Preliminary results of this study have been previously reported and published as an oral presentation conference abstract(19).
Screening and Recruitment
Figure 1: Consort flow diagram
A total of 106 patients were identified as potentially eligible from the weekly MDT list, based on the pre-screening (i.e. eligible diagnosis) (figure 1). However, 36 (34%) were excluded after MDT discussion. Service re-configuration changes implemented after the study was planned meant that these patients were offered and received treatment at other hospital sites closer to their home. As this feasibility trial had approval for a single site only, these patients were not approached for participation. Of the remaining 70 patients who met the criteria for pre-screening based on diagnosis, a total of 26 [0.37, 95% CI 0.27 to 0.49] patients were subsequently excluded following eligibility assessment; eleven [0.42, 95% CI 0.26 to 0.61] patients had unsatisfactory proficiency in English determined by the requirement for an interpreter at consultations. Six [0.23, 95% CI 0.11 to 0.42] patients presented with medical co-morbidities and pre-existing dysphagia unrelated to HNC. Nine [0.35, 95% CI 0.19 to 0.54] patients had to have a change in treatment (for example, total laryngectomy, or treatment with palliative intent). Of the remaining 44 eligible patients, five [0.11, 95% CI 0.05 to 0.24] were not approached in the clinic as they left prior to be seen by the researcher. Seven [0.18, 95% CI 0.08 to 0.33] of the 39 patients approached declined participation. There was some overlap in the reasons provided, but the main stated reasons were: not wanting to be involved in anything extra while having treatment (3/7) [0.43, 95% CI 0.16 to 0.75]; concern about additional hospital visits (2/7), [0.29,95% CI 0.08 to 0.64] and pre-existing participation in another trial (2/7) [0.29,95% CI 0.08 to 0.64]. None of the patients reported declining based on anything in the patient information leaflet or any concerns about the intervention itself. Thus, of the 39 eligible patients approached during the “face to face” clinic visit, 32 [0.82, 95% CI 0.67 to 0.91] were recruited.
Figure 2. Accrual into SIP SMART
Figure 2 illustrates accrual to the SIP SMART trial over an 8-month duration from mid April –mid December (spanning 9 calendar months). The pattern shows a steady rate of recruitment following an initial slower start. The target recruitment (based on prior knowledge of referral patterns into the centre) of an average of four patients a month was shown to be feasible.
Patient characteristics
Table 1. Patient characteristics
Participant demographics
|
Intervention group (n=16)
|
Care as usual group(n=16)
|
Age mean (SD)
|
58.56 (12.41)
|
55.19 (9.45)
|
Gender n (%)
Male
Female
|
15 (94%)
1 (6%)
|
12 (75%)
4 (25%)
|
Ethnicity n (%)
White
Asian/Asian British
Black/ Black British
Chinese
Other
|
11 (69%)
3 (19%)
0 (0%)
2 (12%)
0 (0%)
|
12 (75%)
1 (6%)
1 (6%)
0 (0%)
2 (13%)
|
Marital status n (%)
Married
Single/separated
Widowed
Co-habiting
Divorced
|
4 (25%)
5 (31%)
1 (6%)
2 (13%)
4 (25%)
|
10 (63%)
5 (31%)
0 (0%)
1 (6.%)
0 (0%)
|
Employment status n (%)
Full-time
Part-time
Self-employed
Not employed
Retired
|
7 (44%)
3 (19%)
1 (6%)
3 (19%)
2 (12%)
|
1 (6%)
1 (6%)
5 (31%)
6 (38%)
3 (19%)
|
Occupation n (%)
Manager/director
Graduate professional
Associate professional/technical
Admin/secretarial
Skilled trade
Sales/customer services
Caring/leisure
Other
|
2 (13%)
4 (25%)
1 (6%)
1 (6%)
2 (13%)
1 (6%)
4 (25%)
1 (6%)
|
1 (6%)
0 (0%)
0 (0%)
2 (13%)
5 (31%)
1 (6%)
6 (38%)
1 (6%)
|
AJCC numeric tumour stage
III
IV
|
6 (37.5%)
10 (62.5%)
|
6 (37.5%)
10 (62.5%)
|
Tumour site n (%)
Oral cavity
Nasopharynx
Oropharynx
Hypopharynx/larynx
|
5 (31%)
1 (6%)
9 (56%)
1 (6%)
|
3 (19%)
0 (0%)
10 (62%)
3 (19%)
|
Cancer treatment n (%)
Surgery
Radiotherapy
Surgery & radiotherapy
Radiotherapy & chemotherapy
All three
|
1 (6%)
4 (25%)
1 (6%)
9 (56%)
1 (6%)
|
0 (0%)
1 (6.%)
2 (12%)
11 (69%)
2 (13%)
|
Other treatment n (%)
Nasogastric tube
Gastrostomy tube
Neither/NA
|
4 (25%)
10 (62%)
2 (13%)
|
4 (25%)
11 (69%)
1 (6%)
|
Weight Baseline mean (SD)
|
72.46 (15.37)
|
78.46 (15.36)
|
BMI Baseline mean (SD)
|
24.71 (3.77)
|
27.14 (4.19)
|
The baseline characteristics of all patients recruited to the study are displayed in Table 1. Despite the small sample size, the groups were generally well balanced, and any differences likely due to chance. The average age of participants across the groups was 57 years with five times as many males than females recruited, consistent with the general disease prevalence. The sample included the most common curative options for treatment, with stratification by first line treatment ensuring that a balance between the groups was achieved. Baseline weight and body mass index were comparable across groups. Information regarding baseline smoking and alcohol history has not been included in the summary table as this information was inconsistently reported on the case report forms.
Patient acceptability of participation and randomisation
Table 2. Patient responses to *questionnaire on acceptability to participation and randomisation (* Jenkins et al, 2000)
No
|
Statement
|
Agree
|
Unsure
|
Disagree
|
|
|
%
|
CAU
|
INT
|
%
|
CAU
|
INT
|
%
|
CAU
|
INT
|
1
|
I thought the trial/study offered the best treatment available
|
100
|
12
|
12
|
-
|
-
|
-
|
-
|
-
|
-
|
2
|
I believed the benefits of treatment in the trial would outweigh the side effects.
|
71
|
6
|
11
|
29
|
6
|
1
|
-
|
-
|
-
|
3
|
I was satisfied that either treatment in the trial would be suitable.
|
83
|
10
|
10
|
13
|
1
|
2
|
4
|
1
|
-
|
4
|
I was worried that my illness would get worse unless I joined the trial.
|
17
|
2
|
2
|
8
|
2
|
-
|
75
|
8
|
10
|
5
|
The idea of randomization worried me.
|
17
|
4
|
-
|
13
|
1
|
2
|
70
|
7
|
10
|
6
|
I wanted a doctor to choose my treatment rather than randomized by computer
|
54
|
8
|
5
|
21
|
4
|
1
|
25
|
0
|
6
|
7
|
The doctor told me what I needed to know about the trial.
|
67
|
9
|
7
|
13
|
2
|
1
|
20
|
1
|
4
|
8
|
I trusted the doctor treating me.
|
96
|
12
|
11
|
-
|
-
|
-
|
4
|
-
|
1
|
9
|
I was given too much information to read about the trial.
|
42
|
7
|
3
|
13
|
-
|
3
|
45
|
5
|
6
|
10
|
I was given enough information to read about the trial.
|
88
|
11
|
10
|
8
|
1
|
1
|
4
|
-
|
1
|
11
|
I knew I could leave the trial at any time and still be treated.
|
100
|
12
|
12
|
-
|
-
|
-
|
-
|
-
|
-
|
12
|
I did not feel able to say no.
|
4
|
1
|
-
|
8
|
2
|
-
|
88
|
9
|
12
|
13
|
I wanted to help with the doctor’s research
|
100
|
12
|
12
|
-
|
-
|
-
|
-
|
-
|
-
|
14
|
I feel that others with my illness will benefit from the results of the trial.
|
100
|
12
|
12
|
-
|
-
|
-
|
-
|
-
|
-
|
15
|
The doctor wanted me to join the trial.
|
38
|
4
|
5
|
37
|
5
|
4
|
25
|
3
|
3
|
16
|
Others, for example, family or friends, wanted me to join the trial.
|
42
|
7
|
3
|
12
|
3
|
-
|
46
|
2
|
9
|
Note: % = overall percentage from all respondents, CAU = number in care as usual group, INT = number in intervention group
Table 2 shows responses from 24 patients who returned the questionnaire, categorised into agree, unsure, disagree with the given statements. Patients reported high agreement with several statements indicating they felt that the trial offered the best treatment (SLT intervention) available (100%), they were satisfied that either treatment group would be suitable (83%), were given sufficient information about the trial (88%) and were aware that they could leave the trial at any time without their care being compromised (100%). Given a choice, just over half of patients indicated that they wanted their doctor or health care professional to choose their treatment rather than being randomised by a computer. Four patients (17%) reported being worried by the idea of randomisation. One patient reported feeling unable to decline participation but was aware that it was possible to leave the trial at any stage. All patients indicated that they wanted to help with the research, and felt that other patients would benefit from the results.
Adherence to intervention
Figure 3. Patient reported adherence (INT = intervention, CAU = usual care)
The completeness of adherence data was 77% (24) at 1 month, 70% (21) at 3 months, and 83% (24) at 6 months, after accounting for the three deaths. Figure 3 shows the percentage of patients within each group who demonstrated satisfactory to good adherence based on the responses to the adherence form.
Satisfactory to good adherence was greater than the 35% minimum threshold for the intervention group across all time-points thus meeting the stipulated criterion. The results also show reports of good adherence from the usual care group, who were given a generic exercise sheet.
Candidate outcome measures, data completeness and sample size estimation
Table 3. Completeness of swallowing outcome measures across time-points for both groups
Measure
|
Obtained
Intervention [INT] no (%)
|
Obtained
Care as usual [CAU] no(%)
|
Comments or reasons for
non-completion – (researcher diary)
|
Baseline (T0)
[expected INT = 16, CAU =16]
FACT
MDADI
PSS (normalcy of diet)
100mL WST
FIGS (swallowing)
MIO (jaw opening)
|
16 (100)
16 (100)
16 (100)
16 (100)
16 (100)
16 (100)
|
15 (94)
15 (94)
16 (100)
16 (100)
16 (100)
16 (100)
|
One patient did not complete questionnaires at appointment, and did not respond to requests to return them at next visit.
|
One Month (T1)
[expected INT =16, CAU = 15]
FACT
MDADI
PSS
100mL WST
FIGS
MIO
|
13 (81)
13 (81)
14 (88)
*13 (81)
14 (88)
14 (88)
|
15 (100)
15 (100)
14 (93)
*10 (67)
14 (93)
12 (80)
|
*100mL WST - Clinical notes indicate that several patients reported that they were unable to drink continuously at 1-month post treatment, or clinicians felt it was unsafe to ask them to do so. These patients were given a score of zero and still rated complete.
|
Three Months (T2)
[expected INT = 15, CAU=15]
FACT
MDADI
PSS
100mL WST
FIGS
MIO
|
11 (73)
11 (73)
11 (73)
*8 (53)
11 (73)
11 (73)
|
12 (75)
12 (75)
8 (53)
*8 (53)
8 (53)
8 (53)
|
*100mL WST – As above. Additionally, it was noted that clinicians sometimes did not attempt this measure if the water cooler in the clinic was not working (n=3) – availability of resources.
Four patients did not attend their 3-month follow-up visit, but did not wish to drop out of the study.
|
Six Months (T3)
[expected INT =15, CAU =14]
FACT
MDADI
PSS
100mL WST
FIGS
MIO
**MBS Impairment Profile Score
PAS Score
|
12 (80)
12 (80)
13 (87)
13 (87)
13 (87)
13 (87)
9 (60) 87
9 (60) 87
|
13 (93)
13 (93)
12 (86)
12 (86)
12 (86)
12 (86)
10 (71) 93
10 (71) 93
|
Improvement in collection of most measures in comparison to 3-months as patients were attending for their MBS swallow assessment at the same time as clinical measures were taken.
Due to an unavoidable technical issue (power surge causing loss of exams) **MBS exams could not be rated for 7 patients (INT =4, CAU =3), although for both groups > 70% completed the procedure.
3 patients declined to have the 6-month MBS (INT=2, CAU =1)
|
Notes: FACT = Functional Assessment of Cancer Therapy; MDADI = MD Anderson Dysphagia Inventory; PSS = Performance Status Scale; WST = Water Swallow Test; FIGS = Functional Intra-oral Glasgow Scale; MIO = Maximum Incisor Opening, MBS = Modified Barium Swallow; PAS = Penetration-Aspiration Scale.
Table 3 shows that the minimum threshold of 70% was achieved for most measures within both study groups except for the water swallow test (WST). Three patients declined to have the 6-month MBS; two reported that their swallowing had improved and they could not attend the appointment within the required timeframe and the other reported that he did not wish to have further exposure to radiation after completing his treatment.
For the remaining outcome measures, complete datasets across the four time-points were available for eight patients in the intervention group and 11 in CAU. Excluding the three patients who died, this figure represents an overall total of 66%, [95% CI 0.47 - 0.80]. If only the baseline and the 6-month endpoint are considered, the figure increases to 86%, [95% CI 0.70-.0.94]. This means that despite having multiple outcome measures, data completeness was at 86% when considering the baseline and the 6-month endpoint. While none of the patients dropped out of the study, loss of data due to missing information was 14% [95% CI 0.06 – 0.30] when considering the baseline and 6-month endpoint.
Table 4. Main swallow-related outcome measures and effect sizes (mean, standard deviation and 95% confidence intervals at baseline and final time-point included)
Outcome measure
|
Group
|
Baseline (T0) mean
(±95% CI) SD
|
Between-group effect size
(Cohen’s d)
|
6-months (T3) mean
(±95% CI) SD
|
Between-group effect size
(Cohen’s d)
|
FACT-H&N and MDADI: [n(baseline) intervention=16 CAU=15; n (T3) intervention=12 CAU=13]
|
FACT-H&N total score
|
Intervention
Care as usual
|
104.38 (91.74-117.02) 23.71
101.28 (89.03-113.54) 22.13
|
0.14
|
89.98 (79.24-100.72) 16.9
76.9 (68.02-85.78) 14.69
|
0.83
|
MDADI Composite
|
Intervention
Care as usual
|
83.49 (75.17-91.81) 15.62
77.82 (68.93-86.72) 16.06
|
0.36
|
69.74 (56.42-83.05) 20.95
59.35 (52.03-66.67) 12.11
|
0.61
|
Other measures: [n(baseline) intervention=16 CAU=16; n (T3) intervention=13 CAU=12]
|
MIO - Jaw opening
|
Intervention
Care as usual
|
46.25 (39.44-53.06) 12.78
47.81 (43.83-51.99) 9.84
|
-0.15
|
43.00 (37.24-48.76) 9.53
34.33 (24.56-44.1) 15.38
|
0.68
|
PSS HN Normalcy of diet
|
Intervention
Care as usual
|
70.00 (56.24-83.76) 25.82
75.63 (62.43-88.82) 24.76
|
-0.22
|
70.00 (52.21-87.79) 29.44
60.83 (41.21-80.46) 30.88
|
0.30
|
FIGS Swallowing
|
Intervention
Care as usual
|
4.25 (3.84-4.66) .78
4.50 (4.11-4.89) .73
|
-0.35
|
4.15 (3.42-4.89) 1.21
3.50 (2.71-4.29) 1.24
|
0.53
|
MBS and PAS at T3: (intervention =9 CAU =10) – not performed for both groups at baseline
|
PAS
|
Intervention
Care as usual
|
|
|
3.67 (1.24-6.1) 3.16
3.3 (1.36-5.24) 2.71
|
0.13
|
MBS Imp (composite)
|
Intervention
Care as usual
|
|
|
6.44 (4.49-8.38) 2.53
5.96 (5.05 – 6.86) 1.27
|
0.24
|
Table 4 shows that effect sizes (20)were large (≥ 0.8) for the Functional Assessment of Cancer Therapy (FACT-HN) and moderate (≥ 0.5) for the MD Anderson Dysphagia Inventory (MDADI), Maximal Incisor Opening (MIO) and Functional Intra-oral Glasgow Scale (FIGS).
Sample size calculation:
Due to the low numbers available for analysis on the MBS composite score at 6-months, it was not possible to reliably estimate sample size for this measure. The effect size of the PSS was found to be small, meaning that sample size of 340 (+82, accounting for 24% attrition) would be required to detect clinically important changes. The MDADI met most criteria for a primary outcome measure and showed a moderate effect size. On balance from the measures available, the MDADI was considered the most suitable choice for a primary outcome (see Discussion). Sample size was therefore estimated for the MDADI patient reported outcome. Using the data obtained from the feasibility study, the sample size required if the MDADI were the primary outcome would be 86 (43 in each group). Based on this feasibility study, this figure will need to be inflated by 10% for attrition due to death and an estimated 14% due to missing data. A sample size of 106 will therefore be required for a future 2-arm parallel group trial.
Harms
There were no unexpected serious adverse events related to the trial.