Study characteristics
After removal of duplicates, 8341 references were identified through the initial search. An additional eight references were added by checking the reference list of previously published literature reviews [18, 19]. Of 75 references eligible for full-text screening, 22 met inclusion criteria and were included in the final analysis (Fig. 1).
Of the included studies, 19 (86%) were randomized controlled trials and three (14%) were sequential two-arm cohort studies. Most studies (n = 20, 90%) were conducted in an outpatient clinic setting [16, 17, 26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43]. One study [44] was performed at a hospice and one study at an inpatient clinic [45]. Patients with various cancer types (including lung, breast, colorectal, gynaecologic, prostate, head and neck, lymphatic and prostate) and treatment modalities participated in the individual studies. The number of included patients ranged between 43 and 766 across the studies (Tables 1 and Appendix Table 6).
Table 1 Characteristics of included studies The majority of the studies were conducted in the USA (n = 8), followed by the UK (n = 5) and the Netherlands (n = 3). Great diversity was seen in types of PROMs (n = 20), and several studies used more than one PROM (Table 2). The European Organisation for Research and Treatment of Cancer Quality of Life C30 (EORTC QLQ-C30) (n = 6), general symptoms on a numeric scale (n = 6) and the Hospital Anxiety and Depression Scale (HADS) (n = 3) were the most commonly used PROMs in the included studies.
Table 2 Characteristics of intervention Risk of bias
Figure 2 summarizes the risk of bias of the included studies. Risk of bias (selection, performance, detection, attrition, reporting and other types of bias) was assessed using the Cochrane Risk of Bias Tool [25]. For the risk of bias assessment of each individual study, see Appendix Table 7. Random sequence generation risk of bias was as expected high in the three non-randomized controlled trials (RCTs) [30, 38, 42]. Allocation concealment was maintained in nine studies; in seven studies, this was not reported; and in five studies, there was a high risk of bias. All included studies were rated as high risk regarding performance bias as blinding of participants and personnel was not possible due to the nature of a PROM intervention. Twelve studies (57%) were rated as low risk for detection bias. Ten studies (48%) reported high rates of drop-out or loss to follow-up and were therefore rated as high risk of bias due to incomplete outcome data. Reporting bias was unclear in almost all studies.
PROM as intervention, with or without feedback to patients or health care professionals, compared with not using a PROM
Of the 22 included studies, 15 studies were identified that compared the use of a PROM as the intervention to no PROM intervention [16, 26, 29,30,31,32,33, 36, 38, 43, 44]. Of these fifteen studies, in one study, the intervention was merely the use of a PROM [31]. In 14 studies, the use of the PROM was supplemented with feedback about the results [16, 26, 29, 30, 32, 33, 36, 38,39,40,41,42,43,44]. In case feedback was provided, it was provided either to the HCP, the patient or both. In seven studies, the feedback was only available to the HCPs [36, 38,39,40,41,42, 44]. In five studies, the feedback was provided to both patients and HCPs [16, 26, 29, 30, 32]. In two studies, the results were available for HCPs if patients proactively shared the feedback with them (Table 3) [33, 43].
Table 3 Main findings and outcome assessment comparing PROM as intervention, with or without feedback to patients or health care professionals, to not using a PROM Patient outcomes—survival/mortality
Only one study included survival as an outcome [26]. Basch et al. (2016) found that overall survival after 1 year was 76% in the PROM group versus 68% in the non-PROM group (p = 0.05). The study identified two important subgroups: computer-experienced and computer-inexperienced patients. Only one study included survival as an outcome [26]. Notably, in the subgroup of computer-inexperienced patients within the intervention group, the survival rate was significantly higher compared with the computer-inexperienced patients within the control group (p = 0.02). For the patients who were computer-experienced, no significant difference in survival was identified between the intervention and control group (p = 0.45). The authors suggested that computer-inexperienced patients may have less-developed health communication skills and thereby benefit more from a structured program that incorporates self-reporting via PROMs.
Patient outcomes—morbidity and symptoms
Three studies evaluated symptoms as an outcome [31, 32, 44]. In the study of McMillan et al. (2011), patients with various cancer diagnoses who were admitted in a hospice filled out five PROMs that focused on symptoms, spiritual needs and HRQoL (i.e. PPS, MSAS, HQLI-4, CES-D, SNI, SPMSQ). The only outcome with a significant difference between the intervention group and control group was depression, as measured by the CES-D. While the depression scores declined significantly (p = 0.023) over time in both the intervention and control group, indicating an improvement in mental health, decline was significantly larger in the intervention group than in the control group (p = 0.027) [44].
In the study by Hoekstra et al. (2006), cancer patients receiving palliative treatment were randomized to either completing a symptom-based PROM (ten symptoms on a numeric 1–10 scale) every week at home or not completing a PROM. The symptoms evaluated were fatigue, pain, lack of appetite, shortness of breath, coughing, sleeplessness, nausea, constipation, diarrhoea and vomiting. It is unclear whether results were reported to the HCP and the patients. Significant differences were only identified for vomiting and constipation. The prevalence of these symptoms was lower in the intervention group [31].
Kearney et al. (2009) compared the use of a PROM (CTCAE and Chemotherapy Symptom Assessment Scale integrated into one questionnaire) for 14 days after a cycle of chemotherapy in breast, lung and colorectal cancer patients to care as usual without a PROM and found that the intervention group had a higher prevalence of hand-foot syndrome. Furthermore, the severity of hand-foot syndrome and associated levels of distress were also significantly higher. Other symptoms reported in the PROM did not differ significantly between the intervention and control group [32].
Patient outcomes—HRQoL
Studies performed by Basch et al. (2016), Velikova et al. (2004) and Williams et al. (2013) showed a positive effect on HRQoL when using a PROM compared with no PROM [26, 40, 42]. In these studies, patients in the intervention group reported a significant and clinically relevant improvement in HRQoL over time. Six studies did not identify significant difference between HRQoL scores between the intervention and control group [16, 29, 30, 36, 43, 44]. Mills et al. (2009) and Taenzer et al. (2000) found a negative effect when using a PROM [33, 38]. In the study of Mills et al. (2009), patients in the control group (i.e. no PROM) scored better on a lung-specific HRQoL scale, which included physical well-being, social/family well-being, emotional well-being, functional well-being and seven lung cancer-specific symptoms (p = 0.04) [33]. In the study of Taenzer et al. (2000), patients in the control group scored better on two specific HRQoL sub-domains, physical functioning and role functioning (p < 0.05 and p < 0.01, respectively). In all other HRQoL domains, no significant differences were found between the PROM intervention and control group [38]. A lower reported HRQoL in the intervention group may be explained by the increased attention to HRQoL domains, resulting in more recognition and reporting of specific symptoms.
Patient experiences—patient satisfaction
In three studies, patients expressed the usefulness of a PROM [16, 29, 41]. They stated that PROMs were useful to tell their physician how they were feeling and that they represented an accurate representation of their functioning and well-being. The majority of patients, 79, 85 and 86%, respectively, were confident in that a PROM increased the awareness of their physician regarding their HRQoL and symptoms [16, 29, 41].
Four studies described no significant difference in patient satisfaction between patients completing a PROM and those who did not complete a PROM as part of their cancer care [30, 33, 36, 38].
Process indicators
The study performed by Basch et al. (2016) reported on emergency visits and hospital admissions. The intervention group (i.e. patients receiving a PROM with feedback) reported statistically significant fewer emergency visits and hospital admissions than the control group (i.e. no PROM)[26].
Evaluation of doctor-patient communication was described in four studies comparing a PROM intervention to no PROM [16, 29, 30, 41]. In three studies, doctor-patient communication was rated better using a PROM [16, 30, 41], and one study [29] did not find any differences in the doctor-patient communication between the intervention and control group. In the study performed by Hilarius et al. (2008), more HRQoL topics were discussed in the intervention group than in the control group, but this did not lead to differences in patient management activities (e.g. referral, medication prescription, test ordering and modification chemotherapy) between both groups [30].
PROM as intervention with feedback to patients or health care professionals, compared with a control group in which PROMs were used without giving feedback to patients or health care professionals about the results
Seven studies were identified that compared use of a PROM with feedback to patients or health care professionals to use of a PROM without feedback to patients or health care professionals [17, 27, 28, 34, 35, 37, 45] (Table 4).
Table 4 Main findings and outcome assessment comparing PROM as intervention with feedback to patients or health care professionals, compared with a control group in which PROMs were used without giving feedback to patients or health care professionals about the results Patient outcomes—morbidity and symptoms
Five studies evaluated symptoms as an outcome [17, 28, 34, 37, 45]. In the study performed by Cleeland et al. (2011), an email was forwarded to the health care professionals in case a pre-set threshold of an alarming symptom (e.g. pain, distress, disturbed sleep, shortness of breath and constipation) was exceeded. Approximately 12% fewer emails regarding alarming symptoms were forwarded in the intervention group compared with the control group [28]. Ruland et al. (2010) found that in 75 leukaemia and lymphoma cancer patients who received feedback after completing a PROM, more symptoms had decreased compared with patients who had not received feedback about the findings of the PROM (ten of 19 symptoms vs two of 19 symptoms decreased). Of these ten symptoms, discomfort, eating/drinking, sleep/rest and sexuality were statistically significant in favour of the intervention group [45]. A favourable effect of using a PROM with feedback (compared with no feedback) was also seen in the study conducted by Strasser et al. (2016). The symptom distress score (including nine different symptoms rated on a 1–10 Likert scale) was significantly lower for the intervention than the control group over time (p = 0.003) [37]. In the study by Mooney et al. (2014), patients with various cancer types treated with chemotherapy were randomized to either reporting presence and severity of chemotherapy-related symptoms (rated on a 1–10 Likert scale) using an automated phone system with feedback to their physician or solely reporting chemotherapy-related symptoms using the identical automated phone system but without any feedback. No significant differences in symptom severity and distress scores were seen between the intervention and control group [34].
Patient outcomes—HRQoL
Only one article assessed HRQoL when comparing a PROM with or without feedback. Strasser et al. (2016) found a small, albeit significant, higher HRQoL in the group receiving a PROM with feedback. However, the difference between intervention group and control group was not considered clinically meaningful [37].
Patient experiences—patient satisfaction
Two studies focused on the comparison of PROMs with feedback and without feedback reported on patient satisfaction [28, 34]. In the study of Cleeland et al. (2011), higher levels of patient satisfaction were seen in the intervention group, compared with patients in the control group (p < 0.03). Mooney et al. (2014) found that 79% of patients were confident that the information they reported in the PROM would be noticed by their physician.
Process indicators
Mooney et al. (2014) reported that when health care professionals initiated the contact, more topics were discussed compared with when patients initiated contact [34]. In the study by Ruland et al. (2010), patients in the intervention group (i.e. PROM with feedback) and the control group (i.e. PROM without feedback) were asked to rate nineteen symptoms on a 1–10 Likert scale. In the intervention group, patients needed less symptom management in seventeen of nineteen symptoms. In six of these seventeen symptoms, significantly less symptom management or treatment was seen. In contrast to the intervention group, patients in the control group needed more symptom management in fourteen of nineteen symptoms, indicating that patients had greater needs for support in managing their symptoms. Significantly more symptom management was seen in six of these fourteen symptoms and all involved psychological needs—energy, sleep/rest, sexuality, mood/feelings, maintaining control over my situation and relationships [45]. In the study by Mooney et al. (2014), health care professionals treated both patients in the intervention group (i.e. receiving results of a PROM) and in the control group (i.e. not receiving results of a PROM). They found that the majority of health care professionals were satisfied with the PROM system and receiving alert reports, while 15% were not satisfied and did not read any of the received alert reports [34]. Berry et al. (2011) and Nicklasson et al. (2014) found no differences in consultation length between the group that received feedback on PROM results and the group that did not. Consultation time was not prolonged when health care professionals received feedback about PROM results.