Symptom duration
In total, 67.9% (n = 6179/9103), 24.4% (n = 2218/9103), and 7.8% (n = 706/9103) of participants had cough (as a dominant symptom), sore throat, and otitis media, respectively.
Data on time to symptom resolution were available for 11 studies (n = 8607). A flowchart showing inclusion and exclusion is available in Supplementary Figure S1. Median time to complete symptom resolution for these patients was 9 days (interquartile range [IQR] 4–27 days), with 90% of participants recovering within 28 days. Median time to complete symptom resolution was 15 (IQR 7–28) days for cough (90% recovered within 28 days), 4 (IQR 3–7) days for sore throat (90% recovered within 11 days), and 4 (IQR 2–8) days for otitis media studies (90% recovered within 10 days).
Participants who took antibiotics (48.3%; n = 3637/7529) had shorter median illness duration compared with those who did not take antibiotics (median: 9 [IQR 4–28] days versus 11 [IQR 5–28] days). However, 90% recovery was the same for those who took antibiotics and those who did not (28 days). A similar pattern was observed for those who took antibiotics during the study period (64.5% symptom free at 15 days; n = 2346/3637) and those who did not take antibiotics (65.9% symptom free at 15 days; n = 2565/3892). Over the first 28 days post-consultation, a larger percentage of people still had cough symptoms compared with other symptoms such as sleep disturbance, feeling generally unwell, and interference with normal activities (Figure 1).
Figure 1. Symptom resolution over the first 28 days after index consultation: a) for those who took antibiotics (n = 3637/7529); and b) for those who did not take antibiotics (n = 3892/7529).
Overall, the proportion of people who recovered within 1 week, 2 weeks, 3 weeks, and 4 weeks was 30.7% (n = 2067/6724), 65.8% (n = 3627/5509), 80.5% (n = 4451/5529), and 87.2% (n = 4335/4974), respectively. For adults, the proportion of people who recovered within 1 week, 2 weeks, 3 weeks, and 4 weeks was 26.1% (n = 1199/4594), 58.8% (n = 2043/3473), 76.5% (n = 2655/3470), and 84.4% (n = 2540/3008), respectively. For children, the proportion of people who recovered within 1 week, 2 weeks, 3 weeks, and 4 weeks was 40.8% (n = 868/2130), 77.8% (n = 1584/2036), 87.2% (n = 1796/2059), and 91.3% (n = 1795/1966), respectively. Among those without any lung disease, the proportion of people who recovered within 1 week, 2 weeks, 3 weeks, and 4 weeks was 30.3% (n = 1382/4556), 64.1% (n = 2254/3515), 79.6% (n = 2797/3514), and 86.9% (n = 2712/3121), respectively. Among those with lung disease, the proportion of people who recovered within 1 week, 2 weeks, 3 weeks, and 4 weeks was 25.6% (n = 220/861), 56.2% (n = 367/653), 69.7% (n = 456/654), and 76.8% (n = 468/609), respectively (data not shown).
Symptom trajectories
Data on symptom severity were available for 6436 participants (from seven studies). Three of these studies focused on children and four on a general population. Five of the seven studies were on cough (n = 5314),24,26,28,29,31 one was on sore throat (n = 914),27 and one was on otitis media (n = 208).23
LLCA trajectories
LLCA for cough identified four trajectories with distinct patterns of change in symptom severity over time (see Supplementary Table S3). These trajectories and their distribution within the population were: ‘rapid recovery’ (52.0%; n = 2763/5314), ‘intermediate recovery’ (28.9%; n = 1538/5314), ‘slow progressive improvement’ (12.5%; n = 663/5314), and ‘slow improvement with initial high symptom burden’ (6.6%; n = 350/5314) (Figure 2). Time to symptom resolution to below moderate level for 90% of participants was 6 days for the ‘rapid recovery’ trajectory, 10 days for the ‘intermediate recovery’ trajectory, and 27 days for the ‘slow progressive improvement’ and ‘slow improvement with initial high symptom burden’ trajectories.
Figure 2. Symptom trajectories based on longitudinal latent class analyses for a) cough; b) sore throat; and c) acute otitis media.
Baseline characteristics of the symptom trajectories for cough are presented in Table 1. Participants with ‘rapid recovery’ were generally younger and had slightly shorter duration of illness before the index consultation than participants in the remaining trajectory groups. A higher proportion of participants with ‘rapid recovery’ trajectory had lower baseline severity compared with the remaining trajectories. A lower proportion of participants with ‘rapid recovery’ were female or had any lung disease compared with the remaining trajectories. Similar results were obtained for adults and children.
Table 1. Baseline characteristics of all participants (adults and children) according to symptom trajectories for cough studies, n= 5314a
Similar, although faster, trajectories were observed for sore throat and otitis media (see Supplementary Appendix S1).
Faster versus slower recovery
To increase clinical usefulness, individuals were re-grouped into two groups each for cough and sore throat: ‘faster recovery (symptom recovery to below moderate levels within 10 days)’ and ‘slower recovery (symptom recovery to below moderate levels within 10 days)’.
Associations with baseline characteristics for cough
Older age (16–64 and >64 compared with <16 years) was associated with higher odds of slower recovery (odds ratios [OR] 2.57, 95% confidence interval [CI] = 1.72 to 3.85 and 3.17, 95% CI = 2.05 to 4.90, respectively). Median and above baseline severity was associated with ‘slower recovery’ compared with ‘faster recovery’: (OR 1.51, 95% CI = 1.12 to 2.03). Presence of lung disease (OR 1.78, 95% CI = 1.44 to 2.21) was also associated with ‘slower recovery’ compared with ‘faster recovery’. Median and above prior duration of illness (that is, ≥7 days) (OR 1.99, 95% CI = 1.68 to 2.37) were also associated with ‘slower recovery’ compared with ‘faster recovery’. Compared with no antibiotic prescribing, immediate antibiotic prescribing was associated with lower odds of slower recovery (OR 0.82, 95% CI = 0.68 to 0.98) (Table 2).
Table 2. Logistic regression analysis determining associations with slower (>10 days) versus faster recovery (≤10 days) for cougha
Associations with re-consultation and admission to hospital.
Re-consultation data were available for 5788 out of 6436 (89.9%) participants with symptom diary data (4714, 870, and 204 for cough, sore throat, and otitis media, respectively). Of these, 1685 (29.1%) re-consulted. Rates of re-consultation were 25.9% (n = 1015/3915) and 48.8% (n = 516/1057) in the ‘faster’ and ‘slower’ recovery group, respectively. Compared with the ‘faster recovery’ group, those in the ‘slower recovery’ group had increased odds of re-consultation (OR 2.15, 95% CI = 1.78 to 2.60).
Hospital admission data were available for 5915/6436 participants (91.9%). Hospital admission rates were 0.7% (n = 29/4021) and 2.5% (n = 27/1079) in the ‘faster’ and ‘slower’ recovery group, respectively. Compared with the ‘faster recovery’ group, those in the ‘slower recovery’ group had increased odds of admission to hospital (OR 7.42, 95% CI = 3.49 to 15.78).
Associations with the LLCA trajectories and baseline characteristics, re-consultation, and admission to hospital are available in Supplementary Tables S4– S7.
