University of Ghana http://ugspace.ug.edu.gh SCHOOL OF PUBLIC HEALTH COLLEGE OF HEALTH SCIENCES UNIVERSITY OF GHANA LEGON MICHAEL JEROEN ADJABENG (10128395) RISK FACTORS FOR ACUTE RESPIRATORY INFECTIONS IN SHAI- OSUDOKU AND NINGO-PRAMPRAM DISTRICTS IN THE GREATER ACCRA REGION OF GHANA THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF DOCTOR OF PHILOSOPHY IN PUBLIC HEALTH DEGREE EPIDEMIOLOGY AND DISEASE CONTROL DEPARTMENT JULY 2017 University of Ghana http://ugspace.ug.edu.gh DECLARATION I, Michael Jeroen Adjabeng, affirm that except for other people's studies that have been duly acknowledged, this work is the outcome of my own original research, and that this thesis, either in whole or in part has not been submitted elsewhere for another degree . ............. ~' . MICHAEL JEROEN ADJABENG (10128395) ACADEMIC SUPERVISORS ................ ~ . PROF. RICHARD M. K. ADANU FARI (Rtd) PROF. WILLIAM KWABENA AMPOFO Ie ~!~.iC:HOO!_ 0'- PU8L1~ 't';. : ~: ~ J. LT H L : H r;: ,:\ F~ :~;,:JD') LEG o !'oJ :~-.:';::1 University of Ghana http://ugspace.ug.edu.gh DEDICATION I dedicate this thesis to my wife, Mrs. Juliana Teiko Adjabeng, my source of encouragement for this academic pursuit and the Almighty God for his sovereign guidance. 11 University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT I wish to indicate my sincere gratitude to my academic supervisors, Prof. Richard M. K. Adanu, Prof Edwin Afari all of School of Public Health and Prof William Kwabena Ampofo of the Noguchi Memorial Institute for Medical Research for their mentorship, guidance and invaluable inputs through every phase of the development of this dissertation. Mention should be made of Dr. Lawson Ahadzie, former Head of the Disease Surveillance Department of the Ghana Health Service, who has mentored me professionally and remains a source of inspiration. My sincere thanks to Dr. Evelyn Korkor Ansah of the Ghana Health Service Research Development Division and Dr. Margaret Gyapong of the Dodowa Health Research Centre for their guidance in community entry to the study population and orientation on demographic health surveillance. Dr. Meridith McMorrow of the United States Centers for Diseases Prevention and Control made available invaluable reference materials and provided useful comments during the proposal development. Sincere appreciation also goes to Prof. Moses Aikins, Dr. S.O Sackey, Dr. Francis Anto of the School of Public Health for their academic guidance. The Head of the Epidemiology and Disease Control Department, Dr. Bismark Y. Sarfo facilitated the forum for presentations to the faculties where helpful comments were received and ensured timelines were met. My sincere appreciation again to all the lecturers of the School of Public Health for knowledge shared. I need to also mention colleagues such as Mr. Edward Nartey, Ms. Elizabeth Awini, Mr. Dodzi Amelor and Dr. Joseph Opare for their immense support during this study. III University of Ghana http://ugspace.ug.edu.gh I am grateful to Mrs. Wilma Appiah who played a key role in field coordination from the very inception of this study. The District Directors of Health Services of Shai-Osudoku (Dr. Afua Animwaa Asante) and Ningo-Prampram (Mrs. Gifty Of or i) Districts supported the study and provided background information of the study population. Special mention has to be made of the National Influenza Centre in the facilitation of laboratory investigations carried out on study participants, and staff of the Dodowa Health Research Centre for field support. To respondents from the study population, I thank you all for participating in this study to contribute to the body of knowledge on acute respiratory infections in Ghana. Ifinally acknowledge the financial support for tuition, student's stipend and field work from the U.S. CDC Cooperative Grant Award Number: UOIIP000607-04. The results and conclusions in this thesis are those of the student and do not essentially represent the view of the U.S. CDC. iv University of Ghana http://ugspace.ug.edu.gh ABSTRACT Background: Acute Respiratory Infections (ARIs) result in a large public health burden worldwide, especially in developing countries. The populations at greatest risk for developing fatal respiratory infections are the very young, the aged, and the immunocompromised. In developing countries, 30% of all patient consultation and 25% of all paediatric admissions are ARIs. In the Shai-Osudoku (SO) and Ningo-Prampram (NP) districts, ARls have continuously ranked second in the top ten causes of morbidity for hospital attendance. Despite the availability of influenza vaccines, Ghana and other West African countries are yet to establish routine immunization policies due to limited information. This study sought to determine the risk factors of ARls by investigating the characteristics of patients with respiratory illness in the two districts. Methods: A health facility-based case-control study was conducted among residents of Shai-Osudoku (SO) and Ningo-Prampram (NP) Districts. Prospective cases were selected from a facility-based surveillance on Acute Respiratory Infections which captured Influenza- like Illness and Severe Acute Respiratory Infection syndromes in residents of SO and NP Districts from April to November 2016. One hundred and forty-seven influenza virus- positive case-patients and 294 influenza virus-negative control-patients were identified to assess the risk factors of influenza among Influenza-like Illness patients. The study also investigated factors for Severe Acute Respiratory Infection (SARI) by identifying 134 SARI patients and 402 out-patients with non-respiratory illness as controls. Crude and adjusted odds ratios were calculated. A purposive selection logistic regression was used for the adjusted modelling. Results: The study identified Influenza A(H3N2) and Influenza B subtypes as the predominant circulating influenza viruses in the two districts from March to November of v University of Ghana http://ugspace.ug.edu.gh 2016. Study participants had poor knowledge on the causes and prevention of acute respiratory illness. The Osudoku Health Centre in Shai-Osudoku district reported the highest proportion (29.9%, p = 0.01) of influenza-positive III cases. The crude analysis had the highest odds of an influenza-positive infection in the 5 to <15 years age group (OR:7.80; 95%CI: 2.52 - 24.12). Factors associated with influenza virus infection among Ills were Chills (aOR:4.57; 95%CI: l.51 - 13.76) and a recent travel history in past 2 weeks (aOR:3.05; 95%CI: 1.07 - 8.73). For SARI, Males were more at risk (OR:2.13; 95%CI: 1.40 - 3.23). Just as has been found in some studies, the less than 5-year group formed the majority (42.5%; p '''''''..,~ t, •••••~ A •••••• l Figure 3.2: Map of Shai-Osudoku and Ningo-Prampram Districts indicating study health facilities Source: (Dodowa Health Research Centre, 2013) 48 University of Ghana http://ugspace.ug.edu.gh The Shai-Osudoku district has 2 sub districts while Ningo-Prampram has a total of 6 sub districts. The 2 districts are almost entirely rural with the majority of inhabitants economically dependent on subsistence agriculture (Greater Accra Regional Health Administration, 2011). There are 52 health facilities in these two districts. These comprise: one Hospital, one Polyclinic, four Health Centres, thirty-seven Community-based Health Planning and Services (CHPS) compounds and nine private clinics (1 Maternity Home and I Mission Clinic inclusive). The major causes of death in the study area are hypertension and anaemia (Ghana Health Service, 2017). The age-sex structure of the population has a wide base, composed of children at younger ages. The percentage in higher ages reduces gradually in subsequent age groups with a small number of elderly and more females than males at advanced years (Ghana Statistical Service, 2014b,2014a). The two districts have a combined population of 143,092 for 2016 (Centre for Health Information Management, 2016) representing about 3.1 % of the total population of the Greater Accra Region. The SO District accounted for 60,712 (42.4%) of the total population in the two districts. The ratio of males to females was 92: 100. From the Dodowa Health and Demographic Surveillance System database, there were 24,221 (2014) households scattered in 382 communities ("Our History - Dodowa Health Research Centre," 20 IS). Most of the inhabitants in the SO and NP Districts are subsistence farmers, fishermen and traders. There are few skilled artisans, craftsmen and civil servants, mainly migrant workers 49 University of Ghana http://ugspace.ug.edu.gh of government ministries, departments and agencies. The major ethnic group is Dangme (72%), followed by Ewes (13.7%) with other minor ethnic groups forming one percent of the population. Christianity forms the main (>80%) religious faith among the population ("Our History - Dodowa Health Research Centre," 2015). The 2010 census results indicate that 49 percent of all dwelling units live in compound houses. The main building material for outer walls of most of the houses in the District is Cement block!Concrete (57.5 %) and Mud brick! Earth (32.8 %). The use of Earth/Mud for floor was 15.9 %. For main source of lighting, kerosene lamp formed 29.8% and the mains - 59.4%. The key source of fuel for cooking by households is charcoal - 49.3 %, wood - 27.5% % and gas - 17.4%. Those with no toilet facilities formed 44.3% with defaecation in bushes/beaches/open fields. The most common method of solid waste disposal is by burning (33.3%), (Ghana Statistical Service, 2014b, 2014a). 3.3 Variables The independent variables were grouped into individual, socio-economic and environmental factors. The variables grouped as individual included age, sex, pre-existing medical condition, cigarrete smoker and prior treatment history. Under the socio-economic group were occupation, literacy level of caregiver, educational status, marital status, perception of cause of illness and household wealth index. The environmental risk factors included exposure to indoor biomass fuels, number of people in household (overcrowding) and smoking (passive). The key outcome variables of interest were influenza-positive ILl and SARI patients (Table 3.1). 50 University of Ghana http://ugspace.ug.edu.gh Table 3.1: Definition of some variables and their scale of measurement Type of Scale of Variables Operational definition variable measurement ILJ Any person with sudden onset of fever (history/measured) of~37.50C (axillary) and cough and/ or other respiratory signs with Dependent Nominal onset within the last 10 days SARI Any person with sudden onset of fever (history/measured) of ~37.5°C and cough and/ or other respiratory signs with onset within the Dependent Nominal last 10 days and requires hospitalization Individual This is the age in years of the patients Age interviewed as at last birthday Independent Interval Sex This is defined as male or female Independent Nominal Weight (kg) divided by the square of the Independent Ordinal Body Mass Index height in meters (m") Date of Symptom Date of onset of first sign(s) or symptom(s) Independent Interval Onset Pre-existing Medical Asthma, Chronic Heart Disease, Diabetes, Conditions Chronic Liver Disease, Chronic Lung Independent Nominal Disease, HIV, TB Self-administered Self-administered medication before reporting Independent Nominal medication to a health facility History of Smoking History of ever smoked in the past Independent Nominal Socio-economic Based on assets and possessions of Wealth Index Independent Ordinalhouseholds Any income generation activity that an 18- Occupation Independent Nominalyear-old and above engages in Attribution to Curse Perception of ARl attributed to curse Independent Nominal Marital status of respondents aged 18 years or Marital Status Independent Nominalmore 51 University of Ghana http://ugspace.ug.edu.gh Table 3.1 (continued): Definition of some variables and their scale of measurement Environment Exposure to smoke Smokers or persons living with others who smokein their presence Independent Nominal Travel History Travelled outside SO and NP Districts 2 weeksprior to onset of signs or symptoms Independent Nominal Type of cooking Biomass fuel predominantly used for cooking fuel at home (Firewood, Kerosene Stove, Charcoal, LPG) Independent Nominal 3.4 Sampling 3.4.1 Study Population There is an ongoing facility-based surveillance in the SO and NP Districts undertaken by the Ghana Health Service, Noguchi Memorial Institute for Medical Research and the Centres for Disease Control and Prevention. Patients prospectively identified as III or SARI on this platform were selected for the study. The study population were residents of SO and NP Districts with acute respiratory infections seeking medical care. The hospitals involved were Dodowa, Akuse and Battor Hospitals. The rest were the other government health facilities within SO and NP Districts. It is worth stating that although Akuse and Battor Hospitals are located outside the SO and NP Districts, it was included based on the high patronage by some residents of SO and NP Districts. Akuse Hospital and Battor Hospital are known to have about 47.9% and 16.9% respectively of OPD attendances being residents of SO and NP Districts (Adjabeng, 2012). These hospitals were therefore included as sentinel sites under the Dodowa Influenza Population-based Surveillance (DIPS) project. 52 University of Ghana http://ugspace.ug.edu.gh 3.4.2 Sample size One hundred and forty-seven influenza virus-positive III case-patients and 294 influenza virus-negative III controls were selected. 11was assumed that 27.5% of cases and 16% of controls will have a past exposure to a risk factor for influenza-positive ILl, with an expected odds-ratio of 2.0 and power of 80%. All the above parameters hold true using the Fleiss approach (Centers for Disease Control and Prevention, 2013). The study also had 134 SARI cases and 402 unmatched hospital OPD patients without any respiratory illness selected as controls. For this study, it was assumed that 26% of cases and 15% of controls may have a past exposure to a risk factor for SARI, with an expected odds-ratio of 2.0 and power of 80%. Ratio of case to control was 1:3. The study opted to choose 3 controls to a case, based on the parameters on past exposure which made 134 cases and 402 controls more likely to be achieved within the study period. There were some considerations to have an adequate sample size noting the time available for the study to be carried out". Increasing the controls beyond 4 has a little marginal increase in precision (Wacholder, Silverman, McLaughlin, & Mandel, 1992). These parameters hold true using the Fleiss approach (Centers for Disease Control and Prevention, 2013). The 2010 census data for SO and NP districts (formerly Dangme West District) indicated a proportion of I0.2% sleeping 3 or more in a room. This was assumed as the existing exposure rate. 53 University of Ghana http://ugspace.ug.edu.gh The sample size formula for an unmatched case-control study as described by the method of Kesley et al in "Documentation for Sample Size for an Unmatched Case-Control Study" by Sullivan & Soe (2007) formed the basis for arriving at the sample size: r (p: - p2)2 and nz=s r rn n I = number of cases n2 = number of controls Z a/2 = standard normal deviate for 2-tailed test based on alpha level r = ratio of controls to cases PI = proportion of cases with exposure and q, = 1-PI P2= proportion of controls with exposure and q2 = l-p2 3.4.3 Sampling Method OPD patients at the participating health facilities were screened for IUs. The influenza- positive ILls were the case-patients for a separate comparative analysis in a 1:2 unmatched case-control study where influenza-negative Ills were selected as controls out of the pool of influenza-negative samples for ILl patients. All the influenza-negative Ills had their study identification numbers (IDs) entered in a Microsoft Excel Spreadsheet. By using the randomization syntax in Excel, the existing IDs were assigned new randomization numbers and sorted in a descending order any time new controls are added. The first 2 upper controls after arranging the negative results in a descending order were selected in relation to an 54 University of Ghana http://ugspace.ug.edu.gh influenza-positive III (case-patient). Patients who reported on weekends were not recruited as those were non-working days for the field team monitoring the study sites. SARI patients identified as part of ongoing facility-based surveillance were prospectively recruited as cases for a 1:3 unmatched case-control study. For each SARI patient, 3 out- patients without a diagnosis of a respiratory condition and resident of SO and NP Districts about same time of visit as SARI patient were tagged as controls. The proportion of SARI patients who tested positive for influenza was also assessed. Inclusion Criteria All residents of SO and NP Districts attending selected health facilities who met the case definition for III or SARI were illegible for inclusion. Exclusion Criteria ARI patients with very severe signs and symptoms who needed emergency care 3.5 Data Collection Techniques and TooIslInstruments 3.5.1 Data Collection Techniques A questionnaire was administered by trained research assistants/interviewers who were already part of the existing surveillance platform and therefore had some experience in interacting with the indigenous population attending health facilities. The interview was done in the language the respondent was most comfortable with. The questionnaire administered 55 University of Ghana http://ugspace.ug.edu.gh included information on basic household demographics, individual, socio-economic and environmental risk factors of SARI. The same questionnaire was administered to all ILl and SARI patients as well as SARI control-patients. lnfantometers, stadiometers and weighing scales were employed to measure the length, height and weight of patients respectively. Height was measured to the nearest centimeter by positioning a standard metric rule by the wall and in case of small children by a standard measuring tape. Weight was measured to the nearest 100 grams with the help of a standardized weighing machine (Pore et aI., 2010). The BMI was calculated for only participants who were aged 20 years or more as standard practice (Centers of Disease Control and Prevention, 20 11). A digital sphygmomanometer was used to measure the systolic and diastolic readings. Infra- red non-contact thermometers were used for the temperature readings after the digital and mercury thermometers were phased out due to the Ebola Virus Disease scare. Other equipment used was pulse oximeter to measure the pulse and oxygen saturation levels. Collection and Investigation of Laboratory Samples Oropharyngeal and nasopharyngeal swabs were collected from both III and SARI patients as specified by an existing protocol under the Epidemiology, Prevention and Treatment of Influenza and other Respiratory Infections Project ofNMIMR. Trained Research Assistants and Health Workers collected the oropharyngeal and nasopharyngeal swabs from patients who met the inclusion criteria. The collection of nasal and oropharyngeal swabs was to test 56 University of Ghana http://ugspace.ug.edu.gh for presence of influenza viruses. Nasal swabs have an equal or greater sensitivity than oropharyngeal swabs for detection of respiratory viruses. The addition of an oropharyngeal swab to nasal swab can increase detection of influenza viruses (Dawood et aI., 2015). The swabs were placed immediately in viral transport medium and temporary stored at a temperature ranging +2oC to +8oC or liquid nitrogen tanks in a situation where samples were to be kept on site for more than 2 days. Specimens were collected an average of 2 to 3 times weekly and sent to the National Influenza Centre (NIC) located in NMIMR which is a biosafety level-3 facility. At the NIC, specimens were kept in the liquid nitrogen tanks until they were ready for investigations via real-time reverse transcription PCR (rRT-PCR). Viral RNA was extracted from 200 III of VTM using an RNeasy Mini kit (Qiagen, Dusseldorf, Germany) per the manufacture's protocol. The RNA from each sample was tested for specific primers and probes that target Influenza A, Influenza B, pH 1N 1, seasonal influenza A (H 1N 1) and seasonal influenza (H3N2) following the US CDC's protocol. For the samples obtained from ILl patients, no further laboratory tests were carried out irrespective of the initial test outcome being influenza positive or influenza negative as per the existing protocol. 57 University of Ghana http://ugspace.ug.edu.gh Wealth Index Data on house numbers of all the study participants were taken to link the Dodowa Health and Demographic Surveillance System (DHDSS) database for available necessary demographic variables to generate the household wealth index. This data was last updated by the HDSS in 2015. Unfortunately, most participants could not immediately recall their house numbers and therefore another follow up of respondents to the communities was done. The total number of residents in the household was documented to provide information on household sizes which may be akin to overcrowding. WHO Case Definition for ILl and SARI The WHO Global Influenza Surveillance case definition for III states; "Any person with sudden onset of fever (history/measured) of ~38.0oC (axillary) and cough and/ or other respiratory signs with onset within the last 7 days". That for SARI states; "Any person with sudden onset offever (history/measured) of~38.0oC (axillary) and cough and/ or other respiratory signs with onset within the last 7 days and requires hospitalization" (World Health Organization, 2013). Below are the case definitions of ILl and SARI which were used in this study in the two districts from March to November 2016; Study ILl Case Definition: Any person with sudden onset of fever (history/measured) of ~37.50C (axillary) and cough and/ or other respiratory signs with onset within the last 10 days. 58 University of Ghana http://ugspace.ug.edu.gh Study SARI Case Definition: Any person with sudden onset of fever (history/measured) of ~37.50C and cough and/ or other respiratory signs with onset within the last 10 days and requires hospitalization. The temperature was lowered to 37.50C and period of acuteness made 10 days to make the case definition more sensitive. 3.6 Quality Control As part of increasing the validity of the data collection tools, a pre-test of the instruments was carried out in Okwenya, located in adjoining Lower Manya Krobo District. The pre- testing phase looked at ease of administering the data collection tool, reliability of the questions and assessed the data analysis plan. Findings from the pre-test greatly enhanced the validity of the data collection tool and estimated the time spent on data collection and analysis. Revision of the data collection tools and analysis was made. During the administration of the questionnaire, trained research assistants at the different participating health facilities interacted with the field staff to collect specimens and also assess data quality. The Principal Investigator conducted weekly data auditing meetings to check for inconsistencies. Anomalies were rectified immediately. The weighing scales also had standard weights put on them to ensure the readings were reliable. 59 University of Ghana http://ugspace.ug.edu.gh 3.7 Data Processing and Analysis Data were collected from March to November 2016. The completed questionnaires were entered into a database program (Epi-lnfo version 7) where personal identifiers were coded before entries made. Possible errors such as double-entries, blank fields, transpositions (where 19 became 91), range errors (e.g. age of 290 years), typographical errors and inconsistent responses like ticking of males against known female names were all checked and rectified. The data was scanned for outliers, making it possible to identify entry errors. Data was analyzed by the use ofEpi-Info 71 Stata 14.2 software. The two main outcomes of interest, were Influenza-positive ILl case-patients and SARI case- patients. These were coded as "1" whilst their controls; ILl-negative patients and non- respiratory patients at OPD were coded as "0". Similarly, the binary exposure or predictors of interest were coded as "1" and the non-exposed coded as "0". The continuous variables were approached differently. The logistic command was first used for a crude analysis in relation to the outcome. Furthermore, the same continuous variables were then split into categorical variables based on clinical description or available literature of what constituted the normal range. The normal body temperature of a person can range from 36.SoC to 37.2oC " (John Hopkins University, 2012). However, based on the cut-off point used for the fever case definition in this study, the axillary temperature recorded during physical examination, was categorized as <37.SoC ("0") and 2: 37.SoC ("1"). Therefore, the case definitions oflLl and SARI in this study had a more sensitive fever cut-off point of > 37.SoC compared to ~38.0oC and a longer 60 University of Ghana http://ugspace.ug.edu.gh 10 days of onset compared with 7 days of the World Health Organization (WHO)/ US Centers for Disease Control and Prevention (CDC) definitions. This was a deliberate effort by the existing Dodowa Influenza Population-based Surveillance Project to make the case definition more sensitive to capture influenza infections. The marital status variable had 5 possible responses. This was converted to a bivariate variable, that is "married" - "1" and "not married" ("Single", 'Divorced", "Separated" and "Widowed") - "0". The lowest monthly temperatures coincide with the rainy season and are recorded during the months of June to August. To determine the risk of climatic factors (wet season/temperature) on influenza-positive ILl or SARI, patients with date of onset of signs and symptoms in the months of June to August were coded as "1". The other months of March to May and September to November were coded as "0". Another clinical examination was oxygen saturation levels. The human body needs and controls a very detailed and specific balance of oxygen in the blood. Normal blood oxygen levels in humans ranges from 94% to 100% (Philips Medical System, 2002). Blood oxygen levels below 90% may result in hypoxemia (acute respiratory failure) and a need to consult a physician (Konica Minolta Sensing Inc., 2006). Those with readings less than 90% were coded as "1" whereas those 90% or more were coded as "0", 61 University of Ghana http://ugspace.ug.edu.gh Hypertension was defined as a systolic blood pressure of 140 mmHg or greater and/or a diastolic blood pressure of 90 mmHg or greater, drawing from the World Health Organization definition (1999). Patients with systolic blood pressure equal or more than 140 mmHg and/or a diastolic blood pressure of90 mmHg or greater were tagged as hypertensives and coded as "1". Otherwise, they were tagged as non-hypertensives and therefore coded as "0" (World Health Organization, 1999). The Body Mass Index was calculated for those aged 20 years and above by use of their height and weight measurements, classifying the categories into underweight, normal, overweight and obese. The normal category was used as the reference for computation of odds ratios. The wealth index is an amalgamated measure of a household's cumulative living standard and serves as a proxy for measuring household socioeconomic status (SES). Household wealth index was constructed by use of Principal Component Analysis (PCA), as a proxy measure of each household's socio-economic status. This process creates uncorrelated components, with each component being a linear weighted combination of the initial variables (Howe, Hargreaves, & Huttly, 2008). The assets at the household level used for the PCA included the availability of toilet facility, electricity, source of drinking water, land, house, motor, bicycle, sewing machine, television, refrigerator, radio, gas cooker, electric fan, phone, camera and cattle. All the categorical variables were re-coded into binary variables before applying the PCA. From this, households were categorized into five quintiles (i.e., poorest, poorer, poor, less poor and least poor), (Nattey, 2009). The 'least poor' was made the reference category. 62 University of Ghana http://ugspace.ug.edu.gh The influenza vaccination history was requested from all the patients under the study. Ghana does not administer seasonal influenza vaccines on routine basis. The last pandemic influenza vaccination was conducted in year 2010. Since the last vaccination took place in year 20 10, further analysis was limited to respondents aged 5 years and above. Those who had a history of vaccination were coded as "1" and those without any history were coded as "0" . There was descriptive and inferential data analysis. The descriptive data analysis helped summarize the findings from the data collection phase. The inferential statistics allowed drawing conclusions about the significant risk factors. The ILl case-patients and SARIs with their controls were described by person, place, and time characteristics. Descriptive analysis was done and it included the frequency distribution of these exposure variables and also their confidence intervals. The primary study questions analyzed included the risk factors for influenza-positive ILL the percentage of out-patients who were ILls and history of self-medications prior to hospital attendance. Confidence intervals were determined around these estimates. Similar analysis was done for SARI patients as a dependent variable. Using the Shapiro- Wilk test for normal data, all the continuous variables were tested for normality. At a significance level of 0.05, none of the variables was identified as normal therefore all the variables had the median reported instead of the mean. 63 University of Ghana http://ugspace.ug.edu.gh The median test was done to compute the probability of the median in controls being higher than the median in cases. The Chi-square test or the Fisher's exact test was used as appropriate on categorical data to test for associations of risk factors with influenza-positive ILl or SARI. A test of p< 0.05 was considered as statistically significant. Odds Ratio was used to measure the strength of association between the exposure variable and the outcome variable of influenza-positive III or SARI. The effect of potential confounders such as age, sex, number in household, Body Mass Index (BMI), second-hand smoke and Wealth Index were assessed (Chen et al., 2014; Po, FitzGerald, & Carlsten, 2011; Troko et aI., 20 11) through multiple logistic regression modelling. When there are 3 or more variables involved in a logistic regression model, it can be assumed there is no effect modification between any of them. Therefore estimate for the effect of each whilst controlling for the rest could be done (Kirkwood & Sterne, 2003). Factors associated with Influenza positive ILl or SARI atp <0.05 were identified during the crude analysis (unadjusted) and information added to the descriptive analysis. Clinical and statistical significance were some key considerations made in keeping a variable in the logistic regression model with the use of the purposeful selection approach. Multivariate Analysis The purposeful selection began by a crude analysis of each variable. Any variable that had a significant crude analysis test at an arbitrary level ofp= 0.20 was included in the multivariate analysis. This was a drift away from the more traditional cut-off point of 0.05 since known important variables could be missed. In this iterative process of variable selection, covariates University of Ghana http://ugspace.ug.edu.gh were removed from the model if they were non-significant and not a confounder. Significance of the multivariate analysis was evaluated at the 0.10 alpha level and confounding as a change in any remaining parameter estimate greater than 20% as compared to the full model. A change in parameter estimate below or above 20% indicated the excluded variable was important in the sense of adjusting one or more of the variables remaining in the model. At the end of this iterative process, the model contained significant covariates and confounders. At this point all the variables that earlier on failed to meet the criteria of selection for the initial model were added back one at a time, with significant covariates and confounders retained earlier. Any that were significant at 0.10 level were put back into the model and iteratively reduced as before but only for the variables that were additionally added. At the end of this final step, the main effects model remained (Hosmer, Lemeshow, & Sturdivant, 2013). This model/algorithm was opted for because of the comparative analysis done by Bursae et al (2008) which simulated the performance as compared with well documented procedures of Forward, Backward and Stepwise selection. The advantage of using the purposive selection approach was when the analyst was interested in risk factor modelling and not just prediction (Bursae, Gauss, Williams, & Hosmer, 2008). The model was then checked for adequacy and a goodness-of-fit test was performed to assess whether the fitted test was okay for inferential purposes. 65 University of Ghana http://ugspace.ug.edu.gh A limitation to be noted was the period between exposure to risk factors and reporting with illness at the recruiting health facility which could result in recall bias among study participants. 3.8 Ethical Consideration This proposal was submitted to the Institutional Review Boards of the Noguchi Memorial Institute for Medical Research and Dodowa Health Research Centre. 3.8.1 Informed Consent Process The informed consent document was read clearly to study participants so that they understood participation was completely voluntary. Where the English language could not be comprehended, a Research Assistant who understood English and can speak a local dialect was asked to translate to the study participants for their informed consent. A witness was present in such circumstances when a participant could not read the consent form. The witness then endorsed the witness form after a satisfactory consent had been provided. The identified participant who met the inclusion criteria was consented with the endorsement of consent forms before enrollment. If a participant was below 18 years, the parents or guardians gave their consent. In this instance, children aged 12 to 17 years were asked to provide assent as well. Parents or guardians were asked to provide needed responses on behalf of children less than 12 years. 66 University of Ghana http://ugspace.ug.edu.gh The study participant (18 years or more) or the guardian of the study participant thus if the study participant was under the age of 18 years was asked to sign the consent form provided with either a signature or a thumbprint. The consent document included information on the purpose of the study, processes involved in data collection, level of risk associated with the study and contact persons' information in case of need to lodge any complaints later on. These were read to the participant. The adult, guardian or primary caregiver's consent allowed for the following: • Collection of background household information • Asking screening questions from participants who meet the case definition • Proceeding to ask more questions about possible risk factors for respiratory infections Opportunity was given to patients to ask questions. The leaders of the data collection team provided the answers. Each participant was given one copy of the consent form for future reference. This form contained contact details of local investigators. Patients that refused to sign the informed consent document did not participate in this study. 3.8.2 Privacy of Information The names of the study participants have been kept confidential. Access to the survey data was limited to researchers working directly on this study. No personal identifiers were entered into the computer database. Instead, unique codes were entered to identify study participants. To further ensure confidentiality, no respondent name or their titles were used in the report. 67 University of Ghana http://ugspace.ug.edu.gh 3.8.3 Risks and Discomforts to Research Volunteers The participants in this study did not experience more than minimal risk. There were no invasive procedures involved and they did not incur any expenses as a result of their participation in this study. 3.8.4 Special Risks to Pregnant or Potentially Pregnant Volunteers There were no special risks to pregnant or potentially pregnant women in this study. 3.8.5 Safety Precautions and Emergency Procedures There were no additional safety precautions or emergency procedures required for this study. 3.8.6 Benefits to Research Volunteers There were no direct benefits to study participants and they were neither compensated for their time. Understanding risk factors of influenza-positive ILl or SARI in order to focus interventions is necessary. Information obtained from this survey will be used by the Ghana Health Service to design and improve health interventions towards acute respiratory diseases. 3.8.7 Maintenance of Records All completed questionnaires from the study were reviewed for completeness and legibility. Data entry was done and was followed with analysis. Digital records as well as hard copies from the study were stored in a locked cabinet. 68 University of Ghana http://ugspace.ug.edu.gh 4.0 RESULTS 4.1 Background and Individual Risk Factors of ILl Patients There was a total of 85,663 OPD attendances out of which 32,942 were residents of SO and NP districts. Out of the 3,779 respiratory cases observed among residents of SO and NP districts, 1,172 (3l.0%) met the case definition for an Ill. The proportion of ILl among residents that reported at the OPD was 3.5% (1,172 ILls' out of 32,942 residents). The study identified 147 influenza-positive cases and their corresponding 294 control-patients who were selected within the pool of influenza-negative patients. Using the Shapiro- Wilk test for normal data, all the continuous variables were tested for normality. At a significance level of 0.05, none of the independent variables was observed to be normal. The median age of cases and controls was 9 years (IQR: 3 -24 years) and 7 years (IQR: 2- 36 years) respectively. There was a 50% chance that the median of controls was higher than the median of cases and therefore not significant (Figure 4.1). 69 University of Ghana http://ugspace.ug.edu.gh control case • • Graphs by RECODE of N1StudyPart (Type Part) Figure 4.1: Boxplot on Age (years) ofILI case-patients and controls There were 8 reporting facilities involved with 3 of them being hospitals. Osudoku Health Centre and Akuse Hospital together contributed 46.0% of total ILl cases and controls. The highest proportion of cases (29.9%) was reported from Osudoku Health Centre which was followed by Akuse Government Hospital with 23 cases (15.6%). The identification of influenza positive cases from Osudoku Health Centre is high for a health centre. There were however more controls (24.5%) in Akuse Hospital than Osudoku Health Center (Table 4.1). Table 4.1: Distribution of Influenza-positive cases and controls by reporting facility Health Facility Cases (%) Control (%) Total (%) Agomeda CHPS 12 (8.16) 14 (4.76) 26 (5.90) Akuse Hosp 23 (15.65) 72 (24.49) 95 (21.54) Battor Hosp 1 (0.68) 18 (6.12) 19 (4.31) Lekpongunor CHPS 9 (6.12) 10(3.40) 19 (4.31) OldNingoHC 22(14.97) 51 (17.35) 73(16.55) Osudoku HC 44 (29.93) 64 (21.77) 108 (24.49) Prampram HC 17 (11.56) 38 (12.93) 55 (12.47) SO District Hosp 19 (12.93) 27 (9.18) 46 (10.43) Total 147 (100.00) 294 (l00.00) Pearson Chi2 (7) = 18.3650 (P = 0.010) 70 University of Ghana http://ugspace.ug.edu.gh ILl data were collected within the period of April to October 2016. The total number of residents from SO and NP Districts was less than 50% of total OPD attendance among the study facilities (Table 4.2). Table 4.2 Summary ofOPD attendance by ILl patients Number Proportion (%) Total OPD Attendance 85,663 Number of Patients from SO and NP Districts 32,942 (38.4%) Number of Respiratory Patients from SO and NP Districts 3,779 (11.4%) Proportion of SO and NP Respiratory Patients meeting the IIIcase 1,172 (31.0%) definition Number of ILl Patients Screened 448 (38.2%) Number Consented 441 (98.4%) The ILl study participants were from 106 communities within Shai-Osudoku and Ningo- Pram pram districts. The highest proportions were 96 (22.1 %) participants from Asutsuare and 29 (6.7%) from Old Ningo (Appendix VII). There was variability in the influenza organisms isolated from cases over the period of the study. Influenza virus type A(H3N2) appeared to be the predominant strain from April to August, 2016. This was replaced by Influenza Virus Type B from August to October, 20 16 (Figure 4.2). 71 University of Ghana http://ugspace.ug.edu.gh Monthly Distribution of Confirmed Influenza Types among III case-patients April to October 2016 40 IIIFLUB 35 IIIA(H3N2) 30 mA(HINI)pdm09 025 c FLU A NST ~ 20 C" ~ 15 10 5 0- I April May June July Aug Sep Oct Months Figure 4.2: Monthly Distribution of Confirmed Influenza Types among ILl case- patients The median weight of cases was 28 kg whereas that of controls was 25 kg (p=0.60). The median pulse rate was 109 bpm and 104 bpm among the cases and controls respectively. There was the same median number of persons (3) sleeping in the same room with either cases or controls but median number of persons living in the household was 6 among controls as compared with 5 among cases. The median of the oxygen saturation level was 98% and 97% among cases and controls respectively. The median oxygen saturation level of influenza-positive Ills had a probability of 56% being more than that of the influenza- negative ILls. The diastolic and systolic median readings were lower in cases than in controls. For the systolic, median for ILl case-patients was 120 mmHg and that of controls was 128.5 mmHg. The systolic median in controls had a 62% (p=O.OI) probability to be higher than that of the 72 University of Ghana http://ugspace.ug.edu.gh systolic median among cases. On the other hand, diastolic median was 70 mmHg and 80 mmHg respectively among the influenza-positive cases and influenza-negative controls. The diastolic median among controls had a proportion of 66% (p=O.OOI) chance to be higher than that of the cases (Table 4.3). Table 4.3 Univariate distribution (median) on selected variables ofILI cases and controls Name of Variable No. Median IQR Range P-value for Proportion proportion (var in test cont>case) Age (Yrs) Controls 294 7 2 - 36 2 months, 85 years 0.85 0.505 Cases 147 9 3 - 24 3 months, 86 years Pulse Rate Controls 287 104 86 - 124 10, 171 0.20 0.463 Cases 146 109 94 - 122 58,160 Oxygen Saturation Controls 285 97 85 - 99 52,100 0.04 0.442 Cases 146 98 94 - 99 70,108 Respiratory Rate Controls 289 24 20 - 30 15,67 0.24 0.466 Cases 146 26 22 - 30 15,84 Number in Household Controls 294 6 4 - 8 1,30 0.42 0.523 Cases 147 5 4 - 7 1,40 Number Sleeping in same room Controls 292 3 2-4 I, 21 0.005 0.40 Cases 144 3 2-4 1,30 BP Systolic Controls 126 128.5 110-145 85,211 0.01 0.621 Cases 49 120 106 - 130 60,194 BP Diastolic Controls 126 80 (70 - 90) 47,125 0.001 0.660 Cases 48 70 60 - 80 50, 100 73 University of Ghana http://ugspace.ug.edu.gh Table 4.4 presents the results of the association of the various variables grouped under individual risk factors and an outcome of influenza-positive ILl. The independent variables of Smoke, BMI and Exposure to Smoke were independently not associated with an Influenza-positive ILl at p=0.05 significance level. The BMI had the 'normal' category as the reference compared with 'underweight', 'overweight' and 'obese'. Vaccination had a zero cell on whether ILl positive cases had received vaccination in the past, however 11 (3.74%) controls indicated a history of vaccination. Vaccination status as a variable was therefore omitted from further analysis. There were more females than males in the case (55.1 %) and control (67.4%) groups although it was not significantly associated with a positive influenza infection. The highest proportion of participants belonged to the age group of less than 5 years. Fifty-two case- patients (35.4%,p= 0.034) were below 5 years of age. This was significant in comparison to the reference group of 60 or more years. The variable age categorized into the standard WHO age groupings was significantly associated with a positive influenza associated ILl with the age group 60 years or more as the reference group (World Health Organization, 2013). The only exception was the 45 to <60-year group. Children less than 5 years had odds of3.25 times more to have an influenza positive infection among ILls as compared with the reference age group of> 60 years (Table 4.4). 74 University of Ghana http://ugspace.ug.edu.gh Table 4.4 Association of Individual background risk factors with Influenza Positive ILl Case Patients and Controls Risk Factors Cases, Control, Total, N OR (95% c.u PValue n(%) n(%) (%) Sex Female 8] (55.1) 171 (67.41) 252(57.1) 1.00 0.54 Male 66 (44.9) 123 (32.59) 189 (42.9) 1.13 (0.74 - 1.72) Age <5 52 (35.37) 128 (43.5) 180 (40.82) 3.25 (1.09 - 9.64) 0.034 5 to <15 39 (26.53) 40 (13.61) 79 (17.91) 7.80 (2.52 - 24.12) <0.00] 15 to <30 25(17.01) 40 (13.61) 65 (14.74) 5.00 (1.57 - 15.84) <0.001 30 to <45 15 (10.20) 24 (8.16) 39 (8.84) 5.00 (1.47 -16.99) 0.0]0 45 to <60 12 (8.16) 30 (10.20) 42 (9.52) 3.2 (0.92 - 11.01) 0.065 >= 60 4 (2.72) 32 (10.88) 36(8.16) 1.00 Smoker No 140 (97.90) 280 (96.55) 420 (97.00) 1.00 0.559 Yes 3 (2.10) 10 (3.45) 13 (3.00) 0.60 (0.10 - 2.38) Exposure to Smoke No 117 (85.40) 249 (86.76) 366 (86.32) 1.00 0.703 Yes 20 (14.60) 38 (13.24) 58 (13.68) 1.12 (0.59 - 2.07) BMI underweight I (2.38) 9 (7.69) 10(6.29) 0.33 (0.03 - 2.96) 0.282 normal 17 (40.48) 52 (44.44) 69 (43.40) 1.00 overweight 17 (40.48) 31 (26.50) 48 (30.19) 1.67 (0.74 - 3.78) obesity 7 (16.67) 25 (21.37) 32 (20.13) 0.85 (0.31 - 2.34) Vaccination Status Not Vaccinated 147(100.0) 283 (96.26) 430 (97.51) 1.00 0.019 Vaccinated 0(0.00) 11 (3.74) II (2.49) Zero cell (F) (F) - Fishers Exact Test Chills was not a significant sign associated with influenza-positive ILls as the 95%CI contained 1.0. Diarrhoea also appeared protective (OR: 0.49; 95%CI: 0.22 - 0.99). The rest of the clinical signs and symptoms presented at the health facility can be described as not associated with the influenza-positive ILl during the crude analysis (Table 4.5). Pulse Rate, sore throat, runny nose, abdominal pain, chest pain, vomit, shortness of breath, difficulty in breathing, and myalgia were all not significant at the p=0.05 at the crude or adjusted analysis 75 University of Ghana http://ugspace.ug.edu.gh phases. Due to the fact that Fever and Cough formed a basis of a definition for an ILl, it was common to both cases and controls and were therefore omitted from the analysis. Table 4.5 Association of Signs and Symptoms with Influenza-positive ILl Case Patients and Controls Predictors Cases, Control, Total, OR (95% C.I.) P Value n(%} n(%} N(%} Pulse Rate Cut Off No 56 (38.36) 125 (43.71) 181 (41.90) 1.00 0.286 Yes 90 (61.64) 161 (56.29) 251 (58.10) 1.24 (0.81 - 1.91) Sorethroat No 97 (66.90) 201 (69.07) 298 (68.35) 1.00 0.66 Yes 48 (33.10) 90 (30.93) 138 (31.65) 1.10 (0.70 - 1.72) Runny Nose No 30 (20.41) 74 (25.34) 104 (23.69) 1.00 0.251 Yes 117 (79.59) 218 (74.66) 335 (76.31) 1.32 (0.80 I - 2.22) Chills No 57 (39.04) 145 (49.66) 202 (46.12) 1.00 0.036 Yes 89 (60.96) 147 (50.34) 236 (53.88) 1.54 (1.01 - 2.35) Diarrhoea No 134(91.78) 248 (84.64) 382 (87.02) 1.00 0.036 Yes 12 (8.22) 45 (15.36) 57 (12.98) 0.49 (0.22 - 0.99) Abdominal Pain No 109 (74.66) 217 (74.06) 326 (74.26) 1.00 0.893 Yes 37 (25.34) 76 (25.94) 113(25.74) 0.96 (0.59 - 1.56) Chest Pain No 84 (57.14) 179 (61.51) 263 (60.05) 1.00 0.378 Yes 63 (42.86) 112 (38.49) 175 (39.95) 1.19 (0.73 - 1.82) Vomit No 112 (76.71) 222 (75.77) 334 (76.08) 1.00 0.827 Yes 34 (23.29) 71 (24.23) 105 (23.92) 094 (0.54 - 1.54) Shortness of Breath No 120 (81.63) 229 (78.42) 349 (79.50) 1.00 0.432 Yes 27 (18.37) 63 (21.58) 50 (20.50) 0.81 (0.47- 1.38) Difficulty in Breathing No 127 (86.39) 251 (85.96) 378 (86.10) 1.00 0.901 Yes 20 (13.61) 41 (14.04) 61 (13.90) 0.96 (0.51 -1.76) Myalgia No 83 (56.46) 156 (53.42) 239 (54.44) 1.00 0.546 Yes 64 (43.54) 136 (46.58) 200 (45.56) 0.88 (0.58 - 1.34) Crude analysis on the presence of existing medical conditions being a risk factor for influenza-positive ILl did not indicate any significant associations: Asthma, Chronic Heart 76 University of Ghana http://ugspace.ug.edu.gh Disease and Diabetes. It is worth noting that due to zero cells obtained in the crude analysis of variables such as existence of Chronic Liver Disease, HIV, TB, Malnutrition and Vaccination Status, they were omitted from further analysis (Table 4.6). Table 4.6 Association of existing medical conditions as risk factors for ILl Case Patients and Controls Risk Cases, Control, Total, OR (95% Cl.) p-value Factors n{%} n{%} N {%} Asthma No 144 (97.96) 278 (94.56) 422 (95.96) 1.00 0.135 Yes 3 (2.04) 16 (5.44) 19 (4.31) 0.36 (0.06 - 1.29) (F) CHD No 142 (96.60) 271 (92.18) 413 (93.65) 1.00 0.096 Yes 5 (3.40) 23 (7.82) 28 (6.35) 0.41 (0.12 -1.14) (F) Diabetes No 146 (99.32) 287 (97.62) 433 (98.19) 1.00 0.279 Yes I (0.68) 7 (2.38) 8(1.81) 0.28 (0.00 - 2.22) (F) - Fishers Exact Test 4.2 Socio-economic risk factors relating to Influenza positive ILls Only occupational status out of the exposure variables considered under socioeconomic factors was significant at p < O.OOI(F).Students formed the highest proportion (28.3%) of cases and 30.5% of controls followed by artisans who were 26.9% among the case-patients, compared with 4.2% of artisans in the control group. Out of the various categories of occupation indicated, the artisan group showed they were 6.24 times more likely to be influenza positive than the controls using unemployed category as the reference group. Farming, Trading, Civil Servants and students were not associated with a positive influenza infection. 77 University of Ghana http://ugspace.ug.edu.gh There was no significant relationship between marital status and outcome of influenza- positive infection [OR:1.31 (95%CI: 0.62 - 2.78)] for respondents above 18 years. Other variables such as Literacy level of care giver, attribution of illness to curse, perception on causes of ARI and wealth index were not associated with an influenza-positive ILl (Table 4.7). The wealth index had 'least poor' as the reference category compared with other categories such as 'poorest', 'poorer', 'poor' and' less poor'. 78 University of Ghana http://ugspace.ug.edu.gh Table 4.7 Summary of socio-economic factors for Influenza-Positive ILls Risk Factors Cases Control Total OR P Value Educational Status No Education 8 (12.7) 14 (10.2) 22 (11.0) 1.00 0.848 Primary 49 (77.9) 108 (78.8 157 (78.5) 0.79 (0.31 - 2.02) Secondary 3 (4.7) 10(7.3) 13 (6.5) 0.52 (0.10 - 2.58) Tertiary 3 (4.7) 5 (3.6) 8 (4.0) 1.0(0.]9-5.76) Literacy Level of Care Givers Above primary 52(61.9) 91 (58.0) 143 (59.3) 1.00 0.553 Primary and below 32 (38.1) 66 (42.0) 98 (40.7) 0.77 (0.38 - 1.56) Marital Status Not Married 20 (44.44) 61 (51.26) 81 (49.39) 1.00 0.436 Married 25 (55.56) 58 (48.74) 83 (50.61) ] .31 (0.62 - 2.78) Occupation (2:18yrs) Artisan 12 (26.09) 5 (4.24) 17 (10.37) 6.24 (1.20 - 32.39) <0.001 (F) Civil Servant 5 (10.87) II (9.32) 16(9.76) 1.18 (0.26 - 5.29 Farmer 4 (8.70) 28(23.73) 32 (19.51) 0.37 (0.08 - 1.68) Trader 4 (8.70) 10 (8.47) 14 (8.54) ] .04 (0.21 - 5.03) Student 13 (28.26) 36 (30.51) 49 (29.88) 0.93 (0.27 - 3.18) Others 3 (6.52) 15 (12.71) 18 (10.98) 0.52 (0.09 - 2.70) Unemployed 5 (10.87) 13(11.02) 18 (10.98) 1.00 (Ref) Attribution of ARI to Curse No 140 (95.24) 272 412 1.00 0.315 (F) (92.52) (93.42) Yes 7 (4.76) 22 (7.48) 29 (6.58) 0.61 (0.21 - 1.54) 0.277 Wealth Index Poorest 30 (20.41) 57 (19.66) 87(19.91) 1.14 (0.61 -2.13) 0.801 Poorer 33 (22.45) 61 (21.03) 94 (21.51) 1.17 (0.63 - 2.16) Poor 24 (16.33) 58 (20.00) 82(18.76) 0.89 (0.46 - 1.72) Less Poor 31 (21.09) 51 (17.59) 82 (18.76) ] .32 (0.70 - 2.47) Least Poor 29 (19.73) 63 (21.72) 92 (2] .05) 1.00 Health education on preventive measures concerning respiratory diseases will be informed by the perception of causes of ILls. The majority of respondents for cases (77.5%) and controls (74.8%) did not perceive any known cause ofILI. They responded "Don't know". 79 University of Ghana http://ugspace.ug.edu.gh The second most mentioned perception was the "Weather" in both cases (7.5%) and also the controls (7.5%), (p = 0.326), (Appendix VIU). Closely linked to perception of causes of ILl were ways that it could be prevented. Similar to the observations made with the perception of causes, about 78% of either cases or their controls mentioned "Don't know" when asked how to prevent Ills. Only about 8.9% of total cases together mentioned; "taking medicine", "cough etiquette", "avoid dust" and "wearing nose mask/protective clothing". This can be compared with 10.3% of controls mentioning the same preventive measures (p = 0.148), (Appendix IX). 4.3 Environmental Risk Factors for Influenza-positive ILls The independent variables of Travel History, Onset of illness within June to August (wet season), Exposure to secondary cigarrete smoke, 3 or more persons sleeping in a room and 6 or more persons in a household were considered as environmental risk factors for an influenza positive infection. The others were the use of the following domestic fuels for household cooking: firewood, charcoal, LPG and kerosene. Only the onset of illness during June to August, the wet season (OR: 0.62, 95% CI: 0.41 - 0.95), and 3 or more persons sleeping in a room (OR: 1.71, 95% CI: 1.09 - 2.72) were independently significantly associated with an influenza positive infection at p = 0.05. The onset of signs and symptoms in the wet season appeared to be 'protective' against a positive influenza infection. Such patients had a 38% chance reduction in being diagnosed as influenza positive (Table 4.8). 80 University of Ghana http://ugspace.ug.edu.gh Table 4.8 Crude Anal~sis of Environmental Risk Factors and Influenza-~ositive ILl Risk Factors Cases Control Total OR (!J5%C/) PValue Travel History Not Travelled 125 (87.4) 255 (88.2) 380 (88.00) 1.00 0.805 Travelled 18 (12.6) 34 (11.8) 52 (12.00) 1.08 (0.55 - 2.05) Wet Season No 78 (53.06) 122(41.50) 200 (45.35) l.00 0.021 Yes 69 (46.94) 172 (58.50) 241 (54.65) 0.62 (0.41 - 0.95) Exp. to Sec Smoker No 117 (85.4) 249 (86.76) 366 (86.32) 1.00 0.703 Yes 20 (14.6) 38 (13.24) 58 (13.68) 1.12 (0.59 - 2.07) ~ 3 Sleeping in Room No 40 (27.2]) 115(39.12) 155(35.15) 1.00 0.014 Yes ]07 (72.79) 179 (60.88) 286 (64.85) l.7] (1.09 - 2.72) ~ 6 Living in Household No 77 (52.38) 140 (47.62) 217 (49.21) 1.00 0.346 Yes 70 (47.62) 154 (52.38) 224 (50.79) 0.52 (0.54 - ] .25) Firewood No 87 (59.18) 166 (56.46) 253 (57.37) 1.00 0.586 Yes 60 (40.82) 128 (43.54) 128 (43.54) 0.89 (0.58 - 1.36) Charcoal No 12 (8.16) 34 (11.56) 46 (10.43) 1.00 0.271 Yes 135 (91.84) 260 (88.44) 395 (89.57) 1.47 (0.71 - 3.22) Liquefied Petroleum Gas No 84 (57.14) 182 (6l.90) 266 (60.32) l.00 0.335 Yes 63 (42.86) 112(38.10) 175 (39.68) 1.21 (0.79 - 1.85) Kerosene No 1 (0.68) 292 (99.32) 438 (99.32) 1.00 1.000 (F) Yes ]46 (99.32} 2 (0.68} 3 (0.68) 1.00 (0.01 - 19.35) 4.3.1 Adjusted Logistic Regression for Risk Factors of ILl After adjusting for covariates and confounders in the multivariate regression, chills (OR: 4.S7, 9S%Cl: 1.S1 - 13.76) and history of travel (OR:3.0S, 9S%Cl: 1.07 - 8.73) were significant factors for influenza-positive Ills at p < 0.05 in the presence of other covariates. Important covariates were ages IS to 60 years, Educational Status, Runny Nose, Chills, Wet Season, Chronic Heart Disease and Occupation (Table 4.9). 81 University of Ghana http://ugspace.ug.edu.gh Table 4.9 Final Output Table of Crude and Adjusted Logistic Regression Model for Influenza-positive ILl Risk Factors Variables Crude (95% CI) Adjusted (95% CI) Odds Ratio P>z Odds Ratio P>z Travelled 1.08 (0.55 - 2.05) 0.805 3.05 (1.07 - 8.73) 0.037 Age (years) <5 3.25 (1.09 - 9.64) 0.034 - - 5 to 14 7.80 (2.52 - 24.12) <0.001 - - IS to 29 5.00 (1.48 - 16.78) <0.001 2.81 (0.57-13.71) 0.201 30 to 44 5.00 (1.36 - 18.29) <0.001 5.26 (0.99 -27.92) 0.281 45 to 60 3.20 (0.89 - 11.44) <0.01 2.53 (0.51 - 12.59) 0.255 >=60 (Ref) 1.00 1.00 Educ. Status Primary 0.79 (0.31 - 2.02) 0.62 0.81 (0.22 - 3.01) 0.764 Secondary 0.52 (0. I0 - 2.58) 0.42 0.28 (0.04 - 1.94) 0.199 Tertiary 1.00 (0.19-5.76) 0.95 0.50 (0.05 - 4.56) 0.546 Runny Nose 1.32 (0.80 - 2.35) 0.251 1.24 (0.51 - 3.06) 0.626 Chills 1.54 (1.00 - 2.35) 0.036 4.57 (LSI -13.76) 0.007 Wet Season 0.62 (0.41 - 0.95) 0.021 0.69 (0.29 - 1.61) 0.398 Chronic Heart Disease 0.41 (0.12-1.14) 0.096 (F) 0.56 (0.12 - 2.54) 0.455 Occupation Artisan 6.24 (1.20 - 32.59) 0.012 2.55 (0.32 - 20.27) 0.375 Civil Servant 1.18 (0.26 - 5.29) 0.827 0.55 (0.08 - 3.57) 0.533 - Farmer 0.37 (0.08 - 1.68) 0.181 0.18 (0.03-1.19) 0.076 Trader 1.04 (0.21 - 5.03) 0.961 0.42 (0.06 - 2.96) 0.389 Student 0.93 (0.27 - 3.18) 0.919 0.38 (0.07 - 1.91) 0.242 Others 0.52 (0.09 - 2.70) 0.429 0.23 (0.03 - 1.77) 0.161 The final model was predictive of an influenza-positive ILl patient as it could not be rejected with a Goodness-of-fit test (Prob > Chi2 = 0.4334). The area under the Receiver Operating Characteristics curve (ROC curve) was 0.8037 and can be described as good in separating the Influenza-positive Ills and Influenza negative ILls based on the independent variables 82 University of Ghana http://ugspace.ug.edu.gh of age, Educational Status, Runny Nose, Onset in Wet Season, CHO and Occupation in the final model (Figure 4.3). oo .?;- ~ c'"~O . Q) (/) Number of Groups = 10 0.0 N HL Chi 2(8) = 8.00 ci Prob> Chi2 = 0.4334 Correctly classified = 78.62% 0.00 0.25 0.50 0.75 1.00 1 - Spe cifi city Area under ROC curve = 0.8037 Observations = 159, (ROC Area: 0.8037; 95% CI: 0.728, 0.879) Figure 4.3: Goodness of fit test (ROC Curve) for Influenza Positive ILls The test had a high specificity of92.17% (Table 4.10). Table 4.10 Sensitivity and Specificity of Adjusted ILl Predictive Model Classified D -0 Total + 19 8 27 - 27 109 136 Total 46 117 163 Sensitivity = 43.18%, Specificity = 92.17%, PPY = 67.86%, NPY = 80.92% 4.4 Background Characteristics of Severe Acute Respiratory Infections Twenty-six (19.4%) of the 134 SARI cases identified during the period were positive for an influenza virus. Fifty four percent of the total positives were due to Influenza A(H3N2), (Table 4.11). 83 University of Ghana http://ugspace.ug.edu.gh Table 4.11 Laboratory Results of SARI Case-Patients The number of SARI cases identified over the study period increased gradually from March till it peaked in June to August of same year. From the chart (Figure 4.4), it is evident that Flu B virus replaced the A(H3N2) and A(HIN1)pdm09 viruses during the second part of the year, beginning in July 2016. A similar observation was made among the ILl positive case- patients. Monthly Reported SARI Patients and Laboratory Results - 2016 30 i!!l Patients 25 .A(H3N2) >-.20 l1li A(H1Nl)pdm09 (.) c III Flu B ~ 15 0"" ~case) Controls 402 Age (Yrs) 36 20 - 56 <0.001 0.707Cases 134 6 2 - 35 Axillary Temp Controls 399 37.00 36.5 - 37.7 <0.001 0.327Cases 134 37.75 36.8 - 38.4 Controls Pulse Rate 397 93 81 - 108 <0.001 0.323Cases 134 108 96 - 120 Oxygen Saturation Controls 297 84 76 - 98 0.304 0.513Cases 134 86 77 - 98 Respiratory Rate Controls 397 22 20 - 25 <0.001 0.305 Cases 134 28 22 - 32 Number in Household Controls 402 7 5 - 10 0.758 0.491Cases 134 8 4 -II Number Sleeping in Controls 395 2 2-3 <0.001 0.384 same room Cases 130 3 2-4 Controls 402 3 2 -5 0.012 0.429 Duration Cases 134 4 3-5 4.5 Individual Risk Factors for SARI The proportion of males (50.7%) and females (49.3%) among the SARI cases were almost equal. However, females formed 67.4% of SARI controls (p=<0.000 1). Males had 2.13 higher odds of being classified as a SARI case. When a distribution was done by sex among the 26 influenza SARI positives and 108 influenza SARI negatives, there was no significant relationship between sex and influenza status (OR: 0.65; 95% CI: 0.27 - 1.55). Using age categories of < 5 years, 5 to < 15, 15 to <30, 30 to <45, 45 to <60 and 2: 60 years, respondents less than 5 years among the cases formed the majority of 42.5% compared with 87 University of Ghana http://ugspace.ug.edu.gh 11.4% among controls. (p = <0.0001). The less than 5 years had an odds ratio of 5.83 (95% CI: 2.86 - 11.86), (Refer Table 4.15). Table 4.15 Association ofIndividual background risk factors with SARI Case Patients and Controls Risk Factors Cases, Control, Total, N (%) OR (95% ci.i PValue n(%) n(%) Sex Female 66 (49.25) 271 (67.41) 337 (62.87) 1.00 <0.0001 Male 68 (50.75 131 (32.59) 199(37.13) 2.13 (1.40 - 3.23) Total 134 402 Age (years) agecat60 <5 57 (42.54) 46 (11.44) 103 (19.22) 5.83 (2.86 - 11.86) <0.001 5 to <15 25 (18.66) 33 (8.21) 58 (10.82) 3.56 (1.64 - 7.70) 15 to <30 14(10.45) 87 (21.64) 101 (18.84) 0.75 (0.34 - 1.63) 30 to <45 II (8.21) 85 (21.14) 96 (17.91) 0.60 (0.26 - 1.38) 45 to <60 10(7.46) 10(17.46) 81(15.11) 0.66 (0.28 - 1.54) >= 60 17 (12.69) 80 (19.90) 97 (18.70) 1.00 Total 134 402 Smoke No 123 (91.79) 388 (96.52) 511 (95.34) 1.00 0.024 Yes II (8.21) 14 (3.48) 25 (4.66) 2.47 (1.09 - 5.60) Total 134 402 Self Medication No 69 (51.49) 288 (71.64) 357 (66.60) 1.00 <0.001 Yes 65 (48.51) 114 (28.36) 179 (33.40) 2.37 (1.5 - 3.6) Total 134 402 BMI under 8 (15.69) 39 (12.79) 47 (13.20) 0.99 (0.40 to 2.44) 0.576 normal 102 (33.44) 21(41.18) 123 (34.55) 1.00 overweight 8 (15.69) 59 (19.34) 67 (18.82) 0.65 (0.27 - 1.58) obesity 14 (27.45) 105 (34.43) 119 (33.43) 0.64 (0.31 - 1.34) Total 51 305 Vaccination Status Not Vaccinated 130 (97.01) 383 (95.27) 513 (95.71) 1.00 0.469 (F) Vaccinated 4 (2.99) 19 (4.73) 23 (4.29) 0.62 (0.15 - 1.91) Total 134 392 A high pulse rate (below 60 or above 100 beats per minute), Sore Throat, Diarrhoea, Vomit and Shortness of breath were signs and symptoms significantly associated with a SARI case 88 University of Ghana http://ugspace.ug.edu.gh at p= 0.05 and a 95% Confidence Interval. That of Runny Nose and Chills although were significant at p=0.05 were omitted because the 95% Confidence Interval contained 1.0. Abdominal pain and Difficulty in breathing were not significantly associated with SARI at the crude analysis phase (Refer Table 4.16). 89 University of Ghana http://ugspace.ug.edu.gh Table 4.16 Association of Signs and Symptoms with SARI Case Patients and Controls Risk Factors Cases, Control, Total, OR (95% C.L) P Value n (%} n (%) N(%) Pulse Rate Cut Off 47 (35.07) 251 (63.22) 298 (56.12) 1.00 <0.001 87 (64.93) 146 (36.78) 233 (43.88) 3.18 (2.07 -4.90) Total 134 397 Sorethroat No 83 (62.41) 254 (63.34) 337(63.11) 1.00 0.9174 Yes 50 (37.59) 147 (36.66) 197 (36.89) 1.04 (0.67 - 1.58) Total 133 401 Runny Nose No 58(43.61) 213 (53.52) 271 (51.04) 1.00 0.048 Yes 75 (56.39) 185 (46.48) 260 (48.96) 1.48 (0.98 - 2.25) Total 133 398 Chills No 48 (36.36) 186 (46.50) 234 (43.98) 1.00 0.042 Yes 84 (63.64) 214 (53.50) 298 (56.02) 1.52 (0.99 - 2.33) Total 132 400 Diarrhoea No 95 (71.43) 358 (90.86) 453 (85.96) 1.00 <0.001 Yes 38 (28.57) 36(9.14) 74 (14.04) 3.97 (2.31 - 6.82) Total 133 394 Abdominal Pain No 88 (66.67) 284 (71.36) 372 (70.19) 1.00 0.307 Yes 44 (33.33) 114 (28.64) 158 (29.81) 1.24(0.79 - 1.93) Total 132 398 Vomit No 50 (37.88) 343 (86.18) 408 (76.98) 1.00 <0.001 Yes 82 (62.12) 55 (13.82) 122 (23.02) 2.20 (1.44 - 3.38) Total 132 398 Shortness of Breath No 67 (50.38) 300 (75.38) 367(69.11) 1.00 <0.001 Yes 66 (49.62) 98 (24.62) 164 (30.89) 3.01 (1.95 - 4.63) Total 133 398 Difficulty in Breathing No 79 (59.40) 233 (58.54) 312 (58.76) 1.00 0.862 Yes 54 (40.60) 165 (41.46) 219 (41.24) 0.96 (0.63 - 1.46) Total 133 398 90 University of Ghana http://ugspace.ug.edu.gh For existing medical conditions, only Chronic Heart Disease (CHD) appeared to be associated with a SARI case-patient though appearing protective by having a 60% lower chance of being classified as a SARI case-patient compared to the SARI control-patients (OR: 0.40; 95% CI: 0.21 - 0.73) which does not appear to be medically plausible. This crude analysis needs to be interpreted with caution as the responses were not solely based on clinical notes but self-reported as well. Ninety-two (79.0%) out of the 116 respondents who reported having CHD were through self-reporting. When further comparative crude analysis was done on self-reporting versus clinical notes, the relationship with CHD was not significant (OR: 0.75; 95%CI: 0.21 - 2.57), (Table 4.17). Chest pain can easily be confused with heart disease (Mayo Clinic, 2014). Other underlying conditions assessed such as Chronic Liver Disease, Chronic Lung Disease, Cancer, Immunosuppression, HIV, Tuberculosis and Malnutrition were dropped due to the low numbers in the cells. 91 University of Ghana http://ugspace.ug.edu.gh Table 4.17 Association of existing medical conditions as risk factors for SARI Case Patients Risk Factors Cases, Control, Total, OR (95% c.r.) P Value n (%) n (%) N(%) Asthma No 126 390 (97.01) 516 (96.27) 1.00 0.120 (94.03) Yes 8 (5.97) 12 (2.99) 20 (3.73) 2.06 (0.71 - 5.62) 134 402 Chronic Heart Disease No 118 302 (75.12) 420 (78.36) 1.00 0.001 (88.06) Yes 16 (11.94) 100 (24.88) 116 (21.64) 0.40 (0.21 - 0.73) 134 402 Diabetes No 128 393 (97.76) 521 (97.20) 1.00 0.222 (95.52) (F) Yes 6 (4.48) 9 (2.24) 15 (2.80) 2.04 (0.58 - 6.57) Total 134 402 4.6 Socio-economic Risk Factors for SARI Only the Educational status and Marital status were significant for a classification of a SARI case-patient at a level of p= 0.05. A primary level of education constituted the majority of 73.1% among cases and 56.7% among controls (p= 60 1.00 1.00 Diabetes 2.04 (0.58 - 6.57) 4.52 (1.17 -17.45) 0.028 Sorethroat 1.04 (0.67 - 1.58) 2.51 (1.20-5.24) 0.014 Runny Nose 1.48 (0.98 - 2.25) 0.49 (0.22 - 1.11) 0.089 Diarrhoea 3.97 (2.31 - 6.82) 4.09 (1.38 - 12.08) 0.011 Chills 1.52 (0.99 - 2.33) 4.27 (1.74 - 10.49) 0.002 Asthma 2.06 (0.71 - 5.62) 3.68 (0.88 - 15.40) 0.073 CHD 0.40 (0.21 - 0.73) 0.95 (0.34 - 2.63) 0.926 SOCIOECONOMIC FACTORS Educational Status Primary 2.19 (1.28 - 3.72) 0.89 (0.34 - 2.31) 0.816 Secondary 0.90 (0.40 - 2.06) 0.93 (0.27 - 3.22) 0.919 Tertiary 2.19 (0.71 -7.45) 2.19 (0.36 - 13.29) 0.393 Marital Status 0.54 (0.29 - 0.99) 0.68 (0.31 - 1.49) 0.347 Attribution to Curse 1.33 (0.81 - 2.16) 2.44 (1.08 - 5.54) 0.032 ENVIRONMENTAL FACTORS Wet Season (Jun to Aug) 2.04 (1.34 - 3.10) 2.91 (1.27 - 6.67) 0.011 Liquefied Petroleum Gas 0.95 (0.62 - 1.46) 0.37 (0.15 - 0.93) 0.036 Charcoal 0.43(0.22 - 0.86) 1.90 (0.46 - 7.85) 0.372 2: 3 Sleeping in Room 2.30 (1.50- 3.54) 0.94 (0.47 - 2.24) 0.940 Firewood 0.70 (0.46 - 1.07) 0.59 (0.24 - 1.47) 0.263 Goodness of fit test for SARI Risk Factors The final model was predictive ofa SARI patient as it could not be rejected with a Goodness- of-fit test (Prob > Chi2 = 0.8238). The area under the Receiver Operating Characteristics curve (ROC curve) was 0.8482 and can be described as good in separating the SARI patients and SARI controls based on the independent variables in the final model. (Figure 4.5). 97 University of Ghana http://ugspace.ug.edu.gh oo~ Number of Groups = 10 to No HL Chi 2(8) = 3.48 Prob> Chi2 = 0.9006 Correctly classified = 88.95% o oo ,_--------~----------~--------~----------~ 0.00 0.25 0.50 0.75 1.00 1 - Specificity Area under ROC curve = 0.8482 Figure 4.5: Goodness of fit test (ROC Curve) for SARI Risk Factors Observations = 353, (ROC Area: 0.8482; 95% CI: 0.7968,0.8999) The test had a high specificity of98.68% (Table 4.21). Table 4.21 Sensitivity and Specificity of Adjusted ILl predictive Model Classified D -D Total + 15 4 19 - 35 299 334 Tota.l. 50 303 353 Sensitivity = 30.00%, Specificity = 98.68%, PPY = 78.95%, NPY = 89.52% 98 University of Ghana http://ugspace.ug.edu.gh 5.0 DISCUSSION This study unlike many others sought to establish the risk factors associated with the influenza-positive ILls and SARI in a whole population and not necessarily restricted to children under S years. It made use of an existing surveillance structure under the DIPS protocol which focused on a defined catchment area of SO and NP districts, largely rural districts, a similar design used by Peng, Xu, et al. (20 IS). The discussion has been organized along the lines of individual, socioeconomic and environmental risk factors associated with an ILl positive influenza infection and also a SARI patient. Although the ultimate goal of the Global Influenza Surveillance Programme is to detect circulating influenza viruses, using the surveillance platform to study risk factors associated with influenza-positive ILl and SARI is another way to deploy the surveillance platform. Akuse Government Hospital still featured as one of the health facilities with high attendance by residents of SO and NP Districts (Adjabeng 2011). Of interest was the highest proportion of influenza positive ILl cases (29.9%, p = 0.01) reported by Osudoku Health Centre which is not a hospital. Fifty-nine percent (9S%CI: 4S.22 - 64.77%) of the participants identified at the health centre had their address located in Asutsuare. It was documented that Influenza A(H3N2) and Influenza B have been the predominant circulating subtypes in the study area corroborating Ntiri et al. (2016). That study demonstrated the interchanging predominance for different periods since May 2013 in the study area with the only exception being 2014 when Influenza B was predominant throughout the year. 99 University of Ghana http://ugspace.ug.edu.gh 5.1 Individual Risk Factors for Influenza-positive ILl The highest proportion of participants belonged to the age group ofless than 5 years (35.4%, p= 0.034) reference to the 60 or more years age group which was similar to other studies conducted by Cheng et al (2017), Comas-Garcia et al. (2011) and Taubenberger & Morens (2008) but deaths are known to occur more in the older age groups (WHO 2010). Ntiri et al. (2016) in a study at SO and NP Districts looked at records of 2,322 ILl patients tested from 2014 to 2016, 407 (18%) were positive for influenza. The estimated incidence of influenza-positive ILl was 844 per 100,000 persons (95% CI: 501-1,099). This study did not indicate any significant relationship of sex as a risk factor (OR= ] .13; 95%CJ: 0.74 - 1.72) for influenza infection when female was the reference group. There were more females than males in the case (55.] %) and control (67.4%) groups. In a study conducted at the Agogo Presbyterian Hospital, in Ghana, females constituted 57% of the influenza patients less than 15 years (Hogan et aI., 20] 7). For developed countries, such as the United States of America and Spain, influenza infection was higher in males (up to 60% in the United States) than females of diverse ages (World Health Organization, 20] 0). The standard age groupings « 5 years, 5 to <15 years, 15 to < 30 years, 30 to < 45 years, 45 to < 60 years, ::0:60years) for WHO used was significantly associated (p< 0.00]) with a positive-influenza associated ILls, with the age group 60 years or more as the reference group. The only exception was the 45 to <60-year group. The odds of being an influenza- 100 University of Ghana http://ugspace.ug.edu.gh positive case among children less than S years was 3.2S times higher than the odds of being an Influenza-positive case for the age group of'> 60 years. The highest odds ratio (OR:7.8; 9S% CI: 2.S2-24.12, p < 0.001) was identified in the S to 37.5°C (Axillary) 0 0 0 Mental status change 0 0 0 Other(Specify): 145 University of Ghana http://ugspace.ug.edu.gh Physical Examination Temperature(°C): (Axillary) Blood pressure (mmHg): _ Height(cm): _ Pulse rate:------------ Weight(kg): _ Oxygen saturation: _ Respiratory rate: _ Supplemental oxygen? DYes DNo SARI case: Inability to drink: DYes DNo Inability to be breastfed « I year): DYes D No Chest indrawing «5 years): DYes D No Stridor in a calm child «5 years): DYes DNo Unconscious: DYes DNo Breath sounds: D Vesicular D Bronchial D Vesicular with prolonged expiration D Difficult to comment Rhonchi: D Present D Absent D Difficult to comment Crepitation: D Present D Absent D Difficult to comment Final Clinical Diagnosis Please indicate the final diagnosis: _ Final Outcome D Transferred Out D Still under D Recovered I Discharged D Deceased D Lost to follow-up treatment Date final outcome established __ 1__ 1__ - If person died, date of death __ 1__ 1 _ Length of Stay at the Health Facility Please write how long patient stayed at the health facility: _ Contact details of Interviewer Tel. No. of Name of Interviewer: ---------------- interviewer: -------- Title: ---------------- Date of Interview: (ddimm/yyyy) 146 University of Ghana http://ugspace.ug.edu.gh Confidentiality You are assured that the information collected will be handled with the strictest confidentiality and will be used purely for the purpose of improving the health system. Be assured that all your information will not be shared with third parties not directly involved in the research. We will protect information about you to the best of our ability. Only persons working directly on the study would have access to your records. You will not be named in any report. Compensation Since there is no known human risk attached to participation in this study, there is no compensation. Additional Cost Not applicable Notification of Significant New Findings Not applicable Voluntary Participation and Right to Leave the Research Although there is no known risk associated with the research protocol, notwithstanding, if you feel uneasy and uncomfortable you have the liberty to opt out. Giving consent to participate in this study is absolutely voluntary and not under any coercion or obligation. You are also at liberty to withdraw your participation if you desire to do so at any time without penalty. Termination of Participation by the Researcher Not Applicable Contacts for Additional Information If you have pertinent questions, including the ethical aspects of this study or any problem related to protection of research volunteers or research-related injury, please contact Mr. Michael Adjabeng, the Principal Investigator on 020-815-7618. You may also contact Prof. William Kwabena Ampofo on 020-437-1207 or Dr. Margaret Gyapong on 020-630-1728. Your Rights as a Participant This research has also been reviewed and approved by the Institutional Review Board of Dodowa Health Research Centre. If you have any questions about your rights as a research participant you can contact the IRB Office between the hours of 8 am to 5 pm through the phone number 020 842-0640. 148 University of Ghana http://ugspace.ug.edu.gh APPENDIX II: CONSENT FORM Title: Risk Factors for Acute Respiratory Infections in Shai-Osudoku and Ningo-Prampram Districts Principal Investigator: Mr. Michael Jeroen Adjabeng Address: School of Public Health, College of Health Sciences University of Ghana Legon General Information about Research This research is to help determine the risk factors for acute respiratory infections and influenza-associated respiratory infections in Shai-Osudoku and Ningo-Prampram districts among those seeking medical care. If you (potential study participant) agree to be a study participant, we would like to ask questions about risk factors for acute respiratory illnesses. This group of questions will take about 15-20 minutes. The laboratory specimen collected from those of you with respiratory infections would be investigated at Noguchi Memorial Institute for Medical Research to identify any influenza virus. If the selected person is younger than 18 years, absent or unable to answer for themselves, we will ask parent or caretakers to answer for them. The results of this study will only be available to the people working on this study and will help officials from the Ghana Health Service to reduce the risk factors for Acute Respiratory Infections. This is an invitation for your participation in this research. Possible Risks and Discomforts The information that will be collected includes the background characteristics of respondents and any exposure to risk factors of respiratory infections. There is no known human risk attached to the study protocol. Kindly note that collection of swabs from the mouth or throat may introduce some slight discomfort. Possible Benefits There is no direct benefit to you as an individual, however your responses would help target appropriate public health interventions. Alternatives to Participation Not applicable 147 University of Ghana http://ugspace.ug.edu.gh CONSENT FORM - VOLUNTEER AGREEMENT The above document describing the benefits, risks and procedures for the research title (Risk Factors for Acute Respiratory Infection) has been read and explained to me. J have been given an opportunity to have any questions about the research answered to my satisfaction. I agree to participate as a volunteer. Date Name and signature or mark of volunteer If volunteers cannot read the form themselves, a witness must sign here: I was present while the benefits, risks and procedures were read to the volunteer. All questions were answered and the volunteer has agreed to take part in the research. Date Name and signature of witness I certify that the nature and purpose, the potential benefits, and possible risks associated with participating in this research have been explained to the above individual. Date Name Signature of Person Who Obtained Consent 149 University of Ghana http://ugspace.ug.edu.gh APPENDIX III CHILD ASSENT FORM Introduction My name is Mr. Michael Jeroen Adjabeng and Iam from the Epidemiology and Disease Control Department at School of Public Health. I am conducting a research study entitled Risk Factors for Acute Respiratory Infections in Shai-Osudoku and Ningo-Prampram Districts. ] am asking you to take part in this research study because] am trying to learn more about [risk factors that that promote Acute Respiratory Infections. This will take about 15 to 20 minutes of your time. General Information ]fyou agree to be in this study, you will be asked questions in relation to risk factors for respiratory infections. Trained persons will then collect samples from your mouth and nostril. Possible Benefits Your participation in this study will result in no direct benefit to you as an individual, however your responses would help target appropriate public health interventions. There is also no compensation. Possible Risks and Discomforts However, the risks associated with the study are negligible. Kindly note that collection of swabs from the mouth or throat may introduce some slight discomfort. Voluntary Participation and Right to Leave the Research You can stop participating at any time if you feel uncomfortable. No one will be angry with you if you do not want to participate. Confidentiality Your information will be kept confidential. No one will be able to know how you responded to the questions and your information will be anonymous. Contacts for Additional Information You may ask me any questions about this study. You can call me at any time [Michael Adjabeng, Phone - 020 815-7618] or talk to me the next time you see me. Please talk about this study with your parents before you decide whether or not to participate. ] will also ask permission from your parents before you are enrolled into the study. Even if your parents say "yes" you can still decide not to participate. 150 University of Ghana http://ugspace.ug.edu.gh Your rights as a Participant This research has been reviewed and approved by the Institutional Review Board of Dodowa Health Research Centre. If you have any questions about your rights as a research participant you can contact the IRB Office between the hours of 8am-5pm through the phone number 020 842-0640. lSI University of Ghana http://ugspace.ug.edu.gh CHILD ASSENT - VOLUNTARY AGREEMENT By making a mark or thumb printing below, it means that you understand and know the issues concerning this study. If you do not want to participate in this study, please do not sign this assent form. You and your parents will be given a copy of this form after you have signed it. This assent form which describes the benefits, risks and procedures for the research titled Risk Factors for Sever Acute Respiratory Infections in Shai-Osudoku Ningo-Prampram Districts has been read and or explained to me. I have been given an opportunity to have any questions about the research answered to my satisfaction. I agree to participate. aaaaaa Child's Name:........................... Researcher's Name: . Child's MarklThumbprint......... Researcher's Signature: . Date: . Date: . 152 University of Ghana http://ugspace.ug.edu.gh APPENDIX IV PARENTAL CONSENT Title: Risk Factors for Acute Respiratory Infections in Shai-Osudoku and Ningo- Prampram Districts Principal Investigator: Mr. Michael Jeroen Adjabeng Address: School of Public Health, College of Health Sciences, University of Ghana, Legon General Information about Research This study is to help determine the risk factors for Acute Respiratory Infections among residents of Shai-Osudoku and Ningo-Prampram Districts who seek health care for respiratory illnesses and this includes children. The responses of your child are equally important in this study. The interview would last for about twenty (20) minutes with mainly questions on exposure to risk factors. Depending on the classification of your child as a case or belonging to the comparative group, samples may be collected from the mouth and where possible the nostril. Possible Risks and Discomforts The information that will be collected includes the background characteristics of your child and exposure to risk factors for acute respiratory infections. The risk involved could be described as negligible however there may be some physical discomforts as a result of specimen collection. Possible Benefits There is no direct benefit to your child as an individual, however responses would help in targeting appropriate public health interventions. Alternatives to Participation Not applicable Confidentiality We will protect information about your child to the best of our ability. Your child will not be named in any reports. Some staff of Noguchi Memorial Institute for Medical Research and the Dodowa Health Research Centre may sometimes look at your child's research records. Your child's information will be kept confidential and anonymous. Compensation There will be no compensation. 153 University of Ghana http://ugspace.ug.edu.gh Additional Cost Not applicable Voluntary Participation and Right to Leave the Research You can stop participating at any time without any penalty if your child feels uneasy or uncomfortable. Termination of Participation by the Researcher Not applicable Notification of Significant New Findings Not Applicable Contacts for Additional Information If you have pertinent questions about the research, including the ethical aspects of this study or any problem related to protection of research volunteers or research-related injury, please contact Mr. Michael Adjabeng, the Principal Investigator on 020-815-7618. You may also contact Prof. William Kwabena Ampofo on 020-437-1207 or Dr. Margaret Gyapong on 020- 630-1728. Your Child's Rights as a Participant This research has been reviewed and approved by the Institutional Review Board of Dod ow a Health Research Centre. If you have any questions about your rights as a research participant you can contact the IRB Office between the hours of 8am-5pm through the phone number 020 842-0640. 154 University of Ghana http://ugspace.ug.edu.gh APPENDIX V PARENTAL CONSENT- VOLUNTEER AGREEMENT The above document describing the benefits, risks and procedures for the research title Risk Factors for Acute Respiratory Infections in the Shai-Osudoku and Ningo- Prampram Districts has been read and explained to me. [ have been given an opportunity to have any questions about the research answered to my satisfaction. I agree that my child should participate as a volunteer. Date Name and signature or mark of parent or guardian If volunteers cannot read the form themselves, a witness must sign here: I was present while the benefits, risks and procedures were read to the child's parent or guardian. All questions were answered and the child's parent has agreed that his or her child should take part in the research. Date Name and signature of witness I certify that the nature and purpose, the potential benefits, and possible risks associated with participating in this research have been explained to the above individual. Date Name Signature of Person Who Obtained Consent 155 University of Ghana http://ugspace.ug.edu.gh APPENDIX VI COMMUNITY RESIDENCE OF ILl PATIENTS COMMUNITY No. Percentage ASUTSUARE 96 22.12 OLDNINGO 29 6.68 KASUNYA 18 4.15 DORMELIAM 13 3.00 NATRIKU 13 3.00 OSUWEM 12 2.76 UPPER PRAMPRAM 12 2.76 AYIKUMAH 10 2.30 DAWHENYA 9 2.07 LOWER PRAMPRAM 9 2.07 MANGOTSONY A 9 2.07 ASUTSUARE-JUNCTION 8 1.84 VOLIVO 8 1.84 AGOMEDA 7 1.61 AYETEPA 7 1.61 RAMA TOWN 7 1.61 ATROBINYA 6 1.38 KADJANYA 6 1.38 NGMETSOKOPE 6 1.38 TSEBI (AHWIAM) 6 1.38 ANASISI 5 1.15 DODOWA 5 1.15 SALEM 5 1.15 AMUKOPE 4 0.92 MOYEOKORMOR 4 0.92 NEWNINGO 4 0.92 NYAPIENYA 4 0.92 OLOWE 4 0.92 PRAMPRAM 4 0.92 OTHER COMMUNITIES (77) 104 23.9 TOTAL 434 100 156 University of Ghana http://ugspace.ug.edu.gh APPENDIX VII PERCEPTION OF CAUSE OF ILls Responses Case Percent Control Percent Total Percent Don't know 114 77.55 220 74.83 334 75.74 Weather related (including heat) I I 7.48 22 7.48 33 7.48 Dusty conditions 2 1.36 18 6.12 20 4.54 Smoke (cigarrete, coil, firewood, etc.). 2 1.36 7 2.38 9 2.04 Infected by a neighbor through cough 3 2.04 4 1.36 7 1.59 Malaria 5 3.40 2 0.68 7 1.59 Chemicals/Pesticide 3 2.04 4 1.36 7 1.59 Allergy 2 1.36 3 1.02 5 1.13 Asthma 0 0.00 3 1.02 3 0.68 Headache I 0.68 I 0.34 2 0.45 A Friend Step on my Chest 0 0.00 I 0.34 I 0.23 After Eating Rice 1 0.68 0 0.00 I 0.23 Ageing 0 0.00 I 0.34 I 0.23 Alcoholism 0 0.00 I 0.34 I 0.23 Bacterial Infection I 0.68 0 0.00 1 0.23 Child was Choked 0 0.00 I 0.34 I 0.23 Crowd I 0.68 0 0.00 I 0.23 Drinking of Chilled Water I 0.68 0 0.00 1 0.23 Dry Throat 0 0.00 I 0.34 1 0.23 Family Planning 0 0.00 1 0.34 I 0.23 Fan 0 0.00 I 0.34 1 0.23 Fever 0 0.00 I 0.34 I 0.23 Seasonal Flu 0 0.00 1 0.34 1 0.23 Sleep Under Fan 0 0.00 1 0.34 I 0.23 TOTAL 147 100.00 294 100.00 441 100.00 Pearson chi2(23) = 25.4887 p = 0.326 157 University of Ghana http://ugspace.ug.edu.gh APPENDIX VIII PREVENTION OF ILls PREVENT ILl Cases Percent Control Percent Total Percent Don't Know 116 78.9 228 78.1 344 78.36 Seek Medical Carel Taking Medicine 3 2.0 10 3.4 13 3.0 Cough Etiquette/ Protect Yourself 7 4.8 6 2.1 13 3.0 Avoid Dust 1 0.7 8 2.7 9 2.1 Wearing Nose Mask! Protective Clothing 2 1.4 6 2.1 8 1.8 Eating Balanced Diet and Exercising 3 2.0 2 0.7 5 1.1 Avoid Heat/ Heat 0 0.0 5 1.7 5 1.1 Be Prayerful! Pray for Rain 0 0.0 4 1.4 4 0.9 Must Sleep Under Treated Net 2 1.4 2 0.7 4 0.9 Stay Away from Smoke/Use of Coil 0 0.0 4 1.4 4 0.9 Take Good Care of Yourself/ Personal Hygiene 0 0.0 4 1.4 4 0.9 Drinking Plenty Water 2 1.4 1 0.3 3 0.7 Knowing Your Allergies/What to Eat 2 1.4 I 0.3 3 0.7 Personal Hygiene, Sleeping at Well Ventilated Area 2 1.4 I 0.3 3 0.7 Cook Food Well 2 1.4 0 0.0 2 0.5 Stop Sleeping Under Fan and Drinking Cold Drinks/Water I 0.7 I 0.3 2 0.5 Avoid Cold Areas I 0.7 0 0.0 I 0.2 By Wearing Protective Clothing 0 0.0 1 0.3 1 0.2 Covering the Body with Sweater 0 0.0 I 0.3 I 0.2 Give Proteins 0 0.0 I 0.3 1 0.2 Not Exposing to the Weather and Taking Medication 0 0.0 I 0.3 1 0.2 Stop the Alcohol 0 0.0 I 0.3 1 0.2 Stop the Smoke ] 0.7 0 0.0 I 0.2 Stop Using Firewood and Charcoal and Use Gas ] 0.7 0 0.0 I 0.2 Use Fan 0 0.0 1 0.3 I 0.2 Vaccination 1 0.7 0 0.0 I 0.2 Ventilation 0 0.0 I 0.3 1 0.2 We Should Clean Our Environment 0 0.0 1 0.3 1 0.2 Total 147 100.00% 292 100.0 439 100.0 Pearson chi2(27) = 34.6597 P = 0.148 158 University of Ghana http://ugspace.ug.edu.gh APPENDIX IX PERCEPTION ON CAUSES OF SARI Controls Causes of SARI Cases (%) (%) Total (%) Don't Know 88 (66.17) 338 (84.29) 426 (79.78) Dusty Environment II (8.27) 20 (4.99) 31 (5.81) Weather 9 (6.77) 10(2.49) 19 (3.56) Infection 2 (1.5) 9 (2.24) 11 (2.06) Heat 5 (3.76) 2 (0.50) 7(1.31) Malaria 3 (2.26) 3 (0.75) 6 (1.12) Asthma 1 (0.75) 0(0.00) 1 (0.19) Ageing 0(0.00) 1 (0.25) I (0.19) Ate Something Bad I (0.75) 0(0.00) I (0.19) Bush Burning! Environmental Smoke I (0.75) I (0.25) 2 (0.37) Carrying Heavy Load on the Head 0(0.00) I (0.25) 1 (0.19) Child Walked in The Rains ] (0.75) 0(0.00) I (0.19) Cold Drink! Food I (0.75) 3 (0.75) 4 (0.75) Contaminated Source of Drinking Water 0(0.0) ] (0.25) 1 (0.19) Sitting Outside 1 (0.75) 0(0.00) 1 (0.19) Drinking Unclean Water 0(0.00) 1 (0.25) 1 (0.19) Flies from the Cow 1 (0.75) 0(0.00) 1 (0.19) Inhaling Toxic Smoke 0(0.00) 1 (0.25) 1 (0.19) Pains in The Testis 0(0.00) I (0.25) 1 (0.19) Paracetamol 0(0.00) 1 (0.25) 1 (0.19) Protruding Naval 0(0.00) 1 (0.25) 1 (0.19) Radiation 0(0.00) 1 (0.25) 1 (0.19) Poor Personal Hygiene 2 (1.50) 0(0.00) 2 (0.37) Smoking 2 (1.50) 1 (0.25) 3 (0.56) Spicy Foods 0(0.00) 2 (0.50) 2 (0.37) Stress/ Thinking 1 (0.75) 3 (0.75) 4 (0.75) Use of Pesticides 2 (1.50) 0(0.00) 2(0.37) Drinking Polluted Water 1 (0.75) 0(0.00) 1 (0.19) Total 133 (l00.00) 401 (100.00) 534 (100.00) Pearson chi2(27) = 58.7481 P < 0.001 159 University of Ghana http://ugspace.ug.edu.gh APPENDIX X PREVENTION ON CAUSES OF SARI Controls Responses Cases (%) (%) Total (%) Don't Know 93 (69.92) 349 (86.82) 442 (82.62) Clean and Avoid Dusty Environment 9 (6.77) 16 (3.98) 25 (4.67 Regular Checkup at Clinic/ Taking Medicine 4(3.01) 13 (3.23) 17 (3.18) Good Personal Hygiene 6(4.51) 5 (1.24) 11 (2.06) Prayer/ Pray for Rains 7(5.26) 3(0.75) 10(1.87) Vaccination 2 (1.50) 4 (1.00) 6 (1.12) Use of Coil/ Sleeping in Bednets 3 (2.23) 0(1.00) 3 (0.56) By Covering the Mouth When Coughing 1 (7.5%) I (0.25) 2 (0.37) Don't Drink Cold Water 0(0.0) 0(0.00) 2 (0.37) Stop Smoking 1 (7.5%) 1 (0.25) 2 (0.37 Abstain from Sitting Outside with Babies 1 (7.5%) 0(0.00) 1 (0.19) Avoid Eating Cold Food 0(0.0) 1 (1.25) 1(0.19) Avoid Smoke 1 (7.5%) 0(0.00) 1 (0.19) Avoid the Sun 1 (7.5%) 0(0.00) 1 (0.19) By Sending Infected Person to The Hospital 0(0.0) 1 (1.25) 1«0.19) Covering of the Mouth When Coughing 0(0.0) 1 (1.25) I (0.19) Do not get near to People who Cough 0(0.0) 1 (1.25) 1(0.19) Drinking Plenty Water 0(0.0) 1 (1.25) 1(0.19) Eat Hot Food 1 (7.5%) 0(0.00) 1(0.19) Good Weather Patterns I (7.5%) 0(0.00) 1 (0.19) Sickle Cell Disease 0(0.0) 1 (1.25) 1 (0.19) Staying in Well Ventilated Room 1 (7.5%) 0(0.00) 1 (0.19) Staying Indoors 1 (7.5%) 0(0.00) 1(0.19) Stop Eating Spicy Foods 0(0.0) 1 (1.25) 1 (0.19) To have Rest 0(0.0) 1 (1.25) 1 (0.19) Total 133 (100.0) 402 (100.0) 535 (l00.0) Pearson cht2(25) = 59.5167 Prob < 0.001 160 University of Ghana http://ugspace.ug.edu.gh APPENDIX XI ETHICAL CLEARANCE FROM NOGUCHI IRB NOGUCHt MEMORiAL INSTiTUTE FOR l'AEDICAl RESEARCH Establi.shed 1979A CrNHtitMent of the College of He.alth Sciences INSTmmONAL REVIEWBOARD P~'>st()HlC~: [~o>..Lt, 5k i Leg_oH __:', (ern Fa.\;: 2.556 UGL (JH gO, March, 2017 ETHICAL CLEARANCE .FEDERALWInE ASSURANCE .FWA 00001824 IRnOOOOl276 NMIMR-IRB CI'N 084/15-16 amend. 2017 fORG 000(1)08 On 8th March, 20 J 7 the Noguchi Memorial Institute for Medical. Research (NMTMR) Institutional Review Board (IRB) at a lull board meeting conducted continuing review and amended your protocol titled: ·tITLE (YFPROTOCOL Risk Factors for Acute Respiratory Infections in Shai-Osudoku and Ningo-Prampr-am Disrrtcts PRINCIPA.L INVli:STI.G ATOR MichaetJerocn Adjaheng Please note that a final review report must be submitted to the Board at the completion of the study. Your research records may be audited at any time during or after the implementation. Any modification of this research project must be submitted to the IRS for review and approval prior to implementation. Please report all serious adverse events related to this study to NMI.MR-IRB within seven days verbally and fourteen days in writing. This certificate is valid till 7'll March, 2018. You are to submit annual reports for continuing review. ~nr) " "" ., , " ' ' II; ~'" _ \ .;.~. SIgnature of Chalf;R...G. .: ...~)