Background
Methods
Search strategy and Selection for studies
Data extraction
Data analyses
Results
HRQoL measurement
First authors | HRQoL questionnaire | Timing of assessment | Targeted dimensions | MCID | Statistical approach for dealing with missing data | Statistical approach for HRQoL analysis |
---|---|---|---|---|---|---|
Wu [19] | QLQ-C30; QLQ-LC13 | Randomization and every 3 weeks until disease progression or new cancer treatment | Cough, dyspnoea, pain | 10 points | Distribution of patients whose symptom had improved, remained stable, or worsened; the time to deterioration of symptoms; mixed-effects growth curve model | |
Laurie [20] | QLQ-C30; QLQ-LC13 and two additional questions (hand-foot syndrome and headache) | At baseline and every cycle | ||||
Yang [7] | QLQ-C30; QLQ-LC13 | Random assignment and every 21 days until disease progression | Cough, dyspnoea, pain | 10 | Joint analysis | Time to deterioration; mixed-effects growth curve model |
Wu [19] | FACT-L | Not available | FACT-L total score and TOI | Time to deterioration | ||
Yoshioka [21] | FACT-L; FACT/GOGNTX | At the time of enrollment and at 6 and 9 weeks after the initiation of treatment | Linear mixed model for repeated measures | |||
Lee [22] | QLQ-C30, QLQ-LC14, EUROQOL | Baseline, monthly during the first year, then 18 and 24 months after randomization | Linear mixed model for repeated measures | |||
Gridelli [23] | ||||||
Flotten [24] | QLQ-C30, QLQ-LC13 | Before each cycle, 3 weeks after the last cycle, and then every 8 weeks until 57 weeks | Global health status health, nausea/vomiting, dyspnoea, fatigue | 10 | Mean change from baseline | |
Socinsky [25] | FACT-G, FACTTAXANE | Baseline and on day 1 of each cycle | Peripheral neuropathy, pain, hearing, edema | Mean change from baseline | ||
Groen [26] | QLQ-C30, QLQ-LC13 | Start of chemotherapy and weeks 6,12,16,24,30 | Mean change from baseline | |||
Chen [9] | FACT-L | Baseline and every 6 weeks | 6 for FACT-L total score and TOI; 2 points for LCS LCS | Mean change from baseline | ||
Lara [27] | QLQ-C30 | On day 1 of each odd cycle and at the end of the treatment visit | Mean change from baseline | |||
Koch [12] | QLQ-C30, QLQ-LC13 | Baseline, weeks 3,6,9,12,20,28 | Dyspnoea, fatigue, pain, pain medication, global health status | Group comparisons of scale at each time; mean change from baseline; AUC; rates of symptom palliation | ||
Biesma [28] | QLQ-C30, QLQ-LC13 | Before, during (day 1 and day 8 of each cycle) and after each cycle (weeks 12,15,18) | Global health status | 10 | Mean change from baseline and week 18; linear mixed model for repeated measures | |
Weissman [29] | FACT-L | After each cycle | TOI | Mean change from baseline and after 6 cycles | ||
Okamoto [30] | FACT-L, FACT/GOGNTX | Time of enrollment and at 6 and 9 weeks after initiation of treatment | Linear mixed model for repeated measures | |||
Thongpraset [31] | FACT-L | Baseline, weeks 1 AND 3, 3-weekly until week 18, 6 weekly until progression and at discontinuation | FACT-L Score, TOI, LCS | 6 for FACT-L total score and TOI; 2 points for LCS | If less than 50% of the Fact-l subscale scores were missing, the subscale score was divided by the number of completed items and multiplied by the total number of items on the scale. If 50% or more of the items were missing, that subscale was treated as missing for that patient | Mean change from baseline; Time to worsening; Time to improvement |
Lynch [32] | FACT-L | Baseline, before each treatment cycle and at the end of the therapy | Mean change from baseline; Time to symptomatic disease progression | |||
Takeda [33] | FACT-L | The time of enrollment and at 12 weeks, 18 weeks after initiation of treatment | LCS | Linear mixed-effects model in which the missing data depend on the observed score | Linear mixed model for repeated measures | |
Lee [34] | QLQ-C30, QLQ-LC14 | Random, During each cycle, at the end of the chemotherapy, every 6 months until 24 months | Linear mixed model for repeated measures | |||
Treat [35] | FACT-L | Not available | ||||
Tan [36] | Lung cancer symptom scale | Baseline, at the end of each cycle, just before the next cycle, at the end of the study | ||||
Pirker [37] | QLQ-C30, QLQ-LC13, EuroQoL EQ-5D | Not available | ||||
Gronberg [38] | QLQ-C30, QLQ-LC13 | Weeks 0,3,6,9,12,20,28,36,44,52 | Global health status health, nausea/vomiting, dyspnoea, fatigue | 10 | Last value carried forward for missing value that followed, even after death | Area under the curve |
O’Brien [39] | FACT-L | Baseline, before each cycle and at the end of the treatment | LCS | 2 | Missing scores at week 3 were classified as having less than a 2-point increase in the primary analysis data but classified as missing and excluded from the supplemental analysis | Fisher test for equal proportion of patients achieving at least two points increase |
Langer [40] | FACT-L | Baseline and within 3 days of each treatment | LCS | 2 | Percentage of patients with at least two points improvement at the beginning of cycle 2 | |
Gebbia [41] | QLQ-C30; QLQ-LC13 | Baseline and every cycle |
Statistical analysis of HRQoL
Yes, n (%) | Not clear, n (%) | No, n (%) | |
---|---|---|---|
Targeted dimensions | 13 (48.1) | 0 | 14 (51.9) |
HRQoL hypothesis | 2 (7.4) | 0 | 25 (92.6) |
Procedure to control the type I error | 1 (3.7) | 1 (3.7) | 25 (92.6) |
Minimal clinically important difference | 9 (33.3) | 1 (3.7) | 17 (63) |
Study population | 3 (11.1) | 3 (11.1) | 21 (77.8) |
Number of HRQoL data at subsequent time points | 7 (25.9) | 6 (22.2) | 14 (51.9) |
HRQoL scores at baseline for each group and each dimension | 6 (22.2) | 2 (7.4) | 19 (70.4) |
Profile of missing data at baseline | 1 (3.7) | 2 (7.4) | 24 (88.9) |
Statistical approaches for dealing with missing data | 5 (18.5) | 0 | 22 (81.5) |
Statistical approach for HRQoL analysis | 14 (51.9) | 1 (3.7) | 12 (44.4) |
MCID taken into account in the statistical analysis | 7 (25.9) | 2 (7.4) | 18 (66.7) |
Multivariate analysis | 1 (3.7) | 0 | 26 (96.4) |