This article was downloaded by: [University of Exeter] On: 11 August 2015, At: 12:06 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place, London, SW1P 1WG
Current Medical Research and Opinion Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/icmo20
Prediction model for penile prosthesis implantation for erectile dysfunction management a
b
bc
Robert L. Segal , Stephen B. Camper , Larry Ma
& Arthur L. Burnett
a
a
The James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions Baltimore, MDUSA b
Health Economics & Outcomes Research (HEOR), Endo Pharmaceuticals Inc. Malvern, PAUSA
c
Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania Perelman School of Medicine Philadelphia, PAUSA Published online: 26 May 2015.
Click for updates To cite this article: Robert L. Segal, Stephen B. Camper, Larry Ma & Arthur L. Burnett (2014) Prediction model for penile prosthesis implantation for erectile dysfunction management, Current Medical Research and Opinion, 30:10, 2131-2137 To link to this article: http://dx.doi.org/10.1185/03007995.2014.936188
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Current Medical Research & Opinion 0300-7995 doi:10.1185/03007995.2014.936188
Vol. 30, No. 10, 2014, 2131–2137
Article ST-0125/936188 All rights reserved: reproduction in whole or part not permitted
Downloaded by [University of Exeter] at 12:06 11 August 2015
Original article Prediction model for penile prosthesis implantation for erectile dysfunction management
Robert L. Segal The James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions, Baltimore, MD, USA
Stephen B. Camper Health Economics & Outcomes Research (HEOR), Endo Pharmaceuticals Inc., Malvern, PA, USA
Abstract Objective: Penile prosthesis surgery is indicated based on undesirability, contraindication or ineffectiveness of nonsurgical options for erectile dysfunction. This definitive treatment is often delayed after initial diagnosis. Our objective was to develop a prediction tool based on a patient’s clinical history to determine likelihood of ultimately receiving a penile prosthesis.
Larry Ma Health Economics & Outcomes Research (HEOR), Endo Pharmaceuticals Inc., Malvern, PA, USA Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA
Research design and methods: This retrospective analysis used claims data from Commercial and Medicare supplemental databases. Inclusion criteria were 18 years of age with 1 year of continuous enrollment at the first diagnosis of erectile dysfunction. Patients’ demographics, co-morbidities and erectile dysfunction therapy were derived based on enrollment, medical and prescription histories.
Arthur L. Burnett The James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions, Baltimore, MD, USA Address for correspondence: Arthur L. Burnett MD MBA, 600 North Wolfe Street, Marburg 407, Baltimore, MD 21287, USA. Tel.: +1 410 614 3986; Fax: +1 410 614 3695; [email protected] Keywords: Epidemiology – Health Services – Medical co-morbidities – Risk factors Accepted: 24 May 2014; published online: 1 July 2014 Citation: Curr Med Res Opin 2014; 30:2131–7
Main outcome measures: The Cox proportional hazards model with stepwise selection was used to identify and quantify (using relative risk) factors associated with a future penile prosthesis implant. Co-morbidities and therapies present prior to the index erectile dysfunction diagnosis were analyzed as fixed covariates. Results: Approximately 1% of the dataset’s population (N ¼ 310,303 Commercial, N ¼ 74,315 Medicare, respectively) underwent penile prosthesis implantation during the study period (3928 patients in the overall population: 2405 patients [0.78%] in the Commercial and 1523 patients [2.05%] in the Medicare population). Factors with the greatest predictive strength of penile prosthesis implantation included prostate cancer diagnosis (relative risk: 3.93, 2.29; 95% CI, 3.57–4.34, 2.03–2.6), diabetes mellitus (2.31, 1.23; 2.12–2.52, 1.1–1.37) and previous treatment with first-line therapy (1.39, 1.33; 1.28–1.5, 1.2–1.47) (all P50.01). Conclusion: The presence and extent of specific medical history factors at the time of erectile dysfunction diagnosis predict an individual’s future likelihood of penile prosthesis. Calculating the likelihood of penile prosthesis implantation based on the weight of these factors may assist clinicians with the definition of a care plan and patient counseling. The precision of the model may be limited by factors beyond medical history information that possibly influence the decision to proceed to surgery.
Introduction Erectile dysfunction (ED) is a common problem, with multiple underlying risk factors1–3, which can severely impact the quality of life for the patient as well as his partner. It has been estimated that as the population ages the worldwide ! 2014 Informa UK Ltd www.cmrojournal.com
Prediction model for penile prosthesis implantation Segal et al.
2131
Downloaded by [University of Exeter] at 12:06 11 August 2015
Current Medical Research & Opinion Volume 30, Number 10
October 2014
prevalence of ED will double, to over 300 million men by 20254. For patients experiencing ED, first-line therapy typically includes one or more trials with a phosphodiesterase-5 inhibitor (PDE5i). These agents result in successful penetrative intercourse in the order of 80%5–9. For those in whom oral pharmacotherapy is ineffective or contraindicated, other options include intracavernosal penile injections (ICI), urethral suppositories (MUSE) and/or vacuum erection devices (VEDs). Unfortunately, these modalities are associated with treatment dissatisfaction10–16 and substantial dropout rates17. Implantation of penile prostheses (PP), most commonly the three-piece inflatable device, typically represents the final option for individuals who are dissatisfied with or unresponsive to the above therapies18. These devices are associated with a high degree of patient satisfaction and low rates of adverse events19–21. The time period between onset of ED symptoms and PP implantation (PPI) is variable, and may cover an extended time period22,23. During this time, the patient may experience diminished quality of life, including role limitation due to physical and emotional problems, negative perceptions of health, and relationship hardship. Given the potential of a prolonged negative impact of either untreated or unsuccessfully treated ED on quality of life and the high degree of patient satisfaction reported following PPI, it was of interest to explore strategies that may assist physicians in reducing the time period over which an individual experiences ED. We hypothesized that predicting the likelihood that a patient ultimately receives a PP may qualify as one such strategy. The objective of this study was to determine the capacity of factors and/or combinations of factors identified in an ED patient’s medical history to predict future receipt of a PP.
Patients and methods Data source This was a retrospective cohort study using healthcare claims derived from the MarketScan Commercial and Medicare Supplemental Databases (Truven Health, Ann Arbor, MI, USA), which represents the healthcare experiences of approximately 110 million lives from 2000 to 2010. The sources for the claims are employers, health plans, and Medicare programs. Data include enrollment, medical and prescription records integrated from all care providers. The Commercial database is representative of the commercially insured population in the United States, while the Medicare Supplemental database is representative of Medicare beneficiaries with supplemental coverage, which is roughly 30% of the Medicare population. 2132
Prediction model for penile prosthesis implantation Segal et al.
Patient selection Patients with a claim with a diagnosis of ED (ICD-9-CM code 607.84: Impotence of organic origin) were first identified. The first diagnosis was treated as the index date. Patients had to be 418 years old and required to have 12 months’ continuous enrollment prior to the index ED diagnosis date to qualify as a newly diagnosed patient. Patients were excluded if they did not have both medical and prescription coverage for the entire study period, or had PP at or before the index date.
Patient characteristics Patient demographics such as age, gender and census region were derived based on their enrollment records, and co-morbidities and ED treatments were derived based on medical and prescription histories. The list of co-morbidities, medical/surgical procedures and medication prescriptions (collectively termed ‘Factors’) was constructed based on review of the ED literature. ICD-9 and Current Procedure Terminology (CPT) codes were adapted from the Centers for Medicare and Medicaid (CMS) Chronic Care Warehouse24. Non-surgical ED treatments were classified as first- and second-line: first-line treatments included PDE5i and testosterone replacement therapy, and second-line treatments included ICI, MUSE and VED.
Statistical analysis The cumulative percentage of patients who had received a PP after ED diagnosis was estimated based on the productlimit method, and the Kaplan–Meier curves of time-toprosthesis up to 6 years (72 months) were plotted. The Cox proportional hazards model was used for the analysis of time from the index ED diagnosis to PPI. Patients without PP were censored at the end of their enrollment or end of the analysis period. The model included patient demographics, co-morbidities and ED treatments existent prior to or present at the index ED diagnosis, and they were analyzed as fixed covariates. Stepwise selection was used to identify and quantify (using relative risk) factors associated with future PPI; the criterion for entrance into the model was P50.20 and the criterion for retention in the model was P50.10. Though the emphasis of the study was the estimation of the relative risk, the baseline probability of receiving a PP for an ED patient of a certain age (60 years for Commercial and 70 years for Medicare) without any co-morbidity was also estimated. This was necessary to provide context so as to provide a quantitative estimate of overall probability of requiring a PP based on clinical status. The baseline probability of receiving a PP was calculated from the Cox model via the Nelson–Aalen empirical cumulative www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion
hazard function. Probabilities (absolute risks) of PPI for patients can be modeled based on their age, their respective co-morbidities, and the baseline probability at different time points after being diagnosed with ED.
Results
October 2014
therapy. In general, similar profiles were observed when the overall population was segmented by type of insurance, although the frequencies were typically higher in the Medicare group, most likely reflective of the older age of this cohort.
Factor relative risks
Factor frequency at the time of initial ED diagnosis
Downloaded by [University of Exeter] at 12:06 11 August 2015
Volume 30, Number 10
The mean age at index ED diagnosis was 51.2 years and 71.8 years for the Commercial and Medicare populations, respectively. The average length of continuous enrollment was 5.9 and 7.0 years. The frequencies with which pre-identified co-morbid conditions, use of ED therapies and medical histories were observed at the time of initial ED diagnosis in men who eventually received a PP are presented in Table 1. Conditions with documented impact on the coronary, cerebral and/or peripheral vasculature were common (Table 1). In addition, 39% of the overall population had already received a prescription for first-line ED
A total of 3928 patients (1.02%) in the overall population ultimately received a PP, comprising 2405 (0.78%) in the Commercial and 1523 (2.05%) in the Medicare population. Adjusting for censoring, 1.61% of patients were estimated to have received a PP at 6 years following a diagnosis of ED; the percentages were 1.32% and 2.73% for Commercial and Medicare, respectively (Figure 1). The baseline probabilities of receiving a PP for patients with ED only are presented in Table 2. The relative risks (RRs) for the Commercial and Medicare groups were similar (Table 3). The factor with the highest relative risk when present at the time of ED diagnosis was Peyronie’s disease (RR ¼ 4.39). History of prostate cancer (RR ¼ 2.98), priapism (RR ¼ 2.66), and
Table 1. Patient baseline characteristics, co-morbidities and medical history. Overall Population n ¼ 384,618 Age at ED Diagnosis, Mean (SD) Age Category 18–44 45–54 55–64 65–74 75þ Census Region Northeast Region North Central Region South Region West Region Years of Continuous Enrollment, Mean (SD) Co-morbidities and Medical History Cardiovascular Disease Diabetes Mellitus Hypertension Hyperlipidemia Anxiety and related disorders Depression Psychosexual Dysfunction Low Testosterone Peyronie’s Disease Priapism Prostate Cancer Multiple Sclerosis Polyneuropathy Spinal Cord Injury Radical Prostatectomy Renal Transplant Arterial Bypass Surgery Primary ED Therapy Secondary ED Therapy
55.2 (12.0)
Commercial Population n ¼ 310,303
Medicare Population n ¼ 74,315
51.2 (9.4)
71.8 (6.0)
69,073 (18%) 105,997 (28%) 137,739 (36%) 49,245 (13%) 22,564 (6%)
69,059 (22%) 105,601 (34%) 135,513 (44%) 118 (0%) 12 (0%)
14 (0%) 396 (1%) 2226 (3%) 49,127 (66%) 22,552 (30%)
34,868 (9%) 93,504 (24%) 175,270 (46%) 79,677 (21%) 6.1 (3.0)
29,211 (9%) 71,656 (23%) 147,952 (48%) 60,304 (19%) 5.9 (3.0)
5657 (8%) 21,848 (29%) 27,318 (37%) 19,373 (26%) 7.0 (2.9)
30,013 (7.8%) 80,700 (21.0%) 192,354 (50.0%) 202,156 (52.6%) 26,283 (6.8%) 39,388 (10.2%) 21,595 (5.6%) 27,595 (7.2%) 3194 (0.8%) 656 (0.2%) 36,017 (9.4%) 1292 (0.3%) 11,328 (2.9%) 486 (0.1%) 15,563 (4.0%) 813 (0.2%) 55,271 (14.4%) 148,826 (38.7%) 25,008 (6.5%)
15,785 (5.1%) 58,619 (18.9%) 142,303 (45.9%) 160,252 (51.6%) 23,237 (7.5%) 33,978 (10.9%) 18,807 (6.1%) 23,278 (7.5%) 2763 (0.9%) 576 (0.2%) 21,341 (6.9%) 1132 (0.4%) 7345 (2.4%) 371 (0.1%) 11,827 (3.8%) 637 (0.2%) 30,667 (9.9%) 113,848 (36.7%) 11,786 (3.8%)
14,228 (19.1%) 22,081 (29.7%) 50,051 (67.3%) 41,904 (56.4%) 3046 (4.1%) 5410 (7.3%) 2788 (3.8%) 4317 (5.8%) 431 (0.6%) 80 (0.1%) 14,676 (19.7%) 160 (0.2%) 3983 (5.4%) 115 (0.2%) 3736 (5.0%) 176 (0.2%) 24,594 (33.1%) 34,978 (47.1%) 13,222 (17.8%)
ED, erectile dysfunction
! 2014 Informa UK Ltd www.cmrojournal.com
Prediction model for penile prosthesis implantation Segal et al.
2133
Current Medical Research & Opinion Volume 30, Number 10
October 2014
Cumulative Percentage (%)
4
3
2
Downloaded by [University of Exeter] at 12:06 11 August 2015
1
Population:
Commercial (n=310303)
Medicare (n=74315)
0 0
6
12
18
24
30
36
42
48
54
60
66
72
Months Since ED Diagnosis
Figure 1. Kaplan–Meier estimates of the cumulative percentage of patients with penile prosthesis, all erectile dysfunction (ED) diagnosed patients, by population.
Table 2. Baseline probability of penile prosthesis implantation for patients with ED only.
60 Years Old (Commercial) 70 Years Old (Medicare)
Year 1
Year 2
Year 3
Year 4
Year 5
Year 6
Year 7
Year 8
0.28% 0.99%
0.39% 1.22%
0.46% 1.35%
0.53% 1.50%
0.60% 1.59%
0.67% 1.69%
0.75% 1.74%
0.79% 1.79%
The probabilities shown are from the time of the index diagnosis of ED. ED, erectile dysfunction
receipt of second-line ED therapies (RR ¼ 2.07) were also associated with an increased likelihood. In the Commercial population, patents in the 55–64 years group had the highest likelihood of getting PPI, compared with either younger patients (18–44 years, RR ¼ 0.35; 45– 54 years, RR ¼ 0.76) or older patients (65–74 years, RR ¼ 0.86). In the Medicare population, where most patients are older than 65 years, the RR declines with advancing age, which may reflect a host of ED treatment risk profile factors influencing a decreased utilization of PP surgery in this age category.
Prediction calculation The RRs for each factor contributing to the likelihood of future PPI are multiplicative, and may be employed to predict the likelihood of ultimately undergoing PPI. For example, a 60 year old patient in the Commercial population presenting at ED diagnosis with a history of prostate cancer (RR ¼ 3.93) and diabetes mellitus (DM) (RR ¼ 2.31) and who had already received first-line 2134
Prediction model for penile prosthesis implantation Segal et al.
therapy (RR ¼ 1.39) is 12.6 (3.93 2.31 1.39) times as likely to receive a PP as an ED patient of the same age without these factors. Given that a 60 year old ED patient without any co-morbidities has a baseline probability of receiving a PP of approximately 0.67% at 6 years post ED diagnosis, the patient with the above profile will have an absolute risk of PPI of 8.4% (12.6 0.67%). By contrast, in the Medicare population example, a 70 year old patient with the same co-morbidities would have a 3.7 (2.29 1.23 1.33) times chance of receiving a PP compared to the equivalent ED patient without these factors. Similar calculations may be performed for any combination of co-morbidities.
Discussion The aim of this study was to determine whether a clinically useful predictive model for eventual PPI can be devised based on a patient’s clinical variables. The potential utility of such a tool is to provide the clinician and patient with a sense of magnitude of the risk of PPI for maintaining www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion
Volume 30, Number 10
October 2014
Table 3. Relative risk associated with selected demographic, co-morbidity and medical history variables. Overall Population (n ¼ 384,618)
Downloaded by [University of Exeter] at 12:06 11 August 2015
Variable
Peyronie’s Disease Prostate Cancer Priapism Other Unspecified Disorders of the Penis Second-Line ED Therapy Multiple Sclerosis Spinal Cord Injury Diabetes Mellitus Renal Transplant Arterial Bypass Procedure First-Line ED Therapy Psychosexual Dysfunction, Unspecified Sexual Dysfunction Radical Prostatectomy Depression Polyneuropathy Cardiovascular Disease Hypertension Anxiety and Anxiety-related Diagnoses Dyslipidemia Age Group 18–44 45–54 55–64 65–74 75þ
Commercial (n ¼ 310,303)
Medicare (n ¼ 74,315)
Relative Risk (95% CI)
p Value
Relative Risk (95% CI)
p Value
Relative Risk (95% CI)
p Value
4.39 (3.71, 5.19) 2.98 (2.72, 3.26) 2.66 (1.87, 3.79) 2.27 (1.88, 2.73)
50.001 50.001 50.001 50.001
4.47 (3.68, 5.42) 3.93 (3.57, 4.34) 2.14 (1.41, 3.25) 2.26 (1.82, 2.80)
50.001 50.001 50.001 50.001
3.95 (2.79, 5.59) 2.29 (2.03, 2.60) 4.02 (2.08, 7.75) 2.04 (1.41, 2.95)
50.001 50.001 50.001 50.001
2.07 (1.91, 2.24) 1.95 (1.29, 2.94) 1.87 (1.01, 3.49) 1.84 (1.71, 1.97) 1.69 (1.09, 2.62) 1.52 (1.41, 1.64) 1.42 (1.34, 1.52) 1.41 (1.24, 1.60)
50.001 50.001 0.04 50.001 0.02 50.001 50.001 50.001
3.01 (2.72, 3.34) 1.74 (1.04, 2.89) 2.13 (0.95, 4.74) 2.31 (2.12, 2.52) 1.69 (1.01, 2.82) 1.48 (1.33, 1.65) 1.39 (1.28, 1.50) 1.46 (1.26, 1.69)
50.001 0.03 0.06 50.001 0.04 50.001 50.001 50.001
1.38 (1.23, 1.55) 1.89 (0.94, 3.81)
50.001 0.07
1.23 (1.10, 1.37)
50.001
1.51 (1.35, 1.67) 1.33 (1.20, 1.47) 1.32 (1.04, 1.67)
50.001 50.001 0.02
1.40 (1.25, 1.56) 1.33 (1.20, 1.48) 1.26 (1.09, 1.45) 1.16 (1.05, 1.27) 1.12 (1.05, 1.20) 0.82 (0.70, 0.95)
50.001 50.001 50.001 50.001 50.001 50.001
1.27 (1.12, 1.43) 1.24 (1.03, 1.51) 1.21 (1.06, 1.39) 1.17 (1.07, 1.28)
50.001 0.02 50.001 50.001
1.55 (1.31, 1.84) 1.26 (1.06, 1.51) 1.24 (1.01, 1.52)
50.001 0.01 0.04
0.74 (0.56, 0.99)
0.04
0.72 (0.68, 0.77)
50.001
0.76 (0.69, 0.82)
50.001
0.68 (0.61, 0.75)
50.001
0.28 (0.23, 0.33) 0.67 (0.61, 0.73) Reference 1.19*** (1.10, 1.29) 0.66*** (0.58, 0.75)
50.001 50.001
0.35 (0.29, 0.42) 0.76 (0.69, 0.83) 0.86 (0.12, 6.12)
2.41** (2.02, 2.86) 2.41** (2.02, 2.86) 2.41** (2.02, 2.86) Reference 0.56 (0.49, 0.64)
50.001 50.001 50.001
50.001 50.001
50.001 50.001 Reference 0.88 0.93
50.001
Stepwise regression was used to select the predictors in the final model. The statistical significance of these variables had to be50.2 to be entered, and50.1 to be retained in the multivariate regression. **Age groups 18–44, 45–54 and 55–64 were combined for the Medicare population analysis. ***This reflects a combined effect of advanced age and Medicare insurance as almost all Medicare members are over 65 and almost all Commercial members are under 65. ED, erectile dysfunction
sexual function. Specifically, our focus was to quantitate and assign weight to co-morbid conditions present at the time of ED diagnosis. We demonstrated that each co-morbidity carries a definable relative probability for eventual PPI. We postulate that these risks are multiplicative, and calculating the combined risk will aid decisionmaking regarding whether to pursue PPI and perhaps the timing of such a decision. A significant time delay often exists between the onset of ED symptoms and both presenting to a clinician to discuss the symptoms and instituting treatment. Salonia et al.22 reported on 619 patients seeking medical help for new-onset ED in the PDE5i era, where the mean delay prior to presentation was 30.2 months (median 12.0; range 5–300). This delay seems surprising given the likely waning of the taboo associated with openly discussing ED, as well as the explosion of direct-to-consumer advertising for ED treatments in both electronic and print media25. In another study comprising a cohort of men who underwent PPI, the mean duration of ED prior to surgery was 60 months23. All of these patients ! 2014 Informa UK Ltd www.cmrojournal.com
had previously tried at least one other ED treatment, which conceivably proved unsuccessful enough to warrant PP surgery. It is difficult to determine the exact negative impact that unsuccessful ED therapy has. Beyond the possible treatment complications for each of the therapies26 and financial cost issues27, the opportunity cost of ED on other aspects of life (such as other health problems [i.e., depression, relationship discord] and time missed from work) must be considered. Indeed, in a survey of men who underwent ED therapy post-radical prostatectomy, those who discontinued treatment cited ‘treatment effect below expectation’ and ‘loss of interest in sex’ in 84% and 16% of cases, respectively17. Could this cited loss of interest be the result of this lack of treatment success? PPI typically represents the last resort for ED treatment, when other therapies are ineffective or if the patient is disinclined to pursue them. Multiple factors, including the invasiveness and irreversibility of PPI, or clinician unfamiliarity, may account for this neglect. On this Prediction model for penile prosthesis implantation Segal et al.
2135
Downloaded by [University of Exeter] at 12:06 11 August 2015
Current Medical Research & Opinion Volume 30, Number 10
October 2014
basis, the clinician may fail to mention PPI as a treatment option when counseling a patient for ED. This is in spite of the myriad reports which extol the merits, associated satisfaction and safety of prosthetic devices18–21. The implications of deferring PPI are not fully appreciated. Certain treatments, such as ICI, may induce corporal scarring28 and make subsequent PPI more difficult. Additionally, as the patient ages, co-morbid conditions may make elective surgical intervention less advisable, or preclude surgery altogether. Finally, although age alone should not be the main determinant in deciding for or against PPI29, in our experience older patients take a longer time to recover from surgery and display greater difficulty in learning to operate their devices. As such, delaying PPI may present increasingly challenging management dilemmas. Our research advantages the patient who may elect to forego other ED treatment options and proceed directly to PPI upon understanding that the overall likelihood of ultimately requiring a PP is substantial. This knowledge enables him to avoid possible drawbacks of other ED therapies and derive the benefits of the PP immediately. Notably, this approach does not contradict any ED management guidelines, as no urological society mandates a trial of non-surgical ED therapies prior to proceeding with PPI. This decision may facilitate earlier restoration of erectile function with a treatment option that affords high satisfaction22. The advantages of this study include the large population analyzed and the diversity of the sample. In particular, this methodological approach obviates concerns that an individual clinical practice uniquely determines research outcomes. We do acknowledge possible limitations of this study. The relative likelihood of ultimate PPI depends on the baseline prevalence of surgery for that population, and our ‘absolute’ calculation is based on a relatively small proportion of the sampled population which underwent PPI. This same effect may explain the lower RR for the oldest age category, based on the relatively less frequent rate of PPI performed in our population. A patient’s decision to proceed with PPI may not be solely based on his co-morbidity profile, which limits the precision of this model. Other factors, such as insurance coverage status, relationship issues, concerns about undergoing surgery and cultural mores may influence the decision to proceed with PPI independent of co-morbid status. The factor of Commercial or Medicare insurance was considered in our model but was not included in the final model because it was confounded by age group (i.e., almost all Medicare members are 465). It may be assumed that this tool is most useful for a patient motivated to maintain sexual function whereby the surgery is reimbursed by his insurance provider. We were unable to assess certain factors, such as duration and severity of the analyzed co-morbidities, or other demographic characteristics such as marital 2136
Prediction model for penile prosthesis implantation Segal et al.
status, due to the unavailability of such information in a claims database. Finally, though a patient may be at greater risk for ultimately requiring PPI, this study does not address the likelihood of success with other specific ED treatment modalities. Another issue to discuss is the utility of the overall patient population. Although the Commercial and Medicare samples in our study were representative of their respective populations, the simple pooling of these may not represent a typical American patient diagnosed with ED. However, we elected to present the data from the overall population to give readers the opportunity to assess the robustness of the relative risks associated with various co-morbidities (Table 3). To predict the probability of PPI for American patients diagnosed with ED, baseline probabilities (Table 2) should be used in conjunction with the relative risks for the respective Commercial or Medicare population. We acknowledge that the data is derived from a purely American population source. We feel, however, that because 480% of penile prostheses worldwide are implanted in the United States30, the data presented (particularly from the overall population) can be assumed to apply to ED patients worldwide.
Conclusion The presence of specific medical history factors at the time of ED diagnosis predicts an individual’s future likelihood of PPI. Calculating the likelihood of PPI at the time of initial diagnosis based on the weight of these factors may assist clinicians with the definition of a care plan and with patient counseling. Validation of this model using other study populations would further establish its utility for clinical practice.
Transparency Declaration of funding Editorial support for this paper was funded by Endo Pharmaceuticals Inc. Declaration of financial/other relationships R.L.S. and A.L.B. have disclosed that they have no significant relationships with or financial interests in any commercial companies related to this study or article. S.B.C. and L.M. have disclosed that they are employees of Endo Pharmaceuticals Inc. CMRO peer reviewers on this manuscript have no relevant financial or other relationships to disclose. Acknowledgments The authors thank Angela M. Ginkel, of American Medical Systems Inc. (AMS), Minnetonka, MN, USA, for editorial review support.
www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion
Previous presentation: The Annual Meeting of the American Urological Association (AUA), 4–8 May 2013, San Diego, CA, USA.
Downloaded by [University of Exeter] at 12:06 11 August 2015
References 1. Feldman HA, Goldstein I, Hatzichristou DG, et al. Impotence and its medical and psychosocial correlates: results of the Massachusetts Male Aging Study. J Urol 1994;15:54-61 2. Teles AG, Carreira M, Alarca˜o V, et al. Prevalence, severity, and risk factors for erectile dysfunction in a representative sample of 3,548 Portuguese men aged 40 to 69 years attending primary healthcare centers: results of the Portuguese erectile dysfunction study. J Sex Med 2008;5:1317-24 3. Ahn TY, Park JK, Lee SW, et al. Prevalence and risk factors for erectile dysfunction in Korean men: results of an epidemiological study. J Sex Med 2007;4:1269-76 4. Aytac IA, McKinlay JB, Krane RJ. The likely worldwide increase in erectile dysfunction between 1995 and 2025 and some possible policy consequences. BJU Int 1999;84:50-6 5. Eardley I, Gentile V, Austoni E, et al. Efficacy and safety of tadalafil in a Western European population of men with erectile dysfunction. BJU Int 2004;94:871-7 6. Montorsi F, Hellstrom WJ, Valiquette L, et al. Vardenafil provides reliable efficacy over time in men with erectile dysfunction. Urology 2004;64:1187-95 7. Valiquette L, Montorsi F, Auerbach S; Vardenafil Study Group. First-dose success with vardenafil in men with erectile dysfunction and associated comorbidities: RELY-I. Int J Clin Pract 2006;60:1378-85 8. Chen KK, Paick JS, Ishii N; Vardenafil Study Group. The efficacy and safety of vardenafil in East Asian men with erectile dysfunction. J Sex Med 2007;4:753-61 9. Shabsigh R, Duval S, Shah M, et al. Efficacy of vardenafil for the treatment of erectile dysfunction in men with hypertension: a meta-analysis of clinical trial data. Curr Med Res Opin 2007;23:2453-60 10. Rubio-Aurioles E, Casabe´ A, Torres LO, et al. Efficacy and safety of tadalafil in the treatment of Latin American men with erectile dysfunction: results of integrated analyses. J Sex Med 2008;5:1965-76 11. Carvalheira AA, Pereira NM, Maroco J, Forjaz V. Dropout in the treatment of erectile dysfunction with PDE5: a study on predictors and a qualitative analysis of reasons for discontinuation. J Sex Med 2012;9:2361-9 12. Turner LA, Althof SE, Levine SB, et al. Treating erectile dysfunction with external vacuum devices: impact upon sexual, psychological and marital functioning. J Urol 1990;144:79-82 13. Derouet H, Caspari D, Rohde V, et al. Treatment of erectile dysfunction with external vacuum devices. Andrologia 1999;31(Suppl 1):89-94 14. Prabhu V, Alukal JP, Laze J, et al. Long-term satisfaction and predictors of use of intracorporeal injections for post-prostatectomy erectile dysfunction. J Urol 2013;189:238-42
! 2014 Informa UK Ltd www.cmrojournal.com
Volume 30, Number 10
October 2014
15. Coombs PG, Heck M, Guhring P, et al. A review of outcomes of an intracavernosal injection therapy programme. BJU Int 2012;110: 1787-91 16. Mydlo JH, Volpe MA, MacChia RJ. Results from different patient populations using combined therapy with alprostadil and sildenafil: predictors of satisfaction. BJU Int 2000;86:469-73 17. Salonia A, Gallina A, Zanni G, et al. Acceptance of and discontinuation rate from erectile dysfunction oral treatment in patients following bilateral nervesparing radical prostatectomy. Eur Urol 2008;53:564-70 18. Stephenson RA, Mori M, Hsieh Y, et al. Treatment of erectile dysfunction following therapy for clinically localized prostate cancer: patient reported use and outcomes from the surveillance, epidemiology, and end results prostate cancer outcomes study. J Urol 2005;174:646-50 19. Mulhall JP, Ahmed A, Branch J, Parker M. Serial assessment of efficacy and satisfaction profiles following penile prosthesis surgery. J Urol 2003;169:1429-33 20. Rajpurkar A, Dhabuwala CB. Comparison of satisfaction rates and erectile function in patients treated with sildenafil, intracavernous prostaglandin E1 and penile implant surgery for erectile dysfunction in urology practice. J Urol 2003;170:159-63 21. Bernal RM, Henry GD. Contemporary patient satisfaction rates for three-piece inflatable penile prostheses. Adv Uro 2012;2012:1-5 22. Salonia A, Ferrari M, Sacca A, et al. Delay in seeking medical help in patients with new-onset erectile dysfunction remained high over and despite the PDE5 era – an ecological study. J Sex Med 2012;9:3239-46 23. Menard J, Tremeaux JC, Faix A, et al. Erectile function and sexual satisfaction before and after penile prosthesis implantation in radical prostatectomy patients: a comparison with patients with vasculogenic erectile dysfunction. J Sex Med 2011;8:3479-86 24. Chronic Conditions Data Warehouse. CMS Chronic Condition Data Warehouse Condition Categories. Available at: http://www.ccwdata.org/cs/groups/public/ documents/document/ccw_conditioncategories.pdf [Last accessed 22 March 2013] 25. Conrad P, Leiter V. From Lydia Pinkham to Queen Levitra: direct-to-consumer advertising and medicalisation. Sociol Health Illn 2008;30:825-38 26. Segal R, Burnett AL. Erectile preservation following radical prostatectomy. Ther Adv Urol 2011;3:35-46 27. Sun P, Deftel A, Swindle R, et al. The costs of caring for erectile dysfunction in a managed care setting: evidence from a large national claims database. J Urol 2005;174:1948-52 28. Hwang TI, Yang CR, Ho WL, Chu HW. Histopathological change of corpora cavernosa after long-term intracavernous injection. Eur Urol 1991;20:301-6 29. Villarreal HG, Jones L. Outcomes of and satisfaction with the inflatable penile prosthesis in the elderly male. Adv Urol 2012;2012:240963 30. Wiser H, Kottwitz M, Wilson S, Kohler T. Interesting trends in penile prosthesis usage. Presented at the 2013 Annual Meeting of the American Urological Association, 7 May 2013, San Diego, California, USA
Prediction model for penile prosthesis implantation Segal et al.
2137
Current Medical Research and Opinion Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/icmo20
Prediction model for penile prosthesis implantation for erectile dysfunction management a
b
bc
Robert L. Segal , Stephen B. Camper , Larry Ma
& Arthur L. Burnett
a
a
The James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions Baltimore, MDUSA b
Health Economics & Outcomes Research (HEOR), Endo Pharmaceuticals Inc. Malvern, PAUSA
c
Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania Perelman School of Medicine Philadelphia, PAUSA Published online: 26 May 2015.
Click for updates To cite this article: Robert L. Segal, Stephen B. Camper, Larry Ma & Arthur L. Burnett (2014) Prediction model for penile prosthesis implantation for erectile dysfunction management, Current Medical Research and Opinion, 30:10, 2131-2137 To link to this article: http://dx.doi.org/10.1185/03007995.2014.936188
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Current Medical Research & Opinion 0300-7995 doi:10.1185/03007995.2014.936188
Vol. 30, No. 10, 2014, 2131–2137
Article ST-0125/936188 All rights reserved: reproduction in whole or part not permitted
Downloaded by [University of Exeter] at 12:06 11 August 2015
Original article Prediction model for penile prosthesis implantation for erectile dysfunction management
Robert L. Segal The James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions, Baltimore, MD, USA
Stephen B. Camper Health Economics & Outcomes Research (HEOR), Endo Pharmaceuticals Inc., Malvern, PA, USA
Abstract Objective: Penile prosthesis surgery is indicated based on undesirability, contraindication or ineffectiveness of nonsurgical options for erectile dysfunction. This definitive treatment is often delayed after initial diagnosis. Our objective was to develop a prediction tool based on a patient’s clinical history to determine likelihood of ultimately receiving a penile prosthesis.
Larry Ma Health Economics & Outcomes Research (HEOR), Endo Pharmaceuticals Inc., Malvern, PA, USA Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA
Research design and methods: This retrospective analysis used claims data from Commercial and Medicare supplemental databases. Inclusion criteria were 18 years of age with 1 year of continuous enrollment at the first diagnosis of erectile dysfunction. Patients’ demographics, co-morbidities and erectile dysfunction therapy were derived based on enrollment, medical and prescription histories.
Arthur L. Burnett The James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions, Baltimore, MD, USA Address for correspondence: Arthur L. Burnett MD MBA, 600 North Wolfe Street, Marburg 407, Baltimore, MD 21287, USA. Tel.: +1 410 614 3986; Fax: +1 410 614 3695; [email protected] Keywords: Epidemiology – Health Services – Medical co-morbidities – Risk factors Accepted: 24 May 2014; published online: 1 July 2014 Citation: Curr Med Res Opin 2014; 30:2131–7
Main outcome measures: The Cox proportional hazards model with stepwise selection was used to identify and quantify (using relative risk) factors associated with a future penile prosthesis implant. Co-morbidities and therapies present prior to the index erectile dysfunction diagnosis were analyzed as fixed covariates. Results: Approximately 1% of the dataset’s population (N ¼ 310,303 Commercial, N ¼ 74,315 Medicare, respectively) underwent penile prosthesis implantation during the study period (3928 patients in the overall population: 2405 patients [0.78%] in the Commercial and 1523 patients [2.05%] in the Medicare population). Factors with the greatest predictive strength of penile prosthesis implantation included prostate cancer diagnosis (relative risk: 3.93, 2.29; 95% CI, 3.57–4.34, 2.03–2.6), diabetes mellitus (2.31, 1.23; 2.12–2.52, 1.1–1.37) and previous treatment with first-line therapy (1.39, 1.33; 1.28–1.5, 1.2–1.47) (all P50.01). Conclusion: The presence and extent of specific medical history factors at the time of erectile dysfunction diagnosis predict an individual’s future likelihood of penile prosthesis. Calculating the likelihood of penile prosthesis implantation based on the weight of these factors may assist clinicians with the definition of a care plan and patient counseling. The precision of the model may be limited by factors beyond medical history information that possibly influence the decision to proceed to surgery.
Introduction Erectile dysfunction (ED) is a common problem, with multiple underlying risk factors1–3, which can severely impact the quality of life for the patient as well as his partner. It has been estimated that as the population ages the worldwide ! 2014 Informa UK Ltd www.cmrojournal.com
Prediction model for penile prosthesis implantation Segal et al.
2131
Downloaded by [University of Exeter] at 12:06 11 August 2015
Current Medical Research & Opinion Volume 30, Number 10
October 2014
prevalence of ED will double, to over 300 million men by 20254. For patients experiencing ED, first-line therapy typically includes one or more trials with a phosphodiesterase-5 inhibitor (PDE5i). These agents result in successful penetrative intercourse in the order of 80%5–9. For those in whom oral pharmacotherapy is ineffective or contraindicated, other options include intracavernosal penile injections (ICI), urethral suppositories (MUSE) and/or vacuum erection devices (VEDs). Unfortunately, these modalities are associated with treatment dissatisfaction10–16 and substantial dropout rates17. Implantation of penile prostheses (PP), most commonly the three-piece inflatable device, typically represents the final option for individuals who are dissatisfied with or unresponsive to the above therapies18. These devices are associated with a high degree of patient satisfaction and low rates of adverse events19–21. The time period between onset of ED symptoms and PP implantation (PPI) is variable, and may cover an extended time period22,23. During this time, the patient may experience diminished quality of life, including role limitation due to physical and emotional problems, negative perceptions of health, and relationship hardship. Given the potential of a prolonged negative impact of either untreated or unsuccessfully treated ED on quality of life and the high degree of patient satisfaction reported following PPI, it was of interest to explore strategies that may assist physicians in reducing the time period over which an individual experiences ED. We hypothesized that predicting the likelihood that a patient ultimately receives a PP may qualify as one such strategy. The objective of this study was to determine the capacity of factors and/or combinations of factors identified in an ED patient’s medical history to predict future receipt of a PP.
Patients and methods Data source This was a retrospective cohort study using healthcare claims derived from the MarketScan Commercial and Medicare Supplemental Databases (Truven Health, Ann Arbor, MI, USA), which represents the healthcare experiences of approximately 110 million lives from 2000 to 2010. The sources for the claims are employers, health plans, and Medicare programs. Data include enrollment, medical and prescription records integrated from all care providers. The Commercial database is representative of the commercially insured population in the United States, while the Medicare Supplemental database is representative of Medicare beneficiaries with supplemental coverage, which is roughly 30% of the Medicare population. 2132
Prediction model for penile prosthesis implantation Segal et al.
Patient selection Patients with a claim with a diagnosis of ED (ICD-9-CM code 607.84: Impotence of organic origin) were first identified. The first diagnosis was treated as the index date. Patients had to be 418 years old and required to have 12 months’ continuous enrollment prior to the index ED diagnosis date to qualify as a newly diagnosed patient. Patients were excluded if they did not have both medical and prescription coverage for the entire study period, or had PP at or before the index date.
Patient characteristics Patient demographics such as age, gender and census region were derived based on their enrollment records, and co-morbidities and ED treatments were derived based on medical and prescription histories. The list of co-morbidities, medical/surgical procedures and medication prescriptions (collectively termed ‘Factors’) was constructed based on review of the ED literature. ICD-9 and Current Procedure Terminology (CPT) codes were adapted from the Centers for Medicare and Medicaid (CMS) Chronic Care Warehouse24. Non-surgical ED treatments were classified as first- and second-line: first-line treatments included PDE5i and testosterone replacement therapy, and second-line treatments included ICI, MUSE and VED.
Statistical analysis The cumulative percentage of patients who had received a PP after ED diagnosis was estimated based on the productlimit method, and the Kaplan–Meier curves of time-toprosthesis up to 6 years (72 months) were plotted. The Cox proportional hazards model was used for the analysis of time from the index ED diagnosis to PPI. Patients without PP were censored at the end of their enrollment or end of the analysis period. The model included patient demographics, co-morbidities and ED treatments existent prior to or present at the index ED diagnosis, and they were analyzed as fixed covariates. Stepwise selection was used to identify and quantify (using relative risk) factors associated with future PPI; the criterion for entrance into the model was P50.20 and the criterion for retention in the model was P50.10. Though the emphasis of the study was the estimation of the relative risk, the baseline probability of receiving a PP for an ED patient of a certain age (60 years for Commercial and 70 years for Medicare) without any co-morbidity was also estimated. This was necessary to provide context so as to provide a quantitative estimate of overall probability of requiring a PP based on clinical status. The baseline probability of receiving a PP was calculated from the Cox model via the Nelson–Aalen empirical cumulative www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion
hazard function. Probabilities (absolute risks) of PPI for patients can be modeled based on their age, their respective co-morbidities, and the baseline probability at different time points after being diagnosed with ED.
Results
October 2014
therapy. In general, similar profiles were observed when the overall population was segmented by type of insurance, although the frequencies were typically higher in the Medicare group, most likely reflective of the older age of this cohort.
Factor relative risks
Factor frequency at the time of initial ED diagnosis
Downloaded by [University of Exeter] at 12:06 11 August 2015
Volume 30, Number 10
The mean age at index ED diagnosis was 51.2 years and 71.8 years for the Commercial and Medicare populations, respectively. The average length of continuous enrollment was 5.9 and 7.0 years. The frequencies with which pre-identified co-morbid conditions, use of ED therapies and medical histories were observed at the time of initial ED diagnosis in men who eventually received a PP are presented in Table 1. Conditions with documented impact on the coronary, cerebral and/or peripheral vasculature were common (Table 1). In addition, 39% of the overall population had already received a prescription for first-line ED
A total of 3928 patients (1.02%) in the overall population ultimately received a PP, comprising 2405 (0.78%) in the Commercial and 1523 (2.05%) in the Medicare population. Adjusting for censoring, 1.61% of patients were estimated to have received a PP at 6 years following a diagnosis of ED; the percentages were 1.32% and 2.73% for Commercial and Medicare, respectively (Figure 1). The baseline probabilities of receiving a PP for patients with ED only are presented in Table 2. The relative risks (RRs) for the Commercial and Medicare groups were similar (Table 3). The factor with the highest relative risk when present at the time of ED diagnosis was Peyronie’s disease (RR ¼ 4.39). History of prostate cancer (RR ¼ 2.98), priapism (RR ¼ 2.66), and
Table 1. Patient baseline characteristics, co-morbidities and medical history. Overall Population n ¼ 384,618 Age at ED Diagnosis, Mean (SD) Age Category 18–44 45–54 55–64 65–74 75þ Census Region Northeast Region North Central Region South Region West Region Years of Continuous Enrollment, Mean (SD) Co-morbidities and Medical History Cardiovascular Disease Diabetes Mellitus Hypertension Hyperlipidemia Anxiety and related disorders Depression Psychosexual Dysfunction Low Testosterone Peyronie’s Disease Priapism Prostate Cancer Multiple Sclerosis Polyneuropathy Spinal Cord Injury Radical Prostatectomy Renal Transplant Arterial Bypass Surgery Primary ED Therapy Secondary ED Therapy
55.2 (12.0)
Commercial Population n ¼ 310,303
Medicare Population n ¼ 74,315
51.2 (9.4)
71.8 (6.0)
69,073 (18%) 105,997 (28%) 137,739 (36%) 49,245 (13%) 22,564 (6%)
69,059 (22%) 105,601 (34%) 135,513 (44%) 118 (0%) 12 (0%)
14 (0%) 396 (1%) 2226 (3%) 49,127 (66%) 22,552 (30%)
34,868 (9%) 93,504 (24%) 175,270 (46%) 79,677 (21%) 6.1 (3.0)
29,211 (9%) 71,656 (23%) 147,952 (48%) 60,304 (19%) 5.9 (3.0)
5657 (8%) 21,848 (29%) 27,318 (37%) 19,373 (26%) 7.0 (2.9)
30,013 (7.8%) 80,700 (21.0%) 192,354 (50.0%) 202,156 (52.6%) 26,283 (6.8%) 39,388 (10.2%) 21,595 (5.6%) 27,595 (7.2%) 3194 (0.8%) 656 (0.2%) 36,017 (9.4%) 1292 (0.3%) 11,328 (2.9%) 486 (0.1%) 15,563 (4.0%) 813 (0.2%) 55,271 (14.4%) 148,826 (38.7%) 25,008 (6.5%)
15,785 (5.1%) 58,619 (18.9%) 142,303 (45.9%) 160,252 (51.6%) 23,237 (7.5%) 33,978 (10.9%) 18,807 (6.1%) 23,278 (7.5%) 2763 (0.9%) 576 (0.2%) 21,341 (6.9%) 1132 (0.4%) 7345 (2.4%) 371 (0.1%) 11,827 (3.8%) 637 (0.2%) 30,667 (9.9%) 113,848 (36.7%) 11,786 (3.8%)
14,228 (19.1%) 22,081 (29.7%) 50,051 (67.3%) 41,904 (56.4%) 3046 (4.1%) 5410 (7.3%) 2788 (3.8%) 4317 (5.8%) 431 (0.6%) 80 (0.1%) 14,676 (19.7%) 160 (0.2%) 3983 (5.4%) 115 (0.2%) 3736 (5.0%) 176 (0.2%) 24,594 (33.1%) 34,978 (47.1%) 13,222 (17.8%)
ED, erectile dysfunction
! 2014 Informa UK Ltd www.cmrojournal.com
Prediction model for penile prosthesis implantation Segal et al.
2133
Current Medical Research & Opinion Volume 30, Number 10
October 2014
Cumulative Percentage (%)
4
3
2
Downloaded by [University of Exeter] at 12:06 11 August 2015
1
Population:
Commercial (n=310303)
Medicare (n=74315)
0 0
6
12
18
24
30
36
42
48
54
60
66
72
Months Since ED Diagnosis
Figure 1. Kaplan–Meier estimates of the cumulative percentage of patients with penile prosthesis, all erectile dysfunction (ED) diagnosed patients, by population.
Table 2. Baseline probability of penile prosthesis implantation for patients with ED only.
60 Years Old (Commercial) 70 Years Old (Medicare)
Year 1
Year 2
Year 3
Year 4
Year 5
Year 6
Year 7
Year 8
0.28% 0.99%
0.39% 1.22%
0.46% 1.35%
0.53% 1.50%
0.60% 1.59%
0.67% 1.69%
0.75% 1.74%
0.79% 1.79%
The probabilities shown are from the time of the index diagnosis of ED. ED, erectile dysfunction
receipt of second-line ED therapies (RR ¼ 2.07) were also associated with an increased likelihood. In the Commercial population, patents in the 55–64 years group had the highest likelihood of getting PPI, compared with either younger patients (18–44 years, RR ¼ 0.35; 45– 54 years, RR ¼ 0.76) or older patients (65–74 years, RR ¼ 0.86). In the Medicare population, where most patients are older than 65 years, the RR declines with advancing age, which may reflect a host of ED treatment risk profile factors influencing a decreased utilization of PP surgery in this age category.
Prediction calculation The RRs for each factor contributing to the likelihood of future PPI are multiplicative, and may be employed to predict the likelihood of ultimately undergoing PPI. For example, a 60 year old patient in the Commercial population presenting at ED diagnosis with a history of prostate cancer (RR ¼ 3.93) and diabetes mellitus (DM) (RR ¼ 2.31) and who had already received first-line 2134
Prediction model for penile prosthesis implantation Segal et al.
therapy (RR ¼ 1.39) is 12.6 (3.93 2.31 1.39) times as likely to receive a PP as an ED patient of the same age without these factors. Given that a 60 year old ED patient without any co-morbidities has a baseline probability of receiving a PP of approximately 0.67% at 6 years post ED diagnosis, the patient with the above profile will have an absolute risk of PPI of 8.4% (12.6 0.67%). By contrast, in the Medicare population example, a 70 year old patient with the same co-morbidities would have a 3.7 (2.29 1.23 1.33) times chance of receiving a PP compared to the equivalent ED patient without these factors. Similar calculations may be performed for any combination of co-morbidities.
Discussion The aim of this study was to determine whether a clinically useful predictive model for eventual PPI can be devised based on a patient’s clinical variables. The potential utility of such a tool is to provide the clinician and patient with a sense of magnitude of the risk of PPI for maintaining www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion
Volume 30, Number 10
October 2014
Table 3. Relative risk associated with selected demographic, co-morbidity and medical history variables. Overall Population (n ¼ 384,618)
Downloaded by [University of Exeter] at 12:06 11 August 2015
Variable
Peyronie’s Disease Prostate Cancer Priapism Other Unspecified Disorders of the Penis Second-Line ED Therapy Multiple Sclerosis Spinal Cord Injury Diabetes Mellitus Renal Transplant Arterial Bypass Procedure First-Line ED Therapy Psychosexual Dysfunction, Unspecified Sexual Dysfunction Radical Prostatectomy Depression Polyneuropathy Cardiovascular Disease Hypertension Anxiety and Anxiety-related Diagnoses Dyslipidemia Age Group 18–44 45–54 55–64 65–74 75þ
Commercial (n ¼ 310,303)
Medicare (n ¼ 74,315)
Relative Risk (95% CI)
p Value
Relative Risk (95% CI)
p Value
Relative Risk (95% CI)
p Value
4.39 (3.71, 5.19) 2.98 (2.72, 3.26) 2.66 (1.87, 3.79) 2.27 (1.88, 2.73)
50.001 50.001 50.001 50.001
4.47 (3.68, 5.42) 3.93 (3.57, 4.34) 2.14 (1.41, 3.25) 2.26 (1.82, 2.80)
50.001 50.001 50.001 50.001
3.95 (2.79, 5.59) 2.29 (2.03, 2.60) 4.02 (2.08, 7.75) 2.04 (1.41, 2.95)
50.001 50.001 50.001 50.001
2.07 (1.91, 2.24) 1.95 (1.29, 2.94) 1.87 (1.01, 3.49) 1.84 (1.71, 1.97) 1.69 (1.09, 2.62) 1.52 (1.41, 1.64) 1.42 (1.34, 1.52) 1.41 (1.24, 1.60)
50.001 50.001 0.04 50.001 0.02 50.001 50.001 50.001
3.01 (2.72, 3.34) 1.74 (1.04, 2.89) 2.13 (0.95, 4.74) 2.31 (2.12, 2.52) 1.69 (1.01, 2.82) 1.48 (1.33, 1.65) 1.39 (1.28, 1.50) 1.46 (1.26, 1.69)
50.001 0.03 0.06 50.001 0.04 50.001 50.001 50.001
1.38 (1.23, 1.55) 1.89 (0.94, 3.81)
50.001 0.07
1.23 (1.10, 1.37)
50.001
1.51 (1.35, 1.67) 1.33 (1.20, 1.47) 1.32 (1.04, 1.67)
50.001 50.001 0.02
1.40 (1.25, 1.56) 1.33 (1.20, 1.48) 1.26 (1.09, 1.45) 1.16 (1.05, 1.27) 1.12 (1.05, 1.20) 0.82 (0.70, 0.95)
50.001 50.001 50.001 50.001 50.001 50.001
1.27 (1.12, 1.43) 1.24 (1.03, 1.51) 1.21 (1.06, 1.39) 1.17 (1.07, 1.28)
50.001 0.02 50.001 50.001
1.55 (1.31, 1.84) 1.26 (1.06, 1.51) 1.24 (1.01, 1.52)
50.001 0.01 0.04
0.74 (0.56, 0.99)
0.04
0.72 (0.68, 0.77)
50.001
0.76 (0.69, 0.82)
50.001
0.68 (0.61, 0.75)
50.001
0.28 (0.23, 0.33) 0.67 (0.61, 0.73) Reference 1.19*** (1.10, 1.29) 0.66*** (0.58, 0.75)
50.001 50.001
0.35 (0.29, 0.42) 0.76 (0.69, 0.83) 0.86 (0.12, 6.12)
2.41** (2.02, 2.86) 2.41** (2.02, 2.86) 2.41** (2.02, 2.86) Reference 0.56 (0.49, 0.64)
50.001 50.001 50.001
50.001 50.001
50.001 50.001 Reference 0.88 0.93
50.001
Stepwise regression was used to select the predictors in the final model. The statistical significance of these variables had to be50.2 to be entered, and50.1 to be retained in the multivariate regression. **Age groups 18–44, 45–54 and 55–64 were combined for the Medicare population analysis. ***This reflects a combined effect of advanced age and Medicare insurance as almost all Medicare members are over 65 and almost all Commercial members are under 65. ED, erectile dysfunction
sexual function. Specifically, our focus was to quantitate and assign weight to co-morbid conditions present at the time of ED diagnosis. We demonstrated that each co-morbidity carries a definable relative probability for eventual PPI. We postulate that these risks are multiplicative, and calculating the combined risk will aid decisionmaking regarding whether to pursue PPI and perhaps the timing of such a decision. A significant time delay often exists between the onset of ED symptoms and both presenting to a clinician to discuss the symptoms and instituting treatment. Salonia et al.22 reported on 619 patients seeking medical help for new-onset ED in the PDE5i era, where the mean delay prior to presentation was 30.2 months (median 12.0; range 5–300). This delay seems surprising given the likely waning of the taboo associated with openly discussing ED, as well as the explosion of direct-to-consumer advertising for ED treatments in both electronic and print media25. In another study comprising a cohort of men who underwent PPI, the mean duration of ED prior to surgery was 60 months23. All of these patients ! 2014 Informa UK Ltd www.cmrojournal.com
had previously tried at least one other ED treatment, which conceivably proved unsuccessful enough to warrant PP surgery. It is difficult to determine the exact negative impact that unsuccessful ED therapy has. Beyond the possible treatment complications for each of the therapies26 and financial cost issues27, the opportunity cost of ED on other aspects of life (such as other health problems [i.e., depression, relationship discord] and time missed from work) must be considered. Indeed, in a survey of men who underwent ED therapy post-radical prostatectomy, those who discontinued treatment cited ‘treatment effect below expectation’ and ‘loss of interest in sex’ in 84% and 16% of cases, respectively17. Could this cited loss of interest be the result of this lack of treatment success? PPI typically represents the last resort for ED treatment, when other therapies are ineffective or if the patient is disinclined to pursue them. Multiple factors, including the invasiveness and irreversibility of PPI, or clinician unfamiliarity, may account for this neglect. On this Prediction model for penile prosthesis implantation Segal et al.
2135
Downloaded by [University of Exeter] at 12:06 11 August 2015
Current Medical Research & Opinion Volume 30, Number 10
October 2014
basis, the clinician may fail to mention PPI as a treatment option when counseling a patient for ED. This is in spite of the myriad reports which extol the merits, associated satisfaction and safety of prosthetic devices18–21. The implications of deferring PPI are not fully appreciated. Certain treatments, such as ICI, may induce corporal scarring28 and make subsequent PPI more difficult. Additionally, as the patient ages, co-morbid conditions may make elective surgical intervention less advisable, or preclude surgery altogether. Finally, although age alone should not be the main determinant in deciding for or against PPI29, in our experience older patients take a longer time to recover from surgery and display greater difficulty in learning to operate their devices. As such, delaying PPI may present increasingly challenging management dilemmas. Our research advantages the patient who may elect to forego other ED treatment options and proceed directly to PPI upon understanding that the overall likelihood of ultimately requiring a PP is substantial. This knowledge enables him to avoid possible drawbacks of other ED therapies and derive the benefits of the PP immediately. Notably, this approach does not contradict any ED management guidelines, as no urological society mandates a trial of non-surgical ED therapies prior to proceeding with PPI. This decision may facilitate earlier restoration of erectile function with a treatment option that affords high satisfaction22. The advantages of this study include the large population analyzed and the diversity of the sample. In particular, this methodological approach obviates concerns that an individual clinical practice uniquely determines research outcomes. We do acknowledge possible limitations of this study. The relative likelihood of ultimate PPI depends on the baseline prevalence of surgery for that population, and our ‘absolute’ calculation is based on a relatively small proportion of the sampled population which underwent PPI. This same effect may explain the lower RR for the oldest age category, based on the relatively less frequent rate of PPI performed in our population. A patient’s decision to proceed with PPI may not be solely based on his co-morbidity profile, which limits the precision of this model. Other factors, such as insurance coverage status, relationship issues, concerns about undergoing surgery and cultural mores may influence the decision to proceed with PPI independent of co-morbid status. The factor of Commercial or Medicare insurance was considered in our model but was not included in the final model because it was confounded by age group (i.e., almost all Medicare members are 465). It may be assumed that this tool is most useful for a patient motivated to maintain sexual function whereby the surgery is reimbursed by his insurance provider. We were unable to assess certain factors, such as duration and severity of the analyzed co-morbidities, or other demographic characteristics such as marital 2136
Prediction model for penile prosthesis implantation Segal et al.
status, due to the unavailability of such information in a claims database. Finally, though a patient may be at greater risk for ultimately requiring PPI, this study does not address the likelihood of success with other specific ED treatment modalities. Another issue to discuss is the utility of the overall patient population. Although the Commercial and Medicare samples in our study were representative of their respective populations, the simple pooling of these may not represent a typical American patient diagnosed with ED. However, we elected to present the data from the overall population to give readers the opportunity to assess the robustness of the relative risks associated with various co-morbidities (Table 3). To predict the probability of PPI for American patients diagnosed with ED, baseline probabilities (Table 2) should be used in conjunction with the relative risks for the respective Commercial or Medicare population. We acknowledge that the data is derived from a purely American population source. We feel, however, that because 480% of penile prostheses worldwide are implanted in the United States30, the data presented (particularly from the overall population) can be assumed to apply to ED patients worldwide.
Conclusion The presence of specific medical history factors at the time of ED diagnosis predicts an individual’s future likelihood of PPI. Calculating the likelihood of PPI at the time of initial diagnosis based on the weight of these factors may assist clinicians with the definition of a care plan and with patient counseling. Validation of this model using other study populations would further establish its utility for clinical practice.
Transparency Declaration of funding Editorial support for this paper was funded by Endo Pharmaceuticals Inc. Declaration of financial/other relationships R.L.S. and A.L.B. have disclosed that they have no significant relationships with or financial interests in any commercial companies related to this study or article. S.B.C. and L.M. have disclosed that they are employees of Endo Pharmaceuticals Inc. CMRO peer reviewers on this manuscript have no relevant financial or other relationships to disclose. Acknowledgments The authors thank Angela M. Ginkel, of American Medical Systems Inc. (AMS), Minnetonka, MN, USA, for editorial review support.
www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion
Previous presentation: The Annual Meeting of the American Urological Association (AUA), 4–8 May 2013, San Diego, CA, USA.
Downloaded by [University of Exeter] at 12:06 11 August 2015
References 1. Feldman HA, Goldstein I, Hatzichristou DG, et al. Impotence and its medical and psychosocial correlates: results of the Massachusetts Male Aging Study. J Urol 1994;15:54-61 2. Teles AG, Carreira M, Alarca˜o V, et al. Prevalence, severity, and risk factors for erectile dysfunction in a representative sample of 3,548 Portuguese men aged 40 to 69 years attending primary healthcare centers: results of the Portuguese erectile dysfunction study. J Sex Med 2008;5:1317-24 3. Ahn TY, Park JK, Lee SW, et al. Prevalence and risk factors for erectile dysfunction in Korean men: results of an epidemiological study. J Sex Med 2007;4:1269-76 4. Aytac IA, McKinlay JB, Krane RJ. The likely worldwide increase in erectile dysfunction between 1995 and 2025 and some possible policy consequences. BJU Int 1999;84:50-6 5. Eardley I, Gentile V, Austoni E, et al. Efficacy and safety of tadalafil in a Western European population of men with erectile dysfunction. BJU Int 2004;94:871-7 6. Montorsi F, Hellstrom WJ, Valiquette L, et al. Vardenafil provides reliable efficacy over time in men with erectile dysfunction. Urology 2004;64:1187-95 7. Valiquette L, Montorsi F, Auerbach S; Vardenafil Study Group. First-dose success with vardenafil in men with erectile dysfunction and associated comorbidities: RELY-I. Int J Clin Pract 2006;60:1378-85 8. Chen KK, Paick JS, Ishii N; Vardenafil Study Group. The efficacy and safety of vardenafil in East Asian men with erectile dysfunction. J Sex Med 2007;4:753-61 9. Shabsigh R, Duval S, Shah M, et al. Efficacy of vardenafil for the treatment of erectile dysfunction in men with hypertension: a meta-analysis of clinical trial data. Curr Med Res Opin 2007;23:2453-60 10. Rubio-Aurioles E, Casabe´ A, Torres LO, et al. Efficacy and safety of tadalafil in the treatment of Latin American men with erectile dysfunction: results of integrated analyses. J Sex Med 2008;5:1965-76 11. Carvalheira AA, Pereira NM, Maroco J, Forjaz V. Dropout in the treatment of erectile dysfunction with PDE5: a study on predictors and a qualitative analysis of reasons for discontinuation. J Sex Med 2012;9:2361-9 12. Turner LA, Althof SE, Levine SB, et al. Treating erectile dysfunction with external vacuum devices: impact upon sexual, psychological and marital functioning. J Urol 1990;144:79-82 13. Derouet H, Caspari D, Rohde V, et al. Treatment of erectile dysfunction with external vacuum devices. Andrologia 1999;31(Suppl 1):89-94 14. Prabhu V, Alukal JP, Laze J, et al. Long-term satisfaction and predictors of use of intracorporeal injections for post-prostatectomy erectile dysfunction. J Urol 2013;189:238-42
! 2014 Informa UK Ltd www.cmrojournal.com
Volume 30, Number 10
October 2014
15. Coombs PG, Heck M, Guhring P, et al. A review of outcomes of an intracavernosal injection therapy programme. BJU Int 2012;110: 1787-91 16. Mydlo JH, Volpe MA, MacChia RJ. Results from different patient populations using combined therapy with alprostadil and sildenafil: predictors of satisfaction. BJU Int 2000;86:469-73 17. Salonia A, Gallina A, Zanni G, et al. Acceptance of and discontinuation rate from erectile dysfunction oral treatment in patients following bilateral nervesparing radical prostatectomy. Eur Urol 2008;53:564-70 18. Stephenson RA, Mori M, Hsieh Y, et al. Treatment of erectile dysfunction following therapy for clinically localized prostate cancer: patient reported use and outcomes from the surveillance, epidemiology, and end results prostate cancer outcomes study. J Urol 2005;174:646-50 19. Mulhall JP, Ahmed A, Branch J, Parker M. Serial assessment of efficacy and satisfaction profiles following penile prosthesis surgery. J Urol 2003;169:1429-33 20. Rajpurkar A, Dhabuwala CB. Comparison of satisfaction rates and erectile function in patients treated with sildenafil, intracavernous prostaglandin E1 and penile implant surgery for erectile dysfunction in urology practice. J Urol 2003;170:159-63 21. Bernal RM, Henry GD. Contemporary patient satisfaction rates for three-piece inflatable penile prostheses. Adv Uro 2012;2012:1-5 22. Salonia A, Ferrari M, Sacca A, et al. Delay in seeking medical help in patients with new-onset erectile dysfunction remained high over and despite the PDE5 era – an ecological study. J Sex Med 2012;9:3239-46 23. Menard J, Tremeaux JC, Faix A, et al. Erectile function and sexual satisfaction before and after penile prosthesis implantation in radical prostatectomy patients: a comparison with patients with vasculogenic erectile dysfunction. J Sex Med 2011;8:3479-86 24. Chronic Conditions Data Warehouse. CMS Chronic Condition Data Warehouse Condition Categories. Available at: http://www.ccwdata.org/cs/groups/public/ documents/document/ccw_conditioncategories.pdf [Last accessed 22 March 2013] 25. Conrad P, Leiter V. From Lydia Pinkham to Queen Levitra: direct-to-consumer advertising and medicalisation. Sociol Health Illn 2008;30:825-38 26. Segal R, Burnett AL. Erectile preservation following radical prostatectomy. Ther Adv Urol 2011;3:35-46 27. Sun P, Deftel A, Swindle R, et al. The costs of caring for erectile dysfunction in a managed care setting: evidence from a large national claims database. J Urol 2005;174:1948-52 28. Hwang TI, Yang CR, Ho WL, Chu HW. Histopathological change of corpora cavernosa after long-term intracavernous injection. Eur Urol 1991;20:301-6 29. Villarreal HG, Jones L. Outcomes of and satisfaction with the inflatable penile prosthesis in the elderly male. Adv Urol 2012;2012:240963 30. Wiser H, Kottwitz M, Wilson S, Kohler T. Interesting trends in penile prosthesis usage. Presented at the 2013 Annual Meeting of the American Urological Association, 7 May 2013, San Diego, California, USA
Prediction model for penile prosthesis implantation Segal et al.
2137
