Journal Logo


Saline-Filled Breast Implant Safety and Efficacy: A Multicenter Retrospective Review

Cunningham, Bruce L. M.D., M.S.; Lokeh, Adam M.D.; Gutowski, Karol A. M.D.

Author Information
Plastic and Reconstructive Surgery: May 2000 - Volume 105 - Issue 6 - p 2143-2149
  • Free


Although available since 1965, the saline-filled breast implant was consigned a secondary role in favor of the silicone gel–filled implant for the vast majority of patients undergoing breast augmentation or reconstruction. This changed dramatically in 1992 after the U.S. Food and Drug Administration established a moratorium on silicone gel–filled implants for breast augmentation. Since that time, the saline-filled breast implant has become the prosthesis of choice.

The Food and Drug Administration felt bound to restrict the use of silicone gel–filled implants in large part because of the rapid evolution of breast prosthesis surgery and the paucity of well-controlled, scientific studies of their efficacy and safety. The Food and Drug Administration has since mandated an extensive cooperative pre–market approval review process for the saline-filled implant, partnering with plastic surgeons and implant manufacturers.

In 1996, as part of this pre–market approval process, the senior author (B.L.C.) was commissioned by two U.S. saline implant manufacturers (Mentor Corporation and McGhan Medical Corporation, both in Santa Barbara, Calif.) to perform a multicenter retrospective outcomes study. The goal was to obtain physician-reported and patient-reported data on saline-filled breast implant deflation, capsular contracture, complications, and satisfaction. A pilot study was completed in 1996. 1 It demonstrated low overall deflation (5.5 percent) and complication (4.4 percent) rates and high patient and physician satisfaction.

To extend that pilot study and address several issues raised by an epidemiology panel of the Food and Drug Administration, we were commissioned to conduct a second multicenter retrospective outcomes-based evaluation of the safety and efficacy of saline-filled breast implants. The study design was modified to ensure a 10-year minimum patient follow-up period, to minimize perceived physician selection bias by providing an open invitation to all members of the American Society of Plastic and Reconstructive Surgeons (renamed American Society of Plastic Surgeons in October of 1999) to participate, to strengthen biostatistical support, and to reduce patient selection bias by improving patient tracking for a higher patient participation rate.

Patients and Methods

For the current study, we sent an open letter of invitation to all members of the American Society of Plastic and Reconstructive Surgeons to participate in this multicenter study. We enrolled 11 centers representing geographically diverse plastic surgery practices using a high volume of saline-filled implants. Of these 11 centers, 2 later withdrew and provided no completed data. They did return data culled from physician records but did not conduct patient interviews because of a lack of personnel. The number of patients per participating center was limited to reduce any bias associated with different surgical techniques. Each center was compensated with a research grant of $1500.

Each plastic surgeon was asked to identify and contribute data on 50 patients (range, 11 to 96) who underwent saline-filled breast implant placement from January 1, 1980, through June 30, 1986. A small cohort of 27 patients who underwent placement between January 1, 1977, and December 31, 1980, was included in our analysis. To reduce patient selection bias, each surgeon was asked to contact a consecutive series of patients to complete the outcomes study instrument. Four hundred fifty patients with 882 saline-filled breast implants completed the study.

In our previous study, the inability to successfully locate and contact some patients resulted in gaps in some centers’ consecutive series. To minimize these gaps and to ensure better patient tracking, Equifax (Atlanta, Ga.), a public database company, was employed. Within the selected consecutive series, the names of patients who could not be located were forwarded to Equifax to identify current addresses and telephone numbers. Seven centers were able to report data for 41 to 58 patients of consecutive series of 50 to 90 patients, one center reported data for 96 patients, and one center was able to provide complete data for only 11 patients. Overall, 450 of 878 patients (51.3 percent) identified in consecutive series were contacted and were willing to participate in the study, 37 (4.2 percent) declined to participate, and 391 (44.5 percent) could not be contacted even with the help of Equifax.

Patient enrollment criteria included the use of single-lumen saline-filled breast implants of any manufacturer, in any anatomic location, for any medical indication. Only the first set of saline-filled breast implants, documented as being placed during the study period, was included. Some patients had implants that were placed before the study period, but these were not considered in this analysis. Patients with double-lumen implants and adjustable saline-filled implants with kink-type valves were excluded from this study because of recognized differences in the construction and mechanical properties of these implants.

Personal data were collected from patients by telephone using a standardized outcomes instrument administered by office staff. Patients were read a standardized informed consent statement, invited to participate in the study, and told that they could decline to participate without prejudice. For patients who could be contacted, office staff conducted a medical chart review using a standardized abstraction form. Field audits of the office staff procedures and random data integrity checks were not conducted. The data were compiled on a master sheet that preserved patient confidentiality; this sheet was sent with the completed, numerically coded study forms to the principal investigator (B.L.C.).

The University of Minnesota School of Public Health, through its Divisions of Epidemiology and Biostatistics, helped with study design, data processing, and statistical analysis. Statistical analysis was performed using SAS software. Kaplan-Meier analysis was used to determine actuarial survival statistics at 5 and 10 years for saline-filled breast implants. Odds ratios for risk factors associated with implant deflation and capsular contracture and for variables associated with patient satisfaction were determined by a single stepwise logistic regression analysis. Deflation analysis was performed including and excluding the Surgitek model because of its higher deflation rate. Similarly, odds ratios for implant deflation were calculated including and excluding the Surgitek model.


Mean patient follow-up time was 13.0 years (range, 9.8 to 20.0 years). Most implants (93.9 percent) were placed for augmentation; the remainder were placed for reconstruction. Seventy-four percent were placed in a submammary position; 25.6 percent, subpectorally; and 0.2 percent, subcutaneously. The most commonly used implant manufacturers were Heyer-Schulte (59.7 percent), Mentor (29.2 percent), and Surgitek (8.4 percent) (Table I). The majority of implants (85 percent) had a volume between 190 and 350 cc (mean, 250 cc). Implant filling volume was at the recommended amount in 21.3 percent of implants (of these, 90.9 percent were filled within 25 cc of the recommended volume), 53.7 percent were overfilled (of these, 7.2 percent were overfilled by more than 25 cc), and 25.0 percent were underfilled (of these, 1.9 percent were underfilled by more than 25 cc). Intraluminal steroids were used in 14.2 percent of implants, and intraluminal antibiotics were used in 43.3 percent. Steroids were placed in 0.5 percent of implant pockets, and antibiotics were placed in 24.7 percent of implant pockets.

Table 1
Table 1:
Saline-Filled Breast Implant Types and Deflation Rates

Capsule Formation

By telephone interview, 71 percent of patients rated their reconstructed or augmented breasts as soft and natural; 16 percent, as slightly firm; 8.5 percent, as moderately firm; and 4 percent, as hard. Physician classifications of implants by Baker grade were as follows: I, 79 percent; II, 13 percent; III, 7.2 percent; and IV, 1.2 percent. Closed capsulotomy was performed on 69 patients (105 implants); 49 patients (82 implants) required open capsulotomy. In 20.4 percent of patients (16.6 percent of implants), significant capsule formation occurred (as determined by patient classification of their breast as moderately firm or hard, by a physician-classified Baker grade of III or IV, or by the need for open or closed capsulotomy). Odds ratios for risk factors associated with capsule formation are shown in Table II.

Table 2
Table 2:
Risk Factors for Capsular Contracture


The overall complication rate per implant was 20.2 percent (27.6 percent of patients). Complications per implant and per patient are listed in Table III. The overall patient reoperation rate was 25.8 percent: 12.2 percent for implant replacement or removal, 10.9 percent for open capsulotomy, 0.7 percent for correction of asymmetry, and 2.0 percent for repair of wound complications. The low incidence of wound infection and dehiscence did not allow statistical analysis of the use of intraluminal or implant pocket antibiotics or steroids.

Table 3
Table 3:


Deflation occurred in 73 implants (8.3 percent) according to patient interviews and in 67 implants (7.6 percent) according to physician record reviews. A cohort of patients with implants from a single manufacturer (Surgitek) at only one center accounted for 26 of the 73 total implant deflations (35.6 percent). If this cohort of rarely used implants is excluded from the analysis, the overall patient-reported deflation rate is 5.8 percent. If the 12 implants with an identifiable iatrogenic or traumatic cause of deflation are excluded, the spontaneous deflation rate is 4.3 percent. Of the 67 patients who reported implant deflation, 61 (91 percent) had unilateral deflation, and 6 (9 percent) had bilateral deflation. The total number of deflations reported by implant type is shown in Table I.

Implant actuarial survival is 98.4 percent to 99.8 percent at 5 years and 96.9 percent to 98.9 percent at 10 years (95 percent confidence interval) for the implants when using patient-reported deflation rates, excluding the Surgitek model deflations (Fig. 1). When known, the cause of deflation was attributed to trauma in 2 patients and to mammographic examination in 10 patients. No patients reported closed capsulotomy as the cause of deflation. If the 12 implants with known iatrogenic or traumatic deflations are excluded, implant actuarial survival is 98.8 percent to 99.9 percent at 5 years and 97.9 percent to 99.5 percent at 10 years (95 percent confidence interval) (Fig. 2).

Fig. 1
Fig. 1:
Actuarial survival of all implants except Surgitek model.
Fig. 2
Fig. 2:
Actuarial survival of all implants with spontaneous deflation except Surgitek model.

Seven patients described their implant deflation as painful or uncomfortable. The duration of deflation was less than 24 hours for 37 patients, 1 to 6 days for 18 patients, and 1 to 8 weeks for 5 patients (7 patients did not respond). Most patients (79.6 percent) recalled being informed that their implants could deflate, but only 40 (10.2 percent) reported that this was a significant concern for them.

After deflation, 86.4 percent of patients contacted the original surgeon, 12.1 percent contacted another surgeon or health-care professional, and 1.5 percent did not contact anyone. Contact was made within 1 day of deflation by 75 percent of patients. Of the 67 patients with deflated implants, 58 had their implants replaced, 2 had them removed, and 7 elected to do nothing. Of the 75 patients who underwent implant replacement (for any reason), 30 were done within 1 week and 45 within 1 month of implant deflation.

Replacement costs, according to physicians’ records, were absorbed by the manufacturer for 15 of the 50 patients (30 percent) for whom cost data were available. The surgeon reduced or waived replacement costs for 10 patients (20 percent), and 25 patients (50 percent) were responsible for the entire replacement charge. However, according to the 53 patients who responded to this question, replacement expenses were absorbed by the manufacturer for 9 patients (17 percent), by the surgeon for 14 patients (26 percent), and by the patient in 30 cases (56 percent).

Patient Satisfaction

Overall satisfaction with saline-filled breast implants was high, with 92.9 percent of patients reporting that they were “very satisfied” or “satisfied”; 95.9 percent of all patients were willing to choose them again. Most patients (68 percent) would choose the same size implant if reoperation became necessary, but 20 percent would choose a larger implant, and 12 percent would choose a smaller implant. Patients who desired a larger size had, on average, a 243-cc implant; those who wanted a smaller implant had, on average, a 254-cc implant (analysis of variance, p = 0.28). Of the patients who described their breasts as firm or hard, 26.9 percent did not find them bothersome; 20.0 percent found them not very bothersome; 36.9 percent, somewhat bothersome, and 16.3 percent, very bothersome. Nineteen of 55 dissatisfied patients considered implant removal. Of the 54 patients who had previous silicone-filled implants, 10 preferred their silicone-filled implants to their current saline-filled breast implants. Risk factors for patient dissatisfaction are shown in Table IV.

Table 4
Table 4:
Risk Factors for Patient Dissatisfaction


Deflation rates reported for saline-filled breast implants in the literature have varied widely, ranging from 2 percent to 76 percent. 2,3 Despite large sample sizes, the conclusions that can be drawn from most of these studies are ultimately limited. These studies have often represented only a single clinical practice experience, multiple manufacturing generations, and/or significantly varying clinical settings. Additionally, given the historical lack of implant and patient registries, a migratory patient population, and the inherent difficulties in patient tracking, long-term follow-up has been sporadic at best.

Our previous study, which represented a single manufacturing era and multiple clinical practices, demonstrated an overall deflation rate of 5.5 percent with an average follow-up period of 6 years. 1 One encouraging finding is that despite doubling our average length of follow-up in the current study to 13.0 years, we found a similar deflation rate of 5.8 percent, excluding the Surgitek cohort. When patients with a known iatrogenic or traumatic cause of deflation are excluded, the spontaneous deflation rate decreases to 4.3 percent. More impressively, the 10-year actuarial implant survival rate is 96.9 percent to 98.9 percent (95 percent confidence interval), compared with our earlier findings of 90.2 percent to 95.2 percent (95 percent confidence interval). 1

Of particular interest is our observation of an alarmingly high implant deflation rate (35.1 percent) of the Surgitek saline-filled implant, a device that was not a factor in our previous study. The 74 Surgitek implants evaluated in this study were used in 41 patients at a single clinical practice. Despite our review of the data from this cohort and our investigation of this now-defunct implant model, its frequent failure remains unexplained. No representative of the manufacturer was available to respond to our queries.

However, despite this high failure rate, a review of patients with intact Surgitek implants demonstrated complication and capsular contraction rates statistically consistent with our overall results and those published in the literature. This example illustrates the difficulty in drawing broad conclusions about saline-filled breast implants on the basis of a single center’s experience.

Implant type was identified as the most significant factor for predicting deflation, with the Surgitek implant having a 17-fold greater risk of spontaneous implant deflation compared with other implant models. Confirming results of previous studies, 1,3–5 we found the Heyer-Schulte and Mentor model 1800 had a three-fold higher risk of deflation compared with other saline-filled implants. Additionally, implant size greater than 450 cc was a modest risk factor for deflation.

Despite data published in our previous report, we were unable to demonstrate in our current study a statistically significant effect on deflation caused by use of intraluminal or extraluminal steroids or antibiotics. However, our current study did confirm our previous finding (and that of another center 6) that submammary or subpectoral implant position bears no impact on spontaneous implant failure. We were unable to show a protective effect from modest overfilling of the implant, a procedure previously shown to reduce the risk of spontaneous deflation. 3,7 Significant underfilling of the implants, which, according to our previous study, increases the risk of deflation, 1 was not a statistically significant factor in our current population.

Our latest results may represent a lack of statistical power sufficient to demonstrate minor influences on implant deflation. They may also reflect greater variability in clinical practices, resulting from the broader based enrollment of centers (in contrast to our previous study, in which the senior author recruited participants). Risk factors for implant deflation are shown in Table V.

Table 5
Table 5:
Risk Factors for Implant Deflation

In our previous study, questions about patient selection arose. To minimize selection bias in our current study, participating surgeons were asked to identify a consecutive series of patients beginning on a specified date (January 1, 1980). The 12.1 percent of patients seeking treatment from a different surgeon or health-care professional reflects considerable patient mobility, which may account for the differences between physician-reported and patient-reported deflation rates. The only patient exclusion criterion was use of double-lumen implants or adjustable implants with kink-type valves, which were not allowed because of their recognized mechanical differences. By using the Equifax database, patient enrollment and completion of the interview process increased from 41.5 percent in our previous study to 51.3 percent in this study. Of consecutively identified patients, 26 percent to 82 percent were successfully enrolled by the nine participating centers. Only 37 contacted patients (4.2 percent) were unwilling to participate in the study. Thus, our ability to enroll only 51.3 percent of the patients identified by the centers represents the inherent difficulty of tracking and locating patients, not a significant unwillingness of those contacted to participate.

Our current study’s overall complication and capsular contracture rates reinforce previous findings. 1,8 Despite a 27.8 percent reoperation rate, a full 93 percent of patients were satisfied or very satisfied with the outcome of their breast surgery. Given this high patient satisfaction rate, coupled with the low morbidity and device failure rates, we believe that saline-filled breast implants are safe and effective.


We thank Bruce Lindgren, Ph.D., and Patricia A. Brothen for assisting in data tabulation and statistical analysis. We also thank Mary E. Knatterud, Ph.D., for her editorial contribution.


1. Gutowski, K. A., Mesna, G. T., and Cunningham, B. L. Saline-filled breast implants: A Plastic Surgery Educational Foundation multi-center outcomes study. Plast. Reconstr. Surg. 100:1019, 1997.
2. Lavine, D. M. Saline inflatable prostheses: 14 years’ experience. Aesthetic Plast. Surg. 17:325, 1993.
3. Mladick, R. A. “No-touch” submuscular saline breast augmentation technique. Aesthetic Plast. Surg. 17:183, 1993.
4. Schuster, D. I., and Lavine, D. M. Nine-year experience with subpectoral breast reconstruction after subcutaneous mastectomy in 98 patients utilizing saline-inflatable prostheses. Ann. Plast. Surg. 21:444, 1988.
5. Zollman, W., Chavis, D. D., Wentland, M. J., and Lai, K. D.Breast augmentation mammoplasty using a saline-filled prosthesis: An eighteen-year survey. Pre-publication communication, 1996.
6. Worseg, A., Kuzbari, R., Tairych, G., et al. Long term results of inflatable mammary implants. Br. J. Plast. Surg. 48:183, 1995.
7. Mladick, R. A. Inflatable breast implants. Plast. Reconstr. Surg. 95:600, 1995.
8. McKinney, P., and Tresley, G. Long-term comparison of patients with gel and saline mammary implants. Plast. Reconstr. Surg. 72:27, 1983.
©2000American Society of Plastic Surgeons