companies was unavailable.) Originally, 250 completed interviews were expected. However, only 150 completed interviews were obtained due to interviewer fatigue and time limitations. Every 8th to 25th call was a working phone number. Again, the supervisor monitored the interviewers and ten repeat interviews were conducted. Comparisons between the repeat interviews and original interviews did not reveal many discrepancies. Phone credit was offered as an incentive. Debriefings occurred throughout the day.

Keying and data processing started immediately after the face-to-face interviews and during the phone interviews. Data processors were trained to translate and clean data (e.g., typos and misspellings). Thirty-two cases were later verified by two data processors. Comparisons between the original and re-keyed interviews did not reveal many discrepancies.

There were no outright refusals (failure to answer any questions). Rather, there were 19 incomplete cases (in the phone survey). And there were three call backs, where the respondent did not have time to talk at the moment, but did later that day.

To ensure that the interviewers reached all ten departments by phone, the geo-spread or the number of calls per department were calculated Forty-three percent of the phone interviews reached people in the West department where most of the Haitians reside. The remaining 57% of the calls reached people in the remaining nine departments.

Monica Dashen

I spent a lot of time thinking about the “who” in the aid distribution, as this survey can be readily applied to an immediate or prolonged crisis situation. To “fine tune” this survey, I thought about what the “who” can say. Here the “who” can certainly tell us whether they received aid and the steps needed to obtain it. In contrast, the “who” cannot tell us who delivered the aid, as many people reported that the truck was unmarked. (If the truck was marked, many of the people could not read the logo/words.) Likewise, the “who” cannot tell us how much was pilfered prior to distribution.

Fig. 3.

Monica, Frantzlande and nurse in a hospital.

Besides thinking about those “who” who received aid, I thought about the other “who.” That is, those “who” who were involved in the actual distribution of aid. This idea came from a conversation I had with Dr. Holguin-Veras about his work on the relief efforts after the Haitian earthquake where he found that socially integrated relief agencies were better able to distribute aid than those less integrated.2020 Typically, the “who” involved in aid distribution must have (1) a fleet of trucks able to navigate the narrow and rocky roads of the south, (2) drivers familiar with the areas and (3) people willing to load, unload and parse the vast amount of aid. To find out more about these “who,” I asked an interviewer to call four people involved in the aid distribution and ask them about the availability of aid, trucks, drivers and guides. Unfortunately, these aid distribution people were reluctant to provide any information, as they “do not want to throw stones that may fall back on [their] heads.” (This is a Creole proverb.)

Monica Dashen

As in any project, there are lessons learned. The most obvious lesson is that phone coverage in a crisis situation can be lower than a non-crisis situation. Only 37% of the known phone numbers collected in the south during face-to-face interviews worked. (Phone calls were made immediately after the team left the south to ask a follow-up question.) The most likely reason is that people simply cannot pay their bills. (Mobile tower damage was ruled out as an option due to the interview team mobile phone usage in the field.) Face-to-Face interviews are the best option here. When planning such a survey, the designer should allot more time for face-to-face interviews than phone interviews, as the respondents are upset and talkative (Private interviews are difficult to obtain.) Likewise, the survey designer should allot additional money to hire a guide to provide security and help the field team navigate the backroads. Another lesson learned is that developing questions to assess the situation is difficult. Survey designers should expect to finalize a questionnaire in the field. The authors devised three versions of questions related to a dietary change in habit before hitting the right one.

The final lesson learned is a broader assessment of the sanitation situation (e.g., availability of bleach to treat water) is always needed. Although not directly related to the survey goals, such information paints a boarder picture of the sanitation and disease prevention For instance, in rural areas, cholera is difficult to treat, as it requires truckloads of Intravenous fluid (IV) for those patients who suffer from serve dehydration. Rural clinics simply do not have a ready supply of IV fluid for a large number of patients. Likewise, the roads leading to these clinics are in poor condition, thereby limiting access to large trucks. Such roads and distances make prevention all the more critical. Clean water and good sanitation habits are critical preventative methods. The Center for Disease Control recommends that bleach, a water treatment, should be available in every village, as bleach is easy and cheap to produce in any country.2121

Rachel Green

My key observation is of the benefit of continually refining questions and adjusting to the local situation. While acknowledging the ongoing challenge of how to ask questions on sensitive topics to illicit ongoing participation while collecting useful data and not causing distress for participants.”

Monica Dashen

I wrote in a field note piece in AmstatNews2222 that “looks can be deceiving” regarding the rebuilding process occurring in Haiti after Hurricane Matthew. I now read the OCHA reports with a skeptical eye. I agree that aid is brought into Haiti However, I wonder who actually received the aid.

After talking to some aid officials about our findings, I learned that the aid distribution had some hiccups in the beginning (October, 2016) and that the distribution is more systematic now (February, 2017). To find out if the distribution indeed had changed, Rachel and I designed a follow-up phone survey where the same respondents (interviewed in November, 2016 and again in February, 2017) were asked about (1) aid received, (2) cholera prevention habits and (3) food eaten.

Unfortunately, we had difficulties returning to the field to conduct face-to-face interviews. As a result, we resorted to conducting phone survey interviews. Two interviewers were asked to call 119 known numbers; these numbers were culled from the highly-affected areas only. Of the 119 known numbers, only 32 numbers worked (27%).

There are three findings of interest here. First, when people were asked about aid, only one person (out of 24) received aid. This person received aid right after Hurricane Matthew.

Second, when people were asked about their cholera prevention habits, 17 survey participants (out of 24) relied on aqua tabs and/or bleach to clean their drinking water. Here water can be dirty at its origin (e.g., spring or well) and must be treated at the household level. Only six survey participants could tell us with certainty whether bleach was readily available in their sub commune. Other cholera prevention methods reported by eight people included washing hands, cleaning surfaces and washing food Third, when people were asked about their dietary habits after Matthew, everyone still reported eating a lot of rice Some people reported eating beans, corn and/or pasta too. (We did not ask whether the rice was obtained from a neighbor/friend, grocery store or food stash.)

These pilot survey findings indicate that the aid distribution has not changed. This project reminds me of the potato famine where over a million Irish died of starvation or immigrated even though wheat and other grains were grown on Irish soil and exported to Europe.2323

Rachel Green

As we’ve said earlier, this project relates to some highly complex topics where important work is ongoing. It was interesting to contribute in a small way to these ongoing discussions and thanks again to all those who helped out in the project and thanks especially to those who shared their experiences and stories.

A-1: History of Poverty and Poor Health.

According to UNICEF’s 2015 State of the World’s Children report,2424 only 58% of Haiti’s population had access to an improved drinking water source (water from a well or pipe) and 28% had access to a latrine. Many people resided in poorly constructed concrete homes. Haiti has poor health indices. The life expectancy at birth is 63 years and the infant mortality rate is 52 per 1, 000 live births. An estimated 69 of every 1,000 children born die by the age of 5 years,2525 and 12% of the surviving children are undernourished. Furthermore, Haiti is ranked 163 out of 183 countries in the UN Human Development Index.2626 This development index is based on life expectancy, years of schooling, and standard of living.

A-2: The Cholera Outbreak after the Earthquake in 2010

In October, 2010, the public health ministries were first alerted to a cholera outbreak ten months after the earthquake. Cholera is a water borne bacteria that is highly contagious and deadly at times. The outbreak spread rapidly throughout Haiti and reached portions of the Dominican Republic and Florida (USA).

Unclean drinking water was identified by Center for Disease Control (CDC) officials as the primary source of cholera. Poor medical training and limited medical facilities also contributed to the problem, as some patients were discharged too quickly and died later. Cholera produces serve dehydration due to a sudden onset of vomiting and diarrhea. Training programs for the health care professionals and the general population were designed. The training materials discussed how to treat water, how to make use of ORS for dehydrated patients, and how to make use of bleach to disinfect homes. USAID provided cholera treatment supplies.2727

The number of suspected cholera cases increased after Matthew. A month prior to Matthew, the number of suspected cases was 2,236. A month after Matthew the number increased to 5,100 cases.2828