How do you read ED evidence? Guideline grades, placebo and relative risk explained

In short: most claims about erectile dysfunction rest on three kinds of numbers, and each one is easy to misread. Guideline grades tell you how certain a panel is about the evidence, which is a different thing from how strongly it recommends something. Placebo columns tell you what happened to men who took no active drug. Relative risks compare one group with another and do not tell you your own odds (AUA Erectile Dysfunction Guideline, 2018; NCI, relative risk). This page walks through each one with real examples from the American Urological Association (AUA) guideline, an FDA drug label and a published meta-analysis. It is educational information, not medical advice, and it is the reference page behind our library of ED myths checked against the evidence, where the same tests are applied to one claim at a time.

What do Grade A, B and C mean in the AUA ED guideline?

Grade A, B and C describe how certain the AUA panel is about a body of evidence. The guideline defines them as A (high quality evidence, high certainty), B (moderate quality evidence, moderate certainty) and C (low quality evidence, low certainty) (AUA Guideline, Methodology).

The grade is about the whole body of evidence, not a single study. The guideline says evidence strength covers study quality and also study design, how consistent the findings are across studies, whether sample sizes are adequate, and how well the samples, settings and treatments generalize. Its definitions, paraphrased:

• Grade A: well-conducted, highly generalizable randomized controlled trials (RCTs), or exceptionally strong observational studies with consistent findings.
• Grade B: RCTs with some weaknesses of procedure or generalizability, or moderately strong observational studies with consistent findings.
• Grade C: RCTs with serious deficiencies of procedure or generalizability or extremely small sample sizes, or observational studies that are inconsistent, have small sample sizes, or have other problems that could confound how the data are read.

The guideline also explains what each grade means for the future. With Grade A behind a Strong or Moderate Recommendation, future research is unlikely to change confidence. With Grade B, better evidence could change confidence. With Grade C, better evidence is likely to change confidence (AUA Guideline, Methodology). A Grade C label is a statement about uncertainty. It does not say a treatment is unsafe or does nothing.

Is a Strong Recommendation the same as strong evidence?

No. The AUA rates two separate things, and a statement carries both labels. One is the evidence grade (A, B or C). The other is the type of recommendation: Strong, Moderate or Conditional, which the panel set based on the balance of benefits against risks and burdens. The guideline says all three recommendation types can be supported by any evidence grade (AUA Guideline, Methodology).

The guideline defines the three types this way. Strong and Moderate Recommendations are directive: an action should, or should not, be taken, because the net benefit or net harm is substantial (Strong) or moderate (Moderate). Conditional Recommendations are non-directive and are used when there is no apparent net benefit or harm, or when the balance is unclear. It also notes that Grade C is only rarely used to support a Strong Recommendation.

Real statements from the guideline show how the two scales combine:

Source for every row: the statement list in the AUA ED Guideline.

So a statement can be a real, directive recommendation that rests on low-certainty evidence. The testosterone row is a good example: the panel says the test should be done, and it also says better evidence is likely to change confidence. Our guide to ED and low testosterone covers what that test is for.

What are Clinical Principles and Expert Opinion?

Clinical Principles and Expert Opinion are the labels the AUA uses where evidence was insufficient to grade. The guideline says it provides this guidance where gaps in the evidence existed, with consensus reached by a modified Delphi technique if panel members disagreed. It defines a Clinical Principle as a statement about a part of clinical care that is widely agreed upon by urologists or other clinicians, for which there may or may not be evidence in the medical literature (AUA Guideline, Methodology).

Several familiar statements carry these labels. The guideline’s first statement, that men with ED should have a thorough medical, sexual and psychosocial history, a physical exam and selective lab testing, is a Clinical Principle. So is the statement that men should be counseled that ED is a risk marker for underlying cardiovascular disease. Using validated questionnaires to rate ED severity, and specialized testing for some men, are labeled Expert Opinion, as is the statement that platelet-rich plasma (PRP) therapy should be considered experimental (AUA Guideline, Statements 1, 2, 3, 5 and 25).

The label tells you what the statement rests on: clinical agreement or panel judgment rather than graded research. Our page on how ED is diagnosed explains why the guideline’s diagnosis section rests mainly on this kind of guidance, and our guide to experimental ED therapies shows how the investigational and experimental labels are used.

Why is the evidence for ED pills rated Grade B, not Grade A?

The AUA rated the evidence on outcomes and side effects of PDE5 inhibitor therapy in the general ED population as Grade B, and it explains why in detail. Most of the data came from RCTs that ranged from poor to high quality, with most rated moderate quality, usually because of inadequate information on randomization or blinding. The panel listed strengths: randomization, blinding, placebo control groups and extremely large sample sizes. It also listed two weaknesses. About 70% of studies had industry associations through author affiliations or funding, and about 75% of trials ended at three months or less of follow-up, which the panel called an important gap given that use is likely to continue much longer (AUA Guideline, PDE5 inhibitor discussion).

For special populations, the grade was C, because only limited numbers of randomized studies were available for specific subgroups (AUA Guideline).

This is the kind of reasoning worth looking for whenever you see a grade. The letter is a summary. The explanation behind it tells you what is known and what is still missing.

How do ED trials measure whether a treatment worked?

Most ED trials measure results with questionnaires and diaries that the men fill in themselves, not with a lab test. The most widely used questionnaire is the International Index of Erectile Function (IIEF), a 15-item questionnaire published in 1997 and linguistically validated in 10 languages. Its developers grouped the items into five domains: erectile function, orgasmic function, sexual desire, intercourse satisfaction and overall satisfaction (Rosen et al., Urology 1997).

The Viagra label shows how a drug trial uses these tools. Two IIEF questions, on achieving an erection sufficient for intercourse and keeping it after penetration, served as the primary endpoints. The trials also used a global question about whether erections had improved, daily diaries and an optional partner questionnaire (VIAGRA prescribing information, section 14, DailyMed). So when a page says a certain share of men “improved,” the first question is which measure the number came from.

Sources: Rosen et al., 2011 (six-item erectile function domain); Khera et al., J Sex Med 2023 (6 to 30 scale); Gupta et al., 2011 (IIEF-5); VIAGRA prescribing information, section 14 (global question and diaries).

A “percent improved” figure means little without its placebo partner. On the same label, 24% of men taking placebo answered yes to the global question, a number covered in more detail below.

How big does an improvement on the IIEF have to be to matter?

A change can be statistically significant and still be small in daily life, which is why researchers estimate a minimal clinically important difference (MCID): the smallest change that men themselves rate as a real improvement. Using 17 randomized, placebo-controlled tadalafil trials covering 3,345 men, one research group estimated the MCID for the IIEF erectile function domain at 4 points. It varied with how severe ED was at the start: 2 points for mild, 5 for moderate and 7 for severe. The authors noted that the figures came from tadalafil trials and still need to be replicated with other medicines and with non-drug treatments (Rosen et al., Eur Urol 2011).

That yardstick helps when you read a real result. A 2023 meta-analysis of 11 randomized trials found that aerobic exercise raised erectile function scores by an average of 2.8 points more than in non-exercising control groups, with gains of 2.3, 3.3 and 4.9 points in men with mild, moderate and severe ED (Khera et al., J Sex Med 2023). Putting the two papers side by side is our comparison, not the authors’: the mild-ED gain sits just above the 2-point threshold, while the moderate and severe gains sit below the 5 and 7 point thresholds. That does not make the exercise result meaningless. It is an average across groups, the thresholds were built from drug trials, and exercise has benefits beyond erections. It does show why a headline number needs a “compared with what” before it means much. The exercise paper also lists an author affiliation with a medical device company, the kind of detail the AUA panel tracked when it noted that about 70% of PDE5 inhibitor studies had industry ties. Our guide to exercise and ED covers what the trials tested.

What does the placebo column in a side-effect table tell you?

The placebo column shows how often men who took no active drug reported the same symptom in the same trials. The National Cancer Institute (NCI) defines a placebo as an inactive substance or other intervention that looks the same as, and is given the same way as, the active drug or treatment being tested, and says the effects of the active drug are compared to the effects of the placebo (NCI, placebo).

The FDA label for Viagra (sildenafil) shows how this works. Table 2 covers flexible-dose, placebo-controlled trials lasting two to twenty-six weeks, in which men took the drug as needed. It lists 734 men on Viagra and 725 on placebo (VIAGRA prescribing information, section 6.1, DailyMed):

The placebo numbers are not zero. Some men in the placebo group reported headache, indigestion and a stuffy nose without taking the active drug. Reading both columns together gives a fuller picture than reading either one alone.

The same label reports that in placebo-controlled studies, 2.5% of men stopped Viagra because of side effects, against 2.3% on placebo, a difference it describes as not significant. It also sets an honest limit on these tables: because trials are run under widely varying conditions, side-effect rates from one drug’s trials cannot be directly compared to rates in another drug’s trials, and may not reflect the rates seen in everyday practice (VIAGRA prescribing information, DailyMed). Our guide to sildenafil side effects has the full table.

Do men taking a placebo report better erections too?

Yes, some men in placebo groups do report improvement. In four randomized, double-blind, placebo-controlled fixed-dose studies on the Viagra label, covering 1,797 men over 12 to 24 weeks, 24% of men on placebo reported improved erections in response to a global question. In the sildenafil groups the figure ranged from 63% to 82% across the doses studied (VIAGRA prescribing information, section 14, DailyMed). Those numbers describe the men in those trials. They are not a prediction for any one person.

The AUA guideline draws a lesson from this. It notes that placebo effects reliably occur in trials of PDE5 inhibitors even when the men meet diagnostic criteria for organically caused ED, and says these effects suggest that even men with a physical cause are likely to have unmet needs for psychosocial and relationship support during treatment (AUA Guideline, Statement 6 discussion). It makes a similar point about side effects: men treated with radiotherapy for prostate cancer reported high rates of side effects in placebo groups too, which the panel said suggests they may have heightened sensitivity to body sensations and may have unmet needs for psychosocial support (AUA Guideline, PDE5 inhibitor discussion). Our guide to ED vs performance anxiety covers how the mind and body overlap.

What is the difference between relative risk and absolute risk?

Relative risk compares groups. Absolute risk is the chance of something happening at all. The NCI defines relative risk as a measure of the risk of an event in one group compared to the risk of the same event in another group, and absolute risk as a measure of the risk of a certain event happening (NCI, relative risk; NCI, absolute risk).

A widely quoted ED number shows why the difference matters. A 2011 meta-analysis of 12 prospective cohort studies with 36,744 participants found that men with ED had a combined relative risk of 1.48 for cardiovascular disease compared with the reference group (Dong et al., J Am Coll Cardiol 2011). That is often written as “48% higher risk.” It means higher than the comparison group. It does not mean 48% of men with ED will develop cardiovascular disease. Your own absolute risk depends on where you start. The NCI’s cancer epidemiology division puts it plainly: for clinical and public health uses, the absolute risk of disease is typically more useful than relative risk (NCI DCEG, Absolute Risk Modeling).

The AUA guideline gives an example of both numbers side by side. Discussing a rare eye condition called NAION (sudden vision loss in one eye), it says several studies have suggested PDE5 inhibitor use is associated with an increased risk, but that the absolute risk is small: 3 additional cases per 100,000 men aged 50 or older. It says men in higher-risk groups should be counseled about this small increased risk, including that the absolute risk is extremely low with or without these medicines, and that the association does not imply causation (AUA Guideline, PDE5 inhibitor discussion).

What does a relative risk of 1.0 mean?

A relative risk of 1.0 means the two groups had the same rate. The AUA guideline gives a real example. Its panel pooled three trials that compared early use of a PDE5 inhibitor with placebo in men after radical prostatectomy, and got a pooled relative risk of 1.0, with a 95% confidence interval of 0.66 to 1.64, which it reads as no difference in rates of restored erectile function between the groups (AUA Guideline, Statement 11 discussion).

A number above 1.0 means the event was more common in the first group, and a number below 1.0 means it was less common (NCI, relative risk). The range around the estimate matters too: published results usually report one, as the Dong meta-analysis did for its 1.48 figure (1.25 to 1.74) (Dong et al., 2011).

What is the difference between a randomized trial and an observational study?

A randomized trial assigns treatments by chance, while an observational study watches what people already do. The NCI defines a randomized clinical trial as one in which participants are divided by chance into separate groups that compare treatments, and explains that using chance means the groups will be similar, so the effects can be compared more fairly (NCI, randomized clinical trial). In an observational study, people are observed or outcomes are measured, and no attempt is made to affect the outcome (NCI, observational study).

Sources: NCI, case-control study, cohort study and meta-analysis; VIAGRA prescribing information, section 14; Dong et al., 2011; Pizzol et al., Am J Mens Health 2019; Wu et al., J Sex Med 2022.

The NCI notes that the results of a meta-analysis are usually stronger than those of any single study (NCI, meta-analysis), but pooling cannot fix a weakness the studies share. The cannabis meta-analysis found very high variation between its five case-control studies, and its authors called for longitudinal research to confirm or refute the link (Pizzol et al., 2019). The multiple sclerosis review pooled 16 studies and said that because all of them were observational, the robustness of its evidence may be reduced (Wu et al., 2022).

Many of the questions men search about ED causes can only be studied this way, because it is rarely practical or ethical to assign people to an exposure at random. That is why our pages on porn and ED, masturbation and ED, vasectomy and ED, sleep apnea and ED and conditions often blamed for ED say “linked to” far more often than “causes.” Sometimes the two kinds of study point in different directions: our page on GLP-1 medicines and ED covers a database study and a randomized trial that did, and our page on ED drugs and Alzheimer’s research shows how early insurance-record findings looked once stricter comparisons were run.

Does an association mean one thing causes the other?

Not on its own. The AUA guideline is careful with its words here. It counsels that ED is a risk marker for cardiovascular disease, and defines risk markers as attributes that predict an increased probability of a disease but are not part of the causal pathway (AUA Guideline, Statement 3 discussion). It says the link was originally proposed on the basis of shared clinical risk factors, including high blood pressure, smoking and diabetes, and a presumed overlap in underlying processes such as inflammation, endothelial dysfunction and atherosclerosis. NIDDK lists several of the same conditions, including diabetes, high blood pressure and smoking, among the things that can lead to ED (NIDDK, Symptoms & Causes of ED).

The wording in research papers can be stronger than that. The Dong meta-analysis concludes that ED “significantly increases the risk” of cardiovascular disease, and that the increase is probably independent of conventional risk factors (Dong et al., 2011). The 2018 guideline still frames ED as a risk marker, and that is the framing this site follows. ED does not diagnose heart disease. It is a reason to check. Our guide to what ED can signal about your health covers those links.

The guideline also shows how a panel decides an association is not reliable. Several studies reported a small association between PDE5 inhibitor use and skin cancers. The panel noted that none was designed to rule out important confounders, such as more medical check-ups among men using these medicines, which could find more skin cancers, and possibly more sun exposure. It said the findings fail to convincingly satisfy most of Hill’s criteria for causation, such as strength, consistency and a dose-response gradient, and concluded there is no increased risk of skin cancers reliably associated with PDE5 inhibitor use (AUA Guideline, PDE5 inhibitor discussion).

How current is the evidence behind the AUA ED guideline?

The AUA ED guideline is dated 2018, and its literature search ran from January 1, 1965 to July 29, 2017. The guideline says data from studies published after that cut-off will be incorporated into the next version. The search found 999 relevant publications, and the draft was sent to 35 peer reviewers before approval by AUA committees and its Board of Directors (AUA Guideline, Methodology).

That date is worth knowing when you read any claim that cites the guideline. It reflects the evidence up to mid-2017. Newer studies may exist that the guideline has not yet weighed, and the guideline describes itself as a clinical strategy meant to be read with an evolving understanding of ED.

How can you use this when you talk to a clinician?

You can use these ideas to ask clearer questions, not to reach a diagnosis on your own. The AUA panel calls shared decision-making the cornerstone of the treatment and management of ED. It describes a process in which information about the best available evidence is shared between clinician and patient, and the man, with his partner invited where that fits, expresses his own preferences and values (AUA Guideline, Purpose and Background).

A few questions that follow from this page:

• How strong is the evidence for this option, and is the advice based on research or on clinical agreement?
• In the trials, how did the placebo group do?
• Is this risk number relative or absolute, and what does it mean for someone with my health history?
• Is this an association, or has a cause been shown?

If you are still deciding whether to book that conversation, our guide on when to see a doctor about ED covers the signs that make a visit worth prioritizing, and our overview of ED treatment options shows the grades attached to each one.

Sources

Every clinical statement on this page traces to a primary or guideline source:

• American Urological Association, Erectile Dysfunction (ED) Guideline (2018), web version. Methodology (literature search dates, 999 publications, evidence Grades A, B and C, recommendation types, Clinical Principles and Expert Opinion, peer review); Statements 1 to 5, 8, 11, 14, 18, 23 and 25 with their labels; the Statement 3 risk marker discussion; the PDE5 inhibitor discussion (Grade B rationale, NAION absolute risk, skin cancer and Hill’s criteria, placebo-group side effects after radiotherapy); the Statement 11 pooled relative risk; the Statement 6 discussion on placebo effects; the Purpose and Background on shared decision-making: auanet.org, ED guideline
• VIAGRA (sildenafil citrate) prescribing information, Viatris, revised 11/2023, DailyMed. Section 6.1 (Table 2 as-needed trial side effects with placebo, discontinuation rates, limits on comparing trials) and section 14 (IIEF primary endpoints, global efficacy question, diaries, fixed-dose studies): dailymed.nlm.nih.gov, VIAGRA label
• Dong JY, Zhang YH, Qin LQ. Erectile dysfunction and risk of cardiovascular disease: meta-analysis of prospective cohort studies. J Am Coll Cardiol 2011;58(13). Twelve prospective cohort studies, 36,744 participants, relative risk 1.48 (95% CI 1.25 to 1.74) for cardiovascular disease: pubmed.ncbi.nlm.nih.gov/21920268
• National Cancer Institute, Dictionary of Cancer Terms: relative risk, absolute risk and placebo
• National Cancer Institute, Division of Cancer Epidemiology and Genetics, Absolute Risk Modeling, updated September 8, 2025. Absolute risk is typically more useful than relative risk for clinical and public health uses: dceg.cancer.gov, risk modeling
• Rosen RC, Riley A, Wagner G, et al. The international index of erectile function (IIEF): a multidimensional scale for assessment of erectile dysfunction. Urology 1997;49(6). 15 items, five domains, validated in 10 languages: pubmed.ncbi.nlm.nih.gov/9187685
• Rosen RC, Allen KR, Ni X, Araujo AB. Minimal clinically important differences in the erectile function domain of the International Index of Erectile Function scale. Eur Urol 2011;60(5). 17 tadalafil trials, 3,345 men; MCID 4 points overall, 2, 5 and 7 by baseline severity: pubmed.ncbi.nlm.nih.gov/21855209
• Khera M, Bhattacharyya S, Miller LE. Effect of aerobic exercise on erectile function: systematic review and meta-analysis of randomized controlled trials. J Sex Med 2023;20(12). 11 trials; mean difference 2.8 points on the 6 to 30 IIEF-EF scale; 2.3, 3.3 and 4.9 points by severity: pubmed.ncbi.nlm.nih.gov/37814532
• Gupta BP, et al. The effect of lifestyle modification and cardiovascular risk factor reduction on erectile dysfunction. Arch Intern Med 2011;171(20). Outcome measured with the IIEF-5: pubmed.ncbi.nlm.nih.gov/21911624
• Pizzol D, et al. Relationship between cannabis use and erectile dysfunction: a systematic review and meta-analysis. Am J Mens Health 2019. Five case-control studies, high heterogeneity, call for longitudinal research: pubmed.ncbi.nlm.nih.gov/31795801
• Wu X, et al. Erectile dysfunction in multiple sclerosis: a prevalence meta-analysis and systematic review. J Sex Med 2022;19(8). 16 studies, all observational: pubmed.ncbi.nlm.nih.gov/35697636
• National Cancer Institute, Dictionary of Cancer Terms: randomized clinical trial, observational study, case-control study, cohort study and meta-analysis
• NIDDK (National Institute of Diabetes and Digestive and Kidney Diseases), Symptoms & Causes of Erectile Dysfunction, last reviewed October 2024. Conditions and behaviors that can lead to ED: niddk.nih.gov, ED symptoms and causes