Description of the trials
We identified 15 trials which examined the effect of vitamin C on preventing AF in patients at a high risk of AF [18–32]. See the flow diagram of the searches in Fig. 1. Fourteen trials examined patients undergoing cardiac surgery, either CABG [18–27] or CABG and valvular surgery [28–31]. The 15th trial examined the recurrence of AF after a successful cardioversion [32] (Table 1).
Five trials were carried out in the USA [18, 27–30], 5 in Iran [19, 20, 25, 26, 31], 3 in Greece [21, 22, 32], 1 in Slovenia [23], and 1 in Russia [24]. The total number of participants was 2050. The mean age ranged from 56 to 73 years, and the proportion of males varied from 59 to 100% (median 69%) in the trials that reported sex distribution (Table 1).
Most POAF trials administered 2 g of vitamin C within about 12 h before the operation and 1–2 g/day for 5 days after the operation, and followed patients for the occurrence of AF for the same period. In most trials, vitamin C was administered as tablets, although 5 trials administered it intravenously [21–23, 26, 31]. See the description of dosages in the Additional file 1.
In the cardioversion trial [32], 2 g vitamin C was administered before the cardioversion and thereafter 1 g/day of vitamin C for 7 days. After a successful cardioversion, participants were followed for 7 days for the recurrence of AF.
Because of the selection criteria, all the trials were randomized. Korantzopoulos et al. [32] mentioned that the cardioversion patients bought vitamin C tablets themselves and thus knew the treatment, but the physician who was responsible for cardioversion and follow-up was unaware of the treatment. A placebo was not used in 6 POAF trials [19, 20, 23, 24, 29, 32], but in other trials participants were administered a placebo. Four trials [23, 24, 29, 32] did not report that the physicians in charge of treatments and assessment of outcome were blinded, but in other trials that was the case. None of the trials had substantial or unbalanced drop-out rates. See the Additional file 1 for the details of the methods.
Effect of vitamin C on atrial fibrillation
Figure 2 shows the effect of vitamin C on the occurrence of AF in high risk patients. Over all the 15 trials, vitamin C decreased the risk of AF by 27%. However, there is significant heterogeneity between the 15 trials with I2 = 61% (P = 0.001 in the heterogeneity test), which indicates that no single estimate of effect, such as the 27% mean effect, is consistent with the results of all the 15 trials, and thus the causes of the heterogeneity should be explored.
Because of the variations in the clinical context and in the results, we divided all the trials into 4 subgroups. The 14 trials on POAF were divided into trials carried out a) in the USA, b) in Greece, Slovenia, and Russia, and c) in Iran, and the 15th trial is included as d) the cardioversion trial in Greece (Fig. 2). There is very strong evidence that the estimates calculated for these 4 subgroups are heterogeneous with I2 = 88% (P = 10−5). The high level of heterogeneity is caused by the US trials, which found no effect of vitamin C. If the 5 US trials are removed, the remaining 10 non-US trials are not heterogeneous, I2 = 38% (P = 0.10), and the pooled estimate indicates a 45% decrease (95% CI 35 to 54%; P = 10−10) in the occurrence of AF. If we further remove 4 non-US trials that had some concerns about randomization or blinding [23, 24, 31, 32], the effect estimate remains essentially the same indicating a 48% (95% CI 34 to 60%) decrease in the incidence of AF (see Additional file 2). Thus, there is very strong evidence from the 10 non-US trials that vitamin C decreases the risk of AF in high risk patients.
The 5 POAF trials in Iran found a 51% decrease in the incidence of AF and the 4 POAF trials in Greece, Slovenia, and Russia found a 29% decrease. When these 9 trials were pooled to a group of non-US POAF trials, vitamin C decreased the incidence of AF by 44% (33 to 53%) (Fig. 3).
In a direct comparison of the 5 US POAF trials against the 5 Iran POAF trials, there is very strong evidence of heterogeneity with I2 = 95% (P = 10−5) (Additional file 2). Thus, vitamin C seems to have significantly different effect in these two countries.
Korantzopoulos et al. studied the recurrence of AF within 1 week after a successful cardioversion, finding that vitamin C decreased the rate of AF recurrence by 87% (P = 0.012).
All the POAF trials administered 1–2 g/day of vitamin C. Most trials gave it as tablets, whereas 5 gave it intravenously. These 2 methods of administration lead to different vitamin C levels in the body and we compared the 2 methods among the non-US POAF trials in Fig. 3. Oral administration decreased the occurrence of POAF by 73% and intravenous administration by 36%. There is strong evidence of heterogeneity between the 2 administration methods with I2 = 87% (P = 0.005).
Effect of vitamin C on hospital stay
The effect of vitamin C on the hospital stay in 11 POAF trials is shown in Fig. 4. On average, vitamin C shortened the hospital stay by 10.1%. Because of the heterogeneity found in the effects of vitamin C against POAF, we divided the trials into the US and the non-US trials also in this analysis. The US trials found no effect on hospital stay whereas in the non-US cardiac surgery trials, vitamin C decreased the length of hospital stay by 12.6% (95% CI 8.4 to 16.8%; P = 10−8). Only 1 of the non-US trials had some concerns about blinding [23], and its exclusion had only a small effect on the pooled effect estimate indicating a 13.3% (9.0 to 17.6%) decrease in the length of hospital stay (Additional file 2).
The non-US trials were further divided into oral and intravenous trials. There is strong evidence of heterogeneity between the 3 subgroups with I2 = 78% (P = 0.009) (Fig. 4). In the non-US trials, intravenous administration shortened the length of hospital stay by 16% and oral administration by 7% and there was evidence of heterogeneity between the oral and intravenous non-US trials with I2 = 76% (P = 0.039).
The relative effect, i.e., the effect in percentages, adjusts for baseline variations in the patient groups and is therefore an informative effect measure when pooling trial results in Fig. 4. Nevertheless, since the effect on hospital stay as actual days has more direct practical impact, we also calculated the effect of vitamin C on the days in hospital in Fig. 5. In the non-US trials, intravenous vitamin C shortened hospital stay on average by 1.47 days and oral vitamin C by 0.43 days.
Since oral and intravenous vitamin C differed in their effects on the occurrence of POAF and on the duration of hospital stay, we plotted the estimates of effect in Fig. 6. In the non-US trials, oral vitamin led to a greater effect on POAF occurrence but to a lesser effect on hospital stay compared with intravenous vitamin C. The US trials found no benefit on either outcome.
Effect of vitamin C on ICU stay
The effect of vitamin C on the length of ICU stay in the POAF trials is shown in Fig. 7. Vitamin C shortened ICU stay by 8% (P = 0.002) in the 7 non-US trials and there is no evidence of heterogeneity between the non-US trials (P = 0.1). Three US trials found no effect.