Randomized trial of dyadic-report vs proxy-report and self-report symptom assessment for pediatric patients receiving cancer treatments

Abstract

Background

We validated different approaches to symptom assessment for pediatric cancer patients based on the Symptom Screening in Pediatrics Tool (SSPedi) for self-report (SSPedi and mini-SSPedi), proxy-report (proxy-SSPedi), and structured dyadic-report (co-SSPedi). The objective was to compare co-SSPedi scores vs proxy-report (proxy-SSPedi) and self-report (SSPedi or mini-SSPedi) scores for pediatric patients receiving cancer treatments.

Methods

This was a single-center, randomized crossover study enrolling English-speaking dyads of pediatric patients with cancer or hematopoietic cell transplant recipients 4-18 years old and their guardians. Dyads were randomized to first complete the dyadic-report (co-SSPedi) or self-report (patients: SSPedi or mini-SSPedi) and proxy-report (guardians: proxy-SSPedi). Dyads then crossed over to the alternate approach. Primary analysis compared total SSPedi scores between randomized groups.

Results

We enrolled 420 dyads that were randomized to co-SSPedi first (n = 213) or proxy-SSPedi and self-report SSPedi first (n = 207). Mean total SSPedi scores (± standard deviation) were co-SSPedi (9.6 ± 7.1), proxy-SSPedi (9.7 ± 7.5; P = .950 for comparison vs co-SSPedi), and self-report SSPedi (9.7 ± 8.2; P = .981 for comparison vs co-SSPedi). Co-SSPedi scores were significantly different from proxy-SSPedi for feeling disappointed or sad, feeling cranky or angry, feeling tired, mouth sores, and changes in taste. Co-SSPedi scores were significantly different from self-report SSPedi scores for problems with thinking or remembering things, feeling tired, mouth sores, tingly or numb hands or feet, and diarrhea.

Conclusions

Total co-SSPedi scores were not significantly different compared with proxy-report or self-report scores, although there were differences in specific symptom scores. If different reporter types are used during clinical implementation, specifying reporter type will be important. The study was registered at clinicaltrials.gov (NCT #05012917). Symptoms are common and frequently severely bothersome in pediatric patients with cancer and hematopoietic cell transplant (HCT) recipients (1). To measure the extent of bothersome symptoms, the Symptom Screening in Pediatrics Tool (SSPedi) suite of symptom assessment tools was developed for pediatric patients receiving cancer treatments and currently consists of multiple validated instruments. SSPedi was developed for self-report by patients 8-18 years of age (2,3). Mini-SSPedi was developed for self-report by patients 4 to 7 years of age (4). Proxy-SSPedi was developed for proxy-report by guardians of pediatric patients 2-18 years of age (5). These 3 instruments can be categorized as either self-report (SSPedi or mini-SSPedi) or proxy-report (proxy-SSPedi).

Although symptom self-report and proxy-report may be successful, each approach has limitations. Self-report is generally preferable, particularly when focused on subjective constructs. However, some pediatric patients, especially the youngest patients or patients who are clinically unwell, may have challenges with single or repeated assessment completion. Proxy-report may be more feasible to obtain in settings where self-report is challenging. However, there are well-known differences between proxy-report and self-report assessments (6-8). Furthermore, the same guardian is not always with the pediatric patient when symptom reporting is planned.

We hypothesized that a dyadic approach, where SSPedi is completed by both the child and a guardian together, may be one way to address these challenges. Consequently, we developed and validated a structured dyadic approach called co-SSPedi that allows the guardian to facilitate instrument completion but ensures the pediatric patient voices their rating of each symptom first (8-10). Thus, there are now 3 different SSPedi reporting types, namely self-report, proxy-report, and dyadic-report. As a dyad approach to symptom reporting is relatively new, we hypothesized that a randomized trial of co-SSPedi vs proxy and self-report SSPedi would increase our understanding of how SSPedi scores generated by each type compare to each other.

There are several valid and appropriate approaches to measure symptoms and quality of life for pediatric cancer patients including Patient-Reported Outcomes Measurement Information System (PROMIS) measures (11), Pediatric Patient-reported Outcome Common Terminology Criteria for Adverse Events (Peds-PRO-CTCAE) (12), and several different PedsQL modules (13). SSPedi was chosen as the focus for this study since it is the only validated pediatric symptom screening tool that offers self-report, proxy-report, and dyadic-report options.

The primary objective was to compare co-SSPedi scores vs proxy-report (proxy-SSPedi) and self-report (SSPedi/mini-SSPedi) scores for pediatric patients receiving cancer treatments. The secondary objectives were: 1) to describe co-SSPedi, proxy-SSPedi, and self-report SSPedi variability (represented by score standard deviations); 2) among those randomized to proxy-report and self-report first, to compare within-person severely bothersome scores; and 3) to describe preferences for SSPedi type.

Methods

Study design

This was an open-label, single-center, randomized crossover study enrolling pediatric patients with cancer or HCT recipients at the Hospital for Sick Children in Toronto, Canada. Dyads were randomized to complete either co-SSPedi first or proxy-report (proxy-SSPedi) and self-report (SSPedi or mini-SSPedi) first. After completion of the randomized approach, dyads crossed over to the alternate approach. The study was registered at clinicaltrials.gov (NCT #05012917). This study was approved by the Research Ethics Board at the Hospital for Sick Children, and all participants provided informed consent and assent (as appropriate).

Patients

This study enrolled pediatric patient and guardian dyads. We included English-speaking dyads of a pediatric patient and a guardian where the patient was between 4 and 18 years of age and had a diagnosis of cancer or was a HCT recipient. Exclusion criteria were illness severity, cognitive disability, or other impairment that precluded completion of co-SSPedi, proxy-SSPedi, or self-report SSPedi according to the primary health-care team. Both inpatients and outpatients were eligible.

Instruments

Co-SSPedi or proxy-SSPedi and self-report SSPedi (SSPedi or mini-SSPedi) were completed according to the randomization sequence. All instruments result in a total score that ranges from 0 (no bothersome symptoms) to 60 (worst bothersome symptoms).

SSPedi (8-18 years of age) is a self-report scale that includes the following 15 symptoms: disappointed or sad, scared or worried, cranky or angry, problems thinking, body or face changes, tiredness, mouth sores, headache, other pain, tingling or numbness, throwing up, hunger changes, taste changes, constipation. and diarrhea. Each symptom is rated on a 5-point Likert scale consisting of 0 = “not at all bothered,” 1 = “a little,” 2 = “medium,” 3 = “a lot,” and 4 = “extremely bothered.” The recall period is yesterday or today (2).

Mini-SSPedi (4-7 years of age) is a self-report scale composed of the same 15 symptoms as SSPedi. Based on cognitive interviews conducted with children receiving cancer treatments in this age range, SSPedi was modified to have a recall period of today with a 3-point Likert scale consisting of 0 = “not at all bothered,” 2 = “medium,” and 4 = “extremely bothered.” In other words, the anchors and mid-point are identical to SSPedi. A faces-type scale accompanied the response options, and the symptom descriptions were simplified (4,14).

Proxy-SSPedi is identical to SSPedi except that the leading question refers to “your child” rather than “you.” It was designed for pediatric patients 2-18 years of age. It uses a 5-point Likert scale and has a recall period of yesterday or today (14).

Co-SSPedi is based on either SSPedi (age 8-18) or mini-SSPedi (age 4-7). Co-SSPedi is meant to be completed by both child and guardian together. The format and wording are identical to self-report SSPedi and mini-SSPedi (depending on child age) to maximize the ability of the child to participate in symptom reporting (10). The approach uses a written/audio instruction that asks the dyad to decide who will read the leading question, each symptom, and response options out loud. For each symptom, the child answers how bothered they are by that symptom out loud. The guardian then states if they agree, or they can offer a different score. If the guardian’s score is different from the child’s score, the dyad discusses until they reach consensus. The agreed-upon score is then entered. This process is repeated for the remaining symptoms.

Procedures

Potentially eligible dyads were approached in the inpatient or outpatient setting by a member of the study team. Eligible dyads were invited to participate, and for consenting dyads, demographic information was obtained from the guardian or the patient’s health record.

Dyads were randomized to complete either co-SSPedi first vs guardian proxy-SSPedi and self-report SSPedi first. After completion of SSPedi in the randomized order, the dyad crossed over to the alternate approach. For example, a patient–guardian dyad randomized to complete co-SSPedi first would complete co-SSPedi together, and then, on completion of co-SSPedi, the guardian would complete proxy-SSPedi and the patient would complete SSPedi or mini-SSPedi (depending on age).

The instruments were administered using an electronic platform named Supportive care Prioritization, Assessment and Recommendations for Kids (SPARK) on an iPad (15,16). Two research team members were present for each participating dyad; one focused on the guardian, and the second focused on the child. During completion of co-SSPedi, we recorded who read the instruction, who read the questions, and who held the iPad. Proxy SSPedi and self-report SSPedi (SSPedi or mini-SSPedi) were administered concurrently, with the guardian and child in separate areas of the room to allow for independent completion.

After administration of co-SSPedi, proxy-SSPedi, and self-report SSPedi (with the order being randomized), independent feedback from both patients and guardians was elicited using a semistructured interview. More specifically, we asked about preference for completing SSPedi (co-SSPedi or separately) in general and preference for completing SSPedi if they were reporting symptoms 3 times a week.

Randomization details

The allocation sequence was computer-generated, and it was concealed from all participants, health-care, professionals and research team members. Consenting dyads were randomized 1:1 to the 2 administration approaches using permuted block sizes and stratified by age group (4-7, 8-10, 11-14, and 15-18 years), relapse (yes vs no), and inpatient status (inpatient vs outpatient at the interview) as our previous data suggested that these factors were the most strongly associated with higher total SSPedi scores (17).

Statistical analysis

Demographic variables were compared between randomized groups using Wilcoxon rank sum test for continuous variables and χ² or Fisher exact test for categorical variables.

The primary analysis compared total SSPedi scores between co-SSPedi first vs guardian proxy-SSPedi first and co-SSPedi first vs self-report SSPedi first according to the randomization sequence using linear regression adjusting for the stratification factors of age group, relapse, and inpatient status. Comparisons of individual symptom scores were calculated using ordinal logistic regression adjusting for the stratification factors of age group, relapse, and inpatient status. To determine whether the results varied by age, we separately compared individual symptom scores by SSPedi type according to age group (4-10 and 11-18 years).

For the secondary analysis of co-SSPedi, proxy-SSPedi, SSPedi, and mini-SSPedi variability, SSPedi score standard deviation was described. We also compared the proportion reporting severely bothersome symptoms (scores of 3 or 4 – “a lot” or “extremely” bothersome) between co-SSPedi vs guardian proxy-SSPedi and co-SSPedi vs self-report SSPedi within individuals before and after the crossover using McNemar test. Since we anticipated carryover when co-SSPedi was completed first, we focused on those completing proxy-SSPedi and self-report SSPedi first. Analysis of interview questions regarding preferences was descriptive.

All tests of significance were 2-sided, and we used a P value less than .05 to define statistical significance. Although the primary analysis included 2 comparisons, we did not adjust the P value to account for multiple testing given the exploratory nature of this analysis. Analyses were performed using R Studio, version 3.6.1 (The R Foundation for Statistical Computing).

Sample size was based on a minimal important difference of 2.7 and standard deviation of SSPedi scores of 8.5, which were derived from the SSPedi validation study (2). Assuming 90% power and alpha of 0.05, we required 210 dyads per group, and thus we targeted enrollment of 420 dyads.

Results

Between August 19, 2021, and August 3, 2023, 578 dyads were assessed for eligibility and 420 were enrolled; 213 were randomized to complete co-SSPedi first and 207 were randomized to complete proxy-SSPedi and self-report SSPedi first (Figure 1). Among enrolled participants, all completed the first allocated SSPedi type. However, 4 dyads randomized to complete co-SSPedi first did not complete the alternate SSPedi type (proxy-SSPedi [n = 2] and self-report SSPedi [n = 2]) because the entries failed to save on the device related to a technical issue. Table 1 illustrates the demographic characteristics of the participants stratified by randomized groups. The median patient age was 11.1 years. Most participants were White, had a diagnosis of leukemia, and were on active treatment. Most guardians were mothers.

Figure 1.

Flow diagram of dyad identification, enrollment, and study participation. CONSORT diagram illustrating dyad identification, enrollment, participation, and reasons for exclusion.

Table 1.

Participant demographics stratified by allocation^a

Characteristic	Co-SSPedi first, n (%) n = 213	Guardian proxy- or self-report SSPedi first, n (%) n = 207	Pa
Patient characteristics
Sex, n (%)			.519
Female	92 (43.2%)	82 (39.6%)
Male	121 (56.8%)	125 (60.4%)
Age group in years			.947
4-7	64 (30.0%)	63 (30.4%)
8-10	40 (18.8%)	38 (18.4%)
11-14	64 (30.0%)	58 (28.0%)
15-18	45 (21.1%)	48 (23.2%)
English as first language, n (%)	184 (86.4%)	179 (86.4%)	1.000
Inpatient at interview, n (%)	93 (43.7%)	94 (45.4%)	.793
In school, n (%)	49 (87.5%)	49 (86.0%)	1.000
Race, n (%)			.483
Aboriginal	2 (3.6%)	0 (0)
Arab/West Asian	4 (7.1%)	2 (3.5%)
Asian	13 (23.2%)	21 (36.8%)
Black	1 (1.8%)	5 (8.8%)
Latin American	0 (0.0)	1 (1.8%)
White	27 (48.2%)	26 (45.6%)
Mixed	7 (12.5%)	1 (1.8%)
Unknown/missing	1 (1.8%)	0 (0.0)
Prefer not to say	1 (1.8%)	1 (1.8%)
Diagnosis, n (%)			.578
Leukemia	93 (43.7%)	87 (42.0%)
Lymphoma	35 (16.4%)	25 (12.1%)
Solid tumor	51 (23.9%)	55 (26.6%)
Brain tumor	23 (10.3%)	24 (10.6%)
Not cancer	11 (5.2%)	16 (7.7%)
Metastatic disease, n (%)	47 (22.1%)	44 (21.3%)	.782
Relapse, n (%)	33 (15.5%)	25 (12.1%)	.383
Active treatment, n (%)	159 (74.6%)	149 (72.0%)	.612
Treatments received, n (%)
Chemotherapy	193 (90.6%)	189 (91.3%)	.938
Surgery	56 (26.3%)	58 (28.0%)	.773
Radiotherapy	38 (17.8%)	45 (21.7%)	.379
Stem cell transplantation	29 (13.6%)	27 (13.0%)	.977
Median months from diagnosis (range)	8 (0, 165)	12 (0, 212)	.405
Median days from chemotherapy (range)	2 (0, 4532)	4 (0, 5739)	.340
Reason for visit, n (%)
Chemotherapy or transplant	113 (53.1%)	102 (49.3%)	.499
Surgery	9 (4.2%)	12 (5.8%)	.607
Fever	24 (11.3%)	15 (7.2%)	.211
Bloodwork	9 (4.2%)	7 (3.4%)	.844
Routine visit	50 (23.5%)	56 (27.1%)	.464
Other	22 (10.3%)	30 (14.5%)	.251
Guardian characteristics
Male, n (%)	52 (24.4%)	52 (25.1%)	.956
Median age in years (range)	42.0 (19.0, 63.0)	42.0 (19.0, 76.0)	.837
English as first language	129 (60.6%)	128 (61.8%)	.867
Relationship to patient			.414
Father	52 (24.4%)	49 (23.7%)
Mother	157 (73.7%)	147 (71.0%)
Other	4 (1.9%)	11 (5.4%)
Married	180 (84.5%)	164 (79.2%)	.201
College or university education	178 (83.6%)	166 (80.2%)	.440
House income > $60 000	115 (54.0%)	119 (57.5%)	.533

a

P values calculated using Wilcoxon rank sum test for continuous variables and χ² or Fisher exact test for categorical variables. SSPedi = Symptom Screening in Pediatrics Tool.

For the primary analysis, the mean total SSPedi scores (± standard deviation) by randomized group were as follows: co-SSPedi first (9.6 ± 7.1), proxy-SSPedi first (9.7 ± 7.5; P = .950 for comparison vs co-SSPedi), and self-report SSPedi first (9.7 ± 8.2; P = .981 for comparison vs co-SSPedi). Table 2 shows the specific SSPedi symptom scores by randomized group.

Table 2.

First co-SSPedi, proxy-SSPedi, or self-report SSPedi scores by randomized group

SSPedi scores	Co-SSPedi first (n = 213)	Guardian proxy-SSPedi first (n = 207)	P value Co-SSPedi vs proxy-SSPedia	Self-report SSPedi first (n = 207)	P value Co-SSPedi vs self-report SSPedia
Feeling Disappointed or Sad			.003		0.743
Not at all bothered	125 (58.7%)	91 (44.0%)		124 (59.9%)
A little	45 (21.1%)	57 (27.5%)		41 (19.8%)
Medium	27 (12.7%)	40 (19.3%)		27 (13.0%)
A lot	10 (4.7%)	15 (7.2%)		7 (3.4%)
Extremely bothered	6 (2.8%)	4 (1.9%)		8 (3.9%)
Feeling Scared or Worried			.078		.450
Not at all bothered	119 (55.9%)	91 (44.0%)		122 (58.9%)
A little	45 (21.1%)	72 (34.8%)		46 (22.2%)
Medium	34 (16.0%)	24 (11.6%)		26 (12.6%)
A lot	8 (3.8%)	18 (8.7%)		10 (4.8%)
Extremely bothered	7 (3.3%)	2 (1.0%)		3 (1.4%)
Feeling Cranky or Angry			.002		.098
Not at all bothered	112 (52.6%)	75 (36.2%)		122 (58.9%)
A little	56 (26.3%)	72 (34.8%)		51 (24.6%)
Medium	31 (14.6%)	44 (21.3%)		28 (13.5%)
A lot	13 (6.1%)	10 (4.8%)		2 (1.0%)
Extremely bothered	1 (0.5%)	6 (2.9%)		4 (1.9%)
Problems with Thinking or Remembering Things			.114		.032
Not at all bothered	161 (75.6%)	140 (67.6%)		133 (64.3%)
A little	26 (12.2%)	41 (19.8%)		46 (22.2%)
Medium	13 (6.1%)	18 (8.7%)		16 (7.7%)
A lot	10 (4.7%)	6 (2.9%)		9 (4.3%)
Extremely bothered	3 (1.4%)	2 (1.0%)		3 (1.4%)
Changes in How Your Body or Face Look			.176		.634
Not at all bothered	142 (66.7%)	123 (59.4%)		141 (68.1%)
A little	31 (14.6%)	46 (22.2%)		32 (15.5%)
Medium	22 (10.3%)	19 (9.2%)		20 (9.7%)
A lot	15 (7.0%)	16 (7.7%)		7 (3.4%)
Extremely bothered	3 (1.4%)	3 (1.4%)		7 (3.4%)
Feeling Tired			.008		.012
Not at all bothered	56 (26.3%)	53 (25.6%)		60 (29.0%)
A little	41 (19.2%)	74 (35.7%)		54 (26.1%)
Medium	57 (26.8%)	46 (22.2%)		61 (29.5%)
A lot	34 (16.0%)	29 (14.0%)		21 (10.1%)
Extremely bothered	25 (11.7%)	5 (2.4%)		11 (5.3%)
Mouth Sores			.048		.013
Not at all bothered	188 (88.3%)	167 (80.7%)		163 (78.7%)
A little	11 (5.2%)	21 (10.1%)		19 (9.2%)
Medium	9 (4.2%)	14 (6.8%)		17 (8.2%)
A lot	1 (0.5%)	3 (1.4%)		5 (2.4%)
Extremely bothered	4 (1.9%)	2 (1.0%)		3 (1.4%)
Headache			.933		.329
Not at all bothered	158 (74.2%)	150 (72.5%)		145 (70.0%)
A little	33 (15.5%)	39 (18.8%)		33 (15.9%)
Medium	11 (5.2%)	12 (5.8%)		19 (9.2%)
A lot	4 (1.9%)	5 (2.4%)		6 (2.9%)
Extremely bothered	7 (3.3%)	1 (0.5%)		4 (1.9%)
Hurt or Pain (other than headache)			.140		.132
Not at all bothered	132 (62.0%)	135 (65.2%)		139 (67.1%)
A little	36 (16.9%)	46 (22.2%)		33 (15.9%)
Medium	28 (13.1%)	17 (8.2%)		23 (11.1%)
A lot	10 (4.7%)	8 (3.9%)		6 (2.9%)
Extremely bothered	7 (3.3%)	1 (0.5%)		6 (2.9%)
Tingly or Numb Hands or Feet			.738		.019
Not at all bothered	181 (85.0%)	173 (83.6%)		158 (76.3%)
A little	21 (9.9%)	25 (12.1%)		24 (11.6%)
Medium	10 (4.7%)	8 (3.9%)		12 (5.8%)
A lot	0 (0.0)	1 (0.5%)		6 (2.9%)
Extremely bothered	1 (0.5%)	0 (0.0)		7 (3.4%)
Throwing Up or Feeling Like You May Throw Up			.154		.534
Not at all bothered	140 (65.7%)	147 (71.0%)		126 (60.9%)
A little	28 (13.1%)	26 (12.6%)		36 (17.4%)
Medium	25 (11.7%)	25 (12.1%)		26 (12.6%)
A lot	10 (4.7%)	7 (3.4%)		9 (4.3%)
Extremely bothered	10 (4.7%)	2 (1.0%)		10 (4.8%)
Feeling More or Less Hungry			.091		.104
Not at all bothered	92 (43.2%)	100 (48.3%)		99 (47.8%)
A little	43 (20.2%)	48 (23.2%)		46 (22.2%)
Medium	39 (18.3%)	29 (14.0%)		39 (18.8%)
A lot	21 (9.9%)	22 (10.6%)		12 (5.8%)
Extremely bothered	18 (8.5%)	8 (3.9%)		11 (5.3%)
Changes in Taste			.007		.099
Not at all bothered	166 (77.9%)	134 (64.7%)		145 (70.0%)
A little	16 (7.5%)	39 (18.8%)		26 (12.6%)
Medium	20 (9.4%)	18 (8.7%)		24 (11.6%)
A lot	6 (2.8%)	15 (7.2%)		4 (1.9%)
Extremely bothered	5 (2.3%)	1 (0.5%)		8 (3.9%)
Constipation			.587		.940
Not at all bothered	154 (72.3%)	143 (69.1%)		152 (73.4%)
A little	30 (14.1%)	38 (18.4%)		25 (12.1%)
Medium	19 (8.9%)	14 (6.8%)		16 (7.7%)
A lot	6 (2.8%)	9 (4.3%)		12 (5.8%)
Extremely bothered	4 (1.9%)	3 (1.4%)		2 (1.0%)
Diarrhea			.151		.023
Not at all bothered	185 (86.9%)	169 (81.6%)		163 (78.7%)
A little	15 (7.0%)	23 (11.1%)		21 (10.1%)
Medium	6 (2.8%)	8 (3.9%)		10 (4.8%)
A lot	3 (1.4%)	2 (1.0%)		6 (2.9%)
Extremely bothered	4 (1.9%)	5 (2.4%)		7 (3.4%)

a

Compared using ordinal logistic regression adjusting for the stratification factors of age group, relapse, and inpatient status.

Co-SSPedi scores were significantly different from proxy-SSPedi for 5 symptoms: feeling disappointed or sad, feeling cranky or angry, feeling tired, mouth sores, and changes in taste. Fewer dyads reported severely bothersome symptoms for feeling disappointed or sad, feeling cranky or angry, and changes in taste for co-SSPedi compared with proxy-SSPedi, whereas more dyads reported severely bothersome scores for feeling tired for co-SSPedi compared with proxy-SSPedi. Co-SSPedi scores were significantly different from self-report SSPedi scores for problems with thinking or remembering things, feeling tired, mouth sores, tingly or numb hands or feet, and diarrhea. Fewer dyads reported severely bothersome mouth sores, tingly or numb hands or feet, and diarrhea for co-SSPedi compared with self-report SSPedi, whereas more dyads reported severely bothersome symptoms for feeling tired for co-SSPedi compared with self-report SSPedi. Supplementary Tables 1 and 2 (available online) show the comparison of co-SSPedi vs proxy-SSPedi and co-SSPedi vs self-report SSPedi stratified by age 4-10 years and 11-18 years of age. Substantial differences were not observed by age strata.

Table 3 shows the number of respondents reporting severely bothersome symptoms among those randomized to proxy-SSPedi or self-report SSPedi first. Significantly fewer dyads reported severely bothersome symptoms for feeling scared or worried for co-SSPedi compared with proxy-SSPedi, and significantly fewer dyads reported severely bothersome symptoms for throwing up or feeling like you may throw up for co-SSPedi compared with self-report SSPedi.

Table 3.

Number with severely bothersome symptoms among those where proxy-SSPedi and self-report SSPedi completed first (N = 207)

Outcome measures (%)	Co-SSPedi	Guardian proxy-SSPedi	P value Co-SSPedi vs proxy-SSPedia	Self-report SSPedi	P value Co-SSPedi vs self-report SSPedia
Feeling disappointed or sad	17 (8.2%)	19 (9.2%)	.838	15 (7.2%)	.724
Feeling scared or worried	7 (3.4%)	20 (9.7%)	.004	13 (6.3%)	.149
Feeling cranky or angry	12 (5.8%)	16 (7.7%)	.423	6 (2.9%)	.181
Problems with thinking or remembering things	6 (2.9%)	8 (3.9%)	.723	12 (5.8%)	.114
Changes in how your body or face look	13 (6.3%)	19 (9.2%)	.264	14 (6.8%)	1.000
Feeling tired	33 (15.9%)	34 (16.4%)	1.000	32 (15.5%)	1.000
Mouth sores	5 (2.4%)	5 (2.4%)	1.000	8 (3.9%)	.371
Headache	6 (2.9%)	6 (2.9%)	1.000	10 (4.8%)	.289
Hurt or pain (other than headache)	11 (5.3%)	9 (4.3%)	.773	12 (5.8%)	1.000
Tingly or numb hands or feet	7 (3.4%)	1 (0.5%)	.077	13 (6.3%)	.077
Throwing up or feeling like you may throw up	10 (4.8%)	9 (4.3%)	1.000	19 (9.2%)	.027
Feeling more or less hungry than you usually do	22 (10.6%)	30 (14.5%)	.170	23 (11.1%)	1.000
Changes in taste	9 (4.3%)	16 (7.7%)	.121	12 (5.8%)	.579
Constipation (hard to poop)	10 (4.8%)	12 (5.8%)	.773	14 (6.8%)	.343
Diarrhea (watery, runny poop)	8 (3.9%)	7 (3.4%)	1.000	13 (6.3%)	.267

a

Compared using McNemar test.

Table 4 shows the observed characteristics during co-SSPedi completion. Most commonly, guardians read the instructions and questions. However, the patient read them solely or in part for about one-third of dyads. Preferences for co-SSPedi, proxy-SSPedi, and SSPedi/mini-SSPedi were heterogeneous, both in general and if they were completing SSPedi 3 times weekly. However, most often, co-SSPedi was preferred.

Table 4.

Patient and guardian preferences for SSPedi administration elicited by interview (n = 420)

Characteristics	n (%)
Observed outcomes
Who read instructions for co-SSPedi (%)
Patient	71 (16.9%)
Guardian	275 (65.5%)
Both read some questions	72 (17.1%)
Mini-SSPedi audio	2 (0.5%)
Who read questions for co-SSPedi (%)
Patient	50 (11.9%)
Guardian	269 (64.0%)
Both read some questions	97 (23.1%)
Mini-SSPedi audio	4 (1.0%)
Who held iPad for co-SSPedi (%)
Patient	136 (32.4%)
Guardian	238 (56.7%)
Both	46 (11.0%)
Patient preferences
Preference for completing SSPedi (%)
co-SSPedi	188 (44.8%)
SSPedi/mini-SSPedi	134 (31.9%)
No preference	98 (23.3%)
Complete SSPedi 3 times per week (%)
co-SSPedi	210 (50.0%)
Child self-report only	133 (31.7%)
Guardian proxy-report only	45 (10.7%)
Both but separately	32 (7.6%)
Guardian preferences
Preference for completing SSPedi (%)
co-SSPedi	244 (58.1%)
Proxy-SSPedi	74 (17.6%)
No preference	102 (24.3%)
Complete SSPedi 3 times per week (%)
co-SSPedi	265 (63.1%)
Child self-report only	45 (10.7%)
Guardian proxy-report only	79 (18.8%)
Both but separately	31 (7.4%)

Discussion

In this randomized trial, we showed that total co-SSPedi scores were not significantly different from either total proxy-SSPedi scores or total self-report SSPedi scores. However, when individual symptoms scores were examined, some co-SSPedi scores were significantly different from proxy-report scores and self-report scores with inconsistent findings as to whether fewer or more dyads reported severely bothersome co-SSPedi scores compared with proxy-report or self-report SSPedi scores.

It is interesting that there was heterogeneity in whether fewer or more co-SSPedi scores showed severely bothersome symptoms compared with proxy-report or self-report scores. One interpretation is that co-SSPedi provides a mechanism to help guardians better understand their child’s symptoms and may improve the accuracy of the child’s symptoms if guardians are reminding their child about timeframes or perhaps clarifying the meaning of a symptom. If such findings are replicated, it may strengthen the use of a dyadic approach to symptom reporting in research and in clinical care.

A secondary objective was to describe the variability of SSPedi scores by SSPedi type. Our results showed the smallest variation (standard deviation) for co-SSPedi followed by proxy-SSPedi and SSPedi/mini-SSPedi. We anticipated that co-SSPedi might exhibit lower variability as the dyad approach might reduce extreme values. Our findings support this hypothesis and provide further motivation for the utilization of co-SSPedi, particularly in research where reduced variability may improve the power of clinical trials.

We also found preferences for SSPedi type to be heterogeneous, suggesting that in clinical practice, different SSPedi types should be offered. In some circumstances, co-SSPedi may be an ideal approach, perhaps for younger patients and patients who are more unwell. However, there will be circumstances where patients, such as older teenagers, may prefer to complete SSPedi on their own. In addition, guardians may not be available when symptom reporting is planned, necessitating self-report. If different SSPedi types are used for symptom assessment during routine clinical care, trending of symptom severity may be more challenging. Thus, it will be important to track reporter type for each SSPedi administration. Based on our results, we recommend that for clinical care, participants should be able to choose their SSPedi type. However, for research, where variability reduction is prioritized, SSPedi type should be standardized. We also recommend that a dyadic approach be considered for other patient-reported outcomes in pediatric populations.

SSPedi was designed for self-report for pediatric patients 8 to 18 years of age. Some may argue that adolescents and young adults are a distinct group and that the items and language may need to differ among respondents within this broad age range. Our previous cognitive interviews did not identify such an issue, and SSPedi allows free text to capture additional items. Nonetheless, that SSPedi stops at 18 years of age is problematic, and future research should develop and evaluate a version of SSPedi for young adults.

The strengths of this study are the randomized design, heterogeneous patient population, and large sample size. In addition, this work promotes patient- and family-centered care by emphasizing the role of both patients and guardians in symptom reporting. However, our results should be interpreted in light of study limitations that include conduct among English-speaking dyads only. In addition, the crossover evaluation comparing different SSPedi types was susceptible to carryover since a washout period was not feasible. Finally, this study is limited by its conduct at a single center. Conducting the study at multiple centers would have improved the generalizability of the findings.

In conclusion, total co-SSPedi scores were not significantly different compared with proxy-report or self-report scores, although there were differences in specific symptom scores. If different reporter types are used during clinical implementation, specifying reporter type will be important.

Data availability

The data underlying this article can be shared with appropriate anonymization to protect the privacy of individuals who participated in the study and with appropriate regulatory approvals in place.

Author contributions

Deborah Tomlinson (Conceptualization; Data curation; Investigation; Methodology; Writing—review & editing), Cassandra Tardif-Theriault (Investigation; Project administration; Supervision; Writing—review & editing), Tal Schechter (Conceptualization; Writing—review & editing), Lee Dupuis (Conceptualization; Writing—review & editing), Lillian Sung (Conceptualization; Data curation; Formal analysis; Methodology; Validation; Writing—original draft).

Funding

Not applicable.

Conflicts of interest

None declared.

Acknowledgments

LS is supported by the Canada Research Chair in pediatric oncology supportive care.