Background
Methods
Protocol and registration
Eligibility criteria
Information sources, search strategy and study selection
Data collection process
Data items
:
age; sex; socioeconomic level (variables referring to income, education, occupation); comorbidities (variables referring to additional diseases or disorders co-occurring with the disease of interest or a measure of comorbidity); severity of the disease under treatment (variables referring to different levels of advancement of the disease of interest); medical history (variables referring to past medical events, previous diseases, or genetic disorders); others; and the form of treatment (variables referring to different doses or different exposure periods or different administration form of the same treatment). Measuring the effect of different forms of treatment can be regarded as not heterogeneity in the effect of treatment but as heterogeneity in the treatment per se. Therefore, HTE studies include those where the form of treatment only was looked at; however descriptive findings are also reported excluding these studies.Risk of bias in individual studies
Data analysis
Additional analyses
Results
Study selection
Study characteristics
Study characteristics | All studies (n = 150) | Studies addressing HTE (n = 88) | Studies not addressing HTE (n = 62) | p-value | Studies addressing HTEa,* (n = 111) |
---|---|---|---|---|---|
Sample size, Med(Q1-Q3) | 1633(310–9505) | 3754(850–20,474) | 470(130–2210) | <0.001 | 1235(201–6133) |
Length of follow up (years), Med(Q1-Q3) | 3.5(1.2–7.2) | 5(2–10) | 2(0.5–4.2) | <0.001 | 3(1–6) |
Date of publication | 0.04 | ||||
2000–2013 | 72(48) | 47(53.4) | 23(37) | 52(47) | |
2014–2016 | 78(52) | 41(46.6) | 39(63) | 59(53) | |
Journal type, n(%) | <0.001 | ||||
High impact factor journals (HIJ) | 75(50) | 55(62.5) | 20(32.2) | 52(47) | |
Low impact factor journals (LIJ) | 75(50) | 33(37.5) | 42(67.8) | 59(53) | |
Source of funding, n(%) | 0.85 | ||||
Industry | 100(66.7) | 62(70.5) | 38(61.3) | 69(62) | |
Other | 31(20.7) | 19(21.6) | 10(16.1) | 22(20) | |
NA | 19(12.6) | 7(8) | 14(22.6) | 20(18) | |
Study area, n(%) | <0.001 | ||||
Interventional | 50(32.7) | 19(21.6) | 31(50) | 54(49) | |
Pharmacological | 89(60) | 64(72.7) | 25(40.3) | 56(50) | |
Other | 11(7.3) | 5(5.7) | 6(9.7) | 1(1) | |
Type of primary outcome, n(%) | 0.004 | ||||
Time to event | 65(44) | 48(54.5) | 17(27.4) | 46(41) | |
Binary | 50(32.7) | 24(27.3) | 26(42) | 37(33) | |
Continuous | 35(22) | 16(18.2) | 19(30.6) | 28(26) | |
Main effect of primary outcome, n(%) | 0.6 | ||||
Statistically significant | 96(64.6) | 58(65.9) | 38(61.3) | 69(62) | |
Not statistically significant | 54(35.4) | 30(34.1) | 24(38.7) | 42(38) | |
Newcastle-Ottawa Quality Assessment Score, n(%) | 0.82 | ||||
0–3 | 5(3) | 3(3) | 2(3) | 3(3) | |
4–6 | 88(59) | 54(61) | 34(55) | 64(58) | |
7–8 | 57(38) | 31(35) | 26(29) | 44(40) | |
HTE reporting | 88(58.7) | 88(100) | – | 49(44) | |
Analysis of predictive factors on side effects, n(%) | 2(1.3) | 2(2.3) | – | 2(2) | |
Prespecification of HTE analysis, n(%) | 84(56) | 84(95.4) | – | 46(94) | |
Type of HTE analysis, n(%) | |||||
Subgroup | 2(1.3) | 2(2.3) | – | 2(4) | |
Adjusted | 67(44.7) | 67(76.1) | – | 34(69) | |
Propension | 19(12.7) | 19(21.6) | – | 13(27) | |
Predictive variables studied, n(%) | |||||
Age | 30(20) | 30(34) | – | 30(61) | |
Sexe | 15(10) | 15(17) | – | 15(31) | |
Social level | 4(2.7) | 4(4.5) | – | 4(8) | |
Genetics | 12(8) | 12(13.6) | – | 12(24) | |
Treatment | 66(44) | 66(75) | – | 27(55) | |
Comorbidities | 31(20.7) | 31(35) | – | 31(63) | |
Severity of disease | 13(8.7) | 13(14.7) | – | 13(26) | |
Prognostic variables studied, n(%) | 51(34) | 31(35.2) | 20(3.2) | 0.73 | 31(28) |
Analysis of prognostic factors on side effect, n(%) | 2(1.3) | 1(1.1) | 1(1.6) | 0.71 | 1(1) |
Interaction test, n(%) | 27(18) | 27(31) | 0(0) | – | 8(7) |
Logistic univariate regression analysis | Logistic multivariable regression analysis | |||
---|---|---|---|---|
Study characteristics | Odds ratio (95% CI)* | P-value | Odds ratio (95% CI)* | P-value |
Sample size | 0.99(0.99–1.00) | 0.11 | 1(0.99–1) | 0.58 |
Length of follow up (years) | 1.17(1.07–1.30) | 0.02 | 1.1(0.99–1.24) | 0.07 |
Date of publication (<2014 vs ≥2014) | 0.5(0.25–0.97) | 0.04 | 0.6(0.16–2.17) | 0.42 |
Journal type (high impact factor) | 3.5(1.78–7.5) | <0.001 | 0.4(0.1–1.55) | 0.19 |
Source of funding(industry funded vs non industry funded) | 1 .92(0.82–4.78) | 0.85 | 0.37 | |
Study area (pharmacological vs non pharmacological) | 0.26(0.13–0.51) | <0.001 | 0.25(0.09–0.66) | 0.007 |
Type of primary outcome | ||||
Binary | 1(reference) | – | 1(reference) | – |
Time to event | 3.05(1.41–6.80) | 0.05 | 2.22(0.72–7.12) | 0.17 |
Continuous | 0.91(0.38–2.17) | 0.83 | 0.98(0.26–3.7) | 0.98 |
Main effect of primary outcome (significant vs non significant) | 1.22(0.62–2.40) | 0.56 | 0.97(0.34–2.71) | 0.95 |
Newcastle-Ottawa Quality Assessment Score, n(%) (continuous) | ||||
0–3 4–6 7–8 | ref 1[0.13–6.70] 0.79[0.09–5.14] | ref 0.95 0.81 | ref 1.61(0.17–1.99) 1.75(0.12–2.3) | ref 0.70 0.66 |
Study characteristics | Cohort studies in High impact journals (n = 75) | Cohort studies in low impact journals (n = 75) | p-value* |
---|---|---|---|
Sample size, Median(Interquartile range) | 4176(1274–23,452) | 415(138.5–2586) | <0.001 |
Length of follow up (years), Median(Interquartile range) | 5(2–10) | 2.1(1–5) | 0.002 |
Date of publication | <0.001 | ||
2000–2013 | 57(76) | 13(17) | |
2014–2016 | 17(24) | 62(83) | |
Journal type, n(%) | |||
High impact factor journals (HIJ) | 75(100) | – | |
Low impact factor journals (LIJ) | – | 75(100) | |
Study area, n(%) | 0.81 | ||
Interventional | 23(30.7) | 27(34) | |
Pharmacological | 46(61.3) | 43(57) | |
Other | 6(8) | 5(9) | |
Source of funding, n (%) | 0.15 | ||
Industry | 51(68) | 49(65) | |
Other | 18(24) | 12(16) | |
NA | 6(8) | 14(19) | |
Type of primary outcome, n (%) | 0.001 | ||
Time to event | 44(58.7) | 21(28) | |
Binary | 19(25.3) | 31(41) | |
Continuous | 12(16) | 23(31) | |
Main effect of primary outcome, n (%) | 0.02 | ||
Statistically significant | 55(73) | 41(54.7) | |
Not statistically significant | 20(27) | 34(45.3) | |
Newcastle-Ottawa Quality Assessment Score, n(%) | 0.03 | ||
0–3 | 1(2) | 4(5) | |
4–6 | 46(61) | 42(56) | |
7–8 | 28(37) | 29(39) | |
HTE reporting | 55(73) | 33(44) | <0.001 |
Analysis of predictive factors on side effects, n(%) | 2(4) | 0(0) | 0.53 |
Prespecification of HTE, n(%) | 54(98) | 30(91) | 0.15 |
Type of HTE analysis, n(%) | 0.44 | ||
Subgroup | 1(2) | 1(3) | |
Adjusted | 40(73) | 27(82) | |
Propension | 14(25) | 5(15) | |
Predictive variables studied, n(%) | |||
Age | 23(42) | 7(21) | 0.06 |
Sexe | 12(22) | 3(9) | 0.15 |
Social level | 2(4) | 2(6) | 0.63 |
Genetics | 2(4) | 1(3) | 0.03 |
Treatment | 44(80) | 22(67) | 0.21 |
Comorbidities | 23(42) | 8(24) | 0.11 |
Severity of disease | 9(16) | 4(12) | 0.76 |
Prognostic variables studied, n(%) | 27(36) | 24(73) | 0.7 |
Analysis of prognostic factors on side effect, n(%) | 1(1) | 1(3) | 1 |
Interaction test, n(%) | 21(28) | 6(8) | 0.06 |
Study characteristics | Pharmacological cohort studies (n = 89) | Non pharmacological cohort studies (n = 61) | p-value* |
---|---|---|---|
Sample size, Median(Interquartile range) | 3434(490–20,482) | 1041(151–2714) | <0.001 |
Length of follow up (years), Median(Interquartile range) | 4(1.7–8.3) | 3(1–6) | 0.14 |
Date of publication | 0.67 | ||
2000–2013 | 44(49) | 27(45) | |
2014–2016 | 45(51) | 34(55) | |
Journal type, n(%) | 0.74 | ||
High impact factor journals (HIJ) | 45(51.1) | 29(48) | |
Low impact factor journals (LIJ) | 44(48.9) | 32(52) | |
Study area, n(%) | |||
Non pharmacological | – | 61(100) | |
Pharmacological | 89(100) | – | |
Source of funding, n(%) | 0.04 | ||
Industry | 61(68) | 39(63) | |
Other | 15(17) | 13(22) | |
NA | 13(15) | 9(15) | |
Type of primary outcome, n(%) | 0.37 | ||
Time to event | 42(47) | 24(39) | |
Binary | 30(33) | 19(31) | |
Continuous | 17(20) | 18(30) | |
Main effect of primary outcome, n(%) | 0.55 | ||
Statistically significant | 53(60) | 42(68) | |
Not statistically significant | 36(40) | 19(32) | |
Newcastle-Ottawa Quality Assessment Score, n(%) | 0.5 | ||
0–3 | 3(4) | 2(3) | |
4–6 | 52(58) | 36(59) | |
7–8 | 34(38) | 23(38) | |
Analysis of predictive factors on side effects, n(%) | 1(2) | 1(2) | 0.49 |
HTE reporting | 63(71) | 24(39) | <0.001 |
Prespecification of HTE, n(%) | 59(66) | 24(100) | 0.59 |
Type of HTE analysis, n(%) | 0.40 | ||
Subgroup | 2(3) | 0(0) | |
Adjusted | 50(79) | 16(67) | |
Propension | 11(18) | 8(33) | |
Predictive variables studied, n(%) | |||
Age | 19(30) | 10(42) | 0.19 |
Sexe | 11(17) | 4(17) | 1 |
Social level | 2(3) | 2(8) | 0.33 |
Genetics | 9(14) | 3(12) | 1 |
Treatment | 53(84) | 13(54) | 0.002 |
Comorbidities | 19(30) | 11(46) | 0.13 |
Severity of disease | 9(14) | 3(12) | 0.18 |
Prognostic variables studied, n(%) | 30(48) | 21(34) | 0.77 |
Analysis of prognostic factors on side effect, n(%) | 1(2) | 1(2) | 0.52 |
Interaction test, n(%) | 20(22) | 7(11) | 0.27 |
Study characteristics | All studies (n = 57) | Studies addressing HTE (n = 31) | Studies not addressing HTE (n = 26) | p-value* |
---|---|---|---|---|
Sample size, Med(Q1-Q3) | 2438(415–19,486) | 7484(1259–70,336) | 470(166–2453) | <0.001 |
Length of follow up (years), Med(Q1-Q3) | 3(1–7) | 5(2.2–11.8) | 1(0.1–4) | 0.002 |
Date of publication | 0.2 | |||
2000–2013 | 28(49) | 18(58) | 10(38.5) | |
2014–2016 | 29(51) | 13(42) | 16(61.5) | |
Journal type, n(%) | <0.001 | |||
High impact factor journals (HIJ) | 28(49) | 22(71) | 6(23) | |
Low impact factor journals (LIJ) | 29(51) | 9(29) | 20(77) | |
Source of funding, n(%) | 0.68 | |||
Industry | 36(63) | 23(74) | 13(19) | |
Other | 10(17) | 5(16) | 5(50) | |
NA | 11(20) | 3(10) | 8(31) | |
Study area, n(%) | 0.02 | |||
Medical | 34(60) | 23(74) | 11(42) | |
Non medical | 23(40) | 8(26) | 15(58) | |
Type of primary outcome, n(%) | 0.05 | |||
Time to event | 26(46) | 18(58) | 8(31) | |
Binary | 16(28) | 5(16) | 11(42) | |
Continuous | 15(26) | 8(26) | 7(27) | |
Main effect of primary outcome, n(%) | 1 | |||
Statistically significant | 36(63) | 20(65) | 16(62) | |
Not statistically significant | 21(37) | 11(35) | 10(38) | |
Analysis of predictive factors on side effects, n(%) | 0(0) | 0(0) | – | |
Prespecification of HTE analysis, n(%) | 30(53) | 30(97) | – | |
Type of HTE analysis, n(%) | ||||
Subgroup | 1(2) | 1(3) | – | |
Adjusted | 23(40) | 23(74) | – | |
Propension | 7(12) | 7(23) | – | |
Predictive variables studied, n(%) | ||||
Age | 13(23) | 18(58) | – | |
Sexe | 7(12) | 7(23) | – | |
Social level | 2(3) | 2(6) | – | |
Genetics | 3(5) | 3(10) | – | |
Treatment | 27(47) | 27(87) | – | |
Comorbidities | 8(14) | 8(26) | – | |
Severity of disease | 6(11) | 6(19) | – | |
Prognostic variables studied, n(%) | 20(35) | 10(32) | 10() | 0.78 |
Analysis of prognostic factors on side effect, n(%) | 0(0) | 0(0) | 0(0) | – |
Interaction test, n(%) | 6(10) | 6(19) | 0(0) | – |