Overall characteristics of ANDEMIA study participants

Between 1 February 2018, and 26 May 2022, 19,700 patients were enrolled in the ANDEMIA study, including 5,529 (28.1%) from CIV; 4,802 (24.4%) from BF; 5,937 (30.1%) from DRC and 3,432 (17.4%) from RSA. Across all countries, 36.6% (n = 7,203) of patients presented with AFDUC, 33.9% (n = 6,676) with RTI, 25.8% (n = 5,085) with GI and 3.7% (n = 736) with both RTI and GI. Approximately half of enrolled patients were children under the age of five and 48.6% were female. More were recruited in health facilities from urban areas (n = 12,527, 63.6%) than from rural areas (n = 7,173, 36.4%), although 44.0% (8,624/19,617) of overall enrolled patients with data on place of residence reported living in a village (this included 2,413 patients recruited from urban health facilities). A greater proportion of adult patients or legal guardians of minors reported no education in CIV (n = 2,987, 54.0%) and BF (n = 3,166, 66.4%) compared to DRC (n = 876, 14.8%) and RSA (n = 63, 1.9%). During the study period, approximately half of patients were enrolled before the COVID-19 pandemic (n = 10,604, 53.8%). Although data showed a hospitalization rate of 98.6% (11,009/11,186), data on hospitalization were largely incomplete for CIV (51.8%; 2,861 missing entries) DRC (47.3%; 2,809 missing entries) and BF (29.3%, 1,408 missing entries). A summary table of patient characteristics by country is provided in the supplementary text (see Additional File 6).

Characteristics of study participants reporting antibiotic use

Among 19,700 ANDEMIA patients, 7,258 (36.8%) reported antibiotic use in the ten days prior to study enrolment. During the study period, slightly more than half of patients with self-reported antibiotic use were enrolled before the COVID-19 pandemic (n = 4,319, 59.5%). In CIV, BF, and DRC, more than 80% of patients with reported antibiotic use presented to urban health facilities, in contrast to RSA where the presentation to urban and rural health facilities was fairly balanced including 51.8% presenting to rural and 48.2% to urban facilities. About half of the patients with reported antibiotic use from BF (49.9%) and from RSA (52.1%) reported living in a village in contrast to 37.5% in CIV and 27.7% in DRC. Slightly more than half of enrolled patients with reported antibiotic use were children under the age of five (n = 4,077, 57.0%) and 45.9% were female. A greater proportion of adult patients or legal guardians of minors reported no education in CIV (n = 354, 53.7%) and BF (n = 1,388, 56.6%) compared to DRC (n = 315, 16.6%) and RSA (n = 42, 1.9%). Approximately half of the patients or legal guardians of minors who reported antibiotic use were unemployed, ranging from 21.5% (n = 495) in BF to 72.6% (n = 1,592) in RSA. Among those with data on hospitalization, 99.3% of the patients with reported antibiotic use were currently hospitalized at enrolment, although data on hospitalization was incomplete, particularly in CIV (45.4%, 299/659 missing entries) and DRC (19.9%, 378/1,903 missing entries) (see Table 1).

Table 1 Characteristics of study participants reporting antibiotic use ten days prior to study enrolment by country

Differences in characteristics of patients who did and did not report antibiotic use across all patients enrolled in the study period can be seen in Additional file 7.

Among the 7,258 patients who reported antibiotic use in the ten days prior to study enrolment, 3,009 (41.5%) were enrolled with RTI, 2,198 (30.3%) with AFDUC, and 1,642 (22.6%) with GI (Tables 1 and 2). Over 90% (n = 6,957) of patients who reported antibiotic use had taken the last dose of the reported antibiotic within 2 days prior to study enrolment. Across all syndromes, 45.7% (3,276/7,170) reported symptom onset in the 4–7 days prior to enrolment. Among those with the GI syndrome, 48.2% (757/1,569) reported onset of symptoms in the 0–3 days prior to enrolment.

Half of the patients (2,676/5,325, 50.5%) with available BMI/z-score data who reported antibiotic use were of normal weight, 32.8% (1,749/5,325) were underweight, and 16.9% (900/5,325) were overweight or obese. In particular, approximately half of patients with a GI or GI/RTI syndrome who reported antibiotic use were underweight (514/1,100, 46.7%; and 136/261, 52.1% respectively). Among those with treatment data, 20.0% (1,405/7,036) of patients who reported antibiotic use also reported use of antimalarials and 51.5% (3,601/6,992) reported use of other medication such as analgesics, dietary supplements, antiparasitic drugs (other than antimalarials) and corticosteroids, with only minor differences between the syndrome enrolment (Table 2).

Table 2 Characteristics of study participants reporting antibiotic use in the ten days prior to enrolment by syndromeMost commonly reported antibiotics

Among 7,258 patients who reported antibiotic use, 9,695 antibiotics were reported in the ten days prior to study enrolment. The most common antibiotic reported was ceftriaxone (31.7%, n = 3,071), a WHO Watch antibiotic, followed by amoxicillin (15.1%, n = 1,466) and amoxicillin/clavulanic acid (11.1%, n = 1,075), both WHO Access antibiotics. Among patients with the AFDUC syndrome, ceftriaxone was the most commonly reported antibiotic, ranging from 29.0% in CIV (64/221) to 62.6% (652/1,042) in BF. Among patients with GI, ceftriaxone (ranging from 21.4% (52/243) in CIV to 58.6% (246/420) in BF), and metronidazole (ranging from 13.7% (113/826) in RSA to 18.9% in DRC and CIV (141/746 and 46/243 respectively) were the most commonly reported antibiotics. Among those with RTI, amoxicillin/clavulanic acid was the most commonly reported antibiotic in CIV (32.3%; 112/347) and BF (41.4%; 541/1,307). Apart from ceftriaxone, other Watch group antibiotics such as ciprofloxacin (in particular, 6.1% (169/2,751) of patients with AFDUC), cefotaxime, cefixime, ofloxacin and erythromycin were reported (Fig. 1). In total, only three (0.03%) carbapenems (imipenem once, meropenem twice) were reported. Of the 66 reported antibiotics belonging to the Not recommended group, 63 (95.5%) were reported from DRC. A majority of the 63 Not recommended group antibiotics reported in DRC came from the fixed-dose-combination metronidazole/norfloxacin (71.4% or 45/63) and primarily among patients with the GI and GI/RTI syndrome (66.7% or 42/63).

Fig. 1

Top five most commonly reported antibiotics overall and by country, syndrome and WHO AWaRe classification. Legend: Access antibiotics are colored in green, Watch antibiotics are colored in yellow, and antibiotics not included in the top five reported antibiotics are grouped as “other” and colored in blue. Sulf./Trim. = Sulfamethoxazole/Trimethoprim; Amoxicillin/Clav. = Amoxicillin/clavulanic acid*. AFDUC: acute febrile disease of unknown cause; GI: gastrointestinal infection; RTI: respiratory tract infection; Combination of GI/RTI cases not shown separately (n = 592). CIV: Côte d’Ivoire, BF: Burkina Faso, DRC: The Democratic Republic of the Congo, RSA: The Republic of South Africa

Among the top five antibiotics reported across countries, only Access and Watch antibiotics were found. Out of all 9,695 reported antibiotics, 54.7% (n = 5,299) were Access, 44.7% (n = 4,330) were Watch, 0.7% (n = 66) were Not recommended, and 0% (n = 0) were Reserve group antibiotics. The overall ratio of Access to Watch antibiotics was 1.2. This varied by country, urban and rural health facilities and syndrome, (Fig. 2) ranging from 0.4 among rural facilities in BF to 4.7 among urban facilities in CIV. Among those with the AFDUC syndrome, approximately half or less of reported antibiotics were Access antibiotics. Among those with the GI syndrome, a greater number of patient reported antibiotics were Access antibiotics, especially across the urban health facilities in CIV and rural health care facilities in DRC and RSA which reported greater than 60% of Access group antibiotics. Among those with the RTI syndrome, the greatest number of Access antibiotics were.

reported including all health care facilities in CIV and RSA as well as rural facilities in DRC reporting more than 70% Access antibiotics (Fig. 2).

Fig. 2

Proportional antibiotic use according to WHO AWaRe classification by syndrome, country and health facility location. Legend: AFDUC: acute fever of unknown cause; GI: gastrointestinal infection; RTI: respiratory tract infection; CIV: Ivory Coast; BF: Burkina Faso; DRC: The Democratic Republic of the Congo; RSA: Republic of South Africa; urb.: urban health centre; rur.: rural health centre

Before the COVID-19 pandemic, 40.7% (2,352/5,783) of all reported antibiotics were from the Watch group. This increased to 50.6% (1,978/3,912) during the pandemic (see Additional File 9 for these results by country). The relative increase in the reported use of Watch antibiotics during the pandemic were most pronounced among those with the AFDUC syndrome, from 54.4% (809/1,487) reported antibiotics in AFDUC before the pandemic to 67.6% (855/1,264) during the pandemic, although results varied by country and location. The increase in the reported use of Watch antibiotics from before to during the pandemic were less pronounced among those with GI and RTI, although differences by country and location were also present, particularly in rural RSA where the reported use of Watch among patients with RTI went from 17.3% (84/485) before the pandemic to 42.3% (200/473) during the pandemic.

Due to a missing or an invalid reported date of the last dose, 611 reported antibiotics were excluded from the main analysis. Among these excluded antibiotics, metronidazole was reported in 47.6% (n = 291). A sensitivity analysis of the most commonly reported antibiotics by country and WHO AWaRe classification including these additional antibiotics (n = 10,306) is reported in the supplementary text (see Additional File 10).

Routes of administration and number of reported antibiotics

Of 9,695 total antibiotics reported, 51.6% (n = 4,998) were classified as parenteral use only. Across patients in all countries and by syndrome, these antibiotics included most commonly ceftriaxone, gentamicin, cefotaxime and ampicillin. In BF, ceftriaxone accounted for 90.2% (1,378/1,527) of reported antibiotics classified as parenteral formulations. Of the parenteral classified antibiotics, the last dose was reported on the same day of study enrollment in 52.3% (n = 2,615) of cases, one day prior in 41.9% (n = 2,095) and 3–10 days prior in 5.8% (n = 288) of cases.

The number of different antibiotic substances reported varied considerably between countries and between urban or rural location of the health facilities. While urban DRC reported 37 different antibiotic substances, of which 12 were classified as parenteral, rural CIV, BF and DRC each reported 14 different antibiotic substances, of which two (rural CIV) and four (rural BF and rural DRC) were classified as parenteral formulations (see Additional File 11).

Among the 7,258 patients who received antibiotics, 69.0% (n = 5,010) reported the use of only one antibiotic, followed by 26.2% (n = 1,904) who received two antibiotics, and 4.7% (n = 344) had received three or four antibiotics. In BF, 83.0% (n = 2,051) reported only one antibiotic ten days prior to study enrolment, in contrast to patients in RSA where 52.2% (n = 1,161) reported only one antibiotic. Furthermore, more patients enrolled at urban health care facilities received only one antibiotic (73.2%, 5,284/3,867) before enrolment compared to those at rural health facilities (57.9%, 1,143/1,974) (Fig. 3).

Fig. 3

Number of antibiotics reported per patient in the ten days prior to study inclusion. Legend: CIV: Côte d’Ivoire; BF: Burkina Faso; DRC: Democratic Republic of the Congo; RSA: Republic of South Africa; AFDUC: acute fever of unknown cause; GI: gastrointestinal infection; RTI: respiratory tract infection

The sensitivity analysis assessing the number of antibiotics reported among patients who reported antibiotic use on the same day as study enrolment only (as opposed to the full time period of ten days prior) showed similar trends.