Figure 1 gives a breakdown of the studies identified (7). Searches identified 1416 records and, after the removal of duplicates, 869 unique records were screened. Of these, 243 were sought for retrieval and 52 records were found to be eligible for inclusion. No additional records were identified by hand searching the references of nine reviews of n-of-1 trials.
Study and population characteristics
Table 1 gives the study and population characteristics. Eight studies (15.4%) were identified as being conducted in rare conditions. Twelve (23.1%) of the included articles were protocols for n-of-1 studies and the remainder (n = 40, 76.9%) reported on the results of completed studies. They spanned a range of health and disease areas, with the most common being neuropsychiatric conditions (n = 16, 30.8%). Four studies (7.7%) were not conducted in a specific disease area e.g. blood transfusion dependent patients.
Most of the studies included adult participants (n = 40, 76.9%). Twenty-three (44.2%) included adults aged 65 years and over (“older adults”) and nine (17.3%) included participants aged under 18 years. Studies were mostly of series of n-of-1 trials; 11 (21.2%) studies included a single participant only. The median (IQR) number of participants randomised was ten (1–20) and the number of participants completing the studies was seven (1–12). Of the 12 included study protocols, 10 (83.3%) reported a target sample size. The median (IQR) target sample size was 35 (12–43).
Design characteristics
Table 2 gives the design characteristics for the studies. The studies were mostly of pharmaceutical interventions (n = 35, 67.3%) and were mostly placebo controlled (n = 38, 73.1%).
The median (IQR) number of health technologies compared in the studies was two (2–2). Studies most commonly compared two health technologies (n = 43, 82.7%). Seven studies (13.5%) compared three health technologies, and one compared four (3.8%). The median (IQR) number of periods was six (5–8); 22 studies had six periods (43.3%). The median (IQR) period length was 14 days (5–28).
The majority of studies blinded participants to treatment allocation (n = 41, 78.8%). Of those that did not incorporate blinding, one was of a pharmaceutical intervention (9.1%). The others were of intervention types that are typically more difficult to blind e.g. behavioural and dietary interventions.
Washout periods were used in 23 studies (44.3%). The median (IQR) washout length was 7 days (2–14). The median (IQR) total study duration was 84 days (42–203).
The majority of studies used a patient reported outcome measure (PROM) as the primary outcome (PROMs, n = 37, 71.2%).
Table 1
Study characteristics and population of included studies (n = 52)
Study and Population Characteristics | n (%) or median (IQR) |
Protocol | |
Yes | 12 (23.1) |
No | 40 (76.9) |
Funding | |
Yes | 32 (61.5) |
No | 8 (15.4) |
Not Given | 12 (23.1) |
Country | |
UK | 5 (9.6) |
Netherlands | 6 (11.5) |
France | 3 (5.8) |
China | 7 (13.5) |
Australia | 7 (13.5) |
USA | 9 (17.3) |
Canada | 6 (11.5) |
Finland | 1 (1.9) |
Germany | 1 (1.9) |
Italy | 2 (3.8) |
Norway | 1 (1.9) |
Portugal | 1 (1.9) |
Sweden | 1 (1.9) |
Korea | 1 (1.9) |
Colombia | 1 (1.9) |
Health/disease area | |
Cancer | 2 (3.8) |
Cardiovascular | 4 (7.7) |
Chronic pain | 2 (3.8) |
Gastrointestinal | 3 (5.8) |
Genitourinary | 2 (3.8) |
High cholesterol | 1 (1.9) |
Musculoskeletal | 4 (7.7) |
Neuropsychiatric | 16 (30.8) |
Non-specific | 4 (7.7) |
Nutrition | 1 (1.9) |
Other | 3 (5.8) |
Physical activity | 3 (5.8) |
Pulmonary/respiratory | 5 (9.6) |
Thyroid disorder | 1 (1.9) |
Diabetes | 1 (1.9) |
Rare condition | |
Yes | 8 (15.4) |
No | 44 (84.6) |
Number of participants randomised | 10 (7–20) |
Number of participants completing | 7 (1–12) |
Age of included participants* | |
Children | 9 (17.3) |
Adults | 40 (76.3) |
Older adults (65 + years) | 23 (44.2) |
Target sample size (protocols only) | 35 (12–43) |
* Categories are not mutually exclusive | |
Table 2
Design characteristics of included studies (n = 52)
Design characteristics | n (%) or median (IQR) |
Intervention type | |
Pharmaceutical | 35 (67.3) |
Medical device | 4 (7.7) |
Surgery | 1 (1.9) |
Behavioural | 3 (5.8) |
Other | 9 (17.3) |
Comparator type* | |
Placebo | 38 (73.1) |
Active treatment | 12 (23.1) |
No intervention | 4 (7.7) |
Number of health technologies** | 2 (2–2) |
2 | 43 (82.7) |
3 | 7 (13.5) |
4 | 2 (3.9) |
Number of periods | 6 (5–8) |
Period length (days) | 14 (5–28) |
Blinding | |
Yes | 41 (78.8) |
No | 11 (21.2) |
Washout period | |
Yes | 23 (44.2) |
No | 29 (55.8) |
Washout period length (days) | 7 (2–14) |
Total study duration (days) | 84 (42–203) |
Primary outcome measurement | |
Behavioural test | 2 (3.8) |
Clinical assessment | 1 (1.9) |
Patient reported outcome | 37 (71.2) |
Physical activity measure | 2 (3.8) |
Physiological | 9 (17.3) |
Lab parameter and patient reported | 1 (1.9) |
Primary outcome measured at regular intervals | |
Yes | 33 (63.5) |
No | 19 (36.5) |
Primary outcome measurement frequency | |
More than daily | 2 (3.8) |
Daily | 15 (28.8) |
Weekly | 9 (17.3) |
Every 10 days | 1 (1.9) |
Fortnightly | 2 (3.8) |
Four-weekly | 3 (5.8) |
Eight-weekly | 1 (1.9) |
Not applicable | 19 (36.5) |
* Categories are not mutually exclusive ** Includes placebo and no intervention |
Analysis characteristics
Table 3 gives the analysis characteristics for the study. The included studies used a range of approaches for their analysis. The most common of these were: regression models (n = 14, 26.9%); t-tests (n = 13, 25.0%); Bayesian approaches (n = 8, 15.4%) and non-parametric analyses (n = 8, 15.4%). Eight studies stated they did not use formal analysis methods (15.4%) and methods of analysis were not reported in six studies (11.5%).
The assessment of effect for treatment was undertaken in terms of statistical significance in 18 studies (34.6%), clinical significance in 19 studies (36.5%) and not reported in 15 studies (28.8%).
The majority of studies combined the data from multiple n-of-1 trials or, in the case of study protocols, planned to do so (n = 32, 61.5%). There were nine studies which involved multiple participants but did not combine the data across n-of-1 trials (17.3%). Combining data from multiple n-of-1 trials was not possible in trials including only one patient (n = 11, 21.2%).
Table 3
Analysis characteristics of included studies (n = 52)
Analysis Characteristics | n (%) |
Definition of treatment response | |
Statistical | 18 (34.6) |
Clinical | 19 (36.5) |
Not Given | 15 (28.8) |
Analysis approach* | |
Bayesian | 8 (15.4) |
Non-parametric | 8 (15.4) |
Graph or visual inspection | 2 (3.8) |
t-test | 13 (25.0) |
Regression model | 14 (26.9) |
No formal statistical analysis | 8 (15.4) |
Other | 1 (1.9) |
Not Given | 6 (11.5) |
Individual data pooled | |
Yes | 21 (40.4) |
No | 9 (17.3) |
Planned | 11 (21.2) |
N/A (single patient) | 11 (21.2) |
Numerical results reported | |
Yes | 25 (48.1) |
No | 27 (51.9) |
P values reported | |
Yes | 17 (32.7) |
No | 35 (67.3) |
*Categories are not mutually exclusive | |
Studies in rare conditions
Table 4 summarises the study characteristics of the trials that were undertaken in rare conditions. Generally the study designs were consistent with n-of-1 trials for all n-of-1 studies. Additional file 2 gives more details on these studies.
Table 4
n-of-1 trials in rare conditions identified by the review.
Study (first author) | Condition | Intervention type | Number of periods | Number of health technologies | Comparator(s) | Primary outcome measurement |
Bech (8) | Renal magnesium wasting | Pharmaceutical | 9 | 3 | Active treatment | PROM |
Benhamou (9) | Highly unstable type 1 diabetes (brittle diabetes) | Medical device | 4 | 2 | Active treatment | Physiological parameter |
Jensen (10) | Bronchopulmonary dysplasia | Other (enteral feeding method) | 4 | 2 | Active treatment | Physiological parameter |
Lipka (11) | Myasthenia gravis | Pharmaceutical | 6 | 2 | Placebo | PROM |
Stunnenberg (12) | Nondystrophic myotonia | Pharmaceutical | Up to 8 | 2 | Placebo | PROM |
Vrinten (13) | Myasthenia gravis | Pharmaceutical | 6 | 2 | Placebo | PROM |
Weijma (14) | Hyperkalaemic periodic paralysis | Pharmaceutical | 2 | 2 | Placebo | PROM |
Weng (15) | Myasthenia gravis | Pharmaceutical | 6 | 2 | Placebo | PROM |
PROM: Patient reported outcome measure |