AMR KNUST Basic

Author
Affiliation

Emmanuel Konadu

Kwame Nkrumah University of Science and Technology

METHODS/MATERIALS

Measures

The knowledge assessment consisted of 10 questions, categorized into two areas: knowledge about antibiotic resistance (four questions), and knowledge about antibiotic use (six questions) . Each question related to antibiotics awarded one mark if answered correctly, and zero if incorrect. The maximum possible score for the assessment was 10 and minimum was 0. The scores were later transformed into percentages. To categorise the knowledge, we used the Bloom’s cut-off point; 80–100% being good knowledge level, 60–79% being moderate knowledge level, and <60% and below for poor knowledge level (Seid and Hussen 2018).

Data Analysis

Descriptive statistics were used to summarise the study variables; categorical variables were presented in frequency and percentages, and continuous variables were expressed using the median and interquartile ranges. To evaluate the significant difference in overall knowledge levels between the pre-test and post-test, a McNemar test was applied. A simple linear regression model was also used using the difference between scores (post test minus pre test) as the dependent variable and the independent variables to estimate crude beta coefficients. From this approach, we were able to quantify the impact of each of the independent variables on observed change in knowledge. After adjusting for confounding, variables with a p-value < 0.10 in the unadjusted analysis were included as covariates in a multivariate linear regression to obtain adjusted beta coefficient.

Despite the study design inherently violating the assumption of independence in linear regression due to repeated measures on the same individuals, the difference in scores was used as the dependent variable to address this issue (Allison 1990). By analyzing the difference directly, we effectively reduced the within-subject correlation, focusing on the net change attributable to the independent variables. This approach assumes that individual-specific effects are implicitly controlled since each participant serves as their own baseline, mitigating some of the dependency concerns. Significance level was set at 5%. All analyses were carried out using R Programming Language version 4.4.0(2024)

RESULTS

Sociodemographic characteristics of respondents

The median age of the respondents was 13 years, with more than one-third of the respondents in JHS 1. More than half of the respondents were females. Regarding their living situation, almost ninety-four percent of the respondents resided with their parents. The parents’ median age was 46 and 41 years for the father and mother, respectively. More than three-quarters of the respondents’ fathers had a tertiary education, and nearly two-thirds of the respondents’ mothers also obtained tertiary education as the highest educational qualification. One-quarters of respondents residing with their guardians had their guardians have tertiary education (Table 1).

Table 1: Sociodemographics characteristics of the respondents
Characteristic N = 6111
Class
    JHS 1 228 (37.32%)
    JHS 2 212 (34.70%)
    JHS 3 171 (27.99%)
Age, years
    Median (Q1, Q3) 13.00 (12.00, 13.00)
Gender
    Female 340 (55.65%)
    Male 271 (44.35%)
Living Situation
    Extended Family 31 (5.23%)
    Immediate Family 1 (0.17%)
    Legal Guardian 4 (0.67%)
    Parents 557 (93.93%)
    Missing 18
Father's Age, years
    Median (Q1, Q3) 46 (40, 50)
    Missing 9
Father's Educational Level
    Basic 15 (2.66%)
    Secondary 39 (6.91%)
    Technical 82 (14.54%)
    Tertiary 428 (75.89%)
    Missing 47
Mother's Age
    Median (Q1, Q3) 41 (35, 45)
    Missing 10
Mother's Educational Level
    Basic 34 (5.95%)
    Secondary 74 (12.96%)
    Technical 95 (16.64%)
    Tertiary 368 (64.45%)
    Missing 40
Guardian Educational Level
    Basic 1 (2.44%)
    Secondary 6 (14.63%)
    Technical 9 (21.95%)
    Tertiary 25 (60.98%)
    Missing 570
1 n (%)

Antibiotic Resistance Awareness and Source of Information

More than ninety per cent of the respondents were aware of antibiotic resistance. Health workers (61.0%) were the most cited source of information concerning antimicrobial resistance, followed by an educational campaign (30%). Ninety-three per cent of the respondents knew the definition of antibiotic resistance. Two-thirds of the respondents had first-degree family members being healthcare professionals (Table 2).

Table 2 : Summary distribution for information and source of antibiotics resistance
Characteristic N = 6111
Information about Antibiotic Resistance 558 (94.58%)
    Missing 21
Source of Information (Health Professional) 339 (60.75%)
    Missing 53
Source of Information (Educational Campaign) 166 (29.75%)
    Missing 53
Source of Information (Media - TV,Radio, Social Media, etc) 66 (11.83%)
    Missing 53
Source of Information (Textbook - School Curriculum) 32 (5.73%)
    Missing 53
Source of Information (Family ) 11 (1.97%)
    Missing 53
Source of Information (Friend) 13 (2.33%)
    Missing 53
Antibiotic Resistance Explanation 557 (92.99%)
    Missing 12
First-degree family healthcare occupation 381 (65.58%)
    Missing 30
1 n (%)

Perception of Respondents towards Antimicrobial Use

Concern about antibiotic resistance showed a borderline significant increase, with positive responses rising from 71.0% at the pre-test to 75.6% at the post-test (p = 0.050). Furthermore, awareness of antibiotic resistance as a global issue remained essentially unchanged, with 73.0% of participants at the pre-test and 75.5% at the post-test recognising its global nature (p = 0.3). Similarly, perceptions of the limited impact of antibiotic resistance showed no significant variation, with positive responses remaining stable at 64.2% pre-intervention and 63.3% post-intervention (p = 0.8). However, a considerable improvement was observed in participants acknowledging their responsibility in mitigating antibiotic resistance. Positive responses increased from 67.4% at the pre-test to 85.1% at the post-test (p < 0.001).

Perception towards antimicrobial resistance
Characteristic Pre Test
N = 6111
Post Test
N = 6111
p-value2
Stopping Antibiotics When Better

<0.001
    Negative 223 (36.5%) 111 (18.2%)
    Positive 388 (63.5%) 500 (81.8%)
Missed Doses and Resistance

0.12
    Negative 267 (43.7%) 241 (39.4%)
    Positive 344 (56.3%) 370 (60.6%)
Antibiotics for Viral Infections?

0.013
    Negative 314 (51.4%) 269 (44.0%)
    Positive 297 (48.6%) 342 (56.0%)
Unconcerned About Antibiotic Resistance

0.050
    Negative 177 (29.0%) 149 (24.4%)
    Positive 434 (71.0%) 462 (75.6%)
Antibiotic Resistance Beyond Ghana?

0.3
    Negative 165 (27.0%) 150 (24.5%)
    Positive 446 (73.0%) 461 (75.5%)
Antibiotic Resistance Limited Impact?

0.8
    Negative 219 (35.8%) 224 (36.7%)
    Positive 392 (64.2%) 387 (63.3%)
Personal Role in Antibiotic Resistance

<0.001
    Negative 199 (32.6%) 91 (14.9%)
    Positive 412 (67.4%) 520 (85.1%)
1 n (%)
2 McNemar’s Chi-squared test with continuity correction

Knowledge of Respondents on Antimicrobial and Antimicrobial Resistance

Knowledge of Respondents on Antimicrobial Resistance

After the educational intervention, knowledge was significantly changed across all the questions assessing antimicrobial and antimicrobial resistance. Regarding antibiotic resistance, there was a significant increase in the knowledge in that domain from 59.2% to 78.4%, p-value <0.001. Also, there was a more than a quarter increase in knowledge in understanding of the risks of self-medicating with antibiotics without consulting a health professional or purchasing them without a prescription improved markedly (57.6% to 83.6%, p-value <0.001). There was also a substantial increase in knowledge about the risks of using leftover antibiotics from previous infections, rising from 47.6% to 76.9% (p < 0.001). Finally, knowledge regarding the risk of treatment failure due to not completing antibiotics as instructed by a doctor improved by nearly 20 percentage points, from 70.9% to 90.5% (p < 0.001).

Knowledge on Antimicrobial Resistance
Characteristic Pre Test
N = 6111
Post Test
N = 6111
p-value2
Antibiotic Resistance: Loss of Effectiveness

<0.001
    Correct 362 (59.2%) 479 (78.4%)
    Wrong 249 (40.8%) 132 (21.6%)
Antibiotic Resistance Risk: Leftover Medication

<0.001
    Correct 291 (47.6%) 470 (76.9%)
    Wrong 320 (52.4%) 141 (23.1%)
Antibiotic Treatment Failure: Incomplete Course

<0.001
    Correct 433 (70.9%) 553 (90.5%)
    Wrong 178 (29.1%) 58 (9.5%)
Antibiotic Resistance Risk: Self-Medication

<0.001
    Correct 352 (57.6%) 511 (83.6%)
    Wrong 259 (42.4%) 100 (16.4%)
1 n (%)
2 McNemar’s Chi-squared test with continuity correction

Knowledge of Respondents on Antibiotic Use

The intervention led to significant improvements in respondents’ knowledge of antibiotic use. Familiarity with antibiotics showed a marked increase, with correct responses rising from 43.7% at pre-test to 94.8% at the post-test (p < 0.001). Awareness that antibiotics are ineffective against viral infections also improved significantly, with correct responses increasing from 34.4% to 56.1% (p < 0.001). Similarly, the proportion of respondents correctly identifying that antibiotics are not appropriate for treating cough and wheezing rose from 39.8% at pre-test to 59.6% at post-test (p < 0.001). Knowledge about the spread of antibiotic-resistant bacteria also demonstrated significant gains, with correct responses increasing from 46.3% to 69.4% (p < 0.001). Furthermore, the understanding that low-dose antibiotics are not beneficial improved significantly, with correct responses rising from 38.3% at pre-test to 44.8% at post-test (p = 0.007). These findings underscore substantial progress in participants’ knowledge across critical aspects of antibiotic use and resistance

Knowledge on Antibiotic Use
Characteristic Pre Test
N = 6111
Post Test
N = 6111
p-value2
Familiar with Antibiotics

<0.001
    Correct 267 (43.7%) 579 (94.8%)
    Wrong 344 (56.3%) 32 (5.2%)
Antibiotics Don't Treat Viruses

<0.001
    Correct 210 (34.4%) 343 (56.1%)
    Wrong 401 (65.6%) 268 (43.9%)
Early Antibiotics Prevent Infection?

0.037
    Correct 171 (28.0%) 201 (32.9%)
    Wrong 440 (72.0%) 410 (67.1%)
Antibiotics for Cough and Wheezing?

<0.001
    Correct 243 (39.8%) 364 (59.6%)
    Wrong 368 (60.2%) 247 (40.4%)
Low Dose Antibiotics Beneficial?

0.007
    Correct 234 (38.3%) 274 (44.8%)
    Wrong 377 (61.7%) 337 (55.2%)
Antibiotic Resistant Bacteria Spread

<0.001
    Correct 283 (46.3%) 424 (69.4%)
    Wrong 328 (53.7%) 187 (30.6%)
1 n (%)
2 McNemar’s Chi-squared test with continuity correction

Overall Knowledge of Participants After Intervention

The educational intervention led to significant improvements in knowledge across the knowledge levels. A McNemar test demonstrated a significant overall change (( ^2(3) = 445.08 ), ( p = p = <0.001), with a moderate effect size} = 0.38 ), CI({95%}) [0.35, 0.40]).

Among participants initially categorized as Poor, 41% significantly improved to Moderate, and 24% significantly improved to Good ( p = <0.001). For those in the Moderate category, 38% significantly improved to Good ( p = <0.001). Among participants in the Good category, the majority (62%) retained their Good scores, with a small but significant decline of 34% to Moderate (p = <0.001 ).

Figure 4:Overall Knowledge of Respondents on Antimicrobial Resistance and Antibiotics. Changes in knowledge categories (Poor, Moderate, and Good) before and after the educational intervention. The McNemar test showed significant overall shifts in knowledge scores with a moderate effect size of 0.38. Significant improvements were observed among participants initially in the Poor category, with 41% moving to Moderate and 24% to Good, and among those in the Moderate category, with 38% improving to Good. Significant declines from Good to Moderate (34%) were also observed. The p-values for subgroup transitions are indicated above each bar.

Determinants of Knowledge on on Antimicrobial and Antimicrobial Resistance

In the simple linear regression model, being in JHS 2 was associated with a significant increase in knowledge relative to being in JHS 1 (β = 0.66, 95% CI: 0.24, 1.1, p = 0.002). Male respondents had significantly lower knowledge scores than females (β = -0.67, 95% CI: -1.0, -0.32, p < 0.001). Media sources such as TV, radio, and social media were associated with lower knowledge scores compared to those who did not report these sources (β = -0.57, 95% CI: -1.1, -0.05, p = 0.033).

In the multivariate linear regression model, gender and information sources remained significant. Male respondents continued to demonstrate lower knowledge scores than females (β = -0.51, 95% CI: -1.0, -0.01, p = 0.045). Media as a source of information was still associated with lower knowledge scores (β = -0.56, 95% CI: -1.1, -0.04, p = 0.035). Educational campaigns were also found to negatively influence knowledge scores (β = -1.4, 95% CI: -2.6, -0.17, p = 0.025).

Table 3: Simple Linear regression of change in knowledge
Characteristic N Beta 95% CI1 p-value
Class 611


    JHS 1
Reference Reference
    JHS 2
0.66 0.24, 1.1 0.002
    JHS 3
0.01 -0.42, 0.45 0.948
Age, years 611 -0.04 -0.20, 0.13 0.663
Gender 611


    Female
Reference Reference
    Male
-0.67 -1.0, -0.32 <0.001
Father's Age, years 591 -0.01 -0.02, 0.00 0.108
Father's Educational Level 551


    Basic
Reference Reference
    Secondary
0.68 -0.56, 1.9 0.283
    Technical
0.34 -0.77, 1.5 0.545
    Tertiary
0.67 -0.37, 1.7 0.208
Mother's Age 596 -0.01 -0.02, 0.00 0.166
Mother's Educational Level 548


    Basic
Reference Reference
    Secondary
-0.10 -0.98, 0.78 0.830
    Technical
-0.13 -1.0, 0.73 0.764
    Tertiary
0.03 -0.75, 0.81 0.939
Staying with biological parents 611


    Staying with biological parents
Reference Reference
    No
0.60 -0.79, 2.0 0.395
    Yes
0.37 -0.82, 1.6 0.542
Living Situation 597


    Extended Family
Reference Reference
    Immediate Family
0.07 -2.6, 2.7 0.960
    Legal Guardian
1.1 -1.6, 3.7 0.432
    Parents
-0.13 -0.97, 0.70 0.754
Source of Information (Health Professional) 244


    No
Reference Reference
    Yes
-0.01 -0.54, 0.51 0.958
Source of Information (Educational Campaign) 244


    No
Reference Reference
    Yes
-1.2 -2.4, 0.03 0.056
Source of Information (Media - TV,Radio, Social Media, etc) 244


    No
Reference Reference
    Yes
-0.57 -1.1, -0.05 0.033
Source of Information (Textbook - School Curriculum) 244


    No
Reference Reference
    Yes
0.03 -0.65, 0.71 0.931
Source of Information (Family ) 244


    No
Reference Reference
    Yes
0.02 -0.96, 1.0 0.971
Source of Information (Friend) 244


    No
Reference Reference
    Yes
0.90 -0.13, 1.9 0.086
First-degree family healthcare occupation 579


    No
Reference Reference
    Yes
0.00 -0.37, 0.36 0.994
1 CI = Confidence Interval
Table 4: Multivariate linear regression
Characteristic Beta 95% CI1 p-value
Class


    JHS 1 Reference Reference
    JHS 2 0.57 -0.04, 1.2 0.065
    JHS 3 0.25 -0.38, 0.87 0.437
Gender


    Female Reference Reference
    Male -0.51 -1.0, -0.01 0.045
Source of Information (Media - TV,Radio, Social Media, etc)


    No Reference Reference
    Yes -0.56 -1.1, -0.04 0.035
Source of Information (Educational Campaign)


    No Reference Reference
    Yes -1.4 -2.6, -0.17 0.025
Source of Information (Friend)


    No Reference Reference
    Yes 0.85 -0.18, 1.9 0.107
1 CI = Confidence Interval

References

Allison, Paul D. 1990. “Change Scores as Dependent Variables in Regression Analysis.” Sociological Methodology 20: 93. https://doi.org/10.2307/271083.
R Core Team. 2024. “R: A Language and Environment for Statistical Computing.” https://www.R-project.org/.
Seid, Mohammed Assen, and Mohammed Seid Hussen. 2018. “Knowledge and Attitude Towards Antimicrobial Resistance Among Final Year Undergraduate Paramedical Students at University of Gondar, Ethiopia.” BMC Infectious Diseases 18 (1). https://doi.org/10.1186/s12879-018-3199-1.