A descriptive cross-sectional survey design was employed to study HIV-positive women in the Central Region, which is in the southern part of Ghana. The region has been associated with high rates of teenage pregnancy, HIV and Acquired Immunodeficiency Syndrome (AIDS) [
14]. In 2014, the Central Region recorded HIV prevalence of 1.4% which was slightly higher than the national average of 1.37 in the 2014 HIV Sentinel Survey [
15]. The study was conducted among HIV-positive women receiving antiretroviral therapy in health care facilities in the Central Region of Ghana. The Health Information Unit of the Central Regional Health Directorate suggested that a total of 3483 women between the ages of 20 to 65 years of age were receiving HIV/AIDS care [
16].
Sample size determination and sampling method
In determining the sample size the following conditions were considered: the research design was a cross-sectional survey and therefore required a larger sample size. Confidence levels were estimated at 95%, confidence interval or margin of error of plus or minus 4%, and variability probability of 50%. The 50% variability probability was based on the assumption that the population will split 50/50 on the question [
17]. Based on these parameters, a sample size of 600 was needed. This was increased to 660 as a larger sample size reduces the degree of uncertainty [
18]. Simple random sampling was used to select six of eleven health facilities responsible for providing care for people living with HIV in the Central Region to participate in this study. They were: Cape Coast Teaching Hospital, Abura Dunkwa Hospital, Swedru Government Hospital, Our Lady of Grace, Assikuma, St Francis Xavier, Assin Foso, and Saltpond Government Hospital. In order to obtain the required number of participants from each facility, the probability proportionate to size sampling was used to determine the proportions from each facility to be included in the sample. Accidental quota sampling was used to obtain the number of participants from each participating facility. Therefore, HIV-positive women within 20 and 65 years were selected until the desired sample size was reached.
Data collection
A questionnaire was used to collect relevant data for the study. This was adapted from Mupepi et al. and Hassani et al. [
5,
19]. Selection of measures was guided by the Health Belief Model (HBM) and Theory of Planned Behaviour (TPB) [
20,
21]. These theories attempt to explain the category of people who will take appropriate action to prevent disease and predict their likelihood or intention towards preventive health-related behaviour. Measures based on these models have been used extensively in assessing cervical cancer screening behaviour among healthy as well as disadvantaged and marginalised populations [
5,
22‐
24].
The measures for perceived susceptibility, perceived seriousness, perceived benefits, cues (prompts, triggers or reminders about cervical cancer screening) influencing cervical cancer screening, and perceived barriers were guided by the HBM. These constructs were measured on a four point-Likert scale. The use of a four point Likert scale is highly justified. Garland explained that social desirability bias, a situation where respondents give answers that put them in good light can be reduced by deleting or eliminating the mid-point from Likert scales [
25]. Therefore, the responses to the Likert scale items were: strongly agree (SA), agree (A), disagree (D), and strongly disagree (SD). For positive statements, (SA) had a score of 4, (A) = 3, (D) = 2 and (SD) = 1. The reverse score was used for negative statements.
The details of questions used to assess the different parameters that were assessed have been described. For perceived susceptibility, the questions assessed were I worry about developing cervical cancer; I had a relative with cancer, so I may get cervical cancer; I don’t think I can get cervical cancer; I had multiple sexual partners in the past, so I may get cervical cancer; I have been in a polygamous relationship in the past, so I may get cervical cancer; I may not get cervical cancer because I already have HIV; I think I may get cervical cancer sometime in my life; and I do not feel at risk of getting cervical cancer. The items used to measure perceived seriousness included cervical cancer makes life worse; cervical cancer makes it difficult to have sex; cervical cancer patients may die within a short time; cervical cancer is more serious than other diseases; the problems caused by cervical cancer remain for a long time; cervical cancer makes patients anxious all the time; many women with cervical cancer may have complications; and cervical cancer mainly affects sexually active women.
In the same way, perceived benefits was measured with the following items: screening can find cervical changes before they become cancer; if cervical changes are found early they are easily treatable; screening test will help a woman to know if she has cervical cancer; screening cannot prevent the spread of cervical cancer; screening cannot save the patient’s life; cervical cancer screening may cause infertility; and if I had cervical cancer screening, I would never get cancer. The items on the perceived barriers to cervical cancer screening subscale comprised: I know where to go for cervical cancer screening; I don’t have much information about cervical cancer; the screening centres are too far from where I live; cervical cancer screening would be embarrassing; cervical cancer screening would be painful; cervical cancer screening is against my religious beliefs; I am afraid of knowing that I have cancer; and I cannot afford the cost of screening.
The items that constituted cues about cervical cancer screening were: education on the need for cervical cancer screening at the hospital will encourage me to have the test; If a relative suffers from cervical cancer, it will discourage me from obtaining the screening; If I hear about cervical cancer screening on the radio and television, I will go for the test; If I am referred for cervical cancer screening by my doctor, I will obtain the test; and If a doctor/nurse reminds me, I will obtain the test.
The instrument was shown to experts in the area of cervical cancer to judge it against the purpose and hypotheses the study sought to answer. To achieve internal reliability, a pilot study was conducted to ensure that the questions were appropriate and would be understood by the study participants. The pilot-test was conducted among 100 HIV-positive women in Effia Nkwanta Hospital, Takoradi, Western Region of Ghana. The pilot study was useful in identifying the type of training the data collectors might require before embarking on the main study [
18]. Preliminary analysis was conducted to ensure that the data collected would answer the hypotheses. The Cronbach’s alpha internal consistency indices were obtained for the subscales were .824 for perceived susceptibility, .820 for perceived seriousness, .798 for perceived benefits, .809 for cues about cervical cancer screening and .795 for perceived barriers.
Ethical approval for the study was obtained from the Institutional Review Board of the University of Cape Coast and Ethical Review Committee of the Ghana Health Service. Approval was also obtained from the Ethical Review Committee of the Cape Coast Teaching Hospital. Permission was also sought from the Central Regional Health Directorate to use all the HIV/AIDS clinics randomly selected for the study in the Central Region of Ghana. Written informed consent was obtained from the participants before embarking on the data collection.
Participants were recruited at the hospitals, health centres and by word of mouth by professional nurses who had been trained in the care and management of HIV/AIDS patients. The benefit of participating in the research process was explained to the participants without providing misleading information or exaggerating about potential benefits. An interviewer-administered questionnaire was used. The independent variables for the study were; perceived susceptibility, perceived seriousness, perceived benefits, perceived barriers, and cues about cervical cancer screening. The dependent variable was intention to obtain cervical cancer screening. HIV-positive women who met the inclusion criteria were approached and interviewed after obtaining their full consent. Six nurses who could speak both English and Fante or Twi (native languages spoken by most people in the Central Region of Ghana) and had worked or were currently working with women with HIV were trained to assist with the data collection. The data were collected on special clinic days designated for caring for people with HIV/AIDS in the hospitals selected for the study. The interviews took place in the counseling rooms in the HIV/AIDS clinics from March to May, 2016. The data were analysed using frequencies, percentages and binary logistic regression. Respondents with a score of less than 75% for cues, perceived seriousness, perceived benefits, perceived susceptibility, and perceived barriers were considered to have low perception about the construct being measured while those with scores of 75% or more had high perception [
26].