Study location and design

The West Gonja Municipal is located west of Tamale, Ghana's Northern Regional Capital, between latitude 80 321 and 100 2 1 north and longitude 10 5 1 and 20 581 west with Damongo as its administrative capital [16]. The municipal has about 53,700 people with an expected pregnant population of 2173 and 12,882 women of fertility age. This study was part of a wider study to explore anemia and its associated factors among nomadic Fulani pregnant women in the West Gonja Municipality of Ghana employing a community-based exploratory sequential mixed methods approach [Unpublished]. This study used a community-based cross-sectional study design to determine the adherence to and predictors of iron-folate supplementation among nomadic Fulani pregnant women in the West Gonja Municipality of Ghana. Owing to the nature of cross-sectional designs in investigating the magnitude of a phenomena [17], it will help in determining the current adherence levels of IF supplementation among nomadic Fulani pregnant women.

Source and study population

All reproductive-aged nomadic Fulani pregnant women between 15 and 49 years who were living in the municipality were the population of interest. The study population was self-identified nomadic Fulani pregnant women aged 15–49 years, possessing maternal and child health (MCH) record book, who were given IF supplements within 16 weeks before/during the data collection period and agreed to participate (February–July, 2022). Fulani pregnant women (FPW) who disagreed to participate in the study, severely ill and were living in the study area less than six months were excluded.

Sample size and sampling technique

A total of 130 Fulani pregnant women were recruited for the study. The Fulani population are without a defined number and are not captured in national surveys and other records (thus, no sampling frame is available or can be constructed) because they are mostly a mobile and remote group (who are socially and medically disadvantaged), thus defined as a “hard-to-reach population” [18]. They are theorized to be a small population group, mostly scattered around community peripheries and at undefined locations. Random population samples are often inadequate to accumulate adequate samples of hard-to-reach groups such as the Fulani for population surveys or clinical trials [19].

Research has further shown difficulties in sampling groups defined as “hidden, remote or hard-to-reach”, largely comprising persons who avoid being identified [19]. Research also shows that, there are difficulties in sampling groups with low population numbers such as Indigenous people and so forth [19]. Thus, population-based probability sampling methods are a time consuming and costly strategy for sampling socially and medically disenfranchised populations [19].

Common methods for sampling hard-to-reach populations include non-probability-based approaches (such as census approach, snowball sampling) [20]. Therefore, a census and snowball/referral chain/social network recruitment approach was employed to select eligible respondents from the West Gonja Municipality. The census approach was used to select all eligible respondents because the sample population is theorized to be small without a sampling frame. As suggested by Raifman et al. [21], hard-to-reach populations may be difficult to completely survey even with an assumed sampling frame. They further alluded that, hard-to-reach-populations frequently constitute a small percentage of the general populace and are drifting away from society or socially unnoticed, possibly stemming from their experiences with social marginalization [21]. The referral chain method was used because of the difficult nature of recruiting a large-enough sample. This process relied on peer referrals, where enumerators chose initial respondents (seeds) who recruited their peers, then themselves recruited their peers through referrals, and so on until all eligible respondents were recruited. Similarly, Shedlin et al. [22] employed snowball sampling methods in a community-based study to recruit hard-to-reach respondents. In this study, two community partners/local peers/known community members recommended by the West Gonja Municipal Health Directorate led the enumerators to well-connected members of Fulani communities to aid rapport-building to increase the likelihood of people participating in the study. The community survey was conducted in the Busunu sub-municipality and Canteen sub-municipality due to their higher percentages of Fulani populations [9]. Ten (Kojope, Langantre, Busunu, Achubunyor, Mempeasem, Solepe, Kidendilipa, Kpiri, Sagiss, Kotito (Sore) number 3) out of thirteen (Kojope, Langantre, Busunu, Achubunyor, Mempeasem, Solepe, Kidendilipa, Kpiri, Sagiss, Kotito (Sore)number 1, Kotito (Sore) number 2 and Kotito (Sore) number 3, Jonokponto) dominant Fulani communities according to the municipal health directorate were selected from the sub-municipalities. The 3 communities that were not sampled was due to difficulty to access Fulani pregnant women, distance and poor road network.

Peer chain referrals were helpful since the recruitment process was mainly dependent on Fulani pregnant women, who probably have a better knowledge of the population than the enumerators. Though, community survey methods and referral chain approaches could lead to oversampling, Kalton et al. [23] and Kalsbeek et al. [24] argue that they could be applicable to low prevalence population groups, like the Fulani. Moreover, Platt et al. [25] compared a probability-based approach in sampling hard-to-reach-populations with snowball sampling to reach high risk HIV respondents and found that although referral chain sampling was more expensive, it led to increased response rates. Due to the fact that hard-to-reach populations may hide from being recognized as members of a group or refuse to take part in studies because they fear the law, distrust researchers [26], and believe that their involvement will not serve them personally or their community and could instead lead to discrimination, harassment or exploitation, multiple reports [27, 28] suggest that engaging local partners to assist in research could boost response rates and build trust.

Operational definitionsHousehold

A household was defined as a husband, wife/wives, and/or dependents (children, grandparents and other close relations) living together.

Adherence to IFAS

Respondents were asked if they took iron-folate tablets in the past week before the study and responses were dichotomized into yes or no. Fulani pregnant women who took at least 65% of the expected dose of the iron-folate tablets in the past week before the study, equivalent to intake of at least a tablet daily for 4 days in the week uninterruptedly or intake of 20 tablets in per month every day without forfeiting the recommended doses [29] were said to have adhered to IF supplementation regimen.

Anemia in pregnancy

Anemia was defined as hemoglobin (Hb) levels less than 11 g/dl [30].

Gestational age

Gestational age of 12 weeks or less, larger than 12 weeks but less than 24 weeks, and greater than 24 weeks but less than 42 weeks denoted the first, second, and third pregnancy trimesters, respectively.

IFAS knowledge

Five questions on importance of IFAS, 7 on likely negative effects of IFAS, 6 on management of the likely negative effects, 6 on consequences of reduced iron/folate intake, 7 on anemia signs and symptoms, 7 on dietary intake that boosts hemoglobin states, 1 on possible frequency of ingestion of IF supplements, and 1 on possible duration for IF supplement intake, were asked on IFAS knowledge and responses of the respondents were ranked on 40 Likert scale. A correct response (based on literature) was given a score of one, while a wrong response was allotted a score of zero. If the Fulani pregnant women recorded median score and above, they had good knowledge and Fulani pregnant women who recorded below the median score, had poor knowledge. This categorization has been seen in similar studies [6, 8, 31].

Biological sample collection

The finger was wiped with a swab of alcohol after approval was sought. The clean finger was punctured using a lancet. A microcuvette was employed in collecting the sample of blood from punctured site. Blood drops were steadily dipped in the well of the hemoglobin (Hb) strip of the Hb meter. Hb assay was carried out instantly using URIT 12 \(\circledR \), a battery-operated hand carrier Hb meter. Respondents were updated of their Hb status on-site. Individuals with Hb < 7 g/dL were counselled to request healthcare treatment.

Data collection procedure and quality control

A detailed review of literature [8, 31, 32] and responses from professional reviews were used to create the questionnaire. A face-to-face administered questionnaire was used to elicit replies. The questionnaire contained socio-demographic factors, obstetric factors, health care system related factors, and client related factors. Hb concentration was estimated using URIT 12 ® and current gestational age was estimated using a hand carrier ultrasound apparatus. Data on ANC attendance, gestational age at first antenatal attendance, Hb at first antenatal attendance, IFAS among others during pregnancy were gathered from MCH record books. A two-day training was organized for enumerators (four enumerators who were fluent in Hausa/Fulfulde, other native Ghanaian languages, and English). Two community partners were also recruited to aid rapport building, in order to get more people to participate in the study. Back-translated questions (Hausa/Fulfulde to English language and back) were used to inspect questionnaire accuracy, and needed amendments were made. Questionnaires were pretested in a close community with 20 respondents who shared comparable characteristics to the study's respondents. The flow of the questions was assessed for inclusiveness and fine-tuned as desired. On the interview day, FPW who agreed to contribute to the study had their Hb levels tested. Day-to-day, complete questions were assessed for wholeness, and remaining questions were addressed. Respondents who were unable to answer questions on their own were assisted. For authentication and reviews, information on the questionnaire was compared to typed data on the computer. Individual interviews lasted from 20 and 45 min.

Data processing and analysis

Excel spreadsheet was used to enter and clean data, which was then transported to Statistical Package for Social Services Version 25. Descriptive statistics were produced for categorical and continuous data. To identify the independent predictors of IFAS in pregnancy, factors that were significant in the bivariable model using Chi-square test (χ2) were included into a multivariable logistic regression model. All statistical tests were two-tailed with P-value ≤ 0.05 deemed significant level across all models.