SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

Objective

To determine whether smoking reduces the progression of osteoarthritis (OA).

Methods

Observational studies examining smoking and progression of OA were systematically searched through Medline (1948–), EMBase (1980–), Web of Science, PubMed, and Google and relevant references. The search was last updated in May 2012. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were directly retrieved or calculated. Current standards for reporting meta-analyses of observational studies (Meta-Analysis of Observational Studies in Epidemiology) were followed. Quality-related aspects such as study design, setting, sample selection, definition of progression, and confounding bias were recorded. Stratified and meta-regression analyses were undertaken to examine the covariates.

Results

Sixteen studies (976,564 participants) were identified from the literature. Overall, there was no significant association between smoking and progression of OA (OR 0.92; 95% CI 0.83, 1.02). There was moderate heterogeneity of results (I2 = 57.3%, P = 0.0024). Subgroup analyses showed some associations of marginal significance; however, meta-regression did not confirm any significant results.

Conclusion

There is no compelling evidence that smoking has a protective effect on the progression of OA. The results concur with a previous meta-analysis published by this group that showed no association between smoking and incidence of OA. Taken together, smoking does not appear to reduce either the incidence or progression of OA.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

Osteoarthritis (OA) is a common, complex disorder with multiple risk factors. It carries a large community health burden, especially for hip and knee OA ([1]). The first studies that examined smoking as a possible risk factor reported a negative association between smoking and the development of OA ([2, 3]) and subsequently with the progression of OA ([4-7]). However, this negative association was not supported by a number of other studies ([8-14]). Evidence about the possible mechanism for the negative association between smoking and OA is currently sparse. In contrast, there is some evidence that smoking is detrimental for both cartilage and bone, resulting in uncertain effects on the progression of OA ([15, 16]).

Our group recently published a meta-analysis showing no overall association between smoking and the incidence of OA ([17]). Smoking was found to have a positive association with OA only in case–control studies where controls were recruited from a hospital setting (e.g., a cardiovascular clinic) where smoking-related conditions predominate in the population from which the controls were selected. As a result, the control group had a higher exposure to smoking, which led to the negative association with OA ([17]).

This meta-analysis attempts to address the related question of whether smoking is associated with the progression of OA. Differentiating incident from progressive OA is clinically relevant because they are related to different stages of the disease (development and progression) and prevention (primary or secondary prevention) of the outcomes ([18, 19]). Some studies have found that risk factors for development and progression of OA may differ ([20]). Whether smoking has different effects on the different stages in the natural history of the disease remains unknown. The objective of this study was to examine the association between smoking and the progression of OA through a meta-analysis of observational studies.

Box 1. Significance & Innovations

  • This meta-analysis finds no compelling evidence that smoking has a protective effect on the progression of osteoarthritis (OA).
  • The results concur with a previous meta-analysis published by this group that showed no association between smoking and the incidence of OA.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

Literature search

A systematic literature search was carried out in December 2011 and updated in May 2012 using Medline (1948–), EMBase (1980–), Web of Science, PubMed, and Google. The structured search strategy included 1) osteoarthritis and synonyms, 2) smoking and synonyms, 3) types of observational studies, and 4) progression and synonyms (see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.?????/abstract). To capture any studies not included by this search, a further keyword search was performed by a second author (MH) and a review of bibliographies of retrieved studies was performed.

Inclusion/exclusion criteria

We included published observational studies of progression of OA, including smoking as a primary or secondary exposure (i.e., covariate or confounding factor). There was no language restriction. Studies of OA at all sites were included. Progression of OA was defined as either radiographic progression from Kellgren/Lawrence grade ≥2 ([21]) to a higher grade or cartilage loss in established OA observed by magnetic resonance imaging (MRI) scanning ([14]). We also included total joint replacement (TJR) as a surrogate marker for the progression of OA ([22]). In addition, a group of studies that focused on cartilage loss measured by MRI in normal subjects was analyzed separately to examine the association between smoking and change of cartilage volume irrespective of OA.

Review articles and editorials and reports regarding other recognized joint diseases, such as rheumatoid arthritis, were excluded. In the case of several reports available about a single population, the study with the “better” methodology was used, i.e., the study with the larger sample size or longer observation, or in the instance of a case–control study nested within a cohort, the cohort study was selected. One author (FP) read all of the abstracts to determine eligibility; a random sample of 20 of these abstracts was verified by a second author (MH).

Data extraction

Data were fully extracted and assessed by a single investigator (FP). Major outcome measures that included assessment of study methods such as definition of progression, study design, and setting were double extracted by 2 authors (MH, WZ) blinded to the results of the first author (FP). Disagreements were discussed and ratified.

Odds ratios (ORs), relative risks, or hazard ratios and their respective 95% confidence intervals (95% CIs) were extracted (or calculated from available data or following correspondence with the authors of the articles) to allow further analysis. For simplicity, we used the term OR to address all relative risk measures. Where available, adjusted ORs (adjusted for age, sex, body mass index [BMI], etc.) were used.

Study characteristics were recorded (see Supplementary Appendix B, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21954/abstract). Parameters recorded included study design, mean age, sex split, mean BMI, length of followup, setting (hospital or community), joint involved, whether smoking was a primary exposure, diagnosis of OA, definition of OA progression, definition of smoking, and country of study. Funding sources for each study were carefully examined to see whether they included tobacco industry funding or affiliation.

Quality assessment

Study design, sample size, setting (hospital or community), measures of exposure (smoking) and outcome (progression), adjustment for confounding factors (age, sex, BMI, etc.), and source of funding were assessed. Quality scoring for studies was not performed because it is not possible to assign equal weight to different quality aspects of studies. However, current consensus standards for reporting meta-analysis of observational studies (Meta-Analysis of Observational Studies in Epidemiology) ([23]) were adhered to and assessment of heterogeneity was performed.

For case–control studies, the setting was defined according to the source population of the control group. If the control subjects were identified from the general population (e.g., from the electoral roll or “healthy” subjects registered with general practices), the study was defined as community based. By contrast, if the controls were patients attending the hospital, the study was classified as hospital based.

Statistical analysis

ORs and 95% CIs were used to represent the association between smoking and the progression of OA. The distribution of ORs and 95% CIs was represented using a forest plot. Cochran's Q statistic was used to estimate the P value for heterogeneity and the I2 statistic was calculated to demonstrate the degree of heterogeneity, i.e., the percentage of variation across the studies that is not due to chance. When statistical pooling was required, the random-effects model was used. Publication bias was assessed using a funnel plot and the Begg-Mazumdar and Egger's tests. Stratified (subgroup) analysis was undertaken as appropriate to examine the sensitivity of the estimate to different quality aspects, or effect modification. The analysis was undertaken using StatsDirect, version 2.6.1.

A meta-regression was performed to examine the impact of a subgroup variable on the association (i.e., to identify an effect modifier) given the adjustment for other covariates. The natural logarithms of ORs were used as the dependent variables and the study-level variables were used as the independent variables. Considering the significance in the subgroup analyses and clinical relevance, we included cohort study (yes = 1, no = 0), community setting (yes = 1, no = 0), knee OA (yes = 1, no = 0), radiographic progression (yes = 1, no = 0), and confounding adjustment for age, sex, and/or BMI (yes = 1, no = 0) as the exploratory factors in the meta-regression. The random-effects meta-regression was applied. The relative OR between yes and no of each exploratory variable was converted directly from the partial regression coefficient to present the effect modification (i.e., the interaction between the exploratory variable and the association). We also calculated the adjusted OR for each group to compare with the OR from the subgroup analysis. This was obtained from the antilogarithm value of the sum of the intercept and the partial regression coefficient in the model. The analysis was undertaken using Stata, version 11.

MRI cartilage loss in healthy subjects

There was only one observational study in the published literature that assessed the effect of smoking on cartilage loss in patients with OA by MRI scanning ([14]). Three further reports from 2 studies of cartilage loss assessed by MRI scanning in normal subjects were retrieved from the literature ([24-26]). Because they did not meet the inclusion criteria (i.e., non-OA subjects), they were not included in the main analysis, but formed a subgroup of MRI cartilage loss in normal subjects. Percentage cartilage loss per year between ever smoking and no smoking was compared. Additional data were obtained from the authors to give the overall percentage loss of the cartilage annually from one study where the measure was not reported in all subjects ([25]). All studies addressed the knee. Data on different cartilage compartments within the knee were combined for each sample, and then a summary meta-analysis of the 2 study populations was performed.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

Characteristics of the studies

In total, 1,123 citations were retrieved from the literature, and after reading the abstracts, 1,080 were excluded because they were either irrelevant or duplicate articles. Of the 43 remaining eligible abstracts, full-text copies were obtained for further assessment; 16 of these met the inclusion criteria ([3-14, 27-30]) and had data for meta-analysis and 27 were excluded. Of the 27 excluded studies, 5 had insufficient data and no further data were supplied by the authors after contacting them, 5 recruited subjects who did not have OA, 9 assessed the incidence and not the progression of OA, a further 6 that addressed the progression of OA did not meet our criteria for progression, and 2 were duplicate populations (Figure 1). The characteristics of the included studies are briefly described in Table 1 (with more detailed information in Supplementary Appendix B, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21954/abstract).

image

Figure 1. Flow diagram for study selection. OA = osteoarthritis.

Download figure to PowerPoint

Table 1. Characteristics of included studies*
 RadiologicMRI cartilage loss/defectTotal joint replacementTotal
  1. MRI = magnetic resonance imaging; BMI = body mass index; OA = osteoarthritis.

  2. a

    Data missing from 4 studies (total population 323,499).

  3. b

    Data missing from 1 study (total population 100).

  4. c

    Data missing from 7 studies (total population 467,029).

  5. d

    Some studies reported >1 site.

No. of studies511016
No. of subjects, cases/total population studied4,009/6,749159/1598,715/969,65612,883/976,564
Mean age, yearsa71.16862.262.3
Women, %b68.1059.159.2
Mean BMI, kg/m2c24.23123.123.1
Mean length of followup, years4.82.54.37.2
Hospital based, no. (%)0 (0)1 (100)4 (40)5 (31.3)
Setting, no. (%)    
Cohort4 (80)1 (100)6 (60)11 (69)
Cross-sectional1 (20)0 (0)0 (0)1 (6)
Case–control0 (0)0 (0)4 (40)4 (25)
OA sites, no. (%)d    
Knee5 (100)1 (100)2 (20)8 (50)
Hip0 (0)0 (0)4 (40)4 (25)
Knee and hip0 (0)0 (0)4 (40)4 (25)
Smoking, no. (%)    
Primary exposure4 (80)1 (100)7 (70)12 (75)
Secondary exposure1 (20)0 (0)3 (30)4 (25)

Test for publication bias

The funnel plot showed a symmetric distribution of studies for the association between smoking and progression of OA (P > 0.20) (Figure 2), suggesting that there is no publication bias in favor of either a positive or a negative association.

image

Figure 2. Funnel plot showing publication bias for the overall meta-analysis. Tests for publication bias: Begg-Mazumdar test: Kendall's τ = 0.133333, P = 0.5056; Egger's test: bias = 0.916148 (95% confidence interval −0.697169, 2.529465), P = 0.243.

Download figure to PowerPoint

Association between smoking and progression of OA

The overall summary results showed no association between smoking and progression of OA (OR 0.92; 95% CI 0.83, 1.02) (Figure 3). However, there was moderate heterogeneity (I2 = 57.3%, P = 0.0024).

image

Figure 3. Forest plot. Noncombinability of studies: Cochran's Q = 35.13605, 15 df, P = 0.0024; I2 (inconsistency) = 57.3% (95% confidence interval 13.1%, 74.2%). MRI = magnetic resonance imaging; pts = patients.

Download figure to PowerPoint

Subgroup analyses

Subgroup analysis (Table 2) according to a single study-level variable showed equivocal results. For example, community-based studies showed a significant negative association between smoking and the progression of OA (OR 0.89 [95% CI 0.82, 0.95], I2 = 21.4%) ([4, 5, 7-13, 27, 28]), whereas cohort studies demonstrated no association (OR 0.94 [95% CI 0.83, 1.06], I2 = 55.6%) ([4, 5, 7-14, 28]). Although a negative association was seen in the studies of radiographically defined OA progression (OR 0.88 [95% CI 0.79, 0.99], I2 = 0%) ([8, 12, 13, 27, 28]), a positive association was observed in those studies with MRI-defined OA progression (OR 2.14; 95% CI 1.25, 3.66) ([14]).

Table 2. Subgroup analysis: association between smoking and progression of OA according to the study-level variables*
 No. of studiesNo. of subjects, cases/total population studiedOR (95% CI)Test for heterogeneity
I2, %P
  1. OA = osteoarthritis; OR = odds ratio; 95% CI = 95% confidence interval; N/A = not applicable; MRI = magnetic resonance imaging; TJR = total joint replacement; TKR = total knee replacement; THR = total hip replacement.

  2. a

    Significant negative association.

  3. b

    Three reports included male-only populations, 4 included female-only populations, and 8 included mixed populations. Of the reports on mixed populations, 3 reported separate data for each sex (and are therefore included in both the male and female subgroups) and 5 did not (and are therefore included only in the mixed subgroup), and 1 did not report sex at all (and is therefore excluded from the table).

  4. c

    Four studies reported both a “past” vs. “current” smoking status as well as an “ever” smoking status (which combined both past and current smokers) and are therefore included in >1 subgroup.

  5. d

    Significant positive association.

  6. e

    Of the 10 reports of TJRs, 4 reports included THR and TKR. Of these, 2 reported THR and TKR data separately (and are therefore included in both THR and TKR subgroups) and 2 did not (and are therefore reported only in the THR and TKR combined subgroup).

  7. f

    Adjusted for age, body mass index, and sex.

Overall1612,883/976,5640.92 (0.83, 1.02)57.30.0024
Study design     
Cohort1111,317/974,9980.94 (0.83, 1.06)55.60.0127
Cross-sectional11,383/1,3830.85 (0.74, 0.96)aN/AN/A
Case–control41,566/1,5660.85 (0.59, 1.24)750.0073
Study setting     
Hospital51,725/1,7251.00 (0.66, 1.51)81.50.0002
Community1111,158/974,8390.89 (0.82, 0.95)a21.40.2398
Sexb     
Male64,067/347,0950.84 (0.69, 1.03)73.40.0021
Female74,707/622,6201.03 (0.91, 1.16)22.40.2586
Mixed52,782/4,2190.90 (0.80, 1.01)00.4005
Smoking measured as:     
Primary exposure129,909/485,6930.92 (0.82, 1.03)63.60.0015
Secondary exposure42,974/490,8710.89 (0.65, 1.23)34.80.2037
Definition of smokingc     
Ever95,859/826,2410.85 (0.75, 0.97)a56.50.0185
Past42,189/125,0561.11 (0.95, 1.30)13.70.3239
Current72,981/148,8771.05 (0.81, 1.38)39.80.1263
OA sites     
Knee105,237/817,5180.92 (0.81, 1.04)46.60.0508
Hip64,827/933,3660.95 (0.79, 1.14)80.30.0001
Definition of progression     
Radiographic54,009/6,7490.88 (0.79, 0.99)a00.4518
MRI1159/1592.14 (1.25, 3.66)dN/AN/A
TJRe108,715/969,6560.89 (0.79, 1.00)590.009
TKR42,067/812,0560.86 (0.78, 0.96)a00.4622
THR64,827/933,3660.96 (0.79, 1.14)80.30.0001
THR and TKR combined21,821/34,9580.83 (0.69, 1.00)N/A0.0762
Confounding     
Unadjusted75,104/495,0451.00 (0.83, 1.21)45.80.0863
Adjustedf97,779/481,5190.88 (0.78, 0.99)a60.60.0092
Sample size     
<1,000114,306/159,6900.96 (0.81, 1.13)66.60.0009
>1,00058,504/816,8740.88 (0.80, 0.96)a16.60.309

MRI cartilage loss in normal subjects

A meta-analysis of the 2 further studies of cartilage loss assessed by MRI scanning accorded with the only study of OA knees and showed a significantly positive association between smoking and cartilage loss in smokers compared to nonsmokers ([25, 26, 31]), i.e., there was an increased percentage cartilage loss per year in ever smokers compared to nonsmokers (n = 596; pooled change in percentage cartilage volume per year −0.46 [95% CI −0.71, −0.21]).

Meta-regression

Five exploratory variables were dichotomized (yes = 1, no = 0) and included in this analysis, including cohort study, community study, knee OA, radiographic progression, and confounding adjustment for age, sex, and/or BMI. Apart from knee OA, all showed somewhat statistically significant ORs in the subgroups. We included knee OA because this is the most common joint affected by the disease. We did not include smoking and types of smoking (ever, past, and current) because these overlapped with the definition of smoking. We did not include sample size in the model because it is less clinically useful. Of 5 variables, none statistically modified the association between smoking and progression of OA. The relative ORs varied at ∼1, but none were statistically significant (Table 3). Given the adjustment for the other covariates, the overall OR from the meta-regression was 0.94 (95% CI 0.65, 1.35). The statistically significant ORs found in the subgroup analyses in the community setting (OR 0.89; 95% CI 0.82, 0.95), radiographic progression (OR 0.88; 95% CI 0.79, 0.99), and confounding adjustment (OR 0.88; 95% CI 0.78, 0.99) groups no longer existed (see Supplementary Appendix C, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21954/abstract).

Table 3. Meta-regression analysis*
 ln OR (relative OR)SE95% CI (relative OR)tP
  1. ln OR = natural logarithm of the odds ratio (OR) between smoking and the progression of osteoarthritis (OA); relative OR = relative change of the OR due to the exploratory variable (e.g., the risk of the progression associated with smoking decreased 26% [relative OR 0.74] in the community-based studies, but this was not statistically significant); 95% CI = 95% confidence interval; t = t-test for significance.

Intercept−0.079 (0.92)0.179−0.479, 0.32 (0.62–1.38)−0.440.668
Cohort study0.352 (1.42)0.256−0.218, 0.922 (0.80–2.51)1.380.199
Community setting−0.298 (0.74)0.335−1.043, 0.448 (0.35–1.57)−0.890.394
Knee OA0.152 (1.16)0.273−0.455, 0.759 (0.63–2.14)0.560.589
Radiographic progression−0.015 (0.99)0.353−0.801, 0.772 (0.45–2.16)0.040.968
Confounding adjustment−0.111 (0.89)0.174−0.498, 0.32 (0.61–1.32)−0.440.668

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

Our meta-analysis of 16 observational studies of progression of OA with 976,564 participants shows no negative association between smoking and the progression of OA. Some statistically negative associations were observed in the subgroup analyses, e.g., in community-based studies, radiographic OA studies, and studies adjusted for confounding factors (age, sex, and/or BMI, etc.). However, these associations were marginal. The ORs ranged from 0.85–0.96, with the upper limit of the 95% CIs very close to 1. This suggests that even if there is truly a statistically negative association, it is not clinically significant ([32]). In addition, they were not confirmed in the meta-regression when multiple covariates were adjusted, suggesting that smoking is very unlikely to be beneficial for the progression of OA.

This meta-analysis follows our previous meta-analysis, which examined the association between smoking and the incidence of OA ([17]). Both meta-analyses accord in finding no association between smoking and the incidence or progression of OA. Unlike our previous study, we did not observe the negative association in hospital-based case–control studies; therefore, we were unable to attribute the bias ([17]). However, according to the marginal statistical significance and the distribution of the study outcomes around a null association (i.e., OR 1) in the forest plot (Figure 3), the negative association in some studies is more likely to be a random error.

There has been no previous meta-analysis of whether smoking is associated with the progression of OA. Previous large cohort studies have provided conflicting conclusions, with a negative association found in some (the Framingham study in the US [27], the Swedish construction workers study [7], the Million Women Study in the UK [5], and a large community-based cohort in Australia [4]), but not others (the Clearwater [8] and Nurses Health [9] Studies in the US, the Chingford Study [10] in the UK, the Zoetermeer Study in The Netherlands [28], and 2 recent community-based cohorts in Japan [[12, 13]]). Most recently, a very large community-based cohort of more than 23,500 people in Sweden followed up for 30 years showed no association ([11]).

One study showed a significant positive association of smoking with the progression of OA, but this was the only study using MRI assessment of cartilage loss as the definition of OA progression ([14]). This accorded with 2 other studies (with 3 reports [[25, 26, 31]]) of cartilage loss assessed by MRI, albeit in subjects mostly with normal knees, in which smokers showed significantly increased total knee cartilage volume loss compared to nonsmokers even over a short (2–3 years) time period.

The mechanisms by which smoking potentially might affect the progression of OA are unclear. There are conflicting in vitro data of the effect of smoking on chondrocytes at a cellular level. Nicotine may have a beneficial effect on chondrocyte function ([33, 34]), whereas in other studies, components of tobacco have a deleterious effect on chondrocyte function ([15, 35-38]). Smoking has a detrimental effect on bone health ([39]), including new bone growth after fracture ([40]), so it may inhibit osteophyte formation. Smoking is known to be associated with progression of other chronic musculoskeletal conditions, such as low back pain, and intervertebral disc disease ([41-43]) as well as inflammatory diseases such as rheumatoid arthritis ([44, 45]), but the mechanism has not been elucidated. Finally, smoking may be seen as a wider proxy marker of a person's social demographics, lifestyle, weight, etc., which may be involved in the different pathways during the development and progression of OA.

This meta-analysis has several limitations. First, the meta-analysis included the smaller number of studies (16 studies) in the published literature that address the question of the association of smoking with the progression of OA compared with the incidence of OA (44 studies). This may influence the power of the meta-regression analysis, especially when multiple exploratory factors were included. Sixteen studies are obviously underpowered to examine 5 covariates. The results are likely to be false-positive if the analysis detects any significant effect modifier ([46]). Fortunately, no significant variable was identified in this analysis, so the conclusion may be drawn in conjunction with the main analyses. Second, progression of OA itself is difficult to define, and there is also no consensus in the literature of what constitutes progression or the best way to measure it (radiographs, MRI, surrogate of clinical and radiologic progression such as TJR) ([19]). This heterogeneity in the definition of OA progression may account for some of the heterogeneity of the results. Third, there was also diversity in the study design (cohort, case–control, or cross-sectional), study setting (hospital or community), OA site (hip or knee only was included), and definition of smoking (ever, past, or current), which may make the overall meta-analysis suboptimal. We therefore undertook a number of subgroup analyses and meta-regression analysis to examine the contribution of each of these factors and to give an adjusted overall OR. Furthermore, using TJR as a surrogate marker for OA progression may not be legitimate because it is in fact not truly a progression marker (but a measure of severe OA) and is influenced by numerous other factors apart from OA itself.

However, using TJR as a progression measure has some advantages. It will avoid index event bias, a paradox that is often caused by the selection of people with an index event (e.g., OA) to investigate risk factors of the subsequent event (e.g., progression). The paradox has been observed in OA where BMI appears to be a stronger risk factor for development than for the progression of OA ([17]). It has also been observed in other conditions (e.g., stroke, cardiac events) in which factors that appear to predispose to the development of the condition paradoxically are reported not to be associated with its recurrence or progression ([39]). The index event bias normally dilutes the association between the risk factors and the recurrent or progressive events because the selection of the cases with the index event means (to some extent) the selection of the risk factors. The best way to avoid this bias is to recruit cohorts of healthy subjects (rather than after an index event has occurred) and follow up with them for both index and subsequent outcomes (e.g., both development and progression) of interest. This is often not possible; therefore, an alternative approach is to use a severe marker of the disease such as TJR to bypass the bias ([19]). Several other caveats apply to all meta-analyses of this sort: despite rigorous attempts to search the literature, it is possible that some eligible data were not included, and language bias may apply because English search terms were used, although returned studies in all languages were considered eligible.

In summary, this meta-analysis shows no compelling evidence to support the reported negative association between smoking and progression of OA. It aligns with the results of the previous meta-analysis published by this group that showed no association between smoking and the incidence of OA.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Zhang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Pearce, Hui, Doherty, Zhang.

Acquisition of data. Pearce, Hui, Ding, Zhang.

Analysis and interpretation of data. Pearce, Zhang.

Acknowledgments

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

We would like to thank Joanna Ramowski for her help collecting the studies and Helen Richardson for logistic support. We are in debt to Kun Zou for his pertinent help in validating the random-effects meta-regression analysis using Stata. We would also like to thank Umesh Kadam for providing us with the study data that were not published in the studies and the rheumatology department of Nottingham University Hospitals for support for Dr. Pearce's research leave (one day per week for research).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information
  • 1
    Woolf AD, Pfleger B.Burden of major musculoskeletal conditions.Bull World Health Organ2003;81:64656.
  • 2
    Anderson JJ, Felson DT.Factors associated with osteoarthritis of the knee in the first national Health and Nutrition Examination Survey (HANES I): evidence for an association with overweight, race, and physical demands of work.Am J Epidemiol1988;128:17989.
  • 3
    Cooper C, Inskip H, Croft P, Campbell L, Smith G, McLaren M, et al.Individual risk factors for hip osteoarthritis: obesity, hip injury, and physical activity.Am J Epidemiol1998;147:51622.
  • 4
    Mnatzaganian G, Ryan P, Norman PE, Davidson DC, Hiller JE.Smoking, body weight, physical exercise, and risk of lower limb total joint replacement in a population-based cohort of men.Arthritis Rheum2011;63:252330.
  • 5
    Liu B, Balkwill A, Banks E, Cooper C, Green J, Beral V.Relationship of height, weight and body mass index to the risk of hip and knee replacements in middle-aged women.Rheumatology (Oxford)2007;46:8617.
  • 6
    Sandmark H, Hogstedt C, Lewold S, Vingard E.Osteoarthrosis of the knee in men and women in association with overweight, smoking, and hormone therapy.Ann Rheum Dis1999;58:1515.
  • 7
    Jarvholm B, Lewold S, Malchau H, Vingard E.Age, bodyweight, smoking habits and the risk of severe osteoarthritis in the hip and knee in men.Eur J Epidemiol2005;20:53742.
  • 8
    Wilder FV, Hall BJ, Barrett JP.Smoking and osteoarthritis: is there an association? The Clearwater Osteoarthritis Study.Osteoarthritis Cartilage2003;11:2935.
  • 9
    Karlson EW, Mandl LA, Aweh GN, Sangha O, Liang MH, Grodstein F.Total hip replacement due to osteoarthritis: the importance of age, obesity, and other modifiable risk factors.Am J Med2003;114:938.
  • 10
    Nicholls AS, Kiran A, Javaid MK, Hart DJ, Spector TD, Carr AJ, et al.Change in body mass index during middle age affects risk of total knee arthroplasty due to osteoarthritis: a 19-year prospective study of 1003 women.Knee2012;19:3169.
  • 11
    Kadam U, Holmberg A, Blagojevic M, Nilsson P, Akesson K.Risk factors for cardiovascular disease and future osteoarthritis-related arthroplasty: a population-based cohort study in men and women from Malmö, Sweden.Scand J Rheumatol2011;40:47885.
  • 12
    Muraki S, Akune T, Oka H, Kawaguchi H, Nakamura K, Yoshimura N.Risk factors for the incidence and progress rate of radiographic knee osteoarthritis in Japanese men and women: the ROAD study [abstract].Osteoarthritis Cartilage2011;19 Suppl:S150.
  • 13
    Nishimura A, Hasegawa M, Kato K, Yamada T, Uchida A, Sudo A.Risk factors for the incidence and progression of radiographic osteoarthritis of the knee among Japanese.Int Orthop2011;35:83943.
  • 14
    Amin S, Niu J, Guermazi A, Grigoryan M, Hunter DJ, Clancy M, et al.Cigarette smoking and the risk for cartilage loss and knee pain in men with knee osteoarthritis.Ann Rheum Dis2007;66:1822.
  • 15
    Gullihorn L, Karpman R, Lippiello L.Differential effects of nicotine and smoke condensate on bone cell metabolic activity.J Orthop Trauma2005;19:1722.
  • 16
    Haverstock BD, Mandracchia VJ.Cigarette smoking and bone healing: implications in foot and ankle surgery.J Foot Ankle Surg1998;37:6974.
  • 17
    Hui M, Doherty M, Zhang W.Does smoking protect against osteoarthritis? Meta-analysis of observational studies.Ann Rheum Dis2011;70:12317.
  • 18
    Hernborg JS, Nilsson BE.The natural course of untreated osteoarthritis of the knee.Clin Orthop Relat Res1977;123:1307.
  • 19
    Spector TD, Dacre JE, Harris PA, Huskisson EC.Radiological progression of osteoarthritis: an 11 year follow up study of the knee.Ann Rheum Dis1992;51:110710.
  • 20
    Hochberg MC.Do risk factors for incident hip osteoarthritis (OA) differ from those for progression of hip OA?J Rheumatol2004;31 Suppl:7780.
  • 21
    Kellgren JH, Lawrence JS.Radiological assessment of osteo-arthrosis.Ann Rheum Dis1957;16:494502.
  • 22
    Zhang Y, Niu J, Felson DT, Choi HK, Nevitt M, Neogi T.Methodologic challenges in studying risk factors for progression of knee osteoarthritis [review].Arthritis Care Res (Hoboken)2010;62:152732.
  • 23
    Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al, for the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) Group.Meta-analysis of observational studies in epidemiology: a proposal for reporting.JAMA2000;283:200812.
  • 24
    Davies-Tuck ML, Wluka AE, Forbes A, Wang Y, English DR, Giles GG, et al.Smoking is associated with increased cartilage loss and persistence of bone marrow lesions over 2 years in community-based individuals.Rheumatology (Oxford)2009;48:122731.
  • 25
    Ding C, Cicuttini F, Blizzard L, Jones G.Smoking interacts with family history with regard to change in knee cartilage volume and cartilage defect development.Arthritis Rheum2007;56:15218.
  • 26
    Ding C, Martel-Pelletier J, Pelletier JP, Abram F, Raynauld JP, Cicuttini F, et al.Two-year prospective longitudinal study exploring the factors associated with change in femoral cartilage volume in a cohort largely without knee radiographic osteoarthritis.Osteoarthritis Cartilage2008;16:4439.
  • 27
    Felson DT, Anderson JJ, Naimark A, Hannan MT, Kannel WB, Meenan RF.Does smoking protect against osteoarthritis?Arthritis Rheum1989;32:16672.
  • 28
    Schouten JS, van den Ouweland FA, Valkenburg HA.A 12 year follow up study in the general population on prognostic factors of cartilage loss in osteoarthritis of the knee.Ann Rheum Dis1992;51:9327.
  • 29
    Kraus JF, D'Ambrosia RD, Smith EG, Van Meter J, Borhani NO, Franti CE, et al.An epidemiological study of severe osteoarthritis.Orthopedics1978;1:3742.
  • 30
    Vingard E, Alfredsson L, Malchau H.Lifestyle factors and hip arthrosis: a case referent study of body mass index, smoking and hormone therapy in 503 Swedish women.Acta Orthop Scand1997;68:21620.
  • 31
    Racunica TL, Szramka M, Wluka AE, Wang Y, English DR, Giles GG, et al.A positive association of smoking and articular knee joint cartilage in healthy people.Osteoarthritis Cartilage2007;15:58790.
  • 32
    Kaul S, Diamond GA.How to avoid commonly encountered limitations of published clinical trials.J Am Coll Cardiol2010;55:41527.
  • 33
    Gullahorn L, Lippiello L, Karpman R.Smoking and osteoarthritis: differential effect of nicotine on human chondrocyte glycosaminoglycan and collagen synthesis.Osteoarthritis Cartilage2005;13:9423.
  • 34
    Ying X, Cheng S, Shen Y, Cheng X, An RF, Wang W, et al.Nicotine promotes proliferation and collagen synthesis of chondrocytes isolated from normal human and osteoarthritis patients.Mol Cell Biochem2012;359:2639.
  • 35
    Akmal M, Kesani A, Anand B, Singh A, Wiseman M, Goodship A.Effect of nicotine on spinal disc cells: a cellular mechanism for disc degeneration.Spine (Phila Pa 1976)2004;29:56875.
  • 36
    Iwahashi M, Matsuzaki H, Tokuhashi Y, Wakabayashi K, Uematsu Y.Mechanism of intervertebral disc degeneration caused by nicotine in rabbits to explicate intervertebral disc disorders caused by smoking.Spine (Phila Pa 1976)2002;27:1396401.
  • 37
    Oda H, Matsuzaki H, Tokuhashi Y, Wakabayashi K, Uematsu Y, Iwahashi M.Degeneration of intervertebral discs due to smoking: experimental assessment in a rat-smoking model.J Orthop Sci2004;9:13541.
  • 38
    Uematsu Y, Matuzaki H, Iwahashi M.Effects of nicotine on the intervertebral disc: an experimental study in rabbits.J Orthop Sci2001;6:17782.
  • 39
    Kanis JA, Johnell O, Oden A, Johansson H, De Laet C, Eisman JA, et al.Smoking and fracture risk: a meta-analysis.Osteoporos Int2005;16:15562.
  • 40
    Schmitz MA, Finnegan M, Natarajan R, Champine J.Effect of smoking on tibial shaft fracture healing.Clin Orthop Relat Res1999;365:184200.
  • 41
    Deyo RA, Bass JE.Lifestyle and low-back pain: the influence of smoking and obesity.Spine (Phila Pa 1976)1989;14:5016.
  • 42
    Fogelholm RR, Alho AV.Smoking and intervertebral disc degeneration.Med Hypotheses2001;56:53739.
  • 43
    Goldberg MS, Scott SC, Mayo NE.A review of the association between cigarette smoking and the development of nonspecific back pain and related outcomes.Spine (Phila Pa 1976)2000;25:9951014.
  • 44
    Albano SA, Santana-Sahagun E, Weisman MH.Cigarette smoking and rheumatoid arthritis.Semin Arthritis Rheum2001;31:14659.
  • 45
    Heliovaara M, Aho K, Aromaa A, Knekt P, Reunanen A.Smoking and risk of rheumatoid arthritis.J Rheumatol1993;20:18305.
  • 46
    Thompson S, Higgins J.How should meta-regression analyses be undertaken and interpreted.Stat Med2002;21:155973.

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgments
  9. REFERENCES
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ACR_21954_sm_SupplApp1.doc26KSupplementary Data
ACR_21954_sm_SupplApp2.xlsx16KSupplementary Data
ACR_21954_sm_SupplApp3.doc29KSupplementary Data

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.