Study selection and characteristics
Post deduplication, 1148 results underwent abstract and title screening, with 177 undergoing full-text review (Fig. 1). After full-text review, six studies were excluded due to high risk of bias, leaving a final 32 studies included. Twenty-seven studies measured FCR (Table 1) with a total of 8715 patients [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41] and 18 PSA anxiety (Table 2) combining 9953 patients [11, 17,18,19,20,21,22, 26, 27, 30, 36, 38, 39, 41,42,43,44,45]. Twelve studies measured both FCR and PSA anxiety domains [11, 17,18,19,20,21,22, 26, 30, 36, 38, 39]. Studies drew populations from varied countries with 13 from Europe, 11 from North America, 6 from Australia and 2 from Asia. Dates of publication ranged from 2003 to 2021, with 18 studies utilising a longitudinal design and 14 a cross-sectional design.
Fear of cancer recurrence
Nine studies measured prevalence of significant FCR ranging from 4.1 to 45% (median: 13%, IQR = 29.3) in a total of 6210 patients [18, 23, 25, 28, 32, 34, 38, 39, 46]. Only two longitudinal studies measured the evolution of significant prevalence over time, with one demonstrating a small decrease from 45% at treatment to 39% by 18 months suffering with high FCR . The other showed prevalences of high FCR at 7 years post radical prostatectomy at 5% increasing to 7% at 16 years post surgery. Nilsson et al.  found higher prevalence of FCR in active surveillance (AS) patients compared to radical prostatectomy (RP). The prevalence of low risk of bias studies had a median of 5.7% (IQR = 2.92) [32, 34, 46] compared to a median prevalence of 29.4 (IQR = 27.5) for moderate risk of bias studies [18, 23, 25, 28, 38, 39].
Severity and progression over time
There were seven different measures used to measure FCR, with the memorial anxiety prostate cancer fear of recurrence (MAXPC FoR) subscale most commonly used with 12 out of the 27 studies using it. This is scored out of 12 with higher scores signifying higher levels of FCR. Average scores of 1.5 on each question or a combined score of > 6 have been previously classified as significant . Five studies measured FCR levels with this subscale at diagnosis with mean scores ranging from 0.8 to 5.88 (median = 3.35) and no mean scores at diagnosis above the cut-off [22, 24, 26, 36, 39]. Eight studies measured FCR scores at 6 months to a year after diagnosis with mean scores ranging from 1 to 4.89 (median = 3.6, IQR = 3.8). Lastly, seven evaluated FCR levels at greater than a year post diagnosis with mean scores ranging from 2.6 to 4.29 (median = 3.1, IQR = 0.42) [19, 22, 24, 26, 29, 33, 38]. Nine studies evaluating the progression of FCR severity over time demonstrated scores to be highest at diagnosis with a decrease over time [17, 19, 22,23,24, 33, 36, 37, 39]. This decrease was however often very small with only a only a single study identifying this change to be statistically significant between 0 and 2 months .
Predictors of fear of cancer recurrence
Few factors associated with FCR were consistently evaluated. Most studies measured FCR for prostate cancer patients as a whole, with few distinguishing between treatment groups. Kendel et al.  however compared patients on active surveillance, radical prostatectomy and patients who discontinued active surveillance, identifying active surveillance patients specifically to possess higher levels of FCR. Conversely, Nilsson et al. found RP patients to have higher FCR than AS patients  and Mehta et al. found  higher levels of FCR in radiotherapy than RP. Separately with a metastatic disease at diagnosis  was also found to be associated with higher FCR as compared to those with localised cancer diagnosis. Evaluating patient characteristics, higher levels of FCR were seen in gay or bisexual patients in two studies [19, 31]. Additionally, seven studies found a negative association between age and FCR levels [18, 20, 25, 30, 36, 38, 41], implying younger patients display higher levels of FCR.
Relationship to other outcomes
FCR was consistently associated with other quality of life and mental health measures (Table 3). In particular, FCR was associated with higher scores on the Hospital Anxiety and Depression Scale (HADS) questionnaire in all four studies identified [18, 20, 30, 38, 41]. Similarly, three studies identified an association between FCR and worse outcomes in quality of life scales such as the functional assessment of cancer therapy (FACT-P) and short form 12 (SF 12) [18, 38, 39]. Lastly, Van de Wal et al. found a positive association between FCR and urinary, bowel and hormonal symptoms alongside distress scores .
Four studies measured prevalence of significant PSA anxiety ranging from 1.2 to 27.9% (median = 22.75) totalling 5416 patients [11, 38, 39, 43]. Anderson et al.  found a prevalence of 1.2% of significant PSA anxiety for patients undergoing active surveillance, in comparison to findings from Meissner et al. of 3% in radical prostatectomy patients in their large cohort of 4719 patients . Higher prevalences were seen in other studies of 27% and 15% for those receiving hormone treatment or radiotherapy respectively . Only one low risk of bias studies measured PSA anxiety prevalence finding 3% . This was compared to a median prevalence of 20.2 for moderate risk of bias studies [38, 39, 43].
Severity and progression over time
Like FCR, the MAXPC PSA subscale was the most used measure to measure PSA anxiety severity with 16 of the 18 studies using it, with a total of three measures used overall. A cut-off of 4.5 is considered significant for this subscale. Overall scores were low in studies. At diagnosis, five studies measured PSA anxiety with the MAXPC PSA [19, 22, 36, 39, 44] with scores ranging from 0.25 to 2.43 (median = 0.79, IQR = 1.20). Four studies measured PSA anxiety between 6 and 10 months with scores of between 0.3 and 1.27 [17, 27, 36, 39] and eight studies at over a year post diagnosis (range = 0.27–2.08, median = 0.54, IQR = 0.85) with ongoing = low scores event at up to 10 years follow-up. Throughout the survivorship trajectory, no studies identified mean scores above the cut-off. Trends across longitudinal studies showed a low peak PSA anxiety at diagnosis, remaining low at up to 3 years after [17, 19, 22, 36].
Predictors of PSA anxiety
Due to the nature of active surveillance, more studies specifically looked at this population [17, 19, 22, 38, 39] which stayed below significance after a year when evaluated longitudinally [17, 22]. Overall, few studies compared between cancer characteristics and different treatments. However, where evaluated, no association was observed between cancer stage and PSA anxiety [30, 36]. Similar to FCR, gay and bisexual patients were found to have higher levels than heterosexual patients . The relationship with age and PSA anxiety demonstrated mixed findings. Four studies identified nil association [20, 30, 38, 41]; however, two [36, 44] reported a negative association suggesting younger age to be associated with PSA anxiety.
Relationship to other outcomes
The strongest correlation between PSA anxiety and any other outcome was to overall HADS scores. All three studies identified [20, 30, 38] found a positive association between these scores, implying an association between PSA anxiety and wider mental health conditions such as depression and generalised anxiety. Furthermore, Anderson et al.  found a positive association between PSA anxiety and two other measures: FACT-P and the state trait anxiety inventory (STAI), further highlighting the association with generalised anxiety and suffering with other prostate-related symptoms. Interestingly, a close positive association between FCR and PSA anxiety was found in all three studies evaluating this [18, 30, 38]. These two mental wellbeing issues therefore seem to be closely interlinked, with patients possessing higher FCR having higher PSA anxiety . Lastly, one study found a negative association between bowel, hormonal, urinary symptoms and PSA anxiety, although this correlation was small (β < 0.1) .
Risk of bias assessment
Overall, studies demonstrated good internal and external validity. JBI checklist scores ranged from 4 to 9 (median = 6, IQR = 3). Eight studies demonstrated a low risk of bias, 24 a moderate risk of bias and six a high risk of bias and were excluded from final inclusion (online resource 5 and 6). Common concerns identified included high dropout rates of participants in cohort studies (n = 7), a lack of strategies to address these dropouts (n = 17) and no consideration of confounders in study design (n = 22). Furthermore, many studies demonstrated a poor wider representation of the prostate cancer population, focussing only on certain demographics or introduced selection bias through focussed electronic recruitment on social media or through email dispersal.