Does Smoking Affect Implant Survivorship in Total Hip Arthroplasty?

Abstract

This retrospective study aimed to explore the effects of smoking on hip implant survivorship. The study included 147 patients (165 hips) from 1985 to 1991 who underwent total hip arthroplasty (THA) with a particular uncemented cup, and either a cemented or uncemented femoral component of the same design. Thirty-one patients (34 hips, 21 percent of study group) smoked at the time of surgery. Of 13 components (seven cups, five cemented and one cementless stem) revised for aseptic loosening, eight (8/68, 11.8 percent) were revised in six smokers, and five (5/262, 1.9 percent) were revised in four nonsmokers (p=0.0012). Multivariate covariate analysis revealed a 4.5 times greater risk of implant loosening in smokers (p=0.0662). Based on this preliminary study, further larger studies should be performed to determine the extent that smoking may contribute to THA survivorship.

Dr. Dorr and I have interacted frequently at major national and international meetings and symposia. I have the utmost respect for Dr. Dorr's many contributions to the field of adult reconstructive surgery, his intellectual integrity, his enthusiasm and curiosity, and, finally, his sincere and on-going effort to help his fellow man. I look forward to many more years of continued academic interaction and friendship with Larry.

Bill Capello, M.D.

INTRODUCTION

Nearly 50 million or 23 percent of people in the United States are smokers.¹ Health risks associated with smoking are numerous and well documented. Smokers are at increased risk of heart attack, stroke, respiratory disease and some cancers. In addition, there is a growing body of evidence of deleterious effects of smoking on the musculoskeletal system.² Smoking has been shown to increase the risks of developing osteoporosis, ^3–6 hip fracture,⁷ and exercise-related injuries.^8–9 In addition, smoking has been shown to have a detrimental effect on fracture and wound healing.^10–12 In one study of 202 total hip and knee arthroplasties, patients who smoked cigarettes had longer surgical and anesthetic times and consumed more health resources.¹³ Recently, Møller et al, in a study of 811 consecutive patients who underwent hip or knee arthroplasty, found that smoking was the single most important risk factor for the development of postoperative complications, particularly those relating to wound healing, cardiopulmonary complications, and the requirement of intensive care postoperatively.¹⁴

Cigarette smoking has been shown to interfere with bone metabolism, revascularization and bone formation.^2,15–16 In particular, nicotine decreases bone healing strength as well as revascularization of bone grafts, and slows the production of osteoblasts or the bone-forming cells necessary for fracture healing.^3,17–18 Willert and Buchhorn recently described a three-phase, overlapping healing process in the osseointegration of cemented and noncemented hip implants in which they describe that primary fixation is achieved during implantation, whereas secondary fixation is a result of repair and bone remodeling during the healing process, comparable to fracture healing.¹⁹ Results from one retrieval study implicated vascular injury and subsequent diminished blood supply at the implant-host interface as a reason for insufficient osseointegration in total hip implants.²⁰ Since both cemented and cementless hip implants must be stabilized long-term by a bond between living bone and the prosthesis, the question arises about the effect of smoking on hip implant survivorship over time. Therefore, this study aims at examining the effects of smoking on hip implant survivorship for cemented and cementless implants.

MATERIALS AND METHODS

Patients

One hundred eight-seven consecutive patients (205 hips) meeting our inclusion criteria were culled from the database of a single arthroplasty surgeon (WC) at a large midwestern medical center for participation in this retrospective study. All patients underwent primary total hip arthroplasty between 1985 and 1991 and had a diagnosis of osteoarthritis or avascular necrosis of the hip. In addition, the implants used were limited to the Omnifit® porous-coated acetabular component and the Omnifit® cemented or porous-coated femoral component (Stryker Howmedica Osteonics Corporation, Allendale, NJ). These components were selected because of their reported low rates of aseptic loosening at an average ten years follow-up.^21,22 The component and diagnostic limitations were intended to diminish the effects of extraneous factors on the primary independent variable in this study, that is, the effect of smoking on implant survivorship.

Of the 187 patients (205 hips), 35 patients (35 hips) died and another five patients (five hips) were lost before five-year minimum follow-up, leaving 147 patients (164 hips), or 80 percent of the total group for participation in the study. Statistical comparisons between those who had less than five-year follow-up (40 patients, 40 hips) versus those in the study group who had a minimum five-year follow-up indicated that those in the lost-to-follow-up group were, on average, eight years older at the time of their THA; had a greater percentage of cemented versus cementless stems; and had more postoperative complications. Average length of follow-up in the dropout group was 20 months (range, one to 61 months), and during that time period there were no component revisions and one reoperation, an incision and drainage secondary to sepsis. The percentage of smokers did not differ between the dropout and study groups.

116 patients (131 hips) were categorized as either smokers or non-smokers at the time of THA. 31 patients (34 hips) or 21 percent of the group were smokers. Of the 147 patients (165 hips) with a minimum of five-year follow-up, 39 patients (42 hips) died at an average of 10.6 years after THA (range, 5.2 to 17.3 years). Attempts to contact a living relative who had first-hand knowledge of the smoking habits, as well as reoperation or revision surgery in the deceased patient group, were futile. Of the 108 patients (123 hips) in living patients, a valid follow-up interview was conducted with 93 patients (106 hips) or 86 percent of the living patients. Within the study group, there were no differences in demographic variables between those who completed the telephone interview and those who did not. All patients who participated in the follow-up aspect of this study gave informed consent to participate. Our Institutional Review Board approved this study.

All patients were routinely treated with prophylactic antibiotics prior to and 48 hours after THA, and with 325mg enteric-coated aspirin post-THA for deep vein thrombosis (DVT) prophylaxis.

Statistical Analysis

Descriptive statistics were calculated on all study variables for the total study group. Comparisons between groups (dropouts versus study group, smokers versus nonsmokers, and revision for aseptic loosening versus stable implants) were performed using a student's t-test for continuous variables and the Fisher exact test for nonparametric variables. A p-value of 0.05 was considered statistically significant. Kaplan-Meier survivorship analyses for two samples (smokers versus nonsmokers) were run using two different endpoints, (1) revision for aseptic loosening of one or both components, or (2) revision for aseptic loosening of one or both components or reoperation for osteolysis. Cox's F test was used to determine statistical differences between survivorship endpoints between smokers and nonsmokers. Multivariate Cox regression modeling, using the variables of age and body mass index at the time of surgery, gender, diagnosis, stem fixation, alcohol use, and smoking was performed to determine the estimated hazard risk regarding the endpoint of component revision for aseptic loosening.

RESULTS

Clinical Outcomes

The study group consisted of 147 patients (165 hips) with a minimum of five-year follow-up. Average age at time of the index arthroplasty was 60.5 years (range, 21 to 83). Fifty-four percent of the hip arthroplasties (89 of 165 hips) were done in females, and sixty-one percent (100 of 165) had cemented femoral components. Seventy-nine percent had a diagnosis of osteoarthritis (131 of 165 hips). Of the remaining 27 patients (34 hips) with a diagnosis of avascular necrosis, six patients (nine hips) were post renal transplant, six patients (seven hips) were steroid-related, five patients (seven hips) were alcohol-related, and the remaining 10 patients (10 hips) were post-traumatic, post femoral neck fracture, post radiation necrosis, or of unknown etiology. Average body mass index at the time of arthroplasty was 28 (range, 18 to 55) with 37 hips (22 percent) in patients having a body mass index greater than 30. Seventy-one hips (43 percent) were done in non-drinkers, with the remaining reporting either occasional or regular use of alcohol socially. Breakdown of demographic characteristics by smoking and nonsmoking is seen in Table 1. As a group, the smokers included more males, more patients with avascular necrosis, and more patients with cementless stems. Smokers were also younger and used alcohol more often than did nonsmokers.

TABLE 1. Comparison of Demographic Characteristics Between Smokers And Nonsmokers.

Variable	Nonsmokers	Smokers
Patients/Hips	116/131	31/34
Deceased	34	8
Gender	54 male, 77 female	22 male, 12 female *
Age (years)	62.7	52.2 ***
Diagnosis	110 OA, 21 AVN	21 OA, 13 AVN**
Body Mass Index	28.0	27.9
Stem Type	89 C, 42 CL	11 C, 23 CL***
Alcohol Use	61 none 40 occasional 21 regular	10 none 9 occasional 13 regular *

OA=Osteoarthritis

AVN=Avascular Necrosis

C=Cemented

CL=Cementless (porous-coated)

^*p < 0.05

^**p < 0.01

^***p < 0.001

At the time of the primary THA, 28 of the 31 smokers consumed an average 1.2 packs of cigarettes per day (range, 0.25 to two packs per day) for an average 34 years (range, two to 60 years). Two patients smoked only cigars or pipes, and one patient chewed tobacco. As part of the follow-up interview, patients were asked to verify their smoking status since the time of their THA. No nonsmokers had taken up smoking, and five smokers had quit smoking between one and 14 years after THA.

Patients representing 50 THAs (30 percent) had no comorbid conditions at the time of surgery. The remaining 70% had comorbid conditions. The most frequently reported condition was hypertension (73 hips, 44 percent), followed by heart disease (32 hips, 19 percent), and gastrointestinal problems (20 hips, 12 percent). Eighteen of 31 (42 percent) smokers had one or more comorbid conditions compared to 77 of 116 (33 percent) of nonsmokers (p=0.26). There were no differences in the number or type of comorbid conditions between the two groups.

Average length of hospitalization for the study group was 9.2 days (range, six to 36 days). Seven patients underwent staged bilateral THA between seven and 11 days apart within the same hospitalization, accounting for the majority of the longer hospitalizations. One patient had a hospitalization greater than 18 days. That patient was a nonsmoker who developed an infection postoperatively and underwent incision and drainage 14 days post-THA and subsequently remained hospitalized for intravenous antibiotics. There was no difference in average number of hospital days between smokers and nonsmokers (8.9 and 9.2 days, respectively, p=0.58).

Twenty-four patients experienced a perioperative complication including three intraoperative fractures, one wound infection, one postoperative dislocation, and one femoral nerve injury. Twenty perioperative complications occurred in nonsmokers and four occurred in smokers. Complications in the smokers included two of the intraoperative fractures, the femoral nerve injury, and postoperative atrial fibrillation. A complication occurred in nine of 34 (26 percent) hips in smokers compared to 24 of 131 (18 percent) of nonsmokers (p=0.20).

Radiographic Outcomes

Radiographic review was completed on the most recent films on all non-revised components. Five-year minimum radiographs were unavailable on 15 hips (nine percent); therefore, the average radiographic follow-up was 10.5 years (range, one to 17 years). One hundred forty-nine (90 percent) cups remained stable. One cup was stable fibrous, and three cups were unstable and unrevised to date. On the femoral side, 150 (91 percent) stems were stable, and five cementless stems were stable fibrous. There were no radiographically loose cemented or cementless stems. Two of the five stable fibrous stems were in smokers. The remaining stable fibrous stems and stable fibrous or unstable cups were in nonsmokers.

Survivorship

Thirty-five patients (24 percent) underwent 45 additional procedures on the index THA. Thirteen patients with cementless stems underwent 15 reoperations for acetabular and/or proximal femoral osteolysis involving exchange of the femoral head and polyethylene liner and bone grafting of accessible bone lesions. Average time to reoperation for osteolysis was 127 months (range, 57 to 189). Ten of the 13 patients were male, and all were under the age of 60 at the time of the index procedure. Twenty-three smokers and 42 nonsmokers had cementless stems. Of that cementless group, three smokers and ten nonsmokers under went reoperation for osteolysis with no difference between groups (p=0.30).

There were ten reoperations in seven patients due to sepsis (six incision and drainage, and four component removal). One smoker underwent two incision and drainage procedures and another smoker underwent an incision and drainage and subsequent component removal. The remaining procedures related to sepsis were in nonsmokers with no difference between groups (p=0.59). There were five reoperations secondary to dislocation; three femoral head and polyethylene liner exchanges and two cup revisions. Three patients underwent removal of heterotopic bone due to pain, and one patient had a stem revision secondary to stem fracture. The number of patients in each group (smokers and nonsmokers) who underwent a surgical procedure, either reoperation or revision, are summarized by reason for reoperation or revision procedure in Table 2. Although there was a trend toward a greater number of reoperations for dislocation in the smoking group, there was no difference in the number of patients who suffered a dislocation during the follow-up period. In addition to the five reoperations or revisions due to dislocation, another 11 patients suffered a dislocation treated successfully with closed reduction. In total, five smokers and 11 nonsmokers suffered a dislocation (p=0.23).

TABLE 2. Comparison of Reoperations and Revisions Between Smokers and Nonsmokers.

Reason for Procedure	Smokers (N=31 patients)	Nonsmokers (N=116 patients)	P value
Infection	2	5	0.456
Dislocation	3	2	0.063
Osteolysis	3	10	0.545
Aseptic Loosening	6	4	0.0063**

^**p < 0.01

Thirteen components in ten patients were revised for aseptic loosening (seven cups, five cemented stems, and one cementless stem). One patient had both the cup and stem revised at a single operation. Another bilateral THA patient had both cups revised at a single operation and a third patient had a cup revision and approximately five years later had a stem revision.

The aseptic revision rate for the porous-coated acetabular components was seven of 165 (4.2 percent). The aseptic revision rate for the cemented stems was five of 100 (5 percent) and one of 65 (1.5 percent) for the porous-coated stems. Although there was no difference with regard to revision rate for aseptic loosening of porous coated stems, there was a significant difference between smokers and nonsmokers with regard to cemented stem revisions for aseptic loosening. Two of 11 cemented stems were revised for aseptic loosening in smokers compared to three of 89 cemented stems in nonsmokers (p=0.0335). The aseptic loosening rates for acetabular plus femoral components were eight of 68 components (11.8 percent) in smokers compared to five of 262 components (1.9 percent) in nonsmokers (p=0.0012). Aseptic revision rates by patient, regardless of whether one or both components were revised, was six of 31 smokers (19.4 percent) compared to four of 116 nonsmokers (2.5 percent) (p=0.0063). The component mechanical failure rates (component revision for aseptic loosening plus radiographically loose cup and/ or stem) for each group are seen in Table 3.

TABLE 3. Mechanical Failure Rates Between Smokers and Nonsmokers.

	Smokers	Nonsmokers	p value
Cup Revision for Aseptic Loosening	5	2
Radiographically Loose Cup	0	3
Acetabular MFR	5/34 (14.7%)	5/131 (3.8%)	0.032*
Stem Revision for Aseptic Loosening	2 C, 1 CL	3 C, 0 CL
Radiographically Loose Stem	0	0
Femoral MFR	3/34 (8.8%)	3/131 (2.3%)	0.103
Cup or Stem Revision for Aseptic Loosening	8	5
Radiographically Loose Cup or Stem	0	3
Combined Component MFR	8/68 (11.8%)	8/262 (3.1%)	0.007**

MFR=Mechanical Failure Rate

C=Cement

CL=Cementless (porous-coated)

^*p < 0.05

^**p < 0.01

Kaplan-Meier survivorship with an endpoint of cup and/or stem revision for aseptic loosening was 93.7 percent ± 4.3 percent, 75.3 percent ± 9.1 percent, and 37.6 percent ± 27.0 percent at 10, 15, and 18 years for smokers compared to 96.7 percent ± 1.6 percent at each of the same time intervals for nonsmokers (F=7.0886, p=0.00081) (see Figure 1). Analyses of demographic factors between patients who had a revision for aseptic loosening (n=10) and those who did not (n=137) found no differences in age, body mass, number of comorbidities, alcohol use or stem type, with a statistical trend toward more males (p=0.05) and more diagnoses of avascular necrosis (p=0.09) in the revised group. The only statistically significant variable was smoking (p=0.006).

Figure 1. Kaplan-Meier Survivorship. Endpoint: Component Revision for Aseptic Loosening.

Kaplan-Meier survivorship using the endpoint of component revision for aseptic loosening. Solid line represents survivorship for nonsmokers. Dashed line represents survivorship for smokers.

Results of multivariate Cox regression modeling using the covariates of gender, age and body mass index at the time of the index procedure, diagnosis, stem fixation, alcohol use and smoking found that smokers had a 4.5 times higher risk hazard than did non-smokers with regard to the endpoint of component revision for aseptic loosening (p=0.0662). Male gender yielded an 11.3 times greater risk (p=0.0391), and no other factors approached statistical significance.

DISCUSSION

There have been no prior studies examining the effect of smoking on intermediate- or long-term implant survivorship in THA. We hypothesized that there would be an increased aseptic loosening rate in smokers, regardless of whether the implant was cemented or cementless, and this hypothesis was supported, within the limitations of this study design. Interestingly, there was also no difference between smokers and nonsmokers with regard to reoperation rates for osteolysis in the case of cementless implants.

Although smokers had a higher percentage of implant failure due to aseptic loosening than did non-smokers, a larger proportion of smokers also had other demographic features which have previously been reported as associated with higher failures rates, specifically younger age, male gender and a diagnosis of avascular necrosis of the hip.^31–33 Since those reported relationships are correlational in nature, it is possible that the higher failure rates were incorrectly attributed to those factors if smoking history was not considered. In multivariate analysis, controlling for other covariates, smokers were found to have a 4.5 times greater hazard risk than did non-smokers with regard to the endpoint of component revision for aseptic loosening in this study. The only variable that posed a significantly greater risk was male gender.

This retrospective study is the first to explore the possible deleterious effect of smoking on total hip implant survivorship, and although the results indicated only a statistical trend in multivariate analysis, the percentage differences in failure noted between smokers and nonsmokers are compelling and, we believe, worthy of further study. Until factors contributing to implant failure, including smoking, are studied in greater depth, it is recommended that the arthroplasty surgeon consider smoking status only in conjunction with other accepted risk factors such as age, gender, and bone quality when selecting hip implants for a given patient.

Smoking alone may or may not be a major contributor to implant failure, but in combination with other known or unknown variables, may be worthy of at least discussion between the surgeon and the patient. As an example, although cemented stems failed at a higher rate in smokers than in nonsmokers and there was no difference between groups with respect to revision for aseptic loosening of the porous coated stem in this study, because of the limited number of cases in this study, at this time the authors would not recommend the use of cementless implants over cemented implants based purely on whether the patient is a smoker or not.

Two relatively recent studies^13–14 examined the effect of smoking on the incidence of perioperative complications following total hip or knee arthroplasty with somewhat differing results. Lavernia and his colleagues¹³ found that smokers had more comorbidities, longer surgical times, and higher hospital charges with no difference in the percent of complications, whereas Møller and her colleagues¹⁴ found that smokers had a higher risk of postoperative complications, in particular woundand cardiopulmonary-related ones. In this study, we found no differences in the average number or type of comorbid conditions or in the incidence of postoperative complications.

The effect of smoking has been studied extensively relative to fracture and osteotomy healing and in bony fusions. In an animal model, Lau and colleagues²³ found the incidence of nonunion or delayed union to be twice as great in smokers, and Schmitz and colleagues²⁴ reported that grade I open tibial shaft fractures took 62 percent longer to heal in smokers. Similarly, Chen and colleagues¹⁸ found that smokers had longer healing times and more nonunions after ulna-shortening osteotomy, and Hilibrand et al²⁵ found that smoking had a significant negative impact on healing and clinical recovery following anterior cervical arthrodesis with strutgrafting. It was this growing body of literature on the negative effects of smoking on various orthopaedic procedures that prompted the question of the effect of smoking on hip implant survivorship over time.

The process of osseointegration of cemented and cementless implants was described by Willert and Buchhorn¹⁹ as occurring in three overlapping phases involving the destruction and necrosis of bone followed by repair with integration of the implant into the bone and finally a stabilization and adaptation of the implant to load transfers and other factors. Furthermore, primary fixation is achieved during implantation, whereas secondary fixation is a result of repair and bone remodeling during the healing process, a process comparable to that of fracture healing. Al-Saffar,²⁶ in a histologic study of osseointegration of implants, concluded that a negative balance between the rate of bone growth and resorption around the prosthesis is central to the pathogenesis of aseptic loosening. Yet another histologic study of osseointegration focused on the microvasculature around implants and concluded that vascular injury and decreased blood supply at the host-implant interface may be one reason for insufficient implant osseointegration and loosening.²⁰ These studies may provide theoretical support for the clinical results seen in this study. However, although osteolysis can lead to implant loosening, osteolysis may be more specifically related to particle disease whereas aseptic loosening may be more multifactorial.

In this study, to control as many extraneous variables as possible, we chose to limit the surgeon to a single experienced arthroplasty surgeon; to limit the diagnoses to osteoarthritis and avascular necrosis only; and to limit the implants in the study to those which have documented good outcomes at least ten years post-implantation. In our own published work, the aseptic revision rates for the porous-coated cup and stem were 1.3 and 2.6 percent, respectively, at an average ten-year follow-up, and 2.0 percent for the cemented stem at an average nine-year follow-up.^21–22 In this study, at an average 13-year follow-up and with the population limited to a diagnosis of osteoarthritis or avascular necrosis of the hip, the aseptic loosening rates were 4.2 percent, 1.5 percent, and 5.0 percent for the porous-coated cup, porous-coated and cemented stems, respectively. These failure rates for the porous-coated cup and cemented stem, although higher than those reported for the same components at an average three-to-four year shorter time period, include revision surgeries performed outside of the primary institution and are in concert with other reports of implant loosening at greater than ten-year follow-up.^27–30 In fact, five of the seven porous-coated cups that failed in this study group were revised between ten and 14 years post implantation.

While the overall component aseptic loosening rates are in concert with other implants in the literature, there is a striking difference in the aseptic loosening rates between smokers and nonsmokers in this study. Acetabular aseptic revision rates were 14.7 percent for smokers and 1.5 percent for nonsmokers. On the femoral side, the aseptic revision rates were 8.8 percent for smokers and 2.3 percent for nonsmokers. Differences in the numbers of smokers (n=34 hips) and nonsmokers (n=131 hips) and the relatively overall small sample sizes may magnify the percentage differences, however the proportion of smokers in this study population (31 of 147 patients, 21 percent) is similar to the 23 percent reported by the National Center for Health Statistics in 2000.¹

Studies of the relationship between smoking, drinking, and the development of femoral head necrosis indicate a cumulative effect of alcohol use and an increased risk in current smokers with no cumulative increased risk.³⁴ Because of the small sample number of smokers in this study, no attempt was made to analyze cumulative dose effects of smoking on aseptic loosening. However, anecdotally, of the five patients who quit smoking at various times in the follow-up period, the three who quit at one, two and four years postoperatively had no component failures. One patient who quit smoking nine years postoperatively had a cup revision for aseptic loosening at 17 years postoperative, and one patient who quit smoking at 14 years postoperatively had undergone revision for aseptic loosening of both the cup and stem at 10.5 years postoperative. Because of the time elapsed between implantation and failure, and time of last follow-up, many other factors could have intervened which may or may not have affected our results. However, there is no known reason to suspect that intervening variables differed between the smokers and nonsmokers in this study.

Limitations of this study are the small sample size, the inability to obtain smoking and outside revision data on the deceased patients, and the retrospective nature of obtaining the smoking histories of the patients. Nonetheless, follow-up interviews were obtained on 86 percent of the living patients and no statistical differences were found between those patients who completed the follow-up interview and those who did not due to death or inability to contact. The authors acknowledge the limitations of this study and consider this work a preliminary effort in determining the potential deleterious effect of smoking on total hip arthroplasty outcome. The authors recommend that future studies utilize a prospective design and a much larger number of subjects to allow for the study of multiple contributing variables including both patient and implant factors.

In summary, in this series of 147 patients (165 hips) followed an average 13 years after total hip arthroplasty, there was a significantly greater rate of component revision for aseptic loosening in patients who smoked at the time of THA and who continued to smoke for years afterward. Further study is needed to determine the extent to which smoking contributes to aseptic loosening in relationship to other known contributing factors, and whether there is a smoking dose relationship. For now, the American Academy of Orthopaedic Surgeons Fact Sheet for patients preparing for joint replacement surgery currently states that patients who smoke should try to cut down or quit prior to the arthroplasty as smoking changes blood flow patterns, delays healing, and slows recovery.³⁵